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 -Wdiv-by-zero -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
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
465 -mfix7000 -mno-crt0 -mflush-func=@var{func} -mno-flush-func}
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}
593 -mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
594 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
595 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
596 -mstack-align -mdata-align -mconst-align @gol
597 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
598 -melf -maout -melinux -mlinux -sim -sim2}
600 @item Code Generation Options
601 @xref{Code Gen Options,,Options for Code Generation Conventions}.
603 -fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
604 -ffixed-@var{reg} -fexceptions @gol
605 -fnon-call-exceptions -funwind-tables -fasynchronous-unwind-tables @gol
606 -finhibit-size-directive -finstrument-functions @gol
607 -fcheck-memory-usage -fprefix-function-name @gol
608 -fno-common -fno-ident -fno-gnu-linker @gol
609 -fpcc-struct-return -fpic -fPIC @gol
610 -freg-struct-return -fshared-data -fshort-enums @gol
611 -fshort-double -fvolatile @gol
612 -fvolatile-global -fvolatile-static @gol
613 -fverbose-asm -fpack-struct -fstack-check @gol
614 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
615 -fargument-alias -fargument-noalias @gol
616 -fargument-noalias-global -fleading-underscore}
620 * Overall Options:: Controlling the kind of output:
621 an executable, object files, assembler files,
622 or preprocessed source.
623 * C Dialect Options:: Controlling the variant of C language compiled.
624 * C++ Dialect Options:: Variations on C++.
625 * Objective-C Dialect Options:: Variations on Objective-C.
626 * Language Independent Options:: Controlling how diagnostics should be
628 * Warning Options:: How picky should the compiler be?
629 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
630 * Optimize Options:: How much optimization?
631 * Preprocessor Options:: Controlling header files and macro definitions.
632 Also, getting dependency information for Make.
633 * Assembler Options:: Passing options to the assembler.
634 * Link Options:: Specifying libraries and so on.
635 * Directory Options:: Where to find header files and libraries.
636 Where to find the compiler executable files.
637 * Spec Files:: How to pass switches to sub-processes.
638 * Target Options:: Running a cross-compiler, or an old version of GCC.
641 @node Overall Options
642 @section Options Controlling the Kind of Output
644 Compilation can involve up to four stages: preprocessing, compilation
645 proper, assembly and linking, always in that order. The first three
646 stages apply to an individual source file, and end by producing an
647 object file; linking combines all the object files (those newly
648 compiled, and those specified as input) into an executable file.
650 @cindex file name suffix
651 For any given input file, the file name suffix determines what kind of
656 C source code which must be preprocessed.
659 C source code which should not be preprocessed.
662 C++ source code which should not be preprocessed.
665 Objective-C source code. Note that you must link with the library
666 @file{libobjc.a} to make an Objective-C program work.
669 Objective-C source code which should not be preprocessed.
672 C header file (not to be compiled or linked).
676 @itemx @var{file}.cxx
677 @itemx @var{file}.cpp
678 @itemx @var{file}.c++
680 C++ source code which must be preprocessed. Note that in @samp{.cxx},
681 the last two letters must both be literally @samp{x}. Likewise,
682 @samp{.C} refers to a literal capital C@.
685 @itemx @var{file}.for
686 @itemx @var{file}.FOR
687 Fortran source code which should not be preprocessed.
690 @itemx @var{file}.fpp
691 @itemx @var{file}.FPP
692 Fortran source code which must be preprocessed (with the traditional
696 Fortran source code which must be preprocessed with a RATFOR
697 preprocessor (not included with GCC)@.
699 @xref{Overall Options,,Options Controlling the Kind of Output, g77,
700 Using and Porting GNU Fortran}, for more details of the handling of
703 @c FIXME: Descriptions of Java file types.
710 Ada source code file which contains a library unit declaration (a
711 declaration of a package, subprogram, or generic, or a generic
712 instantiation), or a library unit renaming declaration (a package,
713 generic, or subprogram renaming declaration). Such files are also
716 @itemx @var{file}.adb
717 Ada source code file containing a library unit body (a subprogram or
718 package body). Such files are also called @dfn{bodies}.
720 @c GCC also knows about some suffixes for languages not yet included:
726 @itemx @var{file}.chi
727 CHILL source code (preprocessed with the traditional preprocessor).
733 Assembler code which must be preprocessed.
736 An object file to be fed straight into linking.
737 Any file name with no recognized suffix is treated this way.
741 You can specify the input language explicitly with the @option{-x} option:
744 @item -x @var{language}
745 Specify explicitly the @var{language} for the following input files
746 (rather than letting the compiler choose a default based on the file
747 name suffix). This option applies to all following input files until
748 the next @option{-x} option. Possible values for @var{language} are:
750 c c-header cpp-output
752 objective-c objc-cpp-output
753 assembler assembler-with-cpp
756 f77 f77-cpp-input ratfor
759 @c Also f77-version, for internal use only.
762 Turn off any specification of a language, so that subsequent files are
763 handled according to their file name suffixes (as they are if @option{-x}
764 has not been used at all).
766 @item -pass-exit-codes
767 @opindex pass-exit-codes
768 Normally the @command{gcc} program will exit with the code of 1 if any
769 phase of the compiler returns a non-success return code. If you specify
770 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
771 numerically highest error produced by any phase that returned an error
775 If you only want some of the stages of compilation, you can use
776 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
777 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
778 @command{gcc} is to stop. Note that some combinations (for example,
779 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
784 Compile or assemble the source files, but do not link. The linking
785 stage simply is not done. The ultimate output is in the form of an
786 object file for each source file.
788 By default, the object file name for a source file is made by replacing
789 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
791 Unrecognized input files, not requiring compilation or assembly, are
796 Stop after the stage of compilation proper; do not assemble. The output
797 is in the form of an assembler code file for each non-assembler input
800 By default, the assembler file name for a source file is made by
801 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
803 Input files that don't require compilation are ignored.
807 Stop after the preprocessing stage; do not run the compiler proper. The
808 output is in the form of preprocessed source code, which is sent to the
811 Input files which don't require preprocessing are ignored.
813 @cindex output file option
816 Place output in file @var{file}. This applies regardless to whatever
817 sort of output is being produced, whether it be an executable file,
818 an object file, an assembler file or preprocessed C code.
820 Since only one output file can be specified, it does not make sense to
821 use @option{-o} when compiling more than one input file, unless you are
822 producing an executable file as output.
824 If @option{-o} is not specified, the default is to put an executable file
825 in @file{a.out}, the object file for @file{@var{source}.@var{suffix}} in
826 @file{@var{source}.o}, its assembler file in @file{@var{source}.s}, and
827 all preprocessed C source on standard output.
831 Print (on standard error output) the commands executed to run the stages
832 of compilation. Also print the version number of the compiler driver
833 program and of the preprocessor and the compiler proper.
837 Use pipes rather than temporary files for communication between the
838 various stages of compilation. This fails to work on some systems where
839 the assembler is unable to read from a pipe; but the GNU assembler has
844 Print (on the standard output) a description of the command line options
845 understood by @command{gcc}. If the @option{-v} option is also specified
846 then @option{--help} will also be passed on to the various processes
847 invoked by @command{gcc}, so that they can display the command line options
848 they accept. If the @option{-W} option is also specified then command
849 line options which have no documentation associated with them will also
854 Print (on the standard output) a description of target specific command
855 line options for each tool.
859 @section Compiling C++ Programs
861 @cindex suffixes for C++ source
862 @cindex C++ source file suffixes
863 C++ source files conventionally use one of the suffixes @samp{.C},
864 @samp{.cc}, @samp{.cpp}, @samp{.c++}, @samp{.cp}, or @samp{.cxx};
865 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
866 files with these names and compiles them as C++ programs even if you
867 call the compiler the same way as for compiling C programs (usually with
868 the name @command{gcc}).
872 However, C++ programs often require class libraries as well as a
873 compiler that understands the C++ language---and under some
874 circumstances, you might want to compile programs from standard input,
875 or otherwise without a suffix that flags them as C++ programs.
876 @command{g++} is a program that calls GCC with the default language
877 set to C++, and automatically specifies linking against the C++
878 library. On many systems, @command{g++} is also
879 installed with the name @command{c++}.
881 @cindex invoking @command{g++}
882 When you compile C++ programs, you may specify many of the same
883 command-line options that you use for compiling programs in any
884 language; or command-line options meaningful for C and related
885 languages; or options that are meaningful only for C++ programs.
886 @xref{C Dialect Options,,Options Controlling C Dialect}, for
887 explanations of options for languages related to C@.
888 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
889 explanations of options that are meaningful only for C++ programs.
891 @node C Dialect Options
892 @section Options Controlling C Dialect
893 @cindex dialect options
894 @cindex language dialect options
895 @cindex options, dialect
897 The following options control the dialect of C (or languages derived
898 from C, such as C++ and Objective-C) that the compiler accepts:
905 In C mode, support all ISO C89 programs. In C++ mode,
906 remove GNU extensions that conflict with ISO C++.
908 This turns off certain features of GCC that are incompatible with ISO
909 C89 (when compiling C code), or of standard C++ (when compiling C++ code),
910 such as the @code{asm} and @code{typeof} keywords, and
911 predefined macros such as @code{unix} and @code{vax} that identify the
912 type of system you are using. It also enables the undesirable and
913 rarely used ISO trigraph feature. For the C compiler,
914 it disables recognition of C++ style @samp{//} comments as well as
915 the @code{inline} keyword.
917 The alternate keywords @code{__asm__}, @code{__extension__},
918 @code{__inline__} and @code{__typeof__} continue to work despite
919 @option{-ansi}. You would not want to use them in an ISO C program, of
920 course, but it is useful to put them in header files that might be included
921 in compilations done with @option{-ansi}. Alternate predefined macros
922 such as @code{__unix__} and @code{__vax__} are also available, with or
923 without @option{-ansi}.
925 The @option{-ansi} option does not cause non-ISO programs to be
926 rejected gratuitously. For that, @option{-pedantic} is required in
927 addition to @option{-ansi}. @xref{Warning Options}.
929 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
930 option is used. Some header files may notice this macro and refrain
931 from declaring certain functions or defining certain macros that the
932 ISO standard doesn't call for; this is to avoid interfering with any
933 programs that might use these names for other things.
935 Functions which would normally be built in but do not have semantics
936 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
937 functions with @option{-ansi} is used. @xref{Other Builtins,,Other
938 built-in functions provided by GCC}, for details of the functions
943 Determine the language standard. This option is currently only
944 supported when compiling C@. A value for this option must be provided;
950 ISO C89 (same as @option{-ansi}).
953 ISO C89 as modified in amendment 1.
959 ISO C99. Note that this standard is not yet fully supported; see
960 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
961 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
964 Default, ISO C89 plus GNU extensions (including some C99 features).
968 ISO C99 plus GNU extensions. When ISO C99 is fully implemented in GCC,
969 this will become the default. The name @samp{gnu9x} is deprecated.
973 Even when this option is not specified, you can still use some of the
974 features of newer standards in so far as they do not conflict with
975 previous C standards. For example, you may use @code{__restrict__} even
976 when @option{-std=c99} is not specified.
978 The @option{-std} options specifying some version of ISO C have the same
979 effects as @option{-ansi}, except that features that were not in ISO C89
980 but are in the specified version (for example, @samp{//} comments and
981 the @code{inline} keyword in ISO C99) are not disabled.
983 @xref{Standards,,Language Standards Supported by GCC}, for details of
984 these standard versions.
986 @item -aux-info @var{filename}
988 Output to the given filename prototyped declarations for all functions
989 declared and/or defined in a translation unit, including those in header
990 files. This option is silently ignored in any language other than C@.
992 Besides declarations, the file indicates, in comments, the origin of
993 each declaration (source file and line), whether the declaration was
994 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
995 @samp{O} for old, respectively, in the first character after the line
996 number and the colon), and whether it came from a declaration or a
997 definition (@samp{C} or @samp{F}, respectively, in the following
998 character). In the case of function definitions, a K&R-style list of
999 arguments followed by their declarations is also provided, inside
1000 comments, after the declaration.
1004 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1005 keyword, so that code can use these words as identifiers. You can use
1006 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1007 instead. @option{-ansi} implies @option{-fno-asm}.
1009 In C++, this switch only affects the @code{typeof} keyword, since
1010 @code{asm} and @code{inline} are standard keywords. You may want to
1011 use the @option{-fno-gnu-keywords} flag instead, which has the same
1012 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1013 switch only affects the @code{asm} and @code{typeof} keywords, since
1014 @code{inline} is a standard keyword in ISO C99.
1017 @opindex fno-builtin
1018 @cindex built-in functions
1019 Don't recognize built-in functions that do not begin with
1020 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1021 functions provided by GCC}, for details of the functions affected,
1022 including those which are not built-in functions when @option{-ansi} or
1023 @option{-std} options for strict ISO C conformance are used because they
1024 do not have an ISO standard meaning.
1026 GCC normally generates special code to handle certain built-in functions
1027 more efficiently; for instance, calls to @code{alloca} may become single
1028 instructions that adjust the stack directly, and calls to @code{memcpy}
1029 may become inline copy loops. The resulting code is often both smaller
1030 and faster, but since the function calls no longer appear as such, you
1031 cannot set a breakpoint on those calls, nor can you change the behavior
1032 of the functions by linking with a different library.
1034 In C++, @option{-fno-builtin} is always in effect. The @option{-fbuiltin}
1035 option has no effect. Therefore, in C++, the only way to get the
1036 optimization benefits of built-in functions is to call the function
1037 using the @samp{__builtin_} prefix. The GNU C++ Standard Library uses
1038 built-in functions to implement many functions (like
1039 @code{std::strchr}), so that you automatically get efficient code.
1043 @cindex hosted environment
1045 Assert that compilation takes place in a hosted environment. This implies
1046 @option{-fbuiltin}. A hosted environment is one in which the
1047 entire standard library is available, and in which @code{main} has a return
1048 type of @code{int}. Examples are nearly everything except a kernel.
1049 This is equivalent to @option{-fno-freestanding}.
1051 @item -ffreestanding
1052 @opindex ffreestanding
1053 @cindex hosted environment
1055 Assert that compilation takes place in a freestanding environment. This
1056 implies @option{-fno-builtin}. A freestanding environment
1057 is one in which the standard library may not exist, and program startup may
1058 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1059 This is equivalent to @option{-fno-hosted}.
1061 @xref{Standards,,Language Standards Supported by GCC}, for details of
1062 freestanding and hosted environments.
1066 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1067 options for strict ISO C conformance) implies @option{-trigraphs}.
1069 @cindex traditional C language
1070 @cindex C language, traditional
1072 @opindex traditional
1073 Attempt to support some aspects of traditional C compilers.
1078 All @code{extern} declarations take effect globally even if they
1079 are written inside of a function definition. This includes implicit
1080 declarations of functions.
1083 The newer keywords @code{typeof}, @code{inline}, @code{signed}, @code{const}
1084 and @code{volatile} are not recognized. (You can still use the
1085 alternative keywords such as @code{__typeof__}, @code{__inline__}, and
1089 Comparisons between pointers and integers are always allowed.
1092 Integer types @code{unsigned short} and @code{unsigned char} promote
1093 to @code{unsigned int}.
1096 Out-of-range floating point literals are not an error.
1099 Certain constructs which ISO regards as a single invalid preprocessing
1100 number, such as @samp{0xe-0xd}, are treated as expressions instead.
1103 String ``constants'' are not necessarily constant; they are stored in
1104 writable space, and identical looking constants are allocated
1105 separately. (This is the same as the effect of
1106 @option{-fwritable-strings}.)
1108 @cindex @code{longjmp} and automatic variables
1110 All automatic variables not declared @code{register} are preserved by
1111 @code{longjmp}. Ordinarily, GNU C follows ISO C: automatic variables
1112 not declared @code{volatile} may be clobbered.
1117 @cindex escape sequences, traditional
1118 The character escape sequences @samp{\x} and @samp{\a} evaluate as the
1119 literal characters @samp{x} and @samp{a} respectively. Without
1120 @w{@option{-traditional}}, @samp{\x} is a prefix for the hexadecimal
1121 representation of a character, and @samp{\a} produces a bell.
1124 You may wish to use @option{-fno-builtin} as well as @option{-traditional}
1125 if your program uses names that are normally GNU C built-in functions for
1126 other purposes of its own.
1128 You cannot use @option{-traditional} if you include any header files that
1129 rely on ISO C features. Some vendors are starting to ship systems with
1130 ISO C header files and you cannot use @option{-traditional} on such
1131 systems to compile files that include any system headers.
1133 The @option{-traditional} option also enables @option{-traditional-cpp}.
1135 @item -traditional-cpp
1136 @opindex traditional-cpp
1137 Attempt to support some aspects of traditional C preprocessors.
1138 See the GNU CPP manual for details.
1140 @item -fcond-mismatch
1141 @opindex fcond-mismatch
1142 Allow conditional expressions with mismatched types in the second and
1143 third arguments. The value of such an expression is void. This option
1144 is not supported for C++.
1146 @item -funsigned-char
1147 @opindex funsigned-char
1148 Let the type @code{char} be unsigned, like @code{unsigned char}.
1150 Each kind of machine has a default for what @code{char} should
1151 be. It is either like @code{unsigned char} by default or like
1152 @code{signed char} by default.
1154 Ideally, a portable program should always use @code{signed char} or
1155 @code{unsigned char} when it depends on the signedness of an object.
1156 But many programs have been written to use plain @code{char} and
1157 expect it to be signed, or expect it to be unsigned, depending on the
1158 machines they were written for. This option, and its inverse, let you
1159 make such a program work with the opposite default.
1161 The type @code{char} is always a distinct type from each of
1162 @code{signed char} or @code{unsigned char}, even though its behavior
1163 is always just like one of those two.
1166 @opindex fsigned-char
1167 Let the type @code{char} be signed, like @code{signed char}.
1169 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1170 the negative form of @option{-funsigned-char}. Likewise, the option
1171 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1173 @item -fsigned-bitfields
1174 @itemx -funsigned-bitfields
1175 @itemx -fno-signed-bitfields
1176 @itemx -fno-unsigned-bitfields
1177 @opindex fsigned-bitfields
1178 @opindex funsigned-bitfields
1179 @opindex fno-signed-bitfields
1180 @opindex fno-unsigned-bitfields
1181 These options control whether a bit-field is signed or unsigned, when the
1182 declaration does not use either @code{signed} or @code{unsigned}. By
1183 default, such a bit-field is signed, because this is consistent: the
1184 basic integer types such as @code{int} are signed types.
1186 However, when @option{-traditional} is used, bit-fields are all unsigned
1189 @item -fwritable-strings
1190 @opindex fwritable-strings
1191 Store string constants in the writable data segment and don't uniquize
1192 them. This is for compatibility with old programs which assume they can
1193 write into string constants. The option @option{-traditional} also has
1196 Writing into string constants is a very bad idea; ``constants'' should
1199 @item -fallow-single-precision
1200 @opindex fallow-single-precision
1201 Do not promote single precision math operations to double precision,
1202 even when compiling with @option{-traditional}.
1204 Traditional K&R C promotes all floating point operations to double
1205 precision, regardless of the sizes of the operands. On the
1206 architecture for which you are compiling, single precision may be faster
1207 than double precision. If you must use @option{-traditional}, but want
1208 to use single precision operations when the operands are single
1209 precision, use this option. This option has no effect when compiling
1210 with ISO or GNU C conventions (the default).
1213 @opindex fshort-wchar
1214 Override the underlying type for @samp{wchar_t} to be @samp{short
1215 unsigned int} instead of the default for the target. This option is
1216 useful for building programs to run under WINE@.
1219 @node C++ Dialect Options
1220 @section Options Controlling C++ Dialect
1222 @cindex compiler options, C++
1223 @cindex C++ options, command line
1224 @cindex options, C++
1225 This section describes the command-line options that are only meaningful
1226 for C++ programs; but you can also use most of the GNU compiler options
1227 regardless of what language your program is in. For example, you
1228 might compile a file @code{firstClass.C} like this:
1231 g++ -g -frepo -O -c firstClass.C
1235 In this example, only @option{-frepo} is an option meant
1236 only for C++ programs; you can use the other options with any
1237 language supported by GCC@.
1239 Here is a list of options that are @emph{only} for compiling C++ programs:
1242 @item -fno-access-control
1243 @opindex fno-access-control
1244 Turn off all access checking. This switch is mainly useful for working
1245 around bugs in the access control code.
1249 Check that the pointer returned by @code{operator new} is non-null
1250 before attempting to modify the storage allocated. The current Working
1251 Paper requires that @code{operator new} never return a null pointer, so
1252 this check is normally unnecessary.
1254 An alternative to using this option is to specify that your
1255 @code{operator new} does not throw any exceptions; if you declare it
1256 @samp{throw()}, G++ will check the return value. See also @samp{new
1259 @item -fconserve-space
1260 @opindex fconserve-space
1261 Put uninitialized or runtime-initialized global variables into the
1262 common segment, as C does. This saves space in the executable at the
1263 cost of not diagnosing duplicate definitions. If you compile with this
1264 flag and your program mysteriously crashes after @code{main()} has
1265 completed, you may have an object that is being destroyed twice because
1266 two definitions were merged.
1268 This option is no longer useful on most targets, now that support has
1269 been added for putting variables into BSS without making them common.
1271 @item -fno-const-strings
1272 @opindex fno-const-strings
1273 Give string constants type @code{char *} instead of type @code{const
1274 char *}. By default, G++ uses type @code{const char *} as required by
1275 the standard. Even if you use @option{-fno-const-strings}, you cannot
1276 actually modify the value of a string constant, unless you also use
1277 @option{-fwritable-strings}.
1279 This option might be removed in a future release of G++. For maximum
1280 portability, you should structure your code so that it works with
1281 string constants that have type @code{const char *}.
1283 @item -fdollars-in-identifiers
1284 @opindex fdollars-in-identifiers
1285 Accept @samp{$} in identifiers. You can also explicitly prohibit use of
1286 @samp{$} with the option @option{-fno-dollars-in-identifiers}. (GNU C allows
1287 @samp{$} by default on most target systems, but there are a few exceptions.)
1288 Traditional C allowed the character @samp{$} to form part of
1289 identifiers. However, ISO C and C++ forbid @samp{$} in identifiers.
1291 @item -fno-elide-constructors
1292 @opindex fno-elide-constructors
1293 The C++ standard allows an implementation to omit creating a temporary
1294 which is only used to initialize another object of the same type.
1295 Specifying this option disables that optimization, and forces G++ to
1296 call the copy constructor in all cases.
1298 @item -fno-enforce-eh-specs
1299 @opindex fno-enforce-eh-specs
1300 Don't check for violation of exception specifications at runtime. This
1301 option violates the C++ standard, but may be useful for reducing code
1302 size in production builds, much like defining @samp{NDEBUG}. The compiler
1303 will still optimize based on the exception specifications.
1305 @item -fexternal-templates
1306 @opindex fexternal-templates
1308 Cause @samp{#pragma interface} and @samp{implementation} to apply to
1309 template instantiation; template instances are emitted or not according
1310 to the location of the template definition. @xref{Template
1311 Instantiation}, for more information.
1313 This option is deprecated.
1315 @item -falt-external-templates
1316 @opindex falt-external-templates
1317 Similar to @option{-fexternal-templates}, but template instances are
1318 emitted or not according to the place where they are first instantiated.
1319 @xref{Template Instantiation}, for more information.
1321 This option is deprecated.
1324 @itemx -fno-for-scope
1326 @opindex fno-for-scope
1327 If @option{-ffor-scope} is specified, the scope of variables declared in
1328 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1329 as specified by the C++ standard.
1330 If @option{-fno-for-scope} is specified, the scope of variables declared in
1331 a @i{for-init-statement} extends to the end of the enclosing scope,
1332 as was the case in old versions of G++, and other (traditional)
1333 implementations of C++.
1335 The default if neither flag is given to follow the standard,
1336 but to allow and give a warning for old-style code that would
1337 otherwise be invalid, or have different behavior.
1339 @item -fno-gnu-keywords
1340 @opindex fno-gnu-keywords
1341 Do not recognize @code{typeof} as a keyword, so that code can use this
1342 word as an identifier. You can use the keyword @code{__typeof__} instead.
1343 @option{-ansi} implies @option{-fno-gnu-keywords}.
1345 @item -fno-implicit-templates
1346 @opindex fno-implicit-templates
1347 Never emit code for non-inline templates which are instantiated
1348 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1349 @xref{Template Instantiation}, for more information.
1351 @item -fno-implicit-inline-templates
1352 @opindex fno-implicit-inline-templates
1353 Don't emit code for implicit instantiations of inline templates, either.
1354 The default is to handle inlines differently so that compiles with and
1355 without optimization will need the same set of explicit instantiations.
1357 @item -fno-implement-inlines
1358 @opindex fno-implement-inlines
1359 To save space, do not emit out-of-line copies of inline functions
1360 controlled by @samp{#pragma implementation}. This will cause linker
1361 errors if these functions are not inlined everywhere they are called.
1363 @item -fms-extensions
1364 @opindex fms-extensions
1365 Disable pedantic warnings about constructs used in MFC, such as implicit
1366 int and getting a pointer to member function via non-standard syntax.
1368 @item -fno-nonansi-builtins
1369 @opindex fno-nonansi-builtins
1370 Disable built-in declarations of functions that are not mandated by
1371 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1372 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1374 @item -fno-operator-names
1375 @opindex fno-operator-names
1376 Do not treat the operator name keywords @code{and}, @code{bitand},
1377 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1378 synonyms as keywords.
1380 @item -fno-optional-diags
1381 @opindex fno-optional-diags
1382 Disable diagnostics that the standard says a compiler does not need to
1383 issue. Currently, the only such diagnostic issued by G++ is the one for
1384 a name having multiple meanings within a class.
1387 @opindex fpermissive
1388 Downgrade messages about nonconformant code from errors to warnings. By
1389 default, G++ effectively sets @option{-pedantic-errors} without
1390 @option{-pedantic}; this option reverses that. This behavior and this
1391 option are superseded by @option{-pedantic}, which works as it does for GNU C@.
1395 Enable automatic template instantiation at link time. This option also
1396 implies @option{-fno-implicit-templates}. @xref{Template
1397 Instantiation}, for more information.
1401 Disable generation of information about every class with virtual
1402 functions for use by the C++ runtime type identification features
1403 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1404 of the language, you can save some space by using this flag. Note that
1405 exception handling uses the same information, but it will generate it as
1410 Emit statistics about front-end processing at the end of the compilation.
1411 This information is generally only useful to the G++ development team.
1413 @item -ftemplate-depth-@var{n}
1414 @opindex ftemplate-depth
1415 Set the maximum instantiation depth for template classes to @var{n}.
1416 A limit on the template instantiation depth is needed to detect
1417 endless recursions during template class instantiation. ANSI/ISO C++
1418 conforming programs must not rely on a maximum depth greater than 17.
1420 @item -fuse-cxa-atexit
1421 @opindex fuse-cxa-atexit
1422 Register destructors for objects with static storage duration with the
1423 @code{__cxa_atexit} function rather than the @code{atexit} function.
1424 This option is required for fully standards-compliant handling of static
1425 destructors, but will only work if your C library supports
1426 @code{__cxa_atexit}.
1430 Emit special relocations for vtables and virtual function references
1431 so that the linker can identify unused virtual functions and zero out
1432 vtable slots that refer to them. This is most useful with
1433 @option{-ffunction-sections} and @option{-Wl,--gc-sections}, in order to
1434 also discard the functions themselves.
1436 This optimization requires GNU as and GNU ld. Not all systems support
1437 this option. @option{-Wl,--gc-sections} is ignored without @option{-static}.
1441 Do not use weak symbol support, even if it is provided by the linker.
1442 By default, G++ will use weak symbols if they are available. This
1443 option exists only for testing, and should not be used by end-users;
1444 it will result in inferior code and has no benefits. This option may
1445 be removed in a future release of G++.
1449 Do not search for header files in the standard directories specific to
1450 C++, but do still search the other standard directories. (This option
1451 is used when building the C++ library.)
1454 In addition, these optimization, warning, and code generation options
1455 have meanings only for C++ programs:
1458 @item -fno-default-inline
1459 @opindex fno-default-inline
1460 Do not assume @samp{inline} for functions defined inside a class scope.
1461 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1462 functions will have linkage like inline functions; they just won't be
1465 @item -Wctor-dtor-privacy @r{(C++ only)}
1466 @opindex Wctor-dtor-privacy
1467 Warn when a class seems unusable, because all the constructors or
1468 destructors in a class are private and the class has no friends or
1469 public static member functions.
1471 @item -Wnon-virtual-dtor @r{(C++ only)}
1472 @opindex Wnon-virtual-dtor
1473 Warn when a class declares a non-virtual destructor that should probably
1474 be virtual, because it looks like the class will be used polymorphically.
1476 @item -Wreorder @r{(C++ only)}
1478 @cindex reordering, warning
1479 @cindex warning for reordering of member initializers
1480 Warn when the order of member initializers given in the code does not
1481 match the order in which they must be executed. For instance:
1487 A(): j (0), i (1) @{ @}
1491 Here the compiler will warn that the member initializers for @samp{i}
1492 and @samp{j} will be rearranged to match the declaration order of the
1496 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
1499 @item -Weffc++ @r{(C++ only)}
1501 Warn about violations of various style guidelines from Scott Meyers'
1502 @cite{Effective C++} books. If you use this option, you should be aware
1503 that the standard library headers do not obey all of these guidelines;
1504 you can use @samp{grep -v} to filter out those warnings.
1506 @item -Wno-deprecated @r{(C++ only)}
1507 @opindex Wno-deprecated
1508 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1510 @item -Wno-non-template-friend @r{(C++ only)}
1511 @opindex Wno-non-template-friend
1512 Disable warnings when non-templatized friend functions are declared
1513 within a template. With the advent of explicit template specification
1514 support in G++, if the name of the friend is an unqualified-id (i.e.,
1515 @samp{friend foo(int)}), the C++ language specification demands that the
1516 friend declare or define an ordinary, nontemplate function. (Section
1517 14.5.3). Before G++ implemented explicit specification, unqualified-ids
1518 could be interpreted as a particular specialization of a templatized
1519 function. Because this non-conforming behavior is no longer the default
1520 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
1521 check existing code for potential trouble spots, and is on by default.
1522 This new compiler behavior can be turned off with
1523 @option{-Wno-non-template-friend} which keeps the conformant compiler code
1524 but disables the helpful warning.
1526 @item -Wold-style-cast @r{(C++ only)}
1527 @opindex Wold-style-cast
1528 Warn if an old-style (C-style) cast is used within a C++ program. The
1529 new-style casts (@samp{static_cast}, @samp{reinterpret_cast}, and
1530 @samp{const_cast}) are less vulnerable to unintended effects, and much
1533 @item -Woverloaded-virtual @r{(C++ only)}
1534 @opindex Woverloaded-virtual
1535 @cindex overloaded virtual fn, warning
1536 @cindex warning for overloaded virtual fn
1537 Warn when a function declaration hides virtual functions from a
1538 base class. For example, in:
1545 struct B: public A @{
1550 the @code{A} class version of @code{f} is hidden in @code{B}, and code
1558 will fail to compile.
1560 @item -Wno-pmf-conversions @r{(C++ only)}
1561 @opindex Wno-pmf-conversions
1562 Disable the diagnostic for converting a bound pointer to member function
1565 @item -Wsign-promo @r{(C++ only)}
1566 @opindex Wsign-promo
1567 Warn when overload resolution chooses a promotion from unsigned or
1568 enumeral type to a signed type over a conversion to an unsigned type of
1569 the same size. Previous versions of G++ would try to preserve
1570 unsignedness, but the standard mandates the current behavior.
1572 @item -Wsynth @r{(C++ only)}
1574 @cindex warning for synthesized methods
1575 @cindex synthesized methods, warning
1576 Warn when G++'s synthesis behavior does not match that of cfront. For
1582 A& operator = (int);
1592 In this example, G++ will synthesize a default @samp{A& operator =
1593 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1596 @node Objective-C Dialect Options
1597 @section Options Controlling Objective-C Dialect
1599 @cindex compiler options, Objective-C
1600 @cindex Objective-C options, command line
1601 @cindex options, Objective-C
1602 This section describes the command-line options that are only meaningful
1603 for Objective-C programs; but you can also use most of the GNU compiler
1604 options regardless of what language your program is in. For example,
1605 you might compile a file @code{some_class.m} like this:
1608 gcc -g -fgnu-runtime -O -c some_class.m
1612 In this example, only @option{-fgnu-runtime} is an option meant only for
1613 Objective-C programs; you can use the other options with any language
1616 Here is a list of options that are @emph{only} for compiling Objective-C
1620 @item -fconstant-string-class=@var{class-name}
1621 @opindex fconstant-string-class
1622 Use @var{class-name} as the name of the class to instantiate for each
1623 literal string specified with the syntax @code{@@"@dots{}"}. The default
1624 class name is @code{NXConstantString}.
1627 @opindex fgnu-runtime
1628 Generate object code compatible with the standard GNU Objective-C
1629 runtime. This is the default for most types of systems.
1631 @item -fnext-runtime
1632 @opindex fnext-runtime
1633 Generate output compatible with the NeXT runtime. This is the default
1634 for NeXT-based systems, including Darwin and Mac OS X@.
1638 Dump interface declarations for all classes seen in the source file to a
1639 file named @file{@var{sourcename}.decl}.
1642 @opindex Wno-protocol
1643 Do not warn if methods required by a protocol are not implemented
1644 in the class adopting it.
1648 Warn if a selector has multiple methods of different types defined.
1650 @c not documented because only avail via -Wp
1651 @c @item -print-objc-runtime-info
1655 @node Language Independent Options
1656 @section Options to Control Diagnostic Messages Formatting
1657 @cindex options to control diagnostics formatting
1658 @cindex diagnostic messages
1659 @cindex message formatting
1661 Traditionally, diagnostic messages have been formatted irrespective of
1662 the output device's aspect (e.g.@: its width, @dots{}). The options described
1663 below can be used to control the diagnostic messages formatting
1664 algorithm, e.g.@: how many characters per line, how often source location
1665 information should be reported. Right now, only the C++ front end can
1666 honor these options. However it is expected, in the near future, that
1667 the remaining front ends would be able to digest them correctly.
1670 @item -fmessage-length=@var{n}
1671 @opindex fmessage-length
1672 Try to format error messages so that they fit on lines of about @var{n}
1673 characters. The default is 72 characters for @command{g++} and 0 for the rest of
1674 the front ends supported by GCC@. If @var{n} is zero, then no
1675 line-wrapping will be done; each error message will appear on a single
1678 @opindex fdiagnostics-show-location
1679 @item -fdiagnostics-show-location=once
1680 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
1681 reporter to emit @emph{once} source location information; that is, in
1682 case the message is too long to fit on a single physical line and has to
1683 be wrapped, the source location won't be emitted (as prefix) again,
1684 over and over, in subsequent continuation lines. This is the default
1687 @item -fdiagnostics-show-location=every-line
1688 Only meaningful in line-wrapping mode. Instructs the diagnostic
1689 messages reporter to emit the same source location information (as
1690 prefix) for physical lines that result from the process of breaking
1691 a message which is too long to fit on a single line.
1695 @node Warning Options
1696 @section Options to Request or Suppress Warnings
1697 @cindex options to control warnings
1698 @cindex warning messages
1699 @cindex messages, warning
1700 @cindex suppressing warnings
1702 Warnings are diagnostic messages that report constructions which
1703 are not inherently erroneous but which are risky or suggest there
1704 may have been an error.
1706 You can request many specific warnings with options beginning @samp{-W},
1707 for example @option{-Wimplicit} to request warnings on implicit
1708 declarations. Each of these specific warning options also has a
1709 negative form beginning @samp{-Wno-} to turn off warnings;
1710 for example, @option{-Wno-implicit}. This manual lists only one of the
1711 two forms, whichever is not the default.
1713 These options control the amount and kinds of warnings produced by GCC:
1716 @cindex syntax checking
1718 @opindex fsyntax-only
1719 Check the code for syntax errors, but don't do anything beyond that.
1723 Issue all the warnings demanded by strict ISO C and ISO C++;
1724 reject all programs that use forbidden extensions, and some other
1725 programs that do not follow ISO C and ISO C++. For ISO C, follows the
1726 version of the ISO C standard specified by any @option{-std} option used.
1728 Valid ISO C and ISO C++ programs should compile properly with or without
1729 this option (though a rare few will require @option{-ansi} or a
1730 @option{-std} option specifying the required version of ISO C)@. However,
1731 without this option, certain GNU extensions and traditional C and C++
1732 features are supported as well. With this option, they are rejected.
1734 @option{-pedantic} does not cause warning messages for use of the
1735 alternate keywords whose names begin and end with @samp{__}. Pedantic
1736 warnings are also disabled in the expression that follows
1737 @code{__extension__}. However, only system header files should use
1738 these escape routes; application programs should avoid them.
1739 @xref{Alternate Keywords}.
1741 Some users try to use @option{-pedantic} to check programs for strict ISO
1742 C conformance. They soon find that it does not do quite what they want:
1743 it finds some non-ISO practices, but not all---only those for which
1744 ISO C @emph{requires} a diagnostic, and some others for which
1745 diagnostics have been added.
1747 A feature to report any failure to conform to ISO C might be useful in
1748 some instances, but would require considerable additional work and would
1749 be quite different from @option{-pedantic}. We don't have plans to
1750 support such a feature in the near future.
1752 Where the standard specified with @option{-std} represents a GNU
1753 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
1754 corresponding @dfn{base standard}, the version of ISO C on which the GNU
1755 extended dialect is based. Warnings from @option{-pedantic} are given
1756 where they are required by the base standard. (It would not make sense
1757 for such warnings to be given only for features not in the specified GNU
1758 C dialect, since by definition the GNU dialects of C include all
1759 features the compiler supports with the given option, and there would be
1760 nothing to warn about.)
1762 @item -pedantic-errors
1763 @opindex pedantic-errors
1764 Like @option{-pedantic}, except that errors are produced rather than
1769 Inhibit all warning messages.
1773 Inhibit warning messages about the use of @samp{#import}.
1775 @item -Wchar-subscripts
1776 @opindex Wchar-subscripts
1777 Warn if an array subscript has type @code{char}. This is a common cause
1778 of error, as programmers often forget that this type is signed on some
1783 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
1784 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
1788 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
1789 the arguments supplied have types appropriate to the format string
1790 specified, and that the conversions specified in the format string make
1791 sense. This includes standard functions, and others specified by format
1792 attributes (@pxref{Function Attributes}), in the @code{printf},
1793 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
1794 not in the C standard) families.
1796 The formats are checked against the format features supported by GNU
1797 libc version 2.2. These include all ISO C89 and C99 features, as well
1798 as features from the Single Unix Specification and some BSD and GNU
1799 extensions. Other library implementations may not support all these
1800 features; GCC does not support warning about features that go beyond a
1801 particular library's limitations. However, if @option{-pedantic} is used
1802 with @option{-Wformat}, warnings will be given about format features not
1803 in the selected standard version (but not for @code{strfmon} formats,
1804 since those are not in any version of the C standard). @xref{C Dialect
1805 Options,,Options Controlling C Dialect}.
1807 @option{-Wformat} is included in @option{-Wall}. For more control over some
1808 aspects of format checking, the options @option{-Wno-format-y2k},
1809 @option{-Wno-format-extra-args}, @option{-Wformat-nonliteral},
1810 @option{-Wformat-security} and @option{-Wformat=2} are available, but are
1811 not included in @option{-Wall}.
1813 @item -Wno-format-y2k
1814 @opindex Wno-format-y2k
1815 If @option{-Wformat} is specified, do not warn about @code{strftime}
1816 formats which may yield only a two-digit year.
1818 @item -Wno-format-extra-args
1819 @opindex Wno-format-extra-args
1820 If @option{-Wformat} is specified, do not warn about excess arguments to a
1821 @code{printf} or @code{scanf} format function. The C standard specifies
1822 that such arguments are ignored.
1824 @item -Wformat-nonliteral
1825 @opindex Wformat-nonliteral
1826 If @option{-Wformat} is specified, also warn if the format string is not a
1827 string literal and so cannot be checked, unless the format function
1828 takes its format arguments as a @code{va_list}.
1830 @item -Wformat-security
1831 @opindex Wformat-security
1832 If @option{-Wformat} is specified, also warn about uses of format
1833 functions that represent possible security problems. At present, this
1834 warns about calls to @code{printf} and @code{scanf} functions where the
1835 format string is not a string literal and there are no format arguments,
1836 as in @code{printf (foo);}. This may be a security hole if the format
1837 string came from untrusted input and contains @samp{%n}. (This is
1838 currently a subset of what @option{-Wformat-nonliteral} warns about, but
1839 in future warnings may be added to @option{-Wformat-security} that are not
1840 included in @option{-Wformat-nonliteral}.)
1844 Enable @option{-Wformat} plus format checks not included in
1845 @option{-Wformat}. Currently equivalent to @samp{-Wformat
1846 -Wformat-nonliteral -Wformat-security}.
1848 @item -Wimplicit-int
1849 @opindex Wimplicit-int
1850 Warn when a declaration does not specify a type.
1852 @item -Wimplicit-function-declaration
1853 @itemx -Werror-implicit-function-declaration
1854 @opindex Wimplicit-function-declaration
1855 @opindex Werror-implicit-function-declaration
1856 Give a warning (or error) whenever a function is used before being
1861 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
1865 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
1866 function with external linkage, returning int, taking either zero
1867 arguments, two, or three arguments of appropriate types.
1869 @item -Wmissing-braces
1870 @opindex Wmissing-braces
1871 Warn if an aggregate or union initializer is not fully bracketed. In
1872 the following example, the initializer for @samp{a} is not fully
1873 bracketed, but that for @samp{b} is fully bracketed.
1876 int a[2][2] = @{ 0, 1, 2, 3 @};
1877 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
1881 @opindex Wparentheses
1882 Warn if parentheses are omitted in certain contexts, such
1883 as when there is an assignment in a context where a truth value
1884 is expected, or when operators are nested whose precedence people
1885 often get confused about.
1887 Also warn about constructions where there may be confusion to which
1888 @code{if} statement an @code{else} branch belongs. Here is an example of
1903 In C, every @code{else} branch belongs to the innermost possible @code{if}
1904 statement, which in this example is @code{if (b)}. This is often not
1905 what the programmer expected, as illustrated in the above example by
1906 indentation the programmer chose. When there is the potential for this
1907 confusion, GCC will issue a warning when this flag is specified.
1908 To eliminate the warning, add explicit braces around the innermost
1909 @code{if} statement so there is no way the @code{else} could belong to
1910 the enclosing @code{if}. The resulting code would look like this:
1926 @item -Wsequence-point
1927 @opindex Wsequence-point
1928 Warn about code that may have undefined semantics because of violations
1929 of sequence point rules in the C standard.
1931 The C standard defines the order in which expressions in a C program are
1932 evaluated in terms of @dfn{sequence points}, which represent a partial
1933 ordering between the execution of parts of the program: those executed
1934 before the sequence point, and those executed after it. These occur
1935 after the evaluation of a full expression (one which is not part of a
1936 larger expression), after the evaluation of the first operand of a
1937 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
1938 function is called (but after the evaluation of its arguments and the
1939 expression denoting the called function), and in certain other places.
1940 Other than as expressed by the sequence point rules, the order of
1941 evaluation of subexpressions of an expression is not specified. All
1942 these rules describe only a partial order rather than a total order,
1943 since, for example, if two functions are called within one expression
1944 with no sequence point between them, the order in which the functions
1945 are called is not specified. However, the standards committee have
1946 ruled that function calls do not overlap.
1948 It is not specified when between sequence points modifications to the
1949 values of objects take effect. Programs whose behavior depends on this
1950 have undefined behavior; the C standard specifies that ``Between the
1951 previous and next sequence point an object shall have its stored value
1952 modified at most once by the evaluation of an expression. Furthermore,
1953 the prior value shall be read only to determine the value to be
1954 stored.''. If a program breaks these rules, the results on any
1955 particular implementation are entirely unpredictable.
1957 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
1958 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
1959 diagnosed by this option, and it may give an occasional false positive
1960 result, but in general it has been found fairly effective at detecting
1961 this sort of problem in programs.
1963 The present implementation of this option only works for C programs. A
1964 future implementation may also work for C++ programs.
1966 The C standard is worded confusingly, therefore there is some debate
1967 over the precise meaning of the sequence point rules in subtle cases.
1968 Links to discussions of the problem, including proposed formal
1969 definitions, may be found on our readings page, at
1970 @w{@uref{http://gcc.gnu.org/readings.html}}.
1973 @opindex Wreturn-type
1974 Warn whenever a function is defined with a return-type that defaults to
1975 @code{int}. Also warn about any @code{return} statement with no
1976 return-value in a function whose return-type is not @code{void}.
1978 For C++, a function without return type always produces a diagnostic
1979 message, even when @option{-Wno-return-type} is specified. The only
1980 exceptions are @samp{main} and functions defined in system headers.
1984 Warn whenever a @code{switch} statement has an index of enumeral type
1985 and lacks a @code{case} for one or more of the named codes of that
1986 enumeration. (The presence of a @code{default} label prevents this
1987 warning.) @code{case} labels outside the enumeration range also
1988 provoke warnings when this option is used.
1992 Warn if any trigraphs are encountered that might change the meaning of
1993 the program (trigraphs within comments are not warned about).
1995 @item -Wunused-function
1996 @opindex Wunused-function
1997 Warn whenever a static function is declared but not defined or a
1998 non\-inline static function is unused.
2000 @item -Wunused-label
2001 @opindex Wunused-label
2002 Warn whenever a label is declared but not used.
2004 To suppress this warning use the @samp{unused} attribute
2005 (@pxref{Variable Attributes}).
2007 @item -Wunused-parameter
2008 @opindex Wunused-parameter
2009 Warn whenever a function parameter is unused aside from its declaration.
2011 To suppress this warning use the @samp{unused} attribute
2012 (@pxref{Variable Attributes}).
2014 @item -Wunused-variable
2015 @opindex Wunused-variable
2016 Warn whenever a local variable or non-constant static variable is unused
2017 aside from its declaration
2019 To suppress this warning use the @samp{unused} attribute
2020 (@pxref{Variable Attributes}).
2022 @item -Wunused-value
2023 @opindex Wunused-value
2024 Warn whenever a statement computes a result that is explicitly not used.
2026 To suppress this warning cast the expression to @samp{void}.
2030 All all the above @option{-Wunused} options combined.
2032 In order to get a warning about an unused function parameter, you must
2033 either specify @samp{-W -Wunused} or separately specify
2034 @option{-Wunused-parameter}.
2036 @item -Wuninitialized
2037 @opindex Wuninitialized
2038 Warn if an automatic variable is used without first being initialized or
2039 if a variable may be clobbered by a @code{setjmp} call.
2041 These warnings are possible only in optimizing compilation,
2042 because they require data flow information that is computed only
2043 when optimizing. If you don't specify @option{-O}, you simply won't
2046 These warnings occur only for variables that are candidates for
2047 register allocation. Therefore, they do not occur for a variable that
2048 is declared @code{volatile}, or whose address is taken, or whose size
2049 is other than 1, 2, 4 or 8 bytes. Also, they do not occur for
2050 structures, unions or arrays, even when they are in registers.
2052 Note that there may be no warning about a variable that is used only
2053 to compute a value that itself is never used, because such
2054 computations may be deleted by data flow analysis before the warnings
2057 These warnings are made optional because GCC is not smart
2058 enough to see all the reasons why the code might be correct
2059 despite appearing to have an error. Here is one example of how
2080 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
2081 always initialized, but GCC doesn't know this. Here is
2082 another common case:
2087 if (change_y) save_y = y, y = new_y;
2089 if (change_y) y = save_y;
2094 This has no bug because @code{save_y} is used only if it is set.
2096 @cindex @code{longjmp} warnings
2097 This option also warns when a non-volatile automatic variable might be
2098 changed by a call to @code{longjmp}. These warnings as well are possible
2099 only in optimizing compilation.
2101 The compiler sees only the calls to @code{setjmp}. It cannot know
2102 where @code{longjmp} will be called; in fact, a signal handler could
2103 call it at any point in the code. As a result, you may get a warning
2104 even when there is in fact no problem because @code{longjmp} cannot
2105 in fact be called at the place which would cause a problem.
2107 Some spurious warnings can be avoided if you declare all the functions
2108 you use that never return as @code{noreturn}. @xref{Function
2111 @item -Wreorder @r{(C++ only)}
2113 @cindex reordering, warning
2114 @cindex warning for reordering of member initializers
2115 Warn when the order of member initializers given in the code does not
2116 match the order in which they must be executed. For instance:
2118 @item -Wunknown-pragmas
2119 @opindex Wunknown-pragmas
2120 @cindex warning for unknown pragmas
2121 @cindex unknown pragmas, warning
2122 @cindex pragmas, warning of unknown
2123 Warn when a #pragma directive is encountered which is not understood by
2124 GCC@. If this command line option is used, warnings will even be issued
2125 for unknown pragmas in system header files. This is not the case if
2126 the warnings were only enabled by the @option{-Wall} command line option.
2130 All of the above @samp{-W} options combined. This enables all the
2131 warnings about constructions that some users consider questionable, and
2132 that are easy to avoid (or modify to prevent the warning), even in
2133 conjunction with macros.
2136 @opindex Wno-div-by-zero
2137 @opindex Wdiv-by-zero
2138 Warn about compile-time integer division by zero. This is default. To
2139 inhibit the warning messages, use @option{-Wno-div-by-zero}. Floating
2140 point division by zero is not warned about, as it can be a legitimate
2141 way of obtaining infinities and NaNs.
2144 @opindex Wno-multichar
2146 Warn if a multicharacter constant (@samp{'FOOF'}) is used. This is
2147 default. To inhibit the warning messages, use @option{-Wno-multichar}.
2148 Usually they indicate a typo in the user's code, as they have
2149 implementation-defined values, and should not be used in portable code.
2151 @item -Wsystem-headers
2152 @opindex Wsystem-headers
2153 @cindex warnings from system headers
2154 @cindex system headers, warnings from
2155 Print warning messages for constructs found in system header files.
2156 Warnings from system headers are normally suppressed, on the assumption
2157 that they usually do not indicate real problems and would only make the
2158 compiler output harder to read. Using this command line option tells
2159 GCC to emit warnings from system headers as if they occurred in user
2160 code. However, note that using @option{-Wall} in conjunction with this
2161 option will @emph{not} warn about unknown pragmas in system
2162 headers---for that, @option{-Wunknown-pragmas} must also be used.
2165 The following @option{-W@dots{}} options are not implied by @option{-Wall}.
2166 Some of them warn about constructions that users generally do not
2167 consider questionable, but which occasionally you might wish to check
2168 for; others warn about constructions that are necessary or hard to avoid
2169 in some cases, and there is no simple way to modify the code to suppress
2175 Print extra warning messages for these events:
2179 A function can return either with or without a value. (Falling
2180 off the end of the function body is considered returning without
2181 a value.) For example, this function would evoke such a
2195 An expression-statement or the left-hand side of a comma expression
2196 contains no side effects.
2197 To suppress the warning, cast the unused expression to void.
2198 For example, an expression such as @samp{x[i,j]} will cause a warning,
2199 but @samp{x[(void)i,j]} will not.
2202 An unsigned value is compared against zero with @samp{<} or @samp{<=}.
2205 A comparison like @samp{x<=y<=z} appears; this is equivalent to
2206 @samp{(x<=y ? 1 : 0) <= z}, which is a different interpretation from
2207 that of ordinary mathematical notation.
2210 Storage-class specifiers like @code{static} are not the first things in
2211 a declaration. According to the C Standard, this usage is obsolescent.
2214 The return type of a function has a type qualifier such as @code{const}.
2215 Such a type qualifier has no effect, since the value returned by a
2216 function is not an lvalue. (But don't warn about the GNU extension of
2217 @code{volatile void} return types. That extension will be warned about
2218 if @option{-pedantic} is specified.)
2221 If @option{-Wall} or @option{-Wunused} is also specified, warn about unused
2225 A comparison between signed and unsigned values could produce an
2226 incorrect result when the signed value is converted to unsigned.
2227 (But don't warn if @option{-Wno-sign-compare} is also specified.)
2230 An aggregate has a partly bracketed initializer.
2231 For example, the following code would evoke such a warning,
2232 because braces are missing around the initializer for @code{x.h}:
2235 struct s @{ int f, g; @};
2236 struct t @{ struct s h; int i; @};
2237 struct t x = @{ 1, 2, 3 @};
2241 An aggregate has an initializer which does not initialize all members.
2242 For example, the following code would cause such a warning, because
2243 @code{x.h} would be implicitly initialized to zero:
2246 struct s @{ int f, g, h; @};
2247 struct s x = @{ 3, 4 @};
2252 @opindex Wfloat-equal
2253 Warn if floating point values are used in equality comparisons.
2255 The idea behind this is that sometimes it is convenient (for the
2256 programmer) to consider floating-point values as approximations to
2257 infinitely precise real numbers. If you are doing this, then you need
2258 to compute (by analysing the code, or in some other way) the maximum or
2259 likely maximum error that the computation introduces, and allow for it
2260 when performing comparisons (and when producing output, but that's a
2261 different problem). In particular, instead of testing for equality, you
2262 would check to see whether the two values have ranges that overlap; and
2263 this is done with the relational operators, so equality comparisons are
2266 @item -Wtraditional @r{(C only)}
2267 @opindex Wtraditional
2268 Warn about certain constructs that behave differently in traditional and
2269 ISO C@. Also warn about ISO C constructs that have no traditional C
2270 equivalent, and/or problematic constructs which should be avoided.
2274 Macro parameters that appear within string literals in the macro body.
2275 In traditional C macro replacement takes place within string literals,
2276 but does not in ISO C@.
2279 In traditional C, some preprocessor directives did not exist.
2280 Traditional preprocessors would only consider a line to be a directive
2281 if the @samp{#} appeared in column 1 on the line. Therefore
2282 @option{-Wtraditional} warns about directives that traditional C
2283 understands but would ignore because the @samp{#} does not appear as the
2284 first character on the line. It also suggests you hide directives like
2285 @samp{#pragma} not understood by traditional C by indenting them. Some
2286 traditional implementations would not recognise @samp{#elif}, so it
2287 suggests avoiding it altogether.
2290 A function-like macro that appears without arguments.
2293 The unary plus operator.
2296 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
2297 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
2298 constants.) Note, these suffixes appear in macros defined in the system
2299 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
2300 Use of these macros in user code might normally lead to spurious
2301 warnings, however gcc's integrated preprocessor has enough context to
2302 avoid warning in these cases.
2305 A function declared external in one block and then used after the end of
2309 A @code{switch} statement has an operand of type @code{long}.
2312 A non-@code{static} function declaration follows a @code{static} one.
2313 This construct is not accepted by some traditional C compilers.
2316 The ISO type of an integer constant has a different width or
2317 signedness from its traditional type. This warning is only issued if
2318 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
2319 typically represent bit patterns, are not warned about.
2322 Usage of ISO string concatenation is detected.
2325 Initialization of automatic aggregates.
2328 Identifier conflicts with labels. Traditional C lacks a separate
2329 namespace for labels.
2332 Initialization of unions. If the initializer is zero, the warning is
2333 omitted. This is done under the assumption that the zero initializer in
2334 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
2335 initializer warnings and relies on default initialization to zero in the
2339 Conversions by prototypes between fixed/floating point values and vice
2340 versa. The absence of these prototypes when compiling with traditional
2341 C would cause serious problems. This is a subset of the possible
2342 conversion warnings, for the full set use @option{-Wconversion}.
2347 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
2351 Warn whenever a local variable shadows another local variable, parameter or
2352 global variable or whenever a built-in function is shadowed.
2354 @item -Wlarger-than-@var{len}
2355 @opindex Wlarger-than
2356 Warn whenever an object of larger than @var{len} bytes is defined.
2358 @item -Wpointer-arith
2359 @opindex Wpointer-arith
2360 Warn about anything that depends on the ``size of'' a function type or
2361 of @code{void}. GNU C assigns these types a size of 1, for
2362 convenience in calculations with @code{void *} pointers and pointers
2365 @item -Wbad-function-cast @r{(C only)}
2366 @opindex Wbad-function-cast
2367 Warn whenever a function call is cast to a non-matching type.
2368 For example, warn if @code{int malloc()} is cast to @code{anything *}.
2372 Warn whenever a pointer is cast so as to remove a type qualifier from
2373 the target type. For example, warn if a @code{const char *} is cast
2374 to an ordinary @code{char *}.
2377 @opindex Wcast-align
2378 Warn whenever a pointer is cast such that the required alignment of the
2379 target is increased. For example, warn if a @code{char *} is cast to
2380 an @code{int *} on machines where integers can only be accessed at
2381 two- or four-byte boundaries.
2383 @item -Wwrite-strings
2384 @opindex Wwrite-strings
2385 When compiling C, give string constants the type @code{const
2386 char[@var{length}]} so that
2387 copying the address of one into a non-@code{const} @code{char *}
2388 pointer will get a warning; when compiling C++, warn about the
2389 deprecated conversion from string constants to @code{char *}.
2390 These warnings will help you find at
2391 compile time code that can try to write into a string constant, but
2392 only if you have been very careful about using @code{const} in
2393 declarations and prototypes. Otherwise, it will just be a nuisance;
2394 this is why we did not make @option{-Wall} request these warnings.
2397 @opindex Wconversion
2398 Warn if a prototype causes a type conversion that is different from what
2399 would happen to the same argument in the absence of a prototype. This
2400 includes conversions of fixed point to floating and vice versa, and
2401 conversions changing the width or signedness of a fixed point argument
2402 except when the same as the default promotion.
2404 Also, warn if a negative integer constant expression is implicitly
2405 converted to an unsigned type. For example, warn about the assignment
2406 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
2407 casts like @code{(unsigned) -1}.
2409 @item -Wsign-compare
2410 @opindex Wsign-compare
2411 @cindex warning for comparison of signed and unsigned values
2412 @cindex comparison of signed and unsigned values, warning
2413 @cindex signed and unsigned values, comparison warning
2414 Warn when a comparison between signed and unsigned values could produce
2415 an incorrect result when the signed value is converted to unsigned.
2416 This warning is also enabled by @option{-W}; to get the other warnings
2417 of @option{-W} without this warning, use @samp{-W -Wno-sign-compare}.
2419 @item -Waggregate-return
2420 @opindex Waggregate-return
2421 Warn if any functions that return structures or unions are defined or
2422 called. (In languages where you can return an array, this also elicits
2425 @item -Wstrict-prototypes @r{(C only)}
2426 @opindex Wstrict-prototypes
2427 Warn if a function is declared or defined without specifying the
2428 argument types. (An old-style function definition is permitted without
2429 a warning if preceded by a declaration which specifies the argument
2432 @item -Wmissing-prototypes @r{(C only)}
2433 @opindex Wmissing-prototypes
2434 Warn if a global function is defined without a previous prototype
2435 declaration. This warning is issued even if the definition itself
2436 provides a prototype. The aim is to detect global functions that fail
2437 to be declared in header files.
2439 @item -Wmissing-declarations
2440 @opindex Wmissing-declarations
2441 Warn if a global function is defined without a previous declaration.
2442 Do so even if the definition itself provides a prototype.
2443 Use this option to detect global functions that are not declared in
2446 @item -Wmissing-noreturn
2447 @opindex Wmissing-noreturn
2448 Warn about functions which might be candidates for attribute @code{noreturn}.
2449 Note these are only possible candidates, not absolute ones. Care should
2450 be taken to manually verify functions actually do not ever return before
2451 adding the @code{noreturn} attribute, otherwise subtle code generation
2452 bugs could be introduced. You will not get a warning for @code{main} in
2453 hosted C environments.
2455 @item -Wmissing-format-attribute
2456 @opindex Wmissing-format-attribute
2458 If @option{-Wformat} is enabled, also warn about functions which might be
2459 candidates for @code{format} attributes. Note these are only possible
2460 candidates, not absolute ones. GCC will guess that @code{format}
2461 attributes might be appropriate for any function that calls a function
2462 like @code{vprintf} or @code{vscanf}, but this might not always be the
2463 case, and some functions for which @code{format} attributes are
2464 appropriate may not be detected. This option has no effect unless
2465 @option{-Wformat} is enabled (possibly by @option{-Wall}).
2469 Warn if a structure is given the packed attribute, but the packed
2470 attribute has no effect on the layout or size of the structure.
2471 Such structures may be mis-aligned for little benefit. For
2472 instance, in this code, the variable @code{f.x} in @code{struct bar}
2473 will be misaligned even though @code{struct bar} does not itself
2474 have the packed attribute:
2481 @} __attribute__((packed));
2491 Warn if padding is included in a structure, either to align an element
2492 of the structure or to align the whole structure. Sometimes when this
2493 happens it is possible to rearrange the fields of the structure to
2494 reduce the padding and so make the structure smaller.
2496 @item -Wredundant-decls
2497 @opindex Wredundant-decls
2498 Warn if anything is declared more than once in the same scope, even in
2499 cases where multiple declaration is valid and changes nothing.
2501 @item -Wnested-externs @r{(C only)}
2502 @opindex Wnested-externs
2503 Warn if an @code{extern} declaration is encountered within a function.
2505 @item -Wunreachable-code
2506 @opindex Wunreachable-code
2507 Warn if the compiler detects that code will never be executed.
2509 This option is intended to warn when the compiler detects that at
2510 least a whole line of source code will never be executed, because
2511 some condition is never satisfied or because it is after a
2512 procedure that never returns.
2514 It is possible for this option to produce a warning even though there
2515 are circumstances under which part of the affected line can be executed,
2516 so care should be taken when removing apparently-unreachable code.
2518 For instance, when a function is inlined, a warning may mean that the
2519 line is unreachable in only one inlined copy of the function.
2521 This option is not made part of @option{-Wall} because in a debugging
2522 version of a program there is often substantial code which checks
2523 correct functioning of the program and is, hopefully, unreachable
2524 because the program does work. Another common use of unreachable
2525 code is to provide behaviour which is selectable at compile-time.
2529 Warn if a function can not be inlined and it was declared as inline.
2533 @opindex Wno-long-long
2534 Warn if @samp{long long} type is used. This is default. To inhibit
2535 the warning messages, use @option{-Wno-long-long}. Flags
2536 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
2537 only when @option{-pedantic} flag is used.
2539 @item -Wdisabled-optimization
2540 @opindex Wdisabled-optimization
2541 Warn if a requested optimization pass is disabled. This warning does
2542 not generally indicate that there is anything wrong with your code; it
2543 merely indicates that GCC's optimizers were unable to handle the code
2544 effectively. Often, the problem is that your code is too big or too
2545 complex; GCC will refuse to optimize programs when the optimization
2546 itself is likely to take inordinate amounts of time.
2550 Make all warnings into errors.
2553 @node Debugging Options
2554 @section Options for Debugging Your Program or GCC
2555 @cindex options, debugging
2556 @cindex debugging information options
2558 GCC has various special options that are used for debugging
2559 either your program or GCC:
2564 Produce debugging information in the operating system's native format
2565 (stabs, COFF, XCOFF, or DWARF)@. GDB can work with this debugging
2568 On most systems that use stabs format, @option{-g} enables use of extra
2569 debugging information that only GDB can use; this extra information
2570 makes debugging work better in GDB but will probably make other debuggers
2572 refuse to read the program. If you want to control for certain whether
2573 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
2574 @option{-gxcoff+}, @option{-gxcoff}, @option{-gdwarf-1+}, or @option{-gdwarf-1}
2577 Unlike most other C compilers, GCC allows you to use @option{-g} with
2578 @option{-O}. The shortcuts taken by optimized code may occasionally
2579 produce surprising results: some variables you declared may not exist
2580 at all; flow of control may briefly move where you did not expect it;
2581 some statements may not be executed because they compute constant
2582 results or their values were already at hand; some statements may
2583 execute in different places because they were moved out of loops.
2585 Nevertheless it proves possible to debug optimized output. This makes
2586 it reasonable to use the optimizer for programs that might have bugs.
2588 The following options are useful when GCC is generated with the
2589 capability for more than one debugging format.
2593 Produce debugging information for use by GDB@. This means to use the
2594 most expressive format available (DWARF 2, stabs, or the native format
2595 if neither of those are supported), including GDB extensions if at all
2600 Produce debugging information in stabs format (if that is supported),
2601 without GDB extensions. This is the format used by DBX on most BSD
2602 systems. On MIPS, Alpha and System V Release 4 systems this option
2603 produces stabs debugging output which is not understood by DBX or SDB@.
2604 On System V Release 4 systems this option requires the GNU assembler.
2608 Produce debugging information in stabs format (if that is supported),
2609 using GNU extensions understood only by the GNU debugger (GDB)@. The
2610 use of these extensions is likely to make other debuggers crash or
2611 refuse to read the program.
2615 Produce debugging information in COFF format (if that is supported).
2616 This is the format used by SDB on most System V systems prior to
2621 Produce debugging information in XCOFF format (if that is supported).
2622 This is the format used by the DBX debugger on IBM RS/6000 systems.
2626 Produce debugging information in XCOFF format (if that is supported),
2627 using GNU extensions understood only by the GNU debugger (GDB)@. The
2628 use of these extensions is likely to make other debuggers crash or
2629 refuse to read the program, and may cause assemblers other than the GNU
2630 assembler (GAS) to fail with an error.
2634 Produce debugging information in DWARF version 1 format (if that is
2635 supported). This is the format used by SDB on most System V Release 4
2640 Produce debugging information in DWARF version 1 format (if that is
2641 supported), using GNU extensions understood only by the GNU debugger
2642 (GDB)@. The use of these extensions is likely to make other debuggers
2643 crash or refuse to read the program.
2647 Produce debugging information in DWARF version 2 format (if that is
2648 supported). This is the format used by DBX on IRIX 6.
2651 @itemx -ggdb@var{level}
2652 @itemx -gstabs@var{level}
2653 @itemx -gcoff@var{level}
2654 @itemx -gxcoff@var{level}
2655 @itemx -gdwarf@var{level}
2656 @itemx -gdwarf-2@var{level}
2657 Request debugging information and also use @var{level} to specify how
2658 much information. The default level is 2.
2660 Level 1 produces minimal information, enough for making backtraces in
2661 parts of the program that you don't plan to debug. This includes
2662 descriptions of functions and external variables, but no information
2663 about local variables and no line numbers.
2665 Level 3 includes extra information, such as all the macro definitions
2666 present in the program. Some debuggers support macro expansion when
2667 you use @option{-g3}.
2672 Generate extra code to write profile information suitable for the
2673 analysis program @code{prof}. You must use this option when compiling
2674 the source files you want data about, and you must also use it when
2677 @cindex @code{gprof}
2680 Generate extra code to write profile information suitable for the
2681 analysis program @code{gprof}. You must use this option when compiling
2682 the source files you want data about, and you must also use it when
2688 Generate extra code to write profile information for basic blocks, which will
2689 record the number of times each basic block is executed, the basic block start
2690 address, and the function name containing the basic block. If @option{-g} is
2691 used, the line number and filename of the start of the basic block will also be
2692 recorded. If not overridden by the machine description, the default action is
2693 to append to the text file @file{bb.out}.
2695 This data could be analyzed by a program like @code{tcov}. Note,
2696 however, that the format of the data is not what @code{tcov} expects.
2697 Eventually GNU @code{gprof} should be extended to process this data.
2701 Makes the compiler print out each function name as it is compiled, and
2702 print some statistics about each pass when it finishes.
2705 @opindex ftime-report
2706 Makes the compiler print some statistics about the time consumed by each
2707 pass when it finishes.
2710 @opindex fmem-report
2711 Makes the compiler print some statistics about permanent memory
2712 allocation when it finishes.
2716 Generate extra code to profile basic blocks. Your executable will
2717 produce output that is a superset of that produced when @option{-a} is
2718 used. Additional output is the source and target address of the basic
2719 blocks where a jump takes place, the number of times a jump is executed,
2720 and (optionally) the complete sequence of basic blocks being executed.
2721 The output is appended to file @file{bb.out}.
2723 You can examine different profiling aspects without recompilation. Your
2724 executable will read a list of function names from file @file{bb.in}.
2725 Profiling starts when a function on the list is entered and stops when
2726 that invocation is exited. To exclude a function from profiling, prefix
2727 its name with @samp{-}. If a function name is not unique, you can
2728 disambiguate it by writing it in the form
2729 @samp{/path/filename.d:functionname}. Your executable will write the
2730 available paths and filenames in file @file{bb.out}.
2732 Several function names have a special meaning:
2735 Write source, target and frequency of jumps to file @file{bb.out}.
2736 @item __bb_hidecall__
2737 Exclude function calls from frequency count.
2738 @item __bb_showret__
2739 Include function returns in frequency count.
2741 Write the sequence of basic blocks executed to file @file{bbtrace.gz}.
2742 The file will be compressed using the program @samp{gzip}, which must
2743 exist in your @env{PATH}. On systems without the @samp{popen}
2744 function, the file will be named @file{bbtrace} and will not be
2745 compressed. @strong{Profiling for even a few seconds on these systems
2746 will produce a very large file.} Note: @code{__bb_hidecall__} and
2747 @code{__bb_showret__} will not affect the sequence written to
2751 Here's a short example using different profiling parameters
2752 in file @file{bb.in}. Assume function @code{foo} consists of basic blocks
2753 1 and 2 and is called twice from block 3 of function @code{main}. After
2754 the calls, block 3 transfers control to block 4 of @code{main}.
2756 With @code{__bb_trace__} and @code{main} contained in file @file{bb.in},
2757 the following sequence of blocks is written to file @file{bbtrace.gz}:
2758 0 3 1 2 1 2 4. The return from block 2 to block 3 is not shown, because
2759 the return is to a point inside the block and not to the top. The
2760 block address 0 always indicates, that control is transferred
2761 to the trace from somewhere outside the observed functions. With
2762 @samp{-foo} added to @file{bb.in}, the blocks of function
2763 @code{foo} are removed from the trace, so only 0 3 4 remains.
2765 With @code{__bb_jumps__} and @code{main} contained in file @file{bb.in},
2766 jump frequencies will be written to file @file{bb.out}. The
2767 frequencies are obtained by constructing a trace of blocks
2768 and incrementing a counter for every neighbouring pair of blocks
2769 in the trace. The trace 0 3 1 2 1 2 4 displays the following
2773 Jump from block 0x0 to block 0x3 executed 1 time(s)
2774 Jump from block 0x3 to block 0x1 executed 1 time(s)
2775 Jump from block 0x1 to block 0x2 executed 2 time(s)
2776 Jump from block 0x2 to block 0x1 executed 1 time(s)
2777 Jump from block 0x2 to block 0x4 executed 1 time(s)
2780 With @code{__bb_hidecall__}, control transfer due to call instructions
2781 is removed from the trace, that is the trace is cut into three parts: 0
2782 3 4, 0 1 2 and 0 1 2. With @code{__bb_showret__}, control transfer due
2783 to return instructions is added to the trace. The trace becomes: 0 3 1
2784 2 3 1 2 3 4. Note, that this trace is not the same, as the sequence
2785 written to @file{bbtrace.gz}. It is solely used for counting jump
2788 @item -fprofile-arcs
2789 @opindex fprofile-arcs
2790 Instrument @dfn{arcs} during compilation to generate coverage data
2791 or for profile-directed block ordering. During execution the program
2792 records how many times each branch is executed and how many times it is
2793 taken. When the compiled program exits it saves this data to a file
2794 called @file{@var{sourcename}.da} for each source file.
2796 For profile-directed block ordering, compile the program with
2797 @option{-fprofile-arcs} plus optimization and code generation options,
2798 generate the arc profile information by running the program on a
2799 selected workload, and then compile the program again with the same
2800 optimization and code generation options plus
2801 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
2802 Control Optimization}).
2804 The other use of @option{-fprofile-arcs} is for use with @code{gcov},
2805 when it is used with the @option{-ftest-coverage} option. GCC
2806 supports two methods of determining code coverage: the options that
2807 support @code{gcov}, and options @option{-a} and @option{-ax}, which
2808 write information to text files. The options that support @code{gcov}
2809 do not need to instrument every arc in the program, so a program compiled
2810 with them runs faster than a program compiled with @option{-a}, which
2811 adds instrumentation code to every basic block in the program. The
2812 tradeoff: since @code{gcov} does not have execution counts for all
2813 branches, it must start with the execution counts for the instrumented
2814 branches, and then iterate over the program flow graph until the entire
2815 graph has been solved. Hence, @code{gcov} runs a little more slowly than
2816 a program which uses information from @option{-a} and @option{-ax}.
2818 With @option{-fprofile-arcs}, for each function of your program GCC
2819 creates a program flow graph, then finds a spanning tree for the graph.
2820 Only arcs that are not on the spanning tree have to be instrumented: the
2821 compiler adds code to count the number of times that these arcs are
2822 executed. When an arc is the only exit or only entrance to a block, the
2823 instrumentation code can be added to the block; otherwise, a new basic
2824 block must be created to hold the instrumentation code.
2826 This option makes it possible to estimate branch probabilities and to
2827 calculate basic block execution counts. In general, basic block
2828 execution counts as provided by @option{-a} do not give enough
2829 information to estimate all branch probabilities.
2832 @item -ftest-coverage
2833 @opindex ftest-coverage
2834 Create data files for the @code{gcov} code-coverage utility
2835 (@pxref{Gcov,, @code{gcov}: a GCC Test Coverage Program}).
2836 The data file names begin with the name of your source file:
2839 @item @var{sourcename}.bb
2840 A mapping from basic blocks to line numbers, which @code{gcov} uses to
2841 associate basic block execution counts with line numbers.
2843 @item @var{sourcename}.bbg
2844 A list of all arcs in the program flow graph. This allows @code{gcov}
2845 to reconstruct the program flow graph, so that it can compute all basic
2846 block and arc execution counts from the information in the
2847 @code{@var{sourcename}.da} file.
2850 Use @option{-ftest-coverage} with @option{-fprofile-arcs}; the latter
2851 option adds instrumentation to the program, which then writes
2852 execution counts to another data file:
2855 @item @var{sourcename}.da
2856 Runtime arc execution counts, used in conjunction with the arc
2857 information in the file @code{@var{sourcename}.bbg}.
2860 Coverage data will map better to the source files if
2861 @option{-ftest-coverage} is used without optimization.
2863 @item -d@var{letters}
2865 Says to make debugging dumps during compilation at times specified by
2866 @var{letters}. This is used for debugging the compiler. The file names
2867 for most of the dumps are made by appending a pass number and a word to
2868 the source file name (e.g. @file{foo.c.00.rtl} or @file{foo.c.01.sibling}).
2869 Here are the possible letters for use in @var{letters}, and their meanings:
2874 Annotate the assembler output with miscellaneous debugging information.
2877 Dump after computing branch probabilities, to @file{@var{file}.14.bp}.
2880 Dump after block reordering, to @file{@var{file}.28.bbro}.
2883 Dump after instruction combination, to the file @file{@var{file}.16.combine}.
2886 Dump after the first if conversion, to the file @file{@var{file}.17.ce}.
2889 Dump after delayed branch scheduling, to @file{@var{file}.31.dbr}.
2892 Dump all macro definitions, at the end of preprocessing, in addition to
2896 Dump after SSA optimizations, to @file{@var{file}.04.ssa} and
2897 @file{@var{file}.07.ussa}.
2900 Dump after the second if conversion, to @file{@var{file}.26.ce2}.
2903 Dump after life analysis, to @file{@var{file}.15.life}.
2906 Dump after purging @code{ADDRESSOF} codes, to @file{@var{file}.09.addressof}.
2909 Dump after global register allocation, to @file{@var{file}.21.greg}.
2912 Dump after finalization of EH handling code, to @file{@var{file}.02.eh}.
2915 Dump after post-reload optimizations, to @file{@var{file}.22.postreload}.
2918 Dump after GCSE, to @file{@var{file}.10.gcse}.
2921 Dump after sibling call optimizations, to @file{@var{file}.01.sibling}.
2924 Dump after the first jump optimization, to @file{@var{file}.03.jump}.
2927 Dump after conversion from registers to stack, to @file{@var{file}.32.stack}.
2930 Dump after local register allocation, to @file{@var{file}.20.lreg}.
2933 Dump after loop optimization, to @file{@var{file}.11.loop}.
2936 Dump after performing the machine dependent reorganisation pass, to
2937 @file{@var{file}.30.mach}.
2940 Dump after register renumbering, to @file{@var{file}.25.rnreg}.
2943 Dump after the register move pass, to @file{@var{file}.18.regmove}.
2946 Dump after RTL generation, to @file{@var{file}.00.rtl}.
2949 Dump after the second scheduling pass, to @file{@var{file}.27.sched2}.
2952 Dump after CSE (including the jump optimization that sometimes follows
2953 CSE), to @file{@var{file}.08.cse}.
2956 Dump after the first scheduling pass, to @file{@var{file}.19.sched}.
2959 Dump after the second CSE pass (including the jump optimization that
2960 sometimes follows CSE), to @file{@var{file}.12.cse2}.
2963 Dump after the second flow pass, to @file{@var{file}.23.flow2}.
2966 Dump after SSA dead code elimination, to @file{@var{file}.06.ssadce}.
2969 Dump after the peephole pass, to @file{@var{file}.24.peephole2}.
2972 Produce all the dumps listed above.
2975 Print statistics on memory usage, at the end of the run, to
2979 Annotate the assembler output with a comment indicating which
2980 pattern and alternative was used. The length of each instruction is
2984 Dump the RTL in the assembler output as a comment before each instruction.
2985 Also turns on @option{-dp} annotation.
2988 For each of the other indicated dump files (except for
2989 @file{@var{file}.00.rtl}), dump a representation of the control flow graph
2990 suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
2993 Just generate RTL for a function instead of compiling it. Usually used
2997 Dump debugging information during parsing, to standard error.
3000 @item -fdump-unnumbered
3001 @opindex fdump-unnumbered
3002 When doing debugging dumps (see @option{-d} option above), suppress instruction
3003 numbers and line number note output. This makes it more feasible to
3004 use diff on debugging dumps for compiler invocations with different
3005 options, in particular with and without @option{-g}.
3007 @item -fdump-class-hierarchy @r{(C++ only)}
3008 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
3009 @opindex fdump-class-hierarchy
3010 Dump a representation of each class's hierarchy and virtual function
3011 table layout to a file. The file name is made by appending @file{.class}
3012 to the source file name. If the @samp{-@var{options}} form is used,
3013 @var{options} controls the details of the dump as described for the
3014 @option{-fdump-tree} options.
3016 @item -fdump-tree-@var{switch} @r{(C++ only)}
3017 @itemx -fdump-tree-@var{switch}-@var{options} @r{(C++ only)}
3019 Control the dumping at various stages of processing the intermediate
3020 language tree to a file. The file name is generated by appending a switch
3021 specific suffix to the source file name. If the @samp{-@var{options}}
3022 form is used, @var{options} is a list of @samp{-} separated options that
3023 control the details of the dump. Not all options are applicable to all
3024 dumps, those which are not meaningful will be ignored. The following
3025 options are available
3029 Print the address of each node. Usually this is not meaningful as it
3030 changes according to the environment and source file. Its primary use
3031 is for tying up a dump file with a debug environment.
3033 Inhibit dumping of members of a scope or body of a function merely
3034 because that scope has been reached. Only dump such items when they
3035 are directly reachable by some other path.
3037 Turn on all options.
3040 The following tree dumps are possible:
3043 Dump before any tree based optimization, to @file{@var{file}.original}.
3045 Dump after all tree based optimization, to @file{@var{file}.optimized}.
3047 Dump after function inlining, to @file{@var{file}.inlined}.
3050 @item -fpretend-float
3051 @opindex fpretend-float
3052 When running a cross-compiler, pretend that the target machine uses the
3053 same floating point format as the host machine. This causes incorrect
3054 output of the actual floating constants, but the actual instruction
3055 sequence will probably be the same as GCC would make when running on
3060 Store the usual ``temporary'' intermediate files permanently; place them
3061 in the current directory and name them based on the source file. Thus,
3062 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
3063 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
3064 preprocessed @file{foo.i} output file even though the compiler now
3065 normally uses an integrated preprocessor.
3069 Report the CPU time taken by each subprocess in the compilation
3070 sequence. For C source files, this is the compiler proper and assembler
3071 (plus the linker if linking is done). The output looks like this:
3078 The first number on each line is the ``user time,'' that is time spent
3079 executing the program itself. The second number is ``system time,''
3080 time spent executing operating system routines on behalf of the program.
3081 Both numbers are in seconds.
3083 @item -print-file-name=@var{library}
3084 @opindex print-file-name
3085 Print the full absolute name of the library file @var{library} that
3086 would be used when linking---and don't do anything else. With this
3087 option, GCC does not compile or link anything; it just prints the
3090 @item -print-multi-directory
3091 @opindex print-multi-directory
3092 Print the directory name corresponding to the multilib selected by any
3093 other switches present in the command line. This directory is supposed
3094 to exist in @env{GCC_EXEC_PREFIX}.
3096 @item -print-multi-lib
3097 @opindex print-multi-lib
3098 Print the mapping from multilib directory names to compiler switches
3099 that enable them. The directory name is separated from the switches by
3100 @samp{;}, and each switch starts with an @samp{@@} instead of the
3101 @samp{-}, without spaces between multiple switches. This is supposed to
3102 ease shell-processing.
3104 @item -print-prog-name=@var{program}
3105 @opindex print-prog-name
3106 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
3108 @item -print-libgcc-file-name
3109 @opindex print-libgcc-file-name
3110 Same as @option{-print-file-name=libgcc.a}.
3112 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
3113 but you do want to link with @file{libgcc.a}. You can do
3116 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
3119 @item -print-search-dirs
3120 @opindex print-search-dirs
3121 Print the name of the configured installation directory and a list of
3122 program and library directories gcc will search---and don't do anything else.
3124 This is useful when gcc prints the error message
3125 @samp{installation problem, cannot exec cpp0: No such file or directory}.
3126 To resolve this you either need to put @file{cpp0} and the other compiler
3127 components where gcc expects to find them, or you can set the environment
3128 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
3129 Don't forget the trailing '/'.
3130 @xref{Environment Variables}.
3133 @opindex dumpmachine
3134 Print the compiler's target machine (for example,
3135 @samp{i686-pc-linux-gnu})---and don't do anything else.
3138 @opindex dumpversion
3139 Print the compiler version (for example, @samp{3.0})---and don't do
3144 Print the compiler's built-in specs---and don't do anything else. (This
3145 is used when GCC itself is being built.) @xref{Spec Files}.
3148 @node Optimize Options
3149 @section Options That Control Optimization
3150 @cindex optimize options
3151 @cindex options, optimization
3153 These options control various sorts of optimizations:
3160 Optimize. Optimizing compilation takes somewhat more time, and a lot
3161 more memory for a large function.
3163 Without @option{-O}, the compiler's goal is to reduce the cost of
3164 compilation and to make debugging produce the expected results.
3165 Statements are independent: if you stop the program with a breakpoint
3166 between statements, you can then assign a new value to any variable or
3167 change the program counter to any other statement in the function and
3168 get exactly the results you would expect from the source code.
3170 With @option{-O}, the compiler tries to reduce code size and execution
3171 time, without performing any optimizations that take a great deal of
3176 Optimize even more. GCC performs nearly all supported optimizations
3177 that do not involve a space-speed tradeoff. The compiler does not
3178 perform loop unrolling or function inlining when you specify @option{-O2}.
3179 As compared to @option{-O}, this option increases both compilation time
3180 and the performance of the generated code.
3182 @option{-O2} turns on all optional optimizations except for loop unrolling,
3183 function inlining, and register renaming. It also turns on the
3184 @option{-fforce-mem} option on all machines and frame pointer elimination
3185 on machines where doing so does not interfere with debugging.
3187 Please note the warning under @option{-fgcse} about
3188 invoking @option{-O2} on programs that use computed gotos.
3192 Optimize yet more. @option{-O3} turns on all optimizations specified by
3193 @option{-O2} and also turns on the @option{-finline-functions} and
3194 @option{-frename-registers} options.
3202 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
3203 do not typically increase code size. It also performs further
3204 optimizations designed to reduce code size.
3206 If you use multiple @option{-O} options, with or without level numbers,
3207 the last such option is the one that is effective.
3210 Options of the form @option{-f@var{flag}} specify machine-independent
3211 flags. Most flags have both positive and negative forms; the negative
3212 form of @option{-ffoo} would be @option{-fno-foo}. In the table below,
3213 only one of the forms is listed---the one which is not the default.
3214 You can figure out the other form by either removing @samp{no-} or
3219 @opindex ffloat-store
3220 Do not store floating point variables in registers, and inhibit other
3221 options that might change whether a floating point value is taken from a
3224 @cindex floating point precision
3225 This option prevents undesirable excess precision on machines such as
3226 the 68000 where the floating registers (of the 68881) keep more
3227 precision than a @code{double} is supposed to have. Similarly for the
3228 x86 architecture. For most programs, the excess precision does only
3229 good, but a few programs rely on the precise definition of IEEE floating
3230 point. Use @option{-ffloat-store} for such programs, after modifying
3231 them to store all pertinent intermediate computations into variables.
3233 @item -fno-default-inline
3234 @opindex fno-default-inline
3235 Do not make member functions inline by default merely because they are
3236 defined inside the class scope (C++ only). Otherwise, when you specify
3237 @w{@option{-O}}, member functions defined inside class scope are compiled
3238 inline by default; i.e., you don't need to add @samp{inline} in front of
3239 the member function name.
3241 @item -fno-defer-pop
3242 @opindex fno-defer-pop
3243 Always pop the arguments to each function call as soon as that function
3244 returns. For machines which must pop arguments after a function call,
3245 the compiler normally lets arguments accumulate on the stack for several
3246 function calls and pops them all at once.
3250 Force memory operands to be copied into registers before doing
3251 arithmetic on them. This produces better code by making all memory
3252 references potential common subexpressions. When they are not common
3253 subexpressions, instruction combination should eliminate the separate
3254 register-load. The @option{-O2} option turns on this option.
3257 @opindex fforce-addr
3258 Force memory address constants to be copied into registers before
3259 doing arithmetic on them. This may produce better code just as
3260 @option{-fforce-mem} may.
3262 @item -fomit-frame-pointer
3263 @opindex fomit-frame-pointer
3264 Don't keep the frame pointer in a register for functions that
3265 don't need one. This avoids the instructions to save, set up and
3266 restore frame pointers; it also makes an extra register available
3267 in many functions. @strong{It also makes debugging impossible on
3271 On some machines, such as the VAX, this flag has no effect, because
3272 the standard calling sequence automatically handles the frame pointer
3273 and nothing is saved by pretending it doesn't exist. The
3274 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
3275 whether a target machine supports this flag. @xref{Registers}.
3278 On some machines, such as the VAX, this flag has no effect, because
3279 the standard calling sequence automatically handles the frame pointer
3280 and nothing is saved by pretending it doesn't exist. The
3281 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
3282 whether a target machine supports this flag. @xref{Registers,,Register
3283 Usage, gcc.info, Using and Porting GCC}.
3286 @item -foptimize-sibling-calls
3287 @opindex foptimize-sibling-calls
3288 Optimize sibling and tail recursive calls.
3292 This option generates traps for signed overflow on addition, subtraction,
3293 multiplication operations.
3297 Don't pay attention to the @code{inline} keyword. Normally this option
3298 is used to keep the compiler from expanding any functions inline.
3299 Note that if you are not optimizing, no functions can be expanded inline.
3301 @item -finline-functions
3302 @opindex finline-functions
3303 Integrate all simple functions into their callers. The compiler
3304 heuristically decides which functions are simple enough to be worth
3305 integrating in this way.
3307 If all calls to a given function are integrated, and the function is
3308 declared @code{static}, then the function is normally not output as
3309 assembler code in its own right.
3311 @item -finline-limit=@var{n}
3312 @opindex finline-limit
3313 By default, gcc limits the size of functions that can be inlined. This flag
3314 allows the control of this limit for functions that are explicitly marked as
3315 inline (ie marked with the inline keyword or defined within the class
3316 definition in c++). @var{n} is the size of functions that can be inlined in
3317 number of pseudo instructions (not counting parameter handling). The default
3318 value of @var{n} is 600.
3319 Increasing this value can result in more inlined code at
3320 the cost of compilation time and memory consumption. Decreasing usually makes
3321 the compilation faster and less code will be inlined (which presumably
3322 means slower programs). This option is particularly useful for programs that
3323 use inlining heavily such as those based on recursive templates with C++.
3325 @emph{Note:} pseudo instruction represents, in this particular context, an
3326 abstract measurement of function's size. In no way, it represents a count
3327 of assembly instructions and as such its exact meaning might change from one
3328 release to an another.
3330 @item -fkeep-inline-functions
3331 @opindex fkeep-inline-functions
3332 Even if all calls to a given function are integrated, and the function
3333 is declared @code{static}, nevertheless output a separate run-time
3334 callable version of the function. This switch does not affect
3335 @code{extern inline} functions.
3337 @item -fkeep-static-consts
3338 @opindex fkeep-static-consts
3339 Emit variables declared @code{static const} when optimization isn't turned
3340 on, even if the variables aren't referenced.
3342 GCC enables this option by default. If you want to force the compiler to
3343 check if the variable was referenced, regardless of whether or not
3344 optimization is turned on, use the @option{-fno-keep-static-consts} option.
3346 @item -fmerge-constants
3347 Attempt to merge identical constants (string constants and floating point
3348 constants) accross compilation units.
3350 This option is default for optimized compilation if assembler and linker
3351 support it. Use @option{-fno-merge-constants} to inhibit this behaviour.
3353 @item -fmerge-all-constants
3354 Attempt to merge identical constants and identical variables.
3356 This option implies @option{-fmerge-constants}. In addition to
3357 @option{-fmerge-constants} this considers e.g. even constant initialized
3358 arrays or initialized constant variables with integral or floating point
3359 types. Languages like C or C++ require each non-automatic variable to
3360 have distinct location, so using this option will result in non-conforming
3363 @item -fno-function-cse
3364 @opindex fno-function-cse
3365 Do not put function addresses in registers; make each instruction that
3366 calls a constant function contain the function's address explicitly.
3368 This option results in less efficient code, but some strange hacks
3369 that alter the assembler output may be confused by the optimizations
3370 performed when this option is not used.
3374 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations}, and @*
3375 @option{-fno-trapping-math}.
3377 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
3379 This option should never be turned on by any @option{-O} option since
3380 it can result in incorrect output for programs which depend on
3381 an exact implementation of IEEE or ISO rules/specifications for
3384 @item -fno-math-errno
3385 @opindex fno-math-errno
3386 Do not set ERRNO after calling math functions that are executed
3387 with a single instruction, e.g., sqrt. A program that relies on
3388 IEEE exceptions for math error handling may want to use this flag
3389 for speed while maintaining IEEE arithmetic compatibility.
3391 This option should never be turned on by any @option{-O} option since
3392 it can result in incorrect output for programs which depend on
3393 an exact implementation of IEEE or ISO rules/specifications for
3396 The default is @option{-fmath-errno}.
3398 @item -funsafe-math-optimizations
3399 @opindex funsafe-math-optimizations
3400 Allow optimizations for floating-point arithmetic that (a) assume
3401 that arguments and results are valid and (b) may violate IEEE or
3402 ANSI standards. When used at link-time, it may include libraries
3403 or startup files that change the default FPU control word or other
3404 similar optimizations.
3406 This option should never be turned on by any @option{-O} option since
3407 it can result in incorrect output for programs which depend on
3408 an exact implementation of IEEE or ISO rules/specifications for
3411 The default is @option{-fno-unsafe-math-optimizations}.
3413 @item -fno-trapping-math
3414 @opindex fno-trapping-math
3415 Compile code assuming that floating-point operations cannot generate
3416 user-visible traps. Setting this option may allow faster code
3417 if one relies on ``non-stop'' IEEE arithmetic, for example.
3419 This option should never be turned on by any @option{-O} option since
3420 it can result in incorrect output for programs which depend on
3421 an exact implementation of IEEE or ISO rules/specifications for
3424 The default is @option{-ftrapping-math}.
3427 The following options control specific optimizations. The @option{-O2}
3428 option turns on all of these optimizations except @option{-funroll-loops}
3429 and @option{-funroll-all-loops}. On most machines, the @option{-O} option
3430 turns on the @option{-fthread-jumps} and @option{-fdelayed-branch} options,
3431 but specific machines may handle it differently.
3433 You can use the following flags in the rare cases when ``fine-tuning''
3434 of optimizations to be performed is desired.
3436 Not all of the optimizations performed by GCC have @option{-f} options
3440 @item -fstrength-reduce
3441 @opindex fstrength-reduce
3442 Perform the optimizations of loop strength reduction and
3443 elimination of iteration variables.
3445 @item -fthread-jumps
3446 @opindex fthread-jumps
3447 Perform optimizations where we check to see if a jump branches to a
3448 location where another comparison subsumed by the first is found. If
3449 so, the first branch is redirected to either the destination of the
3450 second branch or a point immediately following it, depending on whether
3451 the condition is known to be true or false.
3453 @item -fcse-follow-jumps
3454 @opindex fcse-follow-jumps
3455 In common subexpression elimination, scan through jump instructions
3456 when the target of the jump is not reached by any other path. For
3457 example, when CSE encounters an @code{if} statement with an
3458 @code{else} clause, CSE will follow the jump when the condition
3461 @item -fcse-skip-blocks
3462 @opindex fcse-skip-blocks
3463 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
3464 follow jumps which conditionally skip over blocks. When CSE
3465 encounters a simple @code{if} statement with no else clause,
3466 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
3467 body of the @code{if}.
3469 @item -frerun-cse-after-loop
3470 @opindex frerun-cse-after-loop
3471 Re-run common subexpression elimination after loop optimizations has been
3474 @item -frerun-loop-opt
3475 @opindex frerun-loop-opt
3476 Run the loop optimizer twice.
3480 Perform a global common subexpression elimination pass.
3481 This pass also performs global constant and copy propagation.
3483 @emph{Note:} When compiling a program using computed gotos, a GCC
3484 extension, you may get better runtime performance if you disable
3485 the global common subexpression elmination pass by adding
3486 @option{-fno-gcse} to the command line.
3490 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
3491 attempt to move loads which are only killed by stores into themselves. This
3492 allows a loop containing a load/store sequence to be changed to a load outside
3493 the loop, and a copy/store within the loop.
3497 When @option{-fgcse-sm} is enabled, A store motion pass is run after global common
3498 subexpression elimination. This pass will attempt to move stores out of loops.
3499 When used in conjunction with @option{-fgcse-lm}, loops containing a load/store sequence
3500 can be changed to a load before the loop and a store after the loop.
3502 @item -fdelete-null-pointer-checks
3503 @opindex fdelete-null-pointer-checks
3504 Use global dataflow analysis to identify and eliminate useless checks
3505 for null pointers. The compiler assumes that dereferencing a null
3506 pointer would have halted the program. If a pointer is checked after
3507 it has already been dereferenced, it cannot be null.
3509 In some environments, this assumption is not true, and programs can
3510 safely dereference null pointers. Use
3511 @option{-fno-delete-null-pointer-checks} to disable this optimization
3512 for programs which depend on that behavior.
3514 @item -fexpensive-optimizations
3515 @opindex fexpensive-optimizations
3516 Perform a number of minor optimizations that are relatively expensive.
3518 @item -foptimize-register-move
3520 @opindex foptimize-register-move
3522 Attempt to reassign register numbers in move instructions and as
3523 operands of other simple instructions in order to maximize the amount of
3524 register tying. This is especially helpful on machines with two-operand
3525 instructions. GCC enables this optimization by default with @option{-O2}
3528 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
3531 @item -fdelayed-branch
3532 @opindex fdelayed-branch
3533 If supported for the target machine, attempt to reorder instructions
3534 to exploit instruction slots available after delayed branch
3537 @item -fschedule-insns
3538 @opindex fschedule-insns
3539 If supported for the target machine, attempt to reorder instructions to
3540 eliminate execution stalls due to required data being unavailable. This
3541 helps machines that have slow floating point or memory load instructions
3542 by allowing other instructions to be issued until the result of the load
3543 or floating point instruction is required.
3545 @item -fschedule-insns2
3546 @opindex fschedule-insns2
3547 Similar to @option{-fschedule-insns}, but requests an additional pass of
3548 instruction scheduling after register allocation has been done. This is
3549 especially useful on machines with a relatively small number of
3550 registers and where memory load instructions take more than one cycle.
3552 @item -ffunction-sections
3553 @itemx -fdata-sections
3554 @opindex ffunction-sections
3555 @opindex fdata-sections
3556 Place each function or data item into its own section in the output
3557 file if the target supports arbitrary sections. The name of the
3558 function or the name of the data item determines the section's name
3561 Use these options on systems where the linker can perform optimizations
3562 to improve locality of reference in the instruction space. HPPA
3563 processors running HP-UX and Sparc processors running Solaris 2 have
3564 linkers with such optimizations. Other systems using the ELF object format
3565 as well as AIX may have these optimizations in the future.
3567 Only use these options when there are significant benefits from doing
3568 so. When you specify these options, the assembler and linker will
3569 create larger object and executable files and will also be slower.
3570 You will not be able to use @code{gprof} on all systems if you
3571 specify this option and you may have problems with debugging if
3572 you specify both this option and @option{-g}.
3574 @item -fcaller-saves
3575 @opindex fcaller-saves
3576 Enable values to be allocated in registers that will be clobbered by
3577 function calls, by emitting extra instructions to save and restore the
3578 registers around such calls. Such allocation is done only when it
3579 seems to result in better code than would otherwise be produced.
3581 This option is always enabled by default on certain machines, usually
3582 those which have no call-preserved registers to use instead.
3584 For all machines, optimization level 2 and higher enables this flag by
3587 @item -funroll-loops
3588 @opindex funroll-loops
3589 Unroll loops whose number of iterations can be determined at compile
3590 time or upon entry to the loop. @option{-funroll-loops} implies both
3591 @option{-fstrength-reduce} and @option{-frerun-cse-after-loop}. This
3592 option makes code larger, and may or may not make it run faster.
3594 @item -funroll-all-loops
3595 @opindex funroll-all-loops
3596 Unroll all loops, even if their number of iterations is uncertain when
3597 the loop is entered. This usually makes programs run more slowly.
3598 @option{-funroll-all-loops} implies the same options as
3599 @option{-funroll-loops},
3602 @item -fmove-all-movables
3603 @opindex fmove-all-movables
3604 Forces all invariant computations in loops to be moved
3607 @item -freduce-all-givs
3608 @opindex freduce-all-givs
3609 Forces all general-induction variables in loops to be
3612 @emph{Note:} When compiling programs written in Fortran,
3613 @option{-fmove-all-movables} and @option{-freduce-all-givs} are enabled
3614 by default when you use the optimizer.
3616 These options may generate better or worse code; results are highly
3617 dependent on the structure of loops within the source code.
3619 These two options are intended to be removed someday, once
3620 they have helped determine the efficacy of various
3621 approaches to improving loop optimizations.
3623 Please let us (@w{@email{gcc@@gcc.gnu.org}} and @w{@email{fortran@@gnu.org}})
3624 know how use of these options affects
3625 the performance of your production code.
3626 We're very interested in code that runs @emph{slower}
3627 when these options are @emph{enabled}.
3630 @itemx -fno-peephole2
3631 @opindex fno-peephole
3632 @opindex fno-peephole2
3633 Disable any machine-specific peephole optimizations. The difference
3634 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
3635 are implemented in the compiler; some targets use one, some use the
3636 other, a few use both.
3638 @item -fbranch-probabilities
3639 @opindex fbranch-probabilities
3640 After running a program compiled with @option{-fprofile-arcs}
3641 (@pxref{Debugging Options,, Options for Debugging Your Program or
3642 @command{gcc}}), you can compile it a second time using
3643 @option{-fbranch-probabilities}, to improve optimizations based on
3644 the number of times each branch was taken. When the program
3645 compiled with @option{-fprofile-arcs} exits it saves arc execution
3646 counts to a file called @file{@var{sourcename}.da} for each source
3647 file The information in this data file is very dependent on the
3648 structure of the generated code, so you must use the same source code
3649 and the same optimization options for both compilations.
3652 With @option{-fbranch-probabilities}, GCC puts a @samp{REG_EXEC_COUNT}
3653 note on the first instruction of each basic block, and a
3654 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
3655 These can be used to improve optimization. Currently, they are only
3656 used in one place: in @file{reorg.c}, instead of guessing which path a
3657 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
3658 exactly determine which path is taken more often.
3661 @item -fno-guess-branch-probability
3662 @opindex fno-guess-branch-probability
3663 Do not guess branch probabilities using a randomized model.
3665 Sometimes gcc will opt to use a randomized model to guess branch
3666 probabilities, when none are available from either profiling feedback
3667 (@option{-fprofile-arcs}) or @samp{__builtin_expect}. This means that
3668 different runs of the compiler on the same program may produce different
3671 In a hard real-time system, people don't want different runs of the
3672 compiler to produce code that has different behavior; minimizing
3673 non-determinism is of paramount import. This switch allows users to
3674 reduce non-determinism, possibly at the expense of inferior
3677 @item -fstrict-aliasing
3678 @opindex fstrict-aliasing
3679 Allows the compiler to assume the strictest aliasing rules applicable to
3680 the language being compiled. For C (and C++), this activates
3681 optimizations based on the type of expressions. In particular, an
3682 object of one type is assumed never to reside at the same address as an
3683 object of a different type, unless the types are almost the same. For
3684 example, an @code{unsigned int} can alias an @code{int}, but not a
3685 @code{void*} or a @code{double}. A character type may alias any other
3688 Pay special attention to code like this:
3701 The practice of reading from a different union member than the one most
3702 recently written to (called ``type-punning'') is common. Even with
3703 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
3704 is accessed through the union type. So, the code above will work as
3705 expected. However, this code might not:
3717 Every language that wishes to perform language-specific alias analysis
3718 should define a function that computes, given an @code{tree}
3719 node, an alias set for the node. Nodes in different alias sets are not
3720 allowed to alias. For an example, see the C front-end function
3721 @code{c_get_alias_set}.
3724 @item -falign-functions
3725 @itemx -falign-functions=@var{n}
3726 @opindex falign-functions
3727 Align the start of functions to the next power-of-two greater than
3728 @var{n}, skipping up to @var{n} bytes. For instance,
3729 @option{-falign-functions=32} aligns functions to the next 32-byte
3730 boundary, but @option{-falign-functions=24} would align to the next
3731 32-byte boundary only if this can be done by skipping 23 bytes or less.
3733 @option{-fno-align-functions} and @option{-falign-functions=1} are
3734 equivalent and mean that functions will not be aligned.
3736 Some assemblers only support this flag when @var{n} is a power of two;
3737 in that case, it is rounded up.
3739 If @var{n} is not specified, use a machine-dependent default.
3741 @item -falign-labels
3742 @itemx -falign-labels=@var{n}
3743 @opindex falign-labels
3744 Align all branch targets to a power-of-two boundary, skipping up to
3745 @var{n} bytes like @option{-falign-functions}. This option can easily
3746 make code slower, because it must insert dummy operations for when the
3747 branch target is reached in the usual flow of the code.
3749 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
3750 are greater than this value, then their values are used instead.
3752 If @var{n} is not specified, use a machine-dependent default which is
3753 very likely to be @samp{1}, meaning no alignment.
3756 @itemx -falign-loops=@var{n}
3757 @opindex falign-loops
3758 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
3759 like @option{-falign-functions}. The hope is that the loop will be
3760 executed many times, which will make up for any execution of the dummy
3763 If @var{n} is not specified, use a machine-dependent default.
3766 @itemx -falign-jumps=@var{n}
3767 @opindex falign-jumps
3768 Align branch targets to a power-of-two boundary, for branch targets
3769 where the targets can only be reached by jumping, skipping up to @var{n}
3770 bytes like @option{-falign-functions}. In this case, no dummy operations
3773 If @var{n} is not specified, use a machine-dependent default.
3777 Perform optimizations in static single assignment form. Each function's
3778 flow graph is translated into SSA form, optimizations are performed, and
3779 the flow graph is translated back from SSA form. Users should not
3780 specify this option, since it is not yet ready for production use.
3784 Perform Sparse Conditional Constant Propagation in SSA form. Requires
3785 @option{-fssa}. Like @option{-fssa}, this is an experimental feature.
3789 Perform aggressive dead-code elimination in SSA form. Requires @option{-fssa}.
3790 Like @option{-fssa}, this is an experimental feature.
3792 @item -fsingle-precision-constant
3793 @opindex fsingle-precision-constant
3794 Treat floating point constant as single precision constant instead of
3795 implicitly converting it to double precision constant.
3797 @item -frename-registers
3798 @opindex frename-registers
3799 Attempt to avoid false dependencies in scheduled code by making use
3800 of registers left over after register allocation. This optimization
3801 will most benefit processors with lots of registers. It can, however,
3802 make debugging impossible, since variables will no longer stay in
3803 a ``home register''.
3805 @item --param @var{name}=@var{value}
3807 In some places, GCC uses various constants to control the amount of
3808 optimization that is done. For example, GCC will not inline functions
3809 that contain more that a certain number of instructions. You can
3810 control some of these constants on the command-line using the
3811 @option{--param} option.
3813 In each case, the @var{value} is an integer. The allowable choices for
3814 @var{name} are given in the following table:
3817 @item max-delay-slot-insn-search
3818 The maximum number of instructions to consider when looking for an
3819 instruction to fill a delay slot. If more than this arbitrary number of
3820 instructions is searched, the time savings from filling the delay slot
3821 will be minimal so stop searching. Increasing values mean more
3822 aggressive optimization, making the compile time increase with probably
3823 small improvement in executable run time.
3825 @item max-delay-slot-live-search
3826 When trying to fill delay slots, the maximum number of instructions to
3827 consider when searching for a block with valid live register
3828 information. Increasing this arbitrarily chosen value means more
3829 aggressive optimization, increasing the compile time. This parameter
3830 should be removed when the delay slot code is rewritten to maintain the
3833 @item max-gcse-memory
3834 The approximate maximum amount of memory that will be allocated in
3835 order to perform the global common subexpression elimination
3836 optimization. If more memory than specified is required, the
3837 optimization will not be done.
3839 @item max-gcse-passes
3840 The maximum number of passes of GCSE to run.
3842 @item max-pending-list-length
3843 The maximum number of pending dependencies scheduling will allow
3844 before flushing the current state and starting over. Large functions
3845 with few branches or calls can create excessively large lists which
3846 needlessly consume memory and resources.
3848 @item max-inline-insns
3849 If an function contains more than this many instructions, it
3850 will not be inlined. This option is precisely equivalent to
3851 @option{-finline-limit}.
3856 @node Preprocessor Options
3857 @section Options Controlling the Preprocessor
3858 @cindex preprocessor options
3859 @cindex options, preprocessor
3861 These options control the C preprocessor, which is run on each C source
3862 file before actual compilation.
3864 If you use the @option{-E} option, nothing is done except preprocessing.
3865 Some of these options make sense only together with @option{-E} because
3866 they cause the preprocessor output to be unsuitable for actual
3870 @item -include @var{file}
3872 Process @var{file} as input before processing the regular input file.
3873 In effect, the contents of @var{file} are compiled first. Any @option{-D}
3874 and @option{-U} options on the command line are always processed before
3875 @option{-include @var{file}}, regardless of the order in which they are
3876 written. All the @option{-include} and @option{-imacros} options are
3877 processed in the order in which they are written.
3879 @item -imacros @var{file}
3881 Process @var{file} as input, discarding the resulting output, before
3882 processing the regular input file. Because the output generated from
3883 @var{file} is discarded, the only effect of @option{-imacros @var{file}}
3884 is to make the macros defined in @var{file} available for use in the
3885 main input. All the @option{-include} and @option{-imacros} options are
3886 processed in the order in which they are written.
3888 @item -idirafter @var{dir}
3890 @cindex second include path
3891 Add the directory @var{dir} to the second include path. The directories
3892 on the second include path are searched when a header file is not found
3893 in any of the directories in the main include path (the one that
3894 @option{-I} adds to).
3896 @item -iprefix @var{prefix}
3898 Specify @var{prefix} as the prefix for subsequent @option{-iwithprefix}
3901 @item -iwithprefix @var{dir}
3902 @opindex iwithprefix
3903 Add a directory to the second include path. The directory's name is
3904 made by concatenating @var{prefix} and @var{dir}, where @var{prefix} was
3905 specified previously with @option{-iprefix}. If you have not specified a
3906 prefix yet, the directory containing the installed passes of the
3907 compiler is used as the default.
3909 @item -iwithprefixbefore @var{dir}
3910 @opindex iwithprefixbefore
3911 Add a directory to the main include path. The directory's name is made
3912 by concatenating @var{prefix} and @var{dir}, as in the case of
3913 @option{-iwithprefix}.
3915 @item -isystem @var{dir}
3917 Add a directory to the beginning of the second include path, marking it
3918 as a system directory, so that it gets the same special treatment as
3919 is applied to the standard system directories.
3923 Do not search the standard system directories for header files. Only
3924 the directories you have specified with @option{-I} options (and the
3925 current directory, if appropriate) are searched. @xref{Directory
3926 Options}, for information on @option{-I}.
3928 By using both @option{-nostdinc} and @option{-I-}, you can limit the include-file
3929 search path to only those directories you specify explicitly.
3933 When searching for a header file in a directory, remap file names if a
3934 file named @file{header.gcc} exists in that directory. This can be used
3935 to work around limitations of file systems with file name restrictions.
3936 The @file{header.gcc} file should contain a series of lines with two
3937 tokens on each line: the first token is the name to map, and the second
3938 token is the actual name to use.
3942 Do not predefine any nonstandard macros. (Including architecture flags).
3946 Run only the C preprocessor. Preprocess all the C source files
3947 specified and output the results to standard output or to the
3948 specified output file.
3952 Tell the preprocessor not to discard comments. Used with the
3957 Tell the preprocessor not to generate @samp{#line} directives.
3958 Used with the @option{-E} option.
3961 @cindex dependencies, make
3964 Instead of outputting the result of preprocessing, output a rule
3965 suitable for @code{make} describing the dependencies of the main source
3966 file. The preprocessor outputs one @code{make} rule containing the
3967 object file name for that source file, a colon, and the names of all the
3968 included files. Unless overridden explicitly, the object file name
3969 consists of the basename of the source file with any suffix replaced with
3970 object file suffix. If there are many included files then the
3971 rule is split into several lines using @samp{\}-newline.
3973 @option{-M} implies @option{-E}.
3977 Like @option{-M}, but mention only the files included with @samp{#include
3978 "@var{file}"}. System header files included with @samp{#include
3979 <@var{file}>} are omitted.
3983 Like @option{-M} but the dependency information is written to a file
3984 rather than stdout. @code{gcc} will use the same file name and
3985 directory as the object file, but with the suffix @file{.d} instead.
3987 This is in addition to compiling the main file as specified---@option{-MD}
3988 does not inhibit ordinary compilation the way @option{-M} does,
3989 unless you also specify @option{-MG}.
3991 With Mach, you can use the utility @code{md} to merge multiple
3992 dependency files into a single dependency file suitable for using with
3993 the @samp{make} command.
3997 Like @option{-MD} except mention only user header files, not system
4000 @item -MF @var{file}
4002 When used with @option{-M} or @option{-MM}, specifies a file to write the
4003 dependencies to. This allows the preprocessor to write the preprocessed
4004 file to stdout normally. If no @option{-MF} switch is given, CPP sends
4005 the rules to stdout and suppresses normal preprocessed output.
4007 Another way to specify output of a @code{make} rule is by setting
4008 the environment variable @env{DEPENDENCIES_OUTPUT} (@pxref{Environment
4013 When used with @option{-M} or @option{-MM}, @option{-MG} says to treat missing
4014 header files as generated files and assume they live in the same
4015 directory as the source file. It suppresses preprocessed output, as a
4016 missing header file is ordinarily an error.
4018 This feature is used in automatic updating of makefiles.
4022 This option instructs CPP to add a phony target for each dependency
4023 other than the main file, causing each to depend on nothing. These
4024 dummy rules work around errors @code{make} gives if you remove header
4025 files without updating the @code{Makefile} to match.
4027 This is typical output:-
4030 /tmp/test.o: /tmp/test.c /tmp/test.h
4035 @item -MQ @var{target}
4036 @item -MT @var{target}
4039 By default CPP uses the main file name, including any path, and appends
4040 the object suffix, normally ``.o'', to it to obtain the name of the
4041 target for dependency generation. With @option{-MT} you can specify a
4042 target yourself, overriding the default one.
4044 If you want multiple targets, you can specify them as a single argument
4045 to @option{-MT}, or use multiple @option{-MT} options.
4047 The targets you specify are output in the order they appear on the
4048 command line. @option{-MQ} is identical to @option{-MT}, except that the
4049 target name is quoted for Make, but with @option{-MT} it isn't. For
4050 example, @option{-MT '$(objpfx)foo.o'} gives
4053 $(objpfx)foo.o: /tmp/foo.c
4056 but @option{-MQ '$(objpfx)foo.o'} gives
4059 $$(objpfx)foo.o: /tmp/foo.c
4062 The default target is automatically quoted, as if it were given with
4067 Print the name of each header file used, in addition to other normal
4070 @item -A@var{question}(@var{answer})
4072 Assert the answer @var{answer} for @var{question}, in case it is tested
4073 with a preprocessing conditional such as @samp{#if
4074 #@var{question}(@var{answer})}. @option{-A-} disables the standard
4075 assertions that normally describe the target machine.
4079 Define macro @var{macro} with the string @samp{1} as its definition.
4081 @item -D@var{macro}=@var{defn}
4082 Define macro @var{macro} as @var{defn}. All instances of @option{-D} on
4083 the command line are processed before any @option{-U} options.
4085 Any @option{-D} and @option{-U} options on the command line are processed in
4086 order, and always before @option{-imacros @var{file}}, regardless of the
4087 order in which they are written.
4091 Undefine macro @var{macro}. @option{-U} options are evaluated after all
4092 @option{-D} options, but before any @option{-include} and @option{-imacros}
4095 Any @option{-D} and @option{-U} options on the command line are processed in
4096 order, and always before @option{-imacros @var{file}}, regardless of the
4097 order in which they are written.
4101 Tell the preprocessor to output only a list of the macro definitions
4102 that are in effect at the end of preprocessing. Used with the @option{-E}
4107 Tell the preprocessing to pass all macro definitions into the output, in
4108 their proper sequence in the rest of the output.
4112 Like @option{-dD} except that the macro arguments and contents are omitted.
4113 Only @samp{#define @var{name}} is included in the output.
4117 Output @samp{#include} directives in addition to the result of
4120 @item -fpreprocessed
4121 @opindex fpreprocessed
4122 Indicate to the preprocessor that the input file has already been
4123 preprocessed. This suppresses things like macro expansion, trigraph
4124 conversion, escaped newline splicing, and processing of most directives.
4125 The preprocessor still recognizes and removes comments, so that you can
4126 pass a file preprocessed with @option{-C} to the compiler without
4127 problems. In this mode the integrated preprocessor is little more than
4128 a tokenizer for the front ends.
4130 @option{-fpreprocessed} is implicit if the input file has one of the
4131 extensions @samp{i}, @samp{ii} or @samp{mi}. These are the extensions
4132 that GCC uses for preprocessed files created by @option{-save-temps}.
4136 Process ISO standard trigraph sequences. These are three-character
4137 sequences, all starting with @samp{??}, that are defined by ISO C to
4138 stand for single characters. For example, @samp{??/} stands for
4139 @samp{\}, so @samp{'??/n'} is a character constant for a newline. By
4140 default, GCC ignores trigraphs, but in standard-conforming modes it
4141 converts them. See the @option{-std} and @option{-ansi} options.
4143 The nine trigraph sequences are
4146 @expansion{} @samp{[}
4149 @expansion{} @samp{]}
4152 @expansion{} @samp{@{}
4155 @expansion{} @samp{@}}
4158 @expansion{} @samp{#}
4161 @expansion{} @samp{\}
4164 @expansion{} @samp{^}
4167 @expansion{} @samp{|}
4170 @expansion{} @samp{~}
4174 Trigraph support is not popular, so many compilers do not implement it
4175 properly. Portable code should not rely on trigraphs being either
4176 converted or ignored.
4178 @item -Wp,@var{option}
4180 Pass @var{option} as an option to the preprocessor. If @var{option}
4181 contains commas, it is split into multiple options at the commas.
4184 @node Assembler Options
4185 @section Passing Options to the Assembler
4187 @c prevent bad page break with this line
4188 You can pass options to the assembler.
4191 @item -Wa,@var{option}
4193 Pass @var{option} as an option to the assembler. If @var{option}
4194 contains commas, it is split into multiple options at the commas.
4198 @section Options for Linking
4199 @cindex link options
4200 @cindex options, linking
4202 These options come into play when the compiler links object files into
4203 an executable output file. They are meaningless if the compiler is
4204 not doing a link step.
4208 @item @var{object-file-name}
4209 A file name that does not end in a special recognized suffix is
4210 considered to name an object file or library. (Object files are
4211 distinguished from libraries by the linker according to the file
4212 contents.) If linking is done, these object files are used as input
4221 If any of these options is used, then the linker is not run, and
4222 object file names should not be used as arguments. @xref{Overall
4226 @item -l@var{library}
4227 @itemx -l @var{library}
4229 Search the library named @var{library} when linking. (The second
4230 alternative with the library as a separate argument is only for
4231 POSIX compliance and is not recommended.)
4233 It makes a difference where in the command you write this option; the
4234 linker searches and processes libraries and object files in the order they
4235 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
4236 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
4237 to functions in @samp{z}, those functions may not be loaded.
4239 The linker searches a standard list of directories for the library,
4240 which is actually a file named @file{lib@var{library}.a}. The linker
4241 then uses this file as if it had been specified precisely by name.
4243 The directories searched include several standard system directories
4244 plus any that you specify with @option{-L}.
4246 Normally the files found this way are library files---archive files
4247 whose members are object files. The linker handles an archive file by
4248 scanning through it for members which define symbols that have so far
4249 been referenced but not defined. But if the file that is found is an
4250 ordinary object file, it is linked in the usual fashion. The only
4251 difference between using an @option{-l} option and specifying a file name
4252 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
4253 and searches several directories.
4257 You need this special case of the @option{-l} option in order to
4258 link an Objective-C program.
4261 @opindex nostartfiles
4262 Do not use the standard system startup files when linking.
4263 The standard system libraries are used normally, unless @option{-nostdlib}
4264 or @option{-nodefaultlibs} is used.
4266 @item -nodefaultlibs
4267 @opindex nodefaultlibs
4268 Do not use the standard system libraries when linking.
4269 Only the libraries you specify will be passed to the linker.
4270 The standard startup files are used normally, unless @option{-nostartfiles}
4271 is used. The compiler may generate calls to memcmp, memset, and memcpy
4272 for System V (and ISO C) environments or to bcopy and bzero for
4273 BSD environments. These entries are usually resolved by entries in
4274 libc. These entry points should be supplied through some other
4275 mechanism when this option is specified.
4279 Do not use the standard system startup files or libraries when linking.
4280 No startup files and only the libraries you specify will be passed to
4281 the linker. The compiler may generate calls to memcmp, memset, and memcpy
4282 for System V (and ISO C) environments or to bcopy and bzero for
4283 BSD environments. These entries are usually resolved by entries in
4284 libc. These entry points should be supplied through some other
4285 mechanism when this option is specified.
4287 @cindex @option{-lgcc}, use with @option{-nostdlib}
4288 @cindex @option{-nostdlib} and unresolved references
4289 @cindex unresolved references and @option{-nostdlib}
4290 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
4291 @cindex @option{-nodefaultlibs} and unresolved references
4292 @cindex unresolved references and @option{-nodefaultlibs}
4293 One of the standard libraries bypassed by @option{-nostdlib} and
4294 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
4295 that GCC uses to overcome shortcomings of particular machines, or special
4296 needs for some languages.
4298 (@xref{Interface,,Interfacing to GCC Output}, for more discussion of
4302 (@xref{Interface,,Interfacing to GCC Output,gcc.info,Porting GCC},
4303 for more discussion of @file{libgcc.a}.)
4305 In most cases, you need @file{libgcc.a} even when you want to avoid
4306 other standard libraries. In other words, when you specify @option{-nostdlib}
4307 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
4308 This ensures that you have no unresolved references to internal GCC
4309 library subroutines. (For example, @samp{__main}, used to ensure C++
4310 constructors will be called; @pxref{Collect2,,@command{collect2}}.)
4314 Remove all symbol table and relocation information from the executable.
4318 On systems that support dynamic linking, this prevents linking with the shared
4319 libraries. On other systems, this option has no effect.
4323 Produce a shared object which can then be linked with other objects to
4324 form an executable. Not all systems support this option. For predictable
4325 results, you must also specify the same set of options that were used to
4326 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
4327 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
4328 needs to build supplementary stub code for constructors to work. On
4329 multi-libbed systems, @samp{gcc -shared} must select the correct support
4330 libraries to link against. Failing to supply the correct flags may lead
4331 to subtle defects. Supplying them in cases where they are not necessary
4334 @item -shared-libgcc
4335 @itemx -static-libgcc
4336 @opindex shared-libgcc
4337 @opindex static-libgcc
4338 On systems that provide @file{libgcc} as a shared library, these options
4339 force the use of either the shared or static version respectively.
4340 If no shared version of @file{libgcc} was built when the compiler was
4341 configured, these options have no effect.
4343 There are several situations in which an application should use the
4344 shared @file{libgcc} instead of the static version. The most common
4345 of these is when the application wishes to throw and catch exceptions
4346 across different shared libraries. In that case, each of the libraries
4347 as well as the application itself should use the shared @file{libgcc}.
4349 Therefore, whenever you specify the @option{-shared} option, the GCC
4350 driver automatically adds @option{-shared-libgcc}, unless you explicitly
4351 specify @option{-static-libgcc}. The G++ driver automatically adds
4352 @option{-shared-libgcc} when you build a main executable as well because
4353 for C++ programs that is typically the right thing to do.
4354 (Exception-handling will not work reliably otherwise.)
4356 However, when linking a main executable written in C, you must
4357 explicitly say @option{-shared-libgcc} if you want to use the shared
4362 Bind references to global symbols when building a shared object. Warn
4363 about any unresolved references (unless overridden by the link editor
4364 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
4367 @item -Xlinker @var{option}
4369 Pass @var{option} as an option to the linker. You can use this to
4370 supply system-specific linker options which GCC does not know how to
4373 If you want to pass an option that takes an argument, you must use
4374 @option{-Xlinker} twice, once for the option and once for the argument.
4375 For example, to pass @option{-assert definitions}, you must write
4376 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
4377 @option{-Xlinker "-assert definitions"}, because this passes the entire
4378 string as a single argument, which is not what the linker expects.
4380 @item -Wl,@var{option}
4382 Pass @var{option} as an option to the linker. If @var{option} contains
4383 commas, it is split into multiple options at the commas.
4385 @item -u @var{symbol}
4387 Pretend the symbol @var{symbol} is undefined, to force linking of
4388 library modules to define it. You can use @option{-u} multiple times with
4389 different symbols to force loading of additional library modules.
4392 @node Directory Options
4393 @section Options for Directory Search
4394 @cindex directory options
4395 @cindex options, directory search
4398 These options specify directories to search for header files, for
4399 libraries and for parts of the compiler:
4404 Add the directory @var{dir} to the head of the list of directories to be
4405 searched for header files. This can be used to override a system header
4406 file, substituting your own version, since these directories are
4407 searched before the system header file directories. However, you should
4408 not use this option to add directories that contain vendor-supplied
4409 system header files (use @option{-isystem} for that). If you use more than
4410 one @option{-I} option, the directories are scanned in left-to-right
4411 order; the standard system directories come after.
4413 If a standard system include directory, or a directory specified with
4414 @option{-isystem}, is also specified with @option{-I}, it will be
4415 searched only in the position requested by @option{-I}. Also, it will
4416 not be considered a system include directory. If that directory really
4417 does contain system headers, there is a good chance that they will
4418 break. For instance, if GCC's installation procedure edited the headers
4419 in @file{/usr/include} to fix bugs, @samp{-I/usr/include} will cause the
4420 original, buggy headers to be found instead of the corrected ones. GCC
4421 will issue a warning when a system include directory is hidden in this
4426 Any directories you specify with @option{-I} options before the @option{-I-}
4427 option are searched only for the case of @samp{#include "@var{file}"};
4428 they are not searched for @samp{#include <@var{file}>}.
4430 If additional directories are specified with @option{-I} options after
4431 the @option{-I-}, these directories are searched for all @samp{#include}
4432 directives. (Ordinarily @emph{all} @option{-I} directories are used
4435 In addition, the @option{-I-} option inhibits the use of the current
4436 directory (where the current input file came from) as the first search
4437 directory for @samp{#include "@var{file}"}. There is no way to
4438 override this effect of @option{-I-}. With @option{-I.} you can specify
4439 searching the directory which was current when the compiler was
4440 invoked. That is not exactly the same as what the preprocessor does
4441 by default, but it is often satisfactory.
4443 @option{-I-} does not inhibit the use of the standard system directories
4444 for header files. Thus, @option{-I-} and @option{-nostdinc} are
4449 Add directory @var{dir} to the list of directories to be searched
4452 @item -B@var{prefix}
4454 This option specifies where to find the executables, libraries,
4455 include files, and data files of the compiler itself.
4457 The compiler driver program runs one or more of the subprograms
4458 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
4459 @var{prefix} as a prefix for each program it tries to run, both with and
4460 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
4462 For each subprogram to be run, the compiler driver first tries the
4463 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
4464 was not specified, the driver tries two standard prefixes, which are
4465 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc-lib/}. If neither of
4466 those results in a file name that is found, the unmodified program
4467 name is searched for using the directories specified in your
4468 @env{PATH} environment variable.
4470 The compiler will check to see if the path provided by the @option{-B}
4471 refers to a directory, and if necessary it will add a directory
4472 separator character at the end of the path.
4474 @option{-B} prefixes that effectively specify directory names also apply
4475 to libraries in the linker, because the compiler translates these
4476 options into @option{-L} options for the linker. They also apply to
4477 includes files in the preprocessor, because the compiler translates these
4478 options into @option{-isystem} options for the preprocessor. In this case,
4479 the compiler appends @samp{include} to the prefix.
4481 The run-time support file @file{libgcc.a} can also be searched for using
4482 the @option{-B} prefix, if needed. If it is not found there, the two
4483 standard prefixes above are tried, and that is all. The file is left
4484 out of the link if it is not found by those means.
4486 Another way to specify a prefix much like the @option{-B} prefix is to use
4487 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
4490 As a special kludge, if the path provided by @option{-B} is
4491 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
4492 9, then it will be replaced by @file{[dir/]include}. This is to help
4493 with boot-strapping the compiler.
4495 @item -specs=@var{file}
4497 Process @var{file} after the compiler reads in the standard @file{specs}
4498 file, in order to override the defaults that the @file{gcc} driver
4499 program uses when determining what switches to pass to @file{cc1},
4500 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
4501 @option{-specs=@var{file}} can be specified on the command line, and they
4502 are processed in order, from left to right.
4508 @section Specifying subprocesses and the switches to pass to them
4510 @command{gcc} is a driver program. It performs its job by invoking a
4511 sequence of other programs to do the work of compiling, assembling and
4512 linking. GCC interprets its command-line parameters and uses these to
4513 deduce which programs it should invoke, and which command-line options
4514 it ought to place on their command lines. This behaviour is controlled
4515 by @dfn{spec strings}. In most cases there is one spec string for each
4516 program that GCC can invoke, but a few programs have multiple spec
4517 strings to control their behaviour. The spec strings built into GCC can
4518 be overridden by using the @option{-specs=} command-line switch to specify
4521 @dfn{Spec files} are plaintext files that are used to construct spec
4522 strings. They consist of a sequence of directives separated by blank
4523 lines. The type of directive is determined by the first non-whitespace
4524 character on the line and it can be one of the following:
4527 @item %@var{command}
4528 Issues a @var{command} to the spec file processor. The commands that can
4532 @item %include <@var{file}>
4534 Search for @var{file} and insert its text at the current point in the
4537 @item %include_noerr <@var{file}>
4538 @cindex %include_noerr
4539 Just like @samp{%include}, but do not generate an error message if the include
4540 file cannot be found.
4542 @item %rename @var{old_name} @var{new_name}
4544 Rename the spec string @var{old_name} to @var{new_name}.
4548 @item *[@var{spec_name}]:
4549 This tells the compiler to create, override or delete the named spec
4550 string. All lines after this directive up to the next directive or
4551 blank line are considered to be the text for the spec string. If this
4552 results in an empty string then the spec will be deleted. (Or, if the
4553 spec did not exist, then nothing will happened.) Otherwise, if the spec
4554 does not currently exist a new spec will be created. If the spec does
4555 exist then its contents will be overridden by the text of this
4556 directive, unless the first character of that text is the @samp{+}
4557 character, in which case the text will be appended to the spec.
4559 @item [@var{suffix}]:
4560 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
4561 and up to the next directive or blank line are considered to make up the
4562 spec string for the indicated suffix. When the compiler encounters an
4563 input file with the named suffix, it will processes the spec string in
4564 order to work out how to compile that file. For example:
4571 This says that any input file whose name ends in @samp{.ZZ} should be
4572 passed to the program @samp{z-compile}, which should be invoked with the
4573 command-line switch @option{-input} and with the result of performing the
4574 @samp{%i} substitution. (See below.)
4576 As an alternative to providing a spec string, the text that follows a
4577 suffix directive can be one of the following:
4580 @item @@@var{language}
4581 This says that the suffix is an alias for a known @var{language}. This is
4582 similar to using the @option{-x} command-line switch to GCC to specify a
4583 language explicitly. For example:
4590 Says that .ZZ files are, in fact, C++ source files.
4593 This causes an error messages saying:
4596 @var{name} compiler not installed on this system.
4600 GCC already has an extensive list of suffixes built into it.
4601 This directive will add an entry to the end of the list of suffixes, but
4602 since the list is searched from the end backwards, it is effectively
4603 possible to override earlier entries using this technique.
4607 GCC has the following spec strings built into it. Spec files can
4608 override these strings or create their own. Note that individual
4609 targets can also add their own spec strings to this list.
4612 asm Options to pass to the assembler
4613 asm_final Options to pass to the assembler post-processor
4614 cpp Options to pass to the C preprocessor
4615 cc1 Options to pass to the C compiler
4616 cc1plus Options to pass to the C++ compiler
4617 endfile Object files to include at the end of the link
4618 link Options to pass to the linker
4619 lib Libraries to include on the command line to the linker
4620 libgcc Decides which GCC support library to pass to the linker
4621 linker Sets the name of the linker
4622 predefines Defines to be passed to the C preprocessor
4623 signed_char Defines to pass to CPP to say whether @code{char} is signed
4625 startfile Object files to include at the start of the link
4628 Here is a small example of a spec file:
4634 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
4637 This example renames the spec called @samp{lib} to @samp{old_lib} and
4638 then overrides the previous definition of @samp{lib} with a new one.
4639 The new definition adds in some extra command-line options before
4640 including the text of the old definition.
4642 @dfn{Spec strings} are a list of command-line options to be passed to their
4643 corresponding program. In addition, the spec strings can contain
4644 @samp{%}-prefixed sequences to substitute variable text or to
4645 conditionally insert text into the command line. Using these constructs
4646 it is possible to generate quite complex command lines.
4648 Here is a table of all defined @samp{%}-sequences for spec
4649 strings. Note that spaces are not generated automatically around the
4650 results of expanding these sequences. Therefore you can concatenate them
4651 together or combine them with constant text in a single argument.
4655 Substitute one @samp{%} into the program name or argument.
4658 Substitute the name of the input file being processed.
4661 Substitute the basename of the input file being processed.
4662 This is the substring up to (and not including) the last period
4663 and not including the directory.
4666 This is the same as @samp{%b}, but include the file suffix (text after
4670 Marks the argument containing or following the @samp{%d} as a
4671 temporary file name, so that that file will be deleted if GCC exits
4672 successfully. Unlike @samp{%g}, this contributes no text to the
4675 @item %g@var{suffix}
4676 Substitute a file name that has suffix @var{suffix} and is chosen
4677 once per compilation, and mark the argument in the same way as
4678 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
4679 name is now chosen in a way that is hard to predict even when previously
4680 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
4681 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
4682 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
4683 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
4684 was simply substituted with a file name chosen once per compilation,
4685 without regard to any appended suffix (which was therefore treated
4686 just like ordinary text), making such attacks more likely to succeed.
4688 @item %u@var{suffix}
4689 Like @samp{%g}, but generates a new temporary file name even if
4690 @samp{%u@var{suffix}} was already seen.
4692 @item %U@var{suffix}
4693 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
4694 new one if there is no such last file name. In the absence of any
4695 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
4696 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
4697 would involve the generation of two distinct file names, one
4698 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
4699 simply substituted with a file name chosen for the previous @samp{%u},
4700 without regard to any appended suffix.
4702 @item %j@var{SUFFIX}
4703 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
4704 writable, and if save-temps is off; otherwise, substitute the name
4705 of a temporary file, just like @samp{%u}. This temporary file is not
4706 meant for communication between processes, but rather as a junk
4709 @item %.@var{SUFFIX}
4710 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
4711 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
4712 terminated by the next space or %.
4715 Marks the argument containing or following the @samp{%w} as the
4716 designated output file of this compilation. This puts the argument
4717 into the sequence of arguments that @samp{%o} will substitute later.
4720 Substitutes the names of all the output files, with spaces
4721 automatically placed around them. You should write spaces
4722 around the @samp{%o} as well or the results are undefined.
4723 @samp{%o} is for use in the specs for running the linker.
4724 Input files whose names have no recognized suffix are not compiled
4725 at all, but they are included among the output files, so they will
4729 Substitutes the suffix for object files. Note that this is
4730 handled specially when it immediately follows @samp{%g, %u, or %U},
4731 because of the need for those to form complete file names. The
4732 handling is such that @samp{%O} is treated exactly as if it had already
4733 been substituted, except that @samp{%g, %u, and %U} do not currently
4734 support additional @var{suffix} characters following @samp{%O} as they would
4735 following, for example, @samp{.o}.
4738 Substitutes the standard macro predefinitions for the
4739 current target machine. Use this when running @code{cpp}.
4742 Like @samp{%p}, but puts @samp{__} before and after the name of each
4743 predefined macro, except for macros that start with @samp{__} or with
4744 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
4748 Substitute a @option{-iprefix} option made from @env{GCC_EXEC_PREFIX}.
4751 Current argument is the name of a library or startup file of some sort.
4752 Search for that file in a standard list of directories and substitute
4753 the full name found.
4756 Print @var{str} as an error message. @var{str} is terminated by a newline.
4757 Use this when inconsistent options are detected.
4760 Output @samp{-} if the input for the current command is coming from a pipe.
4763 Substitute the contents of spec string @var{name} at this point.
4766 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
4768 @item %x@{@var{option}@}
4769 Accumulate an option for @samp{%X}.
4772 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
4776 Output the accumulated assembler options specified by @option{-Wa}.
4779 Output the accumulated preprocessor options specified by @option{-Wp}.
4782 Substitute the major version number of GCC@.
4783 (For version 2.9.5, this is 2.)
4786 Substitute the minor version number of GCC@.
4787 (For version 2.9.5, this is 9.)
4790 Substitute the patch level number of GCC@.
4791 (For version 2.9.5, this is 5.)
4794 Process the @code{asm} spec. This is used to compute the
4795 switches to be passed to the assembler.
4798 Process the @code{asm_final} spec. This is a spec string for
4799 passing switches to an assembler post-processor, if such a program is
4803 Process the @code{link} spec. This is the spec for computing the
4804 command line passed to the linker. Typically it will make use of the
4805 @samp{%L %G %S %D and %E} sequences.
4808 Dump out a @option{-L} option for each directory that GCC believes might
4809 contain startup files. If the target supports multilibs then the
4810 current multilib directory will be prepended to each of these paths.
4813 Output the multilib directory with directory separators replaced with
4814 @samp{_}. If multilib directories are not set, or the multilib directory is
4815 @file{.} then this option emits nothing.
4818 Process the @code{lib} spec. This is a spec string for deciding which
4819 libraries should be included on the command line to the linker.
4822 Process the @code{libgcc} spec. This is a spec string for deciding
4823 which GCC support library should be included on the command line to the linker.
4826 Process the @code{startfile} spec. This is a spec for deciding which
4827 object files should be the first ones passed to the linker. Typically
4828 this might be a file named @file{crt0.o}.
4831 Process the @code{endfile} spec. This is a spec string that specifies
4832 the last object files that will be passed to the linker.
4835 Process the @code{cpp} spec. This is used to construct the arguments
4836 to be passed to the C preprocessor.
4839 Process the @code{signed_char} spec. This is intended to be used
4840 to tell cpp whether a char is signed. It typically has the definition:
4842 %@{funsigned-char:-D__CHAR_UNSIGNED__@}
4846 Process the @code{cc1} spec. This is used to construct the options to be
4847 passed to the actual C compiler (@samp{cc1}).
4850 Process the @code{cc1plus} spec. This is used to construct the options to be
4851 passed to the actual C++ compiler (@samp{cc1plus}).
4854 Substitute the variable part of a matched option. See below.
4855 Note that each comma in the substituted string is replaced by
4859 Substitutes the @code{-S} switch, if that switch was given to GCC@.
4860 If that switch was not specified, this substitutes nothing. Note that
4861 the leading dash is omitted when specifying this option, and it is
4862 automatically inserted if the substitution is performed. Thus the spec
4863 string @samp{%@{foo@}} would match the command-line option @option{-foo}
4864 and would output the command line option @option{-foo}.
4866 @item %W@{@code{S}@}
4867 Like %@{@code{S}@} but mark last argument supplied within as a file to be
4870 @item %@{@code{S}*@}
4871 Substitutes all the switches specified to GCC whose names start
4872 with @code{-S}, but which also take an argument. This is used for
4873 switches like @option{-o}, @option{-D}, @option{-I}, etc.
4874 GCC considers @option{-o foo} as being
4875 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
4876 text, including the space. Thus two arguments would be generated.
4878 @item %@{^@code{S}*@}
4879 Like %@{@code{S}*@}, but don't put a blank between a switch and its
4880 argument. Thus %@{^o*@} would only generate one argument, not two.
4882 @item %@{@code{S}*&@code{T}*@}
4883 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
4884 (the order of @code{S} and @code{T} in the spec is not significant).
4885 There can be any number of ampersand-separated variables; for each the
4886 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
4888 @item %@{<@code{S}@}
4889 Remove all occurrences of @code{-S} from the command line. Note---this
4890 command is position dependent. @samp{%} commands in the spec string
4891 before this option will see @code{-S}, @samp{%} commands in the spec
4892 string after this option will not.
4894 @item %@{@code{S}*:@code{X}@}
4895 Substitutes @code{X} if one or more switches whose names start with
4896 @code{-S} are specified to GCC@. Note that the tail part of the
4897 @code{-S} option (i.e.@: the part matched by the @samp{*}) will be substituted
4898 for each occurrence of @samp{%*} within @code{X}.
4900 @item %@{@code{S}:@code{X}@}
4901 Substitutes @code{X}, but only if the @samp{-S} switch was given to GCC@.
4903 @item %@{!@code{S}:@code{X}@}
4904 Substitutes @code{X}, but only if the @samp{-S} switch was @emph{not} given to GCC@.
4906 @item %@{|@code{S}:@code{X}@}
4907 Like %@{@code{S}:@code{X}@}, but if no @code{S} switch, substitute @samp{-}.
4909 @item %@{|!@code{S}:@code{X}@}
4910 Like %@{!@code{S}:@code{X}@}, but if there is an @code{S} switch, substitute @samp{-}.
4912 @item %@{.@code{S}:@code{X}@}
4913 Substitutes @code{X}, but only if processing a file with suffix @code{S}.
4915 @item %@{!.@code{S}:@code{X}@}
4916 Substitutes @code{X}, but only if @emph{not} processing a file with suffix @code{S}.
4918 @item %@{@code{S}|@code{P}:@code{X}@}
4919 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC@. This may be
4920 combined with @samp{!} and @samp{.} sequences as well, although they
4921 have a stronger binding than the @samp{|}. For example a spec string
4925 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
4928 will output the following command-line options from the following input
4929 command-line options:
4934 -d fred.c -foo -baz -boggle
4935 -d jim.d -bar -baz -boggle
4940 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or
4941 %@{!@code{S}:@code{X}@} construct may contain other nested @samp{%} constructs
4942 or spaces, or even newlines. They are processed as usual, as described
4945 The @option{-O}, @option{-f}, @option{-m}, and @option{-W}
4946 switches are handled specifically in these
4947 constructs. If another value of @option{-O} or the negated form of a @option{-f}, @option{-m}, or
4948 @option{-W} switch is found later in the command line, the earlier switch
4949 value is ignored, except with @{@code{S}*@} where @code{S} is just one
4950 letter, which passes all matching options.
4952 The character @samp{|} at the beginning of the predicate text is used to indicate
4953 that a command should be piped to the following command, but only if @option{-pipe}
4956 It is built into GCC which switches take arguments and which do not.
4957 (You might think it would be useful to generalize this to allow each
4958 compiler's spec to say which switches take arguments. But this cannot
4959 be done in a consistent fashion. GCC cannot even decide which input
4960 files have been specified without knowing which switches take arguments,
4961 and it must know which input files to compile in order to tell which
4964 GCC also knows implicitly that arguments starting in @option{-l} are to be
4965 treated as compiler output files, and passed to the linker in their
4966 proper position among the other output files.
4968 @c man begin OPTIONS
4970 @node Target Options
4971 @section Specifying Target Machine and Compiler Version
4972 @cindex target options
4973 @cindex cross compiling
4974 @cindex specifying machine version
4975 @cindex specifying compiler version and target machine
4976 @cindex compiler version, specifying
4977 @cindex target machine, specifying
4979 By default, GCC compiles code for the same type of machine that you
4980 are using. However, it can also be installed as a cross-compiler, to
4981 compile for some other type of machine. In fact, several different
4982 configurations of GCC, for different target machines, can be
4983 installed side by side. Then you specify which one to use with the
4986 In addition, older and newer versions of GCC can be installed side
4987 by side. One of them (probably the newest) will be the default, but
4988 you may sometimes wish to use another.
4991 @item -b @var{machine}
4993 The argument @var{machine} specifies the target machine for compilation.
4994 This is useful when you have installed GCC as a cross-compiler.
4996 The value to use for @var{machine} is the same as was specified as the
4997 machine type when configuring GCC as a cross-compiler. For
4998 example, if a cross-compiler was configured with @samp{configure
4999 i386v}, meaning to compile for an 80386 running System V, then you
5000 would specify @option{-b i386v} to run that cross compiler.
5002 When you do not specify @option{-b}, it normally means to compile for
5003 the same type of machine that you are using.
5005 @item -V @var{version}
5007 The argument @var{version} specifies which version of GCC to run.
5008 This is useful when multiple versions are installed. For example,
5009 @var{version} might be @samp{2.0}, meaning to run GCC version 2.0.
5011 The default version, when you do not specify @option{-V}, is the last
5012 version of GCC that you installed.
5015 The @option{-b} and @option{-V} options actually work by controlling part of
5016 the file name used for the executable files and libraries used for
5017 compilation. A given version of GCC, for a given target machine, is
5018 normally kept in the directory @file{/usr/local/lib/gcc-lib/@var{machine}/@var{version}}.
5020 Thus, sites can customize the effect of @option{-b} or @option{-V} either by
5021 changing the names of these directories or adding alternate names (or
5022 symbolic links). If in directory @file{/usr/local/lib/gcc-lib/} the
5023 file @file{80386} is a link to the file @file{i386v}, then @option{-b
5024 80386} becomes an alias for @option{-b i386v}.
5026 In one respect, the @option{-b} or @option{-V} do not completely change
5027 to a different compiler: the top-level driver program @command{gcc}
5028 that you originally invoked continues to run and invoke the other
5029 executables (preprocessor, compiler per se, assembler and linker)
5030 that do the real work. However, since no real work is done in the
5031 driver program, it usually does not matter that the driver program
5032 in use is not the one for the specified target. It is common for the
5033 interface to the other executables to change incompatibly between
5034 compiler versions, so unless the version specified is very close to that
5035 of the driver (for example, @option{-V 3.0} with a driver program from GCC
5036 version 3.0.1), use of @option{-V} may not work; for example, using
5037 @option{-V 2.95.2} will not work with a driver program from GCC 3.0.
5039 The only way that the driver program depends on the target machine is
5040 in the parsing and handling of special machine-specific options.
5041 However, this is controlled by a file which is found, along with the
5042 other executables, in the directory for the specified version and
5043 target machine. As a result, a single installed driver program adapts
5044 to any specified target machine, and sufficiently similar compiler
5047 The driver program executable does control one significant thing,
5048 however: the default version and target machine. Therefore, you can
5049 install different instances of the driver program, compiled for
5050 different targets or versions, under different names.
5052 For example, if the driver for version 2.0 is installed as @command{ogcc}
5053 and that for version 2.1 is installed as @command{gcc}, then the command
5054 @command{gcc} will use version 2.1 by default, while @command{ogcc} will use
5055 2.0 by default. However, you can choose either version with either
5056 command with the @option{-V} option.
5058 @node Submodel Options
5059 @section Hardware Models and Configurations
5060 @cindex submodel options
5061 @cindex specifying hardware config
5062 @cindex hardware models and configurations, specifying
5063 @cindex machine dependent options
5065 Earlier we discussed the standard option @option{-b} which chooses among
5066 different installed compilers for completely different target
5067 machines, such as VAX vs.@: 68000 vs.@: 80386.
5069 In addition, each of these target machine types can have its own
5070 special options, starting with @samp{-m}, to choose among various
5071 hardware models or configurations---for example, 68010 vs 68020,
5072 floating coprocessor or none. A single installed version of the
5073 compiler can compile for any model or configuration, according to the
5076 Some configurations of the compiler also support additional special
5077 options, usually for compatibility with other compilers on the same
5081 These options are defined by the macro @code{TARGET_SWITCHES} in the
5082 machine description. The default for the options is also defined by
5083 that macro, which enables you to change the defaults.
5098 * RS/6000 and PowerPC Options::
5101 * i386 and x86-64 Options::
5103 * Intel 960 Options::
5104 * DEC Alpha Options::
5108 * System V Options::
5109 * TMS320C3x/C4x Options::
5117 * S/390 and zSeries Options::
5121 @node M680x0 Options
5122 @subsection M680x0 Options
5123 @cindex M680x0 options
5125 These are the @samp{-m} options defined for the 68000 series. The default
5126 values for these options depends on which style of 68000 was selected when
5127 the compiler was configured; the defaults for the most common choices are
5135 Generate output for a 68000. This is the default
5136 when the compiler is configured for 68000-based systems.
5138 Use this option for microcontrollers with a 68000 or EC000 core,
5139 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
5145 Generate output for a 68020. This is the default
5146 when the compiler is configured for 68020-based systems.
5150 Generate output containing 68881 instructions for floating point.
5151 This is the default for most 68020 systems unless @option{--nfp} was
5152 specified when the compiler was configured.
5156 Generate output for a 68030. This is the default when the compiler is
5157 configured for 68030-based systems.
5161 Generate output for a 68040. This is the default when the compiler is
5162 configured for 68040-based systems.
5164 This option inhibits the use of 68881/68882 instructions that have to be
5165 emulated by software on the 68040. Use this option if your 68040 does not
5166 have code to emulate those instructions.
5170 Generate output for a 68060. This is the default when the compiler is
5171 configured for 68060-based systems.
5173 This option inhibits the use of 68020 and 68881/68882 instructions that
5174 have to be emulated by software on the 68060. Use this option if your 68060
5175 does not have code to emulate those instructions.
5179 Generate output for a CPU32. This is the default
5180 when the compiler is configured for CPU32-based systems.
5182 Use this option for microcontrollers with a
5183 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
5184 68336, 68340, 68341, 68349 and 68360.
5188 Generate output for a 520X ``coldfire'' family cpu. This is the default
5189 when the compiler is configured for 520X-based systems.
5191 Use this option for microcontroller with a 5200 core, including
5192 the MCF5202, MCF5203, MCF5204 and MCF5202.
5197 Generate output for a 68040, without using any of the new instructions.
5198 This results in code which can run relatively efficiently on either a
5199 68020/68881 or a 68030 or a 68040. The generated code does use the
5200 68881 instructions that are emulated on the 68040.
5204 Generate output for a 68060, without using any of the new instructions.
5205 This results in code which can run relatively efficiently on either a
5206 68020/68881 or a 68030 or a 68040. The generated code does use the
5207 68881 instructions that are emulated on the 68060.
5211 Generate output containing Sun FPA instructions for floating point.
5214 @opindex msoft-float
5215 Generate output containing library calls for floating point.
5216 @strong{Warning:} the requisite libraries are not available for all m68k
5217 targets. Normally the facilities of the machine's usual C compiler are
5218 used, but this can't be done directly in cross-compilation. You must
5219 make your own arrangements to provide suitable library functions for
5220 cross-compilation. The embedded targets @samp{m68k-*-aout} and
5221 @samp{m68k-*-coff} do provide software floating point support.
5225 Consider type @code{int} to be 16 bits wide, like @code{short int}.
5228 @opindex mnobitfield
5229 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
5230 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
5234 Do use the bit-field instructions. The @option{-m68020} option implies
5235 @option{-mbitfield}. This is the default if you use a configuration
5236 designed for a 68020.
5240 Use a different function-calling convention, in which functions
5241 that take a fixed number of arguments return with the @code{rtd}
5242 instruction, which pops their arguments while returning. This
5243 saves one instruction in the caller since there is no need to pop
5244 the arguments there.
5246 This calling convention is incompatible with the one normally
5247 used on Unix, so you cannot use it if you need to call libraries
5248 compiled with the Unix compiler.
5250 Also, you must provide function prototypes for all functions that
5251 take variable numbers of arguments (including @code{printf});
5252 otherwise incorrect code will be generated for calls to those
5255 In addition, seriously incorrect code will result if you call a
5256 function with too many arguments. (Normally, extra arguments are
5257 harmlessly ignored.)
5259 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
5260 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
5263 @itemx -mno-align-int
5265 @opindex mno-align-int
5266 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
5267 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
5268 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
5269 Aligning variables on 32-bit boundaries produces code that runs somewhat
5270 faster on processors with 32-bit busses at the expense of more memory.
5272 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
5273 align structures containing the above types differently than
5274 most published application binary interface specifications for the m68k.
5278 Use the pc-relative addressing mode of the 68000 directly, instead of
5279 using a global offset table. At present, this option implies @option{-fpic},
5280 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
5281 not presently supported with @option{-mpcrel}, though this could be supported for
5282 68020 and higher processors.
5284 @item -mno-strict-align
5285 @itemx -mstrict-align
5286 @opindex mno-strict-align
5287 @opindex mstrict-align
5288 Do not (do) assume that unaligned memory references will be handled by
5293 @node M68hc1x Options
5294 @subsection M68hc1x Options
5295 @cindex M68hc1x options
5297 These are the @samp{-m} options defined for the 68hc11 and 68hc12
5298 microcontrollers. The default values for these options depends on
5299 which style of microcontroller was selected when the compiler was configured;
5300 the defaults for the most common choices are given below.
5307 Generate output for a 68HC11. This is the default
5308 when the compiler is configured for 68HC11-based systems.
5314 Generate output for a 68HC12. This is the default
5315 when the compiler is configured for 68HC12-based systems.
5318 @opindex mauto-incdec
5319 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
5324 Consider type @code{int} to be 16 bits wide, like @code{short int}.
5326 @item -msoft-reg-count=@var{count}
5327 @opindex msoft-reg-count
5328 Specify the number of pseudo-soft registers which are used for the
5329 code generation. The maximum number is 32. Using more pseudo-soft
5330 register may or may not result in better code depending on the program.
5331 The default is 4 for 68HC11 and 2 for 68HC12.
5336 @subsection VAX Options
5339 These @samp{-m} options are defined for the VAX:
5344 Do not output certain jump instructions (@code{aobleq} and so on)
5345 that the Unix assembler for the VAX cannot handle across long
5350 Do output those jump instructions, on the assumption that you
5351 will assemble with the GNU assembler.
5355 Output code for g-format floating point numbers instead of d-format.
5359 @subsection SPARC Options
5360 @cindex SPARC options
5362 These @samp{-m} switches are supported on the SPARC:
5367 @opindex mno-app-regs
5369 Specify @option{-mapp-regs} to generate output using the global registers
5370 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
5373 To be fully SVR4 ABI compliant at the cost of some performance loss,
5374 specify @option{-mno-app-regs}. You should compile libraries and system
5375 software with this option.
5380 @opindex mhard-float
5381 Generate output containing floating point instructions. This is the
5387 @opindex msoft-float
5388 Generate output containing library calls for floating point.
5389 @strong{Warning:} the requisite libraries are not available for all SPARC
5390 targets. Normally the facilities of the machine's usual C compiler are
5391 used, but this cannot be done directly in cross-compilation. You must make
5392 your own arrangements to provide suitable library functions for
5393 cross-compilation. The embedded targets @samp{sparc-*-aout} and
5394 @samp{sparclite-*-*} do provide software floating point support.
5396 @option{-msoft-float} changes the calling convention in the output file;
5397 therefore, it is only useful if you compile @emph{all} of a program with
5398 this option. In particular, you need to compile @file{libgcc.a}, the
5399 library that comes with GCC, with @option{-msoft-float} in order for
5402 @item -mhard-quad-float
5403 @opindex mhard-quad-float
5404 Generate output containing quad-word (long double) floating point
5407 @item -msoft-quad-float
5408 @opindex msoft-quad-float
5409 Generate output containing library calls for quad-word (long double)
5410 floating point instructions. The functions called are those specified
5411 in the SPARC ABI@. This is the default.
5413 As of this writing, there are no sparc implementations that have hardware
5414 support for the quad-word floating point instructions. They all invoke
5415 a trap handler for one of these instructions, and then the trap handler
5416 emulates the effect of the instruction. Because of the trap handler overhead,
5417 this is much slower than calling the ABI library routines. Thus the
5418 @option{-msoft-quad-float} option is the default.
5422 @opindex mno-epilogue
5424 With @option{-mepilogue} (the default), the compiler always emits code for
5425 function exit at the end of each function. Any function exit in
5426 the middle of the function (such as a return statement in C) will
5427 generate a jump to the exit code at the end of the function.
5429 With @option{-mno-epilogue}, the compiler tries to emit exit code inline
5430 at every function exit.
5436 With @option{-mflat}, the compiler does not generate save/restore instructions
5437 and will use a ``flat'' or single register window calling convention.
5438 This model uses %i7 as the frame pointer and is compatible with the normal
5439 register window model. Code from either may be intermixed.
5440 The local registers and the input registers (0--5) are still treated as
5441 ``call saved'' registers and will be saved on the stack as necessary.
5443 With @option{-mno-flat} (the default), the compiler emits save/restore
5444 instructions (except for leaf functions) and is the normal mode of operation.
5446 @item -mno-unaligned-doubles
5447 @itemx -munaligned-doubles
5448 @opindex mno-unaligned-doubles
5449 @opindex munaligned-doubles
5450 Assume that doubles have 8 byte alignment. This is the default.
5452 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
5453 alignment only if they are contained in another type, or if they have an
5454 absolute address. Otherwise, it assumes they have 4 byte alignment.
5455 Specifying this option avoids some rare compatibility problems with code
5456 generated by other compilers. It is not the default because it results
5457 in a performance loss, especially for floating point code.
5459 @item -mno-faster-structs
5460 @itemx -mfaster-structs
5461 @opindex mno-faster-structs
5462 @opindex mfaster-structs
5463 With @option{-mfaster-structs}, the compiler assumes that structures
5464 should have 8 byte alignment. This enables the use of pairs of
5465 @code{ldd} and @code{std} instructions for copies in structure
5466 assignment, in place of twice as many @code{ld} and @code{st} pairs.
5467 However, the use of this changed alignment directly violates the Sparc
5468 ABI@. Thus, it's intended only for use on targets where the developer
5469 acknowledges that their resulting code will not be directly in line with
5470 the rules of the ABI@.
5476 These two options select variations on the SPARC architecture.
5478 By default (unless specifically configured for the Fujitsu SPARClite),
5479 GCC generates code for the v7 variant of the SPARC architecture.
5481 @option{-mv8} will give you SPARC v8 code. The only difference from v7
5482 code is that the compiler emits the integer multiply and integer
5483 divide instructions which exist in SPARC v8 but not in SPARC v7.
5485 @option{-msparclite} will give you SPARClite code. This adds the integer
5486 multiply, integer divide step and scan (@code{ffs}) instructions which
5487 exist in SPARClite but not in SPARC v7.
5489 These options are deprecated and will be deleted in a future GCC release.
5490 They have been replaced with @option{-mcpu=xxx}.
5495 @opindex msupersparc
5496 These two options select the processor for which the code is optimised.
5498 With @option{-mcypress} (the default), the compiler optimizes code for the
5499 Cypress CY7C602 chip, as used in the SparcStation/SparcServer 3xx series.
5500 This is also appropriate for the older SparcStation 1, 2, IPX etc.
5502 With @option{-msupersparc} the compiler optimizes code for the SuperSparc cpu, as
5503 used in the SparcStation 10, 1000 and 2000 series. This flag also enables use
5504 of the full SPARC v8 instruction set.
5506 These options are deprecated and will be deleted in a future GCC release.
5507 They have been replaced with @option{-mcpu=xxx}.
5509 @item -mcpu=@var{cpu_type}
5511 Set the instruction set, register set, and instruction scheduling parameters
5512 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
5513 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
5514 @samp{hypersparc}, @samp{sparclite86x}, @samp{f930}, @samp{f934},
5515 @samp{sparclet}, @samp{tsc701}, @samp{v9}, and @samp{ultrasparc}.
5517 Default instruction scheduling parameters are used for values that select
5518 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
5519 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
5521 Here is a list of each supported architecture and their supported
5526 v8: supersparc, hypersparc
5527 sparclite: f930, f934, sparclite86x
5532 @item -mtune=@var{cpu_type}
5534 Set the instruction scheduling parameters for machine type
5535 @var{cpu_type}, but do not set the instruction set or register set that the
5536 option @option{-mcpu=@var{cpu_type}} would.
5538 The same values for @option{-mcpu=@var{cpu_type}} can be used for
5539 @option{-mtune=@var{cpu_type}}, but the only useful values are those
5540 that select a particular cpu implementation. Those are @samp{cypress},
5541 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
5542 @samp{sparclite86x}, @samp{tsc701}, and @samp{ultrasparc}.
5546 These @samp{-m} switches are supported in addition to the above
5547 on the SPARCLET processor.
5550 @item -mlittle-endian
5551 @opindex mlittle-endian
5552 Generate code for a processor running in little-endian mode.
5556 Treat register @code{%g0} as a normal register.
5557 GCC will continue to clobber it as necessary but will not assume
5558 it always reads as 0.
5560 @item -mbroken-saverestore
5561 @opindex mbroken-saverestore
5562 Generate code that does not use non-trivial forms of the @code{save} and
5563 @code{restore} instructions. Early versions of the SPARCLET processor do
5564 not correctly handle @code{save} and @code{restore} instructions used with
5565 arguments. They correctly handle them used without arguments. A @code{save}
5566 instruction used without arguments increments the current window pointer
5567 but does not allocate a new stack frame. It is assumed that the window
5568 overflow trap handler will properly handle this case as will interrupt
5572 These @samp{-m} switches are supported in addition to the above
5573 on SPARC V9 processors in 64-bit environments.
5576 @item -mlittle-endian
5577 @opindex mlittle-endian
5578 Generate code for a processor running in little-endian mode.
5584 Generate code for a 32-bit or 64-bit environment.
5585 The 32-bit environment sets int, long and pointer to 32 bits.
5586 The 64-bit environment sets int to 32 bits and long and pointer
5589 @item -mcmodel=medlow
5590 @opindex mcmodel=medlow
5591 Generate code for the Medium/Low code model: the program must be linked
5592 in the low 32 bits of the address space. Pointers are 64 bits.
5593 Programs can be statically or dynamically linked.
5595 @item -mcmodel=medmid
5596 @opindex mcmodel=medmid
5597 Generate code for the Medium/Middle code model: the program must be linked
5598 in the low 44 bits of the address space, the text segment must be less than
5599 2G bytes, and data segment must be within 2G of the text segment.
5600 Pointers are 64 bits.
5602 @item -mcmodel=medany
5603 @opindex mcmodel=medany
5604 Generate code for the Medium/Anywhere code model: the program may be linked
5605 anywhere in the address space, the text segment must be less than
5606 2G bytes, and data segment must be within 2G of the text segment.
5607 Pointers are 64 bits.
5609 @item -mcmodel=embmedany
5610 @opindex mcmodel=embmedany
5611 Generate code for the Medium/Anywhere code model for embedded systems:
5612 assume a 32-bit text and a 32-bit data segment, both starting anywhere
5613 (determined at link time). Register %g4 points to the base of the
5614 data segment. Pointers are still 64 bits.
5615 Programs are statically linked, PIC is not supported.
5618 @itemx -mno-stack-bias
5619 @opindex mstack-bias
5620 @opindex mno-stack-bias
5621 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
5622 frame pointer if present, are offset by @minus{}2047 which must be added back
5623 when making stack frame references.
5624 Otherwise, assume no such offset is present.
5627 @node Convex Options
5628 @subsection Convex Options
5629 @cindex Convex options
5631 These @samp{-m} options are defined for Convex:
5636 Generate output for C1. The code will run on any Convex machine.
5637 The preprocessor symbol @code{__convex__c1__} is defined.
5641 Generate output for C2. Uses instructions not available on C1.
5642 Scheduling and other optimizations are chosen for max performance on C2.
5643 The preprocessor symbol @code{__convex_c2__} is defined.
5647 Generate output for C32xx. Uses instructions not available on C1.
5648 Scheduling and other optimizations are chosen for max performance on C32.
5649 The preprocessor symbol @code{__convex_c32__} is defined.
5653 Generate output for C34xx. Uses instructions not available on C1.
5654 Scheduling and other optimizations are chosen for max performance on C34.
5655 The preprocessor symbol @code{__convex_c34__} is defined.
5659 Generate output for C38xx. Uses instructions not available on C1.
5660 Scheduling and other optimizations are chosen for max performance on C38.
5661 The preprocessor symbol @code{__convex_c38__} is defined.
5665 Generate code which puts an argument count in the word preceding each
5666 argument list. This is compatible with regular CC, and a few programs
5667 may need the argument count word. GDB and other source-level debuggers
5668 do not need it; this info is in the symbol table.
5671 @opindex mnoargcount
5672 Omit the argument count word. This is the default.
5674 @item -mvolatile-cache
5675 @opindex mvolatile-cache
5676 Allow volatile references to be cached. This is the default.
5678 @item -mvolatile-nocache
5679 @opindex mvolatile-nocache
5680 Volatile references bypass the data cache, going all the way to memory.
5681 This is only needed for multi-processor code that does not use standard
5682 synchronization instructions. Making non-volatile references to volatile
5683 locations will not necessarily work.
5687 Type long is 32 bits, the same as type int. This is the default.
5691 Type long is 64 bits, the same as type long long. This option is useless,
5692 because no library support exists for it.
5695 @node AMD29K Options
5696 @subsection AMD29K Options
5697 @cindex AMD29K options
5699 These @samp{-m} options are defined for the AMD Am29000:
5704 @cindex DW bit (29k)
5705 Generate code that assumes the @code{DW} bit is set, i.e., that byte and
5706 halfword operations are directly supported by the hardware. This is the
5711 Generate code that assumes the @code{DW} bit is not set.
5715 @cindex byte writes (29k)
5716 Generate code that assumes the system supports byte and halfword write
5717 operations. This is the default.
5721 Generate code that assumes the systems does not support byte and
5722 halfword write operations. @option{-mnbw} implies @option{-mndw}.
5726 @cindex memory model (29k)
5727 Use a small memory model that assumes that all function addresses are
5728 either within a single 256 KB segment or at an absolute address of less
5729 than 256k. This allows the @code{call} instruction to be used instead
5730 of a @code{const}, @code{consth}, @code{calli} sequence.
5734 Use the normal memory model: Generate @code{call} instructions only when
5735 calling functions in the same file and @code{calli} instructions
5736 otherwise. This works if each file occupies less than 256 KB but allows
5737 the entire executable to be larger than 256 KB@. This is the default.
5741 Always use @code{calli} instructions. Specify this option if you expect
5742 a single file to compile into more than 256 KB of code.
5746 @cindex processor selection (29k)
5747 Generate code for the Am29050.
5751 Generate code for the Am29000. This is the default.
5753 @item -mkernel-registers
5754 @opindex mkernel-registers
5755 @cindex kernel and user registers (29k)
5756 Generate references to registers @code{gr64-gr95} instead of to
5757 registers @code{gr96-gr127}. This option can be used when compiling
5758 kernel code that wants a set of global registers disjoint from that used
5761 Note that when this option is used, register names in @samp{-f} flags
5762 must use the normal, user-mode, names.
5764 @item -muser-registers
5765 @opindex muser-registers
5766 Use the normal set of global registers, @code{gr96-gr127}. This is the
5770 @itemx -mno-stack-check
5771 @opindex mstack-check
5772 @opindex mno-stack-check
5773 @cindex stack checks (29k)
5774 Insert (or do not insert) a call to @code{__msp_check} after each stack
5775 adjustment. This is often used for kernel code.
5778 @itemx -mno-storem-bug
5779 @opindex mstorem-bug
5780 @opindex mno-storem-bug
5781 @cindex storem bug (29k)
5782 @option{-mstorem-bug} handles 29k processors which cannot handle the
5783 separation of a mtsrim insn and a storem instruction (most 29000 chips
5784 to date, but not the 29050).
5786 @item -mno-reuse-arg-regs
5787 @itemx -mreuse-arg-regs
5788 @opindex mno-reuse-arg-regs
5789 @opindex mreuse-arg-regs
5790 @option{-mno-reuse-arg-regs} tells the compiler to only use incoming argument
5791 registers for copying out arguments. This helps detect calling a function
5792 with fewer arguments than it was declared with.
5794 @item -mno-impure-text
5795 @itemx -mimpure-text
5796 @opindex mno-impure-text
5797 @opindex mimpure-text
5798 @option{-mimpure-text}, used in addition to @option{-shared}, tells the compiler to
5799 not pass @option{-assert pure-text} to the linker when linking a shared object.
5802 @opindex msoft-float
5803 Generate output containing library calls for floating point.
5804 @strong{Warning:} the requisite libraries are not part of GCC@.
5805 Normally the facilities of the machine's usual C compiler are used, but
5806 this can't be done directly in cross-compilation. You must make your
5807 own arrangements to provide suitable library functions for
5812 Do not generate multm or multmu instructions. This is useful for some embedded
5813 systems which do not have trap handlers for these instructions.
5817 @subsection ARM Options
5820 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
5825 @opindex mapcs-frame
5826 Generate a stack frame that is compliant with the ARM Procedure Call
5827 Standard for all functions, even if this is not strictly necessary for
5828 correct execution of the code. Specifying @option{-fomit-frame-pointer}
5829 with this option will cause the stack frames not to be generated for
5830 leaf functions. The default is @option{-mno-apcs-frame}.
5834 This is a synonym for @option{-mapcs-frame}.
5838 Generate code for a processor running with a 26-bit program counter,
5839 and conforming to the function calling standards for the APCS 26-bit
5840 option. This option replaces the @option{-m2} and @option{-m3} options
5841 of previous releases of the compiler.
5845 Generate code for a processor running with a 32-bit program counter,
5846 and conforming to the function calling standards for the APCS 32-bit
5847 option. This option replaces the @option{-m6} option of previous releases
5851 @c not currently implemented
5852 @item -mapcs-stack-check
5853 @opindex mapcs-stack-check
5854 Generate code to check the amount of stack space available upon entry to
5855 every function (that actually uses some stack space). If there is
5856 insufficient space available then either the function
5857 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
5858 called, depending upon the amount of stack space required. The run time
5859 system is required to provide these functions. The default is
5860 @option{-mno-apcs-stack-check}, since this produces smaller code.
5862 @c not currently implemented
5864 @opindex mapcs-float
5865 Pass floating point arguments using the float point registers. This is
5866 one of the variants of the APCS@. This option is recommended if the
5867 target hardware has a floating point unit or if a lot of floating point
5868 arithmetic is going to be performed by the code. The default is
5869 @option{-mno-apcs-float}, since integer only code is slightly increased in
5870 size if @option{-mapcs-float} is used.
5872 @c not currently implemented
5873 @item -mapcs-reentrant
5874 @opindex mapcs-reentrant
5875 Generate reentrant, position independent code. The default is
5876 @option{-mno-apcs-reentrant}.
5879 @item -mthumb-interwork
5880 @opindex mthumb-interwork
5881 Generate code which supports calling between the ARM and Thumb
5882 instruction sets. Without this option the two instruction sets cannot
5883 be reliably used inside one program. The default is
5884 @option{-mno-thumb-interwork}, since slightly larger code is generated
5885 when @option{-mthumb-interwork} is specified.
5887 @item -mno-sched-prolog
5888 @opindex mno-sched-prolog
5889 Prevent the reordering of instructions in the function prolog, or the
5890 merging of those instruction with the instructions in the function's
5891 body. This means that all functions will start with a recognizable set
5892 of instructions (or in fact one of a choice from a small set of
5893 different function prologues), and this information can be used to
5894 locate the start if functions inside an executable piece of code. The
5895 default is @option{-msched-prolog}.
5898 @opindex mhard-float
5899 Generate output containing floating point instructions. This is the
5903 @opindex msoft-float
5904 Generate output containing library calls for floating point.
5905 @strong{Warning:} the requisite libraries are not available for all ARM
5906 targets. Normally the facilities of the machine's usual C compiler are
5907 used, but this cannot be done directly in cross-compilation. You must make
5908 your own arrangements to provide suitable library functions for
5911 @option{-msoft-float} changes the calling convention in the output file;
5912 therefore, it is only useful if you compile @emph{all} of a program with
5913 this option. In particular, you need to compile @file{libgcc.a}, the
5914 library that comes with GCC, with @option{-msoft-float} in order for
5917 @item -mlittle-endian
5918 @opindex mlittle-endian
5919 Generate code for a processor running in little-endian mode. This is
5920 the default for all standard configurations.
5923 @opindex mbig-endian
5924 Generate code for a processor running in big-endian mode; the default is
5925 to compile code for a little-endian processor.
5927 @item -mwords-little-endian
5928 @opindex mwords-little-endian
5929 This option only applies when generating code for big-endian processors.
5930 Generate code for a little-endian word order but a big-endian byte
5931 order. That is, a byte order of the form @samp{32107654}. Note: this
5932 option should only be used if you require compatibility with code for
5933 big-endian ARM processors generated by versions of the compiler prior to
5936 @item -malignment-traps
5937 @opindex malignment-traps
5938 Generate code that will not trap if the MMU has alignment traps enabled.
5939 On ARM architectures prior to ARMv4, there were no instructions to
5940 access half-word objects stored in memory. However, when reading from
5941 memory a feature of the ARM architecture allows a word load to be used,
5942 even if the address is unaligned, and the processor core will rotate the
5943 data as it is being loaded. This option tells the compiler that such
5944 misaligned accesses will cause a MMU trap and that it should instead
5945 synthesise the access as a series of byte accesses. The compiler can
5946 still use word accesses to load half-word data if it knows that the
5947 address is aligned to a word boundary.
5949 This option is ignored when compiling for ARM architecture 4 or later,
5950 since these processors have instructions to directly access half-word
5953 @item -mno-alignment-traps
5954 @opindex mno-alignment-traps
5955 Generate code that assumes that the MMU will not trap unaligned
5956 accesses. This produces better code when the target instruction set
5957 does not have half-word memory operations (i.e.@: implementations prior to
5960 Note that you cannot use this option to access unaligned word objects,
5961 since the processor will only fetch one 32-bit aligned object from
5964 The default setting for most targets is @option{-mno-alignment-traps}, since
5965 this produces better code when there are no half-word memory
5966 instructions available.
5968 @item -mshort-load-bytes
5969 @itemx -mno-short-load-words
5970 @opindex mshort-load-bytes
5971 @opindex mno-short-load-words
5972 These are deprecated aliases for @option{-malignment-traps}.
5974 @item -mno-short-load-bytes
5975 @itemx -mshort-load-words
5976 @opindex mno-short-load-bytes
5977 @opindex mshort-load-words
5978 This are deprecated aliases for @option{-mno-alignment-traps}.
5982 This option only applies to RISC iX@. Emulate the native BSD-mode
5983 compiler. This is the default if @option{-ansi} is not specified.
5987 This option only applies to RISC iX@. Emulate the native X/Open-mode
5990 @item -mno-symrename
5991 @opindex mno-symrename
5992 This option only applies to RISC iX@. Do not run the assembler
5993 post-processor, @samp{symrename}, after code has been assembled.
5994 Normally it is necessary to modify some of the standard symbols in
5995 preparation for linking with the RISC iX C library; this option
5996 suppresses this pass. The post-processor is never run when the
5997 compiler is built for cross-compilation.
5999 @item -mcpu=@var{name}
6001 This specifies the name of the target ARM processor. GCC uses this name
6002 to determine what kind of instructions it can emit when generating
6003 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
6004 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
6005 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
6006 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
6007 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
6008 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm8},
6009 @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
6010 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
6011 @samp{arm920t}, @samp{arm940t}, @samp{arm9tdmi}, @samp{arm10tdmi},
6012 @samp{arm1020t}, @samp{xscale}.
6014 @itemx -mtune=@var{name}
6016 This option is very similar to the @option{-mcpu=} option, except that
6017 instead of specifying the actual target processor type, and hence
6018 restricting which instructions can be used, it specifies that GCC should
6019 tune the performance of the code as if the target were of the type
6020 specified in this option, but still choosing the instructions that it
6021 will generate based on the cpu specified by a @option{-mcpu=} option.
6022 For some ARM implementations better performance can be obtained by using
6025 @item -march=@var{name}
6027 This specifies the name of the target ARM architecture. GCC uses this
6028 name to determine what kind of instructions it can emit when generating
6029 assembly code. This option can be used in conjunction with or instead
6030 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
6031 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
6032 @samp{armv5}, @samp{armv5t}, @samp{armv5te}.
6034 @item -mfpe=@var{number}
6035 @itemx -mfp=@var{number}
6038 This specifies the version of the floating point emulation available on
6039 the target. Permissible values are 2 and 3. @option{-mfp=} is a synonym
6040 for @option{-mfpe=}, for compatibility with older versions of GCC@.
6042 @item -mstructure-size-boundary=@var{n}
6043 @opindex mstructure-size-boundary
6044 The size of all structures and unions will be rounded up to a multiple
6045 of the number of bits set by this option. Permissible values are 8 and
6046 32. The default value varies for different toolchains. For the COFF
6047 targeted toolchain the default value is 8. Specifying the larger number
6048 can produce faster, more efficient code, but can also increase the size
6049 of the program. The two values are potentially incompatible. Code
6050 compiled with one value cannot necessarily expect to work with code or
6051 libraries compiled with the other value, if they exchange information
6052 using structures or unions.
6054 @item -mabort-on-noreturn
6055 @opindex mabort-on-noreturn
6056 Generate a call to the function @code{abort} at the end of a
6057 @code{noreturn} function. It will be executed if the function tries to
6061 @itemx -mno-long-calls
6062 @opindex mlong-calls
6063 @opindex mno-long-calls
6064 Tells the compiler to perform function calls by first loading the
6065 address of the function into a register and then performing a subroutine
6066 call on this register. This switch is needed if the target function
6067 will lie outside of the 64 megabyte addressing range of the offset based
6068 version of subroutine call instruction.
6070 Even if this switch is enabled, not all function calls will be turned
6071 into long calls. The heuristic is that static functions, functions
6072 which have the @samp{short-call} attribute, functions that are inside
6073 the scope of a @samp{#pragma no_long_calls} directive and functions whose
6074 definitions have already been compiled within the current compilation
6075 unit, will not be turned into long calls. The exception to this rule is
6076 that weak function definitions, functions with the @samp{long-call}
6077 attribute or the @samp{section} attribute, and functions that are within
6078 the scope of a @samp{#pragma long_calls} directive, will always be
6079 turned into long calls.
6081 This feature is not enabled by default. Specifying
6082 @option{-mno-long-calls} will restore the default behaviour, as will
6083 placing the function calls within the scope of a @samp{#pragma
6084 long_calls_off} directive. Note these switches have no effect on how
6085 the compiler generates code to handle function calls via function
6088 @item -mnop-fun-dllimport
6089 @opindex mnop-fun-dllimport
6090 Disable support for the @code{dllimport} attribute.
6092 @item -msingle-pic-base
6093 @opindex msingle-pic-base
6094 Treat the register used for PIC addressing as read-only, rather than
6095 loading it in the prologue for each function. The run-time system is
6096 responsible for initialising this register with an appropriate value
6097 before execution begins.
6099 @item -mpic-register=@var{reg}
6100 @opindex mpic-register
6101 Specify the register to be used for PIC addressing. The default is R10
6102 unless stack-checking is enabled, when R9 is used.
6104 @item -mpoke-function-name
6105 @opindex mpoke-function-name
6106 Write the name of each function into the text section, directly
6107 preceding the function prologue. The generated code is similar to this:
6111 .ascii "arm_poke_function_name", 0
6114 .word 0xff000000 + (t1 - t0)
6115 arm_poke_function_name
6117 stmfd sp!, @{fp, ip, lr, pc@}
6121 When performing a stack backtrace, code can inspect the value of
6122 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
6123 location @code{pc - 12} and the top 8 bits are set, then we know that
6124 there is a function name embedded immediately preceding this location
6125 and has length @code{((pc[-3]) & 0xff000000)}.
6129 Generate code for the 16-bit Thumb instruction set. The default is to
6130 use the 32-bit ARM instruction set.
6133 @opindex mtpcs-frame
6134 Generate a stack frame that is compliant with the Thumb Procedure Call
6135 Standard for all non-leaf functions. (A leaf function is one that does
6136 not call any other functions.) The default is @option{-mno-tpcs-frame}.
6138 @item -mtpcs-leaf-frame
6139 @opindex mtpcs-leaf-frame
6140 Generate a stack frame that is compliant with the Thumb Procedure Call
6141 Standard for all leaf functions. (A leaf function is one that does
6142 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
6144 @item -mcallee-super-interworking
6145 @opindex mcallee-super-interworking
6146 Gives all externally visible functions in the file being compiled an ARM
6147 instruction set header which switches to Thumb mode before executing the
6148 rest of the function. This allows these functions to be called from
6149 non-interworking code.
6151 @item -mcaller-super-interworking
6152 @opindex mcaller-super-interworking
6153 Allows calls via function pointers (including virtual functions) to
6154 execute correctly regardless of whether the target code has been
6155 compiled for interworking or not. There is a small overhead in the cost
6156 of executing a function pointer if this option is enabled.
6160 @node MN10200 Options
6161 @subsection MN10200 Options
6162 @cindex MN10200 options
6163 These @option{-m} options are defined for Matsushita MN10200 architectures:
6168 Indicate to the linker that it should perform a relaxation optimization pass
6169 to shorten branches, calls and absolute memory addresses. This option only
6170 has an effect when used on the command line for the final link step.
6172 This option makes symbolic debugging impossible.
6175 @node MN10300 Options
6176 @subsection MN10300 Options
6177 @cindex MN10300 options
6178 These @option{-m} options are defined for Matsushita MN10300 architectures:
6183 Generate code to avoid bugs in the multiply instructions for the MN10300
6184 processors. This is the default.
6187 @opindex mno-mult-bug
6188 Do not generate code to avoid bugs in the multiply instructions for the
6193 Generate code which uses features specific to the AM33 processor.
6197 Do not generate code which uses features specific to the AM33 processor. This
6202 Do not link in the C run-time initialization object file.
6206 Indicate to the linker that it should perform a relaxation optimization pass
6207 to shorten branches, calls and absolute memory addresses. This option only
6208 has an effect when used on the command line for the final link step.
6210 This option makes symbolic debugging impossible.
6214 @node M32R/D Options
6215 @subsection M32R/D Options
6216 @cindex M32R/D options
6218 These @option{-m} options are defined for Mitsubishi M32R/D architectures:
6221 @item -mcode-model=small
6222 @opindex mcode-model=small
6223 Assume all objects live in the lower 16MB of memory (so that their addresses
6224 can be loaded with the @code{ld24} instruction), and assume all subroutines
6225 are reachable with the @code{bl} instruction.
6226 This is the default.
6228 The addressability of a particular object can be set with the
6229 @code{model} attribute.
6231 @item -mcode-model=medium
6232 @opindex mcode-model=medium
6233 Assume objects may be anywhere in the 32-bit address space (the compiler
6234 will generate @code{seth/add3} instructions to load their addresses), and
6235 assume all subroutines are reachable with the @code{bl} instruction.
6237 @item -mcode-model=large
6238 @opindex mcode-model=large
6239 Assume objects may be anywhere in the 32-bit address space (the compiler
6240 will generate @code{seth/add3} instructions to load their addresses), and
6241 assume subroutines may not be reachable with the @code{bl} instruction
6242 (the compiler will generate the much slower @code{seth/add3/jl}
6243 instruction sequence).
6246 @opindex msdata=none
6247 Disable use of the small data area. Variables will be put into
6248 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
6249 @code{section} attribute has been specified).
6250 This is the default.
6252 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
6253 Objects may be explicitly put in the small data area with the
6254 @code{section} attribute using one of these sections.
6257 @opindex msdata=sdata
6258 Put small global and static data in the small data area, but do not
6259 generate special code to reference them.
6263 Put small global and static data in the small data area, and generate
6264 special instructions to reference them.
6268 @cindex smaller data references
6269 Put global and static objects less than or equal to @var{num} bytes
6270 into the small data or bss sections instead of the normal data or bss
6271 sections. The default value of @var{num} is 8.
6272 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
6273 for this option to have any effect.
6275 All modules should be compiled with the same @option{-G @var{num}} value.
6276 Compiling with different values of @var{num} may or may not work; if it
6277 doesn't the linker will give an error message---incorrect code will not be
6283 @subsection M88K Options
6284 @cindex M88k options
6286 These @samp{-m} options are defined for Motorola 88k architectures:
6291 Generate code that works well on both the m88100 and the
6296 Generate code that works best for the m88100, but that also
6301 Generate code that works best for the m88110, and may not run
6306 Obsolete option to be removed from the next revision.
6309 @item -midentify-revision
6310 @opindex midentify-revision
6311 @cindex identifying source, compiler (88k)
6312 Include an @code{ident} directive in the assembler output recording the
6313 source file name, compiler name and version, timestamp, and compilation
6316 @item -mno-underscores
6317 @opindex mno-underscores
6318 @cindex underscores, avoiding (88k)
6319 In assembler output, emit symbol names without adding an underscore
6320 character at the beginning of each name. The default is to use an
6321 underscore as prefix on each name.
6323 @item -mocs-debug-info
6324 @itemx -mno-ocs-debug-info
6325 @opindex mocs-debug-info
6326 @opindex mno-ocs-debug-info
6328 @cindex debugging, 88k OCS
6329 Include (or omit) additional debugging information (about registers used
6330 in each stack frame) as specified in the 88open Object Compatibility
6331 Standard, ``OCS''@. This extra information allows debugging of code that
6332 has had the frame pointer eliminated. The default for DG/UX, SVr4, and
6333 Delta 88 SVr3.2 is to include this information; other 88k configurations
6334 omit this information by default.
6336 @item -mocs-frame-position
6337 @opindex mocs-frame-position
6338 @cindex register positions in frame (88k)
6339 When emitting COFF debugging information for automatic variables and
6340 parameters stored on the stack, use the offset from the canonical frame
6341 address, which is the stack pointer (register 31) on entry to the
6342 function. The DG/UX, SVr4, Delta88 SVr3.2, and BCS configurations use
6343 @option{-mocs-frame-position}; other 88k configurations have the default
6344 @option{-mno-ocs-frame-position}.
6346 @item -mno-ocs-frame-position
6347 @opindex mno-ocs-frame-position
6348 @cindex register positions in frame (88k)
6349 When emitting COFF debugging information for automatic variables and
6350 parameters stored on the stack, use the offset from the frame pointer
6351 register (register 30). When this option is in effect, the frame
6352 pointer is not eliminated when debugging information is selected by the
6355 @item -moptimize-arg-area
6356 @opindex moptimize-arg-area
6357 @cindex arguments in frame (88k)
6358 Save space by reorganizing the stack frame. This option generates code
6359 that does not agree with the 88open specifications, but uses less
6362 @itemx -mno-optimize-arg-area
6363 @opindex mno-optimize-arg-area
6364 Do not reorganize the stack frame to save space. This is the default.
6365 The generated conforms to the specification, but uses more memory.
6367 @item -mshort-data-@var{num}
6368 @opindex mshort-data
6369 @cindex smaller data references (88k)
6370 @cindex r0-relative references (88k)
6371 Generate smaller data references by making them relative to @code{r0},
6372 which allows loading a value using a single instruction (rather than the
6373 usual two). You control which data references are affected by
6374 specifying @var{num} with this option. For example, if you specify
6375 @option{-mshort-data-512}, then the data references affected are those
6376 involving displacements of less than 512 bytes.
6377 @option{-mshort-data-@var{num}} is not effective for @var{num} greater
6380 @item -mserialize-volatile
6381 @opindex mserialize-volatile
6382 @itemx -mno-serialize-volatile
6383 @opindex mno-serialize-volatile
6384 @cindex sequential consistency on 88k
6385 Do, or don't, generate code to guarantee sequential consistency
6386 of volatile memory references. By default, consistency is
6389 The order of memory references made by the MC88110 processor does
6390 not always match the order of the instructions requesting those
6391 references. In particular, a load instruction may execute before
6392 a preceding store instruction. Such reordering violates
6393 sequential consistency of volatile memory references, when there
6394 are multiple processors. When consistency must be guaranteed,
6395 GCC generates special instructions, as needed, to force
6396 execution in the proper order.
6398 The MC88100 processor does not reorder memory references and so
6399 always provides sequential consistency. However, by default, GCC
6400 generates the special instructions to guarantee consistency
6401 even when you use @option{-m88100}, so that the code may be run on an
6402 MC88110 processor. If you intend to run your code only on the
6403 MC88100 processor, you may use @option{-mno-serialize-volatile}.
6405 The extra code generated to guarantee consistency may affect the
6406 performance of your application. If you know that you can safely
6407 forgo this guarantee, you may use @option{-mno-serialize-volatile}.
6413 @cindex assembler syntax, 88k
6415 Turn on (@option{-msvr4}) or off (@option{-msvr3}) compiler extensions
6416 related to System V release 4 (SVr4). This controls the following:
6420 Which variant of the assembler syntax to emit.
6422 @option{-msvr4} makes the C preprocessor recognize @samp{#pragma weak}
6423 that is used on System V release 4.
6425 @option{-msvr4} makes GCC issue additional declaration directives used in
6429 @option{-msvr4} is the default for the m88k-motorola-sysv4 and
6430 m88k-dg-dgux m88k configurations. @option{-msvr3} is the default for all
6431 other m88k configurations.
6433 @item -mversion-03.00
6434 @opindex mversion-03.00
6435 This option is obsolete, and is ignored.
6436 @c ??? which asm syntax better for GAS? option there too?
6438 @item -mno-check-zero-division
6439 @itemx -mcheck-zero-division
6440 @opindex mno-check-zero-division
6441 @opindex mcheck-zero-division
6442 @cindex zero division on 88k
6443 Do, or don't, generate code to guarantee that integer division by
6444 zero will be detected. By default, detection is guaranteed.
6446 Some models of the MC88100 processor fail to trap upon integer
6447 division by zero under certain conditions. By default, when
6448 compiling code that might be run on such a processor, GCC
6449 generates code that explicitly checks for zero-valued divisors
6450 and traps with exception number 503 when one is detected. Use of
6451 @option{-mno-check-zero-division} suppresses such checking for code
6452 generated to run on an MC88100 processor.
6454 GCC assumes that the MC88110 processor correctly detects all instances
6455 of integer division by zero. When @option{-m88110} is specified, no
6456 explicit checks for zero-valued divisors are generated, and both
6457 @option{-mcheck-zero-division} and @option{-mno-check-zero-division} are
6460 @item -muse-div-instruction
6461 @opindex muse-div-instruction
6462 @cindex divide instruction, 88k
6463 Use the div instruction for signed integer division on the
6464 MC88100 processor. By default, the div instruction is not used.
6466 On the MC88100 processor the signed integer division instruction
6467 div) traps to the operating system on a negative operand. The
6468 operating system transparently completes the operation, but at a
6469 large cost in execution time. By default, when compiling code
6470 that might be run on an MC88100 processor, GCC emulates signed
6471 integer division using the unsigned integer division instruction
6472 divu), thereby avoiding the large penalty of a trap to the
6473 operating system. Such emulation has its own, smaller, execution
6474 cost in both time and space. To the extent that your code's
6475 important signed integer division operations are performed on two
6476 nonnegative operands, it may be desirable to use the div
6477 instruction directly.
6479 On the MC88110 processor the div instruction (also known as the
6480 divs instruction) processes negative operands without trapping to
6481 the operating system. When @option{-m88110} is specified,
6482 @option{-muse-div-instruction} is ignored, and the div instruction is used
6483 for signed integer division.
6485 Note that the result of dividing @code{INT_MIN} by @minus{}1 is undefined. In
6486 particular, the behavior of such a division with and without
6487 @option{-muse-div-instruction} may differ.
6489 @item -mtrap-large-shift
6490 @itemx -mhandle-large-shift
6491 @opindex mtrap-large-shift
6492 @opindex mhandle-large-shift
6493 @cindex bit shift overflow (88k)
6494 @cindex large bit shifts (88k)
6495 Include code to detect bit-shifts of more than 31 bits; respectively,
6496 trap such shifts or emit code to handle them properly. By default GCC
6497 makes no special provision for large bit shifts.
6499 @item -mwarn-passed-structs
6500 @opindex mwarn-passed-structs
6501 @cindex structure passing (88k)
6502 Warn when a function passes a struct as an argument or result.
6503 Structure-passing conventions have changed during the evolution of the C
6504 language, and are often the source of portability problems. By default,
6505 GCC issues no such warning.
6508 @c break page here to avoid unsightly interparagraph stretch.
6512 @node RS/6000 and PowerPC Options
6513 @subsection IBM RS/6000 and PowerPC Options
6514 @cindex RS/6000 and PowerPC Options
6515 @cindex IBM RS/6000 and PowerPC Options
6517 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
6525 @itemx -mpowerpc-gpopt
6526 @itemx -mno-powerpc-gpopt
6527 @itemx -mpowerpc-gfxopt
6528 @itemx -mno-powerpc-gfxopt
6530 @itemx -mno-powerpc64
6536 @opindex mno-powerpc
6537 @opindex mpowerpc-gpopt
6538 @opindex mno-powerpc-gpopt
6539 @opindex mpowerpc-gfxopt
6540 @opindex mno-powerpc-gfxopt
6542 @opindex mno-powerpc64
6543 GCC supports two related instruction set architectures for the
6544 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
6545 instructions supported by the @samp{rios} chip set used in the original
6546 RS/6000 systems and the @dfn{PowerPC} instruction set is the
6547 architecture of the Motorola MPC5xx, MPC6xx, MPC8xx microprocessors, and
6548 the IBM 4xx microprocessors.
6550 Neither architecture is a subset of the other. However there is a
6551 large common subset of instructions supported by both. An MQ
6552 register is included in processors supporting the POWER architecture.
6554 You use these options to specify which instructions are available on the
6555 processor you are using. The default value of these options is
6556 determined when configuring GCC@. Specifying the
6557 @option{-mcpu=@var{cpu_type}} overrides the specification of these
6558 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
6559 rather than the options listed above.
6561 The @option{-mpower} option allows GCC to generate instructions that
6562 are found only in the POWER architecture and to use the MQ register.
6563 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
6564 to generate instructions that are present in the POWER2 architecture but
6565 not the original POWER architecture.
6567 The @option{-mpowerpc} option allows GCC to generate instructions that
6568 are found only in the 32-bit subset of the PowerPC architecture.
6569 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
6570 GCC to use the optional PowerPC architecture instructions in the
6571 General Purpose group, including floating-point square root. Specifying
6572 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
6573 use the optional PowerPC architecture instructions in the Graphics
6574 group, including floating-point select.
6576 The @option{-mpowerpc64} option allows GCC to generate the additional
6577 64-bit instructions that are found in the full PowerPC64 architecture
6578 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
6579 @option{-mno-powerpc64}.
6581 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
6582 will use only the instructions in the common subset of both
6583 architectures plus some special AIX common-mode calls, and will not use
6584 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
6585 permits GCC to use any instruction from either architecture and to
6586 allow use of the MQ register; specify this for the Motorola MPC601.
6588 @item -mnew-mnemonics
6589 @itemx -mold-mnemonics
6590 @opindex mnew-mnemonics
6591 @opindex mold-mnemonics
6592 Select which mnemonics to use in the generated assembler code. With
6593 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
6594 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
6595 assembler mnemonics defined for the POWER architecture. Instructions
6596 defined in only one architecture have only one mnemonic; GCC uses that
6597 mnemonic irrespective of which of these options is specified.
6599 GCC defaults to the mnemonics appropriate for the architecture in
6600 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
6601 value of these option. Unless you are building a cross-compiler, you
6602 should normally not specify either @option{-mnew-mnemonics} or
6603 @option{-mold-mnemonics}, but should instead accept the default.
6605 @item -mcpu=@var{cpu_type}
6607 Set architecture type, register usage, choice of mnemonics, and
6608 instruction scheduling parameters for machine type @var{cpu_type}.
6609 Supported values for @var{cpu_type} are @samp{rios}, @samp{rios1},
6610 @samp{rsc}, @samp{rios2}, @samp{rs64a}, @samp{601}, @samp{602},
6611 @samp{603}, @samp{603e}, @samp{604}, @samp{604e}, @samp{620},
6612 @samp{630}, @samp{740}, @samp{7400}, @samp{7450}, @samp{750},
6613 @samp{power}, @samp{power2}, @samp{powerpc}, @samp{403}, @samp{505},
6614 @samp{801}, @samp{821}, @samp{823}, and @samp{860} and @samp{common}.
6616 @option{-mcpu=common} selects a completely generic processor. Code
6617 generated under this option will run on any POWER or PowerPC processor.
6618 GCC will use only the instructions in the common subset of both
6619 architectures, and will not use the MQ register. GCC assumes a generic
6620 processor model for scheduling purposes.
6622 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
6623 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
6624 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
6625 types, with an appropriate, generic processor model assumed for
6626 scheduling purposes.
6628 The other options specify a specific processor. Code generated under
6629 those options will run best on that processor, and may not run at all on
6632 The @option{-mcpu} options automatically enable or disable other
6633 @option{-m} options as follows:
6637 @option{-mno-power}, @option{-mno-powerc}
6644 @option{-mpower}, @option{-mno-powerpc}, @option{-mno-new-mnemonics}
6659 @option{-mno-power}, @option{-mpowerpc}, @option{-mnew-mnemonics}
6662 @option{-mpower}, @option{-mpowerpc}, @option{-mnew-mnemonics}
6667 @option{-mno-power}, @option{-mpowerpc}, @option{-mnew-mnemonics}, @option{-msoft-float}
6670 @item -mtune=@var{cpu_type}
6672 Set the instruction scheduling parameters for machine type
6673 @var{cpu_type}, but do not set the architecture type, register usage, or
6674 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
6675 values for @var{cpu_type} are used for @option{-mtune} as for
6676 @option{-mcpu}. If both are specified, the code generated will use the
6677 architecture, registers, and mnemonics set by @option{-mcpu}, but the
6678 scheduling parameters set by @option{-mtune}.
6681 @itemx -mno-fp-in-toc
6682 @itemx -mno-sum-in-toc
6683 @itemx -mminimal-toc
6685 @opindex mno-fp-in-toc
6686 @opindex mno-sum-in-toc
6687 @opindex mminimal-toc
6688 Modify generation of the TOC (Table Of Contents), which is created for
6689 every executable file. The @option{-mfull-toc} option is selected by
6690 default. In that case, GCC will allocate at least one TOC entry for
6691 each unique non-automatic variable reference in your program. GCC
6692 will also place floating-point constants in the TOC@. However, only
6693 16,384 entries are available in the TOC@.
6695 If you receive a linker error message that saying you have overflowed
6696 the available TOC space, you can reduce the amount of TOC space used
6697 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
6698 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
6699 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
6700 generate code to calculate the sum of an address and a constant at
6701 run-time instead of putting that sum into the TOC@. You may specify one
6702 or both of these options. Each causes GCC to produce very slightly
6703 slower and larger code at the expense of conserving TOC space.
6705 If you still run out of space in the TOC even when you specify both of
6706 these options, specify @option{-mminimal-toc} instead. This option causes
6707 GCC to make only one TOC entry for every file. When you specify this
6708 option, GCC will produce code that is slower and larger but which
6709 uses extremely little TOC space. You may wish to use this option
6710 only on files that contain less frequently executed code.
6716 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
6717 @code{long} type, and the infrastructure needed to support them.
6718 Specifying @option{-maix64} implies @option{-mpowerpc64} and
6719 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
6720 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
6725 @opindex mno-xl-call
6726 On AIX, pass floating-point arguments to prototyped functions beyond the
6727 register save area (RSA) on the stack in addition to argument FPRs. The
6728 AIX calling convention was extended but not initially documented to
6729 handle an obscure K&R C case of calling a function that takes the
6730 address of its arguments with fewer arguments than declared. AIX XL
6731 compilers access floating point arguments which do not fit in the
6732 RSA from the stack when a subroutine is compiled without
6733 optimization. Because always storing floating-point arguments on the
6734 stack is inefficient and rarely needed, this option is not enabled by
6735 default and only is necessary when calling subroutines compiled by AIX
6736 XL compilers without optimization.
6740 Support @dfn{AIX Threads}. Link an application written to use
6741 @dfn{pthreads} with special libraries and startup code to enable the
6746 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
6747 application written to use message passing with special startup code to
6748 enable the application to run. The system must have PE installed in the
6749 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
6750 must be overridden with the @option{-specs=} option to specify the
6751 appropriate directory location. The Parallel Environment does not
6752 support threads, so the @option{-mpe} option and the @option{-mthreads}
6753 option are incompatible.
6757 @opindex msoft-float
6758 @opindex mhard-float
6759 Generate code that does not use (uses) the floating-point register set.
6760 Software floating point emulation is provided if you use the
6761 @option{-msoft-float} option, and pass the option to GCC when linking.
6764 @itemx -mno-multiple
6766 @opindex mno-multiple
6767 Generate code that uses (does not use) the load multiple word
6768 instructions and the store multiple word instructions. These
6769 instructions are generated by default on POWER systems, and not
6770 generated on PowerPC systems. Do not use @option{-mmultiple} on little
6771 endian PowerPC systems, since those instructions do not work when the
6772 processor is in little endian mode. The exceptions are PPC740 and
6773 PPC750 which permit the instructions usage in little endian mode.
6779 Generate code that uses (does not use) the load string instructions
6780 and the store string word instructions to save multiple registers and
6781 do small block moves. These instructions are generated by default on
6782 POWER systems, and not generated on PowerPC systems. Do not use
6783 @option{-mstring} on little endian PowerPC systems, since those
6784 instructions do not work when the processor is in little endian mode.
6785 The exceptions are PPC740 and PPC750 which permit the instructions
6786 usage in little endian mode.
6792 Generate code that uses (does not use) the load or store instructions
6793 that update the base register to the address of the calculated memory
6794 location. These instructions are generated by default. If you use
6795 @option{-mno-update}, there is a small window between the time that the
6796 stack pointer is updated and the address of the previous frame is
6797 stored, which means code that walks the stack frame across interrupts or
6798 signals may get corrupted data.
6801 @itemx -mno-fused-madd
6802 @opindex mfused-madd
6803 @opindex mno-fused-madd
6804 Generate code that uses (does not use) the floating point multiply and
6805 accumulate instructions. These instructions are generated by default if
6806 hardware floating is used.
6808 @item -mno-bit-align
6810 @opindex mno-bit-align
6812 On System V.4 and embedded PowerPC systems do not (do) force structures
6813 and unions that contain bit-fields to be aligned to the base type of the
6816 For example, by default a structure containing nothing but 8
6817 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
6818 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
6819 the structure would be aligned to a 1 byte boundary and be one byte in
6822 @item -mno-strict-align
6823 @itemx -mstrict-align
6824 @opindex mno-strict-align
6825 @opindex mstrict-align
6826 On System V.4 and embedded PowerPC systems do not (do) assume that
6827 unaligned memory references will be handled by the system.
6830 @itemx -mno-relocatable
6831 @opindex mrelocatable
6832 @opindex mno-relocatable
6833 On embedded PowerPC systems generate code that allows (does not allow)
6834 the program to be relocated to a different address at runtime. If you
6835 use @option{-mrelocatable} on any module, all objects linked together must
6836 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
6838 @item -mrelocatable-lib
6839 @itemx -mno-relocatable-lib
6840 @opindex mrelocatable-lib
6841 @opindex mno-relocatable-lib
6842 On embedded PowerPC systems generate code that allows (does not allow)
6843 the program to be relocated to a different address at runtime. Modules
6844 compiled with @option{-mrelocatable-lib} can be linked with either modules
6845 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
6846 with modules compiled with the @option{-mrelocatable} options.
6852 On System V.4 and embedded PowerPC systems do not (do) assume that
6853 register 2 contains a pointer to a global area pointing to the addresses
6854 used in the program.
6857 @itemx -mlittle-endian
6859 @opindex mlittle-endian
6860 On System V.4 and embedded PowerPC systems compile code for the
6861 processor in little endian mode. The @option{-mlittle-endian} option is
6862 the same as @option{-mlittle}.
6867 @opindex mbig-endian
6868 On System V.4 and embedded PowerPC systems compile code for the
6869 processor in big endian mode. The @option{-mbig-endian} option is
6870 the same as @option{-mbig}.
6874 On System V.4 and embedded PowerPC systems compile code using calling
6875 conventions that adheres to the March 1995 draft of the System V
6876 Application Binary Interface, PowerPC processor supplement. This is the
6877 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
6879 @item -mcall-sysv-eabi
6880 @opindex mcall-sysv-eabi
6881 Specify both @option{-mcall-sysv} and @option{-meabi} options.
6883 @item -mcall-sysv-noeabi
6884 @opindex mcall-sysv-noeabi
6885 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
6889 On System V.4 and embedded PowerPC systems compile code using calling
6890 conventions that are similar to those used on AIX@. This is the
6891 default if you configured GCC using @samp{powerpc-*-eabiaix}.
6893 @item -mcall-solaris
6894 @opindex mcall-solaris
6895 On System V.4 and embedded PowerPC systems compile code for the Solaris
6899 @opindex mcall-linux
6900 On System V.4 and embedded PowerPC systems compile code for the
6901 Linux-based GNU system.
6904 @opindex mcall-netbsd
6905 On System V.4 and embedded PowerPC systems compile code for the
6906 NetBSD operating system.
6909 @itemx -mno-prototype
6911 @opindex mno-prototype
6912 On System V.4 and embedded PowerPC systems assume that all calls to
6913 variable argument functions are properly prototyped. Otherwise, the
6914 compiler must insert an instruction before every non prototyped call to
6915 set or clear bit 6 of the condition code register (@var{CR}) to
6916 indicate whether floating point values were passed in the floating point
6917 registers in case the function takes a variable arguments. With
6918 @option{-mprototype}, only calls to prototyped variable argument functions
6919 will set or clear the bit.
6923 On embedded PowerPC systems, assume that the startup module is called
6924 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
6925 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
6930 On embedded PowerPC systems, assume that the startup module is called
6931 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
6936 On embedded PowerPC systems, assume that the startup module is called
6937 @file{crt0.o} and the standard C libraries are @file{libads.a} and
6941 @opindex myellowknife
6942 On embedded PowerPC systems, assume that the startup module is called
6943 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
6948 On System V.4 and embedded PowerPC systems, specify that you are
6949 compiling for a VxWorks system.
6953 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
6954 header to indicate that @samp{eabi} extended relocations are used.
6960 On System V.4 and embedded PowerPC systems do (do not) adhere to the
6961 Embedded Applications Binary Interface (eabi) which is a set of
6962 modifications to the System V.4 specifications. Selecting @option{-meabi}
6963 means that the stack is aligned to an 8 byte boundary, a function
6964 @code{__eabi} is called to from @code{main} to set up the eabi
6965 environment, and the @option{-msdata} option can use both @code{r2} and
6966 @code{r13} to point to two separate small data areas. Selecting
6967 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
6968 do not call an initialization function from @code{main}, and the
6969 @option{-msdata} option will only use @code{r13} to point to a single
6970 small data area. The @option{-meabi} option is on by default if you
6971 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
6974 @opindex msdata=eabi
6975 On System V.4 and embedded PowerPC systems, put small initialized
6976 @code{const} global and static data in the @samp{.sdata2} section, which
6977 is pointed to by register @code{r2}. Put small initialized
6978 non-@code{const} global and static data in the @samp{.sdata} section,
6979 which is pointed to by register @code{r13}. Put small uninitialized
6980 global and static data in the @samp{.sbss} section, which is adjacent to
6981 the @samp{.sdata} section. The @option{-msdata=eabi} option is
6982 incompatible with the @option{-mrelocatable} option. The
6983 @option{-msdata=eabi} option also sets the @option{-memb} option.
6986 @opindex msdata=sysv
6987 On System V.4 and embedded PowerPC systems, put small global and static
6988 data in the @samp{.sdata} section, which is pointed to by register
6989 @code{r13}. Put small uninitialized global and static data in the
6990 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
6991 The @option{-msdata=sysv} option is incompatible with the
6992 @option{-mrelocatable} option.
6994 @item -msdata=default
6996 @opindex msdata=default
6998 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
6999 compile code the same as @option{-msdata=eabi}, otherwise compile code the
7000 same as @option{-msdata=sysv}.
7003 @opindex msdata-data
7004 On System V.4 and embedded PowerPC systems, put small global and static
7005 data in the @samp{.sdata} section. Put small uninitialized global and
7006 static data in the @samp{.sbss} section. Do not use register @code{r13}
7007 to address small data however. This is the default behavior unless
7008 other @option{-msdata} options are used.
7012 @opindex msdata=none
7014 On embedded PowerPC systems, put all initialized global and static data
7015 in the @samp{.data} section, and all uninitialized data in the
7016 @samp{.bss} section.
7020 @cindex smaller data references (PowerPC)
7021 @cindex .sdata/.sdata2 references (PowerPC)
7022 On embedded PowerPC systems, put global and static items less than or
7023 equal to @var{num} bytes into the small data or bss sections instead of
7024 the normal data or bss section. By default, @var{num} is 8. The
7025 @option{-G @var{num}} switch is also passed to the linker.
7026 All modules should be compiled with the same @option{-G @var{num}} value.
7029 @itemx -mno-regnames
7031 @opindex mno-regnames
7032 On System V.4 and embedded PowerPC systems do (do not) emit register
7033 names in the assembly language output using symbolic forms.
7038 @subsection IBM RT Options
7040 @cindex IBM RT options
7042 These @samp{-m} options are defined for the IBM RT PC:
7046 @opindex min-line-mul
7047 Use an in-line code sequence for integer multiplies. This is the
7050 @item -mcall-lib-mul
7051 @opindex mcall-lib-mul
7052 Call @code{lmul$$} for integer multiples.
7054 @item -mfull-fp-blocks
7055 @opindex mfull-fp-blocks
7056 Generate full-size floating point data blocks, including the minimum
7057 amount of scratch space recommended by IBM@. This is the default.
7059 @item -mminimum-fp-blocks
7060 @opindex mminimum-fp-blocks
7061 Do not include extra scratch space in floating point data blocks. This
7062 results in smaller code, but slower execution, since scratch space must
7063 be allocated dynamically.
7065 @cindex @file{varargs.h} and RT PC
7066 @cindex @file{stdarg.h} and RT PC
7067 @item -mfp-arg-in-fpregs
7068 @opindex mfp-arg-in-fpregs
7069 Use a calling sequence incompatible with the IBM calling convention in
7070 which floating point arguments are passed in floating point registers.
7071 Note that @code{varargs.h} and @code{stdarg.h} will not work with
7072 floating point operands if this option is specified.
7074 @item -mfp-arg-in-gregs
7075 @opindex mfp-arg-in-gregs
7076 Use the normal calling convention for floating point arguments. This is
7079 @item -mhc-struct-return
7080 @opindex mhc-struct-return
7081 Return structures of more than one word in memory, rather than in a
7082 register. This provides compatibility with the MetaWare HighC (hc)
7083 compiler. Use the option @option{-fpcc-struct-return} for compatibility
7084 with the Portable C Compiler (pcc).
7086 @item -mnohc-struct-return
7087 @opindex mnohc-struct-return
7088 Return some structures of more than one word in registers, when
7089 convenient. This is the default. For compatibility with the
7090 IBM-supplied compilers, use the option @option{-fpcc-struct-return} or the
7091 option @option{-mhc-struct-return}.
7095 @subsection MIPS Options
7096 @cindex MIPS options
7098 These @samp{-m} options are defined for the MIPS family of computers:
7102 @item -march=@var{cpu-type}
7104 Assume the defaults for the machine type @var{cpu-type} when generating
7105 instructions. The choices for @var{cpu-type} are @samp{r2000}, @samp{r3000},
7106 @samp{r3900}, @samp{r4000}, @samp{r4100}, @samp{r4300}, @samp{r4400},
7107 @samp{r4600}, @samp{r4650}, @samp{r5000}, @samp{r6000}, @samp{r8000},
7108 and @samp{orion}. Additionally, the @samp{r2000}, @samp{r3000},
7109 @samp{r4000}, @samp{r5000}, and @samp{r6000} can be abbreviated as
7110 @samp{r2k} (or @samp{r2K}), @samp{r3k}, etc.
7112 @item -mtune=@var{cpu-type}
7114 Assume the defaults for the machine type @var{cpu-type} when scheduling
7115 instructions. The choices for @var{cpu-type} are @samp{r2000}, @samp{r3000},
7116 @samp{r3900}, @samp{r4000}, @samp{r4100}, @samp{r4300}, @samp{r4400},
7117 @samp{r4600}, @samp{r4650}, @samp{r5000}, @samp{r6000}, @samp{r8000},
7118 and @samp{orion}. Additionally, the @samp{r2000}, @samp{r3000},
7119 @samp{r4000}, @samp{r5000}, and @samp{r6000} can be abbreviated as
7120 @samp{r2k} (or @samp{r2K}), @samp{r3k}, etc. While picking a specific
7121 @var{cpu-type} will schedule things appropriately for that particular
7122 chip, the compiler will not generate any code that does not meet level 1
7123 of the MIPS ISA (instruction set architecture) without a @option{-mipsX}
7124 or @option{-mabi} switch being used.
7126 @item -mcpu=@var{cpu-type}
7128 This is identical to specifying both @option{-march} and @option{-mtune}.
7132 Issue instructions from level 1 of the MIPS ISA@. This is the default.
7133 @samp{r3000} is the default @var{cpu-type} at this ISA level.
7137 Issue instructions from level 2 of the MIPS ISA (branch likely, square
7138 root instructions). @samp{r6000} is the default @var{cpu-type} at this
7143 Issue instructions from level 3 of the MIPS ISA (64-bit instructions).
7144 @samp{r4000} is the default @var{cpu-type} at this ISA level.
7148 Issue instructions from level 4 of the MIPS ISA (conditional move,
7149 prefetch, enhanced FPU instructions). @samp{r8000} is the default
7150 @var{cpu-type} at this ISA level.
7154 Assume that 32 32-bit floating point registers are available. This is
7159 Assume that 32 64-bit floating point registers are available. This is
7160 the default when the @option{-mips3} option is used.
7163 @itemx -mno-fused-madd
7164 @opindex mfused-madd
7165 @opindex mno-fused-madd
7166 Generate code that uses (does not use) the floating point multiply and
7167 accumulate instructions, when they are available. These instructions
7168 are generated by default if they are available, but this may be
7169 undesirable if the extra precision causes problems or on certain chips
7170 in the mode where denormals are rounded to zero where denormals
7171 generated by multiply and accumulate instructions cause exceptions
7176 Assume that 32 32-bit general purpose registers are available. This is
7181 Assume that 32 64-bit general purpose registers are available. This is
7182 the default when the @option{-mips3} option is used.
7186 Force int and long types to be 64 bits wide. See @option{-mlong32} for an
7187 explanation of the default, and the width of pointers.
7191 Force long types to be 64 bits wide. See @option{-mlong32} for an
7192 explanation of the default, and the width of pointers.
7196 Force long, int, and pointer types to be 32 bits wide.
7198 If none of @option{-mlong32}, @option{-mlong64}, or @option{-mint64} are set,
7199 the size of ints, longs, and pointers depends on the ABI and ISA chosen.
7200 For @option{-mabi=32}, and @option{-mabi=n32}, ints and longs are 32 bits
7201 wide. For @option{-mabi=64}, ints are 32 bits, and longs are 64 bits wide.
7202 For @option{-mabi=eabi} and either @option{-mips1} or @option{-mips2}, ints
7203 and longs are 32 bits wide. For @option{-mabi=eabi} and higher ISAs, ints
7204 are 32 bits, and longs are 64 bits wide. The width of pointer types is
7205 the smaller of the width of longs or the width of general purpose
7206 registers (which in turn depends on the ISA)@.
7218 Generate code for the indicated ABI@. The default instruction level is
7219 @option{-mips1} for @samp{32}, @option{-mips3} for @samp{n32}, and
7220 @option{-mips4} otherwise. Conversely, with @option{-mips1} or
7221 @option{-mips2}, the default ABI is @samp{32}; otherwise, the default ABI
7226 Generate code for the MIPS assembler, and invoke @file{mips-tfile} to
7227 add normal debug information. This is the default for all
7228 platforms except for the OSF/1 reference platform, using the OSF/rose
7229 object format. If the either of the @option{-gstabs} or @option{-gstabs+}
7230 switches are used, the @file{mips-tfile} program will encapsulate the
7231 stabs within MIPS ECOFF@.
7235 Generate code for the GNU assembler. This is the default on the OSF/1
7236 reference platform, using the OSF/rose object format. Also, this is
7237 the default if the configure option @option{--with-gnu-as} is used.
7239 @item -msplit-addresses
7240 @itemx -mno-split-addresses
7241 @opindex msplit-addresses
7242 @opindex mno-split-addresses
7243 Generate code to load the high and low parts of address constants separately.
7244 This allows GCC to optimize away redundant loads of the high order
7245 bits of addresses. This optimization requires GNU as and GNU ld.
7246 This optimization is enabled by default for some embedded targets where
7247 GNU as and GNU ld are standard.
7253 The @option{-mrnames} switch says to output code using the MIPS software
7254 names for the registers, instead of the hardware names (ie, @var{a0}
7255 instead of @var{$4}). The only known assembler that supports this option
7256 is the Algorithmics assembler.
7262 The @option{-mgpopt} switch says to write all of the data declarations
7263 before the instructions in the text section, this allows the MIPS
7264 assembler to generate one word memory references instead of using two
7265 words for short global or static data items. This is on by default if
7266 optimization is selected.
7272 For each non-inline function processed, the @option{-mstats} switch
7273 causes the compiler to emit one line to the standard error file to
7274 print statistics about the program (number of registers saved, stack
7281 The @option{-mmemcpy} switch makes all block moves call the appropriate
7282 string function (@samp{memcpy} or @samp{bcopy}) instead of possibly
7283 generating inline code.
7286 @itemx -mno-mips-tfile
7287 @opindex mmips-tfile
7288 @opindex mno-mips-tfile
7289 The @option{-mno-mips-tfile} switch causes the compiler not
7290 postprocess the object file with the @file{mips-tfile} program,
7291 after the MIPS assembler has generated it to add debug support. If
7292 @file{mips-tfile} is not run, then no local variables will be
7293 available to the debugger. In addition, @file{stage2} and
7294 @file{stage3} objects will have the temporary file names passed to the
7295 assembler embedded in the object file, which means the objects will
7296 not compare the same. The @option{-mno-mips-tfile} switch should only
7297 be used when there are bugs in the @file{mips-tfile} program that
7298 prevents compilation.
7301 @opindex msoft-float
7302 Generate output containing library calls for floating point.
7303 @strong{Warning:} the requisite libraries are not part of GCC@.
7304 Normally the facilities of the machine's usual C compiler are used, but
7305 this can't be done directly in cross-compilation. You must make your
7306 own arrangements to provide suitable library functions for
7310 @opindex mhard-float
7311 Generate output containing floating point instructions. This is the
7312 default if you use the unmodified sources.
7315 @itemx -mno-abicalls
7317 @opindex mno-abicalls
7318 Emit (or do not emit) the pseudo operations @samp{.abicalls},
7319 @samp{.cpload}, and @samp{.cprestore} that some System V.4 ports use for
7320 position independent code.
7323 @itemx -mno-long-calls
7324 @opindex mlong-calls
7325 @opindex mno-long-calls
7326 Do all calls with the @samp{JALR} instruction, which requires
7327 loading up a function's address into a register before the call.
7328 You need to use this switch, if you call outside of the current
7329 512 megabyte segment to functions that are not through pointers.
7332 @itemx -mno-half-pic
7334 @opindex mno-half-pic
7335 Put pointers to extern references into the data section and load them
7336 up, rather than put the references in the text section.
7338 @item -membedded-pic
7339 @itemx -mno-embedded-pic
7340 @opindex membedded-pic
7341 @opindex mno-embedded-pic
7342 Generate PIC code suitable for some embedded systems. All calls are
7343 made using PC relative address, and all data is addressed using the $gp
7344 register. No more than 65536 bytes of global data may be used. This
7345 requires GNU as and GNU ld which do most of the work. This currently
7346 only works on targets which use ECOFF; it does not work with ELF@.
7348 @item -membedded-data
7349 @itemx -mno-embedded-data
7350 @opindex membedded-data
7351 @opindex mno-embedded-data
7352 Allocate variables to the read-only data section first if possible, then
7353 next in the small data section if possible, otherwise in data. This gives
7354 slightly slower code than the default, but reduces the amount of RAM required
7355 when executing, and thus may be preferred for some embedded systems.
7357 @item -muninit-const-in-rodata
7358 @itemx -mno-uninit-const-in-rodata
7359 @opindex muninit-const-in-rodata
7360 @opindex mno-uninit-const-in-rodata
7361 When used together with @option{-membedded-data}, it will always store uninitialized
7362 const variables in the read-only data section.
7364 @item -msingle-float
7365 @itemx -mdouble-float
7366 @opindex msingle-float
7367 @opindex mdouble-float
7368 The @option{-msingle-float} switch tells gcc to assume that the floating
7369 point coprocessor only supports single precision operations, as on the
7370 @samp{r4650} chip. The @option{-mdouble-float} switch permits gcc to use
7371 double precision operations. This is the default.
7377 Permit use of the @samp{mad}, @samp{madu} and @samp{mul} instructions,
7378 as on the @samp{r4650} chip.
7382 Turns on @option{-msingle-float}, @option{-mmad}, and, at least for now,
7383 @option{-mcpu=r4650}.
7389 Enable 16-bit instructions.
7393 Use the entry and exit pseudo ops. This option can only be used with
7398 Compile code for the processor in little endian mode.
7399 The requisite libraries are assumed to exist.
7403 Compile code for the processor in big endian mode.
7404 The requisite libraries are assumed to exist.
7408 @cindex smaller data references (MIPS)
7409 @cindex gp-relative references (MIPS)
7410 Put global and static items less than or equal to @var{num} bytes into
7411 the small data or bss sections instead of the normal data or bss
7412 section. This allows the assembler to emit one word memory reference
7413 instructions based on the global pointer (@var{gp} or @var{$28}),
7414 instead of the normal two words used. By default, @var{num} is 8 when
7415 the MIPS assembler is used, and 0 when the GNU assembler is used. The
7416 @option{-G @var{num}} switch is also passed to the assembler and linker.
7417 All modules should be compiled with the same @option{-G @var{num}}
7422 Tell the MIPS assembler to not run its preprocessor over user
7423 assembler files (with a @samp{.s} suffix) when assembling them.
7427 Pass an option to gas which will cause nops to be inserted if
7428 the read of the destination register of an mfhi or mflo instruction
7429 occurs in the following two instructions.
7433 Do not include the default crt0.
7435 @item -mflush-func=@var{func}
7436 @itemx -mno-flush-func
7437 @opindex mflush-func
7438 Specifies the function to call to flush the I and D caches, or to not
7439 call any such function. If called, the function must take the same
7440 arguments as the common @code{_flush_func()}, that is, the address of the
7441 memory range for which the cache is being flushed, the size of the
7442 memory range, and the number 3 (to flush both caches). The default
7443 depends on the target gcc was configured for, but commonly is either
7444 @samp{_flush_func} or @samp{__cpu_flush}.
7448 These options are defined by the macro
7449 @code{TARGET_SWITCHES} in the machine description. The default for the
7450 options is also defined by that macro, which enables you to change the
7454 @node i386 and x86-64 Options
7455 @subsection Intel 386 and AMD x86-64 Options
7456 @cindex i386 Options
7457 @cindex x86-64 Options
7458 @cindex Intel 386 Options
7459 @cindex AMD x86-64 Options
7461 These @samp{-m} options are defined for the i386 and x86-64 family of
7465 @item -mcpu=@var{cpu-type}
7467 Assume the defaults for the machine type @var{cpu-type} when scheduling
7468 instructions. The choices for @var{cpu-type} are @samp{i386},
7469 @samp{i486}, @samp{i586}, @samp{i686}, @samp{pentium},
7470 @samp{pentiumpro}, @samp{pentium4}, @samp{k6}, and @samp{athlon}
7472 While picking a specific @var{cpu-type} will schedule things appropriately
7473 for that particular chip, the compiler will not generate any code that
7474 does not run on the i386 without the @option{-march=@var{cpu-type}} option
7475 being used. @samp{i586} is equivalent to @samp{pentium} and @samp{i686}
7476 is equivalent to @samp{pentiumpro}. @samp{k6} and @samp{athlon} are the
7477 AMD chips as opposed to the Intel ones.
7479 @item -march=@var{cpu-type}
7481 Generate instructions for the machine type @var{cpu-type}. The choices
7482 for @var{cpu-type} are the same as for @option{-mcpu}. Moreover,
7483 specifying @option{-march=@var{cpu-type}} implies @option{-mcpu=@var{cpu-type}}.
7492 @opindex mpentiumpro
7493 These options are synonyms for @option{-mcpu=i386}, @option{-mcpu=i486},
7494 @option{-mcpu=pentium}, and @option{-mcpu=pentiumpro} respectively.
7495 These synonyms are deprecated.
7497 @item -mintel-syntax
7498 @opindex mintel-syntax
7499 Emit assembly using Intel syntax opcodes instead of AT&T syntax.
7504 @opindex mno-ieee-fp
7505 Control whether or not the compiler uses IEEE floating point
7506 comparisons. These handle correctly the case where the result of a
7507 comparison is unordered.
7510 @opindex msoft-float
7511 Generate output containing library calls for floating point.
7512 @strong{Warning:} the requisite libraries are not part of GCC@.
7513 Normally the facilities of the machine's usual C compiler are used, but
7514 this can't be done directly in cross-compilation. You must make your
7515 own arrangements to provide suitable library functions for
7518 On machines where a function returns floating point results in the 80387
7519 register stack, some floating point opcodes may be emitted even if
7520 @option{-msoft-float} is used.
7522 @item -mno-fp-ret-in-387
7523 @opindex mno-fp-ret-in-387
7524 Do not use the FPU registers for return values of functions.
7526 The usual calling convention has functions return values of types
7527 @code{float} and @code{double} in an FPU register, even if there
7528 is no FPU@. The idea is that the operating system should emulate
7531 The option @option{-mno-fp-ret-in-387} causes such values to be returned
7532 in ordinary CPU registers instead.
7534 @item -mno-fancy-math-387
7535 @opindex mno-fancy-math-387
7536 Some 387 emulators do not support the @code{sin}, @code{cos} and
7537 @code{sqrt} instructions for the 387. Specify this option to avoid
7538 generating those instructions. This option is the default on FreeBSD@.
7539 As of revision 2.6.1, these instructions are not generated unless you
7540 also use the @option{-funsafe-math-optimizations} switch.
7542 @item -malign-double
7543 @itemx -mno-align-double
7544 @opindex malign-double
7545 @opindex mno-align-double
7546 Control whether GCC aligns @code{double}, @code{long double}, and
7547 @code{long long} variables on a two word boundary or a one word
7548 boundary. Aligning @code{double} variables on a two word boundary will
7549 produce code that runs somewhat faster on a @samp{Pentium} at the
7550 expense of more memory.
7552 @item -m128bit-long-double
7553 @opindex m128bit-long-double
7554 Control the size of @code{long double} type. i386 application binary interface
7555 specify the size to be 12 bytes, while modern architectures (Pentium and newer)
7556 prefer @code{long double} aligned to 8 or 16 byte boundary. This is
7557 impossible to reach with 12 byte long doubles in the array accesses.
7559 @strong{Warning:} if you use the @option{-m128bit-long-double} switch, the
7560 structures and arrays containing @code{long double} will change their size as
7561 well as function calling convention for function taking @code{long double}
7564 @item -m96bit-long-double
7565 @opindex m96bit-long-double
7566 Set the size of @code{long double} to 96 bits as required by the i386
7567 application binary interface. This is the default.
7570 @itemx -mno-svr3-shlib
7571 @opindex msvr3-shlib
7572 @opindex mno-svr3-shlib
7573 Control whether GCC places uninitialized local variables into the
7574 @code{bss} or @code{data} segments. @option{-msvr3-shlib} places them
7575 into @code{bss}. These options are meaningful only on System V Release 3.
7579 Use a different function-calling convention, in which functions that
7580 take a fixed number of arguments return with the @code{ret} @var{num}
7581 instruction, which pops their arguments while returning. This saves one
7582 instruction in the caller since there is no need to pop the arguments
7585 You can specify that an individual function is called with this calling
7586 sequence with the function attribute @samp{stdcall}. You can also
7587 override the @option{-mrtd} option by using the function attribute
7588 @samp{cdecl}. @xref{Function Attributes}.
7590 @strong{Warning:} this calling convention is incompatible with the one
7591 normally used on Unix, so you cannot use it if you need to call
7592 libraries compiled with the Unix compiler.
7594 Also, you must provide function prototypes for all functions that
7595 take variable numbers of arguments (including @code{printf});
7596 otherwise incorrect code will be generated for calls to those
7599 In addition, seriously incorrect code will result if you call a
7600 function with too many arguments. (Normally, extra arguments are
7601 harmlessly ignored.)
7603 @item -mregparm=@var{num}
7605 Control how many registers are used to pass integer arguments. By
7606 default, no registers are used to pass arguments, and at most 3
7607 registers can be used. You can control this behavior for a specific
7608 function by using the function attribute @samp{regparm}.
7609 @xref{Function Attributes}.
7611 @strong{Warning:} if you use this switch, and
7612 @var{num} is nonzero, then you must build all modules with the same
7613 value, including any libraries. This includes the system libraries and
7616 @item -mpreferred-stack-boundary=@var{num}
7617 @opindex mpreferred-stack-boundary
7618 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
7619 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
7620 the default is 4 (16 bytes or 128 bits), except when optimizing for code
7621 size (@option{-Os}), in which case the default is the minimum correct
7622 alignment (4 bytes for x86, and 8 bytes for x86-64).
7624 On Pentium and PentiumPro, @code{double} and @code{long double} values
7625 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
7626 suffer significant run time performance penalties. On Pentium III, the
7627 Streaming SIMD Extension (SSE) data type @code{__m128} suffers similar
7628 penalties if it is not 16 byte aligned.
7630 To ensure proper alignment of this values on the stack, the stack boundary
7631 must be as aligned as that required by any value stored on the stack.
7632 Further, every function must be generated such that it keeps the stack
7633 aligned. Thus calling a function compiled with a higher preferred
7634 stack boundary from a function compiled with a lower preferred stack
7635 boundary will most likely misalign the stack. It is recommended that
7636 libraries that use callbacks always use the default setting.
7638 This extra alignment does consume extra stack space, and generally
7639 increases code size. Code that is sensitive to stack space usage, such
7640 as embedded systems and operating system kernels, may want to reduce the
7641 preferred alignment to @option{-mpreferred-stack-boundary=2}.
7655 These switches enable or disable the use of built-in functions that allow
7656 direct access to the MMX, SSE and 3Dnow extensions of the instruction set.
7658 The following machine modes are available for use with MMX built-in functions
7659 (@pxref{Vector Extensions}): @code{V2SI} for a vector of two 32 bit integers,
7660 @code{V4HI} for a vector of four 16 bit integers, and @code{V8QI} for a
7661 vector of eight 8 bit integers. Some of the built-in functions operate on
7662 MMX registers as a whole 64 bit entity, these use @code{DI} as their mode.
7664 If 3Dnow extensions are enabled, @code{V2SF} is used as a mode for a vector
7665 of two 32 bit floating point values.
7667 If SSE extensions are enabled, @code{V4SF} is used for a vector of four 32 bit
7668 floating point values. Some instructions use a vector of four 32 bit
7669 integers, these use @code{V4SI}. Finally, some instructions operate on an
7670 entire vector register, interpreting it as a 128 bit integer, these use mode
7673 The following built-in functions are made available by @option{-mmmx}:
7675 @item v8qi __builtin_ia32_paddb (v8qi, v8qi)
7676 Generates the @code{paddb} machine instruction.
7677 @item v4hi __builtin_ia32_paddw (v4hi, v4hi)
7678 Generates the @code{paddw} machine instruction.
7679 @item v2si __builtin_ia32_paddd (v2si, v2si)
7680 Generates the @code{paddd} machine instruction.
7681 @item v8qi __builtin_ia32_psubb (v8qi, v8qi)
7682 Generates the @code{psubb} machine instruction.
7683 @item v4hi __builtin_ia32_psubw (v4hi, v4hi)
7684 Generates the @code{psubw} machine instruction.
7685 @item v2si __builtin_ia32_psubd (v2si, v2si)
7686 Generates the @code{psubd} machine instruction.
7688 @item v8qi __builtin_ia32_paddsb (v8qi, v8qi)
7689 Generates the @code{paddsb} machine instruction.
7690 @item v4hi __builtin_ia32_paddsw (v4hi, v4hi)
7691 Generates the @code{paddsw} machine instruction.
7692 @item v8qi __builtin_ia32_psubsb (v8qi, v8qi)
7693 Generates the @code{psubsb} machine instruction.
7694 @item v4hi __builtin_ia32_psubsw (v4hi, v4hi)
7695 Generates the @code{psubsw} machine instruction.
7697 @item v8qi __builtin_ia32_paddusb (v8qi, v8qi)
7698 Generates the @code{paddusb} machine instruction.
7699 @item v4hi __builtin_ia32_paddusw (v4hi, v4hi)
7700 Generates the @code{paddusw} machine instruction.
7701 @item v8qi __builtin_ia32_psubusb (v8qi, v8qi)
7702 Generates the @code{psubusb} machine instruction.
7703 @item v4hi __builtin_ia32_psubusw (v4hi, v4hi)
7704 Generates the @code{psubusw} machine instruction.
7706 @item v4hi __builtin_ia32_pmullw (v4hi, v4hi)
7707 Generates the @code{pmullw} machine instruction.
7708 @item v4hi __builtin_ia32_pmulhw (v4hi, v4hi)
7709 Generates the @code{pmulhw} machine instruction.
7711 @item di __builtin_ia32_pand (di, di)
7712 Generates the @code{pand} machine instruction.
7713 @item di __builtin_ia32_pandn (di,di)
7714 Generates the @code{pandn} machine instruction.
7715 @item di __builtin_ia32_por (di, di)
7716 Generates the @code{por} machine instruction.
7717 @item di __builtin_ia32_pxor (di, di)
7718 Generates the @code{pxor} machine instruction.
7720 @item v8qi __builtin_ia32_pcmpeqb (v8qi, v8qi)
7721 Generates the @code{pcmpeqb} machine instruction.
7722 @item v4hi __builtin_ia32_pcmpeqw (v4hi, v4hi)
7723 Generates the @code{pcmpeqw} machine instruction.
7724 @item v2si __builtin_ia32_pcmpeqd (v2si, v2si)
7725 Generates the @code{pcmpeqd} machine instruction.
7726 @item v8qi __builtin_ia32_pcmpgtb (v8qi, v8qi)
7727 Generates the @code{pcmpgtb} machine instruction.
7728 @item v4hi __builtin_ia32_pcmpgtw (v4hi, v4hi)
7729 Generates the @code{pcmpgtw} machine instruction.
7730 @item v2si __builtin_ia32_pcmpgtd (v2si, v2si)
7731 Generates the @code{pcmpgtd} machine instruction.
7733 @item v8qi __builtin_ia32_punpckhbw (v8qi, v8qi)
7734 Generates the @code{punpckhbw} machine instruction.
7735 @item v4hi __builtin_ia32_punpckhwd (v4hi, v4hi)
7736 Generates the @code{punpckhwd} machine instruction.
7737 @item v2si __builtin_ia32_punpckhdq (v2si, v2si)
7738 Generates the @code{punpckhdq} machine instruction.
7739 @item v8qi __builtin_ia32_punpcklbw (v8qi, v8qi)
7740 Generates the @code{punpcklbw} machine instruction.
7741 @item v4hi __builtin_ia32_punpcklwd (v4hi, v4hi)
7742 Generates the @code{punpcklwd} machine instruction.
7743 @item v2si __builtin_ia32_punpckldq (v2si, v2si)
7744 Generates the @code{punpckldq} machine instruction.
7746 @item v8qi __builtin_ia32_packsswb (v4hi, v4hi)
7747 Generates the @code{packsswb} machine instruction.
7748 @item v4hi __builtin_ia32_packssdw (v2si, v2si)
7749 Generates the @code{packssdw} machine instruction.
7750 @item v8qi __builtin_ia32_packuswb (v4hi, v4hi)
7751 Generates the @code{packuswb} machine instruction.
7755 The following built-in functions are made available either with @option{-msse}, or
7756 with a combination of @option{-m3dnow} and @option{-march=athlon}.
7759 @item v4hi __builtin_ia32_pmulhuw (v4hi, v4hi)
7760 Generates the @code{pmulhuw} machine instruction.
7762 @item v8qi __builtin_ia32_pavgb (v8qi, v8qi)
7763 Generates the @code{pavgb} machine instruction.
7764 @item v4hi __builtin_ia32_pavgw (v4hi, v4hi)
7765 Generates the @code{pavgw} machine instruction.
7766 @item v4hi __builtin_ia32_psadbw (v8qi, v8qi)
7767 Generates the @code{psadbw} machine instruction.
7769 @item v8qi __builtin_ia32_pmaxub (v8qi, v8qi)
7770 Generates the @code{pmaxub} machine instruction.
7771 @item v4hi __builtin_ia32_pmaxsw (v4hi, v4hi)
7772 Generates the @code{pmaxsw} machine instruction.
7773 @item v8qi __builtin_ia32_pminub (v8qi, v8qi)
7774 Generates the @code{pminub} machine instruction.
7775 @item v4hi __builtin_ia32_pminsw (v4hi, v4hi)
7776 Generates the @code{pminsw} machine instruction.
7778 @item int __builtin_ia32_pextrw (v4hi, int)
7779 Generates the @code{pextrw} machine instruction.
7780 @item v4hi __builtin_ia32_pinsrw (v4hi, int, int)
7781 Generates the @code{pinsrw} machine instruction.
7783 @item int __builtin_ia32_pmovmskb (v8qi)
7784 Generates the @code{pmovmskb} machine instruction.
7785 @item void __builtin_ia32_maskmovq (v8qi, v8qi, char *)
7786 Generates the @code{maskmovq} machine instruction.
7787 @item void __builtin_ia32_movntq (di *, di)
7788 Generates the @code{movntq} machine instruction.
7789 @item void __builtin_ia32_sfence (void)
7790 Generates the @code{sfence} machine instruction.
7791 @item void __builtin_ia32_prefetch (char *, int selector)
7792 Generates a prefetch machine instruction, depending on the value of
7793 selector. If @code{selector} is 0, it generates @code{prefetchnta}; for
7794 a value of 1, it generates @code{prefetcht0}; for a value of 2, it generates
7795 @code{prefetcht1}; and for a value of 3 it generates @code{prefetcht2}.
7799 The following built-in functions are available when @option{-msse} is used.
7802 @item int __builtin_ia32_comieq (v4sf, v4sf)
7803 Generates the @code{comiss} machine instruction and performs an equality
7804 comparison. The return value is the truth value of that comparison.
7805 @item int __builtin_ia32_comineq (v4sf, v4sf)
7806 Generates the @code{comiss} machine instruction and performs an inequality
7807 comparison. The return value is the truth value of that comparison.
7808 @item int __builtin_ia32_comilt (v4sf, v4sf)
7809 Generates the @code{comiss} machine instruction and performs a ``less than''
7810 comparison. The return value is the truth value of that comparison.
7811 @item int __builtin_ia32_comile (v4sf, v4sf)
7812 Generates the @code{comiss} machine instruction and performs a ``less or
7813 equal'' comparison. The return value is the truth value of that comparison.
7814 @item int __builtin_ia32_comigt (v4sf, v4sf)
7815 Generates the @code{comiss} machine instruction and performs a ``greater than''
7816 comparison. The return value is the truth value of that comparison.
7817 @item int __builtin_ia32_comige (v4sf, v4sf)
7818 Generates the @code{comiss} machine instruction and performs a ``greater or
7819 equal'' comparison. The return value is the truth value of that comparison.
7821 @item int __builtin_ia32_ucomieq (v4sf, v4sf)
7822 Generates the @code{ucomiss} machine instruction and performs an equality
7823 comparison. The return value is the truth value of that comparison.
7824 @item int __builtin_ia32_ucomineq (v4sf, v4sf)
7825 Generates the @code{ucomiss} machine instruction and performs an inequality
7826 comparison. The return value is the truth value of that comparison.
7827 @item int __builtin_ia32_ucomilt (v4sf, v4sf)
7828 Generates the @code{ucomiss} machine instruction and performs a ``less than''
7829 comparison. The return value is the truth value of that comparison.
7830 @item int __builtin_ia32_ucomile (v4sf, v4sf)
7831 Generates the @code{ucomiss} machine instruction and performs a ``less or
7832 equal'' comparison. The return value is the truth value of that comparison.
7833 @item int __builtin_ia32_ucomigt (v4sf, v4sf)
7834 Generates the @code{ucomiss} machine instruction and performs a ``greater than''
7835 comparison. The return value is the truth value of that comparison.
7836 @item int __builtin_ia32_ucomige (v4sf, v4sf)
7837 Generates the @code{ucomiss} machine instruction and performs a ``greater or
7838 equal'' comparison. The return value is the truth value of that comparison.
7840 @item v4sf __builtin_ia32_addps (v4sf, v4sf)
7841 Generates the @code{addps} machine instruction.
7842 @item v4sf __builtin_ia32_addss (v4sf, v4sf)
7843 Generates the @code{addss} machine instruction.
7844 @item v4sf __builtin_ia32_subps (v4sf, v4sf)
7845 Generates the @code{subps} machine instruction.
7846 @item v4sf __builtin_ia32_subss (v4sf, v4sf)
7847 Generates the @code{subss} machine instruction.
7848 @item v4sf __builtin_ia32_mulps (v4sf, v4sf)
7849 Generates the @code{mulps} machine instruction.
7850 @item v4sf __builtin_ia32_mulss (v4sf, v4sf)
7851 Generates the @code{mulss} machine instruction.
7852 @item v4sf __builtin_ia32_divps (v4sf, v4sf)
7853 Generates the @code{divps} machine instruction.
7854 @item v4sf __builtin_ia32_divss (v4sf, v4sf)
7855 Generates the @code{divss} machine instruction.
7857 @item v4si __builtin_ia32_cmpeqps (v4sf, v4sf)
7858 Generates the @code{cmpeqps} machine instruction.
7859 @item v4si __builtin_ia32_cmplts (v4sf, v4sf)
7860 Generates the @code{cmpltps} machine instruction.
7861 @item v4si __builtin_ia32_cmpleps (v4sf, v4sf)
7862 Generates the @code{cmpleps} machine instruction.
7863 @item v4si __builtin_ia32_cmpgtps (v4sf, v4sf)
7864 Generates the @code{cmpgtps} machine instruction.
7865 @item v4si __builtin_ia32_cmpgeps (v4sf, v4sf)
7866 Generates the @code{cmpgeps} machine instruction.
7867 @item v4si __builtin_ia32_cmpunordps (v4sf, v4sf)
7868 Generates the @code{cmpunodps} machine instruction.
7869 @item v4si __builtin_ia32_cmpneqps (v4sf, v4sf)
7870 Generates the @code{cmpeqps} machine instruction.
7871 @item v4si __builtin_ia32_cmpnltps (v4sf, v4sf)
7872 Generates the @code{cmpltps} machine instruction.
7873 @item v4si __builtin_ia32_cmpnleps (v4sf, v4sf)
7874 Generates the @code{cmpleps} machine instruction.
7875 @item v4si __builtin_ia32_cmpngtps (v4sf, v4sf)
7876 Generates the @code{cmpgtps} machine instruction.
7877 @item v4si __builtin_ia32_cmpngeps (v4sf, v4sf)
7878 Generates the @code{cmpgeps} machine instruction.
7879 @item v4si __builtin_ia32_cmpordps (v4sf, v4sf)
7880 Generates the @code{cmpunodps} machine instruction.
7882 @item v4si __builtin_ia32_cmpeqss (v4sf, v4sf)
7883 Generates the @code{cmpeqss} machine instruction.
7884 @item v4si __builtin_ia32_cmpltss (v4sf, v4sf)
7885 Generates the @code{cmpltss} machine instruction.
7886 @item v4si __builtin_ia32_cmpless (v4sf, v4sf)
7887 Generates the @code{cmpless} machine instruction.
7888 @item v4si __builtin_ia32_cmpgtss (v4sf, v4sf)
7889 Generates the @code{cmpgtss} machine instruction.
7890 @item v4si __builtin_ia32_cmpgess (v4sf, v4sf)
7891 Generates the @code{cmpgess} machine instruction.
7892 @item v4si __builtin_ia32_cmpunordss (v4sf, v4sf)
7893 Generates the @code{cmpunodss} machine instruction.
7894 @item v4si __builtin_ia32_cmpneqss (v4sf, v4sf)
7895 Generates the @code{cmpeqss} machine instruction.
7896 @item v4si __builtin_ia32_cmpnlts (v4sf, v4sf)
7897 Generates the @code{cmpltss} machine instruction.
7898 @item v4si __builtin_ia32_cmpnless (v4sf, v4sf)
7899 Generates the @code{cmpless} machine instruction.
7900 @item v4si __builtin_ia32_cmpngtss (v4sf, v4sf)
7901 Generates the @code{cmpgtss} machine instruction.
7902 @item v4si __builtin_ia32_cmpngess (v4sf, v4sf)
7903 Generates the @code{cmpgess} machine instruction.
7904 @item v4si __builtin_ia32_cmpordss (v4sf, v4sf)
7905 Generates the @code{cmpunodss} machine instruction.
7907 @item v4sf __builtin_ia32_maxps (v4sf, v4sf)
7908 Generates the @code{maxps} machine instruction.
7909 @item v4sf __builtin_ia32_maxsss (v4sf, v4sf)
7910 Generates the @code{maxss} machine instruction.
7911 @item v4sf __builtin_ia32_minps (v4sf, v4sf)
7912 Generates the @code{minps} machine instruction.
7913 @item v4sf __builtin_ia32_minsss (v4sf, v4sf)
7914 Generates the @code{minss} machine instruction.
7916 @item ti __builtin_ia32_andps (ti, ti)
7917 Generates the @code{andps} machine instruction.
7918 @item ti __builtin_ia32_andnps (ti, ti)
7919 Generates the @code{andnps} machine instruction.
7920 @item ti __builtin_ia32_orps (ti, ti)
7921 Generates the @code{orps} machine instruction.
7922 @item ti __builtin_ia32_xorps (ti, ti)
7923 Generates the @code{xorps} machine instruction.
7925 @item v4sf __builtin_ia32_movps (v4sf, v4sf)
7926 Generates the @code{movps} machine instruction.
7927 @item v4sf __builtin_ia32_movhlps (v4sf, v4sf)
7928 Generates the @code{movhlps} machine instruction.
7929 @item v4sf __builtin_ia32_movlhps (v4sf, v4sf)
7930 Generates the @code{movlhps} machine instruction.
7931 @item v4sf __builtin_ia32_unpckhps (v4sf, v4sf)
7932 Generates the @code{unpckhps} machine instruction.
7933 @item v4sf __builtin_ia32_unpcklps (v4sf, v4sf)
7934 Generates the @code{unpcklps} machine instruction.
7936 @item v4sf __builtin_ia32_cvtpi2ps (v4sf, v2si)
7937 Generates the @code{cvtpi2ps} machine instruction.
7938 @item v2si __builtin_ia32_cvtps2pi (v4sf)
7939 Generates the @code{cvtps2pi} machine instruction.
7940 @item v4sf __builtin_ia32_cvtsi2ss (v4sf, int)
7941 Generates the @code{cvtsi2ss} machine instruction.
7942 @item int __builtin_ia32_cvtss2si (v4sf)
7943 Generates the @code{cvtsi2ss} machine instruction.
7944 @item v2si __builtin_ia32_cvttps2pi (v4sf)
7945 Generates the @code{cvttps2pi} machine instruction.
7946 @item int __builtin_ia32_cvttss2si (v4sf)
7947 Generates the @code{cvttsi2ss} machine instruction.
7949 @item v4sf __builtin_ia32_rcpps (v4sf)
7950 Generates the @code{rcpps} machine instruction.
7951 @item v4sf __builtin_ia32_rsqrtps (v4sf)
7952 Generates the @code{rsqrtps} machine instruction.
7953 @item v4sf __builtin_ia32_sqrtps (v4sf)
7954 Generates the @code{sqrtps} machine instruction.
7955 @item v4sf __builtin_ia32_rcpss (v4sf)
7956 Generates the @code{rcpss} machine instruction.
7957 @item v4sf __builtin_ia32_rsqrtss (v4sf)
7958 Generates the @code{rsqrtss} machine instruction.
7959 @item v4sf __builtin_ia32_sqrtss (v4sf)
7960 Generates the @code{sqrtss} machine instruction.
7962 @item v4sf __builtin_ia32_shufps (v4sf, v4sf, int)
7963 Generates the @code{shufps} machine instruction.
7965 @item v4sf __builtin_ia32_loadaps (float *)
7966 Generates the @code{movaps} machine instruction as a load from memory.
7967 @item void __builtin_ia32_storeaps (float *, v4sf)
7968 Generates the @code{movaps} machine instruction as a store to memory.
7969 @item v4sf __builtin_ia32_loadups (float *)
7970 Generates the @code{movups} machine instruction as a load from memory.
7971 @item void __builtin_ia32_storeups (float *, v4sf)
7972 Generates the @code{movups} machine instruction as a store to memory.
7973 @item v4sf __builtin_ia32_loadsss (float *)
7974 Generates the @code{movss} machine instruction as a load from memory.
7975 @item void __builtin_ia32_storess (float *, v4sf)
7976 Generates the @code{movss} machine instruction as a store to memory.
7978 @item v4sf __builtin_ia32_loadhps (v4sf, v2si *)
7979 Generates the @code{movhps} machine instruction as a load from memory.
7980 @item v4sf __builtin_ia32_loadlps (v4sf, v2si *)
7981 Generates the @code{movlps} machine instruction as a load from memory
7982 @item void __builtin_ia32_storehps (v4sf, v2si *)
7983 Generates the @code{movhps} machine instruction as a store to memory.
7984 @item void __builtin_ia32_storelps (v4sf, v2si *)
7985 Generates the @code{movlps} machine instruction as a store to memory.
7987 @item void __builtin_ia32_movntps (float *, v4sf)
7988 Generates the @code{movntps} machine instruction.
7989 @item int __builtin_ia32_movmskps (v4sf)
7990 Generates the @code{movntps} machine instruction.
7992 @item void __builtin_ia32_storeps1 (float *, v4sf)
7993 Generates the @code{movaps} machine instruction as a store to memory.
7994 Before storing, the value is modified with a @code{shufps} instruction
7995 so that the lowest of the four floating point elements is replicated
7996 across the entire vector that is stored.
7997 @item void __builtin_ia32_storerps (float *, v4sf)
7998 Generates the @code{movaps} machine instruction as a store to memory.
7999 Before storing, the value is modified with a @code{shufps} instruction
8000 so that the order of the four floating point elements in the vector is
8002 @item v4sf __builtin_ia32_loadps1 (float *)
8003 Generates a @code{movss} machine instruction to load a floating point
8004 value from memory, and a @code{shufps} instruction to replicate the
8005 loaded value across all four elements of the result vector.
8006 @item v4sf __builtin_ia32_loadrps (float *)
8007 Generates a @code{movaps} machine instruction to load a vector from
8008 memory, and a @code{shufps} instruction to reverse the order of the
8009 four floating point elements in the result vector.
8010 @item v4sf __builtin_ia32_setps (float, float, float, float)
8011 Constructs a vector from four single floating point values. The return
8012 value is equal to the value that would result from storing the four
8013 arguments into consecutive memory locations and then executing a
8014 @code{movaps} to load the vector from memory.
8015 @item v4sf __builtin_ia32_setps1 (float)
8016 Constructs a vector from a single floating point value by replicating
8017 it across all four elements of the result vector.
8021 @itemx -mno-push-args
8023 @opindex mno-push-args
8024 Use PUSH operations to store outgoing parameters. This method is shorter
8025 and usually equally fast as method using SUB/MOV operations and is enabled
8026 by default. In some cases disabling it may improve performance because of
8027 improved scheduling and reduced dependencies.
8029 @item -maccumulate-outgoing-args
8030 @opindex maccumulate-outgoing-args
8031 If enabled, the maximum amount of space required for outgoing arguments will be
8032 computed in the function prologue. This is faster on most modern CPUs
8033 because of reduced dependencies, improved scheduling and reduced stack usage
8034 when preferred stack boundary is not equal to 2. The drawback is a notable
8035 increase in code size. This switch implies @option{-mno-push-args}.
8039 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
8040 on thread-safe exception handling must compile and link all code with the
8041 @option{-mthreads} option. When compiling, @option{-mthreads} defines
8042 @option{-D_MT}; when linking, it links in a special thread helper library
8043 @option{-lmingwthrd} which cleans up per thread exception handling data.
8045 @item -mno-align-stringops
8046 @opindex mno-align-stringops
8047 Do not align destination of inlined string operations. This switch reduces
8048 code size and improves performance in case the destination is already aligned,
8049 but gcc don't know about it.
8051 @item -minline-all-stringops
8052 @opindex minline-all-stringops
8053 By default GCC inlines string operations only when destination is known to be
8054 aligned at least to 4 byte boundary. This enables more inlining, increase code
8055 size, but may improve performance of code that depends on fast memcpy, strlen
8056 and memset for short lengths.
8058 @item -momit-leaf-frame-pointer
8059 @opindex momit-leaf-frame-pointer
8060 Don't keep the frame pointer in a register for leaf functions. This
8061 avoids the instructions to save, set up and restore frame pointers and
8062 makes an extra register available in leaf functions. The option
8063 @option{-fomit-frame-pointer} removes the frame pointer for all functions
8064 which might make debugging harder.
8067 These @samp{-m} switches are supported in addition to the above
8068 on AMD x86-64 processors in 64-bit environments.
8075 Generate code for a 32-bit or 64-bit environment.
8076 The 32-bit environment sets int, long and pointer to 32 bits and
8077 generates code that runs on any i386 system.
8078 The 64-bit environment sets int to 32 bits and long and pointer
8079 to 64 bits and generates code for AMD's x86-64 architecture.
8082 @opindex no-red-zone
8083 Do not use a so called red zone for x86-64 code. The red zone is mandated
8084 by the x86-64 ABI, it is a 128-byte area beyond the location of the
8085 stack pointer that will not be modified by signal or interrupt handlers
8086 and therefore can be used for temporary data without adjusting the stack
8087 pointer. The flag @option{-mno-red-zone} disables this red zone.
8091 @subsection HPPA Options
8092 @cindex HPPA Options
8094 These @samp{-m} options are defined for the HPPA family of computers:
8097 @item -march=@var{architecture-type}
8099 Generate code for the specified architecture. The choices for
8100 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
8101 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
8102 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
8103 architecture option for your machine. Code compiled for lower numbered
8104 architectures will run on higher numbered architectures, but not the
8107 PA 2.0 support currently requires gas snapshot 19990413 or later. The
8108 next release of binutils (current is 2.9.1) will probably contain PA 2.0
8112 @itemx -mpa-risc-1-1
8113 @itemx -mpa-risc-2-0
8114 @opindex mpa-risc-1-0
8115 @opindex mpa-risc-1-1
8116 @opindex mpa-risc-2-0
8117 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
8120 @opindex mbig-switch
8121 Generate code suitable for big switch tables. Use this option only if
8122 the assembler/linker complain about out of range branches within a switch
8125 @item -mjump-in-delay
8126 @opindex mjump-in-delay
8127 Fill delay slots of function calls with unconditional jump instructions
8128 by modifying the return pointer for the function call to be the target
8129 of the conditional jump.
8131 @item -mdisable-fpregs
8132 @opindex mdisable-fpregs
8133 Prevent floating point registers from being used in any manner. This is
8134 necessary for compiling kernels which perform lazy context switching of
8135 floating point registers. If you use this option and attempt to perform
8136 floating point operations, the compiler will abort.
8138 @item -mdisable-indexing
8139 @opindex mdisable-indexing
8140 Prevent the compiler from using indexing address modes. This avoids some
8141 rather obscure problems when compiling MIG generated code under MACH@.
8143 @item -mno-space-regs
8144 @opindex mno-space-regs
8145 Generate code that assumes the target has no space registers. This allows
8146 GCC to generate faster indirect calls and use unscaled index address modes.
8148 Such code is suitable for level 0 PA systems and kernels.
8150 @item -mfast-indirect-calls
8151 @opindex mfast-indirect-calls
8152 Generate code that assumes calls never cross space boundaries. This
8153 allows GCC to emit code which performs faster indirect calls.
8155 This option will not work in the presence of shared libraries or nested
8158 @item -mlong-load-store
8159 @opindex mlong-load-store
8160 Generate 3-instruction load and store sequences as sometimes required by
8161 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
8164 @item -mportable-runtime
8165 @opindex mportable-runtime
8166 Use the portable calling conventions proposed by HP for ELF systems.
8170 Enable the use of assembler directives only GAS understands.
8172 @item -mschedule=@var{cpu-type}
8174 Schedule code according to the constraints for the machine type
8175 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
8176 @samp{7100}, @samp{7100LC}, @samp{7200}, and @samp{8000}. Refer to
8177 @file{/usr/lib/sched.models} on an HP-UX system to determine the
8178 proper scheduling option for your machine.
8181 @opindex mlinker-opt
8182 Enable the optimization pass in the HPUX linker. Note this makes symbolic
8183 debugging impossible. It also triggers a bug in the HPUX 8 and HPUX 9 linkers
8184 in which they give bogus error messages when linking some programs.
8187 @opindex msoft-float
8188 Generate output containing library calls for floating point.
8189 @strong{Warning:} the requisite libraries are not available for all HPPA
8190 targets. Normally the facilities of the machine's usual C compiler are
8191 used, but this cannot be done directly in cross-compilation. You must make
8192 your own arrangements to provide suitable library functions for
8193 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
8194 does provide software floating point support.
8196 @option{-msoft-float} changes the calling convention in the output file;
8197 therefore, it is only useful if you compile @emph{all} of a program with
8198 this option. In particular, you need to compile @file{libgcc.a}, the
8199 library that comes with GCC, with @option{-msoft-float} in order for
8203 @node Intel 960 Options
8204 @subsection Intel 960 Options
8206 These @samp{-m} options are defined for the Intel 960 implementations:
8209 @item -m@var{cpu-type}
8217 Assume the defaults for the machine type @var{cpu-type} for some of
8218 the other options, including instruction scheduling, floating point
8219 support, and addressing modes. The choices for @var{cpu-type} are
8220 @samp{ka}, @samp{kb}, @samp{mc}, @samp{ca}, @samp{cf},
8221 @samp{sa}, and @samp{sb}.
8228 @opindex msoft-float
8229 The @option{-mnumerics} option indicates that the processor does support
8230 floating-point instructions. The @option{-msoft-float} option indicates
8231 that floating-point support should not be assumed.
8233 @item -mleaf-procedures
8234 @itemx -mno-leaf-procedures
8235 @opindex mleaf-procedures
8236 @opindex mno-leaf-procedures
8237 Do (or do not) attempt to alter leaf procedures to be callable with the
8238 @code{bal} instruction as well as @code{call}. This will result in more
8239 efficient code for explicit calls when the @code{bal} instruction can be
8240 substituted by the assembler or linker, but less efficient code in other
8241 cases, such as calls via function pointers, or using a linker that doesn't
8242 support this optimization.
8245 @itemx -mno-tail-call
8247 @opindex mno-tail-call
8248 Do (or do not) make additional attempts (beyond those of the
8249 machine-independent portions of the compiler) to optimize tail-recursive
8250 calls into branches. You may not want to do this because the detection of
8251 cases where this is not valid is not totally complete. The default is
8252 @option{-mno-tail-call}.
8254 @item -mcomplex-addr
8255 @itemx -mno-complex-addr
8256 @opindex mcomplex-addr
8257 @opindex mno-complex-addr
8258 Assume (or do not assume) that the use of a complex addressing mode is a
8259 win on this implementation of the i960. Complex addressing modes may not
8260 be worthwhile on the K-series, but they definitely are on the C-series.
8261 The default is currently @option{-mcomplex-addr} for all processors except
8265 @itemx -mno-code-align
8266 @opindex mcode-align
8267 @opindex mno-code-align
8268 Align code to 8-byte boundaries for faster fetching (or don't bother).
8269 Currently turned on by default for C-series implementations only.
8272 @item -mclean-linkage
8273 @itemx -mno-clean-linkage
8274 @opindex mclean-linkage
8275 @opindex mno-clean-linkage
8276 These options are not fully implemented.
8280 @itemx -mic2.0-compat
8281 @itemx -mic3.0-compat
8283 @opindex mic2.0-compat
8284 @opindex mic3.0-compat
8285 Enable compatibility with iC960 v2.0 or v3.0.
8289 @opindex masm-compat
8291 Enable compatibility with the iC960 assembler.
8293 @item -mstrict-align
8294 @itemx -mno-strict-align
8295 @opindex mstrict-align
8296 @opindex mno-strict-align
8297 Do not permit (do permit) unaligned accesses.
8301 Enable structure-alignment compatibility with Intel's gcc release version
8302 1.3 (based on gcc 1.37). This option implies @option{-mstrict-align}.
8304 @item -mlong-double-64
8305 @opindex mlong-double-64
8306 Implement type @samp{long double} as 64-bit floating point numbers.
8307 Without the option @samp{long double} is implemented by 80-bit
8308 floating point numbers. The only reason we have it because there is
8309 no 128-bit @samp{long double} support in @samp{fp-bit.c} yet. So it
8310 is only useful for people using soft-float targets. Otherwise, we
8311 should recommend against use of it.
8315 @node DEC Alpha Options
8316 @subsection DEC Alpha Options
8318 These @samp{-m} options are defined for the DEC Alpha implementations:
8321 @item -mno-soft-float
8323 @opindex mno-soft-float
8324 @opindex msoft-float
8325 Use (do not use) the hardware floating-point instructions for
8326 floating-point operations. When @option{-msoft-float} is specified,
8327 functions in @file{libgcc.a} will be used to perform floating-point
8328 operations. Unless they are replaced by routines that emulate the
8329 floating-point operations, or compiled in such a way as to call such
8330 emulations routines, these routines will issue floating-point
8331 operations. If you are compiling for an Alpha without floating-point
8332 operations, you must ensure that the library is built so as not to call
8335 Note that Alpha implementations without floating-point operations are
8336 required to have floating-point registers.
8341 @opindex mno-fp-regs
8342 Generate code that uses (does not use) the floating-point register set.
8343 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
8344 register set is not used, floating point operands are passed in integer
8345 registers as if they were integers and floating-point results are passed
8346 in $0 instead of $f0. This is a non-standard calling sequence, so any
8347 function with a floating-point argument or return value called by code
8348 compiled with @option{-mno-fp-regs} must also be compiled with that
8351 A typical use of this option is building a kernel that does not use,
8352 and hence need not save and restore, any floating-point registers.
8356 The Alpha architecture implements floating-point hardware optimized for
8357 maximum performance. It is mostly compliant with the IEEE floating
8358 point standard. However, for full compliance, software assistance is
8359 required. This option generates code fully IEEE compliant code
8360 @emph{except} that the @var{inexact-flag} is not maintained (see below).
8361 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
8362 defined during compilation. The resulting code is less efficient but is
8363 able to correctly support denormalized numbers and exceptional IEEE
8364 values such as not-a-number and plus/minus infinity. Other Alpha
8365 compilers call this option @option{-ieee_with_no_inexact}.
8367 @item -mieee-with-inexact
8368 @opindex mieee-with-inexact
8369 This is like @option{-mieee} except the generated code also maintains
8370 the IEEE @var{inexact-flag}. Turning on this option causes the
8371 generated code to implement fully-compliant IEEE math. In addition to
8372 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
8373 macro. On some Alpha implementations the resulting code may execute
8374 significantly slower than the code generated by default. Since there is
8375 very little code that depends on the @var{inexact-flag}, you should
8376 normally not specify this option. Other Alpha compilers call this
8377 option @option{-ieee_with_inexact}.
8379 @item -mfp-trap-mode=@var{trap-mode}
8380 @opindex mfp-trap-mode
8381 This option controls what floating-point related traps are enabled.
8382 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
8383 The trap mode can be set to one of four values:
8387 This is the default (normal) setting. The only traps that are enabled
8388 are the ones that cannot be disabled in software (e.g., division by zero
8392 In addition to the traps enabled by @samp{n}, underflow traps are enabled
8396 Like @samp{su}, but the instructions are marked to be safe for software
8397 completion (see Alpha architecture manual for details).
8400 Like @samp{su}, but inexact traps are enabled as well.
8403 @item -mfp-rounding-mode=@var{rounding-mode}
8404 @opindex mfp-rounding-mode
8405 Selects the IEEE rounding mode. Other Alpha compilers call this option
8406 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
8411 Normal IEEE rounding mode. Floating point numbers are rounded towards
8412 the nearest machine number or towards the even machine number in case
8416 Round towards minus infinity.
8419 Chopped rounding mode. Floating point numbers are rounded towards zero.
8422 Dynamic rounding mode. A field in the floating point control register
8423 (@var{fpcr}, see Alpha architecture reference manual) controls the
8424 rounding mode in effect. The C library initializes this register for
8425 rounding towards plus infinity. Thus, unless your program modifies the
8426 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
8429 @item -mtrap-precision=@var{trap-precision}
8430 @opindex mtrap-precision
8431 In the Alpha architecture, floating point traps are imprecise. This
8432 means without software assistance it is impossible to recover from a
8433 floating trap and program execution normally needs to be terminated.
8434 GCC can generate code that can assist operating system trap handlers
8435 in determining the exact location that caused a floating point trap.
8436 Depending on the requirements of an application, different levels of
8437 precisions can be selected:
8441 Program precision. This option is the default and means a trap handler
8442 can only identify which program caused a floating point exception.
8445 Function precision. The trap handler can determine the function that
8446 caused a floating point exception.
8449 Instruction precision. The trap handler can determine the exact
8450 instruction that caused a floating point exception.
8453 Other Alpha compilers provide the equivalent options called
8454 @option{-scope_safe} and @option{-resumption_safe}.
8456 @item -mieee-conformant
8457 @opindex mieee-conformant
8458 This option marks the generated code as IEEE conformant. You must not
8459 use this option unless you also specify @option{-mtrap-precision=i} and either
8460 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
8461 is to emit the line @samp{.eflag 48} in the function prologue of the
8462 generated assembly file. Under DEC Unix, this has the effect that
8463 IEEE-conformant math library routines will be linked in.
8465 @item -mbuild-constants
8466 @opindex mbuild-constants
8467 Normally GCC examines a 32- or 64-bit integer constant to
8468 see if it can construct it from smaller constants in two or three
8469 instructions. If it cannot, it will output the constant as a literal and
8470 generate code to load it from the data segment at runtime.
8472 Use this option to require GCC to construct @emph{all} integer constants
8473 using code, even if it takes more instructions (the maximum is six).
8475 You would typically use this option to build a shared library dynamic
8476 loader. Itself a shared library, it must relocate itself in memory
8477 before it can find the variables and constants in its own data segment.
8483 Select whether to generate code to be assembled by the vendor-supplied
8484 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
8498 Indicate whether GCC should generate code to use the optional BWX,
8499 CIX, and MAX instruction sets. The default is to use the instruction sets
8500 supported by the CPU type specified via @option{-mcpu=} option or that
8501 of the CPU on which GCC was built if none was specified.
8503 @item -mcpu=@var{cpu_type}
8505 Set the instruction set, register set, and instruction scheduling
8506 parameters for machine type @var{cpu_type}. You can specify either the
8507 @samp{EV} style name or the corresponding chip number. GCC
8508 supports scheduling parameters for the EV4 and EV5 family of processors
8509 and will choose the default values for the instruction set from
8510 the processor you specify. If you do not specify a processor type,
8511 GCC will default to the processor on which the compiler was built.
8513 Supported values for @var{cpu_type} are
8518 Schedules as an EV4 and has no instruction set extensions.
8522 Schedules as an EV5 and has no instruction set extensions.
8526 Schedules as an EV5 and supports the BWX extension.
8531 Schedules as an EV5 and supports the BWX and MAX extensions.
8535 Schedules as an EV5 (until Digital releases the scheduling parameters
8536 for the EV6) and supports the BWX, CIX, and MAX extensions.
8539 @item -mmemory-latency=@var{time}
8540 @opindex mmemory-latency
8541 Sets the latency the scheduler should assume for typical memory
8542 references as seen by the application. This number is highly
8543 dependent on the memory access patterns used by the application
8544 and the size of the external cache on the machine.
8546 Valid options for @var{time} are
8550 A decimal number representing clock cycles.
8556 The compiler contains estimates of the number of clock cycles for
8557 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
8558 (also called Dcache, Scache, and Bcache), as well as to main memory.
8559 Note that L3 is only valid for EV5.
8564 @node Clipper Options
8565 @subsection Clipper Options
8567 These @samp{-m} options are defined for the Clipper implementations:
8572 Produce code for a C300 Clipper processor. This is the default.
8576 Produce code for a C400 Clipper processor, i.e.@: use floating point
8580 @node H8/300 Options
8581 @subsection H8/300 Options
8583 These @samp{-m} options are defined for the H8/300 implementations:
8588 Shorten some address references at link time, when possible; uses the
8589 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
8590 ld.info, Using ld}, for a fuller description.
8594 Generate code for the H8/300H@.
8598 Generate code for the H8/S@.
8602 Generate code for the H8/S2600. This switch must be used with @option{-ms}.
8606 Make @code{int} data 32 bits by default.
8610 On the H8/300H and H8/S, use the same alignment rules as for the H8/300.
8611 The default for the H8/300H and H8/S is to align longs and floats on 4
8613 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
8614 This option has no effect on the H8/300.
8618 @subsection SH Options
8620 These @samp{-m} options are defined for the SH implementations:
8625 Generate code for the SH1.
8629 Generate code for the SH2.
8633 Generate code for the SH3.
8637 Generate code for the SH3e.
8641 Generate code for the SH4 without a floating-point unit.
8643 @item -m4-single-only
8644 @opindex m4-single-only
8645 Generate code for the SH4 with a floating-point unit that only
8646 supports single-precision arithmetic.
8650 Generate code for the SH4 assuming the floating-point unit is in
8651 single-precision mode by default.
8655 Generate code for the SH4.
8659 Compile code for the processor in big endian mode.
8663 Compile code for the processor in little endian mode.
8667 Align doubles at 64-bit boundaries. Note that this changes the calling
8668 conventions, and thus some functions from the standard C library will
8669 not work unless you recompile it first with @option{-mdalign}.
8673 Shorten some address references at link time, when possible; uses the
8674 linker option @option{-relax}.
8678 Use 32-bit offsets in @code{switch} tables. The default is to use
8683 Enable the use of the instruction @code{fmovd}.
8687 Comply with the calling conventions defined by Hitachi.
8691 Mark the @code{MAC} register as call-clobbered, even if
8692 @option{-mhitachi} is given.
8696 Increase IEEE-compliance of floating-point code.
8700 Dump instruction size and location in the assembly code.
8704 This option is deprecated. It pads structures to multiple of 4 bytes,
8705 which is incompatible with the SH ABI@.
8709 Optimize for space instead of speed. Implied by @option{-Os}.
8713 When generating position-independent code, emit function calls using
8714 the Global Offset Table instead of the Procedure Linkage Table.
8718 Generate a library function call to invalidate instruction cache
8719 entries, after fixing up a trampoline. This library function call
8720 doesn't assume it can write to the whole memory address space. This
8721 is the default when the target is @code{sh-*-linux*}.
8724 @node System V Options
8725 @subsection Options for System V
8727 These additional options are available on System V Release 4 for
8728 compatibility with other compilers on those systems:
8733 Create a shared object.
8734 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
8738 Identify the versions of each tool used by the compiler, in a
8739 @code{.ident} assembler directive in the output.
8743 Refrain from adding @code{.ident} directives to the output file (this is
8746 @item -YP,@var{dirs}
8748 Search the directories @var{dirs}, and no others, for libraries
8749 specified with @option{-l}.
8753 Look in the directory @var{dir} to find the M4 preprocessor.
8754 The assembler uses this option.
8755 @c This is supposed to go with a -Yd for predefined M4 macro files, but
8756 @c the generic assembler that comes with Solaris takes just -Ym.
8759 @node TMS320C3x/C4x Options
8760 @subsection TMS320C3x/C4x Options
8761 @cindex TMS320C3x/C4x Options
8763 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
8767 @item -mcpu=@var{cpu_type}
8769 Set the instruction set, register set, and instruction scheduling
8770 parameters for machine type @var{cpu_type}. Supported values for
8771 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
8772 @samp{c44}. The default is @samp{c40} to generate code for the
8777 @itemx -msmall-memory
8779 @opindex mbig-memory
8781 @opindex msmall-memory
8783 Generates code for the big or small memory model. The small memory
8784 model assumed that all data fits into one 64K word page. At run-time
8785 the data page (DP) register must be set to point to the 64K page
8786 containing the .bss and .data program sections. The big memory model is
8787 the default and requires reloading of the DP register for every direct
8794 Allow (disallow) allocation of general integer operands into the block
8801 Enable (disable) generation of code using decrement and branch,
8802 DBcond(D), instructions. This is enabled by default for the C4x. To be
8803 on the safe side, this is disabled for the C3x, since the maximum
8804 iteration count on the C3x is @math{2^23 + 1} (but who iterates loops more than
8805 @math{2^23} times on the C3x?). Note that GCC will try to reverse a loop so
8806 that it can utilise the decrement and branch instruction, but will give
8807 up if there is more than one memory reference in the loop. Thus a loop
8808 where the loop counter is decremented can generate slightly more
8809 efficient code, in cases where the RPTB instruction cannot be utilised.
8811 @item -mdp-isr-reload
8813 @opindex mdp-isr-reload
8815 Force the DP register to be saved on entry to an interrupt service
8816 routine (ISR), reloaded to point to the data section, and restored on
8817 exit from the ISR@. This should not be required unless someone has
8818 violated the small memory model by modifying the DP register, say within
8825 For the C3x use the 24-bit MPYI instruction for integer multiplies
8826 instead of a library call to guarantee 32-bit results. Note that if one
8827 of the operands is a constant, then the multiplication will be performed
8828 using shifts and adds. If the @option{-mmpyi} option is not specified for the C3x,
8829 then squaring operations are performed inline instead of a library call.
8832 @itemx -mno-fast-fix
8834 @opindex mno-fast-fix
8835 The C3x/C4x FIX instruction to convert a floating point value to an
8836 integer value chooses the nearest integer less than or equal to the
8837 floating point value rather than to the nearest integer. Thus if the
8838 floating point number is negative, the result will be incorrectly
8839 truncated an additional code is necessary to detect and correct this
8840 case. This option can be used to disable generation of the additional
8841 code required to correct the result.
8847 Enable (disable) generation of repeat block sequences using the RPTB
8848 instruction for zero overhead looping. The RPTB construct is only used
8849 for innermost loops that do not call functions or jump across the loop
8850 boundaries. There is no advantage having nested RPTB loops due to the
8851 overhead required to save and restore the RC, RS, and RE registers.
8852 This is enabled by default with @option{-O2}.
8854 @item -mrpts=@var{count}
8858 Enable (disable) the use of the single instruction repeat instruction
8859 RPTS@. If a repeat block contains a single instruction, and the loop
8860 count can be guaranteed to be less than the value @var{count}, GCC will
8861 emit a RPTS instruction instead of a RPTB@. If no value is specified,
8862 then a RPTS will be emitted even if the loop count cannot be determined
8863 at compile time. Note that the repeated instruction following RPTS does
8864 not have to be reloaded from memory each iteration, thus freeing up the
8865 CPU buses for operands. However, since interrupts are blocked by this
8866 instruction, it is disabled by default.
8868 @item -mloop-unsigned
8869 @itemx -mno-loop-unsigned
8870 @opindex mloop-unsigned
8871 @opindex mno-loop-unsigned
8872 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
8873 is @math{2^31 + 1} since these instructions test if the iteration count is
8874 negative to terminate the loop. If the iteration count is unsigned
8875 there is a possibility than the @math{2^31 + 1} maximum iteration count may be
8876 exceeded. This switch allows an unsigned iteration count.
8880 Try to emit an assembler syntax that the TI assembler (asm30) is happy
8881 with. This also enforces compatibility with the API employed by the TI
8882 C3x C compiler. For example, long doubles are passed as structures
8883 rather than in floating point registers.
8889 Generate code that uses registers (stack) for passing arguments to functions.
8890 By default, arguments are passed in registers where possible rather
8891 than by pushing arguments on to the stack.
8893 @item -mparallel-insns
8894 @itemx -mno-parallel-insns
8895 @opindex mparallel-insns
8896 @opindex mno-parallel-insns
8897 Allow the generation of parallel instructions. This is enabled by
8898 default with @option{-O2}.
8900 @item -mparallel-mpy
8901 @itemx -mno-parallel-mpy
8902 @opindex mparallel-mpy
8903 @opindex mno-parallel-mpy
8904 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
8905 provided @option{-mparallel-insns} is also specified. These instructions have
8906 tight register constraints which can pessimize the code generation
8912 @subsection V850 Options
8913 @cindex V850 Options
8915 These @samp{-m} options are defined for V850 implementations:
8919 @itemx -mno-long-calls
8920 @opindex mlong-calls
8921 @opindex mno-long-calls
8922 Treat all calls as being far away (near). If calls are assumed to be
8923 far away, the compiler will always load the functions address up into a
8924 register, and call indirect through the pointer.
8930 Do not optimize (do optimize) basic blocks that use the same index
8931 pointer 4 or more times to copy pointer into the @code{ep} register, and
8932 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
8933 option is on by default if you optimize.
8935 @item -mno-prolog-function
8936 @itemx -mprolog-function
8937 @opindex mno-prolog-function
8938 @opindex mprolog-function
8939 Do not use (do use) external functions to save and restore registers at
8940 the prolog and epilog of a function. The external functions are slower,
8941 but use less code space if more than one function saves the same number
8942 of registers. The @option{-mprolog-function} option is on by default if
8947 Try to make the code as small as possible. At present, this just turns
8948 on the @option{-mep} and @option{-mprolog-function} options.
8952 Put static or global variables whose size is @var{n} bytes or less into
8953 the tiny data area that register @code{ep} points to. The tiny data
8954 area can hold up to 256 bytes in total (128 bytes for byte references).
8958 Put static or global variables whose size is @var{n} bytes or less into
8959 the small data area that register @code{gp} points to. The small data
8960 area can hold up to 64 kilobytes.
8964 Put static or global variables whose size is @var{n} bytes or less into
8965 the first 32 kilobytes of memory.
8969 Specify that the target processor is the V850.
8972 @opindex mbig-switch
8973 Generate code suitable for big switch tables. Use this option only if
8974 the assembler/linker complain about out of range branches within a switch
8979 @subsection ARC Options
8982 These options are defined for ARC implementations:
8987 Compile code for little endian mode. This is the default.
8991 Compile code for big endian mode.
8994 @opindex mmangle-cpu
8995 Prepend the name of the cpu to all public symbol names.
8996 In multiple-processor systems, there are many ARC variants with different
8997 instruction and register set characteristics. This flag prevents code
8998 compiled for one cpu to be linked with code compiled for another.
8999 No facility exists for handling variants that are ``almost identical''.
9000 This is an all or nothing option.
9002 @item -mcpu=@var{cpu}
9004 Compile code for ARC variant @var{cpu}.
9005 Which variants are supported depend on the configuration.
9006 All variants support @option{-mcpu=base}, this is the default.
9008 @item -mtext=@var{text-section}
9009 @itemx -mdata=@var{data-section}
9010 @itemx -mrodata=@var{readonly-data-section}
9014 Put functions, data, and readonly data in @var{text-section},
9015 @var{data-section}, and @var{readonly-data-section} respectively
9016 by default. This can be overridden with the @code{section} attribute.
9017 @xref{Variable Attributes}.
9022 @subsection NS32K Options
9023 @cindex NS32K options
9025 These are the @samp{-m} options defined for the 32000 series. The default
9026 values for these options depends on which style of 32000 was selected when
9027 the compiler was configured; the defaults for the most common choices are
9035 Generate output for a 32032. This is the default
9036 when the compiler is configured for 32032 and 32016 based systems.
9042 Generate output for a 32332. This is the default
9043 when the compiler is configured for 32332-based systems.
9049 Generate output for a 32532. This is the default
9050 when the compiler is configured for 32532-based systems.
9054 Generate output containing 32081 instructions for floating point.
9055 This is the default for all systems.
9059 Generate output containing 32381 instructions for floating point. This
9060 also implies @option{-m32081}. The 32381 is only compatible with the 32332
9061 and 32532 cpus. This is the default for the pc532-netbsd configuration.
9065 Try and generate multiply-add floating point instructions @code{polyF}
9066 and @code{dotF}. This option is only available if the @option{-m32381}
9067 option is in effect. Using these instructions requires changes to
9068 register allocation which generally has a negative impact on
9069 performance. This option should only be enabled when compiling code
9070 particularly likely to make heavy use of multiply-add instructions.
9073 @opindex mnomulti-add
9074 Do not try and generate multiply-add floating point instructions
9075 @code{polyF} and @code{dotF}. This is the default on all platforms.
9078 @opindex msoft-float
9079 Generate output containing library calls for floating point.
9080 @strong{Warning:} the requisite libraries may not be available.
9083 @opindex mnobitfield
9084 Do not use the bit-field instructions. On some machines it is faster to
9085 use shifting and masking operations. This is the default for the pc532.
9089 Do use the bit-field instructions. This is the default for all platforms
9094 Use a different function-calling convention, in which functions
9095 that take a fixed number of arguments return pop their
9096 arguments on return with the @code{ret} instruction.
9098 This calling convention is incompatible with the one normally
9099 used on Unix, so you cannot use it if you need to call libraries
9100 compiled with the Unix compiler.
9102 Also, you must provide function prototypes for all functions that
9103 take variable numbers of arguments (including @code{printf});
9104 otherwise incorrect code will be generated for calls to those
9107 In addition, seriously incorrect code will result if you call a
9108 function with too many arguments. (Normally, extra arguments are
9109 harmlessly ignored.)
9111 This option takes its name from the 680x0 @code{rtd} instruction.
9116 Use a different function-calling convention where the first two arguments
9117 are passed in registers.
9119 This calling convention is incompatible with the one normally
9120 used on Unix, so you cannot use it if you need to call libraries
9121 compiled with the Unix compiler.
9124 @opindex mnoregparam
9125 Do not pass any arguments in registers. This is the default for all
9130 It is OK to use the sb as an index register which is always loaded with
9131 zero. This is the default for the pc532-netbsd target.
9135 The sb register is not available for use or has not been initialized to
9136 zero by the run time system. This is the default for all targets except
9137 the pc532-netbsd. It is also implied whenever @option{-mhimem} or
9138 @option{-fpic} is set.
9142 Many ns32000 series addressing modes use displacements of up to 512MB@.
9143 If an address is above 512MB then displacements from zero can not be used.
9144 This option causes code to be generated which can be loaded above 512MB@.
9145 This may be useful for operating systems or ROM code.
9149 Assume code will be loaded in the first 512MB of virtual address space.
9150 This is the default for all platforms.
9156 @subsection AVR Options
9159 These options are defined for AVR implementations:
9162 @item -mmcu=@var{mcu}
9164 Specify ATMEL AVR instruction set or MCU type.
9166 Instruction set avr1 is for the minimal AVR core, not supported by the C
9167 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
9168 attiny11, attiny12, attiny15, attiny28).
9170 Instruction set avr2 (default) is for the classic AVR core with up to
9171 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
9172 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
9173 at90c8534, at90s8535).
9175 Instruction set avr3 is for the classic AVR core with up to 128K program
9176 memory space (MCU types: atmega103, atmega603).
9178 Instruction set avr4 is for the enhanced AVR core with up to 8K program
9179 memory space (MCU types: atmega83, atmega85).
9181 Instruction set avr5 is for the enhanced AVR core with up to 128K program
9182 memory space (MCU types: atmega161, atmega163, atmega32, at94k).
9186 Output instruction sizes to the asm file.
9188 @item -minit-stack=@var{N}
9189 @opindex minit-stack
9190 Specify the initial stack address, which may be a symbol or numeric value,
9191 @samp{__stack} is the default.
9193 @item -mno-interrupts
9194 @opindex mno-interrupts
9195 Generated code is not compatible with hardware interrupts.
9196 Code size will be smaller.
9198 @item -mcall-prologues
9199 @opindex mcall-prologues
9200 Functions prologues/epilogues expanded as call to appropriate
9201 subroutines. Code size will be smaller.
9203 @item -mno-tablejump
9204 @opindex mno-tablejump
9205 Do not generate tablejump insns which sometimes increase code size.
9208 @opindex mtiny-stack
9209 Change only the low 8 bits of the stack pointer.
9213 @subsection MCore Options
9214 @cindex MCore options
9216 These are the @samp{-m} options defined for the Motorola M*Core
9226 @opindex mno-hardlit
9227 Inline constants into the code stream if it can be done in two
9228 instructions or less.
9236 Use the divide instruction. (Enabled by default).
9238 @item -mrelax-immediate
9239 @itemx -mrelax-immediate
9240 @itemx -mno-relax-immediate
9241 @opindex mrelax-immediate
9242 @opindex mrelax-immediate
9243 @opindex mno-relax-immediate
9244 Allow arbitrary sized immediates in bit operations.
9246 @item -mwide-bitfields
9247 @itemx -mwide-bitfields
9248 @itemx -mno-wide-bitfields
9249 @opindex mwide-bitfields
9250 @opindex mwide-bitfields
9251 @opindex mno-wide-bitfields
9252 Always treat bit-fields as int-sized.
9254 @item -m4byte-functions
9255 @itemx -m4byte-functions
9256 @itemx -mno-4byte-functions
9257 @opindex m4byte-functions
9258 @opindex m4byte-functions
9259 @opindex mno-4byte-functions
9260 Force all functions to be aligned to a four byte boundary.
9262 @item -mcallgraph-data
9263 @itemx -mcallgraph-data
9264 @itemx -mno-callgraph-data
9265 @opindex mcallgraph-data
9266 @opindex mcallgraph-data
9267 @opindex mno-callgraph-data
9268 Emit callgraph information.
9272 @itemx -mno-slow-bytes
9273 @opindex mslow-bytes
9274 @opindex mslow-bytes
9275 @opindex mno-slow-bytes
9276 Prefer word access when reading byte quantities.
9278 @item -mlittle-endian
9279 @itemx -mlittle-endian
9281 @opindex mlittle-endian
9282 @opindex mlittle-endian
9283 @opindex mbig-endian
9284 Generate code for a little endian target.
9292 Generate code for the 210 processor.
9296 @subsection IA-64 Options
9297 @cindex IA-64 Options
9299 These are the @samp{-m} options defined for the Intel IA-64 architecture.
9303 @opindex mbig-endian
9304 Generate code for a big endian target. This is the default for HPUX@.
9306 @item -mlittle-endian
9307 @opindex mlittle-endian
9308 Generate code for a little endian target. This is the default for AIX5
9315 Generate (or don't) code for the GNU assembler. This is the default.
9316 @c Also, this is the default if the configure option @option{--with-gnu-as}
9323 Generate (or don't) code for the GNU linker. This is the default.
9324 @c Also, this is the default if the configure option @option{--with-gnu-ld}
9329 Generate code that does not use a global pointer register. The result
9330 is not position independent code, and violates the IA-64 ABI@.
9332 @item -mvolatile-asm-stop
9333 @itemx -mno-volatile-asm-stop
9334 @opindex mvolatile-asm-stop
9335 @opindex mno-volatile-asm-stop
9336 Generate (or don't) a stop bit immediately before and after volatile asm
9341 Generate code that works around Itanium B step errata.
9343 @item -mregister-names
9344 @itemx -mno-register-names
9345 @opindex mregister-names
9346 @opindex mno-register-names
9347 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
9348 the stacked registers. This may make assembler output more readable.
9354 Disable (or enable) optimizations that use the small data section. This may
9355 be useful for working around optimizer bugs.
9358 @opindex mconstant-gp
9359 Generate code that uses a single constant global pointer value. This is
9360 useful when compiling kernel code.
9364 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
9365 This is useful when compiling firmware code.
9367 @item -minline-divide-min-latency
9368 @opindex minline-divide-min-latency
9369 Generate code for inline divides using the minimum latency algorithm.
9371 @item -minline-divide-max-throughput
9372 @opindex minline-divide-max-throughput
9373 Generate code for inline divides using the maximum throughput algorithm.
9375 @item -mno-dwarf2-asm
9377 @opindex mno-dwarf2-asm
9378 @opindex mdwarf2-asm
9379 Don't (or do) generate assembler code for the DWARF2 line number debugging
9380 info. This may be useful when not using the GNU assembler.
9382 @item -mfixed-range=@var{register-range}
9383 @opindex mfixed-range
9384 Generate code treating the given register range as fixed registers.
9385 A fixed register is one that the register allocator can not use. This is
9386 useful when compiling kernel code. A register range is specified as
9387 two registers separated by a dash. Multiple register ranges can be
9388 specified separated by a comma.
9392 @subsection D30V Options
9393 @cindex D30V Options
9395 These @samp{-m} options are defined for D30V implementations:
9400 Link the @samp{.text}, @samp{.data}, @samp{.bss}, @samp{.strings},
9401 @samp{.rodata}, @samp{.rodata1}, @samp{.data1} sections into external
9402 memory, which starts at location @code{0x80000000}.
9406 Same as the @option{-mextmem} switch.
9410 Link the @samp{.text} section into onchip text memory, which starts at
9411 location @code{0x0}. Also link @samp{.data}, @samp{.bss},
9412 @samp{.strings}, @samp{.rodata}, @samp{.rodata1}, @samp{.data1} sections
9413 into onchip data memory, which starts at location @code{0x20000000}.
9415 @item -mno-asm-optimize
9416 @itemx -masm-optimize
9417 @opindex mno-asm-optimize
9418 @opindex masm-optimize
9419 Disable (enable) passing @option{-O} to the assembler when optimizing.
9420 The assembler uses the @option{-O} option to automatically parallelize
9421 adjacent short instructions where possible.
9423 @item -mbranch-cost=@var{n}
9424 @opindex mbranch-cost
9425 Increase the internal costs of branches to @var{n}. Higher costs means
9426 that the compiler will issue more instructions to avoid doing a branch.
9429 @item -mcond-exec=@var{n}
9431 Specify the maximum number of conditionally executed instructions that
9432 replace a branch. The default is 4.
9435 @node S/390 and zSeries Options
9436 @subsection S/390 and zSeries Options
9437 @cindex S/390 and zSeries Options
9439 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
9444 @opindex mhard-float
9445 @opindex msoft-float
9446 Use (do not use) the hardware floating-point instructions and registers
9447 for floating-point operations. When @option{-msoft-float} is specified,
9448 functions in @file{libgcc.a} will be used to perform floating-point
9449 operations. When @option{-mhard-float} is specified, the compiler
9450 generates IEEE floating-point instructions. This is the default.
9453 @itemx -mno-backchain
9455 @opindex mno-backchain
9456 Generate (or do not generate) code which maintains an explicit
9457 backchain within the stack frame that points to the caller's frame.
9458 This is currently needed to allow debugging. The default is to
9459 generate the backchain.
9462 @itemx -mno-small-exec
9463 @opindex msmall-exec
9464 @opindex mno-small-exec
9465 Generate (or do not generate) code using the @code{bras} instruction
9466 to do subroutine calls.
9467 This only works reliably if the total executable size does not
9468 exceed 64k. The default is to use the @code{basr} instruction instead,
9469 which does not have this limitation.
9475 When @option{-m31} is specified, generate code compliant to the
9476 Linux for S/390 ABI@. When @option{-m64} is specified, generate
9477 code compliant to the Linux for zSeries ABI@. This allows GCC in
9478 particular to generate 64-bit instructions. For the @samp{s390}
9479 targets, the default is @option{-m31}, while the @samp{s390x}
9480 targets default to @option{-m64}.
9486 Generate (or do not generate) code using the @code{mvcle} instruction
9487 to perform block moves. When @option{-mno-mvcle} is specifed,
9488 use a @code{mvc} loop instead. This is the default.
9494 Print (or do not print) additional debug information when compiling.
9495 The default is to not print debug information.
9500 @subsection CRIS Options
9501 @cindex CRIS Options
9503 These options are defined specifically for the CRIS ports.
9506 @item -march=@var{architecture-type}
9507 @itemx -mcpu=@var{architecture-type}
9510 Generate code for the specified architecture. The choices for
9511 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
9512 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX.
9513 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
9516 @item -mtune=@var{architecture-type}
9518 Tune to @var{architecture-type} everything applicable about the generated
9519 code, except for the ABI and the set of available instructions. The
9520 choices for @var{architecture-type} are the same as for
9521 @option{-march=@var{architecture-type}}.
9523 @item -mmax-stack-frame=@var{n}
9524 @opindex mmax-stack-frame
9525 Warn when the stack frame of a function exceeds @var{n} bytes.
9527 @item -melinux-stacksize=@var{n}
9528 @opindex melinux-stacksize
9529 Only available with the @samp{cris-axis-aout} target. Arranges for
9530 indications in the program to the kernel loader that the stack of the
9531 program should be set to @var{n} bytes.
9537 The options @option{-metrax4} and @option{-metrax100} are synonyms for
9538 @option{-march=v3} and @option{-march=v8} respectively.
9542 Enable CRIS-specific verbose debug-related information in the assembly
9543 code. This option also has the effect to turn off the @samp{#NO_APP}
9544 formatted-code indicator to the assembler at the beginning of the
9549 Do not use condition-code results from previous instruction; always emit
9550 compare and test instructions before use of condition codes.
9552 @item -mno-side-effects
9553 @opindex mno-side-effects
9554 Do not emit instructions with side-effects in addressing modes other than
9558 @itemx -mno-stack-align
9560 @itemx -mno-data-align
9561 @itemx -mconst-align
9562 @itemx -mno-const-align
9563 @opindex mstack-align
9564 @opindex mno-stack-align
9565 @opindex mdata-align
9566 @opindex mno-data-align
9567 @opindex mconst-align
9568 @opindex mno-const-align
9569 These options (no-options) arranges (eliminate arrangements) for the
9570 stack-frame, individual data and constants to be aligned for the maximum
9571 single data access size for the chosen CPU model. The default is to
9572 arrange for 32-bit alignment. ABI details such as structure layout are
9573 not affected by these options.
9581 Similar to the stack- data- and const-align options above, these options
9582 arrange for stack-frame, writable data and constants to all be 32-bit,
9583 16-bit or 8-bit aligned. The default is 32-bit alignment.
9585 @item -mno-prologue-epilogue
9586 @itemx -mprologue-epilogue
9587 @opindex mno-prologue-epilogue
9588 @opindex mprologue-epilogue
9589 With @option{-mno-prologue-epilogue}, the normal function prologue and
9590 epilogue that sets up the stack-frame are omitted and no return
9591 instructions or return sequences are generated in the code. Use this
9592 option only together with visual inspection of the compiled code: no
9593 warnings or errors are generated when call-saved registers must be saved,
9594 or storage for local variable needs to be allocated.
9600 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
9601 instruction sequences that load addresses for functions from the PLT part
9602 of the GOT rather than (traditional on other architectures) calls to the
9603 PLT. The default is @option{-mgotplt}.
9607 Legacy no-op option only recognized with the cris-axis-aout target.
9611 Legacy no-op option only recognized with the cris-axis-elf and
9612 cris-axis-linux-gnu targets.
9616 Only recognized with the cris-axis-aout target, where it selects a
9617 GNU/linux-like multilib, include files and instruction set for
9622 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
9626 This option, recognized for the cris-axis-aout and cris-axis-elf arranges
9627 to link with input-output functions from a simulator library. Code,
9628 initialized data and zero-initialized data are allocated consecutively.
9632 Like @option{-sim}, but pass linker options to locate initialized data at
9633 0x40000000 and zero-initialized data at 0x80000000.
9637 @node Code Gen Options
9638 @section Options for Code Generation Conventions
9639 @cindex code generation conventions
9640 @cindex options, code generation
9641 @cindex run-time options
9643 These machine-independent options control the interface conventions
9644 used in code generation.
9646 Most of them have both positive and negative forms; the negative form
9647 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
9648 one of the forms is listed---the one which is not the default. You
9649 can figure out the other form by either removing @samp{no-} or adding
9654 @opindex fexceptions
9655 Enable exception handling. Generates extra code needed to propagate
9656 exceptions. For some targets, this implies GCC will generate frame
9657 unwind information for all functions, which can produce significant data
9658 size overhead, although it does not affect execution. If you do not
9659 specify this option, GCC will enable it by default for languages like
9660 C++ which normally require exception handling, and disable it for
9661 languages like C that do not normally require it. However, you may need
9662 to enable this option when compiling C code that needs to interoperate
9663 properly with exception handlers written in C++. You may also wish to
9664 disable this option if you are compiling older C++ programs that don't
9665 use exception handling.
9667 @item -fnon-call-exceptions
9668 @opindex fnon-call-exceptions
9669 Generate code that allows trapping instructions to throw exceptions.
9670 Note that this requires platform-specific runtime support that does
9671 not exist everywhere. Moreover, it only allows @emph{trapping}
9672 instructions to throw exceptions, i.e.@: memory references or floating
9673 point instructions. It does not allow exceptions to be thrown from
9674 arbitrary signal handlers such as @code{SIGALRM}.
9676 @item -funwind-tables
9677 @opindex funwind-tables
9678 Similar to @option{-fexceptions}, except that it will just generate any needed
9679 static data, but will not affect the generated code in any other way.
9680 You will normally not enable this option; instead, a language processor
9681 that needs this handling would enable it on your behalf.
9683 @item -fasynchronous-unwind-tables
9684 @opindex funwind-tables
9685 Generate unwind table in dwarf2 format, if supported by target machine. The
9686 table is exact at each instruction boundary, so it can be used for stack
9687 unwinding from asynchronous events (such as debugger or garbage collector).
9689 @item -fpcc-struct-return
9690 @opindex fpcc-struct-return
9691 Return ``short'' @code{struct} and @code{union} values in memory like
9692 longer ones, rather than in registers. This convention is less
9693 efficient, but it has the advantage of allowing intercallability between
9694 GCC-compiled files and files compiled with other compilers.
9696 The precise convention for returning structures in memory depends
9697 on the target configuration macros.
9699 Short structures and unions are those whose size and alignment match
9700 that of some integer type.
9702 @item -freg-struct-return
9703 @opindex freg-struct-return
9704 Return @code{struct} and @code{union} values in registers when possible.
9705 This is more efficient for small structures than
9706 @option{-fpcc-struct-return}.
9708 If you specify neither @option{-fpcc-struct-return} nor
9709 @option{-freg-struct-return}, GCC defaults to whichever convention is
9710 standard for the target. If there is no standard convention, GCC
9711 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
9712 the principal compiler. In those cases, we can choose the standard, and
9713 we chose the more efficient register return alternative.
9716 @opindex fshort-enums
9717 Allocate to an @code{enum} type only as many bytes as it needs for the
9718 declared range of possible values. Specifically, the @code{enum} type
9719 will be equivalent to the smallest integer type which has enough room.
9721 @item -fshort-double
9722 @opindex fshort-double
9723 Use the same size for @code{double} as for @code{float}.
9726 @opindex fshared-data
9727 Requests that the data and non-@code{const} variables of this
9728 compilation be shared data rather than private data. The distinction
9729 makes sense only on certain operating systems, where shared data is
9730 shared between processes running the same program, while private data
9731 exists in one copy per process.
9735 In C, allocate even uninitialized global variables in the data section of the
9736 object file, rather than generating them as common blocks. This has the
9737 effect that if the same variable is declared (without @code{extern}) in
9738 two different compilations, you will get an error when you link them.
9739 The only reason this might be useful is if you wish to verify that the
9740 program will work on other systems which always work this way.
9744 Ignore the @samp{#ident} directive.
9746 @item -fno-gnu-linker
9747 @opindex fno-gnu-linker
9748 Do not output global initializations (such as C++ constructors and
9749 destructors) in the form used by the GNU linker (on systems where the GNU
9750 linker is the standard method of handling them). Use this option when
9751 you want to use a non-GNU linker, which also requires using the
9752 @command{collect2} program to make sure the system linker includes
9753 constructors and destructors. (@command{collect2} is included in the GCC
9754 distribution.) For systems which @emph{must} use @command{collect2}, the
9755 compiler driver @command{gcc} is configured to do this automatically.
9757 @item -finhibit-size-directive
9758 @opindex finhibit-size-directive
9759 Don't output a @code{.size} assembler directive, or anything else that
9760 would cause trouble if the function is split in the middle, and the
9761 two halves are placed at locations far apart in memory. This option is
9762 used when compiling @file{crtstuff.c}; you should not need to use it
9766 @opindex fverbose-asm
9767 Put extra commentary information in the generated assembly code to
9768 make it more readable. This option is generally only of use to those
9769 who actually need to read the generated assembly code (perhaps while
9770 debugging the compiler itself).
9772 @option{-fno-verbose-asm}, the default, causes the
9773 extra information to be omitted and is useful when comparing two assembler
9778 Consider all memory references through pointers to be volatile.
9780 @item -fvolatile-global
9781 @opindex fvolatile-global
9782 Consider all memory references to extern and global data items to
9783 be volatile. GCC does not consider static data items to be volatile
9784 because of this switch.
9786 @item -fvolatile-static
9787 @opindex fvolatile-static
9788 Consider all memory references to static data to be volatile.
9792 @cindex global offset table
9794 Generate position-independent code (PIC) suitable for use in a shared
9795 library, if supported for the target machine. Such code accesses all
9796 constant addresses through a global offset table (GOT)@. The dynamic
9797 loader resolves the GOT entries when the program starts (the dynamic
9798 loader is not part of GCC; it is part of the operating system). If
9799 the GOT size for the linked executable exceeds a machine-specific
9800 maximum size, you get an error message from the linker indicating that
9801 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
9802 instead. (These maximums are 16k on the m88k, 8k on the Sparc, and 32k
9803 on the m68k and RS/6000. The 386 has no such limit.)
9805 Position-independent code requires special support, and therefore works
9806 only on certain machines. For the 386, GCC supports PIC for System V
9807 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
9808 position-independent.
9812 If supported for the target machine, emit position-independent code,
9813 suitable for dynamic linking and avoiding any limit on the size of the
9814 global offset table. This option makes a difference on the m68k, m88k,
9817 Position-independent code requires special support, and therefore works
9818 only on certain machines.
9820 @item -ffixed-@var{reg}
9822 Treat the register named @var{reg} as a fixed register; generated code
9823 should never refer to it (except perhaps as a stack pointer, frame
9824 pointer or in some other fixed role).
9826 @var{reg} must be the name of a register. The register names accepted
9827 are machine-specific and are defined in the @code{REGISTER_NAMES}
9828 macro in the machine description macro file.
9830 This flag does not have a negative form, because it specifies a
9833 @item -fcall-used-@var{reg}
9835 Treat the register named @var{reg} as an allocable register that is
9836 clobbered by function calls. It may be allocated for temporaries or
9837 variables that do not live across a call. Functions compiled this way
9838 will not save and restore the register @var{reg}.
9840 It is an error to used this flag with the frame pointer or stack pointer.
9841 Use of this flag for other registers that have fixed pervasive roles in
9842 the machine's execution model will produce disastrous results.
9844 This flag does not have a negative form, because it specifies a
9847 @item -fcall-saved-@var{reg}
9848 @opindex fcall-saved
9849 Treat the register named @var{reg} as an allocable register saved by
9850 functions. It may be allocated even for temporaries or variables that
9851 live across a call. Functions compiled this way will save and restore
9852 the register @var{reg} if they use it.
9854 It is an error to used this flag with the frame pointer or stack pointer.
9855 Use of this flag for other registers that have fixed pervasive roles in
9856 the machine's execution model will produce disastrous results.
9858 A different sort of disaster will result from the use of this flag for
9859 a register in which function values may be returned.
9861 This flag does not have a negative form, because it specifies a
9865 @opindex fpack-struct
9866 Pack all structure members together without holes. Usually you would
9867 not want to use this option, since it makes the code suboptimal, and
9868 the offsets of structure members won't agree with system libraries.
9870 @item -fcheck-memory-usage
9871 @opindex fcheck-memory-usage
9872 Generate extra code to check each memory access. GCC will generate
9873 code that is suitable for a detector of bad memory accesses such as
9876 Normally, you should compile all, or none, of your code with this option.
9878 If you do mix code compiled with and without this option,
9879 you must ensure that all code that has side effects
9880 and that is called by code compiled with this option
9881 is, itself, compiled with this option.
9882 If you do not, you might get erroneous messages from the detector.
9884 If you use functions from a library that have side-effects (such as
9885 @code{read}), you might not be able to recompile the library and
9886 specify this option. In that case, you can enable the
9887 @option{-fprefix-function-name} option, which requests GCC to encapsulate
9888 your code and make other functions look as if they were compiled with
9889 @option{-fcheck-memory-usage}. This is done by calling ``stubs'',
9890 which are provided by the detector. If you cannot find or build
9891 stubs for every function you call, you might have to specify
9892 @option{-fcheck-memory-usage} without @option{-fprefix-function-name}.
9894 If you specify this option, you can not use the @code{asm} or
9895 @code{__asm__} keywords in functions with memory checking enabled. GCC
9896 cannot understand what the @code{asm} statement may do, and therefore
9897 cannot generate the appropriate code, so it will reject it. However, if
9898 you specify the function attribute @code{no_check_memory_usage}
9899 (@pxref{Function Attributes}), GCC will disable memory checking within a
9900 function; you may use @code{asm} statements inside such functions. You
9901 may have an inline expansion of a non-checked function within a checked
9902 function; in that case GCC will not generate checks for the inlined
9903 function's memory accesses.
9905 If you move your @code{asm} statements to non-checked inline functions
9906 and they do access memory, you can add calls to the support code in your
9907 inline function, to indicate any reads, writes, or copies being done.
9908 These calls would be similar to those done in the stubs described above.
9910 @item -fprefix-function-name
9911 @opindex fprefix-function-name
9912 Request GCC to add a prefix to the symbols generated for function names.
9913 GCC adds a prefix to the names of functions defined as well as
9914 functions called. Code compiled with this option and code compiled
9915 without the option can't be linked together, unless stubs are used.
9917 If you compile the following code with @option{-fprefix-function-name}
9919 extern void bar (int);
9928 GCC will compile the code as if it was written:
9930 extern void prefix_bar (int);
9934 return prefix_bar (a + 5);
9937 This option is designed to be used with @option{-fcheck-memory-usage}.
9939 @item -finstrument-functions
9940 @opindex finstrument-functions
9941 Generate instrumentation calls for entry and exit to functions. Just
9942 after function entry and just before function exit, the following
9943 profiling functions will be called with the address of the current
9944 function and its call site. (On some platforms,
9945 @code{__builtin_return_address} does not work beyond the current
9946 function, so the call site information may not be available to the
9947 profiling functions otherwise.)
9950 void __cyg_profile_func_enter (void *this_fn,
9952 void __cyg_profile_func_exit (void *this_fn,
9956 The first argument is the address of the start of the current function,
9957 which may be looked up exactly in the symbol table.
9959 This instrumentation is also done for functions expanded inline in other
9960 functions. The profiling calls will indicate where, conceptually, the
9961 inline function is entered and exited. This means that addressable
9962 versions of such functions must be available. If all your uses of a
9963 function are expanded inline, this may mean an additional expansion of
9964 code size. If you use @samp{extern inline} in your C code, an
9965 addressable version of such functions must be provided. (This is
9966 normally the case anyways, but if you get lucky and the optimizer always
9967 expands the functions inline, you might have gotten away without
9968 providing static copies.)
9970 A function may be given the attribute @code{no_instrument_function}, in
9971 which case this instrumentation will not be done. This can be used, for
9972 example, for the profiling functions listed above, high-priority
9973 interrupt routines, and any functions from which the profiling functions
9974 cannot safely be called (perhaps signal handlers, if the profiling
9975 routines generate output or allocate memory).
9978 @opindex fstack-check
9979 Generate code to verify that you do not go beyond the boundary of the
9980 stack. You should specify this flag if you are running in an
9981 environment with multiple threads, but only rarely need to specify it in
9982 a single-threaded environment since stack overflow is automatically
9983 detected on nearly all systems if there is only one stack.
9985 Note that this switch does not actually cause checking to be done; the
9986 operating system must do that. The switch causes generation of code
9987 to ensure that the operating system sees the stack being extended.
9989 @item -fstack-limit-register=@var{reg}
9990 @itemx -fstack-limit-symbol=@var{sym}
9991 @itemx -fno-stack-limit
9992 @opindex fstack-limit-register
9993 @opindex fstack-limit-symbol
9994 @opindex fno-stack-limit
9995 Generate code to ensure that the stack does not grow beyond a certain value,
9996 either the value of a register or the address of a symbol. If the stack
9997 would grow beyond the value, a signal is raised. For most targets,
9998 the signal is raised before the stack overruns the boundary, so
9999 it is possible to catch the signal without taking special precautions.
10001 For instance, if the stack starts at absolute address @samp{0x80000000}
10002 and grows downwards, you can use the flags
10003 @option{-fstack-limit-symbol=__stack_limit} and
10004 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
10005 of 128KB@. Note that this may only work with the GNU linker.
10007 @cindex aliasing of parameters
10008 @cindex parameters, aliased
10009 @item -fargument-alias
10010 @itemx -fargument-noalias
10011 @itemx -fargument-noalias-global
10012 @opindex fargument-alias
10013 @opindex fargument-noalias
10014 @opindex fargument-noalias-global
10015 Specify the possible relationships among parameters and between
10016 parameters and global data.
10018 @option{-fargument-alias} specifies that arguments (parameters) may
10019 alias each other and may alias global storage.@*
10020 @option{-fargument-noalias} specifies that arguments do not alias
10021 each other, but may alias global storage.@*
10022 @option{-fargument-noalias-global} specifies that arguments do not
10023 alias each other and do not alias global storage.
10025 Each language will automatically use whatever option is required by
10026 the language standard. You should not need to use these options yourself.
10028 @item -fleading-underscore
10029 @opindex fleading-underscore
10030 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
10031 change the way C symbols are represented in the object file. One use
10032 is to help link with legacy assembly code.
10034 Be warned that you should know what you are doing when invoking this
10035 option, and that not all targets provide complete support for it.
10040 @node Environment Variables
10041 @section Environment Variables Affecting GCC
10042 @cindex environment variables
10044 @c man begin ENVIRONMENT
10046 This section describes several environment variables that affect how GCC
10047 operates. Some of them work by specifying directories or prefixes to use
10048 when searching for various kinds of files. Some are used to specify other
10049 aspects of the compilation environment.
10052 Note that you can also specify places to search using options such as
10053 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
10054 take precedence over places specified using environment variables, which
10055 in turn take precedence over those specified by the configuration of GCC@.
10059 Note that you can also specify places to search using options such as
10060 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
10061 take precedence over places specified using environment variables, which
10062 in turn take precedence over those specified by the configuration of GCC@.
10069 @c @itemx LC_COLLATE
10071 @c @itemx LC_MONETARY
10072 @c @itemx LC_NUMERIC
10077 @c @findex LC_COLLATE
10078 @findex LC_MESSAGES
10079 @c @findex LC_MONETARY
10080 @c @findex LC_NUMERIC
10084 These environment variables control the way that GCC uses
10085 localization information that allow GCC to work with different
10086 national conventions. GCC inspects the locale categories
10087 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
10088 so. These locale categories can be set to any value supported by your
10089 installation. A typical value is @samp{en_UK} for English in the United
10092 The @env{LC_CTYPE} environment variable specifies character
10093 classification. GCC uses it to determine the character boundaries in
10094 a string; this is needed for some multibyte encodings that contain quote
10095 and escape characters that would otherwise be interpreted as a string
10098 The @env{LC_MESSAGES} environment variable specifies the language to
10099 use in diagnostic messages.
10101 If the @env{LC_ALL} environment variable is set, it overrides the value
10102 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
10103 and @env{LC_MESSAGES} default to the value of the @env{LANG}
10104 environment variable. If none of these variables are set, GCC
10105 defaults to traditional C English behavior.
10109 If @env{TMPDIR} is set, it specifies the directory to use for temporary
10110 files. GCC uses temporary files to hold the output of one stage of
10111 compilation which is to be used as input to the next stage: for example,
10112 the output of the preprocessor, which is the input to the compiler
10115 @item GCC_EXEC_PREFIX
10116 @findex GCC_EXEC_PREFIX
10117 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
10118 names of the subprograms executed by the compiler. No slash is added
10119 when this prefix is combined with the name of a subprogram, but you can
10120 specify a prefix that ends with a slash if you wish.
10122 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
10123 an appropriate prefix to use based on the pathname it was invoked with.
10125 If GCC cannot find the subprogram using the specified prefix, it
10126 tries looking in the usual places for the subprogram.
10128 The default value of @env{GCC_EXEC_PREFIX} is
10129 @file{@var{prefix}/lib/gcc-lib/} where @var{prefix} is the value
10130 of @code{prefix} when you ran the @file{configure} script.
10132 Other prefixes specified with @option{-B} take precedence over this prefix.
10134 This prefix is also used for finding files such as @file{crt0.o} that are
10137 In addition, the prefix is used in an unusual way in finding the
10138 directories to search for header files. For each of the standard
10139 directories whose name normally begins with @samp{/usr/local/lib/gcc-lib}
10140 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
10141 replacing that beginning with the specified prefix to produce an
10142 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
10143 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
10144 These alternate directories are searched first; the standard directories
10147 @item COMPILER_PATH
10148 @findex COMPILER_PATH
10149 The value of @env{COMPILER_PATH} is a colon-separated list of
10150 directories, much like @env{PATH}. GCC tries the directories thus
10151 specified when searching for subprograms, if it can't find the
10152 subprograms using @env{GCC_EXEC_PREFIX}.
10155 @findex LIBRARY_PATH
10156 The value of @env{LIBRARY_PATH} is a colon-separated list of
10157 directories, much like @env{PATH}. When configured as a native compiler,
10158 GCC tries the directories thus specified when searching for special
10159 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
10160 using GCC also uses these directories when searching for ordinary
10161 libraries for the @option{-l} option (but directories specified with
10162 @option{-L} come first).
10164 @item C_INCLUDE_PATH
10165 @itemx CPLUS_INCLUDE_PATH
10166 @itemx OBJC_INCLUDE_PATH
10167 @findex C_INCLUDE_PATH
10168 @findex CPLUS_INCLUDE_PATH
10169 @findex OBJC_INCLUDE_PATH
10170 @c @itemx OBJCPLUS_INCLUDE_PATH
10171 These environment variables pertain to particular languages. Each
10172 variable's value is a colon-separated list of directories, much like
10173 @env{PATH}. When GCC searches for header files, it tries the
10174 directories listed in the variable for the language you are using, after
10175 the directories specified with @option{-I} but before the standard header
10178 @item DEPENDENCIES_OUTPUT
10179 @findex DEPENDENCIES_OUTPUT
10180 @cindex dependencies for make as output
10181 If this variable is set, its value specifies how to output dependencies
10182 for Make based on the header files processed by the compiler. This
10183 output looks much like the output from the @option{-M} option
10184 (@pxref{Preprocessor Options}), but it goes to a separate file, and is
10185 in addition to the usual results of compilation.
10187 The value of @env{DEPENDENCIES_OUTPUT} can be just a file name, in
10188 which case the Make rules are written to that file, guessing the target
10189 name from the source file name. Or the value can have the form
10190 @samp{@var{file} @var{target}}, in which case the rules are written to
10191 file @var{file} using @var{target} as the target name.
10195 @cindex locale definition
10196 This variable is used to pass locale information to the compiler. One way in
10197 which this information is used is to determine the character set to be used
10198 when character literals, string literals and comments are parsed in C and C++.
10199 When the compiler is configured to allow multibyte characters,
10200 the following values for @env{LANG} are recognized:
10204 Recognize JIS characters.
10206 Recognize SJIS characters.
10208 Recognize EUCJP characters.
10211 If @env{LANG} is not defined, or if it has some other value, then the
10212 compiler will use mblen and mbtowc as defined by the default locale to
10213 recognize and translate multibyte characters.
10218 @node Running Protoize
10219 @section Running Protoize
10221 The program @code{protoize} is an optional part of GCC@. You can use
10222 it to add prototypes to a program, thus converting the program to ISO
10223 C in one respect. The companion program @code{unprotoize} does the
10224 reverse: it removes argument types from any prototypes that are found.
10226 When you run these programs, you must specify a set of source files as
10227 command line arguments. The conversion programs start out by compiling
10228 these files to see what functions they define. The information gathered
10229 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
10231 After scanning comes actual conversion. The specified files are all
10232 eligible to be converted; any files they include (whether sources or
10233 just headers) are eligible as well.
10235 But not all the eligible files are converted. By default,
10236 @code{protoize} and @code{unprotoize} convert only source and header
10237 files in the current directory. You can specify additional directories
10238 whose files should be converted with the @option{-d @var{directory}}
10239 option. You can also specify particular files to exclude with the
10240 @option{-x @var{file}} option. A file is converted if it is eligible, its
10241 directory name matches one of the specified directory names, and its
10242 name within the directory has not been excluded.
10244 Basic conversion with @code{protoize} consists of rewriting most
10245 function definitions and function declarations to specify the types of
10246 the arguments. The only ones not rewritten are those for varargs
10249 @code{protoize} optionally inserts prototype declarations at the
10250 beginning of the source file, to make them available for any calls that
10251 precede the function's definition. Or it can insert prototype
10252 declarations with block scope in the blocks where undeclared functions
10255 Basic conversion with @code{unprotoize} consists of rewriting most
10256 function declarations to remove any argument types, and rewriting
10257 function definitions to the old-style pre-ISO form.
10259 Both conversion programs print a warning for any function declaration or
10260 definition that they can't convert. You can suppress these warnings
10263 The output from @code{protoize} or @code{unprotoize} replaces the
10264 original source file. The original file is renamed to a name ending
10265 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
10266 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
10267 for DOS) file already exists, then the source file is simply discarded.
10269 @code{protoize} and @code{unprotoize} both depend on GCC itself to
10270 scan the program and collect information about the functions it uses.
10271 So neither of these programs will work until GCC is installed.
10273 Here is a table of the options you can use with @code{protoize} and
10274 @code{unprotoize}. Each option works with both programs unless
10278 @item -B @var{directory}
10279 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
10280 usual directory (normally @file{/usr/local/lib}). This file contains
10281 prototype information about standard system functions. This option
10282 applies only to @code{protoize}.
10284 @item -c @var{compilation-options}
10285 Use @var{compilation-options} as the options when running @code{gcc} to
10286 produce the @samp{.X} files. The special option @option{-aux-info} is
10287 always passed in addition, to tell @code{gcc} to write a @samp{.X} file.
10289 Note that the compilation options must be given as a single argument to
10290 @code{protoize} or @code{unprotoize}. If you want to specify several
10291 @code{gcc} options, you must quote the entire set of compilation options
10292 to make them a single word in the shell.
10294 There are certain @code{gcc} arguments that you cannot use, because they
10295 would produce the wrong kind of output. These include @option{-g},
10296 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
10297 the @var{compilation-options}, they are ignored.
10300 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
10301 systems) instead of @samp{.c}. This is convenient if you are converting
10302 a C program to C++. This option applies only to @code{protoize}.
10305 Add explicit global declarations. This means inserting explicit
10306 declarations at the beginning of each source file for each function
10307 that is called in the file and was not declared. These declarations
10308 precede the first function definition that contains a call to an
10309 undeclared function. This option applies only to @code{protoize}.
10311 @item -i @var{string}
10312 Indent old-style parameter declarations with the string @var{string}.
10313 This option applies only to @code{protoize}.
10315 @code{unprotoize} converts prototyped function definitions to old-style
10316 function definitions, where the arguments are declared between the
10317 argument list and the initial @samp{@{}. By default, @code{unprotoize}
10318 uses five spaces as the indentation. If you want to indent with just
10319 one space instead, use @option{-i " "}.
10322 Keep the @samp{.X} files. Normally, they are deleted after conversion
10326 Add explicit local declarations. @code{protoize} with @option{-l} inserts
10327 a prototype declaration for each function in each block which calls the
10328 function without any declaration. This option applies only to
10332 Make no real changes. This mode just prints information about the conversions
10333 that would have been done without @option{-n}.
10336 Make no @samp{.save} files. The original files are simply deleted.
10337 Use this option with caution.
10339 @item -p @var{program}
10340 Use the program @var{program} as the compiler. Normally, the name
10341 @file{gcc} is used.
10344 Work quietly. Most warnings are suppressed.
10347 Print the version number, just like @option{-v} for @code{gcc}.
10350 If you need special compiler options to compile one of your program's
10351 source files, then you should generate that file's @samp{.X} file
10352 specially, by running @code{gcc} on that source file with the
10353 appropriate options and the option @option{-aux-info}. Then run
10354 @code{protoize} on the entire set of files. @code{protoize} will use
10355 the existing @samp{.X} file because it is newer than the source file.
10359 gcc -Dfoo=bar file1.c -aux-info file1.X
10364 You need to include the special files along with the rest in the
10365 @code{protoize} command, even though their @samp{.X} files already
10366 exist, because otherwise they won't get converted.
10368 @xref{Protoize Caveats}, for more information on how to use
10369 @code{protoize} successfully.