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 -mm32rx -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 -maltivec -mno-altivec @gol
428 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
429 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
430 -mnew-mnemonics -mold-mnemonics @gol
431 -mfull-toc -mminimal-toc -mno-fop-in-toc -mno-sum-in-toc @gol
432 -m64 -m32 -mxl-call -mno-xl-call -mthreads -mpe @gol
433 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
434 -mstring -mno-string -mupdate -mno-update @gol
435 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
436 -mstrict-align -mno-strict-align -mrelocatable @gol
437 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
438 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
439 -mcall-aix -mcall-sysv -mcall-netbsd @gol
441 -mprototype -mno-prototype @gol
442 -msim -mmvme -mads -myellowknife -memb -msdata @gol
443 -msdata=@var{opt} -mvxworks -G @var{num}}
447 -mcall-lib-mul -mfp-arg-in-fpregs -mfp-arg-in-gregs @gol
448 -mfull-fp-blocks -mhc-struct-return -min-line-mul @gol
449 -mminimum-fp-blocks -mnohc-struct-return}
453 -mabicalls -march=@var{cpu-type} -mtune=@var{cpu=type} @gol
454 -mcpu=@var{cpu-type} -membedded-data -muninit-const-in-rodata @gol
455 -membedded-pic -mfp32 -mfp64 -mfused-madd -mno-fused-madd @gol
456 -mgas -mgp32 -mgp64 @gol
457 -mgpopt -mhalf-pic -mhard-float -mint64 -mips1 @gol
458 -mips2 -mips3 -mips4 -mlong64 -mlong32 -mlong-calls -mmemcpy @gol
459 -mmips-as -mmips-tfile -mno-abicalls @gol
460 -mno-embedded-data -mno-uninit-const-in-rodata @gol
461 -mno-embedded-pic -mno-gpopt -mno-long-calls @gol
462 -mno-memcpy -mno-mips-tfile -mno-rnames -mno-stats @gol
463 -mrnames -msoft-float @gol
464 -m4650 -msingle-float -mmad @gol
465 -mstats -EL -EB -G @var{num} -nocpp @gol
466 -mabi=32 -mabi=n32 -mabi=64 -mabi=eabi @gol
467 -mfix7000 -mno-crt0 -mflush-func=@var{func} -mno-flush-func}
469 @emph{i386 and x86-64 Options}
471 -mcpu=@var{cpu-type} -march=@var{cpu-type} @gol
472 -mintel-syntax -mieee-fp -mno-fancy-math-387 @gol
473 -mno-fp-ret-in-387 -msoft-float -msvr3-shlib @gol
474 -mno-wide-multiply -mrtd -malign-double @gol
475 -mpreferred-stack-boundary=@var{num} @gol
476 -mmmx -msse -m3dnow @gol
477 -mthreads -mno-align-stringops -minline-all-stringops @gol
478 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
479 -m96bit-long-double -mregparm=@var{num} -momit-leaf-frame-pointer @gol
485 -march=@var{architecture-type} @gol
486 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
487 -mfast-indirect-calls -mgas -mjump-in-delay @gol
488 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
489 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
490 -mno-jump-in-delay -mno-long-load-store @gol
491 -mno-portable-runtime -mno-soft-float @gol
492 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
493 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
494 -mschedule=@var{cpu-type} -mspace-regs}
496 @emph{Intel 960 Options}
498 -m@var{cpu-type} -masm-compat -mclean-linkage @gol
499 -mcode-align -mcomplex-addr -mleaf-procedures @gol
500 -mic-compat -mic2.0-compat -mic3.0-compat @gol
501 -mintel-asm -mno-clean-linkage -mno-code-align @gol
502 -mno-complex-addr -mno-leaf-procedures @gol
503 -mno-old-align -mno-strict-align -mno-tail-call @gol
504 -mnumerics -mold-align -msoft-float -mstrict-align @gol
507 @emph{DEC Alpha Options}
509 -mfp-regs -mno-fp-regs -mno-soft-float -msoft-float @gol
510 -malpha-as -mgas @gol
511 -mieee -mieee-with-inexact -mieee-conformant @gol
512 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
513 -mtrap-precision=@var{mode} -mbuild-constants @gol
514 -mcpu=@var{cpu-type} @gol
515 -mbwx -mno-bwx -mcix -mno-cix -mmax -mno-max @gol
516 -mmemory-latency=@var{time}}
518 @emph{Clipper Options}
522 @emph{H8/300 Options}
524 -mrelax -mh -ms -mint32 -malign-300}
528 -m1 -m2 -m3 -m3e @gol
529 -m4-nofpu -m4-single-only -m4-single -m4 @gol
530 -mb -ml -mdalign -mrelax @gol
531 -mbigtable -mfmovd -mhitachi -mnomacsave @gol
532 -mieee -misize -mpadstruct -mspace @gol
533 -mprefergot -musermode}
535 @emph{System V Options}
537 -Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
542 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
543 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
545 @emph{TMS320C3x/C4x Options}
547 -mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm @gol
548 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload @gol
549 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned @gol
550 -mparallel-insns -mparallel-mpy -mpreserve-float}
554 -mlong-calls -mno-long-calls -mep -mno-ep @gol
555 -mprolog-function -mno-prolog-function -mspace @gol
556 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
561 -m32032 -m32332 -m32532 -m32081 -m32381 @gol
562 -mmult-add -mnomult-add -msoft-float -mrtd -mnortd @gol
563 -mregparam -mnoregparam -msb -mnosb @gol
564 -mbitfield -mnobitfield -mhimem -mnohimem}
568 -mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts @gol
569 -mcall-prologues -mno-tablejump -mtiny-stack}
573 -mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
574 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
575 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
576 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
577 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
581 -mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
582 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
587 -mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
588 -mvolatile-asm-stop -mb-step -mregister-names -mno-sdata @gol
589 -mconstant-gp -mauto-pic -minline-divide-min-latency @gol
590 -minline-divide-max-throughput -mno-dwarf2-asm @gol
591 -mfixed-range=@var{register-range}}
593 @emph{S/390 and zSeries Options}
595 -mhard-float -msoft-float -mbackchain -mno-backchain @gol
596 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
597 -m64 -m31 -mdebug -mno-debug}
601 -mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
602 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
603 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
604 -mstack-align -mdata-align -mconst-align @gol
605 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
606 -melf -maout -melinux -mlinux -sim -sim2}
608 @item Code Generation Options
609 @xref{Code Gen Options,,Options for Code Generation Conventions}.
611 -fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
612 -ffixed-@var{reg} -fexceptions @gol
613 -fnon-call-exceptions -funwind-tables -fasynchronous-unwind-tables @gol
614 -finhibit-size-directive -finstrument-functions @gol
615 -fcheck-memory-usage -fprefix-function-name @gol
616 -fno-common -fno-ident -fno-gnu-linker @gol
617 -fpcc-struct-return -fpic -fPIC @gol
618 -freg-struct-return -fshared-data -fshort-enums @gol
619 -fshort-double -fvolatile @gol
620 -fvolatile-global -fvolatile-static @gol
621 -fverbose-asm -fpack-struct -fstack-check @gol
622 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
623 -fargument-alias -fargument-noalias @gol
624 -fargument-noalias-global -fleading-underscore}
628 * Overall Options:: Controlling the kind of output:
629 an executable, object files, assembler files,
630 or preprocessed source.
631 * C Dialect Options:: Controlling the variant of C language compiled.
632 * C++ Dialect Options:: Variations on C++.
633 * Objective-C Dialect Options:: Variations on Objective-C.
634 * Language Independent Options:: Controlling how diagnostics should be
636 * Warning Options:: How picky should the compiler be?
637 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
638 * Optimize Options:: How much optimization?
639 * Preprocessor Options:: Controlling header files and macro definitions.
640 Also, getting dependency information for Make.
641 * Assembler Options:: Passing options to the assembler.
642 * Link Options:: Specifying libraries and so on.
643 * Directory Options:: Where to find header files and libraries.
644 Where to find the compiler executable files.
645 * Spec Files:: How to pass switches to sub-processes.
646 * Target Options:: Running a cross-compiler, or an old version of GCC.
649 @node Overall Options
650 @section Options Controlling the Kind of Output
652 Compilation can involve up to four stages: preprocessing, compilation
653 proper, assembly and linking, always in that order. The first three
654 stages apply to an individual source file, and end by producing an
655 object file; linking combines all the object files (those newly
656 compiled, and those specified as input) into an executable file.
658 @cindex file name suffix
659 For any given input file, the file name suffix determines what kind of
664 C source code which must be preprocessed.
667 C source code which should not be preprocessed.
670 C++ source code which should not be preprocessed.
673 Objective-C source code. Note that you must link with the library
674 @file{libobjc.a} to make an Objective-C program work.
677 Objective-C source code which should not be preprocessed.
680 C header file (not to be compiled or linked).
684 @itemx @var{file}.cxx
685 @itemx @var{file}.cpp
686 @itemx @var{file}.c++
688 C++ source code which must be preprocessed. Note that in @samp{.cxx},
689 the last two letters must both be literally @samp{x}. Likewise,
690 @samp{.C} refers to a literal capital C@.
693 @itemx @var{file}.for
694 @itemx @var{file}.FOR
695 Fortran source code which should not be preprocessed.
698 @itemx @var{file}.fpp
699 @itemx @var{file}.FPP
700 Fortran source code which must be preprocessed (with the traditional
704 Fortran source code which must be preprocessed with a RATFOR
705 preprocessor (not included with GCC)@.
707 @xref{Overall Options,,Options Controlling the Kind of Output, g77,
708 Using and Porting GNU Fortran}, for more details of the handling of
711 @c FIXME: Descriptions of Java file types.
718 Ada source code file which contains a library unit declaration (a
719 declaration of a package, subprogram, or generic, or a generic
720 instantiation), or a library unit renaming declaration (a package,
721 generic, or subprogram renaming declaration). Such files are also
724 @itemx @var{file}.adb
725 Ada source code file containing a library unit body (a subprogram or
726 package body). Such files are also called @dfn{bodies}.
728 @c GCC also knows about some suffixes for languages not yet included:
734 @itemx @var{file}.chi
735 CHILL source code (preprocessed with the traditional preprocessor).
741 Assembler code which must be preprocessed.
744 An object file to be fed straight into linking.
745 Any file name with no recognized suffix is treated this way.
749 You can specify the input language explicitly with the @option{-x} option:
752 @item -x @var{language}
753 Specify explicitly the @var{language} for the following input files
754 (rather than letting the compiler choose a default based on the file
755 name suffix). This option applies to all following input files until
756 the next @option{-x} option. Possible values for @var{language} are:
758 c c-header cpp-output
760 objective-c objc-cpp-output
761 assembler assembler-with-cpp
764 f77 f77-cpp-input ratfor
769 Turn off any specification of a language, so that subsequent files are
770 handled according to their file name suffixes (as they are if @option{-x}
771 has not been used at all).
773 @item -pass-exit-codes
774 @opindex pass-exit-codes
775 Normally the @command{gcc} program will exit with the code of 1 if any
776 phase of the compiler returns a non-success return code. If you specify
777 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
778 numerically highest error produced by any phase that returned an error
782 If you only want some of the stages of compilation, you can use
783 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
784 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
785 @command{gcc} is to stop. Note that some combinations (for example,
786 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
791 Compile or assemble the source files, but do not link. The linking
792 stage simply is not done. The ultimate output is in the form of an
793 object file for each source file.
795 By default, the object file name for a source file is made by replacing
796 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
798 Unrecognized input files, not requiring compilation or assembly, are
803 Stop after the stage of compilation proper; do not assemble. The output
804 is in the form of an assembler code file for each non-assembler input
807 By default, the assembler file name for a source file is made by
808 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
810 Input files that don't require compilation are ignored.
814 Stop after the preprocessing stage; do not run the compiler proper. The
815 output is in the form of preprocessed source code, which is sent to the
818 Input files which don't require preprocessing are ignored.
820 @cindex output file option
823 Place output in file @var{file}. This applies regardless to whatever
824 sort of output is being produced, whether it be an executable file,
825 an object file, an assembler file or preprocessed C code.
827 Since only one output file can be specified, it does not make sense to
828 use @option{-o} when compiling more than one input file, unless you are
829 producing an executable file as output.
831 If @option{-o} is not specified, the default is to put an executable file
832 in @file{a.out}, the object file for @file{@var{source}.@var{suffix}} in
833 @file{@var{source}.o}, its assembler file in @file{@var{source}.s}, and
834 all preprocessed C source on standard output.
838 Print (on standard error output) the commands executed to run the stages
839 of compilation. Also print the version number of the compiler driver
840 program and of the preprocessor and the compiler proper.
844 Use pipes rather than temporary files for communication between the
845 various stages of compilation. This fails to work on some systems where
846 the assembler is unable to read from a pipe; but the GNU assembler has
851 Print (on the standard output) a description of the command line options
852 understood by @command{gcc}. If the @option{-v} option is also specified
853 then @option{--help} will also be passed on to the various processes
854 invoked by @command{gcc}, so that they can display the command line options
855 they accept. If the @option{-W} option is also specified then command
856 line options which have no documentation associated with them will also
861 Print (on the standard output) a description of target specific command
862 line options for each tool.
866 @section Compiling C++ Programs
868 @cindex suffixes for C++ source
869 @cindex C++ source file suffixes
870 C++ source files conventionally use one of the suffixes @samp{.C},
871 @samp{.cc}, @samp{.cpp}, @samp{.c++}, @samp{.cp}, or @samp{.cxx};
872 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
873 files with these names and compiles them as C++ programs even if you
874 call the compiler the same way as for compiling C programs (usually with
875 the name @command{gcc}).
879 However, C++ programs often require class libraries as well as a
880 compiler that understands the C++ language---and under some
881 circumstances, you might want to compile programs from standard input,
882 or otherwise without a suffix that flags them as C++ programs.
883 @command{g++} is a program that calls GCC with the default language
884 set to C++, and automatically specifies linking against the C++
885 library. On many systems, @command{g++} is also
886 installed with the name @command{c++}.
888 @cindex invoking @command{g++}
889 When you compile C++ programs, you may specify many of the same
890 command-line options that you use for compiling programs in any
891 language; or command-line options meaningful for C and related
892 languages; or options that are meaningful only for C++ programs.
893 @xref{C Dialect Options,,Options Controlling C Dialect}, for
894 explanations of options for languages related to C@.
895 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
896 explanations of options that are meaningful only for C++ programs.
898 @node C Dialect Options
899 @section Options Controlling C Dialect
900 @cindex dialect options
901 @cindex language dialect options
902 @cindex options, dialect
904 The following options control the dialect of C (or languages derived
905 from C, such as C++ and Objective-C) that the compiler accepts:
912 In C mode, support all ISO C89 programs. In C++ mode,
913 remove GNU extensions that conflict with ISO C++.
915 This turns off certain features of GCC that are incompatible with ISO
916 C89 (when compiling C code), or of standard C++ (when compiling C++ code),
917 such as the @code{asm} and @code{typeof} keywords, and
918 predefined macros such as @code{unix} and @code{vax} that identify the
919 type of system you are using. It also enables the undesirable and
920 rarely used ISO trigraph feature. For the C compiler,
921 it disables recognition of C++ style @samp{//} comments as well as
922 the @code{inline} keyword.
924 The alternate keywords @code{__asm__}, @code{__extension__},
925 @code{__inline__} and @code{__typeof__} continue to work despite
926 @option{-ansi}. You would not want to use them in an ISO C program, of
927 course, but it is useful to put them in header files that might be included
928 in compilations done with @option{-ansi}. Alternate predefined macros
929 such as @code{__unix__} and @code{__vax__} are also available, with or
930 without @option{-ansi}.
932 The @option{-ansi} option does not cause non-ISO programs to be
933 rejected gratuitously. For that, @option{-pedantic} is required in
934 addition to @option{-ansi}. @xref{Warning Options}.
936 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
937 option is used. Some header files may notice this macro and refrain
938 from declaring certain functions or defining certain macros that the
939 ISO standard doesn't call for; this is to avoid interfering with any
940 programs that might use these names for other things.
942 Functions which would normally be built in but do not have semantics
943 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
944 functions with @option{-ansi} is used. @xref{Other Builtins,,Other
945 built-in functions provided by GCC}, for details of the functions
950 Determine the language standard. This option is currently only
951 supported when compiling C@. A value for this option must be provided;
957 ISO C89 (same as @option{-ansi}).
960 ISO C89 as modified in amendment 1.
966 ISO C99. Note that this standard is not yet fully supported; see
967 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
968 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
971 Default, ISO C89 plus GNU extensions (including some C99 features).
975 ISO C99 plus GNU extensions. When ISO C99 is fully implemented in GCC,
976 this will become the default. The name @samp{gnu9x} is deprecated.
980 Even when this option is not specified, you can still use some of the
981 features of newer standards in so far as they do not conflict with
982 previous C standards. For example, you may use @code{__restrict__} even
983 when @option{-std=c99} is not specified.
985 The @option{-std} options specifying some version of ISO C have the same
986 effects as @option{-ansi}, except that features that were not in ISO C89
987 but are in the specified version (for example, @samp{//} comments and
988 the @code{inline} keyword in ISO C99) are not disabled.
990 @xref{Standards,,Language Standards Supported by GCC}, for details of
991 these standard versions.
993 @item -aux-info @var{filename}
995 Output to the given filename prototyped declarations for all functions
996 declared and/or defined in a translation unit, including those in header
997 files. This option is silently ignored in any language other than C@.
999 Besides declarations, the file indicates, in comments, the origin of
1000 each declaration (source file and line), whether the declaration was
1001 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1002 @samp{O} for old, respectively, in the first character after the line
1003 number and the colon), and whether it came from a declaration or a
1004 definition (@samp{C} or @samp{F}, respectively, in the following
1005 character). In the case of function definitions, a K&R-style list of
1006 arguments followed by their declarations is also provided, inside
1007 comments, after the declaration.
1011 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1012 keyword, so that code can use these words as identifiers. You can use
1013 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1014 instead. @option{-ansi} implies @option{-fno-asm}.
1016 In C++, this switch only affects the @code{typeof} keyword, since
1017 @code{asm} and @code{inline} are standard keywords. You may want to
1018 use the @option{-fno-gnu-keywords} flag instead, which has the same
1019 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1020 switch only affects the @code{asm} and @code{typeof} keywords, since
1021 @code{inline} is a standard keyword in ISO C99.
1024 @opindex fno-builtin
1025 @cindex built-in functions
1026 Don't recognize built-in functions that do not begin with
1027 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1028 functions provided by GCC}, for details of the functions affected,
1029 including those which are not built-in functions when @option{-ansi} or
1030 @option{-std} options for strict ISO C conformance are used because they
1031 do not have an ISO standard meaning.
1033 GCC normally generates special code to handle certain built-in functions
1034 more efficiently; for instance, calls to @code{alloca} may become single
1035 instructions that adjust the stack directly, and calls to @code{memcpy}
1036 may become inline copy loops. The resulting code is often both smaller
1037 and faster, but since the function calls no longer appear as such, you
1038 cannot set a breakpoint on those calls, nor can you change the behavior
1039 of the functions by linking with a different library.
1041 In C++, @option{-fno-builtin} is always in effect. The @option{-fbuiltin}
1042 option has no effect. Therefore, in C++, the only way to get the
1043 optimization benefits of built-in functions is to call the function
1044 using the @samp{__builtin_} prefix. The GNU C++ Standard Library uses
1045 built-in functions to implement many functions (like
1046 @code{std::strchr}), so that you automatically get efficient code.
1050 @cindex hosted environment
1052 Assert that compilation takes place in a hosted environment. This implies
1053 @option{-fbuiltin}. A hosted environment is one in which the
1054 entire standard library is available, and in which @code{main} has a return
1055 type of @code{int}. Examples are nearly everything except a kernel.
1056 This is equivalent to @option{-fno-freestanding}.
1058 @item -ffreestanding
1059 @opindex ffreestanding
1060 @cindex hosted environment
1062 Assert that compilation takes place in a freestanding environment. This
1063 implies @option{-fno-builtin}. A freestanding environment
1064 is one in which the standard library may not exist, and program startup may
1065 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1066 This is equivalent to @option{-fno-hosted}.
1068 @xref{Standards,,Language Standards Supported by GCC}, for details of
1069 freestanding and hosted environments.
1073 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1074 options for strict ISO C conformance) implies @option{-trigraphs}.
1076 @cindex traditional C language
1077 @cindex C language, traditional
1079 @opindex traditional
1080 Attempt to support some aspects of traditional C compilers.
1085 All @code{extern} declarations take effect globally even if they
1086 are written inside of a function definition. This includes implicit
1087 declarations of functions.
1090 The newer keywords @code{typeof}, @code{inline}, @code{signed}, @code{const}
1091 and @code{volatile} are not recognized. (You can still use the
1092 alternative keywords such as @code{__typeof__}, @code{__inline__}, and
1096 Comparisons between pointers and integers are always allowed.
1099 Integer types @code{unsigned short} and @code{unsigned char} promote
1100 to @code{unsigned int}.
1103 Out-of-range floating point literals are not an error.
1106 Certain constructs which ISO regards as a single invalid preprocessing
1107 number, such as @samp{0xe-0xd}, are treated as expressions instead.
1110 String ``constants'' are not necessarily constant; they are stored in
1111 writable space, and identical looking constants are allocated
1112 separately. (This is the same as the effect of
1113 @option{-fwritable-strings}.)
1115 @cindex @code{longjmp} and automatic variables
1117 All automatic variables not declared @code{register} are preserved by
1118 @code{longjmp}. Ordinarily, GNU C follows ISO C: automatic variables
1119 not declared @code{volatile} may be clobbered.
1124 @cindex escape sequences, traditional
1125 The character escape sequences @samp{\x} and @samp{\a} evaluate as the
1126 literal characters @samp{x} and @samp{a} respectively. Without
1127 @w{@option{-traditional}}, @samp{\x} is a prefix for the hexadecimal
1128 representation of a character, and @samp{\a} produces a bell.
1131 You may wish to use @option{-fno-builtin} as well as @option{-traditional}
1132 if your program uses names that are normally GNU C built-in functions for
1133 other purposes of its own.
1135 You cannot use @option{-traditional} if you include any header files that
1136 rely on ISO C features. Some vendors are starting to ship systems with
1137 ISO C header files and you cannot use @option{-traditional} on such
1138 systems to compile files that include any system headers.
1140 The @option{-traditional} option also enables @option{-traditional-cpp}.
1142 @item -traditional-cpp
1143 @opindex traditional-cpp
1144 Attempt to support some aspects of traditional C preprocessors.
1145 See the GNU CPP manual for details.
1147 @item -fcond-mismatch
1148 @opindex fcond-mismatch
1149 Allow conditional expressions with mismatched types in the second and
1150 third arguments. The value of such an expression is void. This option
1151 is not supported for C++.
1153 @item -funsigned-char
1154 @opindex funsigned-char
1155 Let the type @code{char} be unsigned, like @code{unsigned char}.
1157 Each kind of machine has a default for what @code{char} should
1158 be. It is either like @code{unsigned char} by default or like
1159 @code{signed char} by default.
1161 Ideally, a portable program should always use @code{signed char} or
1162 @code{unsigned char} when it depends on the signedness of an object.
1163 But many programs have been written to use plain @code{char} and
1164 expect it to be signed, or expect it to be unsigned, depending on the
1165 machines they were written for. This option, and its inverse, let you
1166 make such a program work with the opposite default.
1168 The type @code{char} is always a distinct type from each of
1169 @code{signed char} or @code{unsigned char}, even though its behavior
1170 is always just like one of those two.
1173 @opindex fsigned-char
1174 Let the type @code{char} be signed, like @code{signed char}.
1176 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1177 the negative form of @option{-funsigned-char}. Likewise, the option
1178 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1180 @item -fsigned-bitfields
1181 @itemx -funsigned-bitfields
1182 @itemx -fno-signed-bitfields
1183 @itemx -fno-unsigned-bitfields
1184 @opindex fsigned-bitfields
1185 @opindex funsigned-bitfields
1186 @opindex fno-signed-bitfields
1187 @opindex fno-unsigned-bitfields
1188 These options control whether a bit-field is signed or unsigned, when the
1189 declaration does not use either @code{signed} or @code{unsigned}. By
1190 default, such a bit-field is signed, because this is consistent: the
1191 basic integer types such as @code{int} are signed types.
1193 However, when @option{-traditional} is used, bit-fields are all unsigned
1196 @item -fwritable-strings
1197 @opindex fwritable-strings
1198 Store string constants in the writable data segment and don't uniquize
1199 them. This is for compatibility with old programs which assume they can
1200 write into string constants. The option @option{-traditional} also has
1203 Writing into string constants is a very bad idea; ``constants'' should
1206 @item -fallow-single-precision
1207 @opindex fallow-single-precision
1208 Do not promote single precision math operations to double precision,
1209 even when compiling with @option{-traditional}.
1211 Traditional K&R C promotes all floating point operations to double
1212 precision, regardless of the sizes of the operands. On the
1213 architecture for which you are compiling, single precision may be faster
1214 than double precision. If you must use @option{-traditional}, but want
1215 to use single precision operations when the operands are single
1216 precision, use this option. This option has no effect when compiling
1217 with ISO or GNU C conventions (the default).
1220 @opindex fshort-wchar
1221 Override the underlying type for @samp{wchar_t} to be @samp{short
1222 unsigned int} instead of the default for the target. This option is
1223 useful for building programs to run under WINE@.
1226 @node C++ Dialect Options
1227 @section Options Controlling C++ Dialect
1229 @cindex compiler options, C++
1230 @cindex C++ options, command line
1231 @cindex options, C++
1232 This section describes the command-line options that are only meaningful
1233 for C++ programs; but you can also use most of the GNU compiler options
1234 regardless of what language your program is in. For example, you
1235 might compile a file @code{firstClass.C} like this:
1238 g++ -g -frepo -O -c firstClass.C
1242 In this example, only @option{-frepo} is an option meant
1243 only for C++ programs; you can use the other options with any
1244 language supported by GCC@.
1246 Here is a list of options that are @emph{only} for compiling C++ programs:
1249 @item -fno-access-control
1250 @opindex fno-access-control
1251 Turn off all access checking. This switch is mainly useful for working
1252 around bugs in the access control code.
1256 Check that the pointer returned by @code{operator new} is non-null
1257 before attempting to modify the storage allocated. The current Working
1258 Paper requires that @code{operator new} never return a null pointer, so
1259 this check is normally unnecessary.
1261 An alternative to using this option is to specify that your
1262 @code{operator new} does not throw any exceptions; if you declare it
1263 @samp{throw()}, G++ will check the return value. See also @samp{new
1266 @item -fconserve-space
1267 @opindex fconserve-space
1268 Put uninitialized or runtime-initialized global variables into the
1269 common segment, as C does. This saves space in the executable at the
1270 cost of not diagnosing duplicate definitions. If you compile with this
1271 flag and your program mysteriously crashes after @code{main()} has
1272 completed, you may have an object that is being destroyed twice because
1273 two definitions were merged.
1275 This option is no longer useful on most targets, now that support has
1276 been added for putting variables into BSS without making them common.
1278 @item -fno-const-strings
1279 @opindex fno-const-strings
1280 Give string constants type @code{char *} instead of type @code{const
1281 char *}. By default, G++ uses type @code{const char *} as required by
1282 the standard. Even if you use @option{-fno-const-strings}, you cannot
1283 actually modify the value of a string constant, unless you also use
1284 @option{-fwritable-strings}.
1286 This option might be removed in a future release of G++. For maximum
1287 portability, you should structure your code so that it works with
1288 string constants that have type @code{const char *}.
1290 @item -fdollars-in-identifiers
1291 @opindex fdollars-in-identifiers
1292 Accept @samp{$} in identifiers. You can also explicitly prohibit use of
1293 @samp{$} with the option @option{-fno-dollars-in-identifiers}. (GNU C allows
1294 @samp{$} by default on most target systems, but there are a few exceptions.)
1295 Traditional C allowed the character @samp{$} to form part of
1296 identifiers. However, ISO C and C++ forbid @samp{$} in identifiers.
1298 @item -fno-elide-constructors
1299 @opindex fno-elide-constructors
1300 The C++ standard allows an implementation to omit creating a temporary
1301 which is only used to initialize another object of the same type.
1302 Specifying this option disables that optimization, and forces G++ to
1303 call the copy constructor in all cases.
1305 @item -fno-enforce-eh-specs
1306 @opindex fno-enforce-eh-specs
1307 Don't check for violation of exception specifications at runtime. This
1308 option violates the C++ standard, but may be useful for reducing code
1309 size in production builds, much like defining @samp{NDEBUG}. The compiler
1310 will still optimize based on the exception specifications.
1312 @item -fexternal-templates
1313 @opindex fexternal-templates
1315 Cause @samp{#pragma interface} and @samp{implementation} to apply to
1316 template instantiation; template instances are emitted or not according
1317 to the location of the template definition. @xref{Template
1318 Instantiation}, for more information.
1320 This option is deprecated.
1322 @item -falt-external-templates
1323 @opindex falt-external-templates
1324 Similar to @option{-fexternal-templates}, but template instances are
1325 emitted or not according to the place where they are first instantiated.
1326 @xref{Template Instantiation}, for more information.
1328 This option is deprecated.
1331 @itemx -fno-for-scope
1333 @opindex fno-for-scope
1334 If @option{-ffor-scope} is specified, the scope of variables declared in
1335 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1336 as specified by the C++ standard.
1337 If @option{-fno-for-scope} is specified, the scope of variables declared in
1338 a @i{for-init-statement} extends to the end of the enclosing scope,
1339 as was the case in old versions of G++, and other (traditional)
1340 implementations of C++.
1342 The default if neither flag is given to follow the standard,
1343 but to allow and give a warning for old-style code that would
1344 otherwise be invalid, or have different behavior.
1346 @item -fno-gnu-keywords
1347 @opindex fno-gnu-keywords
1348 Do not recognize @code{typeof} as a keyword, so that code can use this
1349 word as an identifier. You can use the keyword @code{__typeof__} instead.
1350 @option{-ansi} implies @option{-fno-gnu-keywords}.
1352 @item -fno-implicit-templates
1353 @opindex fno-implicit-templates
1354 Never emit code for non-inline templates which are instantiated
1355 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1356 @xref{Template Instantiation}, for more information.
1358 @item -fno-implicit-inline-templates
1359 @opindex fno-implicit-inline-templates
1360 Don't emit code for implicit instantiations of inline templates, either.
1361 The default is to handle inlines differently so that compiles with and
1362 without optimization will need the same set of explicit instantiations.
1364 @item -fno-implement-inlines
1365 @opindex fno-implement-inlines
1366 To save space, do not emit out-of-line copies of inline functions
1367 controlled by @samp{#pragma implementation}. This will cause linker
1368 errors if these functions are not inlined everywhere they are called.
1370 @item -fms-extensions
1371 @opindex fms-extensions
1372 Disable pedantic warnings about constructs used in MFC, such as implicit
1373 int and getting a pointer to member function via non-standard syntax.
1375 @item -fno-nonansi-builtins
1376 @opindex fno-nonansi-builtins
1377 Disable built-in declarations of functions that are not mandated by
1378 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1379 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1381 @item -fno-operator-names
1382 @opindex fno-operator-names
1383 Do not treat the operator name keywords @code{and}, @code{bitand},
1384 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1385 synonyms as keywords.
1387 @item -fno-optional-diags
1388 @opindex fno-optional-diags
1389 Disable diagnostics that the standard says a compiler does not need to
1390 issue. Currently, the only such diagnostic issued by G++ is the one for
1391 a name having multiple meanings within a class.
1394 @opindex fpermissive
1395 Downgrade messages about nonconformant code from errors to warnings. By
1396 default, G++ effectively sets @option{-pedantic-errors} without
1397 @option{-pedantic}; this option reverses that. This behavior and this
1398 option are superseded by @option{-pedantic}, which works as it does for GNU C@.
1402 Enable automatic template instantiation at link time. This option also
1403 implies @option{-fno-implicit-templates}. @xref{Template
1404 Instantiation}, for more information.
1408 Disable generation of information about every class with virtual
1409 functions for use by the C++ runtime type identification features
1410 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1411 of the language, you can save some space by using this flag. Note that
1412 exception handling uses the same information, but it will generate it as
1417 Emit statistics about front-end processing at the end of the compilation.
1418 This information is generally only useful to the G++ development team.
1420 @item -ftemplate-depth-@var{n}
1421 @opindex ftemplate-depth
1422 Set the maximum instantiation depth for template classes to @var{n}.
1423 A limit on the template instantiation depth is needed to detect
1424 endless recursions during template class instantiation. ANSI/ISO C++
1425 conforming programs must not rely on a maximum depth greater than 17.
1427 @item -fuse-cxa-atexit
1428 @opindex fuse-cxa-atexit
1429 Register destructors for objects with static storage duration with the
1430 @code{__cxa_atexit} function rather than the @code{atexit} function.
1431 This option is required for fully standards-compliant handling of static
1432 destructors, but will only work if your C library supports
1433 @code{__cxa_atexit}.
1437 Emit special relocations for vtables and virtual function references
1438 so that the linker can identify unused virtual functions and zero out
1439 vtable slots that refer to them. This is most useful with
1440 @option{-ffunction-sections} and @option{-Wl,--gc-sections}, in order to
1441 also discard the functions themselves.
1443 This optimization requires GNU as and GNU ld. Not all systems support
1444 this option. @option{-Wl,--gc-sections} is ignored without @option{-static}.
1448 Do not use weak symbol support, even if it is provided by the linker.
1449 By default, G++ will use weak symbols if they are available. This
1450 option exists only for testing, and should not be used by end-users;
1451 it will result in inferior code and has no benefits. This option may
1452 be removed in a future release of G++.
1456 Do not search for header files in the standard directories specific to
1457 C++, but do still search the other standard directories. (This option
1458 is used when building the C++ library.)
1461 In addition, these optimization, warning, and code generation options
1462 have meanings only for C++ programs:
1465 @item -fno-default-inline
1466 @opindex fno-default-inline
1467 Do not assume @samp{inline} for functions defined inside a class scope.
1468 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1469 functions will have linkage like inline functions; they just won't be
1472 @item -Wctor-dtor-privacy @r{(C++ only)}
1473 @opindex Wctor-dtor-privacy
1474 Warn when a class seems unusable, because all the constructors or
1475 destructors in a class are private and the class has no friends or
1476 public static member functions.
1478 @item -Wnon-virtual-dtor @r{(C++ only)}
1479 @opindex Wnon-virtual-dtor
1480 Warn when a class declares a non-virtual destructor that should probably
1481 be virtual, because it looks like the class will be used polymorphically.
1483 @item -Wreorder @r{(C++ only)}
1485 @cindex reordering, warning
1486 @cindex warning for reordering of member initializers
1487 Warn when the order of member initializers given in the code does not
1488 match the order in which they must be executed. For instance:
1494 A(): j (0), i (1) @{ @}
1498 Here the compiler will warn that the member initializers for @samp{i}
1499 and @samp{j} will be rearranged to match the declaration order of the
1503 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
1506 @item -Weffc++ @r{(C++ only)}
1508 Warn about violations of various style guidelines from Scott Meyers'
1509 @cite{Effective C++} books. If you use this option, you should be aware
1510 that the standard library headers do not obey all of these guidelines;
1511 you can use @samp{grep -v} to filter out those warnings.
1513 @item -Wno-deprecated @r{(C++ only)}
1514 @opindex Wno-deprecated
1515 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1517 @item -Wno-non-template-friend @r{(C++ only)}
1518 @opindex Wno-non-template-friend
1519 Disable warnings when non-templatized friend functions are declared
1520 within a template. With the advent of explicit template specification
1521 support in G++, if the name of the friend is an unqualified-id (i.e.,
1522 @samp{friend foo(int)}), the C++ language specification demands that the
1523 friend declare or define an ordinary, nontemplate function. (Section
1524 14.5.3). Before G++ implemented explicit specification, unqualified-ids
1525 could be interpreted as a particular specialization of a templatized
1526 function. Because this non-conforming behavior is no longer the default
1527 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
1528 check existing code for potential trouble spots, and is on by default.
1529 This new compiler behavior can be turned off with
1530 @option{-Wno-non-template-friend} which keeps the conformant compiler code
1531 but disables the helpful warning.
1533 @item -Wold-style-cast @r{(C++ only)}
1534 @opindex Wold-style-cast
1535 Warn if an old-style (C-style) cast is used within a C++ program. The
1536 new-style casts (@samp{static_cast}, @samp{reinterpret_cast}, and
1537 @samp{const_cast}) are less vulnerable to unintended effects, and much
1540 @item -Woverloaded-virtual @r{(C++ only)}
1541 @opindex Woverloaded-virtual
1542 @cindex overloaded virtual fn, warning
1543 @cindex warning for overloaded virtual fn
1544 Warn when a function declaration hides virtual functions from a
1545 base class. For example, in:
1552 struct B: public A @{
1557 the @code{A} class version of @code{f} is hidden in @code{B}, and code
1565 will fail to compile.
1567 @item -Wno-pmf-conversions @r{(C++ only)}
1568 @opindex Wno-pmf-conversions
1569 Disable the diagnostic for converting a bound pointer to member function
1572 @item -Wsign-promo @r{(C++ only)}
1573 @opindex Wsign-promo
1574 Warn when overload resolution chooses a promotion from unsigned or
1575 enumeral type to a signed type over a conversion to an unsigned type of
1576 the same size. Previous versions of G++ would try to preserve
1577 unsignedness, but the standard mandates the current behavior.
1579 @item -Wsynth @r{(C++ only)}
1581 @cindex warning for synthesized methods
1582 @cindex synthesized methods, warning
1583 Warn when G++'s synthesis behavior does not match that of cfront. For
1589 A& operator = (int);
1599 In this example, G++ will synthesize a default @samp{A& operator =
1600 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1603 @node Objective-C Dialect Options
1604 @section Options Controlling Objective-C Dialect
1606 @cindex compiler options, Objective-C
1607 @cindex Objective-C options, command line
1608 @cindex options, Objective-C
1609 This section describes the command-line options that are only meaningful
1610 for Objective-C programs; but you can also use most of the GNU compiler
1611 options regardless of what language your program is in. For example,
1612 you might compile a file @code{some_class.m} like this:
1615 gcc -g -fgnu-runtime -O -c some_class.m
1619 In this example, only @option{-fgnu-runtime} is an option meant only for
1620 Objective-C programs; you can use the other options with any language
1623 Here is a list of options that are @emph{only} for compiling Objective-C
1627 @item -fconstant-string-class=@var{class-name}
1628 @opindex fconstant-string-class
1629 Use @var{class-name} as the name of the class to instantiate for each
1630 literal string specified with the syntax @code{@@"@dots{}"}. The default
1631 class name is @code{NXConstantString}.
1634 @opindex fgnu-runtime
1635 Generate object code compatible with the standard GNU Objective-C
1636 runtime. This is the default for most types of systems.
1638 @item -fnext-runtime
1639 @opindex fnext-runtime
1640 Generate output compatible with the NeXT runtime. This is the default
1641 for NeXT-based systems, including Darwin and Mac OS X@.
1645 Dump interface declarations for all classes seen in the source file to a
1646 file named @file{@var{sourcename}.decl}.
1649 @opindex Wno-protocol
1650 Do not warn if methods required by a protocol are not implemented
1651 in the class adopting it.
1655 Warn if a selector has multiple methods of different types defined.
1657 @c not documented because only avail via -Wp
1658 @c @item -print-objc-runtime-info
1662 @node Language Independent Options
1663 @section Options to Control Diagnostic Messages Formatting
1664 @cindex options to control diagnostics formatting
1665 @cindex diagnostic messages
1666 @cindex message formatting
1668 Traditionally, diagnostic messages have been formatted irrespective of
1669 the output device's aspect (e.g.@: its width, @dots{}). The options described
1670 below can be used to control the diagnostic messages formatting
1671 algorithm, e.g.@: how many characters per line, how often source location
1672 information should be reported. Right now, only the C++ front end can
1673 honor these options. However it is expected, in the near future, that
1674 the remaining front ends would be able to digest them correctly.
1677 @item -fmessage-length=@var{n}
1678 @opindex fmessage-length
1679 Try to format error messages so that they fit on lines of about @var{n}
1680 characters. The default is 72 characters for @command{g++} and 0 for the rest of
1681 the front ends supported by GCC@. If @var{n} is zero, then no
1682 line-wrapping will be done; each error message will appear on a single
1685 @opindex fdiagnostics-show-location
1686 @item -fdiagnostics-show-location=once
1687 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
1688 reporter to emit @emph{once} source location information; that is, in
1689 case the message is too long to fit on a single physical line and has to
1690 be wrapped, the source location won't be emitted (as prefix) again,
1691 over and over, in subsequent continuation lines. This is the default
1694 @item -fdiagnostics-show-location=every-line
1695 Only meaningful in line-wrapping mode. Instructs the diagnostic
1696 messages reporter to emit the same source location information (as
1697 prefix) for physical lines that result from the process of breaking
1698 a message which is too long to fit on a single line.
1702 @node Warning Options
1703 @section Options to Request or Suppress Warnings
1704 @cindex options to control warnings
1705 @cindex warning messages
1706 @cindex messages, warning
1707 @cindex suppressing warnings
1709 Warnings are diagnostic messages that report constructions which
1710 are not inherently erroneous but which are risky or suggest there
1711 may have been an error.
1713 You can request many specific warnings with options beginning @samp{-W},
1714 for example @option{-Wimplicit} to request warnings on implicit
1715 declarations. Each of these specific warning options also has a
1716 negative form beginning @samp{-Wno-} to turn off warnings;
1717 for example, @option{-Wno-implicit}. This manual lists only one of the
1718 two forms, whichever is not the default.
1720 These options control the amount and kinds of warnings produced by GCC:
1723 @cindex syntax checking
1725 @opindex fsyntax-only
1726 Check the code for syntax errors, but don't do anything beyond that.
1730 Issue all the warnings demanded by strict ISO C and ISO C++;
1731 reject all programs that use forbidden extensions, and some other
1732 programs that do not follow ISO C and ISO C++. For ISO C, follows the
1733 version of the ISO C standard specified by any @option{-std} option used.
1735 Valid ISO C and ISO C++ programs should compile properly with or without
1736 this option (though a rare few will require @option{-ansi} or a
1737 @option{-std} option specifying the required version of ISO C)@. However,
1738 without this option, certain GNU extensions and traditional C and C++
1739 features are supported as well. With this option, they are rejected.
1741 @option{-pedantic} does not cause warning messages for use of the
1742 alternate keywords whose names begin and end with @samp{__}. Pedantic
1743 warnings are also disabled in the expression that follows
1744 @code{__extension__}. However, only system header files should use
1745 these escape routes; application programs should avoid them.
1746 @xref{Alternate Keywords}.
1748 Some users try to use @option{-pedantic} to check programs for strict ISO
1749 C conformance. They soon find that it does not do quite what they want:
1750 it finds some non-ISO practices, but not all---only those for which
1751 ISO C @emph{requires} a diagnostic, and some others for which
1752 diagnostics have been added.
1754 A feature to report any failure to conform to ISO C might be useful in
1755 some instances, but would require considerable additional work and would
1756 be quite different from @option{-pedantic}. We don't have plans to
1757 support such a feature in the near future.
1759 Where the standard specified with @option{-std} represents a GNU
1760 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
1761 corresponding @dfn{base standard}, the version of ISO C on which the GNU
1762 extended dialect is based. Warnings from @option{-pedantic} are given
1763 where they are required by the base standard. (It would not make sense
1764 for such warnings to be given only for features not in the specified GNU
1765 C dialect, since by definition the GNU dialects of C include all
1766 features the compiler supports with the given option, and there would be
1767 nothing to warn about.)
1769 @item -pedantic-errors
1770 @opindex pedantic-errors
1771 Like @option{-pedantic}, except that errors are produced rather than
1776 Inhibit all warning messages.
1780 Inhibit warning messages about the use of @samp{#import}.
1782 @item -Wchar-subscripts
1783 @opindex Wchar-subscripts
1784 Warn if an array subscript has type @code{char}. This is a common cause
1785 of error, as programmers often forget that this type is signed on some
1790 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
1791 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
1795 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
1796 the arguments supplied have types appropriate to the format string
1797 specified, and that the conversions specified in the format string make
1798 sense. This includes standard functions, and others specified by format
1799 attributes (@pxref{Function Attributes}), in the @code{printf},
1800 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
1801 not in the C standard) families.
1803 The formats are checked against the format features supported by GNU
1804 libc version 2.2. These include all ISO C89 and C99 features, as well
1805 as features from the Single Unix Specification and some BSD and GNU
1806 extensions. Other library implementations may not support all these
1807 features; GCC does not support warning about features that go beyond a
1808 particular library's limitations. However, if @option{-pedantic} is used
1809 with @option{-Wformat}, warnings will be given about format features not
1810 in the selected standard version (but not for @code{strfmon} formats,
1811 since those are not in any version of the C standard). @xref{C Dialect
1812 Options,,Options Controlling C Dialect}.
1814 @option{-Wformat} is included in @option{-Wall}. For more control over some
1815 aspects of format checking, the options @option{-Wno-format-y2k},
1816 @option{-Wno-format-extra-args}, @option{-Wformat-nonliteral},
1817 @option{-Wformat-security} and @option{-Wformat=2} are available, but are
1818 not included in @option{-Wall}.
1820 @item -Wno-format-y2k
1821 @opindex Wno-format-y2k
1822 If @option{-Wformat} is specified, do not warn about @code{strftime}
1823 formats which may yield only a two-digit year.
1825 @item -Wno-format-extra-args
1826 @opindex Wno-format-extra-args
1827 If @option{-Wformat} is specified, do not warn about excess arguments to a
1828 @code{printf} or @code{scanf} format function. The C standard specifies
1829 that such arguments are ignored.
1831 @item -Wformat-nonliteral
1832 @opindex Wformat-nonliteral
1833 If @option{-Wformat} is specified, also warn if the format string is not a
1834 string literal and so cannot be checked, unless the format function
1835 takes its format arguments as a @code{va_list}.
1837 @item -Wformat-security
1838 @opindex Wformat-security
1839 If @option{-Wformat} is specified, also warn about uses of format
1840 functions that represent possible security problems. At present, this
1841 warns about calls to @code{printf} and @code{scanf} functions where the
1842 format string is not a string literal and there are no format arguments,
1843 as in @code{printf (foo);}. This may be a security hole if the format
1844 string came from untrusted input and contains @samp{%n}. (This is
1845 currently a subset of what @option{-Wformat-nonliteral} warns about, but
1846 in future warnings may be added to @option{-Wformat-security} that are not
1847 included in @option{-Wformat-nonliteral}.)
1851 Enable @option{-Wformat} plus format checks not included in
1852 @option{-Wformat}. Currently equivalent to @samp{-Wformat
1853 -Wformat-nonliteral -Wformat-security}.
1855 @item -Wimplicit-int
1856 @opindex Wimplicit-int
1857 Warn when a declaration does not specify a type.
1859 @item -Wimplicit-function-declaration
1860 @itemx -Werror-implicit-function-declaration
1861 @opindex Wimplicit-function-declaration
1862 @opindex Werror-implicit-function-declaration
1863 Give a warning (or error) whenever a function is used before being
1868 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
1872 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
1873 function with external linkage, returning int, taking either zero
1874 arguments, two, or three arguments of appropriate types.
1876 @item -Wmissing-braces
1877 @opindex Wmissing-braces
1878 Warn if an aggregate or union initializer is not fully bracketed. In
1879 the following example, the initializer for @samp{a} is not fully
1880 bracketed, but that for @samp{b} is fully bracketed.
1883 int a[2][2] = @{ 0, 1, 2, 3 @};
1884 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
1888 @opindex Wparentheses
1889 Warn if parentheses are omitted in certain contexts, such
1890 as when there is an assignment in a context where a truth value
1891 is expected, or when operators are nested whose precedence people
1892 often get confused about.
1894 Also warn about constructions where there may be confusion to which
1895 @code{if} statement an @code{else} branch belongs. Here is an example of
1910 In C, every @code{else} branch belongs to the innermost possible @code{if}
1911 statement, which in this example is @code{if (b)}. This is often not
1912 what the programmer expected, as illustrated in the above example by
1913 indentation the programmer chose. When there is the potential for this
1914 confusion, GCC will issue a warning when this flag is specified.
1915 To eliminate the warning, add explicit braces around the innermost
1916 @code{if} statement so there is no way the @code{else} could belong to
1917 the enclosing @code{if}. The resulting code would look like this:
1933 @item -Wsequence-point
1934 @opindex Wsequence-point
1935 Warn about code that may have undefined semantics because of violations
1936 of sequence point rules in the C standard.
1938 The C standard defines the order in which expressions in a C program are
1939 evaluated in terms of @dfn{sequence points}, which represent a partial
1940 ordering between the execution of parts of the program: those executed
1941 before the sequence point, and those executed after it. These occur
1942 after the evaluation of a full expression (one which is not part of a
1943 larger expression), after the evaluation of the first operand of a
1944 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
1945 function is called (but after the evaluation of its arguments and the
1946 expression denoting the called function), and in certain other places.
1947 Other than as expressed by the sequence point rules, the order of
1948 evaluation of subexpressions of an expression is not specified. All
1949 these rules describe only a partial order rather than a total order,
1950 since, for example, if two functions are called within one expression
1951 with no sequence point between them, the order in which the functions
1952 are called is not specified. However, the standards committee have
1953 ruled that function calls do not overlap.
1955 It is not specified when between sequence points modifications to the
1956 values of objects take effect. Programs whose behavior depends on this
1957 have undefined behavior; the C standard specifies that ``Between the
1958 previous and next sequence point an object shall have its stored value
1959 modified at most once by the evaluation of an expression. Furthermore,
1960 the prior value shall be read only to determine the value to be
1961 stored.''. If a program breaks these rules, the results on any
1962 particular implementation are entirely unpredictable.
1964 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
1965 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
1966 diagnosed by this option, and it may give an occasional false positive
1967 result, but in general it has been found fairly effective at detecting
1968 this sort of problem in programs.
1970 The present implementation of this option only works for C programs. A
1971 future implementation may also work for C++ programs.
1973 The C standard is worded confusingly, therefore there is some debate
1974 over the precise meaning of the sequence point rules in subtle cases.
1975 Links to discussions of the problem, including proposed formal
1976 definitions, may be found on our readings page, at
1977 @w{@uref{http://gcc.gnu.org/readings.html}}.
1980 @opindex Wreturn-type
1981 Warn whenever a function is defined with a return-type that defaults to
1982 @code{int}. Also warn about any @code{return} statement with no
1983 return-value in a function whose return-type is not @code{void}.
1985 For C++, a function without return type always produces a diagnostic
1986 message, even when @option{-Wno-return-type} is specified. The only
1987 exceptions are @samp{main} and functions defined in system headers.
1991 Warn whenever a @code{switch} statement has an index of enumeral type
1992 and lacks a @code{case} for one or more of the named codes of that
1993 enumeration. (The presence of a @code{default} label prevents this
1994 warning.) @code{case} labels outside the enumeration range also
1995 provoke warnings when this option is used.
1999 Warn if any trigraphs are encountered that might change the meaning of
2000 the program (trigraphs within comments are not warned about).
2002 @item -Wunused-function
2003 @opindex Wunused-function
2004 Warn whenever a static function is declared but not defined or a
2005 non\-inline static function is unused.
2007 @item -Wunused-label
2008 @opindex Wunused-label
2009 Warn whenever a label is declared but not used.
2011 To suppress this warning use the @samp{unused} attribute
2012 (@pxref{Variable Attributes}).
2014 @item -Wunused-parameter
2015 @opindex Wunused-parameter
2016 Warn whenever a function parameter is unused aside from its declaration.
2018 To suppress this warning use the @samp{unused} attribute
2019 (@pxref{Variable Attributes}).
2021 @item -Wunused-variable
2022 @opindex Wunused-variable
2023 Warn whenever a local variable or non-constant static variable is unused
2024 aside from its declaration
2026 To suppress this warning use the @samp{unused} attribute
2027 (@pxref{Variable Attributes}).
2029 @item -Wunused-value
2030 @opindex Wunused-value
2031 Warn whenever a statement computes a result that is explicitly not used.
2033 To suppress this warning cast the expression to @samp{void}.
2037 All all the above @option{-Wunused} options combined.
2039 In order to get a warning about an unused function parameter, you must
2040 either specify @samp{-W -Wunused} or separately specify
2041 @option{-Wunused-parameter}.
2043 @item -Wuninitialized
2044 @opindex Wuninitialized
2045 Warn if an automatic variable is used without first being initialized or
2046 if a variable may be clobbered by a @code{setjmp} call.
2048 These warnings are possible only in optimizing compilation,
2049 because they require data flow information that is computed only
2050 when optimizing. If you don't specify @option{-O}, you simply won't
2053 These warnings occur only for variables that are candidates for
2054 register allocation. Therefore, they do not occur for a variable that
2055 is declared @code{volatile}, or whose address is taken, or whose size
2056 is other than 1, 2, 4 or 8 bytes. Also, they do not occur for
2057 structures, unions or arrays, even when they are in registers.
2059 Note that there may be no warning about a variable that is used only
2060 to compute a value that itself is never used, because such
2061 computations may be deleted by data flow analysis before the warnings
2064 These warnings are made optional because GCC is not smart
2065 enough to see all the reasons why the code might be correct
2066 despite appearing to have an error. Here is one example of how
2087 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
2088 always initialized, but GCC doesn't know this. Here is
2089 another common case:
2094 if (change_y) save_y = y, y = new_y;
2096 if (change_y) y = save_y;
2101 This has no bug because @code{save_y} is used only if it is set.
2103 @cindex @code{longjmp} warnings
2104 This option also warns when a non-volatile automatic variable might be
2105 changed by a call to @code{longjmp}. These warnings as well are possible
2106 only in optimizing compilation.
2108 The compiler sees only the calls to @code{setjmp}. It cannot know
2109 where @code{longjmp} will be called; in fact, a signal handler could
2110 call it at any point in the code. As a result, you may get a warning
2111 even when there is in fact no problem because @code{longjmp} cannot
2112 in fact be called at the place which would cause a problem.
2114 Some spurious warnings can be avoided if you declare all the functions
2115 you use that never return as @code{noreturn}. @xref{Function
2118 @item -Wreorder @r{(C++ only)}
2120 @cindex reordering, warning
2121 @cindex warning for reordering of member initializers
2122 Warn when the order of member initializers given in the code does not
2123 match the order in which they must be executed. For instance:
2125 @item -Wunknown-pragmas
2126 @opindex Wunknown-pragmas
2127 @cindex warning for unknown pragmas
2128 @cindex unknown pragmas, warning
2129 @cindex pragmas, warning of unknown
2130 Warn when a #pragma directive is encountered which is not understood by
2131 GCC@. If this command line option is used, warnings will even be issued
2132 for unknown pragmas in system header files. This is not the case if
2133 the warnings were only enabled by the @option{-Wall} command line option.
2137 All of the above @samp{-W} options combined. This enables all the
2138 warnings about constructions that some users consider questionable, and
2139 that are easy to avoid (or modify to prevent the warning), even in
2140 conjunction with macros.
2143 @opindex Wno-div-by-zero
2144 @opindex Wdiv-by-zero
2145 Warn about compile-time integer division by zero. This is default. To
2146 inhibit the warning messages, use @option{-Wno-div-by-zero}. Floating
2147 point division by zero is not warned about, as it can be a legitimate
2148 way of obtaining infinities and NaNs.
2151 @opindex Wno-multichar
2153 Warn if a multicharacter constant (@samp{'FOOF'}) is used. This is
2154 default. To inhibit the warning messages, use @option{-Wno-multichar}.
2155 Usually they indicate a typo in the user's code, as they have
2156 implementation-defined values, and should not be used in portable code.
2158 @item -Wsystem-headers
2159 @opindex Wsystem-headers
2160 @cindex warnings from system headers
2161 @cindex system headers, warnings from
2162 Print warning messages for constructs found in system header files.
2163 Warnings from system headers are normally suppressed, on the assumption
2164 that they usually do not indicate real problems and would only make the
2165 compiler output harder to read. Using this command line option tells
2166 GCC to emit warnings from system headers as if they occurred in user
2167 code. However, note that using @option{-Wall} in conjunction with this
2168 option will @emph{not} warn about unknown pragmas in system
2169 headers---for that, @option{-Wunknown-pragmas} must also be used.
2172 The following @option{-W@dots{}} options are not implied by @option{-Wall}.
2173 Some of them warn about constructions that users generally do not
2174 consider questionable, but which occasionally you might wish to check
2175 for; others warn about constructions that are necessary or hard to avoid
2176 in some cases, and there is no simple way to modify the code to suppress
2182 Print extra warning messages for these events:
2186 A function can return either with or without a value. (Falling
2187 off the end of the function body is considered returning without
2188 a value.) For example, this function would evoke such a
2202 An expression-statement or the left-hand side of a comma expression
2203 contains no side effects.
2204 To suppress the warning, cast the unused expression to void.
2205 For example, an expression such as @samp{x[i,j]} will cause a warning,
2206 but @samp{x[(void)i,j]} will not.
2209 An unsigned value is compared against zero with @samp{<} or @samp{<=}.
2212 A comparison like @samp{x<=y<=z} appears; this is equivalent to
2213 @samp{(x<=y ? 1 : 0) <= z}, which is a different interpretation from
2214 that of ordinary mathematical notation.
2217 Storage-class specifiers like @code{static} are not the first things in
2218 a declaration. According to the C Standard, this usage is obsolescent.
2221 The return type of a function has a type qualifier such as @code{const}.
2222 Such a type qualifier has no effect, since the value returned by a
2223 function is not an lvalue. (But don't warn about the GNU extension of
2224 @code{volatile void} return types. That extension will be warned about
2225 if @option{-pedantic} is specified.)
2228 If @option{-Wall} or @option{-Wunused} is also specified, warn about unused
2232 A comparison between signed and unsigned values could produce an
2233 incorrect result when the signed value is converted to unsigned.
2234 (But don't warn if @option{-Wno-sign-compare} is also specified.)
2237 An aggregate has a partly bracketed initializer.
2238 For example, the following code would evoke such a warning,
2239 because braces are missing around the initializer for @code{x.h}:
2242 struct s @{ int f, g; @};
2243 struct t @{ struct s h; int i; @};
2244 struct t x = @{ 1, 2, 3 @};
2248 An aggregate has an initializer which does not initialize all members.
2249 For example, the following code would cause such a warning, because
2250 @code{x.h} would be implicitly initialized to zero:
2253 struct s @{ int f, g, h; @};
2254 struct s x = @{ 3, 4 @};
2259 @opindex Wfloat-equal
2260 Warn if floating point values are used in equality comparisons.
2262 The idea behind this is that sometimes it is convenient (for the
2263 programmer) to consider floating-point values as approximations to
2264 infinitely precise real numbers. If you are doing this, then you need
2265 to compute (by analysing the code, or in some other way) the maximum or
2266 likely maximum error that the computation introduces, and allow for it
2267 when performing comparisons (and when producing output, but that's a
2268 different problem). In particular, instead of testing for equality, you
2269 would check to see whether the two values have ranges that overlap; and
2270 this is done with the relational operators, so equality comparisons are
2273 @item -Wtraditional @r{(C only)}
2274 @opindex Wtraditional
2275 Warn about certain constructs that behave differently in traditional and
2276 ISO C@. Also warn about ISO C constructs that have no traditional C
2277 equivalent, and/or problematic constructs which should be avoided.
2281 Macro parameters that appear within string literals in the macro body.
2282 In traditional C macro replacement takes place within string literals,
2283 but does not in ISO C@.
2286 In traditional C, some preprocessor directives did not exist.
2287 Traditional preprocessors would only consider a line to be a directive
2288 if the @samp{#} appeared in column 1 on the line. Therefore
2289 @option{-Wtraditional} warns about directives that traditional C
2290 understands but would ignore because the @samp{#} does not appear as the
2291 first character on the line. It also suggests you hide directives like
2292 @samp{#pragma} not understood by traditional C by indenting them. Some
2293 traditional implementations would not recognise @samp{#elif}, so it
2294 suggests avoiding it altogether.
2297 A function-like macro that appears without arguments.
2300 The unary plus operator.
2303 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
2304 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
2305 constants.) Note, these suffixes appear in macros defined in the system
2306 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
2307 Use of these macros in user code might normally lead to spurious
2308 warnings, however gcc's integrated preprocessor has enough context to
2309 avoid warning in these cases.
2312 A function declared external in one block and then used after the end of
2316 A @code{switch} statement has an operand of type @code{long}.
2319 A non-@code{static} function declaration follows a @code{static} one.
2320 This construct is not accepted by some traditional C compilers.
2323 The ISO type of an integer constant has a different width or
2324 signedness from its traditional type. This warning is only issued if
2325 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
2326 typically represent bit patterns, are not warned about.
2329 Usage of ISO string concatenation is detected.
2332 Initialization of automatic aggregates.
2335 Identifier conflicts with labels. Traditional C lacks a separate
2336 namespace for labels.
2339 Initialization of unions. If the initializer is zero, the warning is
2340 omitted. This is done under the assumption that the zero initializer in
2341 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
2342 initializer warnings and relies on default initialization to zero in the
2346 Conversions by prototypes between fixed/floating point values and vice
2347 versa. The absence of these prototypes when compiling with traditional
2348 C would cause serious problems. This is a subset of the possible
2349 conversion warnings, for the full set use @option{-Wconversion}.
2354 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
2358 Warn whenever a local variable shadows another local variable, parameter or
2359 global variable or whenever a built-in function is shadowed.
2361 @item -Wlarger-than-@var{len}
2362 @opindex Wlarger-than
2363 Warn whenever an object of larger than @var{len} bytes is defined.
2365 @item -Wpointer-arith
2366 @opindex Wpointer-arith
2367 Warn about anything that depends on the ``size of'' a function type or
2368 of @code{void}. GNU C assigns these types a size of 1, for
2369 convenience in calculations with @code{void *} pointers and pointers
2372 @item -Wbad-function-cast @r{(C only)}
2373 @opindex Wbad-function-cast
2374 Warn whenever a function call is cast to a non-matching type.
2375 For example, warn if @code{int malloc()} is cast to @code{anything *}.
2379 Warn whenever a pointer is cast so as to remove a type qualifier from
2380 the target type. For example, warn if a @code{const char *} is cast
2381 to an ordinary @code{char *}.
2384 @opindex Wcast-align
2385 Warn whenever a pointer is cast such that the required alignment of the
2386 target is increased. For example, warn if a @code{char *} is cast to
2387 an @code{int *} on machines where integers can only be accessed at
2388 two- or four-byte boundaries.
2390 @item -Wwrite-strings
2391 @opindex Wwrite-strings
2392 When compiling C, give string constants the type @code{const
2393 char[@var{length}]} so that
2394 copying the address of one into a non-@code{const} @code{char *}
2395 pointer will get a warning; when compiling C++, warn about the
2396 deprecated conversion from string constants to @code{char *}.
2397 These warnings will help you find at
2398 compile time code that can try to write into a string constant, but
2399 only if you have been very careful about using @code{const} in
2400 declarations and prototypes. Otherwise, it will just be a nuisance;
2401 this is why we did not make @option{-Wall} request these warnings.
2404 @opindex Wconversion
2405 Warn if a prototype causes a type conversion that is different from what
2406 would happen to the same argument in the absence of a prototype. This
2407 includes conversions of fixed point to floating and vice versa, and
2408 conversions changing the width or signedness of a fixed point argument
2409 except when the same as the default promotion.
2411 Also, warn if a negative integer constant expression is implicitly
2412 converted to an unsigned type. For example, warn about the assignment
2413 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
2414 casts like @code{(unsigned) -1}.
2416 @item -Wsign-compare
2417 @opindex Wsign-compare
2418 @cindex warning for comparison of signed and unsigned values
2419 @cindex comparison of signed and unsigned values, warning
2420 @cindex signed and unsigned values, comparison warning
2421 Warn when a comparison between signed and unsigned values could produce
2422 an incorrect result when the signed value is converted to unsigned.
2423 This warning is also enabled by @option{-W}; to get the other warnings
2424 of @option{-W} without this warning, use @samp{-W -Wno-sign-compare}.
2426 @item -Waggregate-return
2427 @opindex Waggregate-return
2428 Warn if any functions that return structures or unions are defined or
2429 called. (In languages where you can return an array, this also elicits
2432 @item -Wstrict-prototypes @r{(C only)}
2433 @opindex Wstrict-prototypes
2434 Warn if a function is declared or defined without specifying the
2435 argument types. (An old-style function definition is permitted without
2436 a warning if preceded by a declaration which specifies the argument
2439 @item -Wmissing-prototypes @r{(C only)}
2440 @opindex Wmissing-prototypes
2441 Warn if a global function is defined without a previous prototype
2442 declaration. This warning is issued even if the definition itself
2443 provides a prototype. The aim is to detect global functions that fail
2444 to be declared in header files.
2446 @item -Wmissing-declarations
2447 @opindex Wmissing-declarations
2448 Warn if a global function is defined without a previous declaration.
2449 Do so even if the definition itself provides a prototype.
2450 Use this option to detect global functions that are not declared in
2453 @item -Wmissing-noreturn
2454 @opindex Wmissing-noreturn
2455 Warn about functions which might be candidates for attribute @code{noreturn}.
2456 Note these are only possible candidates, not absolute ones. Care should
2457 be taken to manually verify functions actually do not ever return before
2458 adding the @code{noreturn} attribute, otherwise subtle code generation
2459 bugs could be introduced. You will not get a warning for @code{main} in
2460 hosted C environments.
2462 @item -Wmissing-format-attribute
2463 @opindex Wmissing-format-attribute
2465 If @option{-Wformat} is enabled, also warn about functions which might be
2466 candidates for @code{format} attributes. Note these are only possible
2467 candidates, not absolute ones. GCC will guess that @code{format}
2468 attributes might be appropriate for any function that calls a function
2469 like @code{vprintf} or @code{vscanf}, but this might not always be the
2470 case, and some functions for which @code{format} attributes are
2471 appropriate may not be detected. This option has no effect unless
2472 @option{-Wformat} is enabled (possibly by @option{-Wall}).
2476 Warn if a structure is given the packed attribute, but the packed
2477 attribute has no effect on the layout or size of the structure.
2478 Such structures may be mis-aligned for little benefit. For
2479 instance, in this code, the variable @code{f.x} in @code{struct bar}
2480 will be misaligned even though @code{struct bar} does not itself
2481 have the packed attribute:
2488 @} __attribute__((packed));
2498 Warn if padding is included in a structure, either to align an element
2499 of the structure or to align the whole structure. Sometimes when this
2500 happens it is possible to rearrange the fields of the structure to
2501 reduce the padding and so make the structure smaller.
2503 @item -Wredundant-decls
2504 @opindex Wredundant-decls
2505 Warn if anything is declared more than once in the same scope, even in
2506 cases where multiple declaration is valid and changes nothing.
2508 @item -Wnested-externs @r{(C only)}
2509 @opindex Wnested-externs
2510 Warn if an @code{extern} declaration is encountered within a function.
2512 @item -Wunreachable-code
2513 @opindex Wunreachable-code
2514 Warn if the compiler detects that code will never be executed.
2516 This option is intended to warn when the compiler detects that at
2517 least a whole line of source code will never be executed, because
2518 some condition is never satisfied or because it is after a
2519 procedure that never returns.
2521 It is possible for this option to produce a warning even though there
2522 are circumstances under which part of the affected line can be executed,
2523 so care should be taken when removing apparently-unreachable code.
2525 For instance, when a function is inlined, a warning may mean that the
2526 line is unreachable in only one inlined copy of the function.
2528 This option is not made part of @option{-Wall} because in a debugging
2529 version of a program there is often substantial code which checks
2530 correct functioning of the program and is, hopefully, unreachable
2531 because the program does work. Another common use of unreachable
2532 code is to provide behaviour which is selectable at compile-time.
2536 Warn if a function can not be inlined and it was declared as inline.
2540 @opindex Wno-long-long
2541 Warn if @samp{long long} type is used. This is default. To inhibit
2542 the warning messages, use @option{-Wno-long-long}. Flags
2543 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
2544 only when @option{-pedantic} flag is used.
2546 @item -Wdisabled-optimization
2547 @opindex Wdisabled-optimization
2548 Warn if a requested optimization pass is disabled. This warning does
2549 not generally indicate that there is anything wrong with your code; it
2550 merely indicates that GCC's optimizers were unable to handle the code
2551 effectively. Often, the problem is that your code is too big or too
2552 complex; GCC will refuse to optimize programs when the optimization
2553 itself is likely to take inordinate amounts of time.
2557 Make all warnings into errors.
2560 @node Debugging Options
2561 @section Options for Debugging Your Program or GCC
2562 @cindex options, debugging
2563 @cindex debugging information options
2565 GCC has various special options that are used for debugging
2566 either your program or GCC:
2571 Produce debugging information in the operating system's native format
2572 (stabs, COFF, XCOFF, or DWARF)@. GDB can work with this debugging
2575 On most systems that use stabs format, @option{-g} enables use of extra
2576 debugging information that only GDB can use; this extra information
2577 makes debugging work better in GDB but will probably make other debuggers
2579 refuse to read the program. If you want to control for certain whether
2580 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
2581 @option{-gxcoff+}, @option{-gxcoff}, @option{-gdwarf-1+}, or @option{-gdwarf-1}
2584 Unlike most other C compilers, GCC allows you to use @option{-g} with
2585 @option{-O}. The shortcuts taken by optimized code may occasionally
2586 produce surprising results: some variables you declared may not exist
2587 at all; flow of control may briefly move where you did not expect it;
2588 some statements may not be executed because they compute constant
2589 results or their values were already at hand; some statements may
2590 execute in different places because they were moved out of loops.
2592 Nevertheless it proves possible to debug optimized output. This makes
2593 it reasonable to use the optimizer for programs that might have bugs.
2595 The following options are useful when GCC is generated with the
2596 capability for more than one debugging format.
2600 Produce debugging information for use by GDB@. This means to use the
2601 most expressive format available (DWARF 2, stabs, or the native format
2602 if neither of those are supported), including GDB extensions if at all
2607 Produce debugging information in stabs format (if that is supported),
2608 without GDB extensions. This is the format used by DBX on most BSD
2609 systems. On MIPS, Alpha and System V Release 4 systems this option
2610 produces stabs debugging output which is not understood by DBX or SDB@.
2611 On System V Release 4 systems this option requires the GNU assembler.
2615 Produce debugging information in stabs format (if that is supported),
2616 using GNU extensions understood only by the GNU debugger (GDB)@. The
2617 use of these extensions is likely to make other debuggers crash or
2618 refuse to read the program.
2622 Produce debugging information in COFF format (if that is supported).
2623 This is the format used by SDB on most System V systems prior to
2628 Produce debugging information in XCOFF format (if that is supported).
2629 This is the format used by the DBX debugger on IBM RS/6000 systems.
2633 Produce debugging information in XCOFF format (if that is supported),
2634 using GNU extensions understood only by the GNU debugger (GDB)@. The
2635 use of these extensions is likely to make other debuggers crash or
2636 refuse to read the program, and may cause assemblers other than the GNU
2637 assembler (GAS) to fail with an error.
2641 Produce debugging information in DWARF version 1 format (if that is
2642 supported). This is the format used by SDB on most System V Release 4
2647 Produce debugging information in DWARF version 1 format (if that is
2648 supported), using GNU extensions understood only by the GNU debugger
2649 (GDB)@. The use of these extensions is likely to make other debuggers
2650 crash or refuse to read the program.
2654 Produce debugging information in DWARF version 2 format (if that is
2655 supported). This is the format used by DBX on IRIX 6.
2658 @itemx -ggdb@var{level}
2659 @itemx -gstabs@var{level}
2660 @itemx -gcoff@var{level}
2661 @itemx -gxcoff@var{level}
2662 @itemx -gdwarf@var{level}
2663 @itemx -gdwarf-2@var{level}
2664 Request debugging information and also use @var{level} to specify how
2665 much information. The default level is 2.
2667 Level 1 produces minimal information, enough for making backtraces in
2668 parts of the program that you don't plan to debug. This includes
2669 descriptions of functions and external variables, but no information
2670 about local variables and no line numbers.
2672 Level 3 includes extra information, such as all the macro definitions
2673 present in the program. Some debuggers support macro expansion when
2674 you use @option{-g3}.
2679 Generate extra code to write profile information suitable for the
2680 analysis program @code{prof}. You must use this option when compiling
2681 the source files you want data about, and you must also use it when
2684 @cindex @code{gprof}
2687 Generate extra code to write profile information suitable for the
2688 analysis program @code{gprof}. You must use this option when compiling
2689 the source files you want data about, and you must also use it when
2695 Generate extra code to write profile information for basic blocks, which will
2696 record the number of times each basic block is executed, the basic block start
2697 address, and the function name containing the basic block. If @option{-g} is
2698 used, the line number and filename of the start of the basic block will also be
2699 recorded. If not overridden by the machine description, the default action is
2700 to append to the text file @file{bb.out}.
2702 This data could be analyzed by a program like @code{tcov}. Note,
2703 however, that the format of the data is not what @code{tcov} expects.
2704 Eventually GNU @code{gprof} should be extended to process this data.
2708 Makes the compiler print out each function name as it is compiled, and
2709 print some statistics about each pass when it finishes.
2712 @opindex ftime-report
2713 Makes the compiler print some statistics about the time consumed by each
2714 pass when it finishes.
2717 @opindex fmem-report
2718 Makes the compiler print some statistics about permanent memory
2719 allocation when it finishes.
2723 Generate extra code to profile basic blocks. Your executable will
2724 produce output that is a superset of that produced when @option{-a} is
2725 used. Additional output is the source and target address of the basic
2726 blocks where a jump takes place, the number of times a jump is executed,
2727 and (optionally) the complete sequence of basic blocks being executed.
2728 The output is appended to file @file{bb.out}.
2730 You can examine different profiling aspects without recompilation. Your
2731 executable will read a list of function names from file @file{bb.in}.
2732 Profiling starts when a function on the list is entered and stops when
2733 that invocation is exited. To exclude a function from profiling, prefix
2734 its name with @samp{-}. If a function name is not unique, you can
2735 disambiguate it by writing it in the form
2736 @samp{/path/filename.d:functionname}. Your executable will write the
2737 available paths and filenames in file @file{bb.out}.
2739 Several function names have a special meaning:
2742 Write source, target and frequency of jumps to file @file{bb.out}.
2743 @item __bb_hidecall__
2744 Exclude function calls from frequency count.
2745 @item __bb_showret__
2746 Include function returns in frequency count.
2748 Write the sequence of basic blocks executed to file @file{bbtrace.gz}.
2749 The file will be compressed using the program @samp{gzip}, which must
2750 exist in your @env{PATH}. On systems without the @samp{popen}
2751 function, the file will be named @file{bbtrace} and will not be
2752 compressed. @strong{Profiling for even a few seconds on these systems
2753 will produce a very large file.} Note: @code{__bb_hidecall__} and
2754 @code{__bb_showret__} will not affect the sequence written to
2758 Here's a short example using different profiling parameters
2759 in file @file{bb.in}. Assume function @code{foo} consists of basic blocks
2760 1 and 2 and is called twice from block 3 of function @code{main}. After
2761 the calls, block 3 transfers control to block 4 of @code{main}.
2763 With @code{__bb_trace__} and @code{main} contained in file @file{bb.in},
2764 the following sequence of blocks is written to file @file{bbtrace.gz}:
2765 0 3 1 2 1 2 4. The return from block 2 to block 3 is not shown, because
2766 the return is to a point inside the block and not to the top. The
2767 block address 0 always indicates, that control is transferred
2768 to the trace from somewhere outside the observed functions. With
2769 @samp{-foo} added to @file{bb.in}, the blocks of function
2770 @code{foo} are removed from the trace, so only 0 3 4 remains.
2772 With @code{__bb_jumps__} and @code{main} contained in file @file{bb.in},
2773 jump frequencies will be written to file @file{bb.out}. The
2774 frequencies are obtained by constructing a trace of blocks
2775 and incrementing a counter for every neighbouring pair of blocks
2776 in the trace. The trace 0 3 1 2 1 2 4 displays the following
2780 Jump from block 0x0 to block 0x3 executed 1 time(s)
2781 Jump from block 0x3 to block 0x1 executed 1 time(s)
2782 Jump from block 0x1 to block 0x2 executed 2 time(s)
2783 Jump from block 0x2 to block 0x1 executed 1 time(s)
2784 Jump from block 0x2 to block 0x4 executed 1 time(s)
2787 With @code{__bb_hidecall__}, control transfer due to call instructions
2788 is removed from the trace, that is the trace is cut into three parts: 0
2789 3 4, 0 1 2 and 0 1 2. With @code{__bb_showret__}, control transfer due
2790 to return instructions is added to the trace. The trace becomes: 0 3 1
2791 2 3 1 2 3 4. Note, that this trace is not the same, as the sequence
2792 written to @file{bbtrace.gz}. It is solely used for counting jump
2795 @item -fprofile-arcs
2796 @opindex fprofile-arcs
2797 Instrument @dfn{arcs} during compilation to generate coverage data
2798 or for profile-directed block ordering. During execution the program
2799 records how many times each branch is executed and how many times it is
2800 taken. When the compiled program exits it saves this data to a file
2801 called @file{@var{sourcename}.da} for each source file.
2803 For profile-directed block ordering, compile the program with
2804 @option{-fprofile-arcs} plus optimization and code generation options,
2805 generate the arc profile information by running the program on a
2806 selected workload, and then compile the program again with the same
2807 optimization and code generation options plus
2808 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
2809 Control Optimization}).
2811 The other use of @option{-fprofile-arcs} is for use with @code{gcov},
2812 when it is used with the @option{-ftest-coverage} option. GCC
2813 supports two methods of determining code coverage: the options that
2814 support @code{gcov}, and options @option{-a} and @option{-ax}, which
2815 write information to text files. The options that support @code{gcov}
2816 do not need to instrument every arc in the program, so a program compiled
2817 with them runs faster than a program compiled with @option{-a}, which
2818 adds instrumentation code to every basic block in the program. The
2819 tradeoff: since @code{gcov} does not have execution counts for all
2820 branches, it must start with the execution counts for the instrumented
2821 branches, and then iterate over the program flow graph until the entire
2822 graph has been solved. Hence, @code{gcov} runs a little more slowly than
2823 a program which uses information from @option{-a} and @option{-ax}.
2825 With @option{-fprofile-arcs}, for each function of your program GCC
2826 creates a program flow graph, then finds a spanning tree for the graph.
2827 Only arcs that are not on the spanning tree have to be instrumented: the
2828 compiler adds code to count the number of times that these arcs are
2829 executed. When an arc is the only exit or only entrance to a block, the
2830 instrumentation code can be added to the block; otherwise, a new basic
2831 block must be created to hold the instrumentation code.
2833 This option makes it possible to estimate branch probabilities and to
2834 calculate basic block execution counts. In general, basic block
2835 execution counts as provided by @option{-a} do not give enough
2836 information to estimate all branch probabilities.
2839 @item -ftest-coverage
2840 @opindex ftest-coverage
2841 Create data files for the @code{gcov} code-coverage utility
2842 (@pxref{Gcov,, @code{gcov}: a GCC Test Coverage Program}).
2843 The data file names begin with the name of your source file:
2846 @item @var{sourcename}.bb
2847 A mapping from basic blocks to line numbers, which @code{gcov} uses to
2848 associate basic block execution counts with line numbers.
2850 @item @var{sourcename}.bbg
2851 A list of all arcs in the program flow graph. This allows @code{gcov}
2852 to reconstruct the program flow graph, so that it can compute all basic
2853 block and arc execution counts from the information in the
2854 @code{@var{sourcename}.da} file.
2857 Use @option{-ftest-coverage} with @option{-fprofile-arcs}; the latter
2858 option adds instrumentation to the program, which then writes
2859 execution counts to another data file:
2862 @item @var{sourcename}.da
2863 Runtime arc execution counts, used in conjunction with the arc
2864 information in the file @code{@var{sourcename}.bbg}.
2867 Coverage data will map better to the source files if
2868 @option{-ftest-coverage} is used without optimization.
2870 @item -d@var{letters}
2872 Says to make debugging dumps during compilation at times specified by
2873 @var{letters}. This is used for debugging the compiler. The file names
2874 for most of the dumps are made by appending a pass number and a word to
2875 the source file name (e.g. @file{foo.c.00.rtl} or @file{foo.c.01.sibling}).
2876 Here are the possible letters for use in @var{letters}, and their meanings:
2881 Annotate the assembler output with miscellaneous debugging information.
2884 Dump after computing branch probabilities, to @file{@var{file}.14.bp}.
2887 Dump after block reordering, to @file{@var{file}.28.bbro}.
2890 Dump after instruction combination, to the file @file{@var{file}.16.combine}.
2893 Dump after the first if conversion, to the file @file{@var{file}.17.ce}.
2896 Dump after delayed branch scheduling, to @file{@var{file}.31.dbr}.
2899 Dump all macro definitions, at the end of preprocessing, in addition to
2903 Dump after SSA optimizations, to @file{@var{file}.04.ssa} and
2904 @file{@var{file}.07.ussa}.
2907 Dump after the second if conversion, to @file{@var{file}.26.ce2}.
2910 Dump after life analysis, to @file{@var{file}.15.life}.
2913 Dump after purging @code{ADDRESSOF} codes, to @file{@var{file}.09.addressof}.
2916 Dump after global register allocation, to @file{@var{file}.21.greg}.
2919 Dump after finalization of EH handling code, to @file{@var{file}.02.eh}.
2922 Dump after post-reload optimizations, to @file{@var{file}.22.postreload}.
2925 Dump after GCSE, to @file{@var{file}.10.gcse}.
2928 Dump after sibling call optimizations, to @file{@var{file}.01.sibling}.
2931 Dump after the first jump optimization, to @file{@var{file}.03.jump}.
2934 Dump after conversion from registers to stack, to @file{@var{file}.32.stack}.
2937 Dump after local register allocation, to @file{@var{file}.20.lreg}.
2940 Dump after loop optimization, to @file{@var{file}.11.loop}.
2943 Dump after performing the machine dependent reorganisation pass, to
2944 @file{@var{file}.30.mach}.
2947 Dump after register renumbering, to @file{@var{file}.25.rnreg}.
2950 Dump after the register move pass, to @file{@var{file}.18.regmove}.
2953 Dump after RTL generation, to @file{@var{file}.00.rtl}.
2956 Dump after the second scheduling pass, to @file{@var{file}.27.sched2}.
2959 Dump after CSE (including the jump optimization that sometimes follows
2960 CSE), to @file{@var{file}.08.cse}.
2963 Dump after the first scheduling pass, to @file{@var{file}.19.sched}.
2966 Dump after the second CSE pass (including the jump optimization that
2967 sometimes follows CSE), to @file{@var{file}.12.cse2}.
2970 Dump after the second flow pass, to @file{@var{file}.23.flow2}.
2973 Dump after SSA dead code elimination, to @file{@var{file}.06.ssadce}.
2976 Dump after the peephole pass, to @file{@var{file}.24.peephole2}.
2979 Produce all the dumps listed above.
2982 Print statistics on memory usage, at the end of the run, to
2986 Annotate the assembler output with a comment indicating which
2987 pattern and alternative was used. The length of each instruction is
2991 Dump the RTL in the assembler output as a comment before each instruction.
2992 Also turns on @option{-dp} annotation.
2995 For each of the other indicated dump files (except for
2996 @file{@var{file}.00.rtl}), dump a representation of the control flow graph
2997 suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
3000 Just generate RTL for a function instead of compiling it. Usually used
3004 Dump debugging information during parsing, to standard error.
3007 @item -fdump-unnumbered
3008 @opindex fdump-unnumbered
3009 When doing debugging dumps (see @option{-d} option above), suppress instruction
3010 numbers and line number note output. This makes it more feasible to
3011 use diff on debugging dumps for compiler invocations with different
3012 options, in particular with and without @option{-g}.
3014 @item -fdump-class-hierarchy @r{(C++ only)}
3015 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
3016 @opindex fdump-class-hierarchy
3017 Dump a representation of each class's hierarchy and virtual function
3018 table layout to a file. The file name is made by appending @file{.class}
3019 to the source file name. If the @samp{-@var{options}} form is used,
3020 @var{options} controls the details of the dump as described for the
3021 @option{-fdump-tree} options.
3023 @item -fdump-tree-@var{switch} @r{(C++ only)}
3024 @itemx -fdump-tree-@var{switch}-@var{options} @r{(C++ only)}
3026 Control the dumping at various stages of processing the intermediate
3027 language tree to a file. The file name is generated by appending a switch
3028 specific suffix to the source file name. If the @samp{-@var{options}}
3029 form is used, @var{options} is a list of @samp{-} separated options that
3030 control the details of the dump. Not all options are applicable to all
3031 dumps, those which are not meaningful will be ignored. The following
3032 options are available
3036 Print the address of each node. Usually this is not meaningful as it
3037 changes according to the environment and source file. Its primary use
3038 is for tying up a dump file with a debug environment.
3040 Inhibit dumping of members of a scope or body of a function merely
3041 because that scope has been reached. Only dump such items when they
3042 are directly reachable by some other path.
3044 Turn on all options.
3047 The following tree dumps are possible:
3050 Dump before any tree based optimization, to @file{@var{file}.original}.
3052 Dump after all tree based optimization, to @file{@var{file}.optimized}.
3054 Dump after function inlining, to @file{@var{file}.inlined}.
3057 @item -fpretend-float
3058 @opindex fpretend-float
3059 When running a cross-compiler, pretend that the target machine uses the
3060 same floating point format as the host machine. This causes incorrect
3061 output of the actual floating constants, but the actual instruction
3062 sequence will probably be the same as GCC would make when running on
3067 Store the usual ``temporary'' intermediate files permanently; place them
3068 in the current directory and name them based on the source file. Thus,
3069 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
3070 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
3071 preprocessed @file{foo.i} output file even though the compiler now
3072 normally uses an integrated preprocessor.
3076 Report the CPU time taken by each subprocess in the compilation
3077 sequence. For C source files, this is the compiler proper and assembler
3078 (plus the linker if linking is done). The output looks like this:
3085 The first number on each line is the ``user time,'' that is time spent
3086 executing the program itself. The second number is ``system time,''
3087 time spent executing operating system routines on behalf of the program.
3088 Both numbers are in seconds.
3090 @item -print-file-name=@var{library}
3091 @opindex print-file-name
3092 Print the full absolute name of the library file @var{library} that
3093 would be used when linking---and don't do anything else. With this
3094 option, GCC does not compile or link anything; it just prints the
3097 @item -print-multi-directory
3098 @opindex print-multi-directory
3099 Print the directory name corresponding to the multilib selected by any
3100 other switches present in the command line. This directory is supposed
3101 to exist in @env{GCC_EXEC_PREFIX}.
3103 @item -print-multi-lib
3104 @opindex print-multi-lib
3105 Print the mapping from multilib directory names to compiler switches
3106 that enable them. The directory name is separated from the switches by
3107 @samp{;}, and each switch starts with an @samp{@@} instead of the
3108 @samp{-}, without spaces between multiple switches. This is supposed to
3109 ease shell-processing.
3111 @item -print-prog-name=@var{program}
3112 @opindex print-prog-name
3113 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
3115 @item -print-libgcc-file-name
3116 @opindex print-libgcc-file-name
3117 Same as @option{-print-file-name=libgcc.a}.
3119 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
3120 but you do want to link with @file{libgcc.a}. You can do
3123 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
3126 @item -print-search-dirs
3127 @opindex print-search-dirs
3128 Print the name of the configured installation directory and a list of
3129 program and library directories gcc will search---and don't do anything else.
3131 This is useful when gcc prints the error message
3132 @samp{installation problem, cannot exec cpp0: No such file or directory}.
3133 To resolve this you either need to put @file{cpp0} and the other compiler
3134 components where gcc expects to find them, or you can set the environment
3135 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
3136 Don't forget the trailing '/'.
3137 @xref{Environment Variables}.
3140 @opindex dumpmachine
3141 Print the compiler's target machine (for example,
3142 @samp{i686-pc-linux-gnu})---and don't do anything else.
3145 @opindex dumpversion
3146 Print the compiler version (for example, @samp{3.0})---and don't do
3151 Print the compiler's built-in specs---and don't do anything else. (This
3152 is used when GCC itself is being built.) @xref{Spec Files}.
3155 @node Optimize Options
3156 @section Options That Control Optimization
3157 @cindex optimize options
3158 @cindex options, optimization
3160 These options control various sorts of optimizations:
3167 Optimize. Optimizing compilation takes somewhat more time, and a lot
3168 more memory for a large function.
3170 Without @option{-O}, the compiler's goal is to reduce the cost of
3171 compilation and to make debugging produce the expected results.
3172 Statements are independent: if you stop the program with a breakpoint
3173 between statements, you can then assign a new value to any variable or
3174 change the program counter to any other statement in the function and
3175 get exactly the results you would expect from the source code.
3177 With @option{-O}, the compiler tries to reduce code size and execution
3178 time, without performing any optimizations that take a great deal of
3183 Optimize even more. GCC performs nearly all supported optimizations
3184 that do not involve a space-speed tradeoff. The compiler does not
3185 perform loop unrolling or function inlining when you specify @option{-O2}.
3186 As compared to @option{-O}, this option increases both compilation time
3187 and the performance of the generated code.
3189 @option{-O2} turns on all optional optimizations except for loop unrolling,
3190 function inlining, and register renaming. It also turns on the
3191 @option{-fforce-mem} option on all machines and frame pointer elimination
3192 on machines where doing so does not interfere with debugging.
3194 Please note the warning under @option{-fgcse} about
3195 invoking @option{-O2} on programs that use computed gotos.
3199 Optimize yet more. @option{-O3} turns on all optimizations specified by
3200 @option{-O2} and also turns on the @option{-finline-functions} and
3201 @option{-frename-registers} options.
3209 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
3210 do not typically increase code size. It also performs further
3211 optimizations designed to reduce code size.
3213 If you use multiple @option{-O} options, with or without level numbers,
3214 the last such option is the one that is effective.
3217 Options of the form @option{-f@var{flag}} specify machine-independent
3218 flags. Most flags have both positive and negative forms; the negative
3219 form of @option{-ffoo} would be @option{-fno-foo}. In the table below,
3220 only one of the forms is listed---the one which is not the default.
3221 You can figure out the other form by either removing @samp{no-} or
3226 @opindex ffloat-store
3227 Do not store floating point variables in registers, and inhibit other
3228 options that might change whether a floating point value is taken from a
3231 @cindex floating point precision
3232 This option prevents undesirable excess precision on machines such as
3233 the 68000 where the floating registers (of the 68881) keep more
3234 precision than a @code{double} is supposed to have. Similarly for the
3235 x86 architecture. For most programs, the excess precision does only
3236 good, but a few programs rely on the precise definition of IEEE floating
3237 point. Use @option{-ffloat-store} for such programs, after modifying
3238 them to store all pertinent intermediate computations into variables.
3240 @item -fno-default-inline
3241 @opindex fno-default-inline
3242 Do not make member functions inline by default merely because they are
3243 defined inside the class scope (C++ only). Otherwise, when you specify
3244 @w{@option{-O}}, member functions defined inside class scope are compiled
3245 inline by default; i.e., you don't need to add @samp{inline} in front of
3246 the member function name.
3248 @item -fno-defer-pop
3249 @opindex fno-defer-pop
3250 Always pop the arguments to each function call as soon as that function
3251 returns. For machines which must pop arguments after a function call,
3252 the compiler normally lets arguments accumulate on the stack for several
3253 function calls and pops them all at once.
3257 Force memory operands to be copied into registers before doing
3258 arithmetic on them. This produces better code by making all memory
3259 references potential common subexpressions. When they are not common
3260 subexpressions, instruction combination should eliminate the separate
3261 register-load. The @option{-O2} option turns on this option.
3264 @opindex fforce-addr
3265 Force memory address constants to be copied into registers before
3266 doing arithmetic on them. This may produce better code just as
3267 @option{-fforce-mem} may.
3269 @item -fomit-frame-pointer
3270 @opindex fomit-frame-pointer
3271 Don't keep the frame pointer in a register for functions that
3272 don't need one. This avoids the instructions to save, set up and
3273 restore frame pointers; it also makes an extra register available
3274 in many functions. @strong{It also makes debugging impossible on
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}.
3285 On some machines, such as the VAX, this flag has no effect, because
3286 the standard calling sequence automatically handles the frame pointer
3287 and nothing is saved by pretending it doesn't exist. The
3288 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
3289 whether a target machine supports this flag. @xref{Registers,,Register
3290 Usage, gcc.info, Using and Porting GCC}.
3293 @item -foptimize-sibling-calls
3294 @opindex foptimize-sibling-calls
3295 Optimize sibling and tail recursive calls.
3299 This option generates traps for signed overflow on addition, subtraction,
3300 multiplication operations.
3304 Don't pay attention to the @code{inline} keyword. Normally this option
3305 is used to keep the compiler from expanding any functions inline.
3306 Note that if you are not optimizing, no functions can be expanded inline.
3308 @item -finline-functions
3309 @opindex finline-functions
3310 Integrate all simple functions into their callers. The compiler
3311 heuristically decides which functions are simple enough to be worth
3312 integrating in this way.
3314 If all calls to a given function are integrated, and the function is
3315 declared @code{static}, then the function is normally not output as
3316 assembler code in its own right.
3318 @item -finline-limit=@var{n}
3319 @opindex finline-limit
3320 By default, gcc limits the size of functions that can be inlined. This flag
3321 allows the control of this limit for functions that are explicitly marked as
3322 inline (ie marked with the inline keyword or defined within the class
3323 definition in c++). @var{n} is the size of functions that can be inlined in
3324 number of pseudo instructions (not counting parameter handling). The default
3325 value of @var{n} is 600.
3326 Increasing this value can result in more inlined code at
3327 the cost of compilation time and memory consumption. Decreasing usually makes
3328 the compilation faster and less code will be inlined (which presumably
3329 means slower programs). This option is particularly useful for programs that
3330 use inlining heavily such as those based on recursive templates with C++.
3332 @emph{Note:} pseudo instruction represents, in this particular context, an
3333 abstract measurement of function's size. In no way, it represents a count
3334 of assembly instructions and as such its exact meaning might change from one
3335 release to an another.
3337 @item -fkeep-inline-functions
3338 @opindex fkeep-inline-functions
3339 Even if all calls to a given function are integrated, and the function
3340 is declared @code{static}, nevertheless output a separate run-time
3341 callable version of the function. This switch does not affect
3342 @code{extern inline} functions.
3344 @item -fkeep-static-consts
3345 @opindex fkeep-static-consts
3346 Emit variables declared @code{static const} when optimization isn't turned
3347 on, even if the variables aren't referenced.
3349 GCC enables this option by default. If you want to force the compiler to
3350 check if the variable was referenced, regardless of whether or not
3351 optimization is turned on, use the @option{-fno-keep-static-consts} option.
3353 @item -fmerge-constants
3354 Attempt to merge identical constants (string constants and floating point
3355 constants) accross compilation units.
3357 This option is default for optimized compilation if assembler and linker
3358 support it. Use @option{-fno-merge-constants} to inhibit this behaviour.
3360 @item -fmerge-all-constants
3361 Attempt to merge identical constants and identical variables.
3363 This option implies @option{-fmerge-constants}. In addition to
3364 @option{-fmerge-constants} this considers e.g. even constant initialized
3365 arrays or initialized constant variables with integral or floating point
3366 types. Languages like C or C++ require each non-automatic variable to
3367 have distinct location, so using this option will result in non-conforming
3370 @item -fno-function-cse
3371 @opindex fno-function-cse
3372 Do not put function addresses in registers; make each instruction that
3373 calls a constant function contain the function's address explicitly.
3375 This option results in less efficient code, but some strange hacks
3376 that alter the assembler output may be confused by the optimizations
3377 performed when this option is not used.
3381 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations}, and @*
3382 @option{-fno-trapping-math}.
3384 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
3386 This option should never be turned on by any @option{-O} option since
3387 it can result in incorrect output for programs which depend on
3388 an exact implementation of IEEE or ISO rules/specifications for
3391 @item -fno-math-errno
3392 @opindex fno-math-errno
3393 Do not set ERRNO after calling math functions that are executed
3394 with a single instruction, e.g., sqrt. A program that relies on
3395 IEEE exceptions for math error handling may want to use this flag
3396 for speed while maintaining IEEE arithmetic compatibility.
3398 This option should never be turned on by any @option{-O} option since
3399 it can result in incorrect output for programs which depend on
3400 an exact implementation of IEEE or ISO rules/specifications for
3403 The default is @option{-fmath-errno}.
3405 @item -funsafe-math-optimizations
3406 @opindex funsafe-math-optimizations
3407 Allow optimizations for floating-point arithmetic that (a) assume
3408 that arguments and results are valid and (b) may violate IEEE or
3409 ANSI standards. When used at link-time, it may include libraries
3410 or startup files that change the default FPU control word or other
3411 similar optimizations.
3413 This option should never be turned on by any @option{-O} option since
3414 it can result in incorrect output for programs which depend on
3415 an exact implementation of IEEE or ISO rules/specifications for
3418 The default is @option{-fno-unsafe-math-optimizations}.
3420 @item -fno-trapping-math
3421 @opindex fno-trapping-math
3422 Compile code assuming that floating-point operations cannot generate
3423 user-visible traps. Setting this option may allow faster code
3424 if one relies on ``non-stop'' IEEE arithmetic, for example.
3426 This option should never be turned on by any @option{-O} option since
3427 it can result in incorrect output for programs which depend on
3428 an exact implementation of IEEE or ISO rules/specifications for
3431 The default is @option{-ftrapping-math}.
3434 The following options control specific optimizations. The @option{-O2}
3435 option turns on all of these optimizations except @option{-funroll-loops}
3436 and @option{-funroll-all-loops}. On most machines, the @option{-O} option
3437 turns on the @option{-fthread-jumps} and @option{-fdelayed-branch} options,
3438 but specific machines may handle it differently.
3440 You can use the following flags in the rare cases when ``fine-tuning''
3441 of optimizations to be performed is desired.
3443 Not all of the optimizations performed by GCC have @option{-f} options
3447 @item -fstrength-reduce
3448 @opindex fstrength-reduce
3449 Perform the optimizations of loop strength reduction and
3450 elimination of iteration variables.
3452 @item -fthread-jumps
3453 @opindex fthread-jumps
3454 Perform optimizations where we check to see if a jump branches to a
3455 location where another comparison subsumed by the first is found. If
3456 so, the first branch is redirected to either the destination of the
3457 second branch or a point immediately following it, depending on whether
3458 the condition is known to be true or false.
3460 @item -fcse-follow-jumps
3461 @opindex fcse-follow-jumps
3462 In common subexpression elimination, scan through jump instructions
3463 when the target of the jump is not reached by any other path. For
3464 example, when CSE encounters an @code{if} statement with an
3465 @code{else} clause, CSE will follow the jump when the condition
3468 @item -fcse-skip-blocks
3469 @opindex fcse-skip-blocks
3470 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
3471 follow jumps which conditionally skip over blocks. When CSE
3472 encounters a simple @code{if} statement with no else clause,
3473 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
3474 body of the @code{if}.
3476 @item -frerun-cse-after-loop
3477 @opindex frerun-cse-after-loop
3478 Re-run common subexpression elimination after loop optimizations has been
3481 @item -frerun-loop-opt
3482 @opindex frerun-loop-opt
3483 Run the loop optimizer twice.
3487 Perform a global common subexpression elimination pass.
3488 This pass also performs global constant and copy propagation.
3490 @emph{Note:} When compiling a program using computed gotos, a GCC
3491 extension, you may get better runtime performance if you disable
3492 the global common subexpression elmination pass by adding
3493 @option{-fno-gcse} to the command line.
3497 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
3498 attempt to move loads which are only killed by stores into themselves. This
3499 allows a loop containing a load/store sequence to be changed to a load outside
3500 the loop, and a copy/store within the loop.
3504 When @option{-fgcse-sm} is enabled, A store motion pass is run after global common
3505 subexpression elimination. This pass will attempt to move stores out of loops.
3506 When used in conjunction with @option{-fgcse-lm}, loops containing a load/store sequence
3507 can be changed to a load before the loop and a store after the loop.
3509 @item -fdelete-null-pointer-checks
3510 @opindex fdelete-null-pointer-checks
3511 Use global dataflow analysis to identify and eliminate useless checks
3512 for null pointers. The compiler assumes that dereferencing a null
3513 pointer would have halted the program. If a pointer is checked after
3514 it has already been dereferenced, it cannot be null.
3516 In some environments, this assumption is not true, and programs can
3517 safely dereference null pointers. Use
3518 @option{-fno-delete-null-pointer-checks} to disable this optimization
3519 for programs which depend on that behavior.
3521 @item -fexpensive-optimizations
3522 @opindex fexpensive-optimizations
3523 Perform a number of minor optimizations that are relatively expensive.
3525 @item -foptimize-register-move
3527 @opindex foptimize-register-move
3529 Attempt to reassign register numbers in move instructions and as
3530 operands of other simple instructions in order to maximize the amount of
3531 register tying. This is especially helpful on machines with two-operand
3532 instructions. GCC enables this optimization by default with @option{-O2}
3535 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
3538 @item -fdelayed-branch
3539 @opindex fdelayed-branch
3540 If supported for the target machine, attempt to reorder instructions
3541 to exploit instruction slots available after delayed branch
3544 @item -fschedule-insns
3545 @opindex fschedule-insns
3546 If supported for the target machine, attempt to reorder instructions to
3547 eliminate execution stalls due to required data being unavailable. This
3548 helps machines that have slow floating point or memory load instructions
3549 by allowing other instructions to be issued until the result of the load
3550 or floating point instruction is required.
3552 @item -fschedule-insns2
3553 @opindex fschedule-insns2
3554 Similar to @option{-fschedule-insns}, but requests an additional pass of
3555 instruction scheduling after register allocation has been done. This is
3556 especially useful on machines with a relatively small number of
3557 registers and where memory load instructions take more than one cycle.
3559 @item -ffunction-sections
3560 @itemx -fdata-sections
3561 @opindex ffunction-sections
3562 @opindex fdata-sections
3563 Place each function or data item into its own section in the output
3564 file if the target supports arbitrary sections. The name of the
3565 function or the name of the data item determines the section's name
3568 Use these options on systems where the linker can perform optimizations
3569 to improve locality of reference in the instruction space. HPPA
3570 processors running HP-UX and Sparc processors running Solaris 2 have
3571 linkers with such optimizations. Other systems using the ELF object format
3572 as well as AIX may have these optimizations in the future.
3574 Only use these options when there are significant benefits from doing
3575 so. When you specify these options, the assembler and linker will
3576 create larger object and executable files and will also be slower.
3577 You will not be able to use @code{gprof} on all systems if you
3578 specify this option and you may have problems with debugging if
3579 you specify both this option and @option{-g}.
3581 @item -fcaller-saves
3582 @opindex fcaller-saves
3583 Enable values to be allocated in registers that will be clobbered by
3584 function calls, by emitting extra instructions to save and restore the
3585 registers around such calls. Such allocation is done only when it
3586 seems to result in better code than would otherwise be produced.
3588 This option is always enabled by default on certain machines, usually
3589 those which have no call-preserved registers to use instead.
3591 For all machines, optimization level 2 and higher enables this flag by
3594 @item -funroll-loops
3595 @opindex funroll-loops
3596 Unroll loops whose number of iterations can be determined at compile
3597 time or upon entry to the loop. @option{-funroll-loops} implies both
3598 @option{-fstrength-reduce} and @option{-frerun-cse-after-loop}. This
3599 option makes code larger, and may or may not make it run faster.
3601 @item -funroll-all-loops
3602 @opindex funroll-all-loops
3603 Unroll all loops, even if their number of iterations is uncertain when
3604 the loop is entered. This usually makes programs run more slowly.
3605 @option{-funroll-all-loops} implies the same options as
3606 @option{-funroll-loops},
3609 @item -fmove-all-movables
3610 @opindex fmove-all-movables
3611 Forces all invariant computations in loops to be moved
3614 @item -freduce-all-givs
3615 @opindex freduce-all-givs
3616 Forces all general-induction variables in loops to be
3619 @emph{Note:} When compiling programs written in Fortran,
3620 @option{-fmove-all-movables} and @option{-freduce-all-givs} are enabled
3621 by default when you use the optimizer.
3623 These options may generate better or worse code; results are highly
3624 dependent on the structure of loops within the source code.
3626 These two options are intended to be removed someday, once
3627 they have helped determine the efficacy of various
3628 approaches to improving loop optimizations.
3630 Please let us (@w{@email{gcc@@gcc.gnu.org}} and @w{@email{fortran@@gnu.org}})
3631 know how use of these options affects
3632 the performance of your production code.
3633 We're very interested in code that runs @emph{slower}
3634 when these options are @emph{enabled}.
3637 @itemx -fno-peephole2
3638 @opindex fno-peephole
3639 @opindex fno-peephole2
3640 Disable any machine-specific peephole optimizations. The difference
3641 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
3642 are implemented in the compiler; some targets use one, some use the
3643 other, a few use both.
3645 @item -fbranch-probabilities
3646 @opindex fbranch-probabilities
3647 After running a program compiled with @option{-fprofile-arcs}
3648 (@pxref{Debugging Options,, Options for Debugging Your Program or
3649 @command{gcc}}), you can compile it a second time using
3650 @option{-fbranch-probabilities}, to improve optimizations based on
3651 the number of times each branch was taken. When the program
3652 compiled with @option{-fprofile-arcs} exits it saves arc execution
3653 counts to a file called @file{@var{sourcename}.da} for each source
3654 file The information in this data file is very dependent on the
3655 structure of the generated code, so you must use the same source code
3656 and the same optimization options for both compilations.
3659 With @option{-fbranch-probabilities}, GCC puts a @samp{REG_EXEC_COUNT}
3660 note on the first instruction of each basic block, and a
3661 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
3662 These can be used to improve optimization. Currently, they are only
3663 used in one place: in @file{reorg.c}, instead of guessing which path a
3664 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
3665 exactly determine which path is taken more often.
3668 @item -fno-guess-branch-probability
3669 @opindex fno-guess-branch-probability
3670 Do not guess branch probabilities using a randomized model.
3672 Sometimes gcc will opt to use a randomized model to guess branch
3673 probabilities, when none are available from either profiling feedback
3674 (@option{-fprofile-arcs}) or @samp{__builtin_expect}. This means that
3675 different runs of the compiler on the same program may produce different
3678 In a hard real-time system, people don't want different runs of the
3679 compiler to produce code that has different behavior; minimizing
3680 non-determinism is of paramount import. This switch allows users to
3681 reduce non-determinism, possibly at the expense of inferior
3684 @item -fstrict-aliasing
3685 @opindex fstrict-aliasing
3686 Allows the compiler to assume the strictest aliasing rules applicable to
3687 the language being compiled. For C (and C++), this activates
3688 optimizations based on the type of expressions. In particular, an
3689 object of one type is assumed never to reside at the same address as an
3690 object of a different type, unless the types are almost the same. For
3691 example, an @code{unsigned int} can alias an @code{int}, but not a
3692 @code{void*} or a @code{double}. A character type may alias any other
3695 Pay special attention to code like this:
3708 The practice of reading from a different union member than the one most
3709 recently written to (called ``type-punning'') is common. Even with
3710 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
3711 is accessed through the union type. So, the code above will work as
3712 expected. However, this code might not:
3724 Every language that wishes to perform language-specific alias analysis
3725 should define a function that computes, given an @code{tree}
3726 node, an alias set for the node. Nodes in different alias sets are not
3727 allowed to alias. For an example, see the C front-end function
3728 @code{c_get_alias_set}.
3731 @item -falign-functions
3732 @itemx -falign-functions=@var{n}
3733 @opindex falign-functions
3734 Align the start of functions to the next power-of-two greater than
3735 @var{n}, skipping up to @var{n} bytes. For instance,
3736 @option{-falign-functions=32} aligns functions to the next 32-byte
3737 boundary, but @option{-falign-functions=24} would align to the next
3738 32-byte boundary only if this can be done by skipping 23 bytes or less.
3740 @option{-fno-align-functions} and @option{-falign-functions=1} are
3741 equivalent and mean that functions will not be aligned.
3743 Some assemblers only support this flag when @var{n} is a power of two;
3744 in that case, it is rounded up.
3746 If @var{n} is not specified, use a machine-dependent default.
3748 @item -falign-labels
3749 @itemx -falign-labels=@var{n}
3750 @opindex falign-labels
3751 Align all branch targets to a power-of-two boundary, skipping up to
3752 @var{n} bytes like @option{-falign-functions}. This option can easily
3753 make code slower, because it must insert dummy operations for when the
3754 branch target is reached in the usual flow of the code.
3756 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
3757 are greater than this value, then their values are used instead.
3759 If @var{n} is not specified, use a machine-dependent default which is
3760 very likely to be @samp{1}, meaning no alignment.
3763 @itemx -falign-loops=@var{n}
3764 @opindex falign-loops
3765 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
3766 like @option{-falign-functions}. The hope is that the loop will be
3767 executed many times, which will make up for any execution of the dummy
3770 If @var{n} is not specified, use a machine-dependent default.
3773 @itemx -falign-jumps=@var{n}
3774 @opindex falign-jumps
3775 Align branch targets to a power-of-two boundary, for branch targets
3776 where the targets can only be reached by jumping, skipping up to @var{n}
3777 bytes like @option{-falign-functions}. In this case, no dummy operations
3780 If @var{n} is not specified, use a machine-dependent default.
3784 Perform optimizations in static single assignment form. Each function's
3785 flow graph is translated into SSA form, optimizations are performed, and
3786 the flow graph is translated back from SSA form. Users should not
3787 specify this option, since it is not yet ready for production use.
3791 Perform Sparse Conditional Constant Propagation in SSA form. Requires
3792 @option{-fssa}. Like @option{-fssa}, this is an experimental feature.
3796 Perform aggressive dead-code elimination in SSA form. Requires @option{-fssa}.
3797 Like @option{-fssa}, this is an experimental feature.
3799 @item -fsingle-precision-constant
3800 @opindex fsingle-precision-constant
3801 Treat floating point constant as single precision constant instead of
3802 implicitly converting it to double precision constant.
3804 @item -frename-registers
3805 @opindex frename-registers
3806 Attempt to avoid false dependencies in scheduled code by making use
3807 of registers left over after register allocation. This optimization
3808 will most benefit processors with lots of registers. It can, however,
3809 make debugging impossible, since variables will no longer stay in
3810 a ``home register''.
3812 @item --param @var{name}=@var{value}
3814 In some places, GCC uses various constants to control the amount of
3815 optimization that is done. For example, GCC will not inline functions
3816 that contain more that a certain number of instructions. You can
3817 control some of these constants on the command-line using the
3818 @option{--param} option.
3820 In each case, the @var{value} is an integer. The allowable choices for
3821 @var{name} are given in the following table:
3824 @item max-delay-slot-insn-search
3825 The maximum number of instructions to consider when looking for an
3826 instruction to fill a delay slot. If more than this arbitrary number of
3827 instructions is searched, the time savings from filling the delay slot
3828 will be minimal so stop searching. Increasing values mean more
3829 aggressive optimization, making the compile time increase with probably
3830 small improvement in executable run time.
3832 @item max-delay-slot-live-search
3833 When trying to fill delay slots, the maximum number of instructions to
3834 consider when searching for a block with valid live register
3835 information. Increasing this arbitrarily chosen value means more
3836 aggressive optimization, increasing the compile time. This parameter
3837 should be removed when the delay slot code is rewritten to maintain the
3840 @item max-gcse-memory
3841 The approximate maximum amount of memory that will be allocated in
3842 order to perform the global common subexpression elimination
3843 optimization. If more memory than specified is required, the
3844 optimization will not be done.
3846 @item max-gcse-passes
3847 The maximum number of passes of GCSE to run.
3849 @item max-pending-list-length
3850 The maximum number of pending dependencies scheduling will allow
3851 before flushing the current state and starting over. Large functions
3852 with few branches or calls can create excessively large lists which
3853 needlessly consume memory and resources.
3855 @item max-inline-insns
3856 If an function contains more than this many instructions, it
3857 will not be inlined. This option is precisely equivalent to
3858 @option{-finline-limit}.
3863 @node Preprocessor Options
3864 @section Options Controlling the Preprocessor
3865 @cindex preprocessor options
3866 @cindex options, preprocessor
3868 These options control the C preprocessor, which is run on each C source
3869 file before actual compilation.
3871 If you use the @option{-E} option, nothing is done except preprocessing.
3872 Some of these options make sense only together with @option{-E} because
3873 they cause the preprocessor output to be unsuitable for actual
3877 @item -include @var{file}
3879 Process @var{file} as input before processing the regular input file.
3880 In effect, the contents of @var{file} are compiled first. Any @option{-D}
3881 and @option{-U} options on the command line are always processed before
3882 @option{-include @var{file}}, regardless of the order in which they are
3883 written. All the @option{-include} and @option{-imacros} options are
3884 processed in the order in which they are written.
3886 @item -imacros @var{file}
3888 Process @var{file} as input, discarding the resulting output, before
3889 processing the regular input file. Because the output generated from
3890 @var{file} is discarded, the only effect of @option{-imacros @var{file}}
3891 is to make the macros defined in @var{file} available for use in the
3892 main input. All the @option{-include} and @option{-imacros} options are
3893 processed in the order in which they are written.
3895 @item -idirafter @var{dir}
3897 @cindex second include path
3898 Add the directory @var{dir} to the second include path. The directories
3899 on the second include path are searched when a header file is not found
3900 in any of the directories in the main include path (the one that
3901 @option{-I} adds to).
3903 @item -iprefix @var{prefix}
3905 Specify @var{prefix} as the prefix for subsequent @option{-iwithprefix}
3908 @item -iwithprefix @var{dir}
3909 @opindex iwithprefix
3910 Add a directory to the second include path. The directory's name is
3911 made by concatenating @var{prefix} and @var{dir}, where @var{prefix} was
3912 specified previously with @option{-iprefix}. If you have not specified a
3913 prefix yet, the directory containing the installed passes of the
3914 compiler is used as the default.
3916 @item -iwithprefixbefore @var{dir}
3917 @opindex iwithprefixbefore
3918 Add a directory to the main include path. The directory's name is made
3919 by concatenating @var{prefix} and @var{dir}, as in the case of
3920 @option{-iwithprefix}.
3922 @item -isystem @var{dir}
3924 Add a directory to the beginning of the second include path, marking it
3925 as a system directory, so that it gets the same special treatment as
3926 is applied to the standard system directories.
3930 Do not search the standard system directories for header files. Only
3931 the directories you have specified with @option{-I} options (and the
3932 current directory, if appropriate) are searched. @xref{Directory
3933 Options}, for information on @option{-I}.
3935 By using both @option{-nostdinc} and @option{-I-}, you can limit the include-file
3936 search path to only those directories you specify explicitly.
3940 When searching for a header file in a directory, remap file names if a
3941 file named @file{header.gcc} exists in that directory. This can be used
3942 to work around limitations of file systems with file name restrictions.
3943 The @file{header.gcc} file should contain a series of lines with two
3944 tokens on each line: the first token is the name to map, and the second
3945 token is the actual name to use.
3949 Do not predefine any nonstandard macros. (Including architecture flags).
3953 Run only the C preprocessor. Preprocess all the C source files
3954 specified and output the results to standard output or to the
3955 specified output file.
3959 Tell the preprocessor not to discard comments. Used with the
3964 Tell the preprocessor not to generate @samp{#line} directives.
3965 Used with the @option{-E} option.
3968 @cindex dependencies, make
3971 Instead of outputting the result of preprocessing, output a rule
3972 suitable for @code{make} describing the dependencies of the main source
3973 file. The preprocessor outputs one @code{make} rule containing the
3974 object file name for that source file, a colon, and the names of all the
3975 included files. Unless overridden explicitly, the object file name
3976 consists of the basename of the source file with any suffix replaced with
3977 object file suffix. If there are many included files then the
3978 rule is split into several lines using @samp{\}-newline.
3980 @option{-M} implies @option{-E}.
3984 Like @option{-M}, but mention only the files included with @samp{#include
3985 "@var{file}"}. System header files included with @samp{#include
3986 <@var{file}>} are omitted.
3990 Like @option{-M} but the dependency information is written to a file
3991 rather than stdout. @code{gcc} will use the same file name and
3992 directory as the object file, but with the suffix @file{.d} instead.
3994 This is in addition to compiling the main file as specified---@option{-MD}
3995 does not inhibit ordinary compilation the way @option{-M} does,
3996 unless you also specify @option{-MG}.
3998 With Mach, you can use the utility @code{md} to merge multiple
3999 dependency files into a single dependency file suitable for using with
4000 the @samp{make} command.
4004 Like @option{-MD} except mention only user header files, not system
4007 @item -MF @var{file}
4009 When used with @option{-M} or @option{-MM}, specifies a file to write the
4010 dependencies to. This allows the preprocessor to write the preprocessed
4011 file to stdout normally. If no @option{-MF} switch is given, CPP sends
4012 the rules to stdout and suppresses normal preprocessed output.
4014 Another way to specify output of a @code{make} rule is by setting
4015 the environment variable @env{DEPENDENCIES_OUTPUT} (@pxref{Environment
4020 When used with @option{-M} or @option{-MM}, @option{-MG} says to treat missing
4021 header files as generated files and assume they live in the same
4022 directory as the source file. It suppresses preprocessed output, as a
4023 missing header file is ordinarily an error.
4025 This feature is used in automatic updating of makefiles.
4029 This option instructs CPP to add a phony target for each dependency
4030 other than the main file, causing each to depend on nothing. These
4031 dummy rules work around errors @code{make} gives if you remove header
4032 files without updating the @code{Makefile} to match.
4034 This is typical output:-
4037 /tmp/test.o: /tmp/test.c /tmp/test.h
4042 @item -MQ @var{target}
4043 @item -MT @var{target}
4046 By default CPP uses the main file name, including any path, and appends
4047 the object suffix, normally ``.o'', to it to obtain the name of the
4048 target for dependency generation. With @option{-MT} you can specify a
4049 target yourself, overriding the default one.
4051 If you want multiple targets, you can specify them as a single argument
4052 to @option{-MT}, or use multiple @option{-MT} options.
4054 The targets you specify are output in the order they appear on the
4055 command line. @option{-MQ} is identical to @option{-MT}, except that the
4056 target name is quoted for Make, but with @option{-MT} it isn't. For
4057 example, @option{-MT '$(objpfx)foo.o'} gives
4060 $(objpfx)foo.o: /tmp/foo.c
4063 but @option{-MQ '$(objpfx)foo.o'} gives
4066 $$(objpfx)foo.o: /tmp/foo.c
4069 The default target is automatically quoted, as if it were given with
4074 Print the name of each header file used, in addition to other normal
4077 @item -A@var{question}(@var{answer})
4079 Assert the answer @var{answer} for @var{question}, in case it is tested
4080 with a preprocessing conditional such as @samp{#if
4081 #@var{question}(@var{answer})}. @option{-A-} disables the standard
4082 assertions that normally describe the target machine.
4086 Define macro @var{macro} with the string @samp{1} as its definition.
4088 @item -D@var{macro}=@var{defn}
4089 Define macro @var{macro} as @var{defn}. All instances of @option{-D} on
4090 the command line are processed before any @option{-U} options.
4092 Any @option{-D} and @option{-U} options on the command line are processed in
4093 order, and always before @option{-imacros @var{file}}, regardless of the
4094 order in which they are written.
4098 Undefine macro @var{macro}. @option{-U} options are evaluated after all
4099 @option{-D} options, but before any @option{-include} and @option{-imacros}
4102 Any @option{-D} and @option{-U} options on the command line are processed in
4103 order, and always before @option{-imacros @var{file}}, regardless of the
4104 order in which they are written.
4108 Tell the preprocessor to output only a list of the macro definitions
4109 that are in effect at the end of preprocessing. Used with the @option{-E}
4114 Tell the preprocessing to pass all macro definitions into the output, in
4115 their proper sequence in the rest of the output.
4119 Like @option{-dD} except that the macro arguments and contents are omitted.
4120 Only @samp{#define @var{name}} is included in the output.
4124 Output @samp{#include} directives in addition to the result of
4127 @item -fpreprocessed
4128 @opindex fpreprocessed
4129 Indicate to the preprocessor that the input file has already been
4130 preprocessed. This suppresses things like macro expansion, trigraph
4131 conversion, escaped newline splicing, and processing of most directives.
4132 The preprocessor still recognizes and removes comments, so that you can
4133 pass a file preprocessed with @option{-C} to the compiler without
4134 problems. In this mode the integrated preprocessor is little more than
4135 a tokenizer for the front ends.
4137 @option{-fpreprocessed} is implicit if the input file has one of the
4138 extensions @samp{i}, @samp{ii} or @samp{mi}. These are the extensions
4139 that GCC uses for preprocessed files created by @option{-save-temps}.
4143 Process ISO standard trigraph sequences. These are three-character
4144 sequences, all starting with @samp{??}, that are defined by ISO C to
4145 stand for single characters. For example, @samp{??/} stands for
4146 @samp{\}, so @samp{'??/n'} is a character constant for a newline. By
4147 default, GCC ignores trigraphs, but in standard-conforming modes it
4148 converts them. See the @option{-std} and @option{-ansi} options.
4150 The nine trigraph sequences are
4153 @expansion{} @samp{[}
4156 @expansion{} @samp{]}
4159 @expansion{} @samp{@{}
4162 @expansion{} @samp{@}}
4165 @expansion{} @samp{#}
4168 @expansion{} @samp{\}
4171 @expansion{} @samp{^}
4174 @expansion{} @samp{|}
4177 @expansion{} @samp{~}
4181 Trigraph support is not popular, so many compilers do not implement it
4182 properly. Portable code should not rely on trigraphs being either
4183 converted or ignored.
4185 @item -Wp,@var{option}
4187 Pass @var{option} as an option to the preprocessor. If @var{option}
4188 contains commas, it is split into multiple options at the commas.
4191 @node Assembler Options
4192 @section Passing Options to the Assembler
4194 @c prevent bad page break with this line
4195 You can pass options to the assembler.
4198 @item -Wa,@var{option}
4200 Pass @var{option} as an option to the assembler. If @var{option}
4201 contains commas, it is split into multiple options at the commas.
4205 @section Options for Linking
4206 @cindex link options
4207 @cindex options, linking
4209 These options come into play when the compiler links object files into
4210 an executable output file. They are meaningless if the compiler is
4211 not doing a link step.
4215 @item @var{object-file-name}
4216 A file name that does not end in a special recognized suffix is
4217 considered to name an object file or library. (Object files are
4218 distinguished from libraries by the linker according to the file
4219 contents.) If linking is done, these object files are used as input
4228 If any of these options is used, then the linker is not run, and
4229 object file names should not be used as arguments. @xref{Overall
4233 @item -l@var{library}
4234 @itemx -l @var{library}
4236 Search the library named @var{library} when linking. (The second
4237 alternative with the library as a separate argument is only for
4238 POSIX compliance and is not recommended.)
4240 It makes a difference where in the command you write this option; the
4241 linker searches and processes libraries and object files in the order they
4242 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
4243 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
4244 to functions in @samp{z}, those functions may not be loaded.
4246 The linker searches a standard list of directories for the library,
4247 which is actually a file named @file{lib@var{library}.a}. The linker
4248 then uses this file as if it had been specified precisely by name.
4250 The directories searched include several standard system directories
4251 plus any that you specify with @option{-L}.
4253 Normally the files found this way are library files---archive files
4254 whose members are object files. The linker handles an archive file by
4255 scanning through it for members which define symbols that have so far
4256 been referenced but not defined. But if the file that is found is an
4257 ordinary object file, it is linked in the usual fashion. The only
4258 difference between using an @option{-l} option and specifying a file name
4259 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
4260 and searches several directories.
4264 You need this special case of the @option{-l} option in order to
4265 link an Objective-C program.
4268 @opindex nostartfiles
4269 Do not use the standard system startup files when linking.
4270 The standard system libraries are used normally, unless @option{-nostdlib}
4271 or @option{-nodefaultlibs} is used.
4273 @item -nodefaultlibs
4274 @opindex nodefaultlibs
4275 Do not use the standard system libraries when linking.
4276 Only the libraries you specify will be passed to the linker.
4277 The standard startup files are used normally, unless @option{-nostartfiles}
4278 is used. The compiler may generate calls to memcmp, memset, and memcpy
4279 for System V (and ISO C) environments or to bcopy and bzero for
4280 BSD environments. These entries are usually resolved by entries in
4281 libc. These entry points should be supplied through some other
4282 mechanism when this option is specified.
4286 Do not use the standard system startup files or libraries when linking.
4287 No startup files and only the libraries you specify will be passed to
4288 the linker. The compiler may generate calls to memcmp, memset, and memcpy
4289 for System V (and ISO C) environments or to bcopy and bzero for
4290 BSD environments. These entries are usually resolved by entries in
4291 libc. These entry points should be supplied through some other
4292 mechanism when this option is specified.
4294 @cindex @option{-lgcc}, use with @option{-nostdlib}
4295 @cindex @option{-nostdlib} and unresolved references
4296 @cindex unresolved references and @option{-nostdlib}
4297 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
4298 @cindex @option{-nodefaultlibs} and unresolved references
4299 @cindex unresolved references and @option{-nodefaultlibs}
4300 One of the standard libraries bypassed by @option{-nostdlib} and
4301 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
4302 that GCC uses to overcome shortcomings of particular machines, or special
4303 needs for some languages.
4305 (@xref{Interface,,Interfacing to GCC Output}, for more discussion of
4309 (@xref{Interface,,Interfacing to GCC Output,gcc.info,Porting GCC},
4310 for more discussion of @file{libgcc.a}.)
4312 In most cases, you need @file{libgcc.a} even when you want to avoid
4313 other standard libraries. In other words, when you specify @option{-nostdlib}
4314 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
4315 This ensures that you have no unresolved references to internal GCC
4316 library subroutines. (For example, @samp{__main}, used to ensure C++
4317 constructors will be called; @pxref{Collect2,,@command{collect2}}.)
4321 Remove all symbol table and relocation information from the executable.
4325 On systems that support dynamic linking, this prevents linking with the shared
4326 libraries. On other systems, this option has no effect.
4330 Produce a shared object which can then be linked with other objects to
4331 form an executable. Not all systems support this option. For predictable
4332 results, you must also specify the same set of options that were used to
4333 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
4334 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
4335 needs to build supplementary stub code for constructors to work. On
4336 multi-libbed systems, @samp{gcc -shared} must select the correct support
4337 libraries to link against. Failing to supply the correct flags may lead
4338 to subtle defects. Supplying them in cases where they are not necessary
4341 @item -shared-libgcc
4342 @itemx -static-libgcc
4343 @opindex shared-libgcc
4344 @opindex static-libgcc
4345 On systems that provide @file{libgcc} as a shared library, these options
4346 force the use of either the shared or static version respectively.
4347 If no shared version of @file{libgcc} was built when the compiler was
4348 configured, these options have no effect.
4350 There are several situations in which an application should use the
4351 shared @file{libgcc} instead of the static version. The most common
4352 of these is when the application wishes to throw and catch exceptions
4353 across different shared libraries. In that case, each of the libraries
4354 as well as the application itself should use the shared @file{libgcc}.
4356 Therefore, whenever you specify the @option{-shared} option, the GCC
4357 driver automatically adds @option{-shared-libgcc}, unless you explicitly
4358 specify @option{-static-libgcc}. The G++ driver automatically adds
4359 @option{-shared-libgcc} when you build a main executable as well because
4360 for C++ programs that is typically the right thing to do.
4361 (Exception-handling will not work reliably otherwise.)
4363 However, when linking a main executable written in C, you must
4364 explicitly say @option{-shared-libgcc} if you want to use the shared
4369 Bind references to global symbols when building a shared object. Warn
4370 about any unresolved references (unless overridden by the link editor
4371 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
4374 @item -Xlinker @var{option}
4376 Pass @var{option} as an option to the linker. You can use this to
4377 supply system-specific linker options which GCC does not know how to
4380 If you want to pass an option that takes an argument, you must use
4381 @option{-Xlinker} twice, once for the option and once for the argument.
4382 For example, to pass @option{-assert definitions}, you must write
4383 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
4384 @option{-Xlinker "-assert definitions"}, because this passes the entire
4385 string as a single argument, which is not what the linker expects.
4387 @item -Wl,@var{option}
4389 Pass @var{option} as an option to the linker. If @var{option} contains
4390 commas, it is split into multiple options at the commas.
4392 @item -u @var{symbol}
4394 Pretend the symbol @var{symbol} is undefined, to force linking of
4395 library modules to define it. You can use @option{-u} multiple times with
4396 different symbols to force loading of additional library modules.
4399 @node Directory Options
4400 @section Options for Directory Search
4401 @cindex directory options
4402 @cindex options, directory search
4405 These options specify directories to search for header files, for
4406 libraries and for parts of the compiler:
4411 Add the directory @var{dir} to the head of the list of directories to be
4412 searched for header files. This can be used to override a system header
4413 file, substituting your own version, since these directories are
4414 searched before the system header file directories. However, you should
4415 not use this option to add directories that contain vendor-supplied
4416 system header files (use @option{-isystem} for that). If you use more than
4417 one @option{-I} option, the directories are scanned in left-to-right
4418 order; the standard system directories come after.
4420 If a standard system include directory, or a directory specified with
4421 @option{-isystem}, is also specified with @option{-I}, it will be
4422 searched only in the position requested by @option{-I}. Also, it will
4423 not be considered a system include directory. If that directory really
4424 does contain system headers, there is a good chance that they will
4425 break. For instance, if GCC's installation procedure edited the headers
4426 in @file{/usr/include} to fix bugs, @samp{-I/usr/include} will cause the
4427 original, buggy headers to be found instead of the corrected ones. GCC
4428 will issue a warning when a system include directory is hidden in this
4433 Any directories you specify with @option{-I} options before the @option{-I-}
4434 option are searched only for the case of @samp{#include "@var{file}"};
4435 they are not searched for @samp{#include <@var{file}>}.
4437 If additional directories are specified with @option{-I} options after
4438 the @option{-I-}, these directories are searched for all @samp{#include}
4439 directives. (Ordinarily @emph{all} @option{-I} directories are used
4442 In addition, the @option{-I-} option inhibits the use of the current
4443 directory (where the current input file came from) as the first search
4444 directory for @samp{#include "@var{file}"}. There is no way to
4445 override this effect of @option{-I-}. With @option{-I.} you can specify
4446 searching the directory which was current when the compiler was
4447 invoked. That is not exactly the same as what the preprocessor does
4448 by default, but it is often satisfactory.
4450 @option{-I-} does not inhibit the use of the standard system directories
4451 for header files. Thus, @option{-I-} and @option{-nostdinc} are
4456 Add directory @var{dir} to the list of directories to be searched
4459 @item -B@var{prefix}
4461 This option specifies where to find the executables, libraries,
4462 include files, and data files of the compiler itself.
4464 The compiler driver program runs one or more of the subprograms
4465 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
4466 @var{prefix} as a prefix for each program it tries to run, both with and
4467 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
4469 For each subprogram to be run, the compiler driver first tries the
4470 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
4471 was not specified, the driver tries two standard prefixes, which are
4472 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc-lib/}. If neither of
4473 those results in a file name that is found, the unmodified program
4474 name is searched for using the directories specified in your
4475 @env{PATH} environment variable.
4477 The compiler will check to see if the path provided by the @option{-B}
4478 refers to a directory, and if necessary it will add a directory
4479 separator character at the end of the path.
4481 @option{-B} prefixes that effectively specify directory names also apply
4482 to libraries in the linker, because the compiler translates these
4483 options into @option{-L} options for the linker. They also apply to
4484 includes files in the preprocessor, because the compiler translates these
4485 options into @option{-isystem} options for the preprocessor. In this case,
4486 the compiler appends @samp{include} to the prefix.
4488 The run-time support file @file{libgcc.a} can also be searched for using
4489 the @option{-B} prefix, if needed. If it is not found there, the two
4490 standard prefixes above are tried, and that is all. The file is left
4491 out of the link if it is not found by those means.
4493 Another way to specify a prefix much like the @option{-B} prefix is to use
4494 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
4497 As a special kludge, if the path provided by @option{-B} is
4498 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
4499 9, then it will be replaced by @file{[dir/]include}. This is to help
4500 with boot-strapping the compiler.
4502 @item -specs=@var{file}
4504 Process @var{file} after the compiler reads in the standard @file{specs}
4505 file, in order to override the defaults that the @file{gcc} driver
4506 program uses when determining what switches to pass to @file{cc1},
4507 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
4508 @option{-specs=@var{file}} can be specified on the command line, and they
4509 are processed in order, from left to right.
4515 @section Specifying subprocesses and the switches to pass to them
4517 @command{gcc} is a driver program. It performs its job by invoking a
4518 sequence of other programs to do the work of compiling, assembling and
4519 linking. GCC interprets its command-line parameters and uses these to
4520 deduce which programs it should invoke, and which command-line options
4521 it ought to place on their command lines. This behaviour is controlled
4522 by @dfn{spec strings}. In most cases there is one spec string for each
4523 program that GCC can invoke, but a few programs have multiple spec
4524 strings to control their behaviour. The spec strings built into GCC can
4525 be overridden by using the @option{-specs=} command-line switch to specify
4528 @dfn{Spec files} are plaintext files that are used to construct spec
4529 strings. They consist of a sequence of directives separated by blank
4530 lines. The type of directive is determined by the first non-whitespace
4531 character on the line and it can be one of the following:
4534 @item %@var{command}
4535 Issues a @var{command} to the spec file processor. The commands that can
4539 @item %include <@var{file}>
4541 Search for @var{file} and insert its text at the current point in the
4544 @item %include_noerr <@var{file}>
4545 @cindex %include_noerr
4546 Just like @samp{%include}, but do not generate an error message if the include
4547 file cannot be found.
4549 @item %rename @var{old_name} @var{new_name}
4551 Rename the spec string @var{old_name} to @var{new_name}.
4555 @item *[@var{spec_name}]:
4556 This tells the compiler to create, override or delete the named spec
4557 string. All lines after this directive up to the next directive or
4558 blank line are considered to be the text for the spec string. If this
4559 results in an empty string then the spec will be deleted. (Or, if the
4560 spec did not exist, then nothing will happened.) Otherwise, if the spec
4561 does not currently exist a new spec will be created. If the spec does
4562 exist then its contents will be overridden by the text of this
4563 directive, unless the first character of that text is the @samp{+}
4564 character, in which case the text will be appended to the spec.
4566 @item [@var{suffix}]:
4567 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
4568 and up to the next directive or blank line are considered to make up the
4569 spec string for the indicated suffix. When the compiler encounters an
4570 input file with the named suffix, it will processes the spec string in
4571 order to work out how to compile that file. For example:
4578 This says that any input file whose name ends in @samp{.ZZ} should be
4579 passed to the program @samp{z-compile}, which should be invoked with the
4580 command-line switch @option{-input} and with the result of performing the
4581 @samp{%i} substitution. (See below.)
4583 As an alternative to providing a spec string, the text that follows a
4584 suffix directive can be one of the following:
4587 @item @@@var{language}
4588 This says that the suffix is an alias for a known @var{language}. This is
4589 similar to using the @option{-x} command-line switch to GCC to specify a
4590 language explicitly. For example:
4597 Says that .ZZ files are, in fact, C++ source files.
4600 This causes an error messages saying:
4603 @var{name} compiler not installed on this system.
4607 GCC already has an extensive list of suffixes built into it.
4608 This directive will add an entry to the end of the list of suffixes, but
4609 since the list is searched from the end backwards, it is effectively
4610 possible to override earlier entries using this technique.
4614 GCC has the following spec strings built into it. Spec files can
4615 override these strings or create their own. Note that individual
4616 targets can also add their own spec strings to this list.
4619 asm Options to pass to the assembler
4620 asm_final Options to pass to the assembler post-processor
4621 cpp Options to pass to the C preprocessor
4622 cc1 Options to pass to the C compiler
4623 cc1plus Options to pass to the C++ compiler
4624 endfile Object files to include at the end of the link
4625 link Options to pass to the linker
4626 lib Libraries to include on the command line to the linker
4627 libgcc Decides which GCC support library to pass to the linker
4628 linker Sets the name of the linker
4629 predefines Defines to be passed to the C preprocessor
4630 signed_char Defines to pass to CPP to say whether @code{char} is signed
4632 startfile Object files to include at the start of the link
4635 Here is a small example of a spec file:
4641 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
4644 This example renames the spec called @samp{lib} to @samp{old_lib} and
4645 then overrides the previous definition of @samp{lib} with a new one.
4646 The new definition adds in some extra command-line options before
4647 including the text of the old definition.
4649 @dfn{Spec strings} are a list of command-line options to be passed to their
4650 corresponding program. In addition, the spec strings can contain
4651 @samp{%}-prefixed sequences to substitute variable text or to
4652 conditionally insert text into the command line. Using these constructs
4653 it is possible to generate quite complex command lines.
4655 Here is a table of all defined @samp{%}-sequences for spec
4656 strings. Note that spaces are not generated automatically around the
4657 results of expanding these sequences. Therefore you can concatenate them
4658 together or combine them with constant text in a single argument.
4662 Substitute one @samp{%} into the program name or argument.
4665 Substitute the name of the input file being processed.
4668 Substitute the basename of the input file being processed.
4669 This is the substring up to (and not including) the last period
4670 and not including the directory.
4673 This is the same as @samp{%b}, but include the file suffix (text after
4677 Marks the argument containing or following the @samp{%d} as a
4678 temporary file name, so that that file will be deleted if GCC exits
4679 successfully. Unlike @samp{%g}, this contributes no text to the
4682 @item %g@var{suffix}
4683 Substitute a file name that has suffix @var{suffix} and is chosen
4684 once per compilation, and mark the argument in the same way as
4685 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
4686 name is now chosen in a way that is hard to predict even when previously
4687 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
4688 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
4689 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
4690 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
4691 was simply substituted with a file name chosen once per compilation,
4692 without regard to any appended suffix (which was therefore treated
4693 just like ordinary text), making such attacks more likely to succeed.
4695 @item %u@var{suffix}
4696 Like @samp{%g}, but generates a new temporary file name even if
4697 @samp{%u@var{suffix}} was already seen.
4699 @item %U@var{suffix}
4700 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
4701 new one if there is no such last file name. In the absence of any
4702 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
4703 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
4704 would involve the generation of two distinct file names, one
4705 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
4706 simply substituted with a file name chosen for the previous @samp{%u},
4707 without regard to any appended suffix.
4709 @item %j@var{SUFFIX}
4710 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
4711 writable, and if save-temps is off; otherwise, substitute the name
4712 of a temporary file, just like @samp{%u}. This temporary file is not
4713 meant for communication between processes, but rather as a junk
4716 @item %.@var{SUFFIX}
4717 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
4718 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
4719 terminated by the next space or %.
4722 Marks the argument containing or following the @samp{%w} as the
4723 designated output file of this compilation. This puts the argument
4724 into the sequence of arguments that @samp{%o} will substitute later.
4727 Substitutes the names of all the output files, with spaces
4728 automatically placed around them. You should write spaces
4729 around the @samp{%o} as well or the results are undefined.
4730 @samp{%o} is for use in the specs for running the linker.
4731 Input files whose names have no recognized suffix are not compiled
4732 at all, but they are included among the output files, so they will
4736 Substitutes the suffix for object files. Note that this is
4737 handled specially when it immediately follows @samp{%g, %u, or %U},
4738 because of the need for those to form complete file names. The
4739 handling is such that @samp{%O} is treated exactly as if it had already
4740 been substituted, except that @samp{%g, %u, and %U} do not currently
4741 support additional @var{suffix} characters following @samp{%O} as they would
4742 following, for example, @samp{.o}.
4745 Substitutes the standard macro predefinitions for the
4746 current target machine. Use this when running @code{cpp}.
4749 Like @samp{%p}, but puts @samp{__} before and after the name of each
4750 predefined macro, except for macros that start with @samp{__} or with
4751 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
4755 Substitute a @option{-iprefix} option made from @env{GCC_EXEC_PREFIX}.
4758 Current argument is the name of a library or startup file of some sort.
4759 Search for that file in a standard list of directories and substitute
4760 the full name found.
4763 Print @var{str} as an error message. @var{str} is terminated by a newline.
4764 Use this when inconsistent options are detected.
4767 Output @samp{-} if the input for the current command is coming from a pipe.
4770 Substitute the contents of spec string @var{name} at this point.
4773 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
4775 @item %x@{@var{option}@}
4776 Accumulate an option for @samp{%X}.
4779 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
4783 Output the accumulated assembler options specified by @option{-Wa}.
4786 Output the accumulated preprocessor options specified by @option{-Wp}.
4789 Substitute the major version number of GCC@.
4790 (For version 2.9.5, this is 2.)
4793 Substitute the minor version number of GCC@.
4794 (For version 2.9.5, this is 9.)
4797 Substitute the patch level number of GCC@.
4798 (For version 2.9.5, this is 5.)
4801 Process the @code{asm} spec. This is used to compute the
4802 switches to be passed to the assembler.
4805 Process the @code{asm_final} spec. This is a spec string for
4806 passing switches to an assembler post-processor, if such a program is
4810 Process the @code{link} spec. This is the spec for computing the
4811 command line passed to the linker. Typically it will make use of the
4812 @samp{%L %G %S %D and %E} sequences.
4815 Dump out a @option{-L} option for each directory that GCC believes might
4816 contain startup files. If the target supports multilibs then the
4817 current multilib directory will be prepended to each of these paths.
4820 Output the multilib directory with directory separators replaced with
4821 @samp{_}. If multilib directories are not set, or the multilib directory is
4822 @file{.} then this option emits nothing.
4825 Process the @code{lib} spec. This is a spec string for deciding which
4826 libraries should be included on the command line to the linker.
4829 Process the @code{libgcc} spec. This is a spec string for deciding
4830 which GCC support library should be included on the command line to the linker.
4833 Process the @code{startfile} spec. This is a spec for deciding which
4834 object files should be the first ones passed to the linker. Typically
4835 this might be a file named @file{crt0.o}.
4838 Process the @code{endfile} spec. This is a spec string that specifies
4839 the last object files that will be passed to the linker.
4842 Process the @code{cpp} spec. This is used to construct the arguments
4843 to be passed to the C preprocessor.
4846 Process the @code{signed_char} spec. This is intended to be used
4847 to tell cpp whether a char is signed. It typically has the definition:
4849 %@{funsigned-char:-D__CHAR_UNSIGNED__@}
4853 Process the @code{cc1} spec. This is used to construct the options to be
4854 passed to the actual C compiler (@samp{cc1}).
4857 Process the @code{cc1plus} spec. This is used to construct the options to be
4858 passed to the actual C++ compiler (@samp{cc1plus}).
4861 Substitute the variable part of a matched option. See below.
4862 Note that each comma in the substituted string is replaced by
4866 Substitutes the @code{-S} switch, if that switch was given to GCC@.
4867 If that switch was not specified, this substitutes nothing. Note that
4868 the leading dash is omitted when specifying this option, and it is
4869 automatically inserted if the substitution is performed. Thus the spec
4870 string @samp{%@{foo@}} would match the command-line option @option{-foo}
4871 and would output the command line option @option{-foo}.
4873 @item %W@{@code{S}@}
4874 Like %@{@code{S}@} but mark last argument supplied within as a file to be
4877 @item %@{@code{S}*@}
4878 Substitutes all the switches specified to GCC whose names start
4879 with @code{-S}, but which also take an argument. This is used for
4880 switches like @option{-o}, @option{-D}, @option{-I}, etc.
4881 GCC considers @option{-o foo} as being
4882 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
4883 text, including the space. Thus two arguments would be generated.
4885 @item %@{^@code{S}*@}
4886 Like %@{@code{S}*@}, but don't put a blank between a switch and its
4887 argument. Thus %@{^o*@} would only generate one argument, not two.
4889 @item %@{@code{S}*&@code{T}*@}
4890 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
4891 (the order of @code{S} and @code{T} in the spec is not significant).
4892 There can be any number of ampersand-separated variables; for each the
4893 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
4895 @item %@{<@code{S}@}
4896 Remove all occurrences of @code{-S} from the command line. Note---this
4897 command is position dependent. @samp{%} commands in the spec string
4898 before this option will see @code{-S}, @samp{%} commands in the spec
4899 string after this option will not.
4901 @item %@{@code{S}*:@code{X}@}
4902 Substitutes @code{X} if one or more switches whose names start with
4903 @code{-S} are specified to GCC@. Note that the tail part of the
4904 @code{-S} option (i.e.@: the part matched by the @samp{*}) will be substituted
4905 for each occurrence of @samp{%*} within @code{X}.
4907 @item %@{@code{S}:@code{X}@}
4908 Substitutes @code{X}, but only if the @samp{-S} switch was given to GCC@.
4910 @item %@{!@code{S}:@code{X}@}
4911 Substitutes @code{X}, but only if the @samp{-S} switch was @emph{not} given to GCC@.
4913 @item %@{|@code{S}:@code{X}@}
4914 Like %@{@code{S}:@code{X}@}, but if no @code{S} switch, substitute @samp{-}.
4916 @item %@{|!@code{S}:@code{X}@}
4917 Like %@{!@code{S}:@code{X}@}, but if there is an @code{S} switch, substitute @samp{-}.
4919 @item %@{.@code{S}:@code{X}@}
4920 Substitutes @code{X}, but only if processing a file with suffix @code{S}.
4922 @item %@{!.@code{S}:@code{X}@}
4923 Substitutes @code{X}, but only if @emph{not} processing a file with suffix @code{S}.
4925 @item %@{@code{S}|@code{P}:@code{X}@}
4926 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC@. This may be
4927 combined with @samp{!} and @samp{.} sequences as well, although they
4928 have a stronger binding than the @samp{|}. For example a spec string
4932 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
4935 will output the following command-line options from the following input
4936 command-line options:
4941 -d fred.c -foo -baz -boggle
4942 -d jim.d -bar -baz -boggle
4947 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or
4948 %@{!@code{S}:@code{X}@} construct may contain other nested @samp{%} constructs
4949 or spaces, or even newlines. They are processed as usual, as described
4952 The @option{-O}, @option{-f}, @option{-m}, and @option{-W}
4953 switches are handled specifically in these
4954 constructs. If another value of @option{-O} or the negated form of a @option{-f}, @option{-m}, or
4955 @option{-W} switch is found later in the command line, the earlier switch
4956 value is ignored, except with @{@code{S}*@} where @code{S} is just one
4957 letter, which passes all matching options.
4959 The character @samp{|} at the beginning of the predicate text is used to indicate
4960 that a command should be piped to the following command, but only if @option{-pipe}
4963 It is built into GCC which switches take arguments and which do not.
4964 (You might think it would be useful to generalize this to allow each
4965 compiler's spec to say which switches take arguments. But this cannot
4966 be done in a consistent fashion. GCC cannot even decide which input
4967 files have been specified without knowing which switches take arguments,
4968 and it must know which input files to compile in order to tell which
4971 GCC also knows implicitly that arguments starting in @option{-l} are to be
4972 treated as compiler output files, and passed to the linker in their
4973 proper position among the other output files.
4975 @c man begin OPTIONS
4977 @node Target Options
4978 @section Specifying Target Machine and Compiler Version
4979 @cindex target options
4980 @cindex cross compiling
4981 @cindex specifying machine version
4982 @cindex specifying compiler version and target machine
4983 @cindex compiler version, specifying
4984 @cindex target machine, specifying
4986 By default, GCC compiles code for the same type of machine that you
4987 are using. However, it can also be installed as a cross-compiler, to
4988 compile for some other type of machine. In fact, several different
4989 configurations of GCC, for different target machines, can be
4990 installed side by side. Then you specify which one to use with the
4993 In addition, older and newer versions of GCC can be installed side
4994 by side. One of them (probably the newest) will be the default, but
4995 you may sometimes wish to use another.
4998 @item -b @var{machine}
5000 The argument @var{machine} specifies the target machine for compilation.
5001 This is useful when you have installed GCC as a cross-compiler.
5003 The value to use for @var{machine} is the same as was specified as the
5004 machine type when configuring GCC as a cross-compiler. For
5005 example, if a cross-compiler was configured with @samp{configure
5006 i386v}, meaning to compile for an 80386 running System V, then you
5007 would specify @option{-b i386v} to run that cross compiler.
5009 When you do not specify @option{-b}, it normally means to compile for
5010 the same type of machine that you are using.
5012 @item -V @var{version}
5014 The argument @var{version} specifies which version of GCC to run.
5015 This is useful when multiple versions are installed. For example,
5016 @var{version} might be @samp{2.0}, meaning to run GCC version 2.0.
5018 The default version, when you do not specify @option{-V}, is the last
5019 version of GCC that you installed.
5022 The @option{-b} and @option{-V} options actually work by controlling part of
5023 the file name used for the executable files and libraries used for
5024 compilation. A given version of GCC, for a given target machine, is
5025 normally kept in the directory @file{/usr/local/lib/gcc-lib/@var{machine}/@var{version}}.
5027 Thus, sites can customize the effect of @option{-b} or @option{-V} either by
5028 changing the names of these directories or adding alternate names (or
5029 symbolic links). If in directory @file{/usr/local/lib/gcc-lib/} the
5030 file @file{80386} is a link to the file @file{i386v}, then @option{-b
5031 80386} becomes an alias for @option{-b i386v}.
5033 In one respect, the @option{-b} or @option{-V} do not completely change
5034 to a different compiler: the top-level driver program @command{gcc}
5035 that you originally invoked continues to run and invoke the other
5036 executables (preprocessor, compiler per se, assembler and linker)
5037 that do the real work. However, since no real work is done in the
5038 driver program, it usually does not matter that the driver program
5039 in use is not the one for the specified target. It is common for the
5040 interface to the other executables to change incompatibly between
5041 compiler versions, so unless the version specified is very close to that
5042 of the driver (for example, @option{-V 3.0} with a driver program from GCC
5043 version 3.0.1), use of @option{-V} may not work; for example, using
5044 @option{-V 2.95.2} will not work with a driver program from GCC 3.0.
5046 The only way that the driver program depends on the target machine is
5047 in the parsing and handling of special machine-specific options.
5048 However, this is controlled by a file which is found, along with the
5049 other executables, in the directory for the specified version and
5050 target machine. As a result, a single installed driver program adapts
5051 to any specified target machine, and sufficiently similar compiler
5054 The driver program executable does control one significant thing,
5055 however: the default version and target machine. Therefore, you can
5056 install different instances of the driver program, compiled for
5057 different targets or versions, under different names.
5059 For example, if the driver for version 2.0 is installed as @command{ogcc}
5060 and that for version 2.1 is installed as @command{gcc}, then the command
5061 @command{gcc} will use version 2.1 by default, while @command{ogcc} will use
5062 2.0 by default. However, you can choose either version with either
5063 command with the @option{-V} option.
5065 @node Submodel Options
5066 @section Hardware Models and Configurations
5067 @cindex submodel options
5068 @cindex specifying hardware config
5069 @cindex hardware models and configurations, specifying
5070 @cindex machine dependent options
5072 Earlier we discussed the standard option @option{-b} which chooses among
5073 different installed compilers for completely different target
5074 machines, such as VAX vs.@: 68000 vs.@: 80386.
5076 In addition, each of these target machine types can have its own
5077 special options, starting with @samp{-m}, to choose among various
5078 hardware models or configurations---for example, 68010 vs 68020,
5079 floating coprocessor or none. A single installed version of the
5080 compiler can compile for any model or configuration, according to the
5083 Some configurations of the compiler also support additional special
5084 options, usually for compatibility with other compilers on the same
5088 These options are defined by the macro @code{TARGET_SWITCHES} in the
5089 machine description. The default for the options is also defined by
5090 that macro, which enables you to change the defaults.
5105 * RS/6000 and PowerPC Options::
5108 * i386 and x86-64 Options::
5110 * Intel 960 Options::
5111 * DEC Alpha Options::
5115 * System V Options::
5116 * TMS320C3x/C4x Options::
5124 * S/390 and zSeries Options::
5129 @node M680x0 Options
5130 @subsection M680x0 Options
5131 @cindex M680x0 options
5133 These are the @samp{-m} options defined for the 68000 series. The default
5134 values for these options depends on which style of 68000 was selected when
5135 the compiler was configured; the defaults for the most common choices are
5143 Generate output for a 68000. This is the default
5144 when the compiler is configured for 68000-based systems.
5146 Use this option for microcontrollers with a 68000 or EC000 core,
5147 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
5153 Generate output for a 68020. This is the default
5154 when the compiler is configured for 68020-based systems.
5158 Generate output containing 68881 instructions for floating point.
5159 This is the default for most 68020 systems unless @option{--nfp} was
5160 specified when the compiler was configured.
5164 Generate output for a 68030. This is the default when the compiler is
5165 configured for 68030-based systems.
5169 Generate output for a 68040. This is the default when the compiler is
5170 configured for 68040-based systems.
5172 This option inhibits the use of 68881/68882 instructions that have to be
5173 emulated by software on the 68040. Use this option if your 68040 does not
5174 have code to emulate those instructions.
5178 Generate output for a 68060. This is the default when the compiler is
5179 configured for 68060-based systems.
5181 This option inhibits the use of 68020 and 68881/68882 instructions that
5182 have to be emulated by software on the 68060. Use this option if your 68060
5183 does not have code to emulate those instructions.
5187 Generate output for a CPU32. This is the default
5188 when the compiler is configured for CPU32-based systems.
5190 Use this option for microcontrollers with a
5191 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
5192 68336, 68340, 68341, 68349 and 68360.
5196 Generate output for a 520X ``coldfire'' family cpu. This is the default
5197 when the compiler is configured for 520X-based systems.
5199 Use this option for microcontroller with a 5200 core, including
5200 the MCF5202, MCF5203, MCF5204 and MCF5202.
5205 Generate output for a 68040, without using any of the new instructions.
5206 This results in code which can run relatively efficiently on either a
5207 68020/68881 or a 68030 or a 68040. The generated code does use the
5208 68881 instructions that are emulated on the 68040.
5212 Generate output for a 68060, without using any of the new instructions.
5213 This results in code which can run relatively efficiently on either a
5214 68020/68881 or a 68030 or a 68040. The generated code does use the
5215 68881 instructions that are emulated on the 68060.
5219 Generate output containing Sun FPA instructions for floating point.
5222 @opindex msoft-float
5223 Generate output containing library calls for floating point.
5224 @strong{Warning:} the requisite libraries are not available for all m68k
5225 targets. Normally the facilities of the machine's usual C compiler are
5226 used, but this can't be done directly in cross-compilation. You must
5227 make your own arrangements to provide suitable library functions for
5228 cross-compilation. The embedded targets @samp{m68k-*-aout} and
5229 @samp{m68k-*-coff} do provide software floating point support.
5233 Consider type @code{int} to be 16 bits wide, like @code{short int}.
5236 @opindex mnobitfield
5237 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
5238 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
5242 Do use the bit-field instructions. The @option{-m68020} option implies
5243 @option{-mbitfield}. This is the default if you use a configuration
5244 designed for a 68020.
5248 Use a different function-calling convention, in which functions
5249 that take a fixed number of arguments return with the @code{rtd}
5250 instruction, which pops their arguments while returning. This
5251 saves one instruction in the caller since there is no need to pop
5252 the arguments there.
5254 This calling convention is incompatible with the one normally
5255 used on Unix, so you cannot use it if you need to call libraries
5256 compiled with the Unix compiler.
5258 Also, you must provide function prototypes for all functions that
5259 take variable numbers of arguments (including @code{printf});
5260 otherwise incorrect code will be generated for calls to those
5263 In addition, seriously incorrect code will result if you call a
5264 function with too many arguments. (Normally, extra arguments are
5265 harmlessly ignored.)
5267 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
5268 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
5271 @itemx -mno-align-int
5273 @opindex mno-align-int
5274 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
5275 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
5276 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
5277 Aligning variables on 32-bit boundaries produces code that runs somewhat
5278 faster on processors with 32-bit busses at the expense of more memory.
5280 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
5281 align structures containing the above types differently than
5282 most published application binary interface specifications for the m68k.
5286 Use the pc-relative addressing mode of the 68000 directly, instead of
5287 using a global offset table. At present, this option implies @option{-fpic},
5288 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
5289 not presently supported with @option{-mpcrel}, though this could be supported for
5290 68020 and higher processors.
5292 @item -mno-strict-align
5293 @itemx -mstrict-align
5294 @opindex mno-strict-align
5295 @opindex mstrict-align
5296 Do not (do) assume that unaligned memory references will be handled by
5301 @node M68hc1x Options
5302 @subsection M68hc1x Options
5303 @cindex M68hc1x options
5305 These are the @samp{-m} options defined for the 68hc11 and 68hc12
5306 microcontrollers. The default values for these options depends on
5307 which style of microcontroller was selected when the compiler was configured;
5308 the defaults for the most common choices are given below.
5315 Generate output for a 68HC11. This is the default
5316 when the compiler is configured for 68HC11-based systems.
5322 Generate output for a 68HC12. This is the default
5323 when the compiler is configured for 68HC12-based systems.
5326 @opindex mauto-incdec
5327 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
5332 Consider type @code{int} to be 16 bits wide, like @code{short int}.
5334 @item -msoft-reg-count=@var{count}
5335 @opindex msoft-reg-count
5336 Specify the number of pseudo-soft registers which are used for the
5337 code generation. The maximum number is 32. Using more pseudo-soft
5338 register may or may not result in better code depending on the program.
5339 The default is 4 for 68HC11 and 2 for 68HC12.
5344 @subsection VAX Options
5347 These @samp{-m} options are defined for the VAX:
5352 Do not output certain jump instructions (@code{aobleq} and so on)
5353 that the Unix assembler for the VAX cannot handle across long
5358 Do output those jump instructions, on the assumption that you
5359 will assemble with the GNU assembler.
5363 Output code for g-format floating point numbers instead of d-format.
5367 @subsection SPARC Options
5368 @cindex SPARC options
5370 These @samp{-m} switches are supported on the SPARC:
5375 @opindex mno-app-regs
5377 Specify @option{-mapp-regs} to generate output using the global registers
5378 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
5381 To be fully SVR4 ABI compliant at the cost of some performance loss,
5382 specify @option{-mno-app-regs}. You should compile libraries and system
5383 software with this option.
5388 @opindex mhard-float
5389 Generate output containing floating point instructions. This is the
5395 @opindex msoft-float
5396 Generate output containing library calls for floating point.
5397 @strong{Warning:} the requisite libraries are not available for all SPARC
5398 targets. Normally the facilities of the machine's usual C compiler are
5399 used, but this cannot be done directly in cross-compilation. You must make
5400 your own arrangements to provide suitable library functions for
5401 cross-compilation. The embedded targets @samp{sparc-*-aout} and
5402 @samp{sparclite-*-*} do provide software floating point support.
5404 @option{-msoft-float} changes the calling convention in the output file;
5405 therefore, it is only useful if you compile @emph{all} of a program with
5406 this option. In particular, you need to compile @file{libgcc.a}, the
5407 library that comes with GCC, with @option{-msoft-float} in order for
5410 @item -mhard-quad-float
5411 @opindex mhard-quad-float
5412 Generate output containing quad-word (long double) floating point
5415 @item -msoft-quad-float
5416 @opindex msoft-quad-float
5417 Generate output containing library calls for quad-word (long double)
5418 floating point instructions. The functions called are those specified
5419 in the SPARC ABI@. This is the default.
5421 As of this writing, there are no sparc implementations that have hardware
5422 support for the quad-word floating point instructions. They all invoke
5423 a trap handler for one of these instructions, and then the trap handler
5424 emulates the effect of the instruction. Because of the trap handler overhead,
5425 this is much slower than calling the ABI library routines. Thus the
5426 @option{-msoft-quad-float} option is the default.
5430 @opindex mno-epilogue
5432 With @option{-mepilogue} (the default), the compiler always emits code for
5433 function exit at the end of each function. Any function exit in
5434 the middle of the function (such as a return statement in C) will
5435 generate a jump to the exit code at the end of the function.
5437 With @option{-mno-epilogue}, the compiler tries to emit exit code inline
5438 at every function exit.
5444 With @option{-mflat}, the compiler does not generate save/restore instructions
5445 and will use a ``flat'' or single register window calling convention.
5446 This model uses %i7 as the frame pointer and is compatible with the normal
5447 register window model. Code from either may be intermixed.
5448 The local registers and the input registers (0--5) are still treated as
5449 ``call saved'' registers and will be saved on the stack as necessary.
5451 With @option{-mno-flat} (the default), the compiler emits save/restore
5452 instructions (except for leaf functions) and is the normal mode of operation.
5454 @item -mno-unaligned-doubles
5455 @itemx -munaligned-doubles
5456 @opindex mno-unaligned-doubles
5457 @opindex munaligned-doubles
5458 Assume that doubles have 8 byte alignment. This is the default.
5460 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
5461 alignment only if they are contained in another type, or if they have an
5462 absolute address. Otherwise, it assumes they have 4 byte alignment.
5463 Specifying this option avoids some rare compatibility problems with code
5464 generated by other compilers. It is not the default because it results
5465 in a performance loss, especially for floating point code.
5467 @item -mno-faster-structs
5468 @itemx -mfaster-structs
5469 @opindex mno-faster-structs
5470 @opindex mfaster-structs
5471 With @option{-mfaster-structs}, the compiler assumes that structures
5472 should have 8 byte alignment. This enables the use of pairs of
5473 @code{ldd} and @code{std} instructions for copies in structure
5474 assignment, in place of twice as many @code{ld} and @code{st} pairs.
5475 However, the use of this changed alignment directly violates the Sparc
5476 ABI@. Thus, it's intended only for use on targets where the developer
5477 acknowledges that their resulting code will not be directly in line with
5478 the rules of the ABI@.
5484 These two options select variations on the SPARC architecture.
5486 By default (unless specifically configured for the Fujitsu SPARClite),
5487 GCC generates code for the v7 variant of the SPARC architecture.
5489 @option{-mv8} will give you SPARC v8 code. The only difference from v7
5490 code is that the compiler emits the integer multiply and integer
5491 divide instructions which exist in SPARC v8 but not in SPARC v7.
5493 @option{-msparclite} will give you SPARClite code. This adds the integer
5494 multiply, integer divide step and scan (@code{ffs}) instructions which
5495 exist in SPARClite but not in SPARC v7.
5497 These options are deprecated and will be deleted in a future GCC release.
5498 They have been replaced with @option{-mcpu=xxx}.
5503 @opindex msupersparc
5504 These two options select the processor for which the code is optimised.
5506 With @option{-mcypress} (the default), the compiler optimizes code for the
5507 Cypress CY7C602 chip, as used in the SparcStation/SparcServer 3xx series.
5508 This is also appropriate for the older SparcStation 1, 2, IPX etc.
5510 With @option{-msupersparc} the compiler optimizes code for the SuperSparc cpu, as
5511 used in the SparcStation 10, 1000 and 2000 series. This flag also enables use
5512 of the full SPARC v8 instruction set.
5514 These options are deprecated and will be deleted in a future GCC release.
5515 They have been replaced with @option{-mcpu=xxx}.
5517 @item -mcpu=@var{cpu_type}
5519 Set the instruction set, register set, and instruction scheduling parameters
5520 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
5521 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
5522 @samp{hypersparc}, @samp{sparclite86x}, @samp{f930}, @samp{f934},
5523 @samp{sparclet}, @samp{tsc701}, @samp{v9}, and @samp{ultrasparc}.
5525 Default instruction scheduling parameters are used for values that select
5526 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
5527 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
5529 Here is a list of each supported architecture and their supported
5534 v8: supersparc, hypersparc
5535 sparclite: f930, f934, sparclite86x
5540 @item -mtune=@var{cpu_type}
5542 Set the instruction scheduling parameters for machine type
5543 @var{cpu_type}, but do not set the instruction set or register set that the
5544 option @option{-mcpu=@var{cpu_type}} would.
5546 The same values for @option{-mcpu=@var{cpu_type}} can be used for
5547 @option{-mtune=@var{cpu_type}}, but the only useful values are those
5548 that select a particular cpu implementation. Those are @samp{cypress},
5549 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
5550 @samp{sparclite86x}, @samp{tsc701}, and @samp{ultrasparc}.
5554 These @samp{-m} switches are supported in addition to the above
5555 on the SPARCLET processor.
5558 @item -mlittle-endian
5559 @opindex mlittle-endian
5560 Generate code for a processor running in little-endian mode.
5564 Treat register @code{%g0} as a normal register.
5565 GCC will continue to clobber it as necessary but will not assume
5566 it always reads as 0.
5568 @item -mbroken-saverestore
5569 @opindex mbroken-saverestore
5570 Generate code that does not use non-trivial forms of the @code{save} and
5571 @code{restore} instructions. Early versions of the SPARCLET processor do
5572 not correctly handle @code{save} and @code{restore} instructions used with
5573 arguments. They correctly handle them used without arguments. A @code{save}
5574 instruction used without arguments increments the current window pointer
5575 but does not allocate a new stack frame. It is assumed that the window
5576 overflow trap handler will properly handle this case as will interrupt
5580 These @samp{-m} switches are supported in addition to the above
5581 on SPARC V9 processors in 64-bit environments.
5584 @item -mlittle-endian
5585 @opindex mlittle-endian
5586 Generate code for a processor running in little-endian mode.
5592 Generate code for a 32-bit or 64-bit environment.
5593 The 32-bit environment sets int, long and pointer to 32 bits.
5594 The 64-bit environment sets int to 32 bits and long and pointer
5597 @item -mcmodel=medlow
5598 @opindex mcmodel=medlow
5599 Generate code for the Medium/Low code model: the program must be linked
5600 in the low 32 bits of the address space. Pointers are 64 bits.
5601 Programs can be statically or dynamically linked.
5603 @item -mcmodel=medmid
5604 @opindex mcmodel=medmid
5605 Generate code for the Medium/Middle code model: the program must be linked
5606 in the low 44 bits of the address space, the text segment must be less than
5607 2G bytes, and data segment must be within 2G of the text segment.
5608 Pointers are 64 bits.
5610 @item -mcmodel=medany
5611 @opindex mcmodel=medany
5612 Generate code for the Medium/Anywhere code model: the program may be linked
5613 anywhere in the address space, the text segment must be less than
5614 2G bytes, and data segment must be within 2G of the text segment.
5615 Pointers are 64 bits.
5617 @item -mcmodel=embmedany
5618 @opindex mcmodel=embmedany
5619 Generate code for the Medium/Anywhere code model for embedded systems:
5620 assume a 32-bit text and a 32-bit data segment, both starting anywhere
5621 (determined at link time). Register %g4 points to the base of the
5622 data segment. Pointers are still 64 bits.
5623 Programs are statically linked, PIC is not supported.
5626 @itemx -mno-stack-bias
5627 @opindex mstack-bias
5628 @opindex mno-stack-bias
5629 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
5630 frame pointer if present, are offset by @minus{}2047 which must be added back
5631 when making stack frame references.
5632 Otherwise, assume no such offset is present.
5635 @node Convex Options
5636 @subsection Convex Options
5637 @cindex Convex options
5639 These @samp{-m} options are defined for Convex:
5644 Generate output for C1. The code will run on any Convex machine.
5645 The preprocessor symbol @code{__convex__c1__} is defined.
5649 Generate output for C2. Uses instructions not available on C1.
5650 Scheduling and other optimizations are chosen for max performance on C2.
5651 The preprocessor symbol @code{__convex_c2__} is defined.
5655 Generate output for C32xx. Uses instructions not available on C1.
5656 Scheduling and other optimizations are chosen for max performance on C32.
5657 The preprocessor symbol @code{__convex_c32__} is defined.
5661 Generate output for C34xx. Uses instructions not available on C1.
5662 Scheduling and other optimizations are chosen for max performance on C34.
5663 The preprocessor symbol @code{__convex_c34__} is defined.
5667 Generate output for C38xx. Uses instructions not available on C1.
5668 Scheduling and other optimizations are chosen for max performance on C38.
5669 The preprocessor symbol @code{__convex_c38__} is defined.
5673 Generate code which puts an argument count in the word preceding each
5674 argument list. This is compatible with regular CC, and a few programs
5675 may need the argument count word. GDB and other source-level debuggers
5676 do not need it; this info is in the symbol table.
5679 @opindex mnoargcount
5680 Omit the argument count word. This is the default.
5682 @item -mvolatile-cache
5683 @opindex mvolatile-cache
5684 Allow volatile references to be cached. This is the default.
5686 @item -mvolatile-nocache
5687 @opindex mvolatile-nocache
5688 Volatile references bypass the data cache, going all the way to memory.
5689 This is only needed for multi-processor code that does not use standard
5690 synchronization instructions. Making non-volatile references to volatile
5691 locations will not necessarily work.
5695 Type long is 32 bits, the same as type int. This is the default.
5699 Type long is 64 bits, the same as type long long. This option is useless,
5700 because no library support exists for it.
5703 @node AMD29K Options
5704 @subsection AMD29K Options
5705 @cindex AMD29K options
5707 These @samp{-m} options are defined for the AMD Am29000:
5712 @cindex DW bit (29k)
5713 Generate code that assumes the @code{DW} bit is set, i.e., that byte and
5714 halfword operations are directly supported by the hardware. This is the
5719 Generate code that assumes the @code{DW} bit is not set.
5723 @cindex byte writes (29k)
5724 Generate code that assumes the system supports byte and halfword write
5725 operations. This is the default.
5729 Generate code that assumes the systems does not support byte and
5730 halfword write operations. @option{-mnbw} implies @option{-mndw}.
5734 @cindex memory model (29k)
5735 Use a small memory model that assumes that all function addresses are
5736 either within a single 256 KB segment or at an absolute address of less
5737 than 256k. This allows the @code{call} instruction to be used instead
5738 of a @code{const}, @code{consth}, @code{calli} sequence.
5742 Use the normal memory model: Generate @code{call} instructions only when
5743 calling functions in the same file and @code{calli} instructions
5744 otherwise. This works if each file occupies less than 256 KB but allows
5745 the entire executable to be larger than 256 KB@. This is the default.
5749 Always use @code{calli} instructions. Specify this option if you expect
5750 a single file to compile into more than 256 KB of code.
5754 @cindex processor selection (29k)
5755 Generate code for the Am29050.
5759 Generate code for the Am29000. This is the default.
5761 @item -mkernel-registers
5762 @opindex mkernel-registers
5763 @cindex kernel and user registers (29k)
5764 Generate references to registers @code{gr64-gr95} instead of to
5765 registers @code{gr96-gr127}. This option can be used when compiling
5766 kernel code that wants a set of global registers disjoint from that used
5769 Note that when this option is used, register names in @samp{-f} flags
5770 must use the normal, user-mode, names.
5772 @item -muser-registers
5773 @opindex muser-registers
5774 Use the normal set of global registers, @code{gr96-gr127}. This is the
5778 @itemx -mno-stack-check
5779 @opindex mstack-check
5780 @opindex mno-stack-check
5781 @cindex stack checks (29k)
5782 Insert (or do not insert) a call to @code{__msp_check} after each stack
5783 adjustment. This is often used for kernel code.
5786 @itemx -mno-storem-bug
5787 @opindex mstorem-bug
5788 @opindex mno-storem-bug
5789 @cindex storem bug (29k)
5790 @option{-mstorem-bug} handles 29k processors which cannot handle the
5791 separation of a mtsrim insn and a storem instruction (most 29000 chips
5792 to date, but not the 29050).
5794 @item -mno-reuse-arg-regs
5795 @itemx -mreuse-arg-regs
5796 @opindex mno-reuse-arg-regs
5797 @opindex mreuse-arg-regs
5798 @option{-mno-reuse-arg-regs} tells the compiler to only use incoming argument
5799 registers for copying out arguments. This helps detect calling a function
5800 with fewer arguments than it was declared with.
5802 @item -mno-impure-text
5803 @itemx -mimpure-text
5804 @opindex mno-impure-text
5805 @opindex mimpure-text
5806 @option{-mimpure-text}, used in addition to @option{-shared}, tells the compiler to
5807 not pass @option{-assert pure-text} to the linker when linking a shared object.
5810 @opindex msoft-float
5811 Generate output containing library calls for floating point.
5812 @strong{Warning:} the requisite libraries are not part of GCC@.
5813 Normally the facilities of the machine's usual C compiler are used, but
5814 this can't be done directly in cross-compilation. You must make your
5815 own arrangements to provide suitable library functions for
5820 Do not generate multm or multmu instructions. This is useful for some embedded
5821 systems which do not have trap handlers for these instructions.
5825 @subsection ARM Options
5828 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
5833 @opindex mapcs-frame
5834 Generate a stack frame that is compliant with the ARM Procedure Call
5835 Standard for all functions, even if this is not strictly necessary for
5836 correct execution of the code. Specifying @option{-fomit-frame-pointer}
5837 with this option will cause the stack frames not to be generated for
5838 leaf functions. The default is @option{-mno-apcs-frame}.
5842 This is a synonym for @option{-mapcs-frame}.
5846 Generate code for a processor running with a 26-bit program counter,
5847 and conforming to the function calling standards for the APCS 26-bit
5848 option. This option replaces the @option{-m2} and @option{-m3} options
5849 of previous releases of the compiler.
5853 Generate code for a processor running with a 32-bit program counter,
5854 and conforming to the function calling standards for the APCS 32-bit
5855 option. This option replaces the @option{-m6} option of previous releases
5859 @c not currently implemented
5860 @item -mapcs-stack-check
5861 @opindex mapcs-stack-check
5862 Generate code to check the amount of stack space available upon entry to
5863 every function (that actually uses some stack space). If there is
5864 insufficient space available then either the function
5865 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
5866 called, depending upon the amount of stack space required. The run time
5867 system is required to provide these functions. The default is
5868 @option{-mno-apcs-stack-check}, since this produces smaller code.
5870 @c not currently implemented
5872 @opindex mapcs-float
5873 Pass floating point arguments using the float point registers. This is
5874 one of the variants of the APCS@. This option is recommended if the
5875 target hardware has a floating point unit or if a lot of floating point
5876 arithmetic is going to be performed by the code. The default is
5877 @option{-mno-apcs-float}, since integer only code is slightly increased in
5878 size if @option{-mapcs-float} is used.
5880 @c not currently implemented
5881 @item -mapcs-reentrant
5882 @opindex mapcs-reentrant
5883 Generate reentrant, position independent code. The default is
5884 @option{-mno-apcs-reentrant}.
5887 @item -mthumb-interwork
5888 @opindex mthumb-interwork
5889 Generate code which supports calling between the ARM and Thumb
5890 instruction sets. Without this option the two instruction sets cannot
5891 be reliably used inside one program. The default is
5892 @option{-mno-thumb-interwork}, since slightly larger code is generated
5893 when @option{-mthumb-interwork} is specified.
5895 @item -mno-sched-prolog
5896 @opindex mno-sched-prolog
5897 Prevent the reordering of instructions in the function prolog, or the
5898 merging of those instruction with the instructions in the function's
5899 body. This means that all functions will start with a recognizable set
5900 of instructions (or in fact one of a choice from a small set of
5901 different function prologues), and this information can be used to
5902 locate the start if functions inside an executable piece of code. The
5903 default is @option{-msched-prolog}.
5906 @opindex mhard-float
5907 Generate output containing floating point instructions. This is the
5911 @opindex msoft-float
5912 Generate output containing library calls for floating point.
5913 @strong{Warning:} the requisite libraries are not available for all ARM
5914 targets. Normally the facilities of the machine's usual C compiler are
5915 used, but this cannot be done directly in cross-compilation. You must make
5916 your own arrangements to provide suitable library functions for
5919 @option{-msoft-float} changes the calling convention in the output file;
5920 therefore, it is only useful if you compile @emph{all} of a program with
5921 this option. In particular, you need to compile @file{libgcc.a}, the
5922 library that comes with GCC, with @option{-msoft-float} in order for
5925 @item -mlittle-endian
5926 @opindex mlittle-endian
5927 Generate code for a processor running in little-endian mode. This is
5928 the default for all standard configurations.
5931 @opindex mbig-endian
5932 Generate code for a processor running in big-endian mode; the default is
5933 to compile code for a little-endian processor.
5935 @item -mwords-little-endian
5936 @opindex mwords-little-endian
5937 This option only applies when generating code for big-endian processors.
5938 Generate code for a little-endian word order but a big-endian byte
5939 order. That is, a byte order of the form @samp{32107654}. Note: this
5940 option should only be used if you require compatibility with code for
5941 big-endian ARM processors generated by versions of the compiler prior to
5944 @item -malignment-traps
5945 @opindex malignment-traps
5946 Generate code that will not trap if the MMU has alignment traps enabled.
5947 On ARM architectures prior to ARMv4, there were no instructions to
5948 access half-word objects stored in memory. However, when reading from
5949 memory a feature of the ARM architecture allows a word load to be used,
5950 even if the address is unaligned, and the processor core will rotate the
5951 data as it is being loaded. This option tells the compiler that such
5952 misaligned accesses will cause a MMU trap and that it should instead
5953 synthesise the access as a series of byte accesses. The compiler can
5954 still use word accesses to load half-word data if it knows that the
5955 address is aligned to a word boundary.
5957 This option is ignored when compiling for ARM architecture 4 or later,
5958 since these processors have instructions to directly access half-word
5961 @item -mno-alignment-traps
5962 @opindex mno-alignment-traps
5963 Generate code that assumes that the MMU will not trap unaligned
5964 accesses. This produces better code when the target instruction set
5965 does not have half-word memory operations (i.e.@: implementations prior to
5968 Note that you cannot use this option to access unaligned word objects,
5969 since the processor will only fetch one 32-bit aligned object from
5972 The default setting for most targets is @option{-mno-alignment-traps}, since
5973 this produces better code when there are no half-word memory
5974 instructions available.
5976 @item -mshort-load-bytes
5977 @itemx -mno-short-load-words
5978 @opindex mshort-load-bytes
5979 @opindex mno-short-load-words
5980 These are deprecated aliases for @option{-malignment-traps}.
5982 @item -mno-short-load-bytes
5983 @itemx -mshort-load-words
5984 @opindex mno-short-load-bytes
5985 @opindex mshort-load-words
5986 This are deprecated aliases for @option{-mno-alignment-traps}.
5990 This option only applies to RISC iX@. Emulate the native BSD-mode
5991 compiler. This is the default if @option{-ansi} is not specified.
5995 This option only applies to RISC iX@. Emulate the native X/Open-mode
5998 @item -mno-symrename
5999 @opindex mno-symrename
6000 This option only applies to RISC iX@. Do not run the assembler
6001 post-processor, @samp{symrename}, after code has been assembled.
6002 Normally it is necessary to modify some of the standard symbols in
6003 preparation for linking with the RISC iX C library; this option
6004 suppresses this pass. The post-processor is never run when the
6005 compiler is built for cross-compilation.
6007 @item -mcpu=@var{name}
6009 This specifies the name of the target ARM processor. GCC uses this name
6010 to determine what kind of instructions it can emit when generating
6011 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
6012 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
6013 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
6014 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
6015 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
6016 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm8},
6017 @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
6018 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
6019 @samp{arm920t}, @samp{arm940t}, @samp{arm9tdmi}, @samp{arm10tdmi},
6020 @samp{arm1020t}, @samp{xscale}.
6022 @itemx -mtune=@var{name}
6024 This option is very similar to the @option{-mcpu=} option, except that
6025 instead of specifying the actual target processor type, and hence
6026 restricting which instructions can be used, it specifies that GCC should
6027 tune the performance of the code as if the target were of the type
6028 specified in this option, but still choosing the instructions that it
6029 will generate based on the cpu specified by a @option{-mcpu=} option.
6030 For some ARM implementations better performance can be obtained by using
6033 @item -march=@var{name}
6035 This specifies the name of the target ARM architecture. GCC uses this
6036 name to determine what kind of instructions it can emit when generating
6037 assembly code. This option can be used in conjunction with or instead
6038 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
6039 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
6040 @samp{armv5}, @samp{armv5t}, @samp{armv5te}.
6042 @item -mfpe=@var{number}
6043 @itemx -mfp=@var{number}
6046 This specifies the version of the floating point emulation available on
6047 the target. Permissible values are 2 and 3. @option{-mfp=} is a synonym
6048 for @option{-mfpe=}, for compatibility with older versions of GCC@.
6050 @item -mstructure-size-boundary=@var{n}
6051 @opindex mstructure-size-boundary
6052 The size of all structures and unions will be rounded up to a multiple
6053 of the number of bits set by this option. Permissible values are 8 and
6054 32. The default value varies for different toolchains. For the COFF
6055 targeted toolchain the default value is 8. Specifying the larger number
6056 can produce faster, more efficient code, but can also increase the size
6057 of the program. The two values are potentially incompatible. Code
6058 compiled with one value cannot necessarily expect to work with code or
6059 libraries compiled with the other value, if they exchange information
6060 using structures or unions.
6062 @item -mabort-on-noreturn
6063 @opindex mabort-on-noreturn
6064 Generate a call to the function @code{abort} at the end of a
6065 @code{noreturn} function. It will be executed if the function tries to
6069 @itemx -mno-long-calls
6070 @opindex mlong-calls
6071 @opindex mno-long-calls
6072 Tells the compiler to perform function calls by first loading the
6073 address of the function into a register and then performing a subroutine
6074 call on this register. This switch is needed if the target function
6075 will lie outside of the 64 megabyte addressing range of the offset based
6076 version of subroutine call instruction.
6078 Even if this switch is enabled, not all function calls will be turned
6079 into long calls. The heuristic is that static functions, functions
6080 which have the @samp{short-call} attribute, functions that are inside
6081 the scope of a @samp{#pragma no_long_calls} directive and functions whose
6082 definitions have already been compiled within the current compilation
6083 unit, will not be turned into long calls. The exception to this rule is
6084 that weak function definitions, functions with the @samp{long-call}
6085 attribute or the @samp{section} attribute, and functions that are within
6086 the scope of a @samp{#pragma long_calls} directive, will always be
6087 turned into long calls.
6089 This feature is not enabled by default. Specifying
6090 @option{-mno-long-calls} will restore the default behaviour, as will
6091 placing the function calls within the scope of a @samp{#pragma
6092 long_calls_off} directive. Note these switches have no effect on how
6093 the compiler generates code to handle function calls via function
6096 @item -mnop-fun-dllimport
6097 @opindex mnop-fun-dllimport
6098 Disable support for the @code{dllimport} attribute.
6100 @item -msingle-pic-base
6101 @opindex msingle-pic-base
6102 Treat the register used for PIC addressing as read-only, rather than
6103 loading it in the prologue for each function. The run-time system is
6104 responsible for initialising this register with an appropriate value
6105 before execution begins.
6107 @item -mpic-register=@var{reg}
6108 @opindex mpic-register
6109 Specify the register to be used for PIC addressing. The default is R10
6110 unless stack-checking is enabled, when R9 is used.
6112 @item -mpoke-function-name
6113 @opindex mpoke-function-name
6114 Write the name of each function into the text section, directly
6115 preceding the function prologue. The generated code is similar to this:
6119 .ascii "arm_poke_function_name", 0
6122 .word 0xff000000 + (t1 - t0)
6123 arm_poke_function_name
6125 stmfd sp!, @{fp, ip, lr, pc@}
6129 When performing a stack backtrace, code can inspect the value of
6130 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
6131 location @code{pc - 12} and the top 8 bits are set, then we know that
6132 there is a function name embedded immediately preceding this location
6133 and has length @code{((pc[-3]) & 0xff000000)}.
6137 Generate code for the 16-bit Thumb instruction set. The default is to
6138 use the 32-bit ARM instruction set.
6141 @opindex mtpcs-frame
6142 Generate a stack frame that is compliant with the Thumb Procedure Call
6143 Standard for all non-leaf functions. (A leaf function is one that does
6144 not call any other functions.) The default is @option{-mno-tpcs-frame}.
6146 @item -mtpcs-leaf-frame
6147 @opindex mtpcs-leaf-frame
6148 Generate a stack frame that is compliant with the Thumb Procedure Call
6149 Standard for all leaf functions. (A leaf function is one that does
6150 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
6152 @item -mcallee-super-interworking
6153 @opindex mcallee-super-interworking
6154 Gives all externally visible functions in the file being compiled an ARM
6155 instruction set header which switches to Thumb mode before executing the
6156 rest of the function. This allows these functions to be called from
6157 non-interworking code.
6159 @item -mcaller-super-interworking
6160 @opindex mcaller-super-interworking
6161 Allows calls via function pointers (including virtual functions) to
6162 execute correctly regardless of whether the target code has been
6163 compiled for interworking or not. There is a small overhead in the cost
6164 of executing a function pointer if this option is enabled.
6168 @node MN10200 Options
6169 @subsection MN10200 Options
6170 @cindex MN10200 options
6171 These @option{-m} options are defined for Matsushita MN10200 architectures:
6176 Indicate to the linker that it should perform a relaxation optimization pass
6177 to shorten branches, calls and absolute memory addresses. This option only
6178 has an effect when used on the command line for the final link step.
6180 This option makes symbolic debugging impossible.
6183 @node MN10300 Options
6184 @subsection MN10300 Options
6185 @cindex MN10300 options
6186 These @option{-m} options are defined for Matsushita MN10300 architectures:
6191 Generate code to avoid bugs in the multiply instructions for the MN10300
6192 processors. This is the default.
6195 @opindex mno-mult-bug
6196 Do not generate code to avoid bugs in the multiply instructions for the
6201 Generate code which uses features specific to the AM33 processor.
6205 Do not generate code which uses features specific to the AM33 processor. This
6210 Do not link in the C run-time initialization object file.
6214 Indicate to the linker that it should perform a relaxation optimization pass
6215 to shorten branches, calls and absolute memory addresses. This option only
6216 has an effect when used on the command line for the final link step.
6218 This option makes symbolic debugging impossible.
6222 @node M32R/D Options
6223 @subsection M32R/D Options
6224 @cindex M32R/D options
6226 These @option{-m} options are defined for Mitsubishi M32R/D architectures:
6231 Generate code for the M32R/X. The default is to generate code for the M32R.
6233 @item -mcode-model=small
6234 @opindex mcode-model=small
6235 Assume all objects live in the lower 16MB of memory (so that their addresses
6236 can be loaded with the @code{ld24} instruction), and assume all subroutines
6237 are reachable with the @code{bl} instruction.
6238 This is the default.
6240 The addressability of a particular object can be set with the
6241 @code{model} attribute.
6243 @item -mcode-model=medium
6244 @opindex mcode-model=medium
6245 Assume objects may be anywhere in the 32-bit address space (the compiler
6246 will generate @code{seth/add3} instructions to load their addresses), and
6247 assume all subroutines are reachable with the @code{bl} instruction.
6249 @item -mcode-model=large
6250 @opindex mcode-model=large
6251 Assume objects may be anywhere in the 32-bit address space (the compiler
6252 will generate @code{seth/add3} instructions to load their addresses), and
6253 assume subroutines may not be reachable with the @code{bl} instruction
6254 (the compiler will generate the much slower @code{seth/add3/jl}
6255 instruction sequence).
6258 @opindex msdata=none
6259 Disable use of the small data area. Variables will be put into
6260 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
6261 @code{section} attribute has been specified).
6262 This is the default.
6264 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
6265 Objects may be explicitly put in the small data area with the
6266 @code{section} attribute using one of these sections.
6269 @opindex msdata=sdata
6270 Put small global and static data in the small data area, but do not
6271 generate special code to reference them.
6275 Put small global and static data in the small data area, and generate
6276 special instructions to reference them.
6280 @cindex smaller data references
6281 Put global and static objects less than or equal to @var{num} bytes
6282 into the small data or bss sections instead of the normal data or bss
6283 sections. The default value of @var{num} is 8.
6284 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
6285 for this option to have any effect.
6287 All modules should be compiled with the same @option{-G @var{num}} value.
6288 Compiling with different values of @var{num} may or may not work; if it
6289 doesn't the linker will give an error message---incorrect code will not be
6295 @subsection M88K Options
6296 @cindex M88k options
6298 These @samp{-m} options are defined for Motorola 88k architectures:
6303 Generate code that works well on both the m88100 and the
6308 Generate code that works best for the m88100, but that also
6313 Generate code that works best for the m88110, and may not run
6318 Obsolete option to be removed from the next revision.
6321 @item -midentify-revision
6322 @opindex midentify-revision
6323 @cindex identifying source, compiler (88k)
6324 Include an @code{ident} directive in the assembler output recording the
6325 source file name, compiler name and version, timestamp, and compilation
6328 @item -mno-underscores
6329 @opindex mno-underscores
6330 @cindex underscores, avoiding (88k)
6331 In assembler output, emit symbol names without adding an underscore
6332 character at the beginning of each name. The default is to use an
6333 underscore as prefix on each name.
6335 @item -mocs-debug-info
6336 @itemx -mno-ocs-debug-info
6337 @opindex mocs-debug-info
6338 @opindex mno-ocs-debug-info
6340 @cindex debugging, 88k OCS
6341 Include (or omit) additional debugging information (about registers used
6342 in each stack frame) as specified in the 88open Object Compatibility
6343 Standard, ``OCS''@. This extra information allows debugging of code that
6344 has had the frame pointer eliminated. The default for DG/UX, SVr4, and
6345 Delta 88 SVr3.2 is to include this information; other 88k configurations
6346 omit this information by default.
6348 @item -mocs-frame-position
6349 @opindex mocs-frame-position
6350 @cindex register positions in frame (88k)
6351 When emitting COFF debugging information for automatic variables and
6352 parameters stored on the stack, use the offset from the canonical frame
6353 address, which is the stack pointer (register 31) on entry to the
6354 function. The DG/UX, SVr4, Delta88 SVr3.2, and BCS configurations use
6355 @option{-mocs-frame-position}; other 88k configurations have the default
6356 @option{-mno-ocs-frame-position}.
6358 @item -mno-ocs-frame-position
6359 @opindex mno-ocs-frame-position
6360 @cindex register positions in frame (88k)
6361 When emitting COFF debugging information for automatic variables and
6362 parameters stored on the stack, use the offset from the frame pointer
6363 register (register 30). When this option is in effect, the frame
6364 pointer is not eliminated when debugging information is selected by the
6367 @item -moptimize-arg-area
6368 @opindex moptimize-arg-area
6369 @cindex arguments in frame (88k)
6370 Save space by reorganizing the stack frame. This option generates code
6371 that does not agree with the 88open specifications, but uses less
6374 @itemx -mno-optimize-arg-area
6375 @opindex mno-optimize-arg-area
6376 Do not reorganize the stack frame to save space. This is the default.
6377 The generated conforms to the specification, but uses more memory.
6379 @item -mshort-data-@var{num}
6380 @opindex mshort-data
6381 @cindex smaller data references (88k)
6382 @cindex r0-relative references (88k)
6383 Generate smaller data references by making them relative to @code{r0},
6384 which allows loading a value using a single instruction (rather than the
6385 usual two). You control which data references are affected by
6386 specifying @var{num} with this option. For example, if you specify
6387 @option{-mshort-data-512}, then the data references affected are those
6388 involving displacements of less than 512 bytes.
6389 @option{-mshort-data-@var{num}} is not effective for @var{num} greater
6392 @item -mserialize-volatile
6393 @opindex mserialize-volatile
6394 @itemx -mno-serialize-volatile
6395 @opindex mno-serialize-volatile
6396 @cindex sequential consistency on 88k
6397 Do, or don't, generate code to guarantee sequential consistency
6398 of volatile memory references. By default, consistency is
6401 The order of memory references made by the MC88110 processor does
6402 not always match the order of the instructions requesting those
6403 references. In particular, a load instruction may execute before
6404 a preceding store instruction. Such reordering violates
6405 sequential consistency of volatile memory references, when there
6406 are multiple processors. When consistency must be guaranteed,
6407 GCC generates special instructions, as needed, to force
6408 execution in the proper order.
6410 The MC88100 processor does not reorder memory references and so
6411 always provides sequential consistency. However, by default, GCC
6412 generates the special instructions to guarantee consistency
6413 even when you use @option{-m88100}, so that the code may be run on an
6414 MC88110 processor. If you intend to run your code only on the
6415 MC88100 processor, you may use @option{-mno-serialize-volatile}.
6417 The extra code generated to guarantee consistency may affect the
6418 performance of your application. If you know that you can safely
6419 forgo this guarantee, you may use @option{-mno-serialize-volatile}.
6425 @cindex assembler syntax, 88k
6427 Turn on (@option{-msvr4}) or off (@option{-msvr3}) compiler extensions
6428 related to System V release 4 (SVr4). This controls the following:
6432 Which variant of the assembler syntax to emit.
6434 @option{-msvr4} makes the C preprocessor recognize @samp{#pragma weak}
6435 that is used on System V release 4.
6437 @option{-msvr4} makes GCC issue additional declaration directives used in
6441 @option{-msvr4} is the default for the m88k-motorola-sysv4 and
6442 m88k-dg-dgux m88k configurations. @option{-msvr3} is the default for all
6443 other m88k configurations.
6445 @item -mversion-03.00
6446 @opindex mversion-03.00
6447 This option is obsolete, and is ignored.
6448 @c ??? which asm syntax better for GAS? option there too?
6450 @item -mno-check-zero-division
6451 @itemx -mcheck-zero-division
6452 @opindex mno-check-zero-division
6453 @opindex mcheck-zero-division
6454 @cindex zero division on 88k
6455 Do, or don't, generate code to guarantee that integer division by
6456 zero will be detected. By default, detection is guaranteed.
6458 Some models of the MC88100 processor fail to trap upon integer
6459 division by zero under certain conditions. By default, when
6460 compiling code that might be run on such a processor, GCC
6461 generates code that explicitly checks for zero-valued divisors
6462 and traps with exception number 503 when one is detected. Use of
6463 @option{-mno-check-zero-division} suppresses such checking for code
6464 generated to run on an MC88100 processor.
6466 GCC assumes that the MC88110 processor correctly detects all instances
6467 of integer division by zero. When @option{-m88110} is specified, no
6468 explicit checks for zero-valued divisors are generated, and both
6469 @option{-mcheck-zero-division} and @option{-mno-check-zero-division} are
6472 @item -muse-div-instruction
6473 @opindex muse-div-instruction
6474 @cindex divide instruction, 88k
6475 Use the div instruction for signed integer division on the
6476 MC88100 processor. By default, the div instruction is not used.
6478 On the MC88100 processor the signed integer division instruction
6479 div) traps to the operating system on a negative operand. The
6480 operating system transparently completes the operation, but at a
6481 large cost in execution time. By default, when compiling code
6482 that might be run on an MC88100 processor, GCC emulates signed
6483 integer division using the unsigned integer division instruction
6484 divu), thereby avoiding the large penalty of a trap to the
6485 operating system. Such emulation has its own, smaller, execution
6486 cost in both time and space. To the extent that your code's
6487 important signed integer division operations are performed on two
6488 nonnegative operands, it may be desirable to use the div
6489 instruction directly.
6491 On the MC88110 processor the div instruction (also known as the
6492 divs instruction) processes negative operands without trapping to
6493 the operating system. When @option{-m88110} is specified,
6494 @option{-muse-div-instruction} is ignored, and the div instruction is used
6495 for signed integer division.
6497 Note that the result of dividing @code{INT_MIN} by @minus{}1 is undefined. In
6498 particular, the behavior of such a division with and without
6499 @option{-muse-div-instruction} may differ.
6501 @item -mtrap-large-shift
6502 @itemx -mhandle-large-shift
6503 @opindex mtrap-large-shift
6504 @opindex mhandle-large-shift
6505 @cindex bit shift overflow (88k)
6506 @cindex large bit shifts (88k)
6507 Include code to detect bit-shifts of more than 31 bits; respectively,
6508 trap such shifts or emit code to handle them properly. By default GCC
6509 makes no special provision for large bit shifts.
6511 @item -mwarn-passed-structs
6512 @opindex mwarn-passed-structs
6513 @cindex structure passing (88k)
6514 Warn when a function passes a struct as an argument or result.
6515 Structure-passing conventions have changed during the evolution of the C
6516 language, and are often the source of portability problems. By default,
6517 GCC issues no such warning.
6520 @c break page here to avoid unsightly interparagraph stretch.
6524 @node RS/6000 and PowerPC Options
6525 @subsection IBM RS/6000 and PowerPC Options
6526 @cindex RS/6000 and PowerPC Options
6527 @cindex IBM RS/6000 and PowerPC Options
6529 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
6537 @itemx -mpowerpc-gpopt
6538 @itemx -mno-powerpc-gpopt
6539 @itemx -mpowerpc-gfxopt
6540 @itemx -mno-powerpc-gfxopt
6542 @itemx -mno-powerpc64
6548 @opindex mno-powerpc
6549 @opindex mpowerpc-gpopt
6550 @opindex mno-powerpc-gpopt
6551 @opindex mpowerpc-gfxopt
6552 @opindex mno-powerpc-gfxopt
6554 @opindex mno-powerpc64
6555 GCC supports two related instruction set architectures for the
6556 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
6557 instructions supported by the @samp{rios} chip set used in the original
6558 RS/6000 systems and the @dfn{PowerPC} instruction set is the
6559 architecture of the Motorola MPC5xx, MPC6xx, MPC8xx microprocessors, and
6560 the IBM 4xx microprocessors.
6562 Neither architecture is a subset of the other. However there is a
6563 large common subset of instructions supported by both. An MQ
6564 register is included in processors supporting the POWER architecture.
6566 You use these options to specify which instructions are available on the
6567 processor you are using. The default value of these options is
6568 determined when configuring GCC@. Specifying the
6569 @option{-mcpu=@var{cpu_type}} overrides the specification of these
6570 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
6571 rather than the options listed above.
6573 The @option{-mpower} option allows GCC to generate instructions that
6574 are found only in the POWER architecture and to use the MQ register.
6575 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
6576 to generate instructions that are present in the POWER2 architecture but
6577 not the original POWER architecture.
6579 The @option{-mpowerpc} option allows GCC to generate instructions that
6580 are found only in the 32-bit subset of the PowerPC architecture.
6581 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
6582 GCC to use the optional PowerPC architecture instructions in the
6583 General Purpose group, including floating-point square root. Specifying
6584 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
6585 use the optional PowerPC architecture instructions in the Graphics
6586 group, including floating-point select.
6588 The @option{-mpowerpc64} option allows GCC to generate the additional
6589 64-bit instructions that are found in the full PowerPC64 architecture
6590 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
6591 @option{-mno-powerpc64}.
6593 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
6594 will use only the instructions in the common subset of both
6595 architectures plus some special AIX common-mode calls, and will not use
6596 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
6597 permits GCC to use any instruction from either architecture and to
6598 allow use of the MQ register; specify this for the Motorola MPC601.
6600 @item -mnew-mnemonics
6601 @itemx -mold-mnemonics
6602 @opindex mnew-mnemonics
6603 @opindex mold-mnemonics
6604 Select which mnemonics to use in the generated assembler code. With
6605 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
6606 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
6607 assembler mnemonics defined for the POWER architecture. Instructions
6608 defined in only one architecture have only one mnemonic; GCC uses that
6609 mnemonic irrespective of which of these options is specified.
6611 GCC defaults to the mnemonics appropriate for the architecture in
6612 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
6613 value of these option. Unless you are building a cross-compiler, you
6614 should normally not specify either @option{-mnew-mnemonics} or
6615 @option{-mold-mnemonics}, but should instead accept the default.
6617 @item -mcpu=@var{cpu_type}
6619 Set architecture type, register usage, choice of mnemonics, and
6620 instruction scheduling parameters for machine type @var{cpu_type}.
6621 Supported values for @var{cpu_type} are @samp{rios}, @samp{rios1},
6622 @samp{rsc}, @samp{rios2}, @samp{rs64a}, @samp{601}, @samp{602},
6623 @samp{603}, @samp{603e}, @samp{604}, @samp{604e}, @samp{620},
6624 @samp{630}, @samp{740}, @samp{7400}, @samp{7450}, @samp{750},
6625 @samp{power}, @samp{power2}, @samp{powerpc}, @samp{403}, @samp{505},
6626 @samp{801}, @samp{821}, @samp{823}, and @samp{860} and @samp{common}.
6628 @option{-mcpu=common} selects a completely generic processor. Code
6629 generated under this option will run on any POWER or PowerPC processor.
6630 GCC will use only the instructions in the common subset of both
6631 architectures, and will not use the MQ register. GCC assumes a generic
6632 processor model for scheduling purposes.
6634 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
6635 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
6636 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
6637 types, with an appropriate, generic processor model assumed for
6638 scheduling purposes.
6640 The other options specify a specific processor. Code generated under
6641 those options will run best on that processor, and may not run at all on
6644 The @option{-mcpu} options automatically enable or disable other
6645 @option{-m} options as follows:
6649 @option{-mno-power}, @option{-mno-powerc}
6656 @option{-mpower}, @option{-mno-powerpc}, @option{-mno-new-mnemonics}
6671 @option{-mno-power}, @option{-mpowerpc}, @option{-mnew-mnemonics}
6674 @option{-mpower}, @option{-mpowerpc}, @option{-mnew-mnemonics}
6679 @option{-mno-power}, @option{-mpowerpc}, @option{-mnew-mnemonics}, @option{-msoft-float}
6682 @item -mtune=@var{cpu_type}
6684 Set the instruction scheduling parameters for machine type
6685 @var{cpu_type}, but do not set the architecture type, register usage, or
6686 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
6687 values for @var{cpu_type} are used for @option{-mtune} as for
6688 @option{-mcpu}. If both are specified, the code generated will use the
6689 architecture, registers, and mnemonics set by @option{-mcpu}, but the
6690 scheduling parameters set by @option{-mtune}.
6695 @opindex mno-altivec
6696 These switches enable or disable the use of built-in functions that
6697 allow access to the AltiVec instruction set. You may also need to set
6698 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
6702 @itemx -mno-fp-in-toc
6703 @itemx -mno-sum-in-toc
6704 @itemx -mminimal-toc
6706 @opindex mno-fp-in-toc
6707 @opindex mno-sum-in-toc
6708 @opindex mminimal-toc
6709 Modify generation of the TOC (Table Of Contents), which is created for
6710 every executable file. The @option{-mfull-toc} option is selected by
6711 default. In that case, GCC will allocate at least one TOC entry for
6712 each unique non-automatic variable reference in your program. GCC
6713 will also place floating-point constants in the TOC@. However, only
6714 16,384 entries are available in the TOC@.
6716 If you receive a linker error message that saying you have overflowed
6717 the available TOC space, you can reduce the amount of TOC space used
6718 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
6719 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
6720 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
6721 generate code to calculate the sum of an address and a constant at
6722 run-time instead of putting that sum into the TOC@. You may specify one
6723 or both of these options. Each causes GCC to produce very slightly
6724 slower and larger code at the expense of conserving TOC space.
6726 If you still run out of space in the TOC even when you specify both of
6727 these options, specify @option{-mminimal-toc} instead. This option causes
6728 GCC to make only one TOC entry for every file. When you specify this
6729 option, GCC will produce code that is slower and larger but which
6730 uses extremely little TOC space. You may wish to use this option
6731 only on files that contain less frequently executed code.
6737 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
6738 @code{long} type, and the infrastructure needed to support them.
6739 Specifying @option{-maix64} implies @option{-mpowerpc64} and
6740 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
6741 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
6746 @opindex mno-xl-call
6747 On AIX, pass floating-point arguments to prototyped functions beyond the
6748 register save area (RSA) on the stack in addition to argument FPRs. The
6749 AIX calling convention was extended but not initially documented to
6750 handle an obscure K&R C case of calling a function that takes the
6751 address of its arguments with fewer arguments than declared. AIX XL
6752 compilers access floating point arguments which do not fit in the
6753 RSA from the stack when a subroutine is compiled without
6754 optimization. Because always storing floating-point arguments on the
6755 stack is inefficient and rarely needed, this option is not enabled by
6756 default and only is necessary when calling subroutines compiled by AIX
6757 XL compilers without optimization.
6761 Support @dfn{AIX Threads}. Link an application written to use
6762 @dfn{pthreads} with special libraries and startup code to enable the
6767 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
6768 application written to use message passing with special startup code to
6769 enable the application to run. The system must have PE installed in the
6770 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
6771 must be overridden with the @option{-specs=} option to specify the
6772 appropriate directory location. The Parallel Environment does not
6773 support threads, so the @option{-mpe} option and the @option{-mthreads}
6774 option are incompatible.
6778 @opindex msoft-float
6779 @opindex mhard-float
6780 Generate code that does not use (uses) the floating-point register set.
6781 Software floating point emulation is provided if you use the
6782 @option{-msoft-float} option, and pass the option to GCC when linking.
6785 @itemx -mno-multiple
6787 @opindex mno-multiple
6788 Generate code that uses (does not use) the load multiple word
6789 instructions and the store multiple word instructions. These
6790 instructions are generated by default on POWER systems, and not
6791 generated on PowerPC systems. Do not use @option{-mmultiple} on little
6792 endian PowerPC systems, since those instructions do not work when the
6793 processor is in little endian mode. The exceptions are PPC740 and
6794 PPC750 which permit the instructions usage in little endian mode.
6800 Generate code that uses (does not use) the load string instructions
6801 and the store string word instructions to save multiple registers and
6802 do small block moves. These instructions are generated by default on
6803 POWER systems, and not generated on PowerPC systems. Do not use
6804 @option{-mstring} on little endian PowerPC systems, since those
6805 instructions do not work when the processor is in little endian mode.
6806 The exceptions are PPC740 and PPC750 which permit the instructions
6807 usage in little endian mode.
6813 Generate code that uses (does not use) the load or store instructions
6814 that update the base register to the address of the calculated memory
6815 location. These instructions are generated by default. If you use
6816 @option{-mno-update}, there is a small window between the time that the
6817 stack pointer is updated and the address of the previous frame is
6818 stored, which means code that walks the stack frame across interrupts or
6819 signals may get corrupted data.
6822 @itemx -mno-fused-madd
6823 @opindex mfused-madd
6824 @opindex mno-fused-madd
6825 Generate code that uses (does not use) the floating point multiply and
6826 accumulate instructions. These instructions are generated by default if
6827 hardware floating is used.
6829 @item -mno-bit-align
6831 @opindex mno-bit-align
6833 On System V.4 and embedded PowerPC systems do not (do) force structures
6834 and unions that contain bit-fields to be aligned to the base type of the
6837 For example, by default a structure containing nothing but 8
6838 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
6839 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
6840 the structure would be aligned to a 1 byte boundary and be one byte in
6843 @item -mno-strict-align
6844 @itemx -mstrict-align
6845 @opindex mno-strict-align
6846 @opindex mstrict-align
6847 On System V.4 and embedded PowerPC systems do not (do) assume that
6848 unaligned memory references will be handled by the system.
6851 @itemx -mno-relocatable
6852 @opindex mrelocatable
6853 @opindex mno-relocatable
6854 On embedded PowerPC systems generate code that allows (does not allow)
6855 the program to be relocated to a different address at runtime. If you
6856 use @option{-mrelocatable} on any module, all objects linked together must
6857 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
6859 @item -mrelocatable-lib
6860 @itemx -mno-relocatable-lib
6861 @opindex mrelocatable-lib
6862 @opindex mno-relocatable-lib
6863 On embedded PowerPC systems generate code that allows (does not allow)
6864 the program to be relocated to a different address at runtime. Modules
6865 compiled with @option{-mrelocatable-lib} can be linked with either modules
6866 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
6867 with modules compiled with the @option{-mrelocatable} options.
6873 On System V.4 and embedded PowerPC systems do not (do) assume that
6874 register 2 contains a pointer to a global area pointing to the addresses
6875 used in the program.
6878 @itemx -mlittle-endian
6880 @opindex mlittle-endian
6881 On System V.4 and embedded PowerPC systems compile code for the
6882 processor in little endian mode. The @option{-mlittle-endian} option is
6883 the same as @option{-mlittle}.
6888 @opindex mbig-endian
6889 On System V.4 and embedded PowerPC systems compile code for the
6890 processor in big endian mode. The @option{-mbig-endian} option is
6891 the same as @option{-mbig}.
6895 On System V.4 and embedded PowerPC systems compile code using calling
6896 conventions that adheres to the March 1995 draft of the System V
6897 Application Binary Interface, PowerPC processor supplement. This is the
6898 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
6900 @item -mcall-sysv-eabi
6901 @opindex mcall-sysv-eabi
6902 Specify both @option{-mcall-sysv} and @option{-meabi} options.
6904 @item -mcall-sysv-noeabi
6905 @opindex mcall-sysv-noeabi
6906 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
6910 On System V.4 and embedded PowerPC systems compile code using calling
6911 conventions that are similar to those used on AIX@. This is the
6912 default if you configured GCC using @samp{powerpc-*-eabiaix}.
6914 @item -mcall-solaris
6915 @opindex mcall-solaris
6916 On System V.4 and embedded PowerPC systems compile code for the Solaris
6920 @opindex mcall-linux
6921 On System V.4 and embedded PowerPC systems compile code for the
6922 Linux-based GNU system.
6925 @opindex mcall-netbsd
6926 On System V.4 and embedded PowerPC systems compile code for the
6927 NetBSD operating system.
6930 @opindex mabi=altivec
6931 Extend the current ABI with AltiVec ABI extensions. This does not
6932 change the default ABI, instead it adds the AltiVec ABI extensions to
6936 @itemx -mno-prototype
6938 @opindex mno-prototype
6939 On System V.4 and embedded PowerPC systems assume that all calls to
6940 variable argument functions are properly prototyped. Otherwise, the
6941 compiler must insert an instruction before every non prototyped call to
6942 set or clear bit 6 of the condition code register (@var{CR}) to
6943 indicate whether floating point values were passed in the floating point
6944 registers in case the function takes a variable arguments. With
6945 @option{-mprototype}, only calls to prototyped variable argument functions
6946 will set or clear the bit.
6950 On embedded PowerPC systems, assume that the startup module is called
6951 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
6952 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
6957 On embedded PowerPC systems, assume that the startup module is called
6958 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
6963 On embedded PowerPC systems, assume that the startup module is called
6964 @file{crt0.o} and the standard C libraries are @file{libads.a} and
6968 @opindex myellowknife
6969 On embedded PowerPC systems, assume that the startup module is called
6970 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
6975 On System V.4 and embedded PowerPC systems, specify that you are
6976 compiling for a VxWorks system.
6980 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
6981 header to indicate that @samp{eabi} extended relocations are used.
6987 On System V.4 and embedded PowerPC systems do (do not) adhere to the
6988 Embedded Applications Binary Interface (eabi) which is a set of
6989 modifications to the System V.4 specifications. Selecting @option{-meabi}
6990 means that the stack is aligned to an 8 byte boundary, a function
6991 @code{__eabi} is called to from @code{main} to set up the eabi
6992 environment, and the @option{-msdata} option can use both @code{r2} and
6993 @code{r13} to point to two separate small data areas. Selecting
6994 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
6995 do not call an initialization function from @code{main}, and the
6996 @option{-msdata} option will only use @code{r13} to point to a single
6997 small data area. The @option{-meabi} option is on by default if you
6998 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
7001 @opindex msdata=eabi
7002 On System V.4 and embedded PowerPC systems, put small initialized
7003 @code{const} global and static data in the @samp{.sdata2} section, which
7004 is pointed to by register @code{r2}. Put small initialized
7005 non-@code{const} global and static data in the @samp{.sdata} section,
7006 which is pointed to by register @code{r13}. Put small uninitialized
7007 global and static data in the @samp{.sbss} section, which is adjacent to
7008 the @samp{.sdata} section. The @option{-msdata=eabi} option is
7009 incompatible with the @option{-mrelocatable} option. The
7010 @option{-msdata=eabi} option also sets the @option{-memb} option.
7013 @opindex msdata=sysv
7014 On System V.4 and embedded PowerPC systems, put small global and static
7015 data in the @samp{.sdata} section, which is pointed to by register
7016 @code{r13}. Put small uninitialized global and static data in the
7017 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
7018 The @option{-msdata=sysv} option is incompatible with the
7019 @option{-mrelocatable} option.
7021 @item -msdata=default
7023 @opindex msdata=default
7025 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
7026 compile code the same as @option{-msdata=eabi}, otherwise compile code the
7027 same as @option{-msdata=sysv}.
7030 @opindex msdata-data
7031 On System V.4 and embedded PowerPC systems, put small global and static
7032 data in the @samp{.sdata} section. Put small uninitialized global and
7033 static data in the @samp{.sbss} section. Do not use register @code{r13}
7034 to address small data however. This is the default behavior unless
7035 other @option{-msdata} options are used.
7039 @opindex msdata=none
7041 On embedded PowerPC systems, put all initialized global and static data
7042 in the @samp{.data} section, and all uninitialized data in the
7043 @samp{.bss} section.
7047 @cindex smaller data references (PowerPC)
7048 @cindex .sdata/.sdata2 references (PowerPC)
7049 On embedded PowerPC systems, put global and static items less than or
7050 equal to @var{num} bytes into the small data or bss sections instead of
7051 the normal data or bss section. By default, @var{num} is 8. The
7052 @option{-G @var{num}} switch is also passed to the linker.
7053 All modules should be compiled with the same @option{-G @var{num}} value.
7056 @itemx -mno-regnames
7058 @opindex mno-regnames
7059 On System V.4 and embedded PowerPC systems do (do not) emit register
7060 names in the assembly language output using symbolic forms.
7065 @subsection IBM RT Options
7067 @cindex IBM RT options
7069 These @samp{-m} options are defined for the IBM RT PC:
7073 @opindex min-line-mul
7074 Use an in-line code sequence for integer multiplies. This is the
7077 @item -mcall-lib-mul
7078 @opindex mcall-lib-mul
7079 Call @code{lmul$$} for integer multiples.
7081 @item -mfull-fp-blocks
7082 @opindex mfull-fp-blocks
7083 Generate full-size floating point data blocks, including the minimum
7084 amount of scratch space recommended by IBM@. This is the default.
7086 @item -mminimum-fp-blocks
7087 @opindex mminimum-fp-blocks
7088 Do not include extra scratch space in floating point data blocks. This
7089 results in smaller code, but slower execution, since scratch space must
7090 be allocated dynamically.
7092 @cindex @file{varargs.h} and RT PC
7093 @cindex @file{stdarg.h} and RT PC
7094 @item -mfp-arg-in-fpregs
7095 @opindex mfp-arg-in-fpregs
7096 Use a calling sequence incompatible with the IBM calling convention in
7097 which floating point arguments are passed in floating point registers.
7098 Note that @code{varargs.h} and @code{stdarg.h} will not work with
7099 floating point operands if this option is specified.
7101 @item -mfp-arg-in-gregs
7102 @opindex mfp-arg-in-gregs
7103 Use the normal calling convention for floating point arguments. This is
7106 @item -mhc-struct-return
7107 @opindex mhc-struct-return
7108 Return structures of more than one word in memory, rather than in a
7109 register. This provides compatibility with the MetaWare HighC (hc)
7110 compiler. Use the option @option{-fpcc-struct-return} for compatibility
7111 with the Portable C Compiler (pcc).
7113 @item -mnohc-struct-return
7114 @opindex mnohc-struct-return
7115 Return some structures of more than one word in registers, when
7116 convenient. This is the default. For compatibility with the
7117 IBM-supplied compilers, use the option @option{-fpcc-struct-return} or the
7118 option @option{-mhc-struct-return}.
7122 @subsection MIPS Options
7123 @cindex MIPS options
7125 These @samp{-m} options are defined for the MIPS family of computers:
7129 @item -march=@var{cpu-type}
7131 Assume the defaults for the machine type @var{cpu-type} when generating
7132 instructions. The choices for @var{cpu-type} are @samp{r2000}, @samp{r3000},
7133 @samp{r3900}, @samp{r4000}, @samp{r4100}, @samp{r4300}, @samp{r4400},
7134 @samp{r4600}, @samp{r4650}, @samp{r5000}, @samp{r6000}, @samp{r8000},
7135 and @samp{orion}. Additionally, the @samp{r2000}, @samp{r3000},
7136 @samp{r4000}, @samp{r5000}, and @samp{r6000} can be abbreviated as
7137 @samp{r2k} (or @samp{r2K}), @samp{r3k}, etc.
7139 @item -mtune=@var{cpu-type}
7141 Assume the defaults for the machine type @var{cpu-type} when scheduling
7142 instructions. The choices for @var{cpu-type} are @samp{r2000}, @samp{r3000},
7143 @samp{r3900}, @samp{r4000}, @samp{r4100}, @samp{r4300}, @samp{r4400},
7144 @samp{r4600}, @samp{r4650}, @samp{r5000}, @samp{r6000}, @samp{r8000},
7145 and @samp{orion}. Additionally, the @samp{r2000}, @samp{r3000},
7146 @samp{r4000}, @samp{r5000}, and @samp{r6000} can be abbreviated as
7147 @samp{r2k} (or @samp{r2K}), @samp{r3k}, etc. While picking a specific
7148 @var{cpu-type} will schedule things appropriately for that particular
7149 chip, the compiler will not generate any code that does not meet level 1
7150 of the MIPS ISA (instruction set architecture) without a @option{-mipsX}
7151 or @option{-mabi} switch being used.
7153 @item -mcpu=@var{cpu-type}
7155 This is identical to specifying both @option{-march} and @option{-mtune}.
7159 Issue instructions from level 1 of the MIPS ISA@. This is the default.
7160 @samp{r3000} is the default @var{cpu-type} at this ISA level.
7164 Issue instructions from level 2 of the MIPS ISA (branch likely, square
7165 root instructions). @samp{r6000} is the default @var{cpu-type} at this
7170 Issue instructions from level 3 of the MIPS ISA (64-bit instructions).
7171 @samp{r4000} is the default @var{cpu-type} at this ISA level.
7175 Issue instructions from level 4 of the MIPS ISA (conditional move,
7176 prefetch, enhanced FPU instructions). @samp{r8000} is the default
7177 @var{cpu-type} at this ISA level.
7181 Assume that 32 32-bit floating point registers are available. This is
7186 Assume that 32 64-bit floating point registers are available. This is
7187 the default when the @option{-mips3} option is used.
7190 @itemx -mno-fused-madd
7191 @opindex mfused-madd
7192 @opindex mno-fused-madd
7193 Generate code that uses (does not use) the floating point multiply and
7194 accumulate instructions, when they are available. These instructions
7195 are generated by default if they are available, but this may be
7196 undesirable if the extra precision causes problems or on certain chips
7197 in the mode where denormals are rounded to zero where denormals
7198 generated by multiply and accumulate instructions cause exceptions
7203 Assume that 32 32-bit general purpose registers are available. This is
7208 Assume that 32 64-bit general purpose registers are available. This is
7209 the default when the @option{-mips3} option is used.
7213 Force int and long types to be 64 bits wide. See @option{-mlong32} for an
7214 explanation of the default, and the width of pointers.
7218 Force long types to be 64 bits wide. See @option{-mlong32} for an
7219 explanation of the default, and the width of pointers.
7223 Force long, int, and pointer types to be 32 bits wide.
7225 If none of @option{-mlong32}, @option{-mlong64}, or @option{-mint64} are set,
7226 the size of ints, longs, and pointers depends on the ABI and ISA chosen.
7227 For @option{-mabi=32}, and @option{-mabi=n32}, ints and longs are 32 bits
7228 wide. For @option{-mabi=64}, ints are 32 bits, and longs are 64 bits wide.
7229 For @option{-mabi=eabi} and either @option{-mips1} or @option{-mips2}, ints
7230 and longs are 32 bits wide. For @option{-mabi=eabi} and higher ISAs, ints
7231 are 32 bits, and longs are 64 bits wide. The width of pointer types is
7232 the smaller of the width of longs or the width of general purpose
7233 registers (which in turn depends on the ISA)@.
7245 Generate code for the indicated ABI@. The default instruction level is
7246 @option{-mips1} for @samp{32}, @option{-mips3} for @samp{n32}, and
7247 @option{-mips4} otherwise. Conversely, with @option{-mips1} or
7248 @option{-mips2}, the default ABI is @samp{32}; otherwise, the default ABI
7253 Generate code for the MIPS assembler, and invoke @file{mips-tfile} to
7254 add normal debug information. This is the default for all
7255 platforms except for the OSF/1 reference platform, using the OSF/rose
7256 object format. If the either of the @option{-gstabs} or @option{-gstabs+}
7257 switches are used, the @file{mips-tfile} program will encapsulate the
7258 stabs within MIPS ECOFF@.
7262 Generate code for the GNU assembler. This is the default on the OSF/1
7263 reference platform, using the OSF/rose object format. Also, this is
7264 the default if the configure option @option{--with-gnu-as} is used.
7266 @item -msplit-addresses
7267 @itemx -mno-split-addresses
7268 @opindex msplit-addresses
7269 @opindex mno-split-addresses
7270 Generate code to load the high and low parts of address constants separately.
7271 This allows GCC to optimize away redundant loads of the high order
7272 bits of addresses. This optimization requires GNU as and GNU ld.
7273 This optimization is enabled by default for some embedded targets where
7274 GNU as and GNU ld are standard.
7280 The @option{-mrnames} switch says to output code using the MIPS software
7281 names for the registers, instead of the hardware names (ie, @var{a0}
7282 instead of @var{$4}). The only known assembler that supports this option
7283 is the Algorithmics assembler.
7289 The @option{-mgpopt} switch says to write all of the data declarations
7290 before the instructions in the text section, this allows the MIPS
7291 assembler to generate one word memory references instead of using two
7292 words for short global or static data items. This is on by default if
7293 optimization is selected.
7299 For each non-inline function processed, the @option{-mstats} switch
7300 causes the compiler to emit one line to the standard error file to
7301 print statistics about the program (number of registers saved, stack
7308 The @option{-mmemcpy} switch makes all block moves call the appropriate
7309 string function (@samp{memcpy} or @samp{bcopy}) instead of possibly
7310 generating inline code.
7313 @itemx -mno-mips-tfile
7314 @opindex mmips-tfile
7315 @opindex mno-mips-tfile
7316 The @option{-mno-mips-tfile} switch causes the compiler not
7317 postprocess the object file with the @file{mips-tfile} program,
7318 after the MIPS assembler has generated it to add debug support. If
7319 @file{mips-tfile} is not run, then no local variables will be
7320 available to the debugger. In addition, @file{stage2} and
7321 @file{stage3} objects will have the temporary file names passed to the
7322 assembler embedded in the object file, which means the objects will
7323 not compare the same. The @option{-mno-mips-tfile} switch should only
7324 be used when there are bugs in the @file{mips-tfile} program that
7325 prevents compilation.
7328 @opindex msoft-float
7329 Generate output containing library calls for floating point.
7330 @strong{Warning:} the requisite libraries are not part of GCC@.
7331 Normally the facilities of the machine's usual C compiler are used, but
7332 this can't be done directly in cross-compilation. You must make your
7333 own arrangements to provide suitable library functions for
7337 @opindex mhard-float
7338 Generate output containing floating point instructions. This is the
7339 default if you use the unmodified sources.
7342 @itemx -mno-abicalls
7344 @opindex mno-abicalls
7345 Emit (or do not emit) the pseudo operations @samp{.abicalls},
7346 @samp{.cpload}, and @samp{.cprestore} that some System V.4 ports use for
7347 position independent code.
7350 @itemx -mno-long-calls
7351 @opindex mlong-calls
7352 @opindex mno-long-calls
7353 Do all calls with the @samp{JALR} instruction, which requires
7354 loading up a function's address into a register before the call.
7355 You need to use this switch, if you call outside of the current
7356 512 megabyte segment to functions that are not through pointers.
7359 @itemx -mno-half-pic
7361 @opindex mno-half-pic
7362 Put pointers to extern references into the data section and load them
7363 up, rather than put the references in the text section.
7365 @item -membedded-pic
7366 @itemx -mno-embedded-pic
7367 @opindex membedded-pic
7368 @opindex mno-embedded-pic
7369 Generate PIC code suitable for some embedded systems. All calls are
7370 made using PC relative address, and all data is addressed using the $gp
7371 register. No more than 65536 bytes of global data may be used. This
7372 requires GNU as and GNU ld which do most of the work. This currently
7373 only works on targets which use ECOFF; it does not work with ELF@.
7375 @item -membedded-data
7376 @itemx -mno-embedded-data
7377 @opindex membedded-data
7378 @opindex mno-embedded-data
7379 Allocate variables to the read-only data section first if possible, then
7380 next in the small data section if possible, otherwise in data. This gives
7381 slightly slower code than the default, but reduces the amount of RAM required
7382 when executing, and thus may be preferred for some embedded systems.
7384 @item -muninit-const-in-rodata
7385 @itemx -mno-uninit-const-in-rodata
7386 @opindex muninit-const-in-rodata
7387 @opindex mno-uninit-const-in-rodata
7388 When used together with @option{-membedded-data}, it will always store uninitialized
7389 const variables in the read-only data section.
7391 @item -msingle-float
7392 @itemx -mdouble-float
7393 @opindex msingle-float
7394 @opindex mdouble-float
7395 The @option{-msingle-float} switch tells gcc to assume that the floating
7396 point coprocessor only supports single precision operations, as on the
7397 @samp{r4650} chip. The @option{-mdouble-float} switch permits gcc to use
7398 double precision operations. This is the default.
7404 Permit use of the @samp{mad}, @samp{madu} and @samp{mul} instructions,
7405 as on the @samp{r4650} chip.
7409 Turns on @option{-msingle-float}, @option{-mmad}, and, at least for now,
7410 @option{-mcpu=r4650}.
7416 Enable 16-bit instructions.
7420 Use the entry and exit pseudo ops. This option can only be used with
7425 Compile code for the processor in little endian mode.
7426 The requisite libraries are assumed to exist.
7430 Compile code for the processor in big endian mode.
7431 The requisite libraries are assumed to exist.
7435 @cindex smaller data references (MIPS)
7436 @cindex gp-relative references (MIPS)
7437 Put global and static items less than or equal to @var{num} bytes into
7438 the small data or bss sections instead of the normal data or bss
7439 section. This allows the assembler to emit one word memory reference
7440 instructions based on the global pointer (@var{gp} or @var{$28}),
7441 instead of the normal two words used. By default, @var{num} is 8 when
7442 the MIPS assembler is used, and 0 when the GNU assembler is used. The
7443 @option{-G @var{num}} switch is also passed to the assembler and linker.
7444 All modules should be compiled with the same @option{-G @var{num}}
7449 Tell the MIPS assembler to not run its preprocessor over user
7450 assembler files (with a @samp{.s} suffix) when assembling them.
7454 Pass an option to gas which will cause nops to be inserted if
7455 the read of the destination register of an mfhi or mflo instruction
7456 occurs in the following two instructions.
7460 Do not include the default crt0.
7462 @item -mflush-func=@var{func}
7463 @itemx -mno-flush-func
7464 @opindex mflush-func
7465 Specifies the function to call to flush the I and D caches, or to not
7466 call any such function. If called, the function must take the same
7467 arguments as the common @code{_flush_func()}, that is, the address of the
7468 memory range for which the cache is being flushed, the size of the
7469 memory range, and the number 3 (to flush both caches). The default
7470 depends on the target gcc was configured for, but commonly is either
7471 @samp{_flush_func} or @samp{__cpu_flush}.
7475 These options are defined by the macro
7476 @code{TARGET_SWITCHES} in the machine description. The default for the
7477 options is also defined by that macro, which enables you to change the
7481 @node i386 and x86-64 Options
7482 @subsection Intel 386 and AMD x86-64 Options
7483 @cindex i386 Options
7484 @cindex x86-64 Options
7485 @cindex Intel 386 Options
7486 @cindex AMD x86-64 Options
7488 These @samp{-m} options are defined for the i386 and x86-64 family of
7492 @item -mcpu=@var{cpu-type}
7494 Assume the defaults for the machine type @var{cpu-type} when scheduling
7495 instructions. The choices for @var{cpu-type} are @samp{i386},
7496 @samp{i486}, @samp{i586}, @samp{i686}, @samp{pentium},
7497 @samp{pentiumpro}, @samp{pentium4}, @samp{k6}, and @samp{athlon}
7499 While picking a specific @var{cpu-type} will schedule things appropriately
7500 for that particular chip, the compiler will not generate any code that
7501 does not run on the i386 without the @option{-march=@var{cpu-type}} option
7502 being used. @samp{i586} is equivalent to @samp{pentium} and @samp{i686}
7503 is equivalent to @samp{pentiumpro}. @samp{k6} and @samp{athlon} are the
7504 AMD chips as opposed to the Intel ones.
7506 @item -march=@var{cpu-type}
7508 Generate instructions for the machine type @var{cpu-type}. The choices
7509 for @var{cpu-type} are the same as for @option{-mcpu}. Moreover,
7510 specifying @option{-march=@var{cpu-type}} implies @option{-mcpu=@var{cpu-type}}.
7519 @opindex mpentiumpro
7520 These options are synonyms for @option{-mcpu=i386}, @option{-mcpu=i486},
7521 @option{-mcpu=pentium}, and @option{-mcpu=pentiumpro} respectively.
7522 These synonyms are deprecated.
7524 @item -mintel-syntax
7525 @opindex mintel-syntax
7526 Emit assembly using Intel syntax opcodes instead of AT&T syntax.
7531 @opindex mno-ieee-fp
7532 Control whether or not the compiler uses IEEE floating point
7533 comparisons. These handle correctly the case where the result of a
7534 comparison is unordered.
7537 @opindex msoft-float
7538 Generate output containing library calls for floating point.
7539 @strong{Warning:} the requisite libraries are not part of GCC@.
7540 Normally the facilities of the machine's usual C compiler are used, but
7541 this can't be done directly in cross-compilation. You must make your
7542 own arrangements to provide suitable library functions for
7545 On machines where a function returns floating point results in the 80387
7546 register stack, some floating point opcodes may be emitted even if
7547 @option{-msoft-float} is used.
7549 @item -mno-fp-ret-in-387
7550 @opindex mno-fp-ret-in-387
7551 Do not use the FPU registers for return values of functions.
7553 The usual calling convention has functions return values of types
7554 @code{float} and @code{double} in an FPU register, even if there
7555 is no FPU@. The idea is that the operating system should emulate
7558 The option @option{-mno-fp-ret-in-387} causes such values to be returned
7559 in ordinary CPU registers instead.
7561 @item -mno-fancy-math-387
7562 @opindex mno-fancy-math-387
7563 Some 387 emulators do not support the @code{sin}, @code{cos} and
7564 @code{sqrt} instructions for the 387. Specify this option to avoid
7565 generating those instructions. This option is the default on FreeBSD@.
7566 As of revision 2.6.1, these instructions are not generated unless you
7567 also use the @option{-funsafe-math-optimizations} switch.
7569 @item -malign-double
7570 @itemx -mno-align-double
7571 @opindex malign-double
7572 @opindex mno-align-double
7573 Control whether GCC aligns @code{double}, @code{long double}, and
7574 @code{long long} variables on a two word boundary or a one word
7575 boundary. Aligning @code{double} variables on a two word boundary will
7576 produce code that runs somewhat faster on a @samp{Pentium} at the
7577 expense of more memory.
7579 @item -m128bit-long-double
7580 @opindex m128bit-long-double
7581 Control the size of @code{long double} type. i386 application binary interface
7582 specify the size to be 12 bytes, while modern architectures (Pentium and newer)
7583 prefer @code{long double} aligned to 8 or 16 byte boundary. This is
7584 impossible to reach with 12 byte long doubles in the array accesses.
7586 @strong{Warning:} if you use the @option{-m128bit-long-double} switch, the
7587 structures and arrays containing @code{long double} will change their size as
7588 well as function calling convention for function taking @code{long double}
7591 @item -m96bit-long-double
7592 @opindex m96bit-long-double
7593 Set the size of @code{long double} to 96 bits as required by the i386
7594 application binary interface. This is the default.
7597 @itemx -mno-svr3-shlib
7598 @opindex msvr3-shlib
7599 @opindex mno-svr3-shlib
7600 Control whether GCC places uninitialized local variables into the
7601 @code{bss} or @code{data} segments. @option{-msvr3-shlib} places them
7602 into @code{bss}. These options are meaningful only on System V Release 3.
7606 Use a different function-calling convention, in which functions that
7607 take a fixed number of arguments return with the @code{ret} @var{num}
7608 instruction, which pops their arguments while returning. This saves one
7609 instruction in the caller since there is no need to pop the arguments
7612 You can specify that an individual function is called with this calling
7613 sequence with the function attribute @samp{stdcall}. You can also
7614 override the @option{-mrtd} option by using the function attribute
7615 @samp{cdecl}. @xref{Function Attributes}.
7617 @strong{Warning:} this calling convention is incompatible with the one
7618 normally used on Unix, so you cannot use it if you need to call
7619 libraries compiled with the Unix compiler.
7621 Also, you must provide function prototypes for all functions that
7622 take variable numbers of arguments (including @code{printf});
7623 otherwise incorrect code will be generated for calls to those
7626 In addition, seriously incorrect code will result if you call a
7627 function with too many arguments. (Normally, extra arguments are
7628 harmlessly ignored.)
7630 @item -mregparm=@var{num}
7632 Control how many registers are used to pass integer arguments. By
7633 default, no registers are used to pass arguments, and at most 3
7634 registers can be used. You can control this behavior for a specific
7635 function by using the function attribute @samp{regparm}.
7636 @xref{Function Attributes}.
7638 @strong{Warning:} if you use this switch, and
7639 @var{num} is nonzero, then you must build all modules with the same
7640 value, including any libraries. This includes the system libraries and
7643 @item -mpreferred-stack-boundary=@var{num}
7644 @opindex mpreferred-stack-boundary
7645 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
7646 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
7647 the default is 4 (16 bytes or 128 bits), except when optimizing for code
7648 size (@option{-Os}), in which case the default is the minimum correct
7649 alignment (4 bytes for x86, and 8 bytes for x86-64).
7651 On Pentium and PentiumPro, @code{double} and @code{long double} values
7652 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
7653 suffer significant run time performance penalties. On Pentium III, the
7654 Streaming SIMD Extension (SSE) data type @code{__m128} suffers similar
7655 penalties if it is not 16 byte aligned.
7657 To ensure proper alignment of this values on the stack, the stack boundary
7658 must be as aligned as that required by any value stored on the stack.
7659 Further, every function must be generated such that it keeps the stack
7660 aligned. Thus calling a function compiled with a higher preferred
7661 stack boundary from a function compiled with a lower preferred stack
7662 boundary will most likely misalign the stack. It is recommended that
7663 libraries that use callbacks always use the default setting.
7665 This extra alignment does consume extra stack space, and generally
7666 increases code size. Code that is sensitive to stack space usage, such
7667 as embedded systems and operating system kernels, may want to reduce the
7668 preferred alignment to @option{-mpreferred-stack-boundary=2}.
7682 These switches enable or disable the use of built-in functions that allow
7683 direct access to the MMX, SSE and 3Dnow extensions of the instruction set.
7685 The following machine modes are available for use with MMX built-in functions
7686 (@pxref{Vector Extensions}): @code{V2SI} for a vector of two 32 bit integers,
7687 @code{V4HI} for a vector of four 16 bit integers, and @code{V8QI} for a
7688 vector of eight 8 bit integers. Some of the built-in functions operate on
7689 MMX registers as a whole 64 bit entity, these use @code{DI} as their mode.
7691 If 3Dnow extensions are enabled, @code{V2SF} is used as a mode for a vector
7692 of two 32 bit floating point values.
7694 If SSE extensions are enabled, @code{V4SF} is used for a vector of four 32 bit
7695 floating point values. Some instructions use a vector of four 32 bit
7696 integers, these use @code{V4SI}. Finally, some instructions operate on an
7697 entire vector register, interpreting it as a 128 bit integer, these use mode
7700 The following built-in functions are made available by @option{-mmmx}:
7702 @item v8qi __builtin_ia32_paddb (v8qi, v8qi)
7703 Generates the @code{paddb} machine instruction.
7704 @item v4hi __builtin_ia32_paddw (v4hi, v4hi)
7705 Generates the @code{paddw} machine instruction.
7706 @item v2si __builtin_ia32_paddd (v2si, v2si)
7707 Generates the @code{paddd} machine instruction.
7708 @item v8qi __builtin_ia32_psubb (v8qi, v8qi)
7709 Generates the @code{psubb} machine instruction.
7710 @item v4hi __builtin_ia32_psubw (v4hi, v4hi)
7711 Generates the @code{psubw} machine instruction.
7712 @item v2si __builtin_ia32_psubd (v2si, v2si)
7713 Generates the @code{psubd} machine instruction.
7715 @item v8qi __builtin_ia32_paddsb (v8qi, v8qi)
7716 Generates the @code{paddsb} machine instruction.
7717 @item v4hi __builtin_ia32_paddsw (v4hi, v4hi)
7718 Generates the @code{paddsw} machine instruction.
7719 @item v8qi __builtin_ia32_psubsb (v8qi, v8qi)
7720 Generates the @code{psubsb} machine instruction.
7721 @item v4hi __builtin_ia32_psubsw (v4hi, v4hi)
7722 Generates the @code{psubsw} machine instruction.
7724 @item v8qi __builtin_ia32_paddusb (v8qi, v8qi)
7725 Generates the @code{paddusb} machine instruction.
7726 @item v4hi __builtin_ia32_paddusw (v4hi, v4hi)
7727 Generates the @code{paddusw} machine instruction.
7728 @item v8qi __builtin_ia32_psubusb (v8qi, v8qi)
7729 Generates the @code{psubusb} machine instruction.
7730 @item v4hi __builtin_ia32_psubusw (v4hi, v4hi)
7731 Generates the @code{psubusw} machine instruction.
7733 @item v4hi __builtin_ia32_pmullw (v4hi, v4hi)
7734 Generates the @code{pmullw} machine instruction.
7735 @item v4hi __builtin_ia32_pmulhw (v4hi, v4hi)
7736 Generates the @code{pmulhw} machine instruction.
7738 @item di __builtin_ia32_pand (di, di)
7739 Generates the @code{pand} machine instruction.
7740 @item di __builtin_ia32_pandn (di,di)
7741 Generates the @code{pandn} machine instruction.
7742 @item di __builtin_ia32_por (di, di)
7743 Generates the @code{por} machine instruction.
7744 @item di __builtin_ia32_pxor (di, di)
7745 Generates the @code{pxor} machine instruction.
7747 @item v8qi __builtin_ia32_pcmpeqb (v8qi, v8qi)
7748 Generates the @code{pcmpeqb} machine instruction.
7749 @item v4hi __builtin_ia32_pcmpeqw (v4hi, v4hi)
7750 Generates the @code{pcmpeqw} machine instruction.
7751 @item v2si __builtin_ia32_pcmpeqd (v2si, v2si)
7752 Generates the @code{pcmpeqd} machine instruction.
7753 @item v8qi __builtin_ia32_pcmpgtb (v8qi, v8qi)
7754 Generates the @code{pcmpgtb} machine instruction.
7755 @item v4hi __builtin_ia32_pcmpgtw (v4hi, v4hi)
7756 Generates the @code{pcmpgtw} machine instruction.
7757 @item v2si __builtin_ia32_pcmpgtd (v2si, v2si)
7758 Generates the @code{pcmpgtd} machine instruction.
7760 @item v8qi __builtin_ia32_punpckhbw (v8qi, v8qi)
7761 Generates the @code{punpckhbw} machine instruction.
7762 @item v4hi __builtin_ia32_punpckhwd (v4hi, v4hi)
7763 Generates the @code{punpckhwd} machine instruction.
7764 @item v2si __builtin_ia32_punpckhdq (v2si, v2si)
7765 Generates the @code{punpckhdq} machine instruction.
7766 @item v8qi __builtin_ia32_punpcklbw (v8qi, v8qi)
7767 Generates the @code{punpcklbw} machine instruction.
7768 @item v4hi __builtin_ia32_punpcklwd (v4hi, v4hi)
7769 Generates the @code{punpcklwd} machine instruction.
7770 @item v2si __builtin_ia32_punpckldq (v2si, v2si)
7771 Generates the @code{punpckldq} machine instruction.
7773 @item v8qi __builtin_ia32_packsswb (v4hi, v4hi)
7774 Generates the @code{packsswb} machine instruction.
7775 @item v4hi __builtin_ia32_packssdw (v2si, v2si)
7776 Generates the @code{packssdw} machine instruction.
7777 @item v8qi __builtin_ia32_packuswb (v4hi, v4hi)
7778 Generates the @code{packuswb} machine instruction.
7782 The following built-in functions are made available either with @option{-msse}, or
7783 with a combination of @option{-m3dnow} and @option{-march=athlon}.
7786 @item v4hi __builtin_ia32_pmulhuw (v4hi, v4hi)
7787 Generates the @code{pmulhuw} machine instruction.
7789 @item v8qi __builtin_ia32_pavgb (v8qi, v8qi)
7790 Generates the @code{pavgb} machine instruction.
7791 @item v4hi __builtin_ia32_pavgw (v4hi, v4hi)
7792 Generates the @code{pavgw} machine instruction.
7793 @item v4hi __builtin_ia32_psadbw (v8qi, v8qi)
7794 Generates the @code{psadbw} machine instruction.
7796 @item v8qi __builtin_ia32_pmaxub (v8qi, v8qi)
7797 Generates the @code{pmaxub} machine instruction.
7798 @item v4hi __builtin_ia32_pmaxsw (v4hi, v4hi)
7799 Generates the @code{pmaxsw} machine instruction.
7800 @item v8qi __builtin_ia32_pminub (v8qi, v8qi)
7801 Generates the @code{pminub} machine instruction.
7802 @item v4hi __builtin_ia32_pminsw (v4hi, v4hi)
7803 Generates the @code{pminsw} machine instruction.
7805 @item int __builtin_ia32_pextrw (v4hi, int)
7806 Generates the @code{pextrw} machine instruction.
7807 @item v4hi __builtin_ia32_pinsrw (v4hi, int, int)
7808 Generates the @code{pinsrw} machine instruction.
7810 @item int __builtin_ia32_pmovmskb (v8qi)
7811 Generates the @code{pmovmskb} machine instruction.
7812 @item void __builtin_ia32_maskmovq (v8qi, v8qi, char *)
7813 Generates the @code{maskmovq} machine instruction.
7814 @item void __builtin_ia32_movntq (di *, di)
7815 Generates the @code{movntq} machine instruction.
7816 @item void __builtin_ia32_sfence (void)
7817 Generates the @code{sfence} machine instruction.
7818 @item void __builtin_ia32_prefetch (char *, int selector)
7819 Generates a prefetch machine instruction, depending on the value of
7820 selector. If @code{selector} is 0, it generates @code{prefetchnta}; for
7821 a value of 1, it generates @code{prefetcht0}; for a value of 2, it generates
7822 @code{prefetcht1}; and for a value of 3 it generates @code{prefetcht2}.
7826 The following built-in functions are available when @option{-msse} is used.
7829 @item int __builtin_ia32_comieq (v4sf, v4sf)
7830 Generates the @code{comiss} machine instruction and performs an equality
7831 comparison. The return value is the truth value of that comparison.
7832 @item int __builtin_ia32_comineq (v4sf, v4sf)
7833 Generates the @code{comiss} machine instruction and performs an inequality
7834 comparison. The return value is the truth value of that comparison.
7835 @item int __builtin_ia32_comilt (v4sf, v4sf)
7836 Generates the @code{comiss} machine instruction and performs a ``less than''
7837 comparison. The return value is the truth value of that comparison.
7838 @item int __builtin_ia32_comile (v4sf, v4sf)
7839 Generates the @code{comiss} machine instruction and performs a ``less or
7840 equal'' comparison. The return value is the truth value of that comparison.
7841 @item int __builtin_ia32_comigt (v4sf, v4sf)
7842 Generates the @code{comiss} machine instruction and performs a ``greater than''
7843 comparison. The return value is the truth value of that comparison.
7844 @item int __builtin_ia32_comige (v4sf, v4sf)
7845 Generates the @code{comiss} machine instruction and performs a ``greater or
7846 equal'' comparison. The return value is the truth value of that comparison.
7848 @item int __builtin_ia32_ucomieq (v4sf, v4sf)
7849 Generates the @code{ucomiss} machine instruction and performs an equality
7850 comparison. The return value is the truth value of that comparison.
7851 @item int __builtin_ia32_ucomineq (v4sf, v4sf)
7852 Generates the @code{ucomiss} machine instruction and performs an inequality
7853 comparison. The return value is the truth value of that comparison.
7854 @item int __builtin_ia32_ucomilt (v4sf, v4sf)
7855 Generates the @code{ucomiss} machine instruction and performs a ``less than''
7856 comparison. The return value is the truth value of that comparison.
7857 @item int __builtin_ia32_ucomile (v4sf, v4sf)
7858 Generates the @code{ucomiss} machine instruction and performs a ``less or
7859 equal'' comparison. The return value is the truth value of that comparison.
7860 @item int __builtin_ia32_ucomigt (v4sf, v4sf)
7861 Generates the @code{ucomiss} machine instruction and performs a ``greater than''
7862 comparison. The return value is the truth value of that comparison.
7863 @item int __builtin_ia32_ucomige (v4sf, v4sf)
7864 Generates the @code{ucomiss} machine instruction and performs a ``greater or
7865 equal'' comparison. The return value is the truth value of that comparison.
7867 @item v4sf __builtin_ia32_addps (v4sf, v4sf)
7868 Generates the @code{addps} machine instruction.
7869 @item v4sf __builtin_ia32_addss (v4sf, v4sf)
7870 Generates the @code{addss} machine instruction.
7871 @item v4sf __builtin_ia32_subps (v4sf, v4sf)
7872 Generates the @code{subps} machine instruction.
7873 @item v4sf __builtin_ia32_subss (v4sf, v4sf)
7874 Generates the @code{subss} machine instruction.
7875 @item v4sf __builtin_ia32_mulps (v4sf, v4sf)
7876 Generates the @code{mulps} machine instruction.
7877 @item v4sf __builtin_ia32_mulss (v4sf, v4sf)
7878 Generates the @code{mulss} machine instruction.
7879 @item v4sf __builtin_ia32_divps (v4sf, v4sf)
7880 Generates the @code{divps} machine instruction.
7881 @item v4sf __builtin_ia32_divss (v4sf, v4sf)
7882 Generates the @code{divss} machine instruction.
7884 @item v4si __builtin_ia32_cmpeqps (v4sf, v4sf)
7885 Generates the @code{cmpeqps} machine instruction.
7886 @item v4si __builtin_ia32_cmplts (v4sf, v4sf)
7887 Generates the @code{cmpltps} machine instruction.
7888 @item v4si __builtin_ia32_cmpleps (v4sf, v4sf)
7889 Generates the @code{cmpleps} machine instruction.
7890 @item v4si __builtin_ia32_cmpgtps (v4sf, v4sf)
7891 Generates the @code{cmpgtps} machine instruction.
7892 @item v4si __builtin_ia32_cmpgeps (v4sf, v4sf)
7893 Generates the @code{cmpgeps} machine instruction.
7894 @item v4si __builtin_ia32_cmpunordps (v4sf, v4sf)
7895 Generates the @code{cmpunodps} machine instruction.
7896 @item v4si __builtin_ia32_cmpneqps (v4sf, v4sf)
7897 Generates the @code{cmpeqps} machine instruction.
7898 @item v4si __builtin_ia32_cmpnltps (v4sf, v4sf)
7899 Generates the @code{cmpltps} machine instruction.
7900 @item v4si __builtin_ia32_cmpnleps (v4sf, v4sf)
7901 Generates the @code{cmpleps} machine instruction.
7902 @item v4si __builtin_ia32_cmpngtps (v4sf, v4sf)
7903 Generates the @code{cmpgtps} machine instruction.
7904 @item v4si __builtin_ia32_cmpngeps (v4sf, v4sf)
7905 Generates the @code{cmpgeps} machine instruction.
7906 @item v4si __builtin_ia32_cmpordps (v4sf, v4sf)
7907 Generates the @code{cmpunodps} machine instruction.
7909 @item v4si __builtin_ia32_cmpeqss (v4sf, v4sf)
7910 Generates the @code{cmpeqss} machine instruction.
7911 @item v4si __builtin_ia32_cmpltss (v4sf, v4sf)
7912 Generates the @code{cmpltss} machine instruction.
7913 @item v4si __builtin_ia32_cmpless (v4sf, v4sf)
7914 Generates the @code{cmpless} machine instruction.
7915 @item v4si __builtin_ia32_cmpgtss (v4sf, v4sf)
7916 Generates the @code{cmpgtss} machine instruction.
7917 @item v4si __builtin_ia32_cmpgess (v4sf, v4sf)
7918 Generates the @code{cmpgess} machine instruction.
7919 @item v4si __builtin_ia32_cmpunordss (v4sf, v4sf)
7920 Generates the @code{cmpunodss} machine instruction.
7921 @item v4si __builtin_ia32_cmpneqss (v4sf, v4sf)
7922 Generates the @code{cmpeqss} machine instruction.
7923 @item v4si __builtin_ia32_cmpnlts (v4sf, v4sf)
7924 Generates the @code{cmpltss} machine instruction.
7925 @item v4si __builtin_ia32_cmpnless (v4sf, v4sf)
7926 Generates the @code{cmpless} machine instruction.
7927 @item v4si __builtin_ia32_cmpngtss (v4sf, v4sf)
7928 Generates the @code{cmpgtss} machine instruction.
7929 @item v4si __builtin_ia32_cmpngess (v4sf, v4sf)
7930 Generates the @code{cmpgess} machine instruction.
7931 @item v4si __builtin_ia32_cmpordss (v4sf, v4sf)
7932 Generates the @code{cmpunodss} machine instruction.
7934 @item v4sf __builtin_ia32_maxps (v4sf, v4sf)
7935 Generates the @code{maxps} machine instruction.
7936 @item v4sf __builtin_ia32_maxsss (v4sf, v4sf)
7937 Generates the @code{maxss} machine instruction.
7938 @item v4sf __builtin_ia32_minps (v4sf, v4sf)
7939 Generates the @code{minps} machine instruction.
7940 @item v4sf __builtin_ia32_minsss (v4sf, v4sf)
7941 Generates the @code{minss} machine instruction.
7943 @item ti __builtin_ia32_andps (ti, ti)
7944 Generates the @code{andps} machine instruction.
7945 @item ti __builtin_ia32_andnps (ti, ti)
7946 Generates the @code{andnps} machine instruction.
7947 @item ti __builtin_ia32_orps (ti, ti)
7948 Generates the @code{orps} machine instruction.
7949 @item ti __builtin_ia32_xorps (ti, ti)
7950 Generates the @code{xorps} machine instruction.
7952 @item v4sf __builtin_ia32_movps (v4sf, v4sf)
7953 Generates the @code{movps} machine instruction.
7954 @item v4sf __builtin_ia32_movhlps (v4sf, v4sf)
7955 Generates the @code{movhlps} machine instruction.
7956 @item v4sf __builtin_ia32_movlhps (v4sf, v4sf)
7957 Generates the @code{movlhps} machine instruction.
7958 @item v4sf __builtin_ia32_unpckhps (v4sf, v4sf)
7959 Generates the @code{unpckhps} machine instruction.
7960 @item v4sf __builtin_ia32_unpcklps (v4sf, v4sf)
7961 Generates the @code{unpcklps} machine instruction.
7963 @item v4sf __builtin_ia32_cvtpi2ps (v4sf, v2si)
7964 Generates the @code{cvtpi2ps} machine instruction.
7965 @item v2si __builtin_ia32_cvtps2pi (v4sf)
7966 Generates the @code{cvtps2pi} machine instruction.
7967 @item v4sf __builtin_ia32_cvtsi2ss (v4sf, int)
7968 Generates the @code{cvtsi2ss} machine instruction.
7969 @item int __builtin_ia32_cvtss2si (v4sf)
7970 Generates the @code{cvtsi2ss} machine instruction.
7971 @item v2si __builtin_ia32_cvttps2pi (v4sf)
7972 Generates the @code{cvttps2pi} machine instruction.
7973 @item int __builtin_ia32_cvttss2si (v4sf)
7974 Generates the @code{cvttsi2ss} machine instruction.
7976 @item v4sf __builtin_ia32_rcpps (v4sf)
7977 Generates the @code{rcpps} machine instruction.
7978 @item v4sf __builtin_ia32_rsqrtps (v4sf)
7979 Generates the @code{rsqrtps} machine instruction.
7980 @item v4sf __builtin_ia32_sqrtps (v4sf)
7981 Generates the @code{sqrtps} machine instruction.
7982 @item v4sf __builtin_ia32_rcpss (v4sf)
7983 Generates the @code{rcpss} machine instruction.
7984 @item v4sf __builtin_ia32_rsqrtss (v4sf)
7985 Generates the @code{rsqrtss} machine instruction.
7986 @item v4sf __builtin_ia32_sqrtss (v4sf)
7987 Generates the @code{sqrtss} machine instruction.
7989 @item v4sf __builtin_ia32_shufps (v4sf, v4sf, int)
7990 Generates the @code{shufps} machine instruction.
7992 @item v4sf __builtin_ia32_loadaps (float *)
7993 Generates the @code{movaps} machine instruction as a load from memory.
7994 @item void __builtin_ia32_storeaps (float *, v4sf)
7995 Generates the @code{movaps} machine instruction as a store to memory.
7996 @item v4sf __builtin_ia32_loadups (float *)
7997 Generates the @code{movups} machine instruction as a load from memory.
7998 @item void __builtin_ia32_storeups (float *, v4sf)
7999 Generates the @code{movups} machine instruction as a store to memory.
8000 @item v4sf __builtin_ia32_loadsss (float *)
8001 Generates the @code{movss} machine instruction as a load from memory.
8002 @item void __builtin_ia32_storess (float *, v4sf)
8003 Generates the @code{movss} machine instruction as a store to memory.
8005 @item v4sf __builtin_ia32_loadhps (v4sf, v2si *)
8006 Generates the @code{movhps} machine instruction as a load from memory.
8007 @item v4sf __builtin_ia32_loadlps (v4sf, v2si *)
8008 Generates the @code{movlps} machine instruction as a load from memory
8009 @item void __builtin_ia32_storehps (v4sf, v2si *)
8010 Generates the @code{movhps} machine instruction as a store to memory.
8011 @item void __builtin_ia32_storelps (v4sf, v2si *)
8012 Generates the @code{movlps} machine instruction as a store to memory.
8014 @item void __builtin_ia32_movntps (float *, v4sf)
8015 Generates the @code{movntps} machine instruction.
8016 @item int __builtin_ia32_movmskps (v4sf)
8017 Generates the @code{movntps} machine instruction.
8019 @item void __builtin_ia32_storeps1 (float *, v4sf)
8020 Generates the @code{movaps} machine instruction as a store to memory.
8021 Before storing, the value is modified with a @code{shufps} instruction
8022 so that the lowest of the four floating point elements is replicated
8023 across the entire vector that is stored.
8024 @item void __builtin_ia32_storerps (float *, v4sf)
8025 Generates the @code{movaps} machine instruction as a store to memory.
8026 Before storing, the value is modified with a @code{shufps} instruction
8027 so that the order of the four floating point elements in the vector is
8029 @item v4sf __builtin_ia32_loadps1 (float *)
8030 Generates a @code{movss} machine instruction to load a floating point
8031 value from memory, and a @code{shufps} instruction to replicate the
8032 loaded value across all four elements of the result vector.
8033 @item v4sf __builtin_ia32_loadrps (float *)
8034 Generates a @code{movaps} machine instruction to load a vector from
8035 memory, and a @code{shufps} instruction to reverse the order of the
8036 four floating point elements in the result vector.
8037 @item v4sf __builtin_ia32_setps (float, float, float, float)
8038 Constructs a vector from four single floating point values. The return
8039 value is equal to the value that would result from storing the four
8040 arguments into consecutive memory locations and then executing a
8041 @code{movaps} to load the vector from memory.
8042 @item v4sf __builtin_ia32_setps1 (float)
8043 Constructs a vector from a single floating point value by replicating
8044 it across all four elements of the result vector.
8048 @itemx -mno-push-args
8050 @opindex mno-push-args
8051 Use PUSH operations to store outgoing parameters. This method is shorter
8052 and usually equally fast as method using SUB/MOV operations and is enabled
8053 by default. In some cases disabling it may improve performance because of
8054 improved scheduling and reduced dependencies.
8056 @item -maccumulate-outgoing-args
8057 @opindex maccumulate-outgoing-args
8058 If enabled, the maximum amount of space required for outgoing arguments will be
8059 computed in the function prologue. This is faster on most modern CPUs
8060 because of reduced dependencies, improved scheduling and reduced stack usage
8061 when preferred stack boundary is not equal to 2. The drawback is a notable
8062 increase in code size. This switch implies @option{-mno-push-args}.
8066 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
8067 on thread-safe exception handling must compile and link all code with the
8068 @option{-mthreads} option. When compiling, @option{-mthreads} defines
8069 @option{-D_MT}; when linking, it links in a special thread helper library
8070 @option{-lmingwthrd} which cleans up per thread exception handling data.
8072 @item -mno-align-stringops
8073 @opindex mno-align-stringops
8074 Do not align destination of inlined string operations. This switch reduces
8075 code size and improves performance in case the destination is already aligned,
8076 but gcc don't know about it.
8078 @item -minline-all-stringops
8079 @opindex minline-all-stringops
8080 By default GCC inlines string operations only when destination is known to be
8081 aligned at least to 4 byte boundary. This enables more inlining, increase code
8082 size, but may improve performance of code that depends on fast memcpy, strlen
8083 and memset for short lengths.
8085 @item -momit-leaf-frame-pointer
8086 @opindex momit-leaf-frame-pointer
8087 Don't keep the frame pointer in a register for leaf functions. This
8088 avoids the instructions to save, set up and restore frame pointers and
8089 makes an extra register available in leaf functions. The option
8090 @option{-fomit-frame-pointer} removes the frame pointer for all functions
8091 which might make debugging harder.
8094 These @samp{-m} switches are supported in addition to the above
8095 on AMD x86-64 processors in 64-bit environments.
8102 Generate code for a 32-bit or 64-bit environment.
8103 The 32-bit environment sets int, long and pointer to 32 bits and
8104 generates code that runs on any i386 system.
8105 The 64-bit environment sets int to 32 bits and long and pointer
8106 to 64 bits and generates code for AMD's x86-64 architecture.
8109 @opindex no-red-zone
8110 Do not use a so called red zone for x86-64 code. The red zone is mandated
8111 by the x86-64 ABI, it is a 128-byte area beyond the location of the
8112 stack pointer that will not be modified by signal or interrupt handlers
8113 and therefore can be used for temporary data without adjusting the stack
8114 pointer. The flag @option{-mno-red-zone} disables this red zone.
8118 @subsection HPPA Options
8119 @cindex HPPA Options
8121 These @samp{-m} options are defined for the HPPA family of computers:
8124 @item -march=@var{architecture-type}
8126 Generate code for the specified architecture. The choices for
8127 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
8128 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
8129 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
8130 architecture option for your machine. Code compiled for lower numbered
8131 architectures will run on higher numbered architectures, but not the
8134 PA 2.0 support currently requires gas snapshot 19990413 or later. The
8135 next release of binutils (current is 2.9.1) will probably contain PA 2.0
8139 @itemx -mpa-risc-1-1
8140 @itemx -mpa-risc-2-0
8141 @opindex mpa-risc-1-0
8142 @opindex mpa-risc-1-1
8143 @opindex mpa-risc-2-0
8144 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
8147 @opindex mbig-switch
8148 Generate code suitable for big switch tables. Use this option only if
8149 the assembler/linker complain about out of range branches within a switch
8152 @item -mjump-in-delay
8153 @opindex mjump-in-delay
8154 Fill delay slots of function calls with unconditional jump instructions
8155 by modifying the return pointer for the function call to be the target
8156 of the conditional jump.
8158 @item -mdisable-fpregs
8159 @opindex mdisable-fpregs
8160 Prevent floating point registers from being used in any manner. This is
8161 necessary for compiling kernels which perform lazy context switching of
8162 floating point registers. If you use this option and attempt to perform
8163 floating point operations, the compiler will abort.
8165 @item -mdisable-indexing
8166 @opindex mdisable-indexing
8167 Prevent the compiler from using indexing address modes. This avoids some
8168 rather obscure problems when compiling MIG generated code under MACH@.
8170 @item -mno-space-regs
8171 @opindex mno-space-regs
8172 Generate code that assumes the target has no space registers. This allows
8173 GCC to generate faster indirect calls and use unscaled index address modes.
8175 Such code is suitable for level 0 PA systems and kernels.
8177 @item -mfast-indirect-calls
8178 @opindex mfast-indirect-calls
8179 Generate code that assumes calls never cross space boundaries. This
8180 allows GCC to emit code which performs faster indirect calls.
8182 This option will not work in the presence of shared libraries or nested
8185 @item -mlong-load-store
8186 @opindex mlong-load-store
8187 Generate 3-instruction load and store sequences as sometimes required by
8188 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
8191 @item -mportable-runtime
8192 @opindex mportable-runtime
8193 Use the portable calling conventions proposed by HP for ELF systems.
8197 Enable the use of assembler directives only GAS understands.
8199 @item -mschedule=@var{cpu-type}
8201 Schedule code according to the constraints for the machine type
8202 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
8203 @samp{7100}, @samp{7100LC}, @samp{7200}, and @samp{8000}. Refer to
8204 @file{/usr/lib/sched.models} on an HP-UX system to determine the
8205 proper scheduling option for your machine.
8208 @opindex mlinker-opt
8209 Enable the optimization pass in the HPUX linker. Note this makes symbolic
8210 debugging impossible. It also triggers a bug in the HPUX 8 and HPUX 9 linkers
8211 in which they give bogus error messages when linking some programs.
8214 @opindex msoft-float
8215 Generate output containing library calls for floating point.
8216 @strong{Warning:} the requisite libraries are not available for all HPPA
8217 targets. Normally the facilities of the machine's usual C compiler are
8218 used, but this cannot be done directly in cross-compilation. You must make
8219 your own arrangements to provide suitable library functions for
8220 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
8221 does provide software floating point support.
8223 @option{-msoft-float} changes the calling convention in the output file;
8224 therefore, it is only useful if you compile @emph{all} of a program with
8225 this option. In particular, you need to compile @file{libgcc.a}, the
8226 library that comes with GCC, with @option{-msoft-float} in order for
8230 @node Intel 960 Options
8231 @subsection Intel 960 Options
8233 These @samp{-m} options are defined for the Intel 960 implementations:
8236 @item -m@var{cpu-type}
8244 Assume the defaults for the machine type @var{cpu-type} for some of
8245 the other options, including instruction scheduling, floating point
8246 support, and addressing modes. The choices for @var{cpu-type} are
8247 @samp{ka}, @samp{kb}, @samp{mc}, @samp{ca}, @samp{cf},
8248 @samp{sa}, and @samp{sb}.
8255 @opindex msoft-float
8256 The @option{-mnumerics} option indicates that the processor does support
8257 floating-point instructions. The @option{-msoft-float} option indicates
8258 that floating-point support should not be assumed.
8260 @item -mleaf-procedures
8261 @itemx -mno-leaf-procedures
8262 @opindex mleaf-procedures
8263 @opindex mno-leaf-procedures
8264 Do (or do not) attempt to alter leaf procedures to be callable with the
8265 @code{bal} instruction as well as @code{call}. This will result in more
8266 efficient code for explicit calls when the @code{bal} instruction can be
8267 substituted by the assembler or linker, but less efficient code in other
8268 cases, such as calls via function pointers, or using a linker that doesn't
8269 support this optimization.
8272 @itemx -mno-tail-call
8274 @opindex mno-tail-call
8275 Do (or do not) make additional attempts (beyond those of the
8276 machine-independent portions of the compiler) to optimize tail-recursive
8277 calls into branches. You may not want to do this because the detection of
8278 cases where this is not valid is not totally complete. The default is
8279 @option{-mno-tail-call}.
8281 @item -mcomplex-addr
8282 @itemx -mno-complex-addr
8283 @opindex mcomplex-addr
8284 @opindex mno-complex-addr
8285 Assume (or do not assume) that the use of a complex addressing mode is a
8286 win on this implementation of the i960. Complex addressing modes may not
8287 be worthwhile on the K-series, but they definitely are on the C-series.
8288 The default is currently @option{-mcomplex-addr} for all processors except
8292 @itemx -mno-code-align
8293 @opindex mcode-align
8294 @opindex mno-code-align
8295 Align code to 8-byte boundaries for faster fetching (or don't bother).
8296 Currently turned on by default for C-series implementations only.
8299 @item -mclean-linkage
8300 @itemx -mno-clean-linkage
8301 @opindex mclean-linkage
8302 @opindex mno-clean-linkage
8303 These options are not fully implemented.
8307 @itemx -mic2.0-compat
8308 @itemx -mic3.0-compat
8310 @opindex mic2.0-compat
8311 @opindex mic3.0-compat
8312 Enable compatibility with iC960 v2.0 or v3.0.
8316 @opindex masm-compat
8318 Enable compatibility with the iC960 assembler.
8320 @item -mstrict-align
8321 @itemx -mno-strict-align
8322 @opindex mstrict-align
8323 @opindex mno-strict-align
8324 Do not permit (do permit) unaligned accesses.
8328 Enable structure-alignment compatibility with Intel's gcc release version
8329 1.3 (based on gcc 1.37). This option implies @option{-mstrict-align}.
8331 @item -mlong-double-64
8332 @opindex mlong-double-64
8333 Implement type @samp{long double} as 64-bit floating point numbers.
8334 Without the option @samp{long double} is implemented by 80-bit
8335 floating point numbers. The only reason we have it because there is
8336 no 128-bit @samp{long double} support in @samp{fp-bit.c} yet. So it
8337 is only useful for people using soft-float targets. Otherwise, we
8338 should recommend against use of it.
8342 @node DEC Alpha Options
8343 @subsection DEC Alpha Options
8345 These @samp{-m} options are defined for the DEC Alpha implementations:
8348 @item -mno-soft-float
8350 @opindex mno-soft-float
8351 @opindex msoft-float
8352 Use (do not use) the hardware floating-point instructions for
8353 floating-point operations. When @option{-msoft-float} is specified,
8354 functions in @file{libgcc.a} will be used to perform floating-point
8355 operations. Unless they are replaced by routines that emulate the
8356 floating-point operations, or compiled in such a way as to call such
8357 emulations routines, these routines will issue floating-point
8358 operations. If you are compiling for an Alpha without floating-point
8359 operations, you must ensure that the library is built so as not to call
8362 Note that Alpha implementations without floating-point operations are
8363 required to have floating-point registers.
8368 @opindex mno-fp-regs
8369 Generate code that uses (does not use) the floating-point register set.
8370 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
8371 register set is not used, floating point operands are passed in integer
8372 registers as if they were integers and floating-point results are passed
8373 in $0 instead of $f0. This is a non-standard calling sequence, so any
8374 function with a floating-point argument or return value called by code
8375 compiled with @option{-mno-fp-regs} must also be compiled with that
8378 A typical use of this option is building a kernel that does not use,
8379 and hence need not save and restore, any floating-point registers.
8383 The Alpha architecture implements floating-point hardware optimized for
8384 maximum performance. It is mostly compliant with the IEEE floating
8385 point standard. However, for full compliance, software assistance is
8386 required. This option generates code fully IEEE compliant code
8387 @emph{except} that the @var{inexact-flag} is not maintained (see below).
8388 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
8389 defined during compilation. The resulting code is less efficient but is
8390 able to correctly support denormalized numbers and exceptional IEEE
8391 values such as not-a-number and plus/minus infinity. Other Alpha
8392 compilers call this option @option{-ieee_with_no_inexact}.
8394 @item -mieee-with-inexact
8395 @opindex mieee-with-inexact
8396 This is like @option{-mieee} except the generated code also maintains
8397 the IEEE @var{inexact-flag}. Turning on this option causes the
8398 generated code to implement fully-compliant IEEE math. In addition to
8399 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
8400 macro. On some Alpha implementations the resulting code may execute
8401 significantly slower than the code generated by default. Since there is
8402 very little code that depends on the @var{inexact-flag}, you should
8403 normally not specify this option. Other Alpha compilers call this
8404 option @option{-ieee_with_inexact}.
8406 @item -mfp-trap-mode=@var{trap-mode}
8407 @opindex mfp-trap-mode
8408 This option controls what floating-point related traps are enabled.
8409 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
8410 The trap mode can be set to one of four values:
8414 This is the default (normal) setting. The only traps that are enabled
8415 are the ones that cannot be disabled in software (e.g., division by zero
8419 In addition to the traps enabled by @samp{n}, underflow traps are enabled
8423 Like @samp{su}, but the instructions are marked to be safe for software
8424 completion (see Alpha architecture manual for details).
8427 Like @samp{su}, but inexact traps are enabled as well.
8430 @item -mfp-rounding-mode=@var{rounding-mode}
8431 @opindex mfp-rounding-mode
8432 Selects the IEEE rounding mode. Other Alpha compilers call this option
8433 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
8438 Normal IEEE rounding mode. Floating point numbers are rounded towards
8439 the nearest machine number or towards the even machine number in case
8443 Round towards minus infinity.
8446 Chopped rounding mode. Floating point numbers are rounded towards zero.
8449 Dynamic rounding mode. A field in the floating point control register
8450 (@var{fpcr}, see Alpha architecture reference manual) controls the
8451 rounding mode in effect. The C library initializes this register for
8452 rounding towards plus infinity. Thus, unless your program modifies the
8453 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
8456 @item -mtrap-precision=@var{trap-precision}
8457 @opindex mtrap-precision
8458 In the Alpha architecture, floating point traps are imprecise. This
8459 means without software assistance it is impossible to recover from a
8460 floating trap and program execution normally needs to be terminated.
8461 GCC can generate code that can assist operating system trap handlers
8462 in determining the exact location that caused a floating point trap.
8463 Depending on the requirements of an application, different levels of
8464 precisions can be selected:
8468 Program precision. This option is the default and means a trap handler
8469 can only identify which program caused a floating point exception.
8472 Function precision. The trap handler can determine the function that
8473 caused a floating point exception.
8476 Instruction precision. The trap handler can determine the exact
8477 instruction that caused a floating point exception.
8480 Other Alpha compilers provide the equivalent options called
8481 @option{-scope_safe} and @option{-resumption_safe}.
8483 @item -mieee-conformant
8484 @opindex mieee-conformant
8485 This option marks the generated code as IEEE conformant. You must not
8486 use this option unless you also specify @option{-mtrap-precision=i} and either
8487 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
8488 is to emit the line @samp{.eflag 48} in the function prologue of the
8489 generated assembly file. Under DEC Unix, this has the effect that
8490 IEEE-conformant math library routines will be linked in.
8492 @item -mbuild-constants
8493 @opindex mbuild-constants
8494 Normally GCC examines a 32- or 64-bit integer constant to
8495 see if it can construct it from smaller constants in two or three
8496 instructions. If it cannot, it will output the constant as a literal and
8497 generate code to load it from the data segment at runtime.
8499 Use this option to require GCC to construct @emph{all} integer constants
8500 using code, even if it takes more instructions (the maximum is six).
8502 You would typically use this option to build a shared library dynamic
8503 loader. Itself a shared library, it must relocate itself in memory
8504 before it can find the variables and constants in its own data segment.
8510 Select whether to generate code to be assembled by the vendor-supplied
8511 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
8525 Indicate whether GCC should generate code to use the optional BWX,
8526 CIX, and MAX instruction sets. The default is to use the instruction sets
8527 supported by the CPU type specified via @option{-mcpu=} option or that
8528 of the CPU on which GCC was built if none was specified.
8530 @item -mcpu=@var{cpu_type}
8532 Set the instruction set, register set, and instruction scheduling
8533 parameters for machine type @var{cpu_type}. You can specify either the
8534 @samp{EV} style name or the corresponding chip number. GCC
8535 supports scheduling parameters for the EV4 and EV5 family of processors
8536 and will choose the default values for the instruction set from
8537 the processor you specify. If you do not specify a processor type,
8538 GCC will default to the processor on which the compiler was built.
8540 Supported values for @var{cpu_type} are
8545 Schedules as an EV4 and has no instruction set extensions.
8549 Schedules as an EV5 and has no instruction set extensions.
8553 Schedules as an EV5 and supports the BWX extension.
8558 Schedules as an EV5 and supports the BWX and MAX extensions.
8562 Schedules as an EV5 (until Digital releases the scheduling parameters
8563 for the EV6) and supports the BWX, CIX, and MAX extensions.
8566 @item -mmemory-latency=@var{time}
8567 @opindex mmemory-latency
8568 Sets the latency the scheduler should assume for typical memory
8569 references as seen by the application. This number is highly
8570 dependent on the memory access patterns used by the application
8571 and the size of the external cache on the machine.
8573 Valid options for @var{time} are
8577 A decimal number representing clock cycles.
8583 The compiler contains estimates of the number of clock cycles for
8584 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
8585 (also called Dcache, Scache, and Bcache), as well as to main memory.
8586 Note that L3 is only valid for EV5.
8591 @node Clipper Options
8592 @subsection Clipper Options
8594 These @samp{-m} options are defined for the Clipper implementations:
8599 Produce code for a C300 Clipper processor. This is the default.
8603 Produce code for a C400 Clipper processor, i.e.@: use floating point
8607 @node H8/300 Options
8608 @subsection H8/300 Options
8610 These @samp{-m} options are defined for the H8/300 implementations:
8615 Shorten some address references at link time, when possible; uses the
8616 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
8617 ld.info, Using ld}, for a fuller description.
8621 Generate code for the H8/300H@.
8625 Generate code for the H8/S@.
8629 Generate code for the H8/S2600. This switch must be used with @option{-ms}.
8633 Make @code{int} data 32 bits by default.
8637 On the H8/300H and H8/S, use the same alignment rules as for the H8/300.
8638 The default for the H8/300H and H8/S is to align longs and floats on 4
8640 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
8641 This option has no effect on the H8/300.
8645 @subsection SH Options
8647 These @samp{-m} options are defined for the SH implementations:
8652 Generate code for the SH1.
8656 Generate code for the SH2.
8660 Generate code for the SH3.
8664 Generate code for the SH3e.
8668 Generate code for the SH4 without a floating-point unit.
8670 @item -m4-single-only
8671 @opindex m4-single-only
8672 Generate code for the SH4 with a floating-point unit that only
8673 supports single-precision arithmetic.
8677 Generate code for the SH4 assuming the floating-point unit is in
8678 single-precision mode by default.
8682 Generate code for the SH4.
8686 Compile code for the processor in big endian mode.
8690 Compile code for the processor in little endian mode.
8694 Align doubles at 64-bit boundaries. Note that this changes the calling
8695 conventions, and thus some functions from the standard C library will
8696 not work unless you recompile it first with @option{-mdalign}.
8700 Shorten some address references at link time, when possible; uses the
8701 linker option @option{-relax}.
8705 Use 32-bit offsets in @code{switch} tables. The default is to use
8710 Enable the use of the instruction @code{fmovd}.
8714 Comply with the calling conventions defined by Hitachi.
8718 Mark the @code{MAC} register as call-clobbered, even if
8719 @option{-mhitachi} is given.
8723 Increase IEEE-compliance of floating-point code.
8727 Dump instruction size and location in the assembly code.
8731 This option is deprecated. It pads structures to multiple of 4 bytes,
8732 which is incompatible with the SH ABI@.
8736 Optimize for space instead of speed. Implied by @option{-Os}.
8740 When generating position-independent code, emit function calls using
8741 the Global Offset Table instead of the Procedure Linkage Table.
8745 Generate a library function call to invalidate instruction cache
8746 entries, after fixing up a trampoline. This library function call
8747 doesn't assume it can write to the whole memory address space. This
8748 is the default when the target is @code{sh-*-linux*}.
8751 @node System V Options
8752 @subsection Options for System V
8754 These additional options are available on System V Release 4 for
8755 compatibility with other compilers on those systems:
8760 Create a shared object.
8761 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
8765 Identify the versions of each tool used by the compiler, in a
8766 @code{.ident} assembler directive in the output.
8770 Refrain from adding @code{.ident} directives to the output file (this is
8773 @item -YP,@var{dirs}
8775 Search the directories @var{dirs}, and no others, for libraries
8776 specified with @option{-l}.
8780 Look in the directory @var{dir} to find the M4 preprocessor.
8781 The assembler uses this option.
8782 @c This is supposed to go with a -Yd for predefined M4 macro files, but
8783 @c the generic assembler that comes with Solaris takes just -Ym.
8786 @node TMS320C3x/C4x Options
8787 @subsection TMS320C3x/C4x Options
8788 @cindex TMS320C3x/C4x Options
8790 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
8794 @item -mcpu=@var{cpu_type}
8796 Set the instruction set, register set, and instruction scheduling
8797 parameters for machine type @var{cpu_type}. Supported values for
8798 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
8799 @samp{c44}. The default is @samp{c40} to generate code for the
8804 @itemx -msmall-memory
8806 @opindex mbig-memory
8808 @opindex msmall-memory
8810 Generates code for the big or small memory model. The small memory
8811 model assumed that all data fits into one 64K word page. At run-time
8812 the data page (DP) register must be set to point to the 64K page
8813 containing the .bss and .data program sections. The big memory model is
8814 the default and requires reloading of the DP register for every direct
8821 Allow (disallow) allocation of general integer operands into the block
8828 Enable (disable) generation of code using decrement and branch,
8829 DBcond(D), instructions. This is enabled by default for the C4x. To be
8830 on the safe side, this is disabled for the C3x, since the maximum
8831 iteration count on the C3x is @math{2^23 + 1} (but who iterates loops more than
8832 @math{2^23} times on the C3x?). Note that GCC will try to reverse a loop so
8833 that it can utilise the decrement and branch instruction, but will give
8834 up if there is more than one memory reference in the loop. Thus a loop
8835 where the loop counter is decremented can generate slightly more
8836 efficient code, in cases where the RPTB instruction cannot be utilised.
8838 @item -mdp-isr-reload
8840 @opindex mdp-isr-reload
8842 Force the DP register to be saved on entry to an interrupt service
8843 routine (ISR), reloaded to point to the data section, and restored on
8844 exit from the ISR@. This should not be required unless someone has
8845 violated the small memory model by modifying the DP register, say within
8852 For the C3x use the 24-bit MPYI instruction for integer multiplies
8853 instead of a library call to guarantee 32-bit results. Note that if one
8854 of the operands is a constant, then the multiplication will be performed
8855 using shifts and adds. If the @option{-mmpyi} option is not specified for the C3x,
8856 then squaring operations are performed inline instead of a library call.
8859 @itemx -mno-fast-fix
8861 @opindex mno-fast-fix
8862 The C3x/C4x FIX instruction to convert a floating point value to an
8863 integer value chooses the nearest integer less than or equal to the
8864 floating point value rather than to the nearest integer. Thus if the
8865 floating point number is negative, the result will be incorrectly
8866 truncated an additional code is necessary to detect and correct this
8867 case. This option can be used to disable generation of the additional
8868 code required to correct the result.
8874 Enable (disable) generation of repeat block sequences using the RPTB
8875 instruction for zero overhead looping. The RPTB construct is only used
8876 for innermost loops that do not call functions or jump across the loop
8877 boundaries. There is no advantage having nested RPTB loops due to the
8878 overhead required to save and restore the RC, RS, and RE registers.
8879 This is enabled by default with @option{-O2}.
8881 @item -mrpts=@var{count}
8885 Enable (disable) the use of the single instruction repeat instruction
8886 RPTS@. If a repeat block contains a single instruction, and the loop
8887 count can be guaranteed to be less than the value @var{count}, GCC will
8888 emit a RPTS instruction instead of a RPTB@. If no value is specified,
8889 then a RPTS will be emitted even if the loop count cannot be determined
8890 at compile time. Note that the repeated instruction following RPTS does
8891 not have to be reloaded from memory each iteration, thus freeing up the
8892 CPU buses for operands. However, since interrupts are blocked by this
8893 instruction, it is disabled by default.
8895 @item -mloop-unsigned
8896 @itemx -mno-loop-unsigned
8897 @opindex mloop-unsigned
8898 @opindex mno-loop-unsigned
8899 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
8900 is @math{2^31 + 1} since these instructions test if the iteration count is
8901 negative to terminate the loop. If the iteration count is unsigned
8902 there is a possibility than the @math{2^31 + 1} maximum iteration count may be
8903 exceeded. This switch allows an unsigned iteration count.
8907 Try to emit an assembler syntax that the TI assembler (asm30) is happy
8908 with. This also enforces compatibility with the API employed by the TI
8909 C3x C compiler. For example, long doubles are passed as structures
8910 rather than in floating point registers.
8916 Generate code that uses registers (stack) for passing arguments to functions.
8917 By default, arguments are passed in registers where possible rather
8918 than by pushing arguments on to the stack.
8920 @item -mparallel-insns
8921 @itemx -mno-parallel-insns
8922 @opindex mparallel-insns
8923 @opindex mno-parallel-insns
8924 Allow the generation of parallel instructions. This is enabled by
8925 default with @option{-O2}.
8927 @item -mparallel-mpy
8928 @itemx -mno-parallel-mpy
8929 @opindex mparallel-mpy
8930 @opindex mno-parallel-mpy
8931 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
8932 provided @option{-mparallel-insns} is also specified. These instructions have
8933 tight register constraints which can pessimize the code generation
8939 @subsection V850 Options
8940 @cindex V850 Options
8942 These @samp{-m} options are defined for V850 implementations:
8946 @itemx -mno-long-calls
8947 @opindex mlong-calls
8948 @opindex mno-long-calls
8949 Treat all calls as being far away (near). If calls are assumed to be
8950 far away, the compiler will always load the functions address up into a
8951 register, and call indirect through the pointer.
8957 Do not optimize (do optimize) basic blocks that use the same index
8958 pointer 4 or more times to copy pointer into the @code{ep} register, and
8959 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
8960 option is on by default if you optimize.
8962 @item -mno-prolog-function
8963 @itemx -mprolog-function
8964 @opindex mno-prolog-function
8965 @opindex mprolog-function
8966 Do not use (do use) external functions to save and restore registers at
8967 the prolog and epilog of a function. The external functions are slower,
8968 but use less code space if more than one function saves the same number
8969 of registers. The @option{-mprolog-function} option is on by default if
8974 Try to make the code as small as possible. At present, this just turns
8975 on the @option{-mep} and @option{-mprolog-function} options.
8979 Put static or global variables whose size is @var{n} bytes or less into
8980 the tiny data area that register @code{ep} points to. The tiny data
8981 area can hold up to 256 bytes in total (128 bytes for byte references).
8985 Put static or global variables whose size is @var{n} bytes or less into
8986 the small data area that register @code{gp} points to. The small data
8987 area can hold up to 64 kilobytes.
8991 Put static or global variables whose size is @var{n} bytes or less into
8992 the first 32 kilobytes of memory.
8996 Specify that the target processor is the V850.
8999 @opindex mbig-switch
9000 Generate code suitable for big switch tables. Use this option only if
9001 the assembler/linker complain about out of range branches within a switch
9006 @subsection ARC Options
9009 These options are defined for ARC implementations:
9014 Compile code for little endian mode. This is the default.
9018 Compile code for big endian mode.
9021 @opindex mmangle-cpu
9022 Prepend the name of the cpu to all public symbol names.
9023 In multiple-processor systems, there are many ARC variants with different
9024 instruction and register set characteristics. This flag prevents code
9025 compiled for one cpu to be linked with code compiled for another.
9026 No facility exists for handling variants that are ``almost identical''.
9027 This is an all or nothing option.
9029 @item -mcpu=@var{cpu}
9031 Compile code for ARC variant @var{cpu}.
9032 Which variants are supported depend on the configuration.
9033 All variants support @option{-mcpu=base}, this is the default.
9035 @item -mtext=@var{text-section}
9036 @itemx -mdata=@var{data-section}
9037 @itemx -mrodata=@var{readonly-data-section}
9041 Put functions, data, and readonly data in @var{text-section},
9042 @var{data-section}, and @var{readonly-data-section} respectively
9043 by default. This can be overridden with the @code{section} attribute.
9044 @xref{Variable Attributes}.
9049 @subsection NS32K Options
9050 @cindex NS32K options
9052 These are the @samp{-m} options defined for the 32000 series. The default
9053 values for these options depends on which style of 32000 was selected when
9054 the compiler was configured; the defaults for the most common choices are
9062 Generate output for a 32032. This is the default
9063 when the compiler is configured for 32032 and 32016 based systems.
9069 Generate output for a 32332. This is the default
9070 when the compiler is configured for 32332-based systems.
9076 Generate output for a 32532. This is the default
9077 when the compiler is configured for 32532-based systems.
9081 Generate output containing 32081 instructions for floating point.
9082 This is the default for all systems.
9086 Generate output containing 32381 instructions for floating point. This
9087 also implies @option{-m32081}. The 32381 is only compatible with the 32332
9088 and 32532 cpus. This is the default for the pc532-netbsd configuration.
9092 Try and generate multiply-add floating point instructions @code{polyF}
9093 and @code{dotF}. This option is only available if the @option{-m32381}
9094 option is in effect. Using these instructions requires changes to
9095 register allocation which generally has a negative impact on
9096 performance. This option should only be enabled when compiling code
9097 particularly likely to make heavy use of multiply-add instructions.
9100 @opindex mnomulti-add
9101 Do not try and generate multiply-add floating point instructions
9102 @code{polyF} and @code{dotF}. This is the default on all platforms.
9105 @opindex msoft-float
9106 Generate output containing library calls for floating point.
9107 @strong{Warning:} the requisite libraries may not be available.
9110 @opindex mnobitfield
9111 Do not use the bit-field instructions. On some machines it is faster to
9112 use shifting and masking operations. This is the default for the pc532.
9116 Do use the bit-field instructions. This is the default for all platforms
9121 Use a different function-calling convention, in which functions
9122 that take a fixed number of arguments return pop their
9123 arguments on return with the @code{ret} instruction.
9125 This calling convention is incompatible with the one normally
9126 used on Unix, so you cannot use it if you need to call libraries
9127 compiled with the Unix compiler.
9129 Also, you must provide function prototypes for all functions that
9130 take variable numbers of arguments (including @code{printf});
9131 otherwise incorrect code will be generated for calls to those
9134 In addition, seriously incorrect code will result if you call a
9135 function with too many arguments. (Normally, extra arguments are
9136 harmlessly ignored.)
9138 This option takes its name from the 680x0 @code{rtd} instruction.
9143 Use a different function-calling convention where the first two arguments
9144 are passed in registers.
9146 This calling convention is incompatible with the one normally
9147 used on Unix, so you cannot use it if you need to call libraries
9148 compiled with the Unix compiler.
9151 @opindex mnoregparam
9152 Do not pass any arguments in registers. This is the default for all
9157 It is OK to use the sb as an index register which is always loaded with
9158 zero. This is the default for the pc532-netbsd target.
9162 The sb register is not available for use or has not been initialized to
9163 zero by the run time system. This is the default for all targets except
9164 the pc532-netbsd. It is also implied whenever @option{-mhimem} or
9165 @option{-fpic} is set.
9169 Many ns32000 series addressing modes use displacements of up to 512MB@.
9170 If an address is above 512MB then displacements from zero can not be used.
9171 This option causes code to be generated which can be loaded above 512MB@.
9172 This may be useful for operating systems or ROM code.
9176 Assume code will be loaded in the first 512MB of virtual address space.
9177 This is the default for all platforms.
9183 @subsection AVR Options
9186 These options are defined for AVR implementations:
9189 @item -mmcu=@var{mcu}
9191 Specify ATMEL AVR instruction set or MCU type.
9193 Instruction set avr1 is for the minimal AVR core, not supported by the C
9194 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
9195 attiny11, attiny12, attiny15, attiny28).
9197 Instruction set avr2 (default) is for the classic AVR core with up to
9198 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
9199 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
9200 at90c8534, at90s8535).
9202 Instruction set avr3 is for the classic AVR core with up to 128K program
9203 memory space (MCU types: atmega103, atmega603).
9205 Instruction set avr4 is for the enhanced AVR core with up to 8K program
9206 memory space (MCU types: atmega83, atmega85).
9208 Instruction set avr5 is for the enhanced AVR core with up to 128K program
9209 memory space (MCU types: atmega161, atmega163, atmega32, at94k).
9213 Output instruction sizes to the asm file.
9215 @item -minit-stack=@var{N}
9216 @opindex minit-stack
9217 Specify the initial stack address, which may be a symbol or numeric value,
9218 @samp{__stack} is the default.
9220 @item -mno-interrupts
9221 @opindex mno-interrupts
9222 Generated code is not compatible with hardware interrupts.
9223 Code size will be smaller.
9225 @item -mcall-prologues
9226 @opindex mcall-prologues
9227 Functions prologues/epilogues expanded as call to appropriate
9228 subroutines. Code size will be smaller.
9230 @item -mno-tablejump
9231 @opindex mno-tablejump
9232 Do not generate tablejump insns which sometimes increase code size.
9235 @opindex mtiny-stack
9236 Change only the low 8 bits of the stack pointer.
9240 @subsection MCore Options
9241 @cindex MCore options
9243 These are the @samp{-m} options defined for the Motorola M*Core
9253 @opindex mno-hardlit
9254 Inline constants into the code stream if it can be done in two
9255 instructions or less.
9263 Use the divide instruction. (Enabled by default).
9265 @item -mrelax-immediate
9266 @itemx -mrelax-immediate
9267 @itemx -mno-relax-immediate
9268 @opindex mrelax-immediate
9269 @opindex mrelax-immediate
9270 @opindex mno-relax-immediate
9271 Allow arbitrary sized immediates in bit operations.
9273 @item -mwide-bitfields
9274 @itemx -mwide-bitfields
9275 @itemx -mno-wide-bitfields
9276 @opindex mwide-bitfields
9277 @opindex mwide-bitfields
9278 @opindex mno-wide-bitfields
9279 Always treat bit-fields as int-sized.
9281 @item -m4byte-functions
9282 @itemx -m4byte-functions
9283 @itemx -mno-4byte-functions
9284 @opindex m4byte-functions
9285 @opindex m4byte-functions
9286 @opindex mno-4byte-functions
9287 Force all functions to be aligned to a four byte boundary.
9289 @item -mcallgraph-data
9290 @itemx -mcallgraph-data
9291 @itemx -mno-callgraph-data
9292 @opindex mcallgraph-data
9293 @opindex mcallgraph-data
9294 @opindex mno-callgraph-data
9295 Emit callgraph information.
9299 @itemx -mno-slow-bytes
9300 @opindex mslow-bytes
9301 @opindex mslow-bytes
9302 @opindex mno-slow-bytes
9303 Prefer word access when reading byte quantities.
9305 @item -mlittle-endian
9306 @itemx -mlittle-endian
9308 @opindex mlittle-endian
9309 @opindex mlittle-endian
9310 @opindex mbig-endian
9311 Generate code for a little endian target.
9319 Generate code for the 210 processor.
9323 @subsection IA-64 Options
9324 @cindex IA-64 Options
9326 These are the @samp{-m} options defined for the Intel IA-64 architecture.
9330 @opindex mbig-endian
9331 Generate code for a big endian target. This is the default for HPUX@.
9333 @item -mlittle-endian
9334 @opindex mlittle-endian
9335 Generate code for a little endian target. This is the default for AIX5
9342 Generate (or don't) code for the GNU assembler. This is the default.
9343 @c Also, this is the default if the configure option @option{--with-gnu-as}
9350 Generate (or don't) code for the GNU linker. This is the default.
9351 @c Also, this is the default if the configure option @option{--with-gnu-ld}
9356 Generate code that does not use a global pointer register. The result
9357 is not position independent code, and violates the IA-64 ABI@.
9359 @item -mvolatile-asm-stop
9360 @itemx -mno-volatile-asm-stop
9361 @opindex mvolatile-asm-stop
9362 @opindex mno-volatile-asm-stop
9363 Generate (or don't) a stop bit immediately before and after volatile asm
9368 Generate code that works around Itanium B step errata.
9370 @item -mregister-names
9371 @itemx -mno-register-names
9372 @opindex mregister-names
9373 @opindex mno-register-names
9374 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
9375 the stacked registers. This may make assembler output more readable.
9381 Disable (or enable) optimizations that use the small data section. This may
9382 be useful for working around optimizer bugs.
9385 @opindex mconstant-gp
9386 Generate code that uses a single constant global pointer value. This is
9387 useful when compiling kernel code.
9391 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
9392 This is useful when compiling firmware code.
9394 @item -minline-divide-min-latency
9395 @opindex minline-divide-min-latency
9396 Generate code for inline divides using the minimum latency algorithm.
9398 @item -minline-divide-max-throughput
9399 @opindex minline-divide-max-throughput
9400 Generate code for inline divides using the maximum throughput algorithm.
9402 @item -mno-dwarf2-asm
9404 @opindex mno-dwarf2-asm
9405 @opindex mdwarf2-asm
9406 Don't (or do) generate assembler code for the DWARF2 line number debugging
9407 info. This may be useful when not using the GNU assembler.
9409 @item -mfixed-range=@var{register-range}
9410 @opindex mfixed-range
9411 Generate code treating the given register range as fixed registers.
9412 A fixed register is one that the register allocator can not use. This is
9413 useful when compiling kernel code. A register range is specified as
9414 two registers separated by a dash. Multiple register ranges can be
9415 specified separated by a comma.
9419 @subsection D30V Options
9420 @cindex D30V Options
9422 These @samp{-m} options are defined for D30V implementations:
9427 Link the @samp{.text}, @samp{.data}, @samp{.bss}, @samp{.strings},
9428 @samp{.rodata}, @samp{.rodata1}, @samp{.data1} sections into external
9429 memory, which starts at location @code{0x80000000}.
9433 Same as the @option{-mextmem} switch.
9437 Link the @samp{.text} section into onchip text memory, which starts at
9438 location @code{0x0}. Also link @samp{.data}, @samp{.bss},
9439 @samp{.strings}, @samp{.rodata}, @samp{.rodata1}, @samp{.data1} sections
9440 into onchip data memory, which starts at location @code{0x20000000}.
9442 @item -mno-asm-optimize
9443 @itemx -masm-optimize
9444 @opindex mno-asm-optimize
9445 @opindex masm-optimize
9446 Disable (enable) passing @option{-O} to the assembler when optimizing.
9447 The assembler uses the @option{-O} option to automatically parallelize
9448 adjacent short instructions where possible.
9450 @item -mbranch-cost=@var{n}
9451 @opindex mbranch-cost
9452 Increase the internal costs of branches to @var{n}. Higher costs means
9453 that the compiler will issue more instructions to avoid doing a branch.
9456 @item -mcond-exec=@var{n}
9458 Specify the maximum number of conditionally executed instructions that
9459 replace a branch. The default is 4.
9462 @node S/390 and zSeries Options
9463 @subsection S/390 and zSeries Options
9464 @cindex S/390 and zSeries Options
9466 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
9471 @opindex mhard-float
9472 @opindex msoft-float
9473 Use (do not use) the hardware floating-point instructions and registers
9474 for floating-point operations. When @option{-msoft-float} is specified,
9475 functions in @file{libgcc.a} will be used to perform floating-point
9476 operations. When @option{-mhard-float} is specified, the compiler
9477 generates IEEE floating-point instructions. This is the default.
9480 @itemx -mno-backchain
9482 @opindex mno-backchain
9483 Generate (or do not generate) code which maintains an explicit
9484 backchain within the stack frame that points to the caller's frame.
9485 This is currently needed to allow debugging. The default is to
9486 generate the backchain.
9489 @itemx -mno-small-exec
9490 @opindex msmall-exec
9491 @opindex mno-small-exec
9492 Generate (or do not generate) code using the @code{bras} instruction
9493 to do subroutine calls.
9494 This only works reliably if the total executable size does not
9495 exceed 64k. The default is to use the @code{basr} instruction instead,
9496 which does not have this limitation.
9502 When @option{-m31} is specified, generate code compliant to the
9503 Linux for S/390 ABI@. When @option{-m64} is specified, generate
9504 code compliant to the Linux for zSeries ABI@. This allows GCC in
9505 particular to generate 64-bit instructions. For the @samp{s390}
9506 targets, the default is @option{-m31}, while the @samp{s390x}
9507 targets default to @option{-m64}.
9513 Generate (or do not generate) code using the @code{mvcle} instruction
9514 to perform block moves. When @option{-mno-mvcle} is specifed,
9515 use a @code{mvc} loop instead. This is the default.
9521 Print (or do not print) additional debug information when compiling.
9522 The default is to not print debug information.
9527 @subsection CRIS Options
9528 @cindex CRIS Options
9530 These options are defined specifically for the CRIS ports.
9533 @item -march=@var{architecture-type}
9534 @itemx -mcpu=@var{architecture-type}
9537 Generate code for the specified architecture. The choices for
9538 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
9539 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX.
9540 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
9543 @item -mtune=@var{architecture-type}
9545 Tune to @var{architecture-type} everything applicable about the generated
9546 code, except for the ABI and the set of available instructions. The
9547 choices for @var{architecture-type} are the same as for
9548 @option{-march=@var{architecture-type}}.
9550 @item -mmax-stack-frame=@var{n}
9551 @opindex mmax-stack-frame
9552 Warn when the stack frame of a function exceeds @var{n} bytes.
9554 @item -melinux-stacksize=@var{n}
9555 @opindex melinux-stacksize
9556 Only available with the @samp{cris-axis-aout} target. Arranges for
9557 indications in the program to the kernel loader that the stack of the
9558 program should be set to @var{n} bytes.
9564 The options @option{-metrax4} and @option{-metrax100} are synonyms for
9565 @option{-march=v3} and @option{-march=v8} respectively.
9569 Enable CRIS-specific verbose debug-related information in the assembly
9570 code. This option also has the effect to turn off the @samp{#NO_APP}
9571 formatted-code indicator to the assembler at the beginning of the
9576 Do not use condition-code results from previous instruction; always emit
9577 compare and test instructions before use of condition codes.
9579 @item -mno-side-effects
9580 @opindex mno-side-effects
9581 Do not emit instructions with side-effects in addressing modes other than
9585 @itemx -mno-stack-align
9587 @itemx -mno-data-align
9588 @itemx -mconst-align
9589 @itemx -mno-const-align
9590 @opindex mstack-align
9591 @opindex mno-stack-align
9592 @opindex mdata-align
9593 @opindex mno-data-align
9594 @opindex mconst-align
9595 @opindex mno-const-align
9596 These options (no-options) arranges (eliminate arrangements) for the
9597 stack-frame, individual data and constants to be aligned for the maximum
9598 single data access size for the chosen CPU model. The default is to
9599 arrange for 32-bit alignment. ABI details such as structure layout are
9600 not affected by these options.
9608 Similar to the stack- data- and const-align options above, these options
9609 arrange for stack-frame, writable data and constants to all be 32-bit,
9610 16-bit or 8-bit aligned. The default is 32-bit alignment.
9612 @item -mno-prologue-epilogue
9613 @itemx -mprologue-epilogue
9614 @opindex mno-prologue-epilogue
9615 @opindex mprologue-epilogue
9616 With @option{-mno-prologue-epilogue}, the normal function prologue and
9617 epilogue that sets up the stack-frame are omitted and no return
9618 instructions or return sequences are generated in the code. Use this
9619 option only together with visual inspection of the compiled code: no
9620 warnings or errors are generated when call-saved registers must be saved,
9621 or storage for local variable needs to be allocated.
9627 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
9628 instruction sequences that load addresses for functions from the PLT part
9629 of the GOT rather than (traditional on other architectures) calls to the
9630 PLT. The default is @option{-mgotplt}.
9634 Legacy no-op option only recognized with the cris-axis-aout target.
9638 Legacy no-op option only recognized with the cris-axis-elf and
9639 cris-axis-linux-gnu targets.
9643 Only recognized with the cris-axis-aout target, where it selects a
9644 GNU/linux-like multilib, include files and instruction set for
9649 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
9653 This option, recognized for the cris-axis-aout and cris-axis-elf arranges
9654 to link with input-output functions from a simulator library. Code,
9655 initialized data and zero-initialized data are allocated consecutively.
9659 Like @option{-sim}, but pass linker options to locate initialized data at
9660 0x40000000 and zero-initialized data at 0x80000000.
9664 @subsection MMIX Options
9665 @cindex MMIX Options
9667 These options are defined for the MMIX:
9671 @itemx -mno-libfuncs
9672 Specify that intrinsic library functions are being compiled, passing all
9673 values in registers, no matter the size.
9677 Generate floating-point comparison instructions that compare with respect
9678 to the @code{rE} epsilon register.
9680 @item -mabi=mmixware
9682 Generate code that passes function parameters and return values that (in
9683 the called function) are seen as registers @code{$0} and up, as opposed to
9684 the GNU ABI which uses global registers @code{$231} and up.
9687 @item -mno-zero-extend
9688 When reading data from memory in sizes shorter than 64 bits, use (do not
9689 use) zero-extending load instructions by default, rather than
9690 sign-extending ones.
9693 @itemx -mno-knuthdiv
9694 Make the result of a division yielding a remainder have the same sign as
9695 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
9696 remainder follows the sign of the dividend. Both methods are
9697 arithmetically valid, the latter being almost exclusively used.
9699 @item -mtoplevel-symbols
9700 @itemx -mno-toplevel-symbols
9701 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
9702 code can be used with the @code{PREFIX} assembly directive.
9705 Generate an executable in the ELF format, rather than the default
9706 @samp{mmo} format used by the @command{mmix} simulator.
9710 @node Code Gen Options
9711 @section Options for Code Generation Conventions
9712 @cindex code generation conventions
9713 @cindex options, code generation
9714 @cindex run-time options
9716 These machine-independent options control the interface conventions
9717 used in code generation.
9719 Most of them have both positive and negative forms; the negative form
9720 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
9721 one of the forms is listed---the one which is not the default. You
9722 can figure out the other form by either removing @samp{no-} or adding
9727 @opindex fexceptions
9728 Enable exception handling. Generates extra code needed to propagate
9729 exceptions. For some targets, this implies GCC will generate frame
9730 unwind information for all functions, which can produce significant data
9731 size overhead, although it does not affect execution. If you do not
9732 specify this option, GCC will enable it by default for languages like
9733 C++ which normally require exception handling, and disable it for
9734 languages like C that do not normally require it. However, you may need
9735 to enable this option when compiling C code that needs to interoperate
9736 properly with exception handlers written in C++. You may also wish to
9737 disable this option if you are compiling older C++ programs that don't
9738 use exception handling.
9740 @item -fnon-call-exceptions
9741 @opindex fnon-call-exceptions
9742 Generate code that allows trapping instructions to throw exceptions.
9743 Note that this requires platform-specific runtime support that does
9744 not exist everywhere. Moreover, it only allows @emph{trapping}
9745 instructions to throw exceptions, i.e.@: memory references or floating
9746 point instructions. It does not allow exceptions to be thrown from
9747 arbitrary signal handlers such as @code{SIGALRM}.
9749 @item -funwind-tables
9750 @opindex funwind-tables
9751 Similar to @option{-fexceptions}, except that it will just generate any needed
9752 static data, but will not affect the generated code in any other way.
9753 You will normally not enable this option; instead, a language processor
9754 that needs this handling would enable it on your behalf.
9756 @item -fasynchronous-unwind-tables
9757 @opindex funwind-tables
9758 Generate unwind table in dwarf2 format, if supported by target machine. The
9759 table is exact at each instruction boundary, so it can be used for stack
9760 unwinding from asynchronous events (such as debugger or garbage collector).
9762 @item -fpcc-struct-return
9763 @opindex fpcc-struct-return
9764 Return ``short'' @code{struct} and @code{union} values in memory like
9765 longer ones, rather than in registers. This convention is less
9766 efficient, but it has the advantage of allowing intercallability between
9767 GCC-compiled files and files compiled with other compilers.
9769 The precise convention for returning structures in memory depends
9770 on the target configuration macros.
9772 Short structures and unions are those whose size and alignment match
9773 that of some integer type.
9775 @item -freg-struct-return
9776 @opindex freg-struct-return
9777 Return @code{struct} and @code{union} values in registers when possible.
9778 This is more efficient for small structures than
9779 @option{-fpcc-struct-return}.
9781 If you specify neither @option{-fpcc-struct-return} nor
9782 @option{-freg-struct-return}, GCC defaults to whichever convention is
9783 standard for the target. If there is no standard convention, GCC
9784 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
9785 the principal compiler. In those cases, we can choose the standard, and
9786 we chose the more efficient register return alternative.
9789 @opindex fshort-enums
9790 Allocate to an @code{enum} type only as many bytes as it needs for the
9791 declared range of possible values. Specifically, the @code{enum} type
9792 will be equivalent to the smallest integer type which has enough room.
9794 @item -fshort-double
9795 @opindex fshort-double
9796 Use the same size for @code{double} as for @code{float}.
9799 @opindex fshared-data
9800 Requests that the data and non-@code{const} variables of this
9801 compilation be shared data rather than private data. The distinction
9802 makes sense only on certain operating systems, where shared data is
9803 shared between processes running the same program, while private data
9804 exists in one copy per process.
9808 In C, allocate even uninitialized global variables in the data section of the
9809 object file, rather than generating them as common blocks. This has the
9810 effect that if the same variable is declared (without @code{extern}) in
9811 two different compilations, you will get an error when you link them.
9812 The only reason this might be useful is if you wish to verify that the
9813 program will work on other systems which always work this way.
9817 Ignore the @samp{#ident} directive.
9819 @item -fno-gnu-linker
9820 @opindex fno-gnu-linker
9821 Do not output global initializations (such as C++ constructors and
9822 destructors) in the form used by the GNU linker (on systems where the GNU
9823 linker is the standard method of handling them). Use this option when
9824 you want to use a non-GNU linker, which also requires using the
9825 @command{collect2} program to make sure the system linker includes
9826 constructors and destructors. (@command{collect2} is included in the GCC
9827 distribution.) For systems which @emph{must} use @command{collect2}, the
9828 compiler driver @command{gcc} is configured to do this automatically.
9830 @item -finhibit-size-directive
9831 @opindex finhibit-size-directive
9832 Don't output a @code{.size} assembler directive, or anything else that
9833 would cause trouble if the function is split in the middle, and the
9834 two halves are placed at locations far apart in memory. This option is
9835 used when compiling @file{crtstuff.c}; you should not need to use it
9839 @opindex fverbose-asm
9840 Put extra commentary information in the generated assembly code to
9841 make it more readable. This option is generally only of use to those
9842 who actually need to read the generated assembly code (perhaps while
9843 debugging the compiler itself).
9845 @option{-fno-verbose-asm}, the default, causes the
9846 extra information to be omitted and is useful when comparing two assembler
9851 Consider all memory references through pointers to be volatile.
9853 @item -fvolatile-global
9854 @opindex fvolatile-global
9855 Consider all memory references to extern and global data items to
9856 be volatile. GCC does not consider static data items to be volatile
9857 because of this switch.
9859 @item -fvolatile-static
9860 @opindex fvolatile-static
9861 Consider all memory references to static data to be volatile.
9865 @cindex global offset table
9867 Generate position-independent code (PIC) suitable for use in a shared
9868 library, if supported for the target machine. Such code accesses all
9869 constant addresses through a global offset table (GOT)@. The dynamic
9870 loader resolves the GOT entries when the program starts (the dynamic
9871 loader is not part of GCC; it is part of the operating system). If
9872 the GOT size for the linked executable exceeds a machine-specific
9873 maximum size, you get an error message from the linker indicating that
9874 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
9875 instead. (These maximums are 16k on the m88k, 8k on the Sparc, and 32k
9876 on the m68k and RS/6000. The 386 has no such limit.)
9878 Position-independent code requires special support, and therefore works
9879 only on certain machines. For the 386, GCC supports PIC for System V
9880 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
9881 position-independent.
9885 If supported for the target machine, emit position-independent code,
9886 suitable for dynamic linking and avoiding any limit on the size of the
9887 global offset table. This option makes a difference on the m68k, m88k,
9890 Position-independent code requires special support, and therefore works
9891 only on certain machines.
9893 @item -ffixed-@var{reg}
9895 Treat the register named @var{reg} as a fixed register; generated code
9896 should never refer to it (except perhaps as a stack pointer, frame
9897 pointer or in some other fixed role).
9899 @var{reg} must be the name of a register. The register names accepted
9900 are machine-specific and are defined in the @code{REGISTER_NAMES}
9901 macro in the machine description macro file.
9903 This flag does not have a negative form, because it specifies a
9906 @item -fcall-used-@var{reg}
9908 Treat the register named @var{reg} as an allocable register that is
9909 clobbered by function calls. It may be allocated for temporaries or
9910 variables that do not live across a call. Functions compiled this way
9911 will not save and restore the register @var{reg}.
9913 It is an error to used this flag with the frame pointer or stack pointer.
9914 Use of this flag for other registers that have fixed pervasive roles in
9915 the machine's execution model will produce disastrous results.
9917 This flag does not have a negative form, because it specifies a
9920 @item -fcall-saved-@var{reg}
9921 @opindex fcall-saved
9922 Treat the register named @var{reg} as an allocable register saved by
9923 functions. It may be allocated even for temporaries or variables that
9924 live across a call. Functions compiled this way will save and restore
9925 the register @var{reg} if they use it.
9927 It is an error to used this flag with the frame pointer or stack pointer.
9928 Use of this flag for other registers that have fixed pervasive roles in
9929 the machine's execution model will produce disastrous results.
9931 A different sort of disaster will result from the use of this flag for
9932 a register in which function values may be returned.
9934 This flag does not have a negative form, because it specifies a
9938 @opindex fpack-struct
9939 Pack all structure members together without holes. Usually you would
9940 not want to use this option, since it makes the code suboptimal, and
9941 the offsets of structure members won't agree with system libraries.
9943 @item -fcheck-memory-usage
9944 @opindex fcheck-memory-usage
9945 Generate extra code to check each memory access. GCC will generate
9946 code that is suitable for a detector of bad memory accesses such as
9949 Normally, you should compile all, or none, of your code with this option.
9951 If you do mix code compiled with and without this option,
9952 you must ensure that all code that has side effects
9953 and that is called by code compiled with this option
9954 is, itself, compiled with this option.
9955 If you do not, you might get erroneous messages from the detector.
9957 If you use functions from a library that have side-effects (such as
9958 @code{read}), you might not be able to recompile the library and
9959 specify this option. In that case, you can enable the
9960 @option{-fprefix-function-name} option, which requests GCC to encapsulate
9961 your code and make other functions look as if they were compiled with
9962 @option{-fcheck-memory-usage}. This is done by calling ``stubs'',
9963 which are provided by the detector. If you cannot find or build
9964 stubs for every function you call, you might have to specify
9965 @option{-fcheck-memory-usage} without @option{-fprefix-function-name}.
9967 If you specify this option, you can not use the @code{asm} or
9968 @code{__asm__} keywords in functions with memory checking enabled. GCC
9969 cannot understand what the @code{asm} statement may do, and therefore
9970 cannot generate the appropriate code, so it will reject it. However, if
9971 you specify the function attribute @code{no_check_memory_usage}
9972 (@pxref{Function Attributes}), GCC will disable memory checking within a
9973 function; you may use @code{asm} statements inside such functions. You
9974 may have an inline expansion of a non-checked function within a checked
9975 function; in that case GCC will not generate checks for the inlined
9976 function's memory accesses.
9978 If you move your @code{asm} statements to non-checked inline functions
9979 and they do access memory, you can add calls to the support code in your
9980 inline function, to indicate any reads, writes, or copies being done.
9981 These calls would be similar to those done in the stubs described above.
9983 @item -fprefix-function-name
9984 @opindex fprefix-function-name
9985 Request GCC to add a prefix to the symbols generated for function names.
9986 GCC adds a prefix to the names of functions defined as well as
9987 functions called. Code compiled with this option and code compiled
9988 without the option can't be linked together, unless stubs are used.
9990 If you compile the following code with @option{-fprefix-function-name}
9992 extern void bar (int);
10001 GCC will compile the code as if it was written:
10003 extern void prefix_bar (int);
10007 return prefix_bar (a + 5);
10010 This option is designed to be used with @option{-fcheck-memory-usage}.
10012 @item -finstrument-functions
10013 @opindex finstrument-functions
10014 Generate instrumentation calls for entry and exit to functions. Just
10015 after function entry and just before function exit, the following
10016 profiling functions will be called with the address of the current
10017 function and its call site. (On some platforms,
10018 @code{__builtin_return_address} does not work beyond the current
10019 function, so the call site information may not be available to the
10020 profiling functions otherwise.)
10023 void __cyg_profile_func_enter (void *this_fn,
10025 void __cyg_profile_func_exit (void *this_fn,
10029 The first argument is the address of the start of the current function,
10030 which may be looked up exactly in the symbol table.
10032 This instrumentation is also done for functions expanded inline in other
10033 functions. The profiling calls will indicate where, conceptually, the
10034 inline function is entered and exited. This means that addressable
10035 versions of such functions must be available. If all your uses of a
10036 function are expanded inline, this may mean an additional expansion of
10037 code size. If you use @samp{extern inline} in your C code, an
10038 addressable version of such functions must be provided. (This is
10039 normally the case anyways, but if you get lucky and the optimizer always
10040 expands the functions inline, you might have gotten away without
10041 providing static copies.)
10043 A function may be given the attribute @code{no_instrument_function}, in
10044 which case this instrumentation will not be done. This can be used, for
10045 example, for the profiling functions listed above, high-priority
10046 interrupt routines, and any functions from which the profiling functions
10047 cannot safely be called (perhaps signal handlers, if the profiling
10048 routines generate output or allocate memory).
10050 @item -fstack-check
10051 @opindex fstack-check
10052 Generate code to verify that you do not go beyond the boundary of the
10053 stack. You should specify this flag if you are running in an
10054 environment with multiple threads, but only rarely need to specify it in
10055 a single-threaded environment since stack overflow is automatically
10056 detected on nearly all systems if there is only one stack.
10058 Note that this switch does not actually cause checking to be done; the
10059 operating system must do that. The switch causes generation of code
10060 to ensure that the operating system sees the stack being extended.
10062 @item -fstack-limit-register=@var{reg}
10063 @itemx -fstack-limit-symbol=@var{sym}
10064 @itemx -fno-stack-limit
10065 @opindex fstack-limit-register
10066 @opindex fstack-limit-symbol
10067 @opindex fno-stack-limit
10068 Generate code to ensure that the stack does not grow beyond a certain value,
10069 either the value of a register or the address of a symbol. If the stack
10070 would grow beyond the value, a signal is raised. For most targets,
10071 the signal is raised before the stack overruns the boundary, so
10072 it is possible to catch the signal without taking special precautions.
10074 For instance, if the stack starts at absolute address @samp{0x80000000}
10075 and grows downwards, you can use the flags
10076 @option{-fstack-limit-symbol=__stack_limit} and
10077 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
10078 of 128KB@. Note that this may only work with the GNU linker.
10080 @cindex aliasing of parameters
10081 @cindex parameters, aliased
10082 @item -fargument-alias
10083 @itemx -fargument-noalias
10084 @itemx -fargument-noalias-global
10085 @opindex fargument-alias
10086 @opindex fargument-noalias
10087 @opindex fargument-noalias-global
10088 Specify the possible relationships among parameters and between
10089 parameters and global data.
10091 @option{-fargument-alias} specifies that arguments (parameters) may
10092 alias each other and may alias global storage.@*
10093 @option{-fargument-noalias} specifies that arguments do not alias
10094 each other, but may alias global storage.@*
10095 @option{-fargument-noalias-global} specifies that arguments do not
10096 alias each other and do not alias global storage.
10098 Each language will automatically use whatever option is required by
10099 the language standard. You should not need to use these options yourself.
10101 @item -fleading-underscore
10102 @opindex fleading-underscore
10103 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
10104 change the way C symbols are represented in the object file. One use
10105 is to help link with legacy assembly code.
10107 Be warned that you should know what you are doing when invoking this
10108 option, and that not all targets provide complete support for it.
10113 @node Environment Variables
10114 @section Environment Variables Affecting GCC
10115 @cindex environment variables
10117 @c man begin ENVIRONMENT
10119 This section describes several environment variables that affect how GCC
10120 operates. Some of them work by specifying directories or prefixes to use
10121 when searching for various kinds of files. Some are used to specify other
10122 aspects of the compilation environment.
10125 Note that you can also specify places to search using options such as
10126 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
10127 take precedence over places specified using environment variables, which
10128 in turn take precedence over those specified by the configuration of GCC@.
10132 Note that you can also specify places to search using options such as
10133 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
10134 take precedence over places specified using environment variables, which
10135 in turn take precedence over those specified by the configuration of GCC@.
10142 @c @itemx LC_COLLATE
10144 @c @itemx LC_MONETARY
10145 @c @itemx LC_NUMERIC
10150 @c @findex LC_COLLATE
10151 @findex LC_MESSAGES
10152 @c @findex LC_MONETARY
10153 @c @findex LC_NUMERIC
10157 These environment variables control the way that GCC uses
10158 localization information that allow GCC to work with different
10159 national conventions. GCC inspects the locale categories
10160 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
10161 so. These locale categories can be set to any value supported by your
10162 installation. A typical value is @samp{en_UK} for English in the United
10165 The @env{LC_CTYPE} environment variable specifies character
10166 classification. GCC uses it to determine the character boundaries in
10167 a string; this is needed for some multibyte encodings that contain quote
10168 and escape characters that would otherwise be interpreted as a string
10171 The @env{LC_MESSAGES} environment variable specifies the language to
10172 use in diagnostic messages.
10174 If the @env{LC_ALL} environment variable is set, it overrides the value
10175 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
10176 and @env{LC_MESSAGES} default to the value of the @env{LANG}
10177 environment variable. If none of these variables are set, GCC
10178 defaults to traditional C English behavior.
10182 If @env{TMPDIR} is set, it specifies the directory to use for temporary
10183 files. GCC uses temporary files to hold the output of one stage of
10184 compilation which is to be used as input to the next stage: for example,
10185 the output of the preprocessor, which is the input to the compiler
10188 @item GCC_EXEC_PREFIX
10189 @findex GCC_EXEC_PREFIX
10190 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
10191 names of the subprograms executed by the compiler. No slash is added
10192 when this prefix is combined with the name of a subprogram, but you can
10193 specify a prefix that ends with a slash if you wish.
10195 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
10196 an appropriate prefix to use based on the pathname it was invoked with.
10198 If GCC cannot find the subprogram using the specified prefix, it
10199 tries looking in the usual places for the subprogram.
10201 The default value of @env{GCC_EXEC_PREFIX} is
10202 @file{@var{prefix}/lib/gcc-lib/} where @var{prefix} is the value
10203 of @code{prefix} when you ran the @file{configure} script.
10205 Other prefixes specified with @option{-B} take precedence over this prefix.
10207 This prefix is also used for finding files such as @file{crt0.o} that are
10210 In addition, the prefix is used in an unusual way in finding the
10211 directories to search for header files. For each of the standard
10212 directories whose name normally begins with @samp{/usr/local/lib/gcc-lib}
10213 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
10214 replacing that beginning with the specified prefix to produce an
10215 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
10216 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
10217 These alternate directories are searched first; the standard directories
10220 @item COMPILER_PATH
10221 @findex COMPILER_PATH
10222 The value of @env{COMPILER_PATH} is a colon-separated list of
10223 directories, much like @env{PATH}. GCC tries the directories thus
10224 specified when searching for subprograms, if it can't find the
10225 subprograms using @env{GCC_EXEC_PREFIX}.
10228 @findex LIBRARY_PATH
10229 The value of @env{LIBRARY_PATH} is a colon-separated list of
10230 directories, much like @env{PATH}. When configured as a native compiler,
10231 GCC tries the directories thus specified when searching for special
10232 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
10233 using GCC also uses these directories when searching for ordinary
10234 libraries for the @option{-l} option (but directories specified with
10235 @option{-L} come first).
10237 @item C_INCLUDE_PATH
10238 @itemx CPLUS_INCLUDE_PATH
10239 @itemx OBJC_INCLUDE_PATH
10240 @findex C_INCLUDE_PATH
10241 @findex CPLUS_INCLUDE_PATH
10242 @findex OBJC_INCLUDE_PATH
10243 @c @itemx OBJCPLUS_INCLUDE_PATH
10244 These environment variables pertain to particular languages. Each
10245 variable's value is a colon-separated list of directories, much like
10246 @env{PATH}. When GCC searches for header files, it tries the
10247 directories listed in the variable for the language you are using, after
10248 the directories specified with @option{-I} but before the standard header
10251 @item DEPENDENCIES_OUTPUT
10252 @findex DEPENDENCIES_OUTPUT
10253 @cindex dependencies for make as output
10254 If this variable is set, its value specifies how to output dependencies
10255 for Make based on the header files processed by the compiler. This
10256 output looks much like the output from the @option{-M} option
10257 (@pxref{Preprocessor Options}), but it goes to a separate file, and is
10258 in addition to the usual results of compilation.
10260 The value of @env{DEPENDENCIES_OUTPUT} can be just a file name, in
10261 which case the Make rules are written to that file, guessing the target
10262 name from the source file name. Or the value can have the form
10263 @samp{@var{file} @var{target}}, in which case the rules are written to
10264 file @var{file} using @var{target} as the target name.
10268 @cindex locale definition
10269 This variable is used to pass locale information to the compiler. One way in
10270 which this information is used is to determine the character set to be used
10271 when character literals, string literals and comments are parsed in C and C++.
10272 When the compiler is configured to allow multibyte characters,
10273 the following values for @env{LANG} are recognized:
10277 Recognize JIS characters.
10279 Recognize SJIS characters.
10281 Recognize EUCJP characters.
10284 If @env{LANG} is not defined, or if it has some other value, then the
10285 compiler will use mblen and mbtowc as defined by the default locale to
10286 recognize and translate multibyte characters.
10291 @node Running Protoize
10292 @section Running Protoize
10294 The program @code{protoize} is an optional part of GCC@. You can use
10295 it to add prototypes to a program, thus converting the program to ISO
10296 C in one respect. The companion program @code{unprotoize} does the
10297 reverse: it removes argument types from any prototypes that are found.
10299 When you run these programs, you must specify a set of source files as
10300 command line arguments. The conversion programs start out by compiling
10301 these files to see what functions they define. The information gathered
10302 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
10304 After scanning comes actual conversion. The specified files are all
10305 eligible to be converted; any files they include (whether sources or
10306 just headers) are eligible as well.
10308 But not all the eligible files are converted. By default,
10309 @code{protoize} and @code{unprotoize} convert only source and header
10310 files in the current directory. You can specify additional directories
10311 whose files should be converted with the @option{-d @var{directory}}
10312 option. You can also specify particular files to exclude with the
10313 @option{-x @var{file}} option. A file is converted if it is eligible, its
10314 directory name matches one of the specified directory names, and its
10315 name within the directory has not been excluded.
10317 Basic conversion with @code{protoize} consists of rewriting most
10318 function definitions and function declarations to specify the types of
10319 the arguments. The only ones not rewritten are those for varargs
10322 @code{protoize} optionally inserts prototype declarations at the
10323 beginning of the source file, to make them available for any calls that
10324 precede the function's definition. Or it can insert prototype
10325 declarations with block scope in the blocks where undeclared functions
10328 Basic conversion with @code{unprotoize} consists of rewriting most
10329 function declarations to remove any argument types, and rewriting
10330 function definitions to the old-style pre-ISO form.
10332 Both conversion programs print a warning for any function declaration or
10333 definition that they can't convert. You can suppress these warnings
10336 The output from @code{protoize} or @code{unprotoize} replaces the
10337 original source file. The original file is renamed to a name ending
10338 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
10339 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
10340 for DOS) file already exists, then the source file is simply discarded.
10342 @code{protoize} and @code{unprotoize} both depend on GCC itself to
10343 scan the program and collect information about the functions it uses.
10344 So neither of these programs will work until GCC is installed.
10346 Here is a table of the options you can use with @code{protoize} and
10347 @code{unprotoize}. Each option works with both programs unless
10351 @item -B @var{directory}
10352 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
10353 usual directory (normally @file{/usr/local/lib}). This file contains
10354 prototype information about standard system functions. This option
10355 applies only to @code{protoize}.
10357 @item -c @var{compilation-options}
10358 Use @var{compilation-options} as the options when running @code{gcc} to
10359 produce the @samp{.X} files. The special option @option{-aux-info} is
10360 always passed in addition, to tell @code{gcc} to write a @samp{.X} file.
10362 Note that the compilation options must be given as a single argument to
10363 @code{protoize} or @code{unprotoize}. If you want to specify several
10364 @code{gcc} options, you must quote the entire set of compilation options
10365 to make them a single word in the shell.
10367 There are certain @code{gcc} arguments that you cannot use, because they
10368 would produce the wrong kind of output. These include @option{-g},
10369 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
10370 the @var{compilation-options}, they are ignored.
10373 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
10374 systems) instead of @samp{.c}. This is convenient if you are converting
10375 a C program to C++. This option applies only to @code{protoize}.
10378 Add explicit global declarations. This means inserting explicit
10379 declarations at the beginning of each source file for each function
10380 that is called in the file and was not declared. These declarations
10381 precede the first function definition that contains a call to an
10382 undeclared function. This option applies only to @code{protoize}.
10384 @item -i @var{string}
10385 Indent old-style parameter declarations with the string @var{string}.
10386 This option applies only to @code{protoize}.
10388 @code{unprotoize} converts prototyped function definitions to old-style
10389 function definitions, where the arguments are declared between the
10390 argument list and the initial @samp{@{}. By default, @code{unprotoize}
10391 uses five spaces as the indentation. If you want to indent with just
10392 one space instead, use @option{-i " "}.
10395 Keep the @samp{.X} files. Normally, they are deleted after conversion
10399 Add explicit local declarations. @code{protoize} with @option{-l} inserts
10400 a prototype declaration for each function in each block which calls the
10401 function without any declaration. This option applies only to
10405 Make no real changes. This mode just prints information about the conversions
10406 that would have been done without @option{-n}.
10409 Make no @samp{.save} files. The original files are simply deleted.
10410 Use this option with caution.
10412 @item -p @var{program}
10413 Use the program @var{program} as the compiler. Normally, the name
10414 @file{gcc} is used.
10417 Work quietly. Most warnings are suppressed.
10420 Print the version number, just like @option{-v} for @code{gcc}.
10423 If you need special compiler options to compile one of your program's
10424 source files, then you should generate that file's @samp{.X} file
10425 specially, by running @code{gcc} on that source file with the
10426 appropriate options and the option @option{-aux-info}. Then run
10427 @code{protoize} on the entire set of files. @code{protoize} will use
10428 the existing @samp{.X} file because it is newer than the source file.
10432 gcc -Dfoo=bar file1.c -aux-info file1.X
10437 You need to include the special files along with the rest in the
10438 @code{protoize} command, even though their @samp{.X} files already
10439 exist, because otherwise they won't get converted.
10441 @xref{Protoize Caveats}, for more information on how to use
10442 @code{protoize} successfully.