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 [@samp{-c}|@samp{-S}|@samp{-E}] [@samp{-std=}@var{standard}]
31 [@samp{-g}] [@samp{-pg}] [@samp{-O}@var{level}]
32 [@samp{-W}@var{warn}...] [@samp{-pedantic}]
33 [@samp{-I}@var{dir}...] [@samp{-L}@var{dir}...]
34 [@samp{-D}@var{macro}[=@var{defn}]...] [@samp{-U}@var{macro}]
35 [@samp{-f}@var{option}...] [@samp{-m}@var{machine-option}...]
36 [@samp{-o} @var{outfile}] @var{infile}...
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 @samp{-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: @samp{-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 @samp{-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, @samp{-fforce-mem},
103 @samp{-fstrength-reduce}, @samp{-Wformat} and so on. Most of
104 these have both positive and negative forms; the negative form of
105 @samp{-ffoo} would be @samp{-fno-foo}. This manual documents
106 only one of these two forms, whichever one is not the default.
111 * Option Summary:: Brief list of all options, without explanations.
112 * Overall Options:: Controlling the kind of output:
113 an executable, object files, assembler files,
114 or preprocessed source.
115 * Invoking G++:: Compiling C++ programs.
116 * C Dialect Options:: Controlling the variant of C language compiled.
117 * C++ Dialect Options:: Variations on C++.
118 * Language Independent Options:: Controlling how diagnostics should be
120 * Warning Options:: How picky should the compiler be?
121 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
122 * Optimize Options:: How much optimization?
123 * Preprocessor Options:: Controlling header files and macro definitions.
124 Also, getting dependency information for Make.
125 * Assembler Options:: Passing options to the assembler.
126 * Link Options:: Specifying libraries and so on.
127 * Directory Options:: Where to find header files and libraries.
128 Where to find the compiler executable files.
129 * Spec Files:: How to pass switches to sub-processes.
130 * Target Options:: Running a cross-compiler, or an old version of GCC.
131 * Submodel Options:: Specifying minor hardware or convention variations,
132 such as 68010 vs 68020.
133 * Code Gen Options:: Specifying conventions for function calls, data layout
135 * Environment Variables:: Env vars that affect GCC.
136 * Running Protoize:: Automatically adding or removing function prototypes.
142 @section Option Summary
144 Here is a summary of all the options, grouped by type. Explanations are
145 in the following sections.
148 @item Overall Options
149 @xref{Overall Options,,Options Controlling the Kind of Output}.
151 -c -S -E -o @var{file} -pipe -pass-exit-codes -x @var{language} @gol
152 -v --target-help --help}
154 @item C Language Options
155 @xref{C Dialect Options,,Options Controlling C Dialect}.
157 -ansi -std=@var{standard} -fno-asm -fno-builtin @gol
158 -fhosted -ffreestanding @gol
159 -trigraphs -traditional -traditional-cpp @gol
160 -fallow-single-precision -fcond-mismatch @gol
161 -fsigned-bitfields -fsigned-char @gol
162 -funsigned-bitfields -funsigned-char @gol
163 -fwritable-strings -fshort-wchar}
165 @item C++ Language Options
166 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
168 -fno-access-control -fcheck-new -fconserve-space @gol
169 -fdollars-in-identifiers -fno-elide-constructors @gol
170 -fno-enforce-eh-specs -fexternal-templates @gol
171 -falt-external-templates @gol
172 -ffor-scope -fno-for-scope -fno-gnu-keywords -fhonor-std @gol
173 -fhuge-objects -fno-implicit-templates @gol
174 -fno-implicit-inline-templates @gol
175 -fno-implement-inlines -fms-extensions @gol
176 -fno-operator-names @gol
177 -fno-optional-diags -fpermissive @gol
178 -frepo -fno-rtti -ftemplate-depth-@var{n} @gol
179 -fuse-cxa-atexit -fvtable-thunks -nostdinc++ @gol
180 -fno-default-inline -Wctor-dtor-privacy @gol
181 -Wnon-virtual-dtor -Wreorder @gol
182 -Weffc++ -Wno-deprecated @gol
183 -Wno-non-template-friend -Wold-style-cast @gol
184 -Woverloaded-virtual -Wno-pmf-conversions @gol
185 -Wsign-promo -Wsynth}
187 @item Language Independent Options
188 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
190 -fmessage-length=@var{n} @gol
191 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]}}
193 @item Warning Options
194 @xref{Warning Options,,Options to Request or Suppress Warnings}.
196 -fsyntax-only -pedantic -pedantic-errors @gol
197 -w -W -Wall -Waggregate-return @gol
198 -Wcast-align -Wcast-qual -Wchar-subscripts -Wcomment @gol
199 -Wconversion -Wdisabled-optimization -Werror @gol
200 -Wfloat-equal -Wformat -Wformat=2 @gol
201 -Wformat-nonliteral -Wformat-security @gol
202 -Wid-clash-@var{len} -Wimplicit -Wimplicit-int @gol
203 -Wimplicit-function-declaration @gol
204 -Werror-implicit-function-declaration @gol
205 -Wimport -Winline @gol
206 -Wlarger-than-@var{len} -Wlong-long @gol
207 -Wmain -Wmissing-declarations @gol
208 -Wmissing-format-attribute -Wmissing-noreturn @gol
209 -Wmultichar -Wno-format-extra-args -Wno-format-y2k @gol
210 -Wno-import -Wpacked -Wpadded @gol
211 -Wparentheses -Wpointer-arith -Wredundant-decls @gol
212 -Wreturn-type -Wsequence-point -Wshadow @gol
213 -Wsign-compare -Wswitch -Wsystem-headers @gol
214 -Wtrigraphs -Wundef -Wuninitialized @gol
215 -Wunknown-pragmas -Wunreachable-code @gol
216 -Wunused -Wunused-function -Wunused-label -Wunused-parameter @gol
217 -Wunused-value -Wunused-variable -Wwrite-strings}
219 @item C-only Warning Options
221 -Wbad-function-cast -Wmissing-prototypes -Wnested-externs @gol
222 -Wstrict-prototypes -Wtraditional}
224 @item Debugging Options
225 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
227 -a -ax -d@var{letters} -fdump-unnumbered -fdump-translation-unit=@var{file} @gol
228 -fdump-class-layout=@var{file} -fpretend-float -fprofile-arcs -ftest-coverage @gol
229 -g -g@var{level} -gcoff -gdwarf -gdwarf-1 -gdwarf-1+ -gdwarf-2 @gol
230 -ggdb -gstabs -gstabs+ -gxcoff -gxcoff+ @gol
231 -p -pg -print-file-name=@var{library} -print-libgcc-file-name @gol
232 -print-prog-name=@var{program} -print-search-dirs -Q @gol
235 @item Optimization Options
236 @xref{Optimize Options,,Options that Control Optimization}.
238 -falign-functions=@var{n} -falign-jumps=@var{n} @gol
239 -falign-labels=@var{n} -falign-loops=@var{n} @gol
240 -fbranch-probabilities -fcaller-saves @gol
241 -fcse-follow-jumps -fcse-skip-blocks -fdata-sections -fdce @gol
242 -fdelayed-branch -fdelete-null-pointer-checks @gol
243 -fexpensive-optimizations -ffast-math -ffloat-store @gol
244 -fforce-addr -fforce-mem -ffunction-sections -fgcse -fgcse-lm -fgcse-sm @gol
245 -finline-functions -finline-limit=@var{n} -fkeep-inline-functions @gol
246 -fkeep-static-consts -fmove-all-movables @gol
247 -fno-default-inline -fno-defer-pop @gol
248 -fno-function-cse -fno-guess-branch-probability @gol
249 -fno-inline -fno-math-errno -fno-peephole @gol
250 -funsafe-math-optimizations -fno-trapping-math @gol
251 -fomit-frame-pointer -foptimize-register-move @gol
252 -foptimize-sibling-calls -freduce-all-givs @gol
253 -fregmove -frename-registers @gol
254 -frerun-cse-after-loop -frerun-loop-opt @gol
255 -fschedule-insns -fschedule-insns2 @gol
256 -fsingle-precision-constant -fssa @gol
257 -fstrength-reduce -fstrict-aliasing -fthread-jumps -ftrapv @gol
258 -funroll-all-loops -funroll-loops @gol
259 --param @var{name}=@var{value}
260 -O -O0 -O1 -O2 -O3 -Os}
262 @item Preprocessor Options
263 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
265 -$ -A@var{question}=@var{answer} -A-@var{question}@r{[}=@var{answer}@r{]} @gol
266 -C -dD -dI -dM -dN @gol
267 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
268 -idirafter @var{dir} @gol
269 -include @var{file} -imacros @var{file} @gol
270 -iprefix @var{file} -iwithprefix @var{dir} @gol
271 -iwithprefixbefore @var{dir} -isystem @var{dir} -isystem-c++ @var{dir} @gol
272 -M -MM -MF -MG -MP -MQ -MT -nostdinc -P -remap @gol
273 -trigraphs -undef -U@var{macro} -Wp\,@var{option}}
275 @item Assembler Option
276 @xref{Assembler Options,,Passing Options to the Assembler}.
281 @xref{Link Options,,Options for Linking}.
283 @var{object-file-name} -l@var{library} @gol
284 -nostartfiles -nodefaultlibs -nostdlib @gol
285 -s -static -static-libgcc -shared -shared-libgcc -symbolic @gol
286 -Wl\,@var{option} -Xlinker @var{option} @gol
289 @item Directory Options
290 @xref{Directory Options,,Options for Directory Search}.
292 -B@var{prefix} -I@var{dir} -I- -L@var{dir} -specs=@var{file}}
295 @c I wrote this xref this way to avoid overfull hbox. -- rms
296 @xref{Target Options}.
298 -b @var{machine} -V @var{version}}
300 @item Machine Dependent Options
301 @xref{Submodel Options,,Hardware Models and Configurations}.
302 @emph{M680x0 Options}
304 -m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
305 -m68060 -mcpu32 -m5200 -m68881 -mbitfield -mc68000 -mc68020 @gol
306 -mfpa -mnobitfield -mrtd -mshort -msoft-float -mpcrel @gol
307 -malign-int -mstrict-align}
309 @emph{M68hc1x Options}
311 -m6811 -m6812 -m68hc11 -m68hc12 @gol
312 -mauto-incdec -mshort -msoft-reg-count=@var{count}}
320 -mcpu=@var{cpu type} @gol
321 -mtune=@var{cpu type} @gol
322 -mcmodel=@var{code model} @gol
324 -mapp-regs -mbroken-saverestore -mcypress @gol
325 -mepilogue -mfaster-structs -mflat @gol
326 -mfpu -mhard-float -mhard-quad-float @gol
327 -mimpure-text -mlive-g0 -mno-app-regs @gol
328 -mno-epilogue -mno-faster-structs -mno-flat -mno-fpu @gol
329 -mno-impure-text -mno-stack-bias -mno-unaligned-doubles @gol
330 -msoft-float -msoft-quad-float -msparclite -mstack-bias @gol
331 -msupersparc -munaligned-doubles -mv8}
333 @emph{Convex Options}
335 -mc1 -mc2 -mc32 -mc34 -mc38 @gol
336 -margcount -mnoargcount @gol
337 -mlong32 -mlong64 @gol
338 -mvolatile-cache -mvolatile-nocache}
340 @emph{AMD29K Options}
342 -m29000 -m29050 -mbw -mnbw -mdw -mndw @gol
343 -mlarge -mnormal -msmall @gol
344 -mkernel-registers -mno-reuse-arg-regs @gol
345 -mno-stack-check -mno-storem-bug @gol
346 -mreuse-arg-regs -msoft-float -mstack-check @gol
347 -mstorem-bug -muser-registers}
351 -mapcs-frame -mno-apcs-frame @gol
352 -mapcs-26 -mapcs-32 @gol
353 -mapcs-stack-check -mno-apcs-stack-check @gol
354 -mapcs-float -mno-apcs-float @gol
355 -mapcs-reentrant -mno-apcs-reentrant @gol
356 -msched-prolog -mno-sched-prolog @gol
357 -mlittle-endian -mbig-endian -mwords-little-endian @gol
358 -malignment-traps -mno-alignment-traps @gol
359 -msoft-float -mhard-float -mfpe @gol
360 -mthumb-interwork -mno-thumb-interwork @gol
361 -mcpu= -march= -mfpe= @gol
362 -mstructure-size-boundary= @gol
363 -mbsd -mxopen -mno-symrename @gol
364 -mabort-on-noreturn @gol
365 -mlong-calls -mno-long-calls @gol
366 -mnop-fun-dllimport -mno-nop-fun-dllimport @gol
367 -msingle-pic-base -mno-single-pic-base @gol
372 -mtpcs-frame -mno-tpcs-frame @gol
373 -mtpcs-leaf-frame -mno-tpcs-leaf-frame @gol
374 -mlittle-endian -mbig-endian @gol
375 -mthumb-interwork -mno-thumb-interwork @gol
376 -mstructure-size-boundary= @gol
377 -mnop-fun-dllimport -mno-nop-fun-dllimport @gol
378 -mcallee-super-interworking -mno-callee-super-interworking @gol
379 -mcaller-super-interworking -mno-caller-super-interworking @gol
380 -msingle-pic-base -mno-single-pic-base @gol
383 @emph{MN10200 Options}
387 @emph{MN10300 Options}
395 @emph{M32R/D Options}
397 -mcode-model=@var{model type} -msdata=@var{sdata type} @gol
402 -m88000 -m88100 -m88110 -mbig-pic @gol
403 -mcheck-zero-division -mhandle-large-shift @gol
404 -midentify-revision -mno-check-zero-division @gol
405 -mno-ocs-debug-info -mno-ocs-frame-position @gol
406 -mno-optimize-arg-area -mno-serialize-volatile @gol
407 -mno-underscores -mocs-debug-info @gol
408 -mocs-frame-position -moptimize-arg-area @gol
409 -mserialize-volatile -mshort-data-@var{num} -msvr3 @gol
410 -msvr4 -mtrap-large-shift -muse-div-instruction @gol
411 -mversion-03.00 -mwarn-passed-structs}
413 @emph{RS/6000 and PowerPC Options}
415 -mcpu=@var{cpu type} @gol
416 -mtune=@var{cpu type} @gol
417 -mpower -mno-power -mpower2 -mno-power2 @gol
418 -mpowerpc -mpowerpc64 -mno-powerpc @gol
419 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
420 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
421 -mnew-mnemonics -mold-mnemonics @gol
422 -mfull-toc -mminimal-toc -mno-fop-in-toc -mno-sum-in-toc @gol
423 -m64 -m32 -mxl-call -mno-xl-call -mthreads -mpe @gol
424 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
425 -mstring -mno-string -mupdate -mno-update @gol
426 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
427 -mstrict-align -mno-strict-align -mrelocatable @gol
428 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
429 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
430 -mcall-aix -mcall-sysv -mprototype -mno-prototype @gol
431 -msim -mmvme -mads -myellowknife -memb -msdata @gol
432 -msdata=@var{opt} -mvxworks -G @var{num}}
436 -mcall-lib-mul -mfp-arg-in-fpregs -mfp-arg-in-gregs @gol
437 -mfull-fp-blocks -mhc-struct-return -min-line-mul @gol
438 -mminimum-fp-blocks -mnohc-struct-return}
442 -mabicalls -mcpu=@var{cpu type}
443 -membedded-data -muninit-const-in-rodata @gol
444 -membedded-pic -mfp32 -mfp64 -mgas -mgp32 -mgp64 @gol
445 -mgpopt -mhalf-pic -mhard-float -mint64 -mips1 @gol
446 -mips2 -mips3 -mips4 -mlong64 -mlong32 -mlong-calls -mmemcpy @gol
447 -mmips-as -mmips-tfile -mno-abicalls @gol
448 -mno-embedded-data -mno-uninit-const-in-rodata -mno-embedded-pic @gol
449 -mno-gpopt -mno-long-calls @gol
450 -mno-memcpy -mno-mips-tfile -mno-rnames -mno-stats @gol
451 -mrnames -msoft-float @gol
452 -m4650 -msingle-float -mmad @gol
453 -mstats -EL -EB -G @var{num} -nocpp @gol
454 -mabi=32 -mabi=n32 -mabi=64 -mabi=eabi @gol
459 -mcpu=@var{cpu type} -march=@var{cpu type} @gol
460 -mintel-syntax -mieee-fp -mno-fancy-math-387 @gol
461 -mno-fp-ret-in-387 -msoft-float -msvr3-shlib @gol
462 -mno-wide-multiply -mrtd -malign-double @gol
463 -malign-jumps=@var{num} -malign-loops=@var{num} @gol
464 -malign-functions=@var{num} -mpreferred-stack-boundary=@var{num} @gol
465 -mthreads -mno-align-stringops -minline-all-stringops @gol
466 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
467 -m96bit-long-double -mregparm=@var{num}}
471 -march=@var{architecture type} @gol
472 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
473 -mfast-indirect-calls -mgas -mjump-in-delay @gol
474 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
475 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
476 -mno-jump-in-delay -mno-long-load-store @gol
477 -mno-portable-runtime -mno-soft-float @gol
478 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
479 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
480 -mschedule=@var{cpu type} -mspace-regs}
482 @emph{Intel 960 Options}
484 -m@var{cpu type} -masm-compat -mclean-linkage @gol
485 -mcode-align -mcomplex-addr -mleaf-procedures @gol
486 -mic-compat -mic2.0-compat -mic3.0-compat @gol
487 -mintel-asm -mno-clean-linkage -mno-code-align @gol
488 -mno-complex-addr -mno-leaf-procedures @gol
489 -mno-old-align -mno-strict-align -mno-tail-call @gol
490 -mnumerics -mold-align -msoft-float -mstrict-align @gol
493 @emph{DEC Alpha Options}
495 -mfp-regs -mno-fp-regs -mno-soft-float -msoft-float @gol
496 -malpha-as -mgas @gol
497 -mieee -mieee-with-inexact -mieee-conformant @gol
498 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
499 -mtrap-precision=@var{mode} -mbuild-constants @gol
500 -mcpu=@var{cpu type} @gol
501 -mbwx -mno-bwx -mcix -mno-cix -mmax -mno-max @gol
502 -mmemory-latency=@var{time}}
504 @emph{Clipper Options}
508 @emph{H8/300 Options}
510 -mrelax -mh -ms -mint32 -malign-300}
514 -m1 -m2 -m3 -m3e @gol
515 -m4-nofpu -m4-single-only -m4-single -m4 @gol
516 -mb -ml -mdalign -mrelax @gol
517 -mbigtable -mfmovd -mhitachi -mnomacsave @gol
518 -misize -mpadstruct -mspace @gol
522 @emph{System V Options}
524 -Qy -Qn -YP\,@var{paths} -Ym\,@var{dir}}
529 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text section} @gol
530 -mdata=@var{data section} -mrodata=@var{readonly data section}}
532 @emph{TMS320C3x/C4x Options}
534 -mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm @gol
535 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload @gol
536 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned @gol
537 -mparallel-insns -mparallel-mpy -mpreserve-float}
541 -mlong-calls -mno-long-calls -mep -mno-ep @gol
542 -mprolog-function -mno-prolog-function -mspace @gol
543 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
548 -m32032 -m32332 -m32532 -m32081 -m32381 -mmult-add -mnomult-add @gol
549 -msoft-float -mrtd -mnortd -mregparam -mnoregparam -msb -mnosb @gol
550 -mbitfield -mnobitfield -mhimem -mnohimem}
554 -mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts @gol
555 -mcall-prologues -mno-tablejump -mtiny-stack}
559 -mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
560 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
561 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
562 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
563 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
567 -mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
568 -mvolatile-asm-stop -mb-step -mregister-names -mno-sdata @gol
569 -mconstant-gp -mauto-pic -minline-divide-min-latency @gol
570 -minline-divide-max-throughput -mno-dwarf2-asm @gol
571 -mfixed-range=@var{register range}}
573 @item Code Generation Options
574 @xref{Code Gen Options,,Options for Code Generation Conventions}.
576 -fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
577 -fexceptions -funwind-tables -ffixed-@var{reg} @gol
578 -finhibit-size-directive -finstrument-functions @gol
579 -fcheck-memory-usage -fprefix-function-name @gol
580 -fno-common -fno-ident -fno-gnu-linker @gol
581 -fpcc-struct-return -fpic -fPIC @gol
582 -freg-struct-return -fshared-data -fshort-enums @gol
583 -fshort-double -fvolatile -fvolatile-global -fvolatile-static @gol
584 -fverbose-asm -fpack-struct -fstack-check @gol
585 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
586 -fargument-alias -fargument-noalias @gol
587 -fargument-noalias-global @gol
588 -fleading-underscore}
592 * Overall Options:: Controlling the kind of output:
593 an executable, object files, assembler files,
594 or preprocessed source.
595 * C Dialect Options:: Controlling the variant of C language compiled.
596 * C++ Dialect Options:: Variations on C++.
597 * Language Independent Options:: Controlling how diagnostics should be
599 * Warning Options:: How picky should the compiler be?
600 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
601 * Optimize Options:: How much optimization?
602 * Preprocessor Options:: Controlling header files and macro definitions.
603 Also, getting dependency information for Make.
604 * Assembler Options:: Passing options to the assembler.
605 * Link Options:: Specifying libraries and so on.
606 * Directory Options:: Where to find header files and libraries.
607 Where to find the compiler executable files.
608 * Spec Files:: How to pass switches to sub-processes.
609 * Target Options:: Running a cross-compiler, or an old version of GCC.
612 @node Overall Options
613 @section Options Controlling the Kind of Output
615 Compilation can involve up to four stages: preprocessing, compilation
616 proper, assembly and linking, always in that order. The first three
617 stages apply to an individual source file, and end by producing an
618 object file; linking combines all the object files (those newly
619 compiled, and those specified as input) into an executable file.
621 @cindex file name suffix
622 For any given input file, the file name suffix determines what kind of
627 C source code which must be preprocessed.
630 C source code which should not be preprocessed.
633 C++ source code which should not be preprocessed.
636 Objective-C source code. Note that you must link with the library
637 @file{libobjc.a} to make an Objective-C program work.
640 Objective-C source code which should not be preprocessed.
643 C header file (not to be compiled or linked).
647 @itemx @var{file}.cxx
648 @itemx @var{file}.cpp
649 @itemx @var{file}.c++
651 C++ source code which must be preprocessed. Note that in @samp{.cxx},
652 the last two letters must both be literally @samp{x}. Likewise,
653 @samp{.C} refers to a literal capital C.
656 @itemx @var{file}.for
657 @itemx @var{file}.FOR
658 Fortran source code which should not be preprocessed.
661 @itemx @var{file}.fpp
662 @itemx @var{file}.FPP
663 Fortran source code which must be preprocessed (with the traditional
667 Fortran source code which must be preprocessed with a RATFOR
668 preprocessor (not included with GCC).
670 @xref{Overall Options,,Options Controlling the Kind of Output, g77,
671 Using and Porting GNU Fortran}, for more details of the handling of
674 @c FIXME: Descriptions of Java file types.
680 @c GCC also knows about some suffixes for languages not yet included:
690 @itemx @var{file}.chi
691 CHILL source code (preprocessed with the traditional preprocessor).
697 Assembler code which must be preprocessed.
700 An object file to be fed straight into linking.
701 Any file name with no recognized suffix is treated this way.
704 You can specify the input language explicitly with the @samp{-x} option:
707 @item -x @var{language}
708 Specify explicitly the @var{language} for the following input files
709 (rather than letting the compiler choose a default based on the file
710 name suffix). This option applies to all following input files until
711 the next @samp{-x} option. Possible values for @var{language} are:
713 c c-header cpp-output
715 objective-c objc-cpp-output
716 assembler assembler-with-cpp
717 f77 f77-cpp-input ratfor
720 @c Also f77-version, for internal use only.
723 Turn off any specification of a language, so that subsequent files are
724 handled according to their file name suffixes (as they are if @samp{-x}
725 has not been used at all).
727 @item -pass-exit-codes
728 Normally the @command{gcc} program will exit with the code of 1 if any
729 phase of the compiler returns a non-success return code. If you specify
730 @samp{-pass-exit-codes}, the @command{gcc} program will instead return with
731 numerically highest error produced by any phase that returned an error
735 If you only want some of the stages of compilation, you can use
736 @samp{-x} (or filename suffixes) to tell @command{gcc} where to start, and
737 one of the options @samp{-c}, @samp{-S}, or @samp{-E} to say where
738 @command{gcc} is to stop. Note that some combinations (for example,
739 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
743 Compile or assemble the source files, but do not link. The linking
744 stage simply is not done. The ultimate output is in the form of an
745 object file for each source file.
747 By default, the object file name for a source file is made by replacing
748 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
750 Unrecognized input files, not requiring compilation or assembly, are
754 Stop after the stage of compilation proper; do not assemble. The output
755 is in the form of an assembler code file for each non-assembler input
758 By default, the assembler file name for a source file is made by
759 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
761 Input files that don't require compilation are ignored.
764 Stop after the preprocessing stage; do not run the compiler proper. The
765 output is in the form of preprocessed source code, which is sent to the
768 Input files which don't require preprocessing are ignored.
770 @cindex output file option
772 Place output in file @var{file}. This applies regardless to whatever
773 sort of output is being produced, whether it be an executable file,
774 an object file, an assembler file or preprocessed C code.
776 Since only one output file can be specified, it does not make sense to
777 use @samp{-o} when compiling more than one input file, unless you are
778 producing an executable file as output.
780 If @samp{-o} is not specified, the default is to put an executable file
781 in @file{a.out}, the object file for @file{@var{source}.@var{suffix}} in
782 @file{@var{source}.o}, its assembler file in @file{@var{source}.s}, and
783 all preprocessed C source on standard output.@refill
786 Print (on standard error output) the commands executed to run the stages
787 of compilation. Also print the version number of the compiler driver
788 program and of the preprocessor and the compiler proper.
791 Use pipes rather than temporary files for communication between the
792 various stages of compilation. This fails to work on some systems where
793 the assembler is unable to read from a pipe; but the GNU assembler has
797 Print (on the standard output) a description of the command line options
798 understood by @command{gcc}. If the @option{-v} option is also specified
799 then @option{--help} will also be passed on to the various processes
800 invoked by @command{gcc}, so that they can display the command line options
801 they accept. If the @option{-W} option is also specified then command
802 line options which have no documentation associated with them will also
806 Print (on the standard output) a description of target specific command
807 line options for each tool.
811 @section Compiling C++ Programs
813 @cindex suffixes for C++ source
814 @cindex C++ source file suffixes
815 C++ source files conventionally use one of the suffixes @samp{.C},
816 @samp{.cc}, @samp{.cpp}, @samp{.c++}, @samp{.cp}, or @samp{.cxx};
817 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
818 files with these names and compiles them as C++ programs even if you
819 call the compiler the same way as for compiling C programs (usually with
820 the name @command{gcc}).
824 However, C++ programs often require class libraries as well as a
825 compiler that understands the C++ language---and under some
826 circumstances, you might want to compile programs from standard input,
827 or otherwise without a suffix that flags them as C++ programs.
828 @command{g++} is a program that calls GCC with the default language
829 set to C++, and automatically specifies linking against the C++
830 library. On many systems, @command{g++} is also
831 installed with the name @command{c++}.
833 @cindex invoking @command{g++}
834 When you compile C++ programs, you may specify many of the same
835 command-line options that you use for compiling programs in any
836 language; or command-line options meaningful for C and related
837 languages; or options that are meaningful only for C++ programs.
838 @xref{C Dialect Options,,Options Controlling C Dialect}, for
839 explanations of options for languages related to C.
840 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
841 explanations of options that are meaningful only for C++ programs.
843 @node C Dialect Options
844 @section Options Controlling C Dialect
845 @cindex dialect options
846 @cindex language dialect options
847 @cindex options, dialect
849 The following options control the dialect of C (or languages derived
850 from C, such as C++ and Objective C) that the compiler accepts:
856 In C mode, support all ISO C89 programs. In C++ mode,
857 remove GNU extensions that conflict with ISO C++.
859 This turns off certain features of GCC that are incompatible with ISO
860 C (when compiling C code), or of standard C++ (when compiling C++ code),
861 such as the @code{asm} and @code{typeof} keywords, and
862 predefined macros such as @code{unix} and @code{vax} that identify the
863 type of system you are using. It also enables the undesirable and
864 rarely used ISO trigraph feature. For the C compiler,
865 it disables recognition of C++ style @samp{//} comments as well as
866 the @code{inline} keyword.
868 The alternate keywords @code{__asm__}, @code{__extension__},
869 @code{__inline__} and @code{__typeof__} continue to work despite
870 @samp{-ansi}. You would not want to use them in an ISO C program, of
871 course, but it is useful to put them in header files that might be included
872 in compilations done with @samp{-ansi}. Alternate predefined macros
873 such as @code{__unix__} and @code{__vax__} are also available, with or
874 without @samp{-ansi}.
876 The @samp{-ansi} option does not cause non-ISO programs to be
877 rejected gratuitously. For that, @samp{-pedantic} is required in
878 addition to @samp{-ansi}. @xref{Warning Options}.
880 The macro @code{__STRICT_ANSI__} is predefined when the @samp{-ansi}
881 option is used. Some header files may notice this macro and refrain
882 from declaring certain functions or defining certain macros that the
883 ISO standard doesn't call for; this is to avoid interfering with any
884 programs that might use these names for other things.
886 Functions which would normally be builtin but do not have semantics
887 defined by ISO C (such as @code{alloca} and @code{ffs}) are not builtin
888 functions with @samp{-ansi} is used. @xref{Other Builtins,,Other
889 built-in functions provided by GNU CC}, for details of the functions
893 Determine the language standard. A value for this option must be provided;
898 Same as @option{-ansi}
901 ISO C as modified in amend. 1
904 ISO C99. Note that this standard is not yet fully supported; see
905 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information.
908 same as @option{-std=iso9899:1990}
911 same as @option{-std=iso9899:1999}
914 default, iso9899:1990 + gnu extensions
917 iso9899:1999 + gnu extensions
920 same as @option{-std=iso9899:1999}, deprecated
923 same as @option{-std=iso9899:1999}, deprecated
926 same as @option{-std=gnu99}, deprecated
930 Even when this option is not specified, you can still use some of the
931 features of newer standards in so far as they do not conflict with
932 previous C standards. For example, you may use @code{__restrict__} even
933 when @option{-std=c99} is not specified.
935 The @option{-std} options specifying some version of ISO C have the same
936 effects as @option{-ansi}, except that features that were not in ISO C89
937 but are in the specified version (for example, @samp{//} comments and
938 the @code{inline} keyword in ISO C99) are not disabled.
940 @xref{Standards,,Language Standards Supported by GCC}, for details of
941 these standard versions.
944 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
945 keyword, so that code can use these words as identifiers. You can use
946 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
947 instead. @samp{-ansi} implies @samp{-fno-asm}.
949 In C++, this switch only affects the @code{typeof} keyword, since
950 @code{asm} and @code{inline} are standard keywords. You may want to
951 use the @samp{-fno-gnu-keywords} flag instead, which has the same
952 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
953 switch only affects the @code{asm} and @code{typeof} keywords, since
954 @code{inline} is a standard keyword in ISO C99.
957 @cindex builtin functions
958 Don't recognize builtin functions that do not begin with
959 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
960 functions provided by GNU CC}, for details of the functions affected,
961 including those which are not builtin functions when @option{-ansi} or
962 @option{-std} options for strict ISO C conformance are used because they
963 do not have an ISO standard meaning.
965 GCC normally generates special code to handle certain builtin functions
966 more efficiently; for instance, calls to @code{alloca} may become single
967 instructions that adjust the stack directly, and calls to @code{memcpy}
968 may become inline copy loops. The resulting code is often both smaller
969 and faster, but since the function calls no longer appear as such, you
970 cannot set a breakpoint on those calls, nor can you change the behavior
971 of the functions by linking with a different library.
974 @cindex hosted environment
976 Assert that compilation takes place in a hosted environment. This implies
977 @samp{-fbuiltin}. A hosted environment is one in which the
978 entire standard library is available, and in which @code{main} has a return
979 type of @code{int}. Examples are nearly everything except a kernel.
980 This is equivalent to @samp{-fno-freestanding}.
983 @cindex hosted environment
985 Assert that compilation takes place in a freestanding environment. This
986 implies @samp{-fno-builtin}. A freestanding environment
987 is one in which the standard library may not exist, and program startup may
988 not necessarily be at @code{main}. The most obvious example is an OS kernel.
989 This is equivalent to @samp{-fno-hosted}.
991 @xref{Standards,,Language Standards Supported by GCC}, for details of
992 freestanding and hosted environments.
995 Support ISO C trigraphs. You don't want to know about this
996 brain-damage. The @option{-ansi} option (and @option{-std} options for
997 strict ISO C conformance) implies @option{-trigraphs}.
999 @cindex traditional C language
1000 @cindex C language, traditional
1002 Attempt to support some aspects of traditional C compilers.
1007 All @code{extern} declarations take effect globally even if they
1008 are written inside of a function definition. This includes implicit
1009 declarations of functions.
1012 The newer keywords @code{typeof}, @code{inline}, @code{signed}, @code{const}
1013 and @code{volatile} are not recognized. (You can still use the
1014 alternative keywords such as @code{__typeof__}, @code{__inline__}, and
1018 Comparisons between pointers and integers are always allowed.
1021 Integer types @code{unsigned short} and @code{unsigned char} promote
1022 to @code{unsigned int}.
1025 Out-of-range floating point literals are not an error.
1028 Certain constructs which ISO regards as a single invalid preprocessing
1029 number, such as @samp{0xe-0xd}, are treated as expressions instead.
1032 String ``constants'' are not necessarily constant; they are stored in
1033 writable space, and identical looking constants are allocated
1034 separately. (This is the same as the effect of
1035 @samp{-fwritable-strings}.)
1037 @cindex @code{longjmp} and automatic variables
1039 All automatic variables not declared @code{register} are preserved by
1040 @code{longjmp}. Ordinarily, GNU C follows ISO C: automatic variables
1041 not declared @code{volatile} may be clobbered.
1046 @cindex escape sequences, traditional
1047 The character escape sequences @samp{\x} and @samp{\a} evaluate as the
1048 literal characters @samp{x} and @samp{a} respectively. Without
1049 @w{@samp{-traditional}}, @samp{\x} is a prefix for the hexadecimal
1050 representation of a character, and @samp{\a} produces a bell.
1053 You may wish to use @samp{-fno-builtin} as well as @samp{-traditional}
1054 if your program uses names that are normally GNU C builtin functions for
1055 other purposes of its own.
1057 You cannot use @samp{-traditional} if you include any header files that
1058 rely on ISO C features. Some vendors are starting to ship systems with
1059 ISO C header files and you cannot use @samp{-traditional} on such
1060 systems to compile files that include any system headers.
1062 The @samp{-traditional} option also enables @samp{-traditional-cpp},
1063 which is described next.
1065 @item -traditional-cpp
1066 Attempt to support some aspects of traditional C preprocessors.
1071 Comments convert to nothing at all, rather than to a space. This allows
1072 traditional token concatenation.
1075 In a preprocessing directive, the @samp{#} symbol must appear as the first
1076 character of a line.
1079 Macro arguments are recognized within string constants in a macro
1080 definition (and their values are stringified, though without additional
1081 quote marks, when they appear in such a context). The preprocessor
1082 always considers a string constant to end at a newline.
1085 @cindex detecting @w{@samp{-traditional}}
1086 The predefined macro @code{__STDC__} is not defined when you use
1087 @samp{-traditional}, but @code{__GNUC__} is (since the GNU extensions
1088 which @code{__GNUC__} indicates are not affected by
1089 @samp{-traditional}). If you need to write header files that work
1090 differently depending on whether @samp{-traditional} is in use, by
1091 testing both of these predefined macros you can distinguish four
1092 situations: GNU C, traditional GNU C, other ISO C compilers, and other
1093 old C compilers. The predefined macro @code{__STDC_VERSION__} is also
1094 not defined when you use @samp{-traditional}. @xref{Standard
1095 Predefined,,Standard Predefined Macros,cpp.info,The C Preprocessor},
1096 for more discussion of these and other predefined macros.
1099 @cindex string constants vs newline
1100 @cindex newline vs string constants
1101 The preprocessor considers a string constant to end at a newline (unless
1102 the newline is escaped with @samp{\}). (Without @w{@samp{-traditional}},
1103 string constants can contain the newline character as typed.)
1106 @item -fcond-mismatch
1107 Allow conditional expressions with mismatched types in the second and
1108 third arguments. The value of such an expression is void. This option
1109 is not supported for C++.
1111 @item -funsigned-char
1112 Let the type @code{char} be unsigned, like @code{unsigned char}.
1114 Each kind of machine has a default for what @code{char} should
1115 be. It is either like @code{unsigned char} by default or like
1116 @code{signed char} by default.
1118 Ideally, a portable program should always use @code{signed char} or
1119 @code{unsigned char} when it depends on the signedness of an object.
1120 But many programs have been written to use plain @code{char} and
1121 expect it to be signed, or expect it to be unsigned, depending on the
1122 machines they were written for. This option, and its inverse, let you
1123 make such a program work with the opposite default.
1125 The type @code{char} is always a distinct type from each of
1126 @code{signed char} or @code{unsigned char}, even though its behavior
1127 is always just like one of those two.
1130 Let the type @code{char} be signed, like @code{signed char}.
1132 Note that this is equivalent to @samp{-fno-unsigned-char}, which is
1133 the negative form of @samp{-funsigned-char}. Likewise, the option
1134 @samp{-fno-signed-char} is equivalent to @samp{-funsigned-char}.
1136 You may wish to use @samp{-fno-builtin} as well as @samp{-traditional}
1137 if your program uses names that are normally GNU C builtin functions for
1138 other purposes of its own.
1140 You cannot use @samp{-traditional} if you include any header files that
1141 rely on ISO C features. Some vendors are starting to ship systems with
1142 ISO C header files and you cannot use @samp{-traditional} on such
1143 systems to compile files that include any system headers.
1145 @item -fsigned-bitfields
1146 @itemx -funsigned-bitfields
1147 @itemx -fno-signed-bitfields
1148 @itemx -fno-unsigned-bitfields
1149 These options control whether a bitfield is signed or unsigned, when the
1150 declaration does not use either @code{signed} or @code{unsigned}. By
1151 default, such a bitfield is signed, because this is consistent: the
1152 basic integer types such as @code{int} are signed types.
1154 However, when @samp{-traditional} is used, bitfields are all unsigned
1157 @item -fwritable-strings
1158 Store string constants in the writable data segment and don't uniquize
1159 them. This is for compatibility with old programs which assume they can
1160 write into string constants. The option @samp{-traditional} also has
1163 Writing into string constants is a very bad idea; ``constants'' should
1166 @item -fallow-single-precision
1167 Do not promote single precision math operations to double precision,
1168 even when compiling with @samp{-traditional}.
1170 Traditional K&R C promotes all floating point operations to double
1171 precision, regardless of the sizes of the operands. On the
1172 architecture for which you are compiling, single precision may be faster
1173 than double precision. If you must use @samp{-traditional}, but want
1174 to use single precision operations when the operands are single
1175 precision, use this option. This option has no effect when compiling
1176 with ISO or GNU C conventions (the default).
1179 Override the underlying type for @samp{wchar_t} to be @samp{short
1180 unsigned int} instead of the default for the target. This option is
1181 useful for building programs to run under WINE.
1184 @node C++ Dialect Options
1185 @section Options Controlling C++ Dialect
1187 @cindex compiler options, C++
1188 @cindex C++ options, command line
1189 @cindex options, C++
1190 This section describes the command-line options that are only meaningful
1191 for C++ programs; but you can also use most of the GNU compiler options
1192 regardless of what language your program is in. For example, you
1193 might compile a file @code{firstClass.C} like this:
1196 g++ -g -frepo -O -c firstClass.C
1200 In this example, only @samp{-frepo} is an option meant
1201 only for C++ programs; you can use the other options with any
1202 language supported by GCC.
1204 Here is a list of options that are @emph{only} for compiling C++ programs:
1207 @item -fno-access-control
1208 Turn off all access checking. This switch is mainly useful for working
1209 around bugs in the access control code.
1212 Check that the pointer returned by @code{operator new} is non-null
1213 before attempting to modify the storage allocated. The current Working
1214 Paper requires that @code{operator new} never return a null pointer, so
1215 this check is normally unnecessary.
1217 An alternative to using this option is to specify that your
1218 @code{operator new} does not throw any exceptions; if you declare it
1219 @samp{throw()}, g++ will check the return value. See also @samp{new
1222 @item -fconserve-space
1223 Put uninitialized or runtime-initialized global variables into the
1224 common segment, as C does. This saves space in the executable at the
1225 cost of not diagnosing duplicate definitions. If you compile with this
1226 flag and your program mysteriously crashes after @code{main()} has
1227 completed, you may have an object that is being destroyed twice because
1228 two definitions were merged.
1230 This option is no longer useful on most targets, now that support has
1231 been added for putting variables into BSS without making them common.
1233 @item -fdollars-in-identifiers
1234 Accept @samp{$} in identifiers. You can also explicitly prohibit use of
1235 @samp{$} with the option @samp{-fno-dollars-in-identifiers}. (GNU C allows
1236 @samp{$} by default on most target systems, but there are a few exceptions.)
1237 Traditional C allowed the character @samp{$} to form part of
1238 identifiers. However, ISO C and C++ forbid @samp{$} in identifiers.
1240 @item -fno-elide-constructors
1241 The C++ standard allows an implementation to omit creating a temporary
1242 which is only used to initialize another object of the same type.
1243 Specifying this option disables that optimization, and forces g++ to
1244 call the copy constructor in all cases.
1246 @item -fno-enforce-eh-specs
1247 Don't check for violation of exception specifications at runtime. This
1248 option violates the C++ standard, but may be useful for reducing code
1249 size in production builds, much like defining @samp{NDEBUG}. The compiler
1250 will still optimize based on the exception specifications.
1252 @item -fexternal-templates
1253 Cause template instantiations to obey @samp{#pragma interface} and
1254 @samp{implementation}; template instances are emitted or not according
1255 to the location of the template definition. @xref{Template
1256 Instantiation}, for more information.
1258 This option is deprecated.
1260 @item -falt-external-templates
1261 Similar to -fexternal-templates, but template instances are emitted or
1262 not according to the place where they are first instantiated.
1263 @xref{Template Instantiation}, for more information.
1265 This option is deprecated.
1268 @itemx -fno-for-scope
1269 If -ffor-scope is specified, the scope of variables declared in
1270 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1271 as specified by the C++ standard.
1272 If -fno-for-scope is specified, the scope of variables declared in
1273 a @i{for-init-statement} extends to the end of the enclosing scope,
1274 as was the case in old versions of gcc, and other (traditional)
1275 implementations of C++.
1277 The default if neither flag is given to follow the standard,
1278 but to allow and give a warning for old-style code that would
1279 otherwise be invalid, or have different behavior.
1281 @item -fno-gnu-keywords
1282 Do not recognize @code{typeof} as a keyword, so that code can use this
1283 word as an identifier. You can use the keyword @code{__typeof__} instead.
1284 @samp{-ansi} implies @samp{-fno-gnu-keywords}.
1287 Treat the @code{namespace std} as a namespace, instead of ignoring
1288 it. For compatibility with earlier versions of g++, the compiler will,
1289 by default, ignore @code{namespace-declarations},
1290 @code{using-declarations}, @code{using-directives}, and
1291 @code{namespace-names}, if they involve @code{std}.
1293 @item -fhuge-objects
1294 Support virtual function calls for objects that exceed the size
1295 representable by a @samp{short int}. Users should not use this flag by
1296 default; if you need to use it, the compiler will tell you so.
1298 This flag is not useful when compiling with -fvtable-thunks.
1300 Like all options that change the ABI, all C++ code, @emph{including
1301 libgcc} must be built with the same setting of this option.
1303 @item -fno-implicit-templates
1304 Never emit code for non-inline templates which are instantiated
1305 implicitly (i.e. by use); only emit code for explicit instantiations.
1306 @xref{Template Instantiation}, for more information.
1308 @item -fno-implicit-inline-templates
1309 Don't emit code for implicit instantiations of inline templates, either.
1310 The default is to handle inlines differently so that compiles with and
1311 without optimization will need the same set of explicit instantiations.
1313 @item -fno-implement-inlines
1314 To save space, do not emit out-of-line copies of inline functions
1315 controlled by @samp{#pragma implementation}. This will cause linker
1316 errors if these functions are not inlined everywhere they are called.
1318 @item -fms-extensions
1319 Disable pedantic warnings about constructs used in MFC, such as implicit
1320 int and getting a pointer to member function via non-standard syntax.
1322 @item -fno-operator-names
1323 Do not treat the operator name keywords @code{and}, @code{bitand},
1324 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1325 synonyms as keywords.
1327 @item -fno-optional-diags
1328 Disable diagnostics that the standard says a compiler does not need to
1329 issue. Currently, the only such diagnostic issued by g++ is the one for
1330 a name having multiple meanings within a class.
1333 Downgrade messages about nonconformant code from errors to warnings. By
1334 default, g++ effectively sets @samp{-pedantic-errors} without
1335 @samp{-pedantic}; this option reverses that. This behavior and this
1336 option are superseded by @samp{-pedantic}, which works as it does for GNU C.
1339 Enable automatic template instantiation. This option also implies
1340 @samp{-fno-implicit-templates}. @xref{Template Instantiation}, for more
1344 Disable generation of information about every class with virtual
1345 functions for use by the C++ runtime type identification features
1346 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1347 of the language, you can save some space by using this flag. Note that
1348 exception handling uses the same information, but it will generate it as
1351 @item -ftemplate-depth-@var{n}
1352 Set the maximum instantiation depth for template classes to @var{n}.
1353 A limit on the template instantiation depth is needed to detect
1354 endless recursions during template class instantiation. ANSI/ISO C++
1355 conforming programs must not rely on a maximum depth greater than 17.
1357 @item -fuse-cxa-atexit
1358 Register destructors for objects with static storage duration with the
1359 @code{__cxa_atexit} function rather than the @code{atexit} function.
1360 This option is required for fully standards-compliant handling of static
1361 destructors, but will only work if your C library supports
1362 @code{__cxa_atexit}.
1364 @item -fvtable-thunks
1365 Use @samp{thunks} to implement the virtual function dispatch table
1366 (@samp{vtable}). The traditional (cfront-style) approach to
1367 implementing vtables was to store a pointer to the function and two
1368 offsets for adjusting the @samp{this} pointer at the call site. Newer
1369 implementations store a single pointer to a @samp{thunk} function which
1370 does any necessary adjustment and then calls the target function.
1372 This option also enables a heuristic for controlling emission of
1373 vtables; if a class has any non-inline virtual functions, the vtable
1374 will be emitted in the translation unit containing the first one of
1377 Like all options that change the ABI, all C++ code, @emph{including
1378 libgcc.a} must be built with the same setting of this option.
1381 Do not search for header files in the standard directories specific to
1382 C++, but do still search the other standard directories. (This option
1383 is used when building the C++ library.)
1386 In addition, these optimization, warning, and code generation options
1387 have meanings only for C++ programs:
1390 @item -fno-default-inline
1391 Do not assume @samp{inline} for functions defined inside a class scope.
1392 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1393 functions will have linkage like inline functions; they just won't be
1396 @item -Wctor-dtor-privacy (C++ only)
1397 Warn when a class seems unusable, because all the constructors or
1398 destructors in a class are private and the class has no friends or
1399 public static member functions.
1401 @item -Wnon-virtual-dtor (C++ only)
1402 Warn when a class declares a non-virtual destructor that should probably
1403 be virtual, because it looks like the class will be used polymorphically.
1405 @item -Wreorder (C++ only)
1406 @cindex reordering, warning
1407 @cindex warning for reordering of member initializers
1408 Warn when the order of member initializers given in the code does not
1409 match the order in which they must be executed. For instance:
1415 A(): j (0), i (1) @{ @}
1419 Here the compiler will warn that the member initializers for @samp{i}
1420 and @samp{j} will be rearranged to match the declaration order of the
1424 The following @samp{-W@dots{}} options are not affected by @samp{-Wall}.
1427 @item -Weffc++ (C++ only)
1428 Warn about violations of various style guidelines from Scott Meyers'
1429 @cite{Effective C++} books. If you use this option, you should be aware
1430 that the standard library headers do not obey all of these guidelines;
1431 you can use @samp{grep -v} to filter out those warnings.
1433 @item -Wno-deprecated (C++ only)
1434 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1436 @item -Wno-non-template-friend (C++ only)
1437 Disable warnings when non-templatized friend functions are declared
1438 within a template. With the advent of explicit template specification
1439 support in g++, if the name of the friend is an unqualified-id (ie,
1440 @samp{friend foo(int)}), the C++ language specification demands that the
1441 friend declare or define an ordinary, nontemplate function. (Section
1442 14.5.3). Before g++ implemented explicit specification, unqualified-ids
1443 could be interpreted as a particular specialization of a templatized
1444 function. Because this non-conforming behavior is no longer the default
1445 behavior for g++, @samp{-Wnon-template-friend} allows the compiler to
1446 check existing code for potential trouble spots, and is on by default.
1447 This new compiler behavior can be turned off with
1448 @samp{-Wno-non-template-friend} which keeps the conformant compiler code
1449 but disables the helpful warning.
1451 @item -Wold-style-cast (C++ only)
1452 Warn if an old-style (C-style) cast is used within a C++ program. The
1453 new-style casts (@samp{static_cast}, @samp{reinterpret_cast}, and
1454 @samp{const_cast}) are less vulnerable to unintended effects.
1456 @item -Woverloaded-virtual (C++ only)
1457 @cindex overloaded virtual fn, warning
1458 @cindex warning for overloaded virtual fn
1459 Warn when a function declaration hides virtual functions from a
1460 base class. For example, in:
1467 struct B: public A @{
1472 the @code{A} class version of @code{f} is hidden in @code{B}, and code
1480 will fail to compile.
1482 @item -Wno-pmf-conversions (C++ only)
1483 Disable the diagnostic for converting a bound pointer to member function
1486 @item -Wsign-promo (C++ only)
1487 Warn when overload resolution chooses a promotion from unsigned or
1488 enumeral type to a signed type over a conversion to an unsigned type of
1489 the same size. Previous versions of g++ would try to preserve
1490 unsignedness, but the standard mandates the current behavior.
1492 @item -Wsynth (C++ only)
1493 @cindex warning for synthesized methods
1494 @cindex synthesized methods, warning
1495 Warn when g++'s synthesis behavior does not match that of cfront. For
1501 A& operator = (int);
1511 In this example, g++ will synthesize a default @samp{A& operator =
1512 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1515 @node Language Independent Options
1516 @section Options to Control Diagnostic Messages Formatting
1517 @cindex options to control diagnostics formatting
1518 @cindex diagnostic messages
1519 @cindex message formatting
1521 Traditionally, diagnostic messages have been formatted irrespective of
1522 the output device's aspect (e.g. its width, ...). The options described
1523 below can be used to control the diagnostic messages formatting
1524 algorithm, e.g. how many characters per line, how often source location
1525 information should be reported. Right now, only the C++ front-end can
1526 honor these options. However it is expected, in the near future, that
1527 the remaining front-ends would be able to digest them correctly.
1530 @item -fmessage-length=@var{n}
1531 Try to format error messages so that they fit on lines of about @var{n}
1532 characters. The default is 72 characters for g++ and 0 for the rest of
1533 the front-ends supported by GCC. If @var{n} is zero, then no
1534 line-wrapping will be done; each error message will appear on a single
1537 @item -fdiagnostics-show-location=once
1538 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
1539 reporter to emit @emph{once} source location information; that is, in
1540 case the message is too long to fit on a single physical line and has to
1541 be wrapped, the source location won't be emitted (as prefix) again,
1542 over and over, in subsequent continuation lines. This is the default
1545 @item -fdiagnostics-show-location=every-line
1546 Only meaningful in line-wrapping mode. Instructs the diagnostic
1547 messages reporter to emit the same source location information (as
1548 prefix) for physical lines that result from the process of breaking a
1549 a message which is too long to fit on a single line.
1553 @node Warning Options
1554 @section Options to Request or Suppress Warnings
1555 @cindex options to control warnings
1556 @cindex warning messages
1557 @cindex messages, warning
1558 @cindex suppressing warnings
1560 Warnings are diagnostic messages that report constructions which
1561 are not inherently erroneous but which are risky or suggest there
1562 may have been an error.
1564 You can request many specific warnings with options beginning @samp{-W},
1565 for example @samp{-Wimplicit} to request warnings on implicit
1566 declarations. Each of these specific warning options also has a
1567 negative form beginning @samp{-Wno-} to turn off warnings;
1568 for example, @samp{-Wno-implicit}. This manual lists only one of the
1569 two forms, whichever is not the default.
1571 These options control the amount and kinds of warnings produced by GCC:
1574 @cindex syntax checking
1576 Check the code for syntax errors, but don't do anything beyond that.
1579 Issue all the warnings demanded by strict ISO C and ISO C++;
1580 reject all programs that use forbidden extensions, and some other
1581 programs that do not follow ISO C and ISO C++. For ISO C, follows the
1582 version of the ISO C standard specified by any @samp{-std} option used.
1584 Valid ISO C and ISO C++ programs should compile properly with or without
1585 this option (though a rare few will require @option{-ansi} or a
1586 @option{-std} option specifying the required version of ISO C). However,
1587 without this option, certain GNU extensions and traditional C and C++
1588 features are supported as well. With this option, they are rejected.
1590 @samp{-pedantic} does not cause warning messages for use of the
1591 alternate keywords whose names begin and end with @samp{__}. Pedantic
1592 warnings are also disabled in the expression that follows
1593 @code{__extension__}. However, only system header files should use
1594 these escape routes; application programs should avoid them.
1595 @xref{Alternate Keywords}.
1597 Some users try to use @samp{-pedantic} to check programs for strict ISO
1598 C conformance. They soon find that it does not do quite what they want:
1599 it finds some non-ISO practices, but not all---only those for which
1600 ISO C @emph{requires} a diagnostic, and some others for which
1601 diagnostics have been added.
1603 A feature to report any failure to conform to ISO C might be useful in
1604 some instances, but would require considerable additional work and would
1605 be quite different from @samp{-pedantic}. We don't have plans to
1606 support such a feature in the near future.
1608 @item -pedantic-errors
1609 Like @samp{-pedantic}, except that errors are produced rather than
1613 Inhibit all warning messages.
1616 Inhibit warning messages about the use of @samp{#import}.
1618 @item -Wchar-subscripts
1619 Warn if an array subscript has type @code{char}. This is a common cause
1620 of error, as programmers often forget that this type is signed on some
1624 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
1625 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
1628 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
1629 the arguments supplied have types appropriate to the format string
1630 specified, and that the conversions specified in the format string make
1631 sense. This includes standard functions, and others specified by format
1632 attributes (@pxref{Function Attributes}), in the @code{printf},
1633 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
1634 not in the C standard) families.
1636 The formats are checked against the format features supported by GNU
1637 libc version 2.2. These include all ISO C89 and C99 features, as well
1638 as features from the Single Unix Specification and some BSD and GNU
1639 extensions. Other library implementations may not support all these
1640 features; GCC does not support warning about features that go beyond a
1641 particular library's limitations. However, if @samp{-pedantic} is used
1642 with @samp{-Wformat}, warnings will be given about format features not
1643 in the selected standard version (but not for @code{strfmon} formats,
1644 since those are not in any version of the C standard). @xref{C Dialect
1645 Options,,Options Controlling C Dialect}.
1647 @samp{-Wformat} is included in @samp{-Wall}. For more control over some
1648 aspects of format checking, the options @samp{-Wno-format-y2k},
1649 @samp{-Wno-format-extra-args}, @samp{-Wformat-nonliteral},
1650 @samp{-Wformat-security} and @samp{-Wformat=2} are available, but are
1651 not included in @samp{-Wall}.
1653 @item -Wno-format-y2k
1654 If @samp{-Wformat} is specified, do not warn about @code{strftime}
1655 formats which may yield only a two-digit year.
1657 @item -Wno-format-extra-args
1658 If @samp{-Wformat} is specified, do not warn about excess arguments to a
1659 @code{printf} or @code{scanf} format function. The C standard specifies
1660 that such arguments are ignored.
1662 @item -Wformat-nonliteral
1663 If @samp{-Wformat} is specified, also warn if the format string is not a
1664 string literal and so cannot be checked, unless the format function
1665 takes its format arguments as a @code{va_list}.
1667 @item -Wformat-security
1668 If @samp{-Wformat} is specified, also warn about uses of format
1669 functions that represent possible security problems. At present, this
1670 warns about calls to @code{printf} and @code{scanf} functions where the
1671 format string is not a string literal and there are no format arguments,
1672 as in @code{printf (foo);}. This may be a security hole if the format
1673 string came from untrusted input and contains @samp{%n}. (This is
1674 currently a subset of what @samp{-Wformat-nonliteral} warns about, but
1675 in future warnings may be added to @samp{-Wformat-security} that are not
1676 included in @samp{-Wformat-nonliteral}.)
1679 Enable @samp{-Wformat} plus format checks not included in
1680 @samp{-Wformat}. Currently equivalent to @samp{-Wformat
1681 -Wformat-nonliteral -Wformat-security}.
1683 @item -Wimplicit-int
1684 Warn when a declaration does not specify a type.
1686 @item -Wimplicit-function-declaration
1687 @itemx -Werror-implicit-function-declaration
1688 Give a warning (or error) whenever a function is used before being
1692 Same as @samp{-Wimplicit-int} and @samp{-Wimplicit-function-}@*
1696 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
1697 function with external linkage, returning int, taking either zero
1698 arguments, two, or three arguments of appropriate types.
1701 Warn if a multicharacter constant (@samp{'FOOF'}) is used. Usually they
1702 indicate a typo in the user's code, as they have implementation-defined
1703 values, and should not be used in portable code.
1706 Warn if parentheses are omitted in certain contexts, such
1707 as when there is an assignment in a context where a truth value
1708 is expected, or when operators are nested whose precedence people
1709 often get confused about.
1711 Also warn about constructions where there may be confusion to which
1712 @code{if} statement an @code{else} branch belongs. Here is an example of
1725 In C, every @code{else} branch belongs to the innermost possible @code{if}
1726 statement, which in this example is @code{if (b)}. This is often not
1727 what the programmer expected, as illustrated in the above example by
1728 indentation the programmer chose. When there is the potential for this
1729 confusion, GNU C will issue a warning when this flag is specified.
1730 To eliminate the warning, add explicit braces around the innermost
1731 @code{if} statement so there is no way the @code{else} could belong to
1732 the enclosing @code{if}. The resulting code would look like this:
1746 @item -Wsequence-point
1747 Warn about code that may have undefined semantics because of violations
1748 of sequence point rules in the C standard.
1750 The C standard defines the order in which expressions in a C program are
1751 evaluated in terms of @dfn{sequence points}, which represent a partial
1752 ordering between the execution of parts of the program: those executed
1753 before the sequence point, and those executed after it. These occur
1754 after the evaluation of a full expression (one which is not part of a
1755 larger expression), after the evaluation of the first operand of a
1756 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
1757 function is called (but after the evaluation of its arguments and the
1758 expression denoting the called function), and in certain other places.
1759 Other than as expressed by the sequence point rules, the order of
1760 evaluation of subexpressions of an expression is not specified. All
1761 these rules describe only a partial order rather than a total order,
1762 since, for example, if two functions are called within one expression
1763 with no sequence point between them, the order in which the functions
1764 are called is not specified. However, the standards committee have
1765 ruled that function calls do not overlap.
1767 It is not specified when between sequence points modifications to the
1768 values of objects take effect. Programs whose behavior depends on this
1769 have undefined behavior; the C standard specifies that ``Between the
1770 previous and next sequence point an object shall have its stored value
1771 modified at most once by the evaluation of an expression. Furthermore,
1772 the prior value shall be read only to determine the value to be
1773 stored.''. If a program breaks these rules, the results on any
1774 particular implementation are entirely unpredictable.
1776 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
1777 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
1778 diagnosed by this option, and it may give an occasional false positive
1779 result, but in general it has been found fairly effective at detecting
1780 this sort of problem in programs.
1782 The present implementation of this option only works for C programs. A
1783 future implementation may also work for C++ programs.
1785 There is some controversy over the precise meaning of the sequence point
1786 rules in subtle cases. Alternative formal definitions may be found in
1787 Clive Feather's ``Annex S''
1788 @w{@uref{http://wwwold.dkuug.dk/JTC1/SC22/WG14/www/docs/n925.htm}} and in
1789 Michael Norrish's thesis
1790 @w{@uref{http://www.cl.cam.ac.uk/users/mn200/PhD/thesis-report.ps.gz}}.
1791 Other discussions are by Raymond Mak
1792 @w{@uref{http://wwwold.dkuug.dk/JTC1/SC22/WG14/www/docs/n926.htm}} and
1794 @w{@uref{http://wwwold.dkuug.dk/JTC1/SC22/WG14/www/docs/n927.htm}}.
1797 Warn whenever a function is defined with a return-type that defaults to
1798 @code{int}. Also warn about any @code{return} statement with no
1799 return-value in a function whose return-type is not @code{void}.
1801 For C++, a function without return type always produces a diagnostic
1802 message, even when @samp{-Wno-return-type} is specified. The only
1803 exceptions are @samp{main} and functions defined in system headers.
1806 Warn whenever a @code{switch} statement has an index of enumeral type
1807 and lacks a @code{case} for one or more of the named codes of that
1808 enumeration. (The presence of a @code{default} label prevents this
1809 warning.) @code{case} labels outside the enumeration range also
1810 provoke warnings when this option is used.
1813 Warn if any trigraphs are encountered that might change the meaning of
1814 the program (trigraphs within comments are not warned about).
1816 @item -Wunused-function
1817 Warn whenever a static function is declared but not defined or a
1818 non\-inline static function is unused.
1820 @item -Wunused-label
1821 Warn whenever a label is declared but not used.
1823 To suppress this warning use the @samp{unused} attribute
1824 (@pxref{Variable Attributes}).
1826 @item -Wunused-parameter
1827 Warn whenever a function parameter is unused aside from its declaration.
1829 To suppress this warning use the @samp{unused} attribute
1830 (@pxref{Variable Attributes}).
1832 @item -Wunused-variable
1833 Warn whenever a local variable or non-constant static variable is unused
1834 aside from its declaration
1836 To suppress this warning use the @samp{unused} attribute
1837 (@pxref{Variable Attributes}).
1839 @item -Wunused-value
1840 Warn whenever a statement computes a result that is explicitly not used.
1842 To suppress this warning cast the expression to @samp{void}.
1845 All all the above @samp{-Wunused} options combined.
1847 In order to get a warning about an unused function parameter, you must
1848 either specify @samp{-W -Wunused} or separately specify
1849 @samp{-Wunused-parameter}.
1851 @item -Wuninitialized
1852 Warn if an automatic variable is used without first being initialized or
1853 if a variable may be clobbered by a @code{setjmp} call.
1855 These warnings are possible only in optimizing compilation,
1856 because they require data flow information that is computed only
1857 when optimizing. If you don't specify @samp{-O}, you simply won't
1860 These warnings occur only for variables that are candidates for
1861 register allocation. Therefore, they do not occur for a variable that
1862 is declared @code{volatile}, or whose address is taken, or whose size
1863 is other than 1, 2, 4 or 8 bytes. Also, they do not occur for
1864 structures, unions or arrays, even when they are in registers.
1866 Note that there may be no warning about a variable that is used only
1867 to compute a value that itself is never used, because such
1868 computations may be deleted by data flow analysis before the warnings
1871 These warnings are made optional because GCC is not smart
1872 enough to see all the reasons why the code might be correct
1873 despite appearing to have an error. Here is one example of how
1892 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
1893 always initialized, but GCC doesn't know this. Here is
1894 another common case:
1899 if (change_y) save_y = y, y = new_y;
1901 if (change_y) y = save_y;
1906 This has no bug because @code{save_y} is used only if it is set.
1908 @cindex @code{longjmp} warnings
1909 This option also warns when a non-volatile automatic variable might be
1910 changed by a call to @code{longjmp}. These warnings as well are possible
1911 only in optimizing compilation.
1913 The compiler sees only the calls to @code{setjmp}. It cannot know
1914 where @code{longjmp} will be called; in fact, a signal handler could
1915 call it at any point in the code. As a result, you may get a warning
1916 even when there is in fact no problem because @code{longjmp} cannot
1917 in fact be called at the place which would cause a problem.
1919 Some spurious warnings can be avoided if you declare all the functions
1920 you use that never return as @code{noreturn}. @xref{Function
1923 @item -Wreorder (C++ only)
1924 @cindex reordering, warning
1925 @cindex warning for reordering of member initializers
1926 Warn when the order of member initializers given in the code does not
1927 match the order in which they must be executed. For instance:
1929 @item -Wunknown-pragmas
1930 @cindex warning for unknown pragmas
1931 @cindex unknown pragmas, warning
1932 @cindex pragmas, warning of unknown
1933 Warn when a #pragma directive is encountered which is not understood by
1934 GCC. If this command line option is used, warnings will even be issued
1935 for unknown pragmas in system header files. This is not the case if
1936 the warnings were only enabled by the @samp{-Wall} command line option.
1939 All of the above @samp{-W} options combined. This enables all the
1940 warnings about constructions that some users consider questionable, and
1941 that are easy to avoid (or modify to prevent the warning), even in
1942 conjunction with macros.
1944 @item -Wsystem-headers
1945 @cindex warnings from system headers
1946 @cindex system headers, warnings from
1947 Print warning messages for constructs found in system header files.
1948 Warnings from system headers are normally suppressed, on the assumption
1949 that they usually do not indicate real problems and would only make the
1950 compiler output harder to read. Using this command line option tells
1951 GCC to emit warnings from system headers as if they occurred in user
1952 code. However, note that using @samp{-Wall} in conjunction with this
1953 option will @emph{not} warn about unknown pragmas in system
1954 headers---for that, @samp{-Wunknown-pragmas} must also be used.
1957 The following @samp{-W@dots{}} options are not implied by @samp{-Wall}.
1958 Some of them warn about constructions that users generally do not
1959 consider questionable, but which occasionally you might wish to check
1960 for; others warn about constructions that are necessary or hard to avoid
1961 in some cases, and there is no simple way to modify the code to suppress
1966 Print extra warning messages for these events:
1970 A function can return either with or without a value. (Falling
1971 off the end of the function body is considered returning without
1972 a value.) For example, this function would evoke such a
1986 An expression-statement or the left-hand side of a comma expression
1987 contains no side effects.
1988 To suppress the warning, cast the unused expression to void.
1989 For example, an expression such as @samp{x[i,j]} will cause a warning,
1990 but @samp{x[(void)i,j]} will not.
1993 An unsigned value is compared against zero with @samp{<} or @samp{<=}.
1996 A comparison like @samp{x<=y<=z} appears; this is equivalent to
1997 @samp{(x<=y ? 1 : 0) <= z}, which is a different interpretation from
1998 that of ordinary mathematical notation.
2001 Storage-class specifiers like @code{static} are not the first things in
2002 a declaration. According to the C Standard, this usage is obsolescent.
2005 The return type of a function has a type qualifier such as @code{const}.
2006 Such a type qualifier has no effect, since the value returned by a
2007 function is not an lvalue. (But don't warn about the GNU extension of
2008 @code{volatile void} return types. That extension will be warned about
2009 if @samp{-pedantic} is specified.)
2012 If @samp{-Wall} or @samp{-Wunused} is also specified, warn about unused
2016 A comparison between signed and unsigned values could produce an
2017 incorrect result when the signed value is converted to unsigned.
2018 (But don't warn if @samp{-Wno-sign-compare} is also specified.)
2021 An aggregate has a partly bracketed initializer.
2022 For example, the following code would evoke such a warning,
2023 because braces are missing around the initializer for @code{x.h}:
2026 struct s @{ int f, g; @};
2027 struct t @{ struct s h; int i; @};
2028 struct t x = @{ 1, 2, 3 @};
2032 An aggregate has an initializer which does not initialize all members.
2033 For example, the following code would cause such a warning, because
2034 @code{x.h} would be implicitly initialized to zero:
2037 struct s @{ int f, g, h; @};
2038 struct s x = @{ 3, 4 @};
2043 Warn if floating point values are used in equality comparisons.
2045 The idea behind this is that sometimes it is convenient (for the
2046 programmer) to consider floating-point values as approximations to
2047 infinitely precise real numbers. If you are doing this, then you need
2048 to compute (by analysing the code, or in some other way) the maximum or
2049 likely maximum error that the computation introduces, and allow for it
2050 when performing comparisons (and when producing output, but that's a
2051 different problem). In particular, instead of testing for equality, you
2052 would check to see whether the two values have ranges that overlap; and
2053 this is done with the relational operators, so equality comparisons are
2056 @item -Wtraditional (C only)
2057 Warn about certain constructs that behave differently in traditional and
2058 ISO C. Also warn about ISO C constructs that have no traditional C
2059 equivalent, and/or problematic constructs which should be avoided.
2063 Macro parameters that appear within string literals in the macro body.
2064 In traditional C macro replacement takes place within string literals,
2065 but does not in ISO C.
2068 In traditional C, some preprocessor directives did not exist.
2069 Traditional preprocessors would only consider a line to be a directive
2070 if the @samp{#} appeared in column 1 on the line. Therefore
2071 @samp{-Wtraditional} warns about directives that traditional C
2072 understands but would ignore because the @samp{#} does not appear as the
2073 first character on the line. It also suggests you hide directives like
2074 @samp{#pragma} not understood by traditional C by indenting them. Some
2075 traditional implementations would not recognise @samp{#elif}, so it
2076 suggests avoiding it altogether.
2079 A function-like macro that appears without arguments.
2082 The unary plus operator.
2085 The `U' integer constant suffix, or the `F' or `L' floating point
2086 constant suffixes. (Traditonal C does support the `L' suffix on integer
2087 constants.) Note, these suffixes appear in macros defined in the system
2088 headers of most modern systems, e.g. the _MIN/_MAX macros in limits.h.
2089 Use of these macros in user code might normally lead to spurious
2090 warnings, however gcc's integrated preprocessor has enough context to
2091 avoid warning in these cases.
2094 A function declared external in one block and then used after the end of
2098 A @code{switch} statement has an operand of type @code{long}.
2101 A non-@code{static} function declaration follows a @code{static} one.
2102 This construct is not accepted by some traditional C compilers.
2105 The ISO type of an integer constant has a different width or
2106 signedness from its traditional type. This warning is only issued if
2107 the base of the constant is ten. I.e. hexadecimal or octal values, which
2108 typically represent bit patterns, are not warned about.
2111 Usage of ISO string concatenation is detected.
2114 Initialization of automatic aggregates.
2117 Identifier conflicts with labels. Traditional C lacks a separate
2118 namespace for labels.
2121 Initialization of unions. If the initializer is zero, the warning is
2122 omitted. This is done under the assumption that the zero initializer in
2123 user code appears conditioned on e.g. @code{__STDC__} to avoid missing
2124 initializer warnings and relies on default initialization to zero in the
2128 Conversions by prototypes. This is similar to @samp{-Wconversion} in
2129 that it warns about width changes and fixed/floating point conversions,
2130 however it does not warn about changes in signedness.
2134 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
2137 Warn whenever a local variable shadows another local variable.
2139 @item -Wid-clash-@var{len}
2140 Warn whenever two distinct identifiers match in the first @var{len}
2141 characters. This may help you prepare a program that will compile
2142 with certain obsolete, brain-damaged compilers.
2144 @item -Wlarger-than-@var{len}
2145 Warn whenever an object of larger than @var{len} bytes is defined.
2147 @item -Wpointer-arith
2148 Warn about anything that depends on the ``size of'' a function type or
2149 of @code{void}. GNU C assigns these types a size of 1, for
2150 convenience in calculations with @code{void *} pointers and pointers
2153 @item -Wbad-function-cast (C only)
2154 Warn whenever a function call is cast to a non-matching type.
2155 For example, warn if @code{int malloc()} is cast to @code{anything *}.
2158 Warn whenever a pointer is cast so as to remove a type qualifier from
2159 the target type. For example, warn if a @code{const char *} is cast
2160 to an ordinary @code{char *}.
2163 Warn whenever a pointer is cast such that the required alignment of the
2164 target is increased. For example, warn if a @code{char *} is cast to
2165 an @code{int *} on machines where integers can only be accessed at
2166 two- or four-byte boundaries.
2168 @item -Wwrite-strings
2169 Give string constants the type @code{const char[@var{length}]} so that
2170 copying the address of one into a non-@code{const} @code{char *}
2171 pointer will get a warning. These warnings will help you find at
2172 compile time code that can try to write into a string constant, but
2173 only if you have been very careful about using @code{const} in
2174 declarations and prototypes. Otherwise, it will just be a nuisance;
2175 this is why we did not make @samp{-Wall} request these warnings.
2178 Warn if a prototype causes a type conversion that is different from what
2179 would happen to the same argument in the absence of a prototype. This
2180 includes conversions of fixed point to floating and vice versa, and
2181 conversions changing the width or signedness of a fixed point argument
2182 except when the same as the default promotion.
2184 Also, warn if a negative integer constant expression is implicitly
2185 converted to an unsigned type. For example, warn about the assignment
2186 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
2187 casts like @code{(unsigned) -1}.
2189 @item -Wsign-compare
2190 @cindex warning for comparison of signed and unsigned values
2191 @cindex comparison of signed and unsigned values, warning
2192 @cindex signed and unsigned values, comparison warning
2193 Warn when a comparison between signed and unsigned values could produce
2194 an incorrect result when the signed value is converted to unsigned.
2195 This warning is also enabled by @samp{-W}; to get the other warnings
2196 of @samp{-W} without this warning, use @samp{-W -Wno-sign-compare}.
2198 @item -Waggregate-return
2199 Warn if any functions that return structures or unions are defined or
2200 called. (In languages where you can return an array, this also elicits
2203 @item -Wstrict-prototypes (C only)
2204 Warn if a function is declared or defined without specifying the
2205 argument types. (An old-style function definition is permitted without
2206 a warning if preceded by a declaration which specifies the argument
2209 @item -Wmissing-prototypes (C only)
2210 Warn if a global function is defined without a previous prototype
2211 declaration. This warning is issued even if the definition itself
2212 provides a prototype. The aim is to detect global functions that fail
2213 to be declared in header files.
2215 @item -Wmissing-declarations
2216 Warn if a global function is defined without a previous declaration.
2217 Do so even if the definition itself provides a prototype.
2218 Use this option to detect global functions that are not declared in
2221 @item -Wmissing-noreturn
2222 Warn about functions which might be candidates for attribute @code{noreturn}.
2223 Note these are only possible candidates, not absolute ones. Care should
2224 be taken to manually verify functions actually do not ever return before
2225 adding the @code{noreturn} attribute, otherwise subtle code generation
2226 bugs could be introduced. You will not get a warning for @code{main} in
2227 hosted C environments.
2229 @item -Wmissing-format-attribute
2230 If @samp{-Wformat} is enabled, also warn about functions which might be
2231 candidates for @code{format} attributes. Note these are only possible
2232 candidates, not absolute ones. GCC will guess that @code{format}
2233 attributes might be appropriate for any function that calls a function
2234 like @code{vprintf} or @code{vscanf}, but this might not always be the
2235 case, and some functions for which @code{format} attributes are
2236 appropriate may not be detected. This option has no effect unless
2237 @samp{-Wformat} is enabled (possibly by @samp{-Wall}).
2240 Warn if a structure is given the packed attribute, but the packed
2241 attribute has no effect on the layout or size of the structure.
2242 Such structures may be mis-aligned for little benefit. For
2243 instance, in this code, the variable @code{f.x} in @code{struct bar}
2244 will be misaligned even though @code{struct bar} does not itself
2245 have the packed attribute:
2252 @} __attribute__((packed));
2261 Warn if padding is included in a structure, either to align an element
2262 of the structure or to align the whole structure. Sometimes when this
2263 happens it is possible to rearrange the fields of the structure to
2264 reduce the padding and so make the structure smaller.
2266 @item -Wredundant-decls
2267 Warn if anything is declared more than once in the same scope, even in
2268 cases where multiple declaration is valid and changes nothing.
2270 @item -Wnested-externs (C only)
2271 Warn if an @code{extern} declaration is encountered within a function.
2273 @item -Wunreachable-code
2274 Warn if the compiler detects that code will never be executed.
2276 This option is intended to warn when the compiler detects that at
2277 least a whole line of source code will never be executed, because
2278 some condition is never satisfied or because it is after a
2279 procedure that never returns.
2281 It is possible for this option to produce a warning even though there
2282 are circumstances under which part of the affected line can be executed,
2283 so care should be taken when removing apparently-unreachable code.
2285 For instance, when a function is inlined, a warning may mean that the
2286 line is unreachable in only one inlined copy of the function.
2288 This option is not made part of @samp{-Wall} because in a debugging
2289 version of a program there is often substantial code which checks
2290 correct functioning of the program and is, hopefully, unreachable
2291 because the program does work. Another common use of unreachable
2292 code is to provide behaviour which is selectable at compile-time.
2295 Warn if a function can not be inlined and it was declared as inline.
2298 Warn if @samp{long long} type is used. This is default. To inhibit
2299 the warning messages, use @samp{-Wno-long-long}. Flags
2300 @samp{-Wlong-long} and @samp{-Wno-long-long} are taken into account
2301 only when @samp{-pedantic} flag is used.
2303 @item -Wdisabled-optimization
2304 Warn if a requested optimization pass is disabled. This warning does
2305 not generally indicate that there is anything wrong with your code; it
2306 merely indicates that GCC's optimizers were unable to handle the code
2307 effectively. Often, the problem is that your code is too big or too
2308 complex; GCC will refuse to optimize programs when the optimization
2309 itself is likely to take inordinate amounts of time.
2312 Make all warnings into errors.
2315 @node Debugging Options
2316 @section Options for Debugging Your Program or GCC
2317 @cindex options, debugging
2318 @cindex debugging information options
2320 GCC has various special options that are used for debugging
2321 either your program or GCC:
2325 Produce debugging information in the operating system's native format
2326 (stabs, COFF, XCOFF, or DWARF). GDB can work with this debugging
2329 On most systems that use stabs format, @samp{-g} enables use of extra
2330 debugging information that only GDB can use; this extra information
2331 makes debugging work better in GDB but will probably make other debuggers
2333 refuse to read the program. If you want to control for certain whether
2334 to generate the extra information, use @samp{-gstabs+}, @samp{-gstabs},
2335 @samp{-gxcoff+}, @samp{-gxcoff}, @samp{-gdwarf-1+}, or @samp{-gdwarf-1}
2338 Unlike most other C compilers, GCC allows you to use @samp{-g} with
2339 @samp{-O}. The shortcuts taken by optimized code may occasionally
2340 produce surprising results: some variables you declared may not exist
2341 at all; flow of control may briefly move where you did not expect it;
2342 some statements may not be executed because they compute constant
2343 results or their values were already at hand; some statements may
2344 execute in different places because they were moved out of loops.
2346 Nevertheless it proves possible to debug optimized output. This makes
2347 it reasonable to use the optimizer for programs that might have bugs.
2349 The following options are useful when GCC is generated with the
2350 capability for more than one debugging format.
2353 Produce debugging information for use by GDB. This means to use the
2354 most expressive format available (DWARF 2, stabs, or the native format
2355 if neither of those are supported), including GDB extensions if at all
2359 Produce debugging information in stabs format (if that is supported),
2360 without GDB extensions. This is the format used by DBX on most BSD
2361 systems. On MIPS, Alpha and System V Release 4 systems this option
2362 produces stabs debugging output which is not understood by DBX or SDB.
2363 On System V Release 4 systems this option requires the GNU assembler.
2366 Produce debugging information in stabs format (if that is supported),
2367 using GNU extensions understood only by the GNU debugger (GDB). The
2368 use of these extensions is likely to make other debuggers crash or
2369 refuse to read the program.
2372 Produce debugging information in COFF format (if that is supported).
2373 This is the format used by SDB on most System V systems prior to
2377 Produce debugging information in XCOFF format (if that is supported).
2378 This is the format used by the DBX debugger on IBM RS/6000 systems.
2381 Produce debugging information in XCOFF format (if that is supported),
2382 using GNU extensions understood only by the GNU debugger (GDB). The
2383 use of these extensions is likely to make other debuggers crash or
2384 refuse to read the program, and may cause assemblers other than the GNU
2385 assembler (GAS) to fail with an error.
2388 Produce debugging information in DWARF version 1 format (if that is
2389 supported). This is the format used by SDB on most System V Release 4
2393 Produce debugging information in DWARF version 1 format (if that is
2394 supported), using GNU extensions understood only by the GNU debugger
2395 (GDB). The use of these extensions is likely to make other debuggers
2396 crash or refuse to read the program.
2399 Produce debugging information in DWARF version 2 format (if that is
2400 supported). This is the format used by DBX on IRIX 6.
2403 @itemx -ggdb@var{level}
2404 @itemx -gstabs@var{level}
2405 @itemx -gcoff@var{level}
2406 @itemx -gxcoff@var{level}
2407 @itemx -gdwarf@var{level}
2408 @itemx -gdwarf-2@var{level}
2409 Request debugging information and also use @var{level} to specify how
2410 much information. The default level is 2.
2412 Level 1 produces minimal information, enough for making backtraces in
2413 parts of the program that you don't plan to debug. This includes
2414 descriptions of functions and external variables, but no information
2415 about local variables and no line numbers.
2417 Level 3 includes extra information, such as all the macro definitions
2418 present in the program. Some debuggers support macro expansion when
2423 Generate extra code to write profile information suitable for the
2424 analysis program @code{prof}. You must use this option when compiling
2425 the source files you want data about, and you must also use it when
2428 @cindex @code{gprof}
2430 Generate extra code to write profile information suitable for the
2431 analysis program @code{gprof}. You must use this option when compiling
2432 the source files you want data about, and you must also use it when
2437 Generate extra code to write profile information for basic blocks, which will
2438 record the number of times each basic block is executed, the basic block start
2439 address, and the function name containing the basic block. If @samp{-g} is
2440 used, the line number and filename of the start of the basic block will also be
2441 recorded. If not overridden by the machine description, the default action is
2442 to append to the text file @file{bb.out}.
2444 This data could be analyzed by a program like @code{tcov}. Note,
2445 however, that the format of the data is not what @code{tcov} expects.
2446 Eventually GNU @code{gprof} should be extended to process this data.
2449 Makes the compiler print out each function name as it is compiled, and
2450 print some statistics about each pass when it finishes.
2453 Generate extra code to profile basic blocks. Your executable will
2454 produce output that is a superset of that produced when @samp{-a} is
2455 used. Additional output is the source and target address of the basic
2456 blocks where a jump takes place, the number of times a jump is executed,
2457 and (optionally) the complete sequence of basic blocks being executed.
2458 The output is appended to file @file{bb.out}.
2460 You can examine different profiling aspects without recompilation. Your
2461 executable will read a list of function names from file @file{bb.in}.
2462 Profiling starts when a function on the list is entered and stops when
2463 that invocation is exited. To exclude a function from profiling, prefix
2464 its name with `-'. If a function name is not unique, you can
2465 disambiguate it by writing it in the form
2466 @samp{/path/filename.d:functionname}. Your executable will write the
2467 available paths and filenames in file @file{bb.out}.
2469 Several function names have a special meaning:
2472 Write source, target and frequency of jumps to file @file{bb.out}.
2473 @item __bb_hidecall__
2474 Exclude function calls from frequency count.
2475 @item __bb_showret__
2476 Include function returns in frequency count.
2478 Write the sequence of basic blocks executed to file @file{bbtrace.gz}.
2479 The file will be compressed using the program @samp{gzip}, which must
2480 exist in your @env{PATH}. On systems without the @samp{popen}
2481 function, the file will be named @file{bbtrace} and will not be
2482 compressed. @strong{Profiling for even a few seconds on these systems
2483 will produce a very large file.} Note: @code{__bb_hidecall__} and
2484 @code{__bb_showret__} will not affect the sequence written to
2488 Here's a short example using different profiling parameters
2489 in file @file{bb.in}. Assume function @code{foo} consists of basic blocks
2490 1 and 2 and is called twice from block 3 of function @code{main}. After
2491 the calls, block 3 transfers control to block 4 of @code{main}.
2493 With @code{__bb_trace__} and @code{main} contained in file @file{bb.in},
2494 the following sequence of blocks is written to file @file{bbtrace.gz}:
2495 0 3 1 2 1 2 4. The return from block 2 to block 3 is not shown, because
2496 the return is to a point inside the block and not to the top. The
2497 block address 0 always indicates, that control is transferred
2498 to the trace from somewhere outside the observed functions. With
2499 @samp{-foo} added to @file{bb.in}, the blocks of function
2500 @code{foo} are removed from the trace, so only 0 3 4 remains.
2502 With @code{__bb_jumps__} and @code{main} contained in file @file{bb.in},
2503 jump frequencies will be written to file @file{bb.out}. The
2504 frequencies are obtained by constructing a trace of blocks
2505 and incrementing a counter for every neighbouring pair of blocks
2506 in the trace. The trace 0 3 1 2 1 2 4 displays the following
2510 Jump from block 0x0 to block 0x3 executed 1 time(s)
2511 Jump from block 0x3 to block 0x1 executed 1 time(s)
2512 Jump from block 0x1 to block 0x2 executed 2 time(s)
2513 Jump from block 0x2 to block 0x1 executed 1 time(s)
2514 Jump from block 0x2 to block 0x4 executed 1 time(s)
2517 With @code{__bb_hidecall__}, control transfer due to call instructions
2518 is removed from the trace, that is the trace is cut into three parts: 0
2519 3 4, 0 1 2 and 0 1 2. With @code{__bb_showret__}, control transfer due
2520 to return instructions is added to the trace. The trace becomes: 0 3 1
2521 2 3 1 2 3 4. Note, that this trace is not the same, as the sequence
2522 written to @file{bbtrace.gz}. It is solely used for counting jump
2525 @item -fprofile-arcs
2526 Instrument @dfn{arcs} during compilation. For each function of your
2527 program, GCC creates a program flow graph, then finds a spanning tree
2528 for the graph. Only arcs that are not on the spanning tree have to be
2529 instrumented: the compiler adds code to count the number of times that these
2530 arcs are executed. When an arc is the only exit or only entrance to a
2531 block, the instrumentation code can be added to the block; otherwise, a
2532 new basic block must be created to hold the instrumentation code.
2534 Since not every arc in the program must be instrumented, programs
2535 compiled with this option run faster than programs compiled with
2536 @samp{-a}, which adds instrumentation code to every basic block in the
2537 program. The tradeoff: since @code{gcov} does not have
2538 execution counts for all branches, it must start with the execution
2539 counts for the instrumented branches, and then iterate over the program
2540 flow graph until the entire graph has been solved. Hence, @code{gcov}
2541 runs a little more slowly than a program which uses information from
2544 @samp{-fprofile-arcs} also makes it possible to estimate branch
2545 probabilities, and to calculate basic block execution counts. In
2546 general, basic block execution counts do not give enough information to
2547 estimate all branch probabilities. When the compiled program exits, it
2548 saves the arc execution counts to a file called
2549 @file{@var{sourcename}.da}. Use the compiler option
2550 @samp{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
2551 Control Optimization}) when recompiling, to optimize using estimated
2552 branch probabilities.
2555 @item -ftest-coverage
2556 Create data files for the @code{gcov} code-coverage utility
2557 (@pxref{Gcov,, @code{gcov}: a GCC Test Coverage Program}).
2558 The data file names begin with the name of your source file:
2561 @item @var{sourcename}.bb
2562 A mapping from basic blocks to line numbers, which @code{gcov} uses to
2563 associate basic block execution counts with line numbers.
2565 @item @var{sourcename}.bbg
2566 A list of all arcs in the program flow graph. This allows @code{gcov}
2567 to reconstruct the program flow graph, so that it can compute all basic
2568 block and arc execution counts from the information in the
2569 @code{@var{sourcename}.da} file (this last file is the output from
2570 @samp{-fprofile-arcs}).
2573 @item -d@var{letters}
2574 Says to make debugging dumps during compilation at times specified by
2575 @var{letters}. This is used for debugging the compiler. The file names
2576 for most of the dumps are made by appending a pass number and a word to
2577 the source file name (e.g. @file{foo.c.00.rtl} or @file{foo.c.01.sibling}).
2578 Here are the possible letters for use in @var{letters}, and their meanings:
2582 Annotate the assembler output with miscellaneous debugging information.
2584 Dump after computing branch probabilities, to @file{@var{file}.11.bp}.
2586 Dump after block reordering, to @file{@var{file}.26.bbro}.
2588 Dump after instruction combination, to the file @file{@var{file}.14.combine}.
2590 Dump after the first if conversion, to the file @file{@var{file}.15.ce}.
2592 Dump after delayed branch scheduling, to @file{@var{file}.29.dbr}.
2594 Dump all macro definitions, at the end of preprocessing, in addition to
2597 Dump after SSA optimizations, to @file{@var{file}.05.ssa} and
2598 @file{@var{file}.06.ussa}.
2600 Dump after the second if conversion, to @file{@var{file}.24.ce2}.
2602 Dump after life analysis, to @file{@var{file}.13.life}.
2604 Dump after purging @code{ADDRESSOF} codes, to @file{@var{file}.04.addressof}.
2606 Dump after global register allocation, to @file{@var{file}.19.greg}.
2608 Dump after post-reload CSE and other optimizations, to @file{@var{file}.20.postreload}.
2610 Dump after GCSE, to @file{@var{file}.08.gcse}.
2612 Dump after sibling call optimizations, to @file{@var{file}.01.sibling}.
2614 Dump after the first jump optimization, to @file{@var{file}.02.jump}.
2616 Dump after the last jump optimization, to @file{@var{file}.27.jump2}.
2618 Dump after conversion from registers to stack, to @file{@var{file}.29.stack}.
2620 Dump after local register allocation, to @file{@var{file}.18.lreg}.
2622 Dump after loop optimization, to @file{@var{file}.09.loop}.
2624 Dump after performing the machine dependent reorganisation pass, to
2625 @file{@var{file}.28.mach}.
2627 Dump after register renumbering, to @file{@var{file}.23.rnreg}.
2629 Dump after the register move pass, to @file{@var{file}.16.regmove}.
2631 Dump after RTL generation, to @file{@var{file}.00.rtl}.
2633 Dump after the second instruction scheduling pass, to
2634 @file{@var{file}.25.sched2}.
2636 Dump after CSE (including the jump optimization that sometimes follows
2637 CSE), to @file{@var{file}.03.cse}.
2639 Dump after the first instruction scheduling pass, to
2640 @file{@var{file}.17.sched}.
2642 Dump after the second CSE pass (including the jump optimization that
2643 sometimes follows CSE), to @file{@var{file}.10.cse2}.
2645 Dump after the second flow pass, to @file{@var{file}.21.flow2}.
2647 Dump after dead code elimination, to @file{@var{file}.06.dce}.
2649 Dump after the peephole pass, to @file{@var{file}.22.peephole2}.
2651 Produce all the dumps listed above.
2653 Print statistics on memory usage, at the end of the run, to
2656 Annotate the assembler output with a comment indicating which
2657 pattern and alternative was used. The length of each instruction is
2660 Dump the RTL in the assembler output as a comment before each instruction.
2661 Also turns on @samp{-dp} annotation.
2663 For each of the other indicated dump files (except for
2664 @file{@var{file}.00.rtl}), dump a representation of the control flow graph
2665 suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
2667 Just generate RTL for a function instead of compiling it. Usually used
2670 Dump debugging information during parsing, to standard error.
2673 @item -fdump-unnumbered
2674 When doing debugging dumps (see -d option above), suppress instruction
2675 numbers and line number note output. This makes it more feasible to
2676 use diff on debugging dumps for compiler invocations with different
2677 options, in particular with and without -g.
2679 @item -fdump-translation-unit=@var{file} (C and C++ only)
2680 Dump a representation of the tree structure for the entire translation
2683 @item -fdump-class_layout=@var{file} (C++ only)
2684 @item -fdump-class_layout (C++ only)
2685 Dump a representation of each class's heirarchy to @var{file}, or
2686 @code{stderr} if not specified.
2688 @item -fpretend-float
2689 When running a cross-compiler, pretend that the target machine uses the
2690 same floating point format as the host machine. This causes incorrect
2691 output of the actual floating constants, but the actual instruction
2692 sequence will probably be the same as GCC would make when running on
2696 Store the usual ``temporary'' intermediate files permanently; place them
2697 in the current directory and name them based on the source file. Thus,
2698 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
2699 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
2700 preprocessed @file{foo.i} output file even though the compiler now
2701 normally uses an integrated preprocessor.
2704 Report the CPU time taken by each subprocess in the compilation
2705 sequence. For C source files, this is the compiler proper and assembler
2706 (plus the linker if linking is done). The output looks like this:
2713 The first number on each line is the ``user time,'' that is time spent
2714 executing the program itself. The second number is ``system time,''
2715 time spent executing operating system routines on behalf of the program.
2716 Both numbers are in seconds.
2718 @item -print-file-name=@var{library}
2719 Print the full absolute name of the library file @var{library} that
2720 would be used when linking---and don't do anything else. With this
2721 option, GCC does not compile or link anything; it just prints the
2724 @item -print-prog-name=@var{program}
2725 Like @samp{-print-file-name}, but searches for a program such as @samp{cpp}.
2727 @item -print-libgcc-file-name
2728 Same as @samp{-print-file-name=libgcc.a}.
2730 This is useful when you use @samp{-nostdlib} or @samp{-nodefaultlibs}
2731 but you do want to link with @file{libgcc.a}. You can do
2734 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
2737 @item -print-search-dirs
2738 Print the name of the configured installation directory and a list of
2739 program and library directories gcc will search---and don't do anything else.
2741 This is useful when gcc prints the error message
2742 @samp{installation problem, cannot exec cpp0: No such file or directory}.
2743 To resolve this you either need to put @file{cpp0} and the other compiler
2744 components where gcc expects to find them, or you can set the environment
2745 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
2746 Don't forget the trailing '/'.
2747 @xref{Environment Variables}.
2750 @node Optimize Options
2751 @section Options That Control Optimization
2752 @cindex optimize options
2753 @cindex options, optimization
2755 These options control various sorts of optimizations:
2760 Optimize. Optimizing compilation takes somewhat more time, and a lot
2761 more memory for a large function.
2763 Without @samp{-O}, the compiler's goal is to reduce the cost of
2764 compilation and to make debugging produce the expected results.
2765 Statements are independent: if you stop the program with a breakpoint
2766 between statements, you can then assign a new value to any variable or
2767 change the program counter to any other statement in the function and
2768 get exactly the results you would expect from the source code.
2770 Without @samp{-O}, the compiler only allocates variables declared
2771 @code{register} in registers. The resulting compiled code is a little
2772 worse than produced by PCC without @samp{-O}.
2774 With @samp{-O}, the compiler tries to reduce code size and execution
2777 When you specify @samp{-O}, the compiler turns on @samp{-fthread-jumps}
2778 and @samp{-fdefer-pop} on all machines. The compiler turns on
2779 @samp{-fdelayed-branch} on machines that have delay slots, and
2780 @samp{-fomit-frame-pointer} on machines that can support debugging even
2781 without a frame pointer. On some machines the compiler also turns
2782 on other flags.@refill
2785 Optimize even more. GCC performs nearly all supported optimizations
2786 that do not involve a space-speed tradeoff. The compiler does not
2787 perform loop unrolling or function inlining when you specify @samp{-O2}.
2788 As compared to @samp{-O}, this option increases both compilation time
2789 and the performance of the generated code.
2791 @samp{-O2} turns on all optional optimizations except for loop unrolling,
2792 function inlining, and register renaming. It also turns on the
2793 @samp{-fforce-mem} option on all machines and frame pointer elimination
2794 on machines where doing so does not interfere with debugging.
2797 Optimize yet more. @samp{-O3} turns on all optimizations specified by
2798 @samp{-O2} and also turns on the @samp{-finline-functions} and
2799 @samp{-frename-registers} options.
2805 Optimize for size. @samp{-Os} enables all @samp{-O2} optimizations that
2806 do not typically increase code size. It also performs further
2807 optimizations designed to reduce code size.
2809 If you use multiple @samp{-O} options, with or without level numbers,
2810 the last such option is the one that is effective.
2813 Options of the form @samp{-f@var{flag}} specify machine-independent
2814 flags. Most flags have both positive and negative forms; the negative
2815 form of @samp{-ffoo} would be @samp{-fno-foo}. In the table below,
2816 only one of the forms is listed---the one which is not the default.
2817 You can figure out the other form by either removing @samp{no-} or
2822 Do not store floating point variables in registers, and inhibit other
2823 options that might change whether a floating point value is taken from a
2826 @cindex floating point precision
2827 This option prevents undesirable excess precision on machines such as
2828 the 68000 where the floating registers (of the 68881) keep more
2829 precision than a @code{double} is supposed to have. Similarly for the
2830 x86 architecture. For most programs, the excess precision does only
2831 good, but a few programs rely on the precise definition of IEEE floating
2832 point. Use @samp{-ffloat-store} for such programs, after modifying
2833 them to store all pertinent intermediate computations into variables.
2835 @item -fno-default-inline
2836 Do not make member functions inline by default merely because they are
2837 defined inside the class scope (C++ only). Otherwise, when you specify
2838 @w{@samp{-O}}, member functions defined inside class scope are compiled
2839 inline by default; i.e., you don't need to add @samp{inline} in front of
2840 the member function name.
2842 @item -fno-defer-pop
2843 Always pop the arguments to each function call as soon as that function
2844 returns. For machines which must pop arguments after a function call,
2845 the compiler normally lets arguments accumulate on the stack for several
2846 function calls and pops them all at once.
2849 Force memory operands to be copied into registers before doing
2850 arithmetic on them. This produces better code by making all memory
2851 references potential common subexpressions. When they are not common
2852 subexpressions, instruction combination should eliminate the separate
2853 register-load. The @samp{-O2} option turns on this option.
2856 Force memory address constants to be copied into registers before
2857 doing arithmetic on them. This may produce better code just as
2858 @samp{-fforce-mem} may.
2860 @item -fomit-frame-pointer
2861 Don't keep the frame pointer in a register for functions that
2862 don't need one. This avoids the instructions to save, set up and
2863 restore frame pointers; it also makes an extra register available
2864 in many functions. @strong{It also makes debugging impossible on
2868 On some machines, such as the Vax, this flag has no effect, because
2869 the standard calling sequence automatically handles the frame pointer
2870 and nothing is saved by pretending it doesn't exist. The
2871 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
2872 whether a target machine supports this flag. @xref{Registers}.@refill
2875 On some machines, such as the Vax, this flag has no effect, because
2876 the standard calling sequence automatically handles the frame pointer
2877 and nothing is saved by pretending it doesn't exist. The
2878 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
2879 whether a target machine supports this flag. @xref{Registers,,Register
2880 Usage, gcc.info, Using and Porting GCC}.@refill
2883 @item -foptimize-sibling-calls
2884 Optimize sibling and tail recursive calls.
2887 This option generates traps for signed overflow on addition, subtraction,
2888 multiplication operations.
2891 Don't pay attention to the @code{inline} keyword. Normally this option
2892 is used to keep the compiler from expanding any functions inline.
2893 Note that if you are not optimizing, no functions can be expanded inline.
2895 @item -finline-functions
2896 Integrate all simple functions into their callers. The compiler
2897 heuristically decides which functions are simple enough to be worth
2898 integrating in this way.
2900 If all calls to a given function are integrated, and the function is
2901 declared @code{static}, then the function is normally not output as
2902 assembler code in its own right.
2904 @item -finline-limit=@var{n}
2905 By default, gcc limits the size of functions that can be inlined. This flag
2906 allows the control of this limit for functions that are explicitly marked as
2907 inline (ie marked with the inline keyword or defined within the class
2908 definition in c++). @var{n} is the size of functions that can be inlined in
2909 number of pseudo instructions (not counting parameter handling). The default
2910 value of n is 10000. Increasing this value can result in more inlined code at
2911 the cost of compilation time and memory consumption. Decreasing usually makes
2912 the compilation faster and less code will be inlined (which presumably
2913 means slower programs). This option is particularly useful for programs that
2914 use inlining heavily such as those based on recursive templates with c++.
2916 @emph{Note:} pseudo instruction represents, in this particular context, an
2917 abstract measurement of function's size. In no way, it represents a count
2918 of assembly instructions and as such its exact meaning might change from one
2919 release to an another.
2921 @item -fkeep-inline-functions
2922 Even if all calls to a given function are integrated, and the function
2923 is declared @code{static}, nevertheless output a separate run-time
2924 callable version of the function. This switch does not affect
2925 @code{extern inline} functions.
2927 @item -fkeep-static-consts
2928 Emit variables declared @code{static const} when optimization isn't turned
2929 on, even if the variables aren't referenced.
2931 GCC enables this option by default. If you want to force the compiler to
2932 check if the variable was referenced, regardless of whether or not
2933 optimization is turned on, use the @samp{-fno-keep-static-consts} option.
2935 @item -fno-function-cse
2936 Do not put function addresses in registers; make each instruction that
2937 calls a constant function contain the function's address explicitly.
2939 This option results in less efficient code, but some strange hacks
2940 that alter the assembler output may be confused by the optimizations
2941 performed when this option is not used.
2944 Sets @samp{-fno-math-errno}, @samp{-funsafe-math-optimizations},
2945 and @samp{-fno-trapping-math}.
2947 This option causes the preprocessor macro __FAST_MATH__ to be defined.
2949 This option should never be turned on by any @samp{-O} option since
2950 it can result in incorrect output for programs which depend on
2951 an exact implementation of IEEE or ISO rules/specifications for
2954 @item -fno-math-errno
2955 Do not set ERRNO after calling math functions that are executed
2956 with a single instruction, e.g., sqrt. A program that relies on
2957 IEEE exceptions for math error handling may want to use this flag
2958 for speed while maintaining IEEE arithmetic compatibility.
2960 This option should never be turned on by any @samp{-O} option since
2961 it can result in incorrect output for programs which depend on
2962 an exact implementation of IEEE or ISO rules/specifications for
2965 The default is @samp{-fmath-errno}. The @samp{-ffast-math} option
2966 sets @samp{-fno-math-errno}.
2968 @item -funsafe-math-optimizations
2969 Allow optimizations for floating-point arithmetic that (a) assume
2970 that arguments and results are valid and (b) may violate IEEE or
2973 This option should never be turned on by any @samp{-O} option since
2974 it can result in incorrect output for programs which depend on
2975 an exact implementation of IEEE or ISO rules/specifications for
2978 The default is @samp{-fno-unsafe-math-optimizations}. The
2979 @samp{-ffast-math} option sets @samp{-funsafe-math-optimizations}.
2981 @item -fno-trapping-math
2982 Compile code assuming that floating-point operations cannot generate
2983 user-visible traps. Setting this option may allow faster code
2984 if one relies on ``non-stop'' IEEE arithmetic, for example.
2986 This option should never be turned on by any @samp{-O} option since
2987 it can result in incorrect output for programs which depend on
2988 an exact implementation of IEEE or ISO rules/specifications for
2991 The default is @samp{-ftrapping-math}. The @samp{-ffast-math}
2992 option sets @samp{-fno-trapping-math}.
2995 @c following causes underfulls.. they don't look great, but we deal.
2997 The following options control specific optimizations. The @samp{-O2}
2998 option turns on all of these optimizations except @samp{-funroll-loops}
2999 and @samp{-funroll-all-loops}. On most machines, the @samp{-O} option
3000 turns on the @samp{-fthread-jumps} and @samp{-fdelayed-branch} options,
3001 but specific machines may handle it differently.
3003 You can use the following flags in the rare cases when ``fine-tuning''
3004 of optimizations to be performed is desired.
3007 @item -fstrength-reduce
3008 Perform the optimizations of loop strength reduction and
3009 elimination of iteration variables.
3011 @item -fthread-jumps
3012 Perform optimizations where we check to see if a jump branches to a
3013 location where another comparison subsumed by the first is found. If
3014 so, the first branch is redirected to either the destination of the
3015 second branch or a point immediately following it, depending on whether
3016 the condition is known to be true or false.
3018 @item -fcse-follow-jumps
3019 In common subexpression elimination, scan through jump instructions
3020 when the target of the jump is not reached by any other path. For
3021 example, when CSE encounters an @code{if} statement with an
3022 @code{else} clause, CSE will follow the jump when the condition
3025 @item -fcse-skip-blocks
3026 This is similar to @samp{-fcse-follow-jumps}, but causes CSE to
3027 follow jumps which conditionally skip over blocks. When CSE
3028 encounters a simple @code{if} statement with no else clause,
3029 @samp{-fcse-skip-blocks} causes CSE to follow the jump around the
3030 body of the @code{if}.
3032 @item -frerun-cse-after-loop
3033 Re-run common subexpression elimination after loop optimizations has been
3036 @item -frerun-loop-opt
3037 Run the loop optimizer twice.
3040 Perform a global common subexpression elimination pass.
3041 This pass also performs global constant and copy propagation.
3044 When -fgcse-lm is enabled, global common subexpression elimination will
3045 attempt to move loads which are only killed by stores into themselves. This
3046 allows a loop containing a load/store sequence to be changed to a load outside
3047 the loop, and a copy/store within the loop.
3050 When -fgcse-sm is enabled, A store motion pass is run after global common
3051 subexpression elimination. This pass will attempt to move stores out of loops.
3052 When used in conjunction with -fgcse-lm, loops containing a load/store sequence
3053 can be changed to a load before the loop and a store after the loop.
3055 @item -fdelete-null-pointer-checks
3056 Use global dataflow analysis to identify and eliminate useless null
3057 pointer checks. Programs which rely on NULL pointer dereferences @emph{not}
3058 halting the program may not work properly with this option. Use
3059 -fno-delete-null-pointer-checks to disable this optimizing for programs
3060 which depend on that behavior.
3062 @item -fexpensive-optimizations
3063 Perform a number of minor optimizations that are relatively expensive.
3065 @item -foptimize-register-move
3067 Attempt to reassign register numbers in move instructions and as
3068 operands of other simple instructions in order to maximize the amount of
3069 register tying. This is especially helpful on machines with two-operand
3070 instructions. GCC enables this optimization by default with @samp{-O2}
3073 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
3076 @item -fdelayed-branch
3077 If supported for the target machine, attempt to reorder instructions
3078 to exploit instruction slots available after delayed branch
3081 @item -fschedule-insns
3082 If supported for the target machine, attempt to reorder instructions to
3083 eliminate execution stalls due to required data being unavailable. This
3084 helps machines that have slow floating point or memory load instructions
3085 by allowing other instructions to be issued until the result of the load
3086 or floating point instruction is required.
3088 @item -fschedule-insns2
3089 Similar to @samp{-fschedule-insns}, but requests an additional pass of
3090 instruction scheduling after register allocation has been done. This is
3091 especially useful on machines with a relatively small number of
3092 registers and where memory load instructions take more than one cycle.
3094 @item -ffunction-sections
3095 @itemx -fdata-sections
3096 Place each function or data item into its own section in the output
3097 file if the target supports arbitrary sections. The name of the
3098 function or the name of the data item determines the section's name
3101 Use these options on systems where the linker can perform optimizations
3102 to improve locality of reference in the instruction space. HPPA
3103 processors running HP-UX and Sparc processors running Solaris 2 have
3104 linkers with such optimizations. Other systems using the ELF object format
3105 as well as AIX may have these optimizations in the future.
3107 Only use these options when there are significant benefits from doing
3108 so. When you specify these options, the assembler and linker will
3109 create larger object and executable files and will also be slower.
3110 You will not be able to use @code{gprof} on all systems if you
3111 specify this option and you may have problems with debugging if
3112 you specify both this option and @samp{-g}.
3114 @item -fcaller-saves
3115 Enable values to be allocated in registers that will be clobbered by
3116 function calls, by emitting extra instructions to save and restore the
3117 registers around such calls. Such allocation is done only when it
3118 seems to result in better code than would otherwise be produced.
3120 This option is always enabled by default on certain machines, usually
3121 those which have no call-preserved registers to use instead.
3123 For all machines, optimization level 2 and higher enables this flag by
3126 @item -funroll-loops
3127 Perform the optimization of loop unrolling. This is only done for loops
3128 whose number of iterations can be determined at compile time or run time.
3129 @samp{-funroll-loops} implies both @samp{-fstrength-reduce} and
3130 @samp{-frerun-cse-after-loop}.
3132 @item -funroll-all-loops
3133 Perform the optimization of loop unrolling. This is done for all loops
3134 and usually makes programs run more slowly. @samp{-funroll-all-loops}
3135 implies @samp{-fstrength-reduce} as well as @samp{-frerun-cse-after-loop}.
3137 @item -fmove-all-movables
3138 Forces all invariant computations in loops to be moved
3141 @item -freduce-all-givs
3142 Forces all general-induction variables in loops to be
3145 @emph{Note:} When compiling programs written in Fortran,
3146 @samp{-fmove-all-movables} and @samp{-freduce-all-givs} are enabled
3147 by default when you use the optimizer.
3149 These options may generate better or worse code; results are highly
3150 dependent on the structure of loops within the source code.
3152 These two options are intended to be removed someday, once
3153 they have helped determine the efficacy of various
3154 approaches to improving loop optimizations.
3156 Please let us (@w{@email{gcc@@gcc.gnu.org}} and @w{@email{fortran@@gnu.org}})
3157 know how use of these options affects
3158 the performance of your production code.
3159 We're very interested in code that runs @emph{slower}
3160 when these options are @emph{enabled}.
3163 Disable any machine-specific peephole optimizations.
3165 @item -fbranch-probabilities
3166 After running a program compiled with @samp{-fprofile-arcs}
3167 (@pxref{Debugging Options,, Options for Debugging Your Program or
3168 @command{gcc}}), you can compile it a second time using
3169 @samp{-fbranch-probabilities}, to improve optimizations based on
3170 guessing the path a branch might take.
3173 With @samp{-fbranch-probabilities}, GCC puts a @samp{REG_EXEC_COUNT}
3174 note on the first instruction of each basic block, and a
3175 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
3176 These can be used to improve optimization. Currently, they are only
3177 used in one place: in @file{reorg.c}, instead of guessing which path a
3178 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
3179 exactly determine which path is taken more often.
3182 @item -fno-guess-branch-probability
3183 Sometimes gcc will opt to guess branch probabilities when none are
3184 available from either profile directed feedback (@samp{-fprofile-arcs})
3185 or @samp{__builtin_expect}. In a hard real-time system, people don't
3186 want different runs of the compiler to produce code that has different
3187 behavior; minimizing non-determinism is of paramount import. This
3188 switch allows users to reduce non-determinism, possibly at the expense
3189 of inferior optimization.
3191 @item -fstrict-aliasing
3192 Allows the compiler to assume the strictest aliasing rules applicable to
3193 the language being compiled. For C (and C++), this activates
3194 optimizations based on the type of expressions. In particular, an
3195 object of one type is assumed never to reside at the same address as an
3196 object of a different type, unless the types are almost the same. For
3197 example, an @code{unsigned int} can alias an @code{int}, but not a
3198 @code{void*} or a @code{double}. A character type may alias any other
3201 Pay special attention to code like this:
3214 The practice of reading from a different union member than the one most
3215 recently written to (called ``type-punning'') is common. Even with
3216 @samp{-fstrict-aliasing}, type-punning is allowed, provided the memory
3217 is accessed through the union type. So, the code above will work as
3218 expected. However, this code might not:
3230 Every language that wishes to perform language-specific alias analysis
3231 should define a function that computes, given an @code{tree}
3232 node, an alias set for the node. Nodes in different alias sets are not
3233 allowed to alias. For an example, see the C front-end function
3234 @code{c_get_alias_set}.
3237 @item -falign-functions
3238 @itemx -falign-functions=@var{n}
3239 Align the start of functions to the next power-of-two greater than
3240 @var{n}, skipping up to @var{n} bytes. For instance,
3241 @samp{-falign-functions=32} aligns functions to the next 32-byte
3242 boundary, but @samp{-falign-functions=24} would align to the next
3243 32-byte boundary only if this can be done by skipping 23 bytes or less.
3245 @samp{-fno-align-functions} and @samp{-falign-functions=1} are
3246 equivalent and mean that functions will not be aligned.
3248 Some assemblers only support this flag when @var{n} is a power of two;
3249 in that case, it is rounded up.
3251 If @var{n} is not specified, use a machine-dependent default.
3253 @item -falign-labels
3254 @itemx -falign-labels=@var{n}
3255 Align all branch targets to a power-of-two boundary, skipping up to
3256 @var{n} bytes like @samp{-falign-functions}. This option can easily
3257 make code slower, because it must insert dummy operations for when the
3258 branch target is reached in the usual flow of the code.
3260 If @samp{-falign-loops} or @samp{-falign-jumps} are applicable and
3261 are greater than this value, then their values are used instead.
3263 If @var{n} is not specified, use a machine-dependent default which is
3264 very likely to be @samp{1}, meaning no alignment.
3267 @itemx -falign-loops=@var{n}
3268 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
3269 like @samp{-falign-functions}. The hope is that the loop will be
3270 executed many times, which will make up for any execution of the dummy
3273 If @var{n} is not specified, use a machine-dependent default.
3276 @itemx -falign-jumps=@var{n}
3277 Align branch targets to a power-of-two boundary, for branch targets
3278 where the targets can only be reached by jumping, skipping up to @var{n}
3279 bytes like @samp{-falign-functions}. In this case, no dummy operations
3282 If @var{n} is not specified, use a machine-dependent default.
3285 Perform optimizations in static single assignment form. Each function's
3286 flow graph is translated into SSA form, optimizations are performed, and
3287 the flow graph is translated back from SSA form. User's should not
3288 specify this option, since it is not yet ready for production use.
3291 Perform dead-code elimination in SSA form. Requires @samp{-fssa}. Like
3292 @samp{-fssa}, this is an experimental feature.
3294 @item -fsingle-precision-constant
3295 Treat floating point constant as single precision constant instead of
3296 implicitly converting it to double precision constant.
3298 @item -frename-registers
3299 Attempt to avoid false dependancies in scheduled code by making use
3300 of registers left over after register allocation. This optimization
3301 will most benefit processors with lots of registers. It can, however,
3302 make debugging impossible, since variables will no longer stay in
3303 a ``home register''.
3305 @item --param @var{name}=@var{value}
3306 In some places, GCC uses various constants to control the amount of
3307 optimization that is done. For example, GCC will not inline functions
3308 that contain more that a certain number of instructions. You can
3309 control some of these constants on the command-line using the
3310 @samp{--param} option.
3312 In each case, the @var{value} is a integer. The allowable choices for
3313 @var{name} are given in the following table:
3316 @item max-inline-insns
3317 If an function contains more than this many instructions, it
3318 will not be inlined. This option is precisely equivalent to
3319 @samp{-finline-limit}.
3324 @node Preprocessor Options
3325 @section Options Controlling the Preprocessor
3326 @cindex preprocessor options
3327 @cindex options, preprocessor
3329 These options control the C preprocessor, which is run on each C source
3330 file before actual compilation.
3332 If you use the @samp{-E} option, nothing is done except preprocessing.
3333 Some of these options make sense only together with @samp{-E} because
3334 they cause the preprocessor output to be unsuitable for actual
3338 @item -include @var{file}
3339 Process @var{file} as input before processing the regular input file.
3340 In effect, the contents of @var{file} are compiled first. Any @samp{-D}
3341 and @samp{-U} options on the command line are always processed before
3342 @samp{-include @var{file}}, regardless of the order in which they are
3343 written. All the @samp{-include} and @samp{-imacros} options are
3344 processed in the order in which they are written.
3346 @item -imacros @var{file}
3347 Process @var{file} as input, discarding the resulting output, before
3348 processing the regular input file. Because the output generated from
3349 @var{file} is discarded, the only effect of @samp{-imacros @var{file}}
3350 is to make the macros defined in @var{file} available for use in the
3351 main input. All the @samp{-include} and @samp{-imacros} options are
3352 processed in the order in which they are written.
3354 @item -idirafter @var{dir}
3355 @cindex second include path
3356 Add the directory @var{dir} to the second include path. The directories
3357 on the second include path are searched when a header file is not found
3358 in any of the directories in the main include path (the one that
3361 @item -iprefix @var{prefix}
3362 Specify @var{prefix} as the prefix for subsequent @samp{-iwithprefix}
3365 @item -iwithprefix @var{dir}
3366 Add a directory to the second include path. The directory's name is
3367 made by concatenating @var{prefix} and @var{dir}, where @var{prefix} was
3368 specified previously with @samp{-iprefix}. If you have not specified a
3369 prefix yet, the directory containing the installed passes of the
3370 compiler is used as the default.
3372 @item -iwithprefixbefore @var{dir}
3373 Add a directory to the main include path. The directory's name is made
3374 by concatenating @var{prefix} and @var{dir}, as in the case of
3375 @samp{-iwithprefix}.
3377 @item -isystem @var{dir}
3378 Add a directory to the beginning of the second include path, marking it
3379 as a system directory, so that it gets the same special treatment as
3380 is applied to the standard system directories.
3383 Do not search the standard system directories for header files. Only
3384 the directories you have specified with @samp{-I} options (and the
3385 current directory, if appropriate) are searched. @xref{Directory
3386 Options}, for information on @samp{-I}.
3388 By using both @samp{-nostdinc} and @samp{-I-}, you can limit the include-file
3389 search path to only those directories you specify explicitly.
3393 When searching for a header file in a directory, remap file names if a
3394 file named @file{header.gcc} exists in that directory. This can be used
3395 to work around limitations of file systems with file name restrictions.
3396 The @file{header.gcc} file should contain a series of lines with two
3397 tokens on each line: the first token is the name to map, and the second
3398 token is the actual name to use.
3401 Do not predefine any nonstandard macros. (Including architecture flags).
3404 Run only the C preprocessor. Preprocess all the C source files
3405 specified and output the results to standard output or to the
3406 specified output file.
3409 Tell the preprocessor not to discard comments. Used with the
3413 Tell the preprocessor not to generate @samp{#line} directives.
3414 Used with the @samp{-E} option.
3417 @cindex dependencies, make
3420 Instead of outputting the result of preprocessing, output a rule
3421 suitable for @code{make} describing the dependencies of the main source
3422 file. The preprocessor outputs one @code{make} rule containing the
3423 object file name for that source file, a colon, and the names of all the
3424 included files. Unless overridden explicitly, the object file name
3425 consists of the basename of the source file with any suffix replaced with
3426 object file suffix. If there are many included files then the
3427 rule is split into several lines using @samp{\}-newline.
3429 @samp{-M} implies @samp{-E}.
3433 Like @samp{-M}, but mention only the files included with @samp{#include
3434 "@var{file}"}. System header files included with @samp{#include
3435 <@var{file}>} are omitted.
3439 Like @samp{-M} but the dependency information is written to a file
3440 rather than stdout. @code{gcc} will use the same file name and
3441 directory as the object file, but with the suffix ".d" instead.
3443 This is in addition to compiling the main file as specified ---
3444 @samp{-MD} does not inhibit ordinary compilation the way @samp{-M} does,
3445 unless you also specify @samp{-MG}.
3447 With Mach, you can use the utility @code{md} to merge multiple
3448 dependency files into a single dependency file suitable for using with
3449 the @samp{make} command.
3453 Like @samp{-MD} except mention only user header files, not system
3456 @item -MF @var{file}
3458 When used with @samp{-M} or @samp{-MM}, specifies a file to write the
3459 dependencies to. This allows the preprocessor to write the preprocessed
3460 file to stdout normally. If no @samp{-MF} switch is given, CPP sends
3461 the rules to stdout and suppresses normal preprocessed output.
3463 Another way to specify output of a @code{make} rule is by setting
3464 the environment variable @env{DEPENDENCIES_OUTPUT} (@pxref{Environment
3469 When used with @samp{-M} or @samp{-MM}, @samp{-MG} says to treat missing
3470 header files as generated files and assume they live in the same
3471 directory as the source file. It suppresses preprocessed output, as a
3472 missing header file is ordinarily an error.
3474 This feature is used in automatic updating of makefiles.
3478 This option instructs CPP to add a phony target for each dependency
3479 other than the main file, causing each to depend on nothing. These
3480 dummy rules work around errors @code{make} gives if you remove header
3481 files without updating the @code{Makefile} to match.
3483 This is typical output:-
3486 /tmp/test.o: /tmp/test.c /tmp/test.h
3491 @item -MQ @var{target}
3492 @item -MT @var{target}
3495 By default CPP uses the main file name, including any path, and appends
3496 the object suffix, normally ``.o'', to it to obtain the name of the
3497 target for dependency generation. With @samp{-MT} you can specify a
3498 target yourself, overriding the default one.
3500 If you want multiple targets, you can specify them as a single argument
3501 to @samp{-MT}, or use multiple @samp{-MT} options.
3503 The targets you specify are output in the order they appear on the
3504 command line. @samp{-MQ} is identical to @samp{-MT}, except that the
3505 target name is quoted for Make, but with @samp{-MT} it isn't. For
3506 example, -MT '$(objpfx)foo.o' gives
3509 $(objpfx)foo.o: /tmp/foo.c
3512 but -MQ '$(objpfx)foo.o' gives
3515 $$(objpfx)foo.o: /tmp/foo.c
3518 The default target is automatically quoted, as if it were given with
3522 Print the name of each header file used, in addition to other normal
3525 @item -A@var{question}(@var{answer})
3526 Assert the answer @var{answer} for @var{question}, in case it is tested
3527 with a preprocessing conditional such as @samp{#if
3528 #@var{question}(@var{answer})}. @samp{-A-} disables the standard
3529 assertions that normally describe the target machine.
3532 Define macro @var{macro} with the string @samp{1} as its definition.
3534 @item -D@var{macro}=@var{defn}
3535 Define macro @var{macro} as @var{defn}. All instances of @samp{-D} on
3536 the command line are processed before any @samp{-U} options.
3538 Any @samp{-D} and @samp{-U} options on the command line are processed in
3539 order, and always before @samp{-imacros @var{file}}, regardless of the
3540 order in which they are written.
3543 Undefine macro @var{macro}. @samp{-U} options are evaluated after all
3544 @samp{-D} options, but before any @samp{-include} and @samp{-imacros}
3547 Any @samp{-D} and @samp{-U} options on the command line are processed in
3548 order, and always before @samp{-imacros @var{file}}, regardless of the
3549 order in which they are written.
3552 Tell the preprocessor to output only a list of the macro definitions
3553 that are in effect at the end of preprocessing. Used with the @samp{-E}
3557 Tell the preprocessing to pass all macro definitions into the output, in
3558 their proper sequence in the rest of the output.
3561 Like @samp{-dD} except that the macro arguments and contents are omitted.
3562 Only @samp{#define @var{name}} is included in the output.
3566 Output @samp{#include} directives in addition to the result of
3571 Process ISO standard trigraph sequences. These are three-character
3572 sequences, all starting with @samp{??}, that are defined by ISO C to
3573 stand for single characters. For example, @samp{??/} stands for
3574 @samp{\}, so @samp{'??/n'} is a character constant for a newline. By
3575 default, GCC ignores trigraphs, but in standard-conforming modes it
3576 converts them. See the @samp{-std} and @samp{-ansi} options.
3578 The nine trigraph sequences are
3609 Trigraph support is not popular, so many compilers do not implement it
3610 properly. Portable code should not rely on trigraphs being either
3611 converted or ignored.
3613 @item -Wp\,@var{option}
3614 Pass @var{option} as an option to the preprocessor. If @var{option}
3615 contains commas, it is split into multiple options at the commas.
3618 @node Assembler Options
3619 @section Passing Options to the Assembler
3621 @c prevent bad page break with this line
3622 You can pass options to the assembler.
3625 @item -Wa\,@var{option}
3626 Pass @var{option} as an option to the assembler. If @var{option}
3627 contains commas, it is split into multiple options at the commas.
3631 @section Options for Linking
3632 @cindex link options
3633 @cindex options, linking
3635 These options come into play when the compiler links object files into
3636 an executable output file. They are meaningless if the compiler is
3637 not doing a link step.
3641 @item @var{object-file-name}
3642 A file name that does not end in a special recognized suffix is
3643 considered to name an object file or library. (Object files are
3644 distinguished from libraries by the linker according to the file
3645 contents.) If linking is done, these object files are used as input
3651 If any of these options is used, then the linker is not run, and
3652 object file names should not be used as arguments. @xref{Overall
3656 @item -l@var{library}
3657 Search the library named @var{library} when linking.
3659 It makes a difference where in the command you write this option; the
3660 linker searches processes libraries and object files in the order they
3661 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
3662 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
3663 to functions in @samp{z}, those functions may not be loaded.
3665 The linker searches a standard list of directories for the library,
3666 which is actually a file named @file{lib@var{library}.a}. The linker
3667 then uses this file as if it had been specified precisely by name.
3669 The directories searched include several standard system directories
3670 plus any that you specify with @samp{-L}.
3672 Normally the files found this way are library files---archive files
3673 whose members are object files. The linker handles an archive file by
3674 scanning through it for members which define symbols that have so far
3675 been referenced but not defined. But if the file that is found is an
3676 ordinary object file, it is linked in the usual fashion. The only
3677 difference between using an @samp{-l} option and specifying a file name
3678 is that @samp{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
3679 and searches several directories.
3682 You need this special case of the @samp{-l} option in order to
3683 link an Objective C program.
3686 Do not use the standard system startup files when linking.
3687 The standard system libraries are used normally, unless @option{-nostdlib}
3688 or @option{-nodefaultlibs} is used.
3690 @item -nodefaultlibs
3691 Do not use the standard system libraries when linking.
3692 Only the libraries you specify will be passed to the linker.
3693 The standard startup files are used normally, unless @option{-nostartfiles}
3694 is used. The compiler may generate calls to memcmp, memset, and memcpy
3695 for System V (and ISO C) environments or to bcopy and bzero for
3696 BSD environments. These entries are usually resolved by entries in
3697 libc. These entry points should be supplied through some other
3698 mechanism when this option is specified.
3701 Do not use the standard system startup files or libraries when linking.
3702 No startup files and only the libraries you specify will be passed to
3703 the linker. The compiler may generate calls to memcmp, memset, and memcpy
3704 for System V (and ISO C) environments or to bcopy and bzero for
3705 BSD environments. These entries are usually resolved by entries in
3706 libc. These entry points should be supplied through some other
3707 mechanism when this option is specified.
3709 @cindex @code{-lgcc}, use with @code{-nostdlib}
3710 @cindex @code{-nostdlib} and unresolved references
3711 @cindex unresolved references and @code{-nostdlib}
3712 @cindex @code{-lgcc}, use with @code{-nodefaultlibs}
3713 @cindex @code{-nodefaultlibs} and unresolved references
3714 @cindex unresolved references and @code{-nodefaultlibs}
3715 One of the standard libraries bypassed by @samp{-nostdlib} and
3716 @samp{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
3717 that GCC uses to overcome shortcomings of particular machines, or special
3718 needs for some languages.
3720 (@xref{Interface,,Interfacing to GCC Output}, for more discussion of
3724 (@xref{Interface,,Interfacing to GCC Output,gcc.info,Porting GCC},
3725 for more discussion of @file{libgcc.a}.)
3727 In most cases, you need @file{libgcc.a} even when you want to avoid
3728 other standard libraries. In other words, when you specify @samp{-nostdlib}
3729 or @samp{-nodefaultlibs} you should usually specify @samp{-lgcc} as well.
3730 This ensures that you have no unresolved references to internal GCC
3731 library subroutines. (For example, @samp{__main}, used to ensure C++
3732 constructors will be called; @pxref{Collect2,,@command{collect2}}.)
3735 Remove all symbol table and relocation information from the executable.
3738 On systems that support dynamic linking, this prevents linking with the shared
3739 libraries. On other systems, this option has no effect.
3742 Produce a shared object which can then be linked with other objects to
3743 form an executable. Not all systems support this option. For predictable
3744 results, you must also specify the same set of options that were used to
3745 generate code (@samp{-fpic}, @samp{-fPIC}, or model suboptions)
3746 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
3747 needs to build supplementary stub code for constructors to work. On
3748 multi-libbed systems, @samp{gcc -shared} must select the correct support
3749 libraries to link against. Failing to supply the correct flags may lead
3750 to subtle defects. Supplying them in cases where they are not necessary
3753 @item -shared-libgcc
3754 @itemx -static-libgcc
3755 On systems that provide @file{libgcc} as a shared library, these options
3756 force the use of either the shared or static version respectively.
3757 If no shared version of @file{libgcc} was built when the compiler was
3758 configured, these options have no effect.
3760 There are several situations in which an application should use the
3761 shared @file{libgcc} instead of the static version. The most common
3762 of these is when the application wishes to throw and catch exceptions
3763 across different shared libraries. In that case, each of the libraries
3764 as well as the application itself should use the shared @file{libgcc}.
3766 Therefore, whenever you specify the @samp{-shared} option, the GCC
3767 driver automatically adds @samp{-shared-libgcc}, unless you explicitly
3768 specify @samp{-static-libgcc}. The G++ driver automatically adds
3769 @samp{-shared-libgcc} when you build a main executable as well because
3770 for C++ programs that is typically the right thing to do.
3771 (Exception-handling will not work reliably otherwise.)
3773 However, when linking a main executable written in C, you must
3774 explicitly say @samp{-shared-libgcc} if you want to use the shared
3778 Bind references to global symbols when building a shared object. Warn
3779 about any unresolved references (unless overridden by the link editor
3780 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
3783 @item -Xlinker @var{option}
3784 Pass @var{option} as an option to the linker. You can use this to
3785 supply system-specific linker options which GCC does not know how to
3788 If you want to pass an option that takes an argument, you must use
3789 @samp{-Xlinker} twice, once for the option and once for the argument.
3790 For example, to pass @samp{-assert definitions}, you must write
3791 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
3792 @samp{-Xlinker "-assert definitions"}, because this passes the entire
3793 string as a single argument, which is not what the linker expects.
3795 @item -Wl\,@var{option}
3796 Pass @var{option} as an option to the linker. If @var{option} contains
3797 commas, it is split into multiple options at the commas.
3799 @item -u @var{symbol}
3800 Pretend the symbol @var{symbol} is undefined, to force linking of
3801 library modules to define it. You can use @samp{-u} multiple times with
3802 different symbols to force loading of additional library modules.
3805 @node Directory Options
3806 @section Options for Directory Search
3807 @cindex directory options
3808 @cindex options, directory search
3811 These options specify directories to search for header files, for
3812 libraries and for parts of the compiler:
3816 Add the directory @var{dir} to the head of the list of directories to be
3817 searched for header files. This can be used to override a system header
3818 file, substituting your own version, since these directories are
3819 searched before the system header file directories. However, you should
3820 not use this option to add directories that contain vendor-supplied
3821 system header files (use @samp{-isystem} for that). If you use more than
3822 one @samp{-I} option, the directories are scanned in left-to-right
3823 order; the standard system directories come after.
3826 Any directories you specify with @samp{-I} options before the @samp{-I-}
3827 option are searched only for the case of @samp{#include "@var{file}"};
3828 they are not searched for @samp{#include <@var{file}>}.
3830 If additional directories are specified with @samp{-I} options after
3831 the @samp{-I-}, these directories are searched for all @samp{#include}
3832 directives. (Ordinarily @emph{all} @samp{-I} directories are used
3835 In addition, the @samp{-I-} option inhibits the use of the current
3836 directory (where the current input file came from) as the first search
3837 directory for @samp{#include "@var{file}"}. There is no way to
3838 override this effect of @samp{-I-}. With @samp{-I.} you can specify
3839 searching the directory which was current when the compiler was
3840 invoked. That is not exactly the same as what the preprocessor does
3841 by default, but it is often satisfactory.
3843 @samp{-I-} does not inhibit the use of the standard system directories
3844 for header files. Thus, @samp{-I-} and @samp{-nostdinc} are
3848 Add directory @var{dir} to the list of directories to be searched
3851 @item -B@var{prefix}
3852 This option specifies where to find the executables, libraries,
3853 include files, and data files of the compiler itself.
3855 The compiler driver program runs one or more of the subprograms
3856 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
3857 @var{prefix} as a prefix for each program it tries to run, both with and
3858 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
3860 For each subprogram to be run, the compiler driver first tries the
3861 @samp{-B} prefix, if any. If that name is not found, or if @samp{-B}
3862 was not specified, the driver tries two standard prefixes, which are
3863 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc-lib/}. If neither of
3864 those results in a file name that is found, the unmodified program
3865 name is searched for using the directories specified in your
3866 @env{PATH} environment variable.
3868 @samp{-B} prefixes that effectively specify directory names also apply
3869 to libraries in the linker, because the compiler translates these
3870 options into @samp{-L} options for the linker. They also apply to
3871 includes files in the preprocessor, because the compiler translates these
3872 options into @samp{-isystem} options for the preprocessor. In this case,
3873 the compiler appends @samp{include} to the prefix.
3875 The run-time support file @file{libgcc.a} can also be searched for using
3876 the @samp{-B} prefix, if needed. If it is not found there, the two
3877 standard prefixes above are tried, and that is all. The file is left
3878 out of the link if it is not found by those means.
3880 Another way to specify a prefix much like the @samp{-B} prefix is to use
3881 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
3884 @item -specs=@var{file}
3885 Process @var{file} after the compiler reads in the standard @file{specs}
3886 file, in order to override the defaults that the @file{gcc} driver
3887 program uses when determining what switches to pass to @file{cc1},
3888 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
3889 @samp{-specs=}@var{file} can be specified on the command line, and they
3890 are processed in order, from left to right.
3896 @section Specifying subprocesses and the switches to pass to them
3898 @command{gcc} is a driver program. It performs its job by invoking a
3899 sequence of other programs to do the work of compiling, assembling and
3900 linking. GCC interprets its command-line parameters and uses these to
3901 deduce which programs it should invoke, and which command-line options
3902 it ought to place on their command lines. This behaviour is controlled
3903 by @dfn{spec strings}. In most cases there is one spec string for each
3904 program that GCC can invoke, but a few programs have multiple spec
3905 strings to control their behaviour. The spec strings built into GCC can
3906 be overridden by using the @samp{-specs=} command-line switch to specify
3909 @dfn{Spec files} are plaintext files that are used to construct spec
3910 strings. They consist of a sequence of directives separated by blank
3911 lines. The type of directive is determined by the first non-whitespace
3912 character on the line and it can be one of the following:
3915 @item %@var{command}
3916 Issues a @var{command} to the spec file processor. The commands that can
3920 @item %include <@var{file}>
3922 Search for @var{file} and insert its text at the current point in the
3925 @item %include_noerr <@var{file}>
3926 @cindex %include_noerr
3927 Just like @samp{%include}, but do not generate an error message if the include
3928 file cannot be found.
3930 @item %rename @var{old_name} @var{new_name}
3932 Rename the spec string @var{old_name} to @var{new_name}.
3936 @item *[@var{spec_name}]:
3937 This tells the compiler to create, override or delete the named spec
3938 string. All lines after this directive up to the next directive or
3939 blank line are considered to be the text for the spec string. If this
3940 results in an empty string then the spec will be deleted. (Or, if the
3941 spec did not exist, then nothing will happened.) Otherwise, if the spec
3942 does not currently exist a new spec will be created. If the spec does
3943 exist then its contents will be overridden by the text of this
3944 directive, unless the first character of that text is the @samp{+}
3945 character, in which case the text will be appended to the spec.
3947 @item [@var{suffix}]:
3948 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
3949 and up to the next directive or blank line are considered to make up the
3950 spec string for the indicated suffix. When the compiler encounters an
3951 input file with the named suffix, it will processes the spec string in
3952 order to work out how to compile that file. For example:
3959 This says that any input file whose name ends in @samp{.ZZ} should be
3960 passed to the program @samp{z-compile}, which should be invoked with the
3961 command-line switch @samp{-input} and with the result of performing the
3962 @samp{%i} substitution. (See below.)
3964 As an alternative to providing a spec string, the text that follows a
3965 suffix directive can be one of the following:
3968 @item @@@var{language}
3969 This says that the suffix is an alias for a known @var{language}. This is
3970 similar to using the @option{-x} command-line switch to GCC to specify a
3971 language explicitly. For example:
3978 Says that .ZZ files are, in fact, C++ source files.
3981 This causes an error messages saying:
3984 @var{name} compiler not installed on this system.
3988 GCC already has an extensive list of suffixes built into it.
3989 This directive will add an entry to the end of the list of suffixes, but
3990 since the list is searched from the end backwards, it is effectively
3991 possible to override earlier entries using this technique.
3995 GCC has the following spec strings built into it. Spec files can
3996 override these strings or create their own. Note that individual
3997 targets can also add their own spec strings to this list.
4000 asm Options to pass to the assembler
4001 asm_final Options to pass to the assembler post-processor
4002 cpp Options to pass to the C preprocessor
4003 cc1 Options to pass to the C compiler
4004 cc1plus Options to pass to the C++ compiler
4005 endfile Object files to include at the end of the link
4006 link Options to pass to the linker
4007 lib Libraries to include on the command line to the linker
4008 libgcc Decides which GCC support library to pass to the linker
4009 linker Sets the name of the linker
4010 predefines Defines to be passed to the C preprocessor
4011 signed_char Defines to pass to CPP to say whether @code{char} is signed by default
4012 startfile Object files to include at the start of the link
4015 Here is a small example of a spec file:
4021 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
4024 This example renames the spec called @samp{lib} to @samp{old_lib} and
4025 then overrides the previous definition of @samp{lib} with a new one.
4026 The new definition adds in some extra command-line options before
4027 including the text of the old definition.
4029 @dfn{Spec strings} are a list of command-line options to be passed to their
4030 corresponding program. In addition, the spec strings can contain
4031 @samp{%}-prefixed sequences to substitute variable text or to
4032 conditionally insert text into the command line. Using these constructs
4033 it is possible to generate quite complex command lines.
4035 Here is a table of all defined @samp{%}-sequences for spec
4036 strings. Note that spaces are not generated automatically around the
4037 results of expanding these sequences. Therefore you can concatenate them
4038 together or combine them with constant text in a single argument.
4042 Substitute one @samp{%} into the program name or argument.
4045 Substitute the name of the input file being processed.
4048 Substitute the basename of the input file being processed.
4049 This is the substring up to (and not including) the last period
4050 and not including the directory.
4053 This is the same as @samp{%b}, but include the file suffix (text after
4057 Marks the argument containing or following the @samp{%d} as a
4058 temporary file name, so that that file will be deleted if GCC exits
4059 successfully. Unlike @samp{%g}, this contributes no text to the
4062 @item %g@var{suffix}
4063 Substitute a file name that has suffix @var{suffix} and is chosen
4064 once per compilation, and mark the argument in the same way as
4065 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
4066 name is now chosen in a way that is hard to predict even when previously
4067 chosen file names are known. For example, @samp{%g.s ... %g.o ... %g.s}
4068 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
4069 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
4070 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
4071 was simply substituted with a file name chosen once per compilation,
4072 without regard to any appended suffix (which was therefore treated
4073 just like ordinary text), making such attacks more likely to succeed.
4075 @item %u@var{suffix}
4076 Like @samp{%g}, but generates a new temporary file name even if
4077 @samp{%u@var{suffix}} was already seen.
4079 @item %U@var{suffix}
4080 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
4081 new one if there is no such last file name. In the absence of any
4082 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
4083 the same suffix @emph{space}, so @samp{%g.s ... %U.s ... %g.s ... %U.s}
4084 would involve the generation of two distinct file names, one
4085 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
4086 simply substituted with a file name chosen for the previous @samp{%u},
4087 without regard to any appended suffix.
4089 @item %j@var{SUFFIX}
4090 Substitutes the name of the HOST_BIT_BUCKET, if any, and if it is
4091 writable, and if save-temps is off; otherwise, substitute the name
4092 of a temporary file, just like @samp{%u}. This temporary file is not
4093 meant for communication between processes, but rather as a junk
4096 @item %.@var{SUFFIX}
4097 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
4098 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
4099 terminated by the next space or %.
4102 Marks the argument containing or following the @samp{%w} as the
4103 designated output file of this compilation. This puts the argument
4104 into the sequence of arguments that @samp{%o} will substitute later.
4107 Substitutes the names of all the output files, with spaces
4108 automatically placed around them. You should write spaces
4109 around the @samp{%o} as well or the results are undefined.
4110 @samp{%o} is for use in the specs for running the linker.
4111 Input files whose names have no recognized suffix are not compiled
4112 at all, but they are included among the output files, so they will
4116 Substitutes the suffix for object files. Note that this is
4117 handled specially when it immediately follows @samp{%g, %u, or %U},
4118 because of the need for those to form complete file names. The
4119 handling is such that @samp{%O} is treated exactly as if it had already
4120 been substituted, except that @samp{%g, %u, and %U} do not currently
4121 support additional @var{suffix} characters following @samp{%O} as they would
4122 following, for example, @samp{.o}.
4125 Substitutes the standard macro predefinitions for the
4126 current target machine. Use this when running @code{cpp}.
4129 Like @samp{%p}, but puts @samp{__} before and after the name of each
4130 predefined macro, except for macros that start with @samp{__} or with
4131 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
4135 Substitute a @samp{-iprefix} option made from GCC_EXEC_PREFIX.
4138 Current argument is the name of a library or startup file of some sort.
4139 Search for that file in a standard list of directories and substitute
4140 the full name found.
4143 Print @var{str} as an error message. @var{str} is terminated by a newline.
4144 Use this when inconsistent options are detected.
4147 Output @samp{-} if the input for the current command is coming from a pipe.
4150 Substitute the contents of spec string @var{name} at this point.
4153 Like @samp{%(...)} but put @samp{__} around @samp{-D} arguments.
4155 @item %x@{@var{option}@}
4156 Accumulate an option for @samp{%X}.
4159 Output the accumulated linker options specified by @samp{-Wl} or a @samp{%x}
4163 Output the accumulated assembler options specified by @samp{-Wa}.
4166 Output the accumulated preprocessor options specified by @samp{-Wp}.
4169 Substitute the major version number of GCC.
4170 (For version 2.9.5, this is 2.)
4173 Substitute the minor version number of GCC.
4174 (For version 2.9.5, this is 9.)
4177 Substitute the patch level number of GCC.
4178 (For version 2.9.5, this is 5.)
4181 Process the @code{asm} spec. This is used to compute the
4182 switches to be passed to the assembler.
4185 Process the @code{asm_final} spec. This is a spec string for
4186 passing switches to an assembler post-processor, if such a program is
4190 Process the @code{link} spec. This is the spec for computing the
4191 command line passed to the linker. Typically it will make use of the
4192 @samp{%L %G %S %D and %E} sequences.
4195 Dump out a @samp{-L} option for each directory that GCC believes might
4196 contain startup files. If the target supports multilibs then the
4197 current multilib directory will be prepended to each of these paths.
4200 Output the multilib directory with directory seperators replaced with
4201 "_". If multilib directories are not set, or the multilib directory is
4202 "." then this option emits nothing.
4205 Process the @code{lib} spec. This is a spec string for deciding which
4206 libraries should be included on the command line to the linker.
4209 Process the @code{libgcc} spec. This is a spec string for deciding
4210 which GCC support library should be included on the command line to the linker.
4213 Process the @code{startfile} spec. This is a spec for deciding which
4214 object files should be the first ones passed to the linker. Typically
4215 this might be a file named @file{crt0.o}.
4218 Process the @code{endfile} spec. This is a spec string that specifies
4219 the last object files that will be passed to the linker.
4222 Process the @code{cpp} spec. This is used to construct the arguments
4223 to be passed to the C preprocessor.
4226 Process the @code{signed_char} spec. This is intended to be used
4227 to tell cpp whether a char is signed. It typically has the definition:
4229 %@{funsigned-char:-D__CHAR_UNSIGNED__@}
4233 Process the @code{cc1} spec. This is used to construct the options to be
4234 passed to the actual C compiler (@samp{cc1}).
4237 Process the @code{cc1plus} spec. This is used to construct the options to be
4238 passed to the actual C++ compiler (@samp{cc1plus}).
4241 Substitute the variable part of a matched option. See below.
4242 Note that each comma in the substituted string is replaced by
4246 Substitutes the @code{-S} switch, if that switch was given to GCC.
4247 If that switch was not specified, this substitutes nothing. Note that
4248 the leading dash is omitted when specifying this option, and it is
4249 automatically inserted if the substitution is performed. Thus the spec
4250 string @samp{%@{foo@}} would match the command-line option @samp{-foo}
4251 and would output the command line option @samp{-foo}.
4253 @item %W@{@code{S}@}
4254 Like %@{@code{S}@} but mark last argument supplied within as a file to be
4257 @item %@{@code{S}*@}
4258 Substitutes all the switches specified to GCC whose names start
4259 with @code{-S}, but which also take an argument. This is used for
4260 switches like @samp{-o, -D, -I}, etc. GCC considers @samp{-o foo} as being
4261 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
4262 text, including the space. Thus two arguments would be generated.
4264 @item %@{^@code{S}*@}
4265 Like %@{@code{S}*@}, but don't put a blank between a switch and its
4266 argument. Thus %@{^o*@} would only generate one argument, not two.
4268 @item %@{@code{S}*&@code{T}*@}
4269 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
4270 (the order of @code{S} and @code{T} in the spec is not significant).
4271 There can be any number of ampersand-separated variables; for each the
4272 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
4274 @item %@{<@code{S}@}
4275 Remove all occurrences of @code{-S} from the command line. Note - this
4276 command is position dependent. @samp{%} commands in the spec string
4277 before this option will see @code{-S}, @samp{%} commands in the spec
4278 string after this option will not.
4280 @item %@{@code{S}*:@code{X}@}
4281 Substitutes @code{X} if one or more switches whose names start with
4282 @code{-S} are specified to GCC. Note that the tail part of the
4283 @code{-S} option (i.e. the part matched by the @samp{*}) will be substituted
4284 for each occurrence of @samp{%*} within @code{X}.
4286 @item %@{@code{S}:@code{X}@}
4287 Substitutes @code{X}, but only if the @samp{-S} switch was given to GCC.
4289 @item %@{!@code{S}:@code{X}@}
4290 Substitutes @code{X}, but only if the @samp{-S} switch was @emph{not} given to GCC.
4292 @item %@{|@code{S}:@code{X}@}
4293 Like %@{@code{S}:@code{X}@}, but if no @code{S} switch, substitute @samp{-}.
4295 @item %@{|!@code{S}:@code{X}@}
4296 Like %@{!@code{S}:@code{X}@}, but if there is an @code{S} switch, substitute @samp{-}.
4298 @item %@{.@code{S}:@code{X}@}
4299 Substitutes @code{X}, but only if processing a file with suffix @code{S}.
4301 @item %@{!.@code{S}:@code{X}@}
4302 Substitutes @code{X}, but only if @emph{not} processing a file with suffix @code{S}.
4304 @item %@{@code{S}|@code{P}:@code{X}@}
4305 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC. This may be
4306 combined with @samp{!} and @samp{.} sequences as well, although they
4307 have a stronger binding than the @samp{|}. For example a spec string
4311 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
4314 will output the following command-line options from the following input
4315 command-line options:
4320 -d fred.c -foo -baz -boggle
4321 -d jim.d -bar -baz -boggle
4326 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or
4327 %@{!@code{S}:@code{X}@} construct may contain other nested @samp{%} constructs
4328 or spaces, or even newlines. They are processed as usual, as described
4331 The @samp{-O, -f, -m, and -W} switches are handled specifically in these
4332 constructs. If another value of @samp{-O} or the negated form of a @samp{-f, -m, or
4333 -W} switch is found later in the command line, the earlier switch
4334 value is ignored, except with @{@code{S}*@} where @code{S} is just one
4335 letter, which passes all matching options.
4337 The character @samp{|} at the beginning of the predicate text is used to indicate
4338 that a command should be piped to the following command, but only if @samp{-pipe}
4341 It is built into GCC which switches take arguments and which do not.
4342 (You might think it would be useful to generalize this to allow each
4343 compiler's spec to say which switches take arguments. But this cannot
4344 be done in a consistent fashion. GCC cannot even decide which input
4345 files have been specified without knowing which switches take arguments,
4346 and it must know which input files to compile in order to tell which
4349 GCC also knows implicitly that arguments starting in @samp{-l} are to be
4350 treated as compiler output files, and passed to the linker in their
4351 proper position among the other output files.
4353 @c man begin OPTIONS
4355 @node Target Options
4356 @section Specifying Target Machine and Compiler Version
4357 @cindex target options
4358 @cindex cross compiling
4359 @cindex specifying machine version
4360 @cindex specifying compiler version and target machine
4361 @cindex compiler version, specifying
4362 @cindex target machine, specifying
4364 By default, GCC compiles code for the same type of machine that you
4365 are using. However, it can also be installed as a cross-compiler, to
4366 compile for some other type of machine. In fact, several different
4367 configurations of GCC, for different target machines, can be
4368 installed side by side. Then you specify which one to use with the
4371 In addition, older and newer versions of GCC can be installed side
4372 by side. One of them (probably the newest) will be the default, but
4373 you may sometimes wish to use another.
4376 @item -b @var{machine}
4377 The argument @var{machine} specifies the target machine for compilation.
4378 This is useful when you have installed GCC as a cross-compiler.
4380 The value to use for @var{machine} is the same as was specified as the
4381 machine type when configuring GCC as a cross-compiler. For
4382 example, if a cross-compiler was configured with @samp{configure
4383 i386v}, meaning to compile for an 80386 running System V, then you
4384 would specify @samp{-b i386v} to run that cross compiler.
4386 When you do not specify @samp{-b}, it normally means to compile for
4387 the same type of machine that you are using.
4389 @item -V @var{version}
4390 The argument @var{version} specifies which version of GCC to run.
4391 This is useful when multiple versions are installed. For example,
4392 @var{version} might be @samp{2.0}, meaning to run GCC version 2.0.
4394 The default version, when you do not specify @samp{-V}, is the last
4395 version of GCC that you installed.
4398 The @samp{-b} and @samp{-V} options actually work by controlling part of
4399 the file name used for the executable files and libraries used for
4400 compilation. A given version of GCC, for a given target machine, is
4401 normally kept in the directory @file{/usr/local/lib/gcc-lib/@var{machine}/@var{version}}.@refill
4403 Thus, sites can customize the effect of @samp{-b} or @samp{-V} either by
4404 changing the names of these directories or adding alternate names (or
4405 symbolic links). If in directory @file{/usr/local/lib/gcc-lib/} the
4406 file @file{80386} is a link to the file @file{i386v}, then @samp{-b
4407 80386} becomes an alias for @samp{-b i386v}.
4409 In one respect, the @samp{-b} or @samp{-V} do not completely change
4410 to a different compiler: the top-level driver program @command{gcc}
4411 that you originally invoked continues to run and invoke the other
4412 executables (preprocessor, compiler per se, assembler and linker)
4413 that do the real work. However, since no real work is done in the
4414 driver program, it usually does not matter that the driver program
4415 in use is not the one for the specified target. It is common for the
4416 interface to the other executables to change incompatibly between
4417 compiler versions, so unless the version specified is very close to that
4418 of the driver (for example, @samp{-V 3.0} with a driver program from GCC
4419 version 3.0.1), use of @samp{-V} may not work; for example, using
4420 @samp{-V 2.95.2} will not work with a driver program from GCC 3.0.
4422 The only way that the driver program depends on the target machine is
4423 in the parsing and handling of special machine-specific options.
4424 However, this is controlled by a file which is found, along with the
4425 other executables, in the directory for the specified version and
4426 target machine. As a result, a single installed driver program adapts
4427 to any specified target machine, and sufficiently similar compiler
4430 The driver program executable does control one significant thing,
4431 however: the default version and target machine. Therefore, you can
4432 install different instances of the driver program, compiled for
4433 different targets or versions, under different names.
4435 For example, if the driver for version 2.0 is installed as @command{ogcc}
4436 and that for version 2.1 is installed as @command{gcc}, then the command
4437 @command{gcc} will use version 2.1 by default, while @command{ogcc} will use
4438 2.0 by default. However, you can choose either version with either
4439 command with the @samp{-V} option.
4441 @node Submodel Options
4442 @section Hardware Models and Configurations
4443 @cindex submodel options
4444 @cindex specifying hardware config
4445 @cindex hardware models and configurations, specifying
4446 @cindex machine dependent options
4448 Earlier we discussed the standard option @samp{-b} which chooses among
4449 different installed compilers for completely different target
4450 machines, such as Vax vs. 68000 vs. 80386.
4452 In addition, each of these target machine types can have its own
4453 special options, starting with @samp{-m}, to choose among various
4454 hardware models or configurations---for example, 68010 vs 68020,
4455 floating coprocessor or none. A single installed version of the
4456 compiler can compile for any model or configuration, according to the
4459 Some configurations of the compiler also support additional special
4460 options, usually for compatibility with other compilers on the same
4464 These options are defined by the macro @code{TARGET_SWITCHES} in the
4465 machine description. The default for the options is also defined by
4466 that macro, which enables you to change the defaults.
4482 * RS/6000 and PowerPC Options::
4487 * Intel 960 Options::
4488 * DEC Alpha Options::
4492 * System V Options::
4493 * TMS320C3x/C4x Options::
4503 @node M680x0 Options
4504 @subsection M680x0 Options
4505 @cindex M680x0 options
4507 These are the @samp{-m} options defined for the 68000 series. The default
4508 values for these options depends on which style of 68000 was selected when
4509 the compiler was configured; the defaults for the most common choices are
4515 Generate output for a 68000. This is the default
4516 when the compiler is configured for 68000-based systems.
4518 Use this option for microcontrollers with a 68000 or EC000 core,
4519 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
4523 Generate output for a 68020. This is the default
4524 when the compiler is configured for 68020-based systems.
4527 Generate output containing 68881 instructions for floating point.
4528 This is the default for most 68020 systems unless @samp{-nfp} was
4529 specified when the compiler was configured.
4532 Generate output for a 68030. This is the default when the compiler is
4533 configured for 68030-based systems.
4536 Generate output for a 68040. This is the default when the compiler is
4537 configured for 68040-based systems.
4539 This option inhibits the use of 68881/68882 instructions that have to be
4540 emulated by software on the 68040. Use this option if your 68040 does not
4541 have code to emulate those instructions.
4544 Generate output for a 68060. This is the default when the compiler is
4545 configured for 68060-based systems.
4547 This option inhibits the use of 68020 and 68881/68882 instructions that
4548 have to be emulated by software on the 68060. Use this option if your 68060
4549 does not have code to emulate those instructions.
4552 Generate output for a CPU32. This is the default
4553 when the compiler is configured for CPU32-based systems.
4555 Use this option for microcontrollers with a
4556 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
4557 68336, 68340, 68341, 68349 and 68360.
4560 Generate output for a 520X "coldfire" family cpu. This is the default
4561 when the compiler is configured for 520X-based systems.
4563 Use this option for microcontroller with a 5200 core, including
4564 the MCF5202, MCF5203, MCF5204 and MCF5202.
4568 Generate output for a 68040, without using any of the new instructions.
4569 This results in code which can run relatively efficiently on either a
4570 68020/68881 or a 68030 or a 68040. The generated code does use the
4571 68881 instructions that are emulated on the 68040.
4574 Generate output for a 68060, without using any of the new instructions.
4575 This results in code which can run relatively efficiently on either a
4576 68020/68881 or a 68030 or a 68040. The generated code does use the
4577 68881 instructions that are emulated on the 68060.
4580 Generate output containing Sun FPA instructions for floating point.
4583 Generate output containing library calls for floating point.
4584 @strong{Warning:} the requisite libraries are not available for all m68k
4585 targets. Normally the facilities of the machine's usual C compiler are
4586 used, but this can't be done directly in cross-compilation. You must
4587 make your own arrangements to provide suitable library functions for
4588 cross-compilation. The embedded targets @samp{m68k-*-aout} and
4589 @samp{m68k-*-coff} do provide software floating point support.
4592 Consider type @code{int} to be 16 bits wide, like @code{short int}.
4595 Do not use the bit-field instructions. The @samp{-m68000}, @samp{-mcpu32}
4596 and @samp{-m5200} options imply @w{@samp{-mnobitfield}}.
4599 Do use the bit-field instructions. The @samp{-m68020} option implies
4600 @samp{-mbitfield}. This is the default if you use a configuration
4601 designed for a 68020.
4604 Use a different function-calling convention, in which functions
4605 that take a fixed number of arguments return with the @code{rtd}
4606 instruction, which pops their arguments while returning. This
4607 saves one instruction in the caller since there is no need to pop
4608 the arguments there.
4610 This calling convention is incompatible with the one normally
4611 used on Unix, so you cannot use it if you need to call libraries
4612 compiled with the Unix compiler.
4614 Also, you must provide function prototypes for all functions that
4615 take variable numbers of arguments (including @code{printf});
4616 otherwise incorrect code will be generated for calls to those
4619 In addition, seriously incorrect code will result if you call a
4620 function with too many arguments. (Normally, extra arguments are
4621 harmlessly ignored.)
4623 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
4624 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
4627 @itemx -mno-align-int
4628 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
4629 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
4630 boundary (@samp{-malign-int}) or a 16-bit boundary (@samp{-mno-align-int}).
4631 Aligning variables on 32-bit boundaries produces code that runs somewhat
4632 faster on processors with 32-bit busses at the expense of more memory.
4634 @strong{Warning:} if you use the @samp{-malign-int} switch, GCC will
4635 align structures containing the above types differently than
4636 most published application binary interface specifications for the m68k.
4639 Use the pc-relative addressing mode of the 68000 directly, instead of
4640 using a global offset table. At present, this option implies -fpic,
4641 allowing at most a 16-bit offset for pc-relative addressing. -fPIC is
4642 not presently supported with -mpcrel, though this could be supported for
4643 68020 and higher processors.
4645 @item -mno-strict-align
4646 @itemx -mstrict-align
4647 @kindex -mstrict-align
4648 Do not (do) assume that unaligned memory references will be handled by
4653 @node M68hc1x Options
4654 @subsection M68hc1x Options
4655 @cindex M68hc1x options
4657 These are the @samp{-m} options defined for the 68hc11 and 68hc12
4658 microcontrollers. The default values for these options depends on
4659 which style of microcontroller was selected when the compiler was configured;
4660 the defaults for the most common choices are given below.
4665 Generate output for a 68HC11. This is the default
4666 when the compiler is configured for 68HC11-based systems.
4670 Generate output for a 68HC12. This is the default
4671 when the compiler is configured for 68HC12-based systems.
4674 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
4678 Consider type @code{int} to be 16 bits wide, like @code{short int}.
4680 @item -msoft-reg-count=@var{count}
4681 Specify the number of pseudo-soft registers which are used for the
4682 code generation. The maximum number is 32. Using more pseudo-soft
4683 register may or may not result in better code depending on the program.
4684 The default is 4 for 68HC11 and 2 for 68HC12.
4689 @subsection VAX Options
4692 These @samp{-m} options are defined for the Vax:
4696 Do not output certain jump instructions (@code{aobleq} and so on)
4697 that the Unix assembler for the Vax cannot handle across long
4701 Do output those jump instructions, on the assumption that you
4702 will assemble with the GNU assembler.
4705 Output code for g-format floating point numbers instead of d-format.
4709 @subsection SPARC Options
4710 @cindex SPARC options
4712 These @samp{-m} switches are supported on the SPARC:
4717 Specify @samp{-mapp-regs} to generate output using the global registers
4718 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
4721 To be fully SVR4 ABI compliant at the cost of some performance loss,
4722 specify @samp{-mno-app-regs}. You should compile libraries and system
4723 software with this option.
4727 Generate output containing floating point instructions. This is the
4732 Generate output containing library calls for floating point.
4733 @strong{Warning:} the requisite libraries are not available for all SPARC
4734 targets. Normally the facilities of the machine's usual C compiler are
4735 used, but this cannot be done directly in cross-compilation. You must make
4736 your own arrangements to provide suitable library functions for
4737 cross-compilation. The embedded targets @samp{sparc-*-aout} and
4738 @samp{sparclite-*-*} do provide software floating point support.
4740 @samp{-msoft-float} changes the calling convention in the output file;
4741 therefore, it is only useful if you compile @emph{all} of a program with
4742 this option. In particular, you need to compile @file{libgcc.a}, the
4743 library that comes with GCC, with @samp{-msoft-float} in order for
4746 @item -mhard-quad-float
4747 Generate output containing quad-word (long double) floating point
4750 @item -msoft-quad-float
4751 Generate output containing library calls for quad-word (long double)
4752 floating point instructions. The functions called are those specified
4753 in the SPARC ABI. This is the default.
4755 As of this writing, there are no sparc implementations that have hardware
4756 support for the quad-word floating point instructions. They all invoke
4757 a trap handler for one of these instructions, and then the trap handler
4758 emulates the effect of the instruction. Because of the trap handler overhead,
4759 this is much slower than calling the ABI library routines. Thus the
4760 @samp{-msoft-quad-float} option is the default.
4764 With @samp{-mepilogue} (the default), the compiler always emits code for
4765 function exit at the end of each function. Any function exit in
4766 the middle of the function (such as a return statement in C) will
4767 generate a jump to the exit code at the end of the function.
4769 With @samp{-mno-epilogue}, the compiler tries to emit exit code inline
4770 at every function exit.
4774 With @samp{-mflat}, the compiler does not generate save/restore instructions
4775 and will use a "flat" or single register window calling convention.
4776 This model uses %i7 as the frame pointer and is compatible with the normal
4777 register window model. Code from either may be intermixed.
4778 The local registers and the input registers (0-5) are still treated as
4779 "call saved" registers and will be saved on the stack as necessary.
4781 With @samp{-mno-flat} (the default), the compiler emits save/restore
4782 instructions (except for leaf functions) and is the normal mode of operation.
4784 @item -mno-unaligned-doubles
4785 @itemx -munaligned-doubles
4786 Assume that doubles have 8 byte alignment. This is the default.
4788 With @samp{-munaligned-doubles}, GCC assumes that doubles have 8 byte
4789 alignment only if they are contained in another type, or if they have an
4790 absolute address. Otherwise, it assumes they have 4 byte alignment.
4791 Specifying this option avoids some rare compatibility problems with code
4792 generated by other compilers. It is not the default because it results
4793 in a performance loss, especially for floating point code.
4795 @item -mno-faster-structs
4796 @itemx -mfaster-structs
4797 With @samp{-mfaster-structs}, the compiler assumes that structures
4798 should have 8 byte alignment. This enables the use of pairs of
4799 @code{ldd} and @code{std} instructions for copies in structure
4800 assignment, in place of twice as many @code{ld} and @code{st} pairs.
4801 However, the use of this changed alignment directly violates the Sparc
4802 ABI. Thus, it's intended only for use on targets where the developer
4803 acknowledges that their resulting code will not be directly in line with
4804 the rules of the ABI.
4808 These two options select variations on the SPARC architecture.
4810 By default (unless specifically configured for the Fujitsu SPARClite),
4811 GCC generates code for the v7 variant of the SPARC architecture.
4813 @samp{-mv8} will give you SPARC v8 code. The only difference from v7
4814 code is that the compiler emits the integer multiply and integer
4815 divide instructions which exist in SPARC v8 but not in SPARC v7.
4817 @samp{-msparclite} will give you SPARClite code. This adds the integer
4818 multiply, integer divide step and scan (@code{ffs}) instructions which
4819 exist in SPARClite but not in SPARC v7.
4821 These options are deprecated and will be deleted in a future GCC release.
4822 They have been replaced with @samp{-mcpu=xxx}.
4826 These two options select the processor for which the code is optimised.
4828 With @samp{-mcypress} (the default), the compiler optimizes code for the
4829 Cypress CY7C602 chip, as used in the SparcStation/SparcServer 3xx series.
4830 This is also appropriate for the older SparcStation 1, 2, IPX etc.
4832 With @samp{-msupersparc} the compiler optimizes code for the SuperSparc cpu, as
4833 used in the SparcStation 10, 1000 and 2000 series. This flag also enables use
4834 of the full SPARC v8 instruction set.
4836 These options are deprecated and will be deleted in a future GCC release.
4837 They have been replaced with @samp{-mcpu=xxx}.
4839 @item -mcpu=@var{cpu_type}
4840 Set the instruction set, register set, and instruction scheduling parameters
4841 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
4842 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
4843 @samp{hypersparc}, @samp{sparclite86x}, @samp{f930}, @samp{f934},
4844 @samp{sparclet}, @samp{tsc701}, @samp{v9}, and @samp{ultrasparc}.
4846 Default instruction scheduling parameters are used for values that select
4847 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
4848 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
4850 Here is a list of each supported architecture and their supported
4855 v8: supersparc, hypersparc
4856 sparclite: f930, f934, sparclite86x
4861 @item -mtune=@var{cpu_type}
4862 Set the instruction scheduling parameters for machine type
4863 @var{cpu_type}, but do not set the instruction set or register set that the
4864 option @samp{-mcpu=}@var{cpu_type} would.
4866 The same values for @samp{-mcpu=}@var{cpu_type} are used for
4867 @samp{-mtune=}@*@var{cpu_type}, though the only useful values are those that
4868 select a particular cpu implementation: @samp{cypress}, @samp{supersparc},
4869 @samp{hypersparc}, @samp{f930}, @samp{f934}, @samp{sparclite86x},
4870 @samp{tsc701}, @samp{ultrasparc}.
4874 These @samp{-m} switches are supported in addition to the above
4875 on the SPARCLET processor.
4878 @item -mlittle-endian
4879 Generate code for a processor running in little-endian mode.
4882 Treat register @code{%g0} as a normal register.
4883 GCC will continue to clobber it as necessary but will not assume
4884 it always reads as 0.
4886 @item -mbroken-saverestore
4887 Generate code that does not use non-trivial forms of the @code{save} and
4888 @code{restore} instructions. Early versions of the SPARCLET processor do
4889 not correctly handle @code{save} and @code{restore} instructions used with
4890 arguments. They correctly handle them used without arguments. A @code{save}
4891 instruction used without arguments increments the current window pointer
4892 but does not allocate a new stack frame. It is assumed that the window
4893 overflow trap handler will properly handle this case as will interrupt
4897 These @samp{-m} switches are supported in addition to the above
4898 on SPARC V9 processors in 64 bit environments.
4901 @item -mlittle-endian
4902 Generate code for a processor running in little-endian mode.
4906 Generate code for a 32 bit or 64 bit environment.
4907 The 32 bit environment sets int, long and pointer to 32 bits.
4908 The 64 bit environment sets int to 32 bits and long and pointer
4911 @item -mcmodel=medlow
4912 Generate code for the Medium/Low code model: the program must be linked
4913 in the low 32 bits of the address space. Pointers are 64 bits.
4914 Programs can be statically or dynamically linked.
4916 @item -mcmodel=medmid
4917 Generate code for the Medium/Middle code model: the program must be linked
4918 in the low 44 bits of the address space, the text segment must be less than
4919 2G bytes, and data segment must be within 2G of the text segment.
4920 Pointers are 64 bits.
4922 @item -mcmodel=medany
4923 Generate code for the Medium/Anywhere code model: the program may be linked
4924 anywhere in the address space, the text segment must be less than
4925 2G bytes, and data segment must be within 2G of the text segment.
4926 Pointers are 64 bits.
4928 @item -mcmodel=embmedany
4929 Generate code for the Medium/Anywhere code model for embedded systems:
4930 assume a 32 bit text and a 32 bit data segment, both starting anywhere
4931 (determined at link time). Register %g4 points to the base of the
4932 data segment. Pointers still 64 bits.
4933 Programs are statically linked, PIC is not supported.
4936 @itemx -mno-stack-bias
4937 With @samp{-mstack-bias}, GCC assumes that the stack pointer, and
4938 frame pointer if present, are offset by -2047 which must be added back
4939 when making stack frame references.
4940 Otherwise, assume no such offset is present.
4943 @node Convex Options
4944 @subsection Convex Options
4945 @cindex Convex options
4947 These @samp{-m} options are defined for Convex:
4951 Generate output for C1. The code will run on any Convex machine.
4952 The preprocessor symbol @code{__convex__c1__} is defined.
4955 Generate output for C2. Uses instructions not available on C1.
4956 Scheduling and other optimizations are chosen for max performance on C2.
4957 The preprocessor symbol @code{__convex_c2__} is defined.
4960 Generate output for C32xx. Uses instructions not available on C1.
4961 Scheduling and other optimizations are chosen for max performance on C32.
4962 The preprocessor symbol @code{__convex_c32__} is defined.
4965 Generate output for C34xx. Uses instructions not available on C1.
4966 Scheduling and other optimizations are chosen for max performance on C34.
4967 The preprocessor symbol @code{__convex_c34__} is defined.
4970 Generate output for C38xx. Uses instructions not available on C1.
4971 Scheduling and other optimizations are chosen for max performance on C38.
4972 The preprocessor symbol @code{__convex_c38__} is defined.
4975 Generate code which puts an argument count in the word preceding each
4976 argument list. This is compatible with regular CC, and a few programs
4977 may need the argument count word. GDB and other source-level debuggers
4978 do not need it; this info is in the symbol table.
4981 Omit the argument count word. This is the default.
4983 @item -mvolatile-cache
4984 Allow volatile references to be cached. This is the default.
4986 @item -mvolatile-nocache
4987 Volatile references bypass the data cache, going all the way to memory.
4988 This is only needed for multi-processor code that does not use standard
4989 synchronization instructions. Making non-volatile references to volatile
4990 locations will not necessarily work.
4993 Type long is 32 bits, the same as type int. This is the default.
4996 Type long is 64 bits, the same as type long long. This option is useless,
4997 because no library support exists for it.
5000 @node AMD29K Options
5001 @subsection AMD29K Options
5002 @cindex AMD29K options
5004 These @samp{-m} options are defined for the AMD Am29000:
5009 @cindex DW bit (29k)
5010 Generate code that assumes the @code{DW} bit is set, i.e., that byte and
5011 halfword operations are directly supported by the hardware. This is the
5016 Generate code that assumes the @code{DW} bit is not set.
5020 @cindex byte writes (29k)
5021 Generate code that assumes the system supports byte and halfword write
5022 operations. This is the default.
5026 Generate code that assumes the systems does not support byte and
5027 halfword write operations. @samp{-mnbw} implies @samp{-mndw}.
5031 @cindex memory model (29k)
5032 Use a small memory model that assumes that all function addresses are
5033 either within a single 256 KB segment or at an absolute address of less
5034 than 256k. This allows the @code{call} instruction to be used instead
5035 of a @code{const}, @code{consth}, @code{calli} sequence.
5039 Use the normal memory model: Generate @code{call} instructions only when
5040 calling functions in the same file and @code{calli} instructions
5041 otherwise. This works if each file occupies less than 256 KB but allows
5042 the entire executable to be larger than 256 KB. This is the default.
5045 Always use @code{calli} instructions. Specify this option if you expect
5046 a single file to compile into more than 256 KB of code.
5050 @cindex processor selection (29k)
5051 Generate code for the Am29050.
5055 Generate code for the Am29000. This is the default.
5057 @item -mkernel-registers
5058 @kindex -mkernel-registers
5059 @cindex kernel and user registers (29k)
5060 Generate references to registers @code{gr64-gr95} instead of to
5061 registers @code{gr96-gr127}. This option can be used when compiling
5062 kernel code that wants a set of global registers disjoint from that used
5065 Note that when this option is used, register names in @samp{-f} flags
5066 must use the normal, user-mode, names.
5068 @item -muser-registers
5069 @kindex -muser-registers
5070 Use the normal set of global registers, @code{gr96-gr127}. This is the
5074 @itemx -mno-stack-check
5075 @kindex -mstack-check
5076 @cindex stack checks (29k)
5077 Insert (or do not insert) a call to @code{__msp_check} after each stack
5078 adjustment. This is often used for kernel code.
5081 @itemx -mno-storem-bug
5082 @kindex -mstorem-bug
5083 @cindex storem bug (29k)
5084 @samp{-mstorem-bug} handles 29k processors which cannot handle the
5085 separation of a mtsrim insn and a storem instruction (most 29000 chips
5086 to date, but not the 29050).
5088 @item -mno-reuse-arg-regs
5089 @itemx -mreuse-arg-regs
5090 @kindex -mreuse-arg-regs
5091 @samp{-mno-reuse-arg-regs} tells the compiler to only use incoming argument
5092 registers for copying out arguments. This helps detect calling a function
5093 with fewer arguments than it was declared with.
5095 @item -mno-impure-text
5096 @itemx -mimpure-text
5097 @kindex -mimpure-text
5098 @samp{-mimpure-text}, used in addition to @samp{-shared}, tells the compiler to
5099 not pass @samp{-assert pure-text} to the linker when linking a shared object.
5102 @kindex -msoft-float
5103 Generate output containing library calls for floating point.
5104 @strong{Warning:} the requisite libraries are not part of GCC.
5105 Normally the facilities of the machine's usual C compiler are used, but
5106 this can't be done directly in cross-compilation. You must make your
5107 own arrangements to provide suitable library functions for
5112 Do not generate multm or multmu instructions. This is useful for some embedded
5113 systems which do not have trap handlers for these instructions.
5117 @subsection ARM Options
5120 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
5125 @kindex -mapcs-frame
5126 Generate a stack frame that is compliant with the ARM Procedure Call
5127 Standard for all functions, even if this is not strictly necessary for
5128 correct execution of the code. Specifying @samp{-fomit-frame-pointer}
5129 with this option will cause the stack frames not to be generated for
5130 leaf functions. The default is @samp{-mno-apcs-frame}.
5134 This is a synonym for @samp{-mapcs-frame}.
5138 Generate code for a processor running with a 26-bit program counter,
5139 and conforming to the function calling standards for the APCS 26-bit
5140 option. This option replaces the @samp{-m2} and @samp{-m3} options
5141 of previous releases of the compiler.
5145 Generate code for a processor running with a 32-bit program counter,
5146 and conforming to the function calling standards for the APCS 32-bit
5147 option. This option replaces the @samp{-m6} option of previous releases
5150 @item -mapcs-stack-check
5151 @kindex -mapcs-stack-check
5152 @kindex -mno-apcs-stack-check
5153 Generate code to check the amount of stack space available upon entry to
5154 every function (that actually uses some stack space). If there is
5155 insufficient space available then either the function
5156 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
5157 called, depending upon the amount of stack space required. The run time
5158 system is required to provide these functions. The default is
5159 @samp{-mno-apcs-stack-check}, since this produces smaller code.
5162 @kindex -mapcs-float
5163 @kindex -mno-apcs-float
5164 Pass floating point arguments using the float point registers. This is
5165 one of the variants of the APCS. This option is recommended if the
5166 target hardware has a floating point unit or if a lot of floating point
5167 arithmetic is going to be performed by the code. The default is
5168 @samp{-mno-apcs-float}, since integer only code is slightly increased in
5169 size if @samp{-mapcs-float} is used.
5171 @item -mapcs-reentrant
5172 @kindex -mapcs-reentrant
5173 @kindex -mno-apcs-reentrant
5174 Generate reentrant, position independent code. This is the equivalent
5175 to specifying the @samp{-fpic} option. The default is
5176 @samp{-mno-apcs-reentrant}.
5178 @item -mthumb-interwork
5179 @kindex -mthumb-interwork
5180 @kindex -mno-thumb-interwork
5181 Generate code which supports calling between the ARM and THUMB
5182 instruction sets. Without this option the two instruction sets cannot
5183 be reliably used inside one program. The default is
5184 @samp{-mno-thumb-interwork}, since slightly larger code is generated
5185 when @samp{-mthumb-interwork} is specified.
5187 @item -mno-sched-prolog
5188 @kindex -mno-sched-prolog
5189 @kindex -msched-prolog
5190 Prevent the reordering of instructions in the function prolog, or the
5191 merging of those instruction with the instructions in the function's
5192 body. This means that all functions will start with a recognizable set
5193 of instructions (or in fact one of a choice from a small set of
5194 different function prologues), and this information can be used to
5195 locate the start if functions inside an executable piece of code. The
5196 default is @samp{-msched-prolog}.
5199 Generate output containing floating point instructions. This is the
5203 Generate output containing library calls for floating point.
5204 @strong{Warning:} the requisite libraries are not available for all ARM
5205 targets. Normally the facilities of the machine's usual C compiler are
5206 used, but this cannot be done directly in cross-compilation. You must make
5207 your own arrangements to provide suitable library functions for
5210 @samp{-msoft-float} changes the calling convention in the output file;
5211 therefore, it is only useful if you compile @emph{all} of a program with
5212 this option. In particular, you need to compile @file{libgcc.a}, the
5213 library that comes with GCC, with @samp{-msoft-float} in order for
5216 @item -mlittle-endian
5217 Generate code for a processor running in little-endian mode. This is
5218 the default for all standard configurations.
5221 Generate code for a processor running in big-endian mode; the default is
5222 to compile code for a little-endian processor.
5224 @item -mwords-little-endian
5225 This option only applies when generating code for big-endian processors.
5226 Generate code for a little-endian word order but a big-endian byte
5227 order. That is, a byte order of the form @samp{32107654}. Note: this
5228 option should only be used if you require compatibility with code for
5229 big-endian ARM processors generated by versions of the compiler prior to
5232 @item -malignment-traps
5233 @kindex -malignment-traps
5234 Generate code that will not trap if the MMU has alignment traps enabled.
5235 On ARM architectures prior to ARMv4, there were no instructions to
5236 access half-word objects stored in memory. However, when reading from
5237 memory a feature of the ARM architecture allows a word load to be used,
5238 even if the address is unaligned, and the processor core will rotate the
5239 data as it is being loaded. This option tells the compiler that such
5240 misaligned accesses will cause a MMU trap and that it should instead
5241 synthesise the access as a series of byte accesses. The compiler can
5242 still use word accesses to load half-word data if it knows that the
5243 address is aligned to a word boundary.
5245 This option is ignored when compiling for ARM architecture 4 or later,
5246 since these processors have instructions to directly access half-word
5249 @item -mno-alignment-traps
5250 @kindex -mno-alignment-traps
5251 Generate code that assumes that the MMU will not trap unaligned
5252 accesses. This produces better code when the target instruction set
5253 does not have half-word memory operations (implementations prior to
5256 Note that you cannot use this option to access unaligned word objects,
5257 since the processor will only fetch one 32-bit aligned object from
5260 The default setting for most targets is -mno-alignment-traps, since
5261 this produces better code when there are no half-word memory
5262 instructions available.
5264 @item -mshort-load-bytes
5265 @kindex -mshort-load-bytes
5266 This is a deprecated alias for @samp{-malignment-traps}.
5268 @item -mno-short-load-bytes
5269 @kindex -mno-short-load-bytes
5270 This is a deprecated alias for @samp{-mno-alignment-traps}.
5272 @item -mshort-load-words
5273 @kindex -mshort-load-words
5274 This is a deprecated alias for @samp{-mno-alignment-traps}.
5276 @item -mno-short-load-words
5277 @kindex -mno-short-load-words
5278 This is a deprecated alias for @samp{-malignment-traps}.
5282 This option only applies to RISC iX. Emulate the native BSD-mode
5283 compiler. This is the default if @samp{-ansi} is not specified.
5287 This option only applies to RISC iX. Emulate the native X/Open-mode
5290 @item -mno-symrename
5291 @kindex -mno-symrename
5292 This option only applies to RISC iX. Do not run the assembler
5293 post-processor, @samp{symrename}, after code has been assembled.
5294 Normally it is necessary to modify some of the standard symbols in
5295 preparation for linking with the RISC iX C library; this option
5296 suppresses this pass. The post-processor is never run when the
5297 compiler is built for cross-compilation.
5301 This specifies the name of the target ARM processor. GCC uses this name
5302 to determine what kind of instructions it can use when generating
5303 assembly code. Permissible names are: arm2, arm250, arm3, arm6, arm60,
5304 arm600, arm610, arm620, arm7, arm7m, arm7d, arm7dm, arm7di, arm7dmi,
5305 arm70, arm700, arm700i, arm710, arm710c, arm7100, arm7500, arm7500fe,
5306 arm7tdmi, arm8, strongarm, strongarm110, strongarm1100, arm8, arm810,
5307 arm9, arm920, arm920t, arm9tdmi.
5309 @itemx -mtune=<name>
5311 This option is very similar to the @samp{-mcpu=} option, except that
5312 instead of specifying the actual target processor type, and hence
5313 restricting which instructions can be used, it specifies that GCC should
5314 tune the performance of the code as if the target were of the type
5315 specified in this option, but still choosing the instructions that it
5316 will generate based on the cpu specified by a @samp{-mcpu=} option.
5317 For some arm implementations better performance can be obtained by using
5322 This specifies the name of the target ARM architecture. GCC uses this
5323 name to determine what kind of instructions it can use when generating
5324 assembly code. This option can be used in conjunction with or instead
5325 of the @samp{-mcpu=} option. Permissible names are: armv2, armv2a,
5326 armv3, armv3m, armv4, armv4t, armv5.
5328 @item -mfpe=<number>
5329 @itemx -mfp=<number>
5332 This specifies the version of the floating point emulation available on
5333 the target. Permissible values are 2 and 3. @samp{-mfp=} is a synonym
5334 for @samp{-mfpe=} to support older versions of GCC.
5336 @item -mstructure-size-boundary=<n>
5337 @kindex -mstructure-size-boundary
5338 The size of all structures and unions will be rounded up to a multiple
5339 of the number of bits set by this option. Permissible values are 8 and
5340 32. The default value varies for different toolchains. For the COFF
5341 targeted toolchain the default value is 8. Specifying the larger number
5342 can produce faster, more efficient code, but can also increase the size
5343 of the program. The two values are potentially incompatible. Code
5344 compiled with one value cannot necessarily expect to work with code or
5345 libraries compiled with the other value, if they exchange information
5346 using structures or unions. Programmers are encouraged to use the 32
5347 value as future versions of the toolchain may default to this value.
5349 @item -mabort-on-noreturn
5350 @kindex -mabort-on-noreturn
5351 @kindex -mnoabort-on-noreturn
5352 Generate a call to the function abort at the end of a noreturn function.
5353 It will be executed if the function tries to return.
5356 @itemx -mno-long-calls
5357 Tells the compiler to perform function calls by first loading the
5358 address of the function into a register and then performing a subroutine
5359 call on this register. This switch is needed if the target function
5360 will lie outside of the 64 megabyte addressing range of the offset based
5361 version of subroutine call instruction.
5363 Even if this switch is enabled, not all function calls will be turned
5364 into long calls. The heuristic is that static functions, functions
5365 which have the @samp{short-call} attribute, functions that are inside
5366 the scope of a @samp{#pragma no_long_calls} directive and functions whose
5367 definitions have already been compiled within the current compilation
5368 unit, will not be turned into long calls. The exception to this rule is
5369 that weak function definitions, functions with the @samp{long-call}
5370 attribute or the @samp{section} attribute, and functions that are within
5371 the scope of a @samp{#pragma long_calls} directive, will always be
5372 turned into long calls.
5374 This feature is not enabled by default. Specifying
5375 @samp{--no-long-calls} will restore the default behaviour, as will
5376 placing the function calls within the scope of a @samp{#pragma
5377 long_calls_off} directive. Note these switches have no effect on how
5378 the compiler generates code to handle function calls via function
5381 @item -mnop-fun-dllimport
5382 @kindex -mnop-fun-dllimport
5383 Disable the support for the @emph{dllimport} attribute.
5385 @item -msingle-pic-base
5386 @kindex -msingle-pic-base
5387 Treat the register used for PIC addressing as read-only, rather than
5388 loading it in the prologue for each function. The run-time system is
5389 responsible for initialising this register with an appropriate value
5390 before execution begins.
5392 @item -mpic-register=<reg>
5393 @kindex -mpic-register=
5394 Specify the register to be used for PIC addressing. The default is R10
5395 unless stack-checking is enabled, when R9 is used.
5400 @subsection Thumb Options
5401 @cindex Thumb Options
5405 @item -mthumb-interwork
5406 @kindex -mthumb-interwork
5407 @kindex -mno-thumb-interwork
5408 Generate code which supports calling between the THUMB and ARM
5409 instruction sets. Without this option the two instruction sets cannot
5410 be reliably used inside one program. The default is
5411 @samp{-mno-thumb-interwork}, since slightly smaller code is generated
5415 @kindex -mtpcs-frame
5416 @kindex -mno-tpcs-frame
5417 Generate a stack frame that is compliant with the Thumb Procedure Call
5418 Standard for all non-leaf functions. (A leaf function is one that does
5419 not call any other functions). The default is @samp{-mno-apcs-frame}.
5421 @item -mtpcs-leaf-frame
5422 @kindex -mtpcs-leaf-frame
5423 @kindex -mno-tpcs-leaf-frame
5424 Generate a stack frame that is compliant with the Thumb Procedure Call
5425 Standard for all leaf functions. (A leaf function is one that does
5426 not call any other functions). The default is @samp{-mno-apcs-leaf-frame}.
5428 @item -mlittle-endian
5429 @kindex -mlittle-endian
5430 Generate code for a processor running in little-endian mode. This is
5431 the default for all standard configurations.
5434 @kindex -mbig-endian
5435 Generate code for a processor running in big-endian mode.
5437 @item -mstructure-size-boundary=<n>
5438 @kindex -mstructure-size-boundary
5439 The size of all structures and unions will be rounded up to a multiple
5440 of the number of bits set by this option. Permissible values are 8 and
5441 32. The default value varies for different toolchains. For the COFF
5442 targeted toolchain the default value is 8. Specifying the larger number
5443 can produced faster, more efficient code, but can also increase the size
5444 of the program. The two values are potentially incompatible. Code
5445 compiled with one value cannot necessarily expect to work with code or
5446 libraries compiled with the other value, if they exchange information
5447 using structures or unions. Programmers are encouraged to use the 32
5448 value as future versions of the toolchain may default to this value.
5450 @item -mnop-fun-dllimport
5451 @kindex -mnop-fun-dllimport
5452 Disable the support for the @emph{dllimport} attribute.
5454 @item -mcallee-super-interworking
5455 @kindex -mcallee-super-interworking
5456 Gives all externally visible functions in the file being compiled an ARM
5457 instruction set header which switches to Thumb mode before executing the
5458 rest of the function. This allows these functions to be called from
5459 non-interworking code.
5461 @item -mcaller-super-interworking
5462 @kindex -mcaller-super-interworking
5463 Allows calls via function pointers (including virtual functions) to
5464 execute correctly regardless of whether the target code has been
5465 compiled for interworking or not. There is a small overhead in the cost
5466 of executing a function pointer if this option is enabled.
5468 @item -msingle-pic-base
5469 @kindex -msingle-pic-base
5470 Treat the register used for PIC addressing as read-only, rather than
5471 loading it in the prologue for each function. The run-time system is
5472 responsible for initialising this register with an appropriate value
5473 before execution begins.
5475 @item -mpic-register=<reg>
5476 @kindex -mpic-register=
5477 Specify the register to be used for PIC addressing. The default is R10.
5481 @node MN10200 Options
5482 @subsection MN10200 Options
5483 @cindex MN10200 options
5484 These @samp{-m} options are defined for Matsushita MN10200 architectures:
5488 Indicate to the linker that it should perform a relaxation optimization pass
5489 to shorten branches, calls and absolute memory addresses. This option only
5490 has an effect when used on the command line for the final link step.
5492 This option makes symbolic debugging impossible.
5495 @node MN10300 Options
5496 @subsection MN10300 Options
5497 @cindex MN10300 options
5498 These @samp{-m} options are defined for Matsushita MN10300 architectures:
5502 Generate code to avoid bugs in the multiply instructions for the MN10300
5503 processors. This is the default.
5506 Do not generate code to avoid bugs in the multiply instructions for the
5510 Generate code which uses features specific to the AM33 processor.
5513 Do not generate code which uses features specific to the AM33 processor. This
5517 Indicate to the linker that it should perform a relaxation optimization pass
5518 to shorten branches, calls and absolute memory addresses. This option only
5519 has an effect when used on the command line for the final link step.
5521 This option makes symbolic debugging impossible.
5525 @node M32R/D Options
5526 @subsection M32R/D Options
5527 @cindex M32R/D options
5529 These @samp{-m} options are defined for Mitsubishi M32R/D architectures:
5532 @item -mcode-model=small
5533 Assume all objects live in the lower 16MB of memory (so that their addresses
5534 can be loaded with the @code{ld24} instruction), and assume all subroutines
5535 are reachable with the @code{bl} instruction.
5536 This is the default.
5538 The addressability of a particular object can be set with the
5539 @code{model} attribute.
5541 @item -mcode-model=medium
5542 Assume objects may be anywhere in the 32 bit address space (the compiler
5543 will generate @code{seth/add3} instructions to load their addresses), and
5544 assume all subroutines are reachable with the @code{bl} instruction.
5546 @item -mcode-model=large
5547 Assume objects may be anywhere in the 32 bit address space (the compiler
5548 will generate @code{seth/add3} instructions to load their addresses), and
5549 assume subroutines may not be reachable with the @code{bl} instruction
5550 (the compiler will generate the much slower @code{seth/add3/jl}
5551 instruction sequence).
5554 Disable use of the small data area. Variables will be put into
5555 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
5556 @code{section} attribute has been specified).
5557 This is the default.
5559 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
5560 Objects may be explicitly put in the small data area with the
5561 @code{section} attribute using one of these sections.
5564 Put small global and static data in the small data area, but do not
5565 generate special code to reference them.
5568 Put small global and static data in the small data area, and generate
5569 special instructions to reference them.
5572 @cindex smaller data references
5573 Put global and static objects less than or equal to @var{num} bytes
5574 into the small data or bss sections instead of the normal data or bss
5575 sections. The default value of @var{num} is 8.
5576 The @samp{-msdata} option must be set to one of @samp{sdata} or @samp{use}
5577 for this option to have any effect.
5579 All modules should be compiled with the same @samp{-G @var{num}} value.
5580 Compiling with different values of @var{num} may or may not work; if it
5581 doesn't the linker will give an error message - incorrect code will not be
5587 @subsection M88K Options
5588 @cindex M88k options
5590 These @samp{-m} options are defined for Motorola 88k architectures:
5595 Generate code that works well on both the m88100 and the
5600 Generate code that works best for the m88100, but that also
5605 Generate code that works best for the m88110, and may not run
5610 Obsolete option to be removed from the next revision.
5613 @item -midentify-revision
5614 @kindex -midentify-revision
5616 @cindex identifying source, compiler (88k)
5617 Include an @code{ident} directive in the assembler output recording the
5618 source file name, compiler name and version, timestamp, and compilation
5621 @item -mno-underscores
5622 @kindex -mno-underscores
5623 @cindex underscores, avoiding (88k)
5624 In assembler output, emit symbol names without adding an underscore
5625 character at the beginning of each name. The default is to use an
5626 underscore as prefix on each name.
5628 @item -mocs-debug-info
5629 @itemx -mno-ocs-debug-info
5630 @kindex -mocs-debug-info
5631 @kindex -mno-ocs-debug-info
5633 @cindex debugging, 88k OCS
5634 Include (or omit) additional debugging information (about registers used
5635 in each stack frame) as specified in the 88open Object Compatibility
5636 Standard, ``OCS''. This extra information allows debugging of code that
5637 has had the frame pointer eliminated. The default for DG/UX, SVr4, and
5638 Delta 88 SVr3.2 is to include this information; other 88k configurations
5639 omit this information by default.
5641 @item -mocs-frame-position
5642 @kindex -mocs-frame-position
5643 @cindex register positions in frame (88k)
5644 When emitting COFF debugging information for automatic variables and
5645 parameters stored on the stack, use the offset from the canonical frame
5646 address, which is the stack pointer (register 31) on entry to the
5647 function. The DG/UX, SVr4, Delta88 SVr3.2, and BCS configurations use
5648 @samp{-mocs-frame-position}; other 88k configurations have the default
5649 @samp{-mno-ocs-frame-position}.
5651 @item -mno-ocs-frame-position
5652 @kindex -mno-ocs-frame-position
5653 @cindex register positions in frame (88k)
5654 When emitting COFF debugging information for automatic variables and
5655 parameters stored on the stack, use the offset from the frame pointer
5656 register (register 30). When this option is in effect, the frame
5657 pointer is not eliminated when debugging information is selected by the
5660 @item -moptimize-arg-area
5661 @itemx -mno-optimize-arg-area
5662 @kindex -moptimize-arg-area
5663 @kindex -mno-optimize-arg-area
5664 @cindex arguments in frame (88k)
5665 Control how function arguments are stored in stack frames.
5666 @samp{-moptimize-arg-area} saves space by optimizing them, but this
5667 conflicts with the 88open specifications. The opposite alternative,
5668 @samp{-mno-optimize-arg-area}, agrees with 88open standards. By default
5669 GCC does not optimize the argument area.
5671 @item -mshort-data-@var{num}
5672 @kindex -mshort-data-@var{num}
5673 @cindex smaller data references (88k)
5674 @cindex r0-relative references (88k)
5675 Generate smaller data references by making them relative to @code{r0},
5676 which allows loading a value using a single instruction (rather than the
5677 usual two). You control which data references are affected by
5678 specifying @var{num} with this option. For example, if you specify
5679 @samp{-mshort-data-512}, then the data references affected are those
5680 involving displacements of less than 512 bytes.
5681 @samp{-mshort-data-@var{num}} is not effective for @var{num} greater
5684 @item -mserialize-volatile
5685 @kindex -mserialize-volatile
5686 @itemx -mno-serialize-volatile
5687 @kindex -mno-serialize-volatile
5688 @cindex sequential consistency on 88k
5689 Do, or don't, generate code to guarantee sequential consistency
5690 of volatile memory references. By default, consistency is
5693 The order of memory references made by the MC88110 processor does
5694 not always match the order of the instructions requesting those
5695 references. In particular, a load instruction may execute before
5696 a preceding store instruction. Such reordering violates
5697 sequential consistency of volatile memory references, when there
5698 are multiple processors. When consistency must be guaranteed,
5699 GNU C generates special instructions, as needed, to force
5700 execution in the proper order.
5702 The MC88100 processor does not reorder memory references and so
5703 always provides sequential consistency. However, by default, GNU
5704 C generates the special instructions to guarantee consistency
5705 even when you use @samp{-m88100}, so that the code may be run on an
5706 MC88110 processor. If you intend to run your code only on the
5707 MC88100 processor, you may use @samp{-mno-serialize-volatile}.
5709 The extra code generated to guarantee consistency may affect the
5710 performance of your application. If you know that you can safely
5711 forgo this guarantee, you may use @samp{-mno-serialize-volatile}.
5717 @cindex assembler syntax, 88k
5719 Turn on (@samp{-msvr4}) or off (@samp{-msvr3}) compiler extensions
5720 related to System V release 4 (SVr4). This controls the following:
5724 Which variant of the assembler syntax to emit.
5726 @samp{-msvr4} makes the C preprocessor recognize @samp{#pragma weak}
5727 that is used on System V release 4.
5729 @samp{-msvr4} makes GCC issue additional declaration directives used in
5733 @samp{-msvr4} is the default for the m88k-motorola-sysv4 and
5734 m88k-dg-dgux m88k configurations. @samp{-msvr3} is the default for all
5735 other m88k configurations.
5737 @item -mversion-03.00
5738 @kindex -mversion-03.00
5739 This option is obsolete, and is ignored.
5740 @c ??? which asm syntax better for GAS? option there too?
5742 @item -mno-check-zero-division
5743 @itemx -mcheck-zero-division
5744 @kindex -mno-check-zero-division
5745 @kindex -mcheck-zero-division
5746 @cindex zero division on 88k
5747 Do, or don't, generate code to guarantee that integer division by
5748 zero will be detected. By default, detection is guaranteed.
5750 Some models of the MC88100 processor fail to trap upon integer
5751 division by zero under certain conditions. By default, when
5752 compiling code that might be run on such a processor, GNU C
5753 generates code that explicitly checks for zero-valued divisors
5754 and traps with exception number 503 when one is detected. Use of
5755 mno-check-zero-division suppresses such checking for code
5756 generated to run on an MC88100 processor.
5758 GNU C assumes that the MC88110 processor correctly detects all
5759 instances of integer division by zero. When @samp{-m88110} is
5760 specified, both @samp{-mcheck-zero-division} and
5761 @samp{-mno-check-zero-division} are ignored, and no explicit checks for
5762 zero-valued divisors are generated.
5764 @item -muse-div-instruction
5765 @kindex -muse-div-instruction
5766 @cindex divide instruction, 88k
5767 Use the div instruction for signed integer division on the
5768 MC88100 processor. By default, the div instruction is not used.
5770 On the MC88100 processor the signed integer division instruction
5771 div) traps to the operating system on a negative operand. The
5772 operating system transparently completes the operation, but at a
5773 large cost in execution time. By default, when compiling code
5774 that might be run on an MC88100 processor, GNU C emulates signed
5775 integer division using the unsigned integer division instruction
5776 divu), thereby avoiding the large penalty of a trap to the
5777 operating system. Such emulation has its own, smaller, execution
5778 cost in both time and space. To the extent that your code's
5779 important signed integer division operations are performed on two
5780 nonnegative operands, it may be desirable to use the div
5781 instruction directly.
5783 On the MC88110 processor the div instruction (also known as the
5784 divs instruction) processes negative operands without trapping to
5785 the operating system. When @samp{-m88110} is specified,
5786 @samp{-muse-div-instruction} is ignored, and the div instruction is used
5787 for signed integer division.
5789 Note that the result of dividing INT_MIN by -1 is undefined. In
5790 particular, the behavior of such a division with and without
5791 @samp{-muse-div-instruction} may differ.
5793 @item -mtrap-large-shift
5794 @itemx -mhandle-large-shift
5795 @kindex -mtrap-large-shift
5796 @kindex -mhandle-large-shift
5797 @cindex bit shift overflow (88k)
5798 @cindex large bit shifts (88k)
5799 Include code to detect bit-shifts of more than 31 bits; respectively,
5800 trap such shifts or emit code to handle them properly. By default GCC
5801 makes no special provision for large bit shifts.
5803 @item -mwarn-passed-structs
5804 @kindex -mwarn-passed-structs
5805 @cindex structure passing (88k)
5806 Warn when a function passes a struct as an argument or result.
5807 Structure-passing conventions have changed during the evolution of the C
5808 language, and are often the source of portability problems. By default,
5809 GCC issues no such warning.
5812 @node RS/6000 and PowerPC Options
5813 @subsection IBM RS/6000 and PowerPC Options
5814 @cindex RS/6000 and PowerPC Options
5815 @cindex IBM RS/6000 and PowerPC Options
5817 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
5825 @itemx -mpowerpc-gpopt
5826 @itemx -mno-powerpc-gpopt
5827 @itemx -mpowerpc-gfxopt
5828 @itemx -mno-powerpc-gfxopt
5830 @itemx -mno-powerpc64
5834 @kindex -mpowerpc-gpopt
5835 @kindex -mpowerpc-gfxopt
5837 GCC supports two related instruction set architectures for the
5838 RS/6000 and PowerPC. The @dfn{POWER} instruction set are those
5839 instructions supported by the @samp{rios} chip set used in the original
5840 RS/6000 systems and the @dfn{PowerPC} instruction set is the
5841 architecture of the Motorola MPC5xx, MPC6xx, MPC8xx microprocessors, and
5842 the IBM 4xx microprocessors.
5844 Neither architecture is a subset of the other. However there is a
5845 large common subset of instructions supported by both. An MQ
5846 register is included in processors supporting the POWER architecture.
5848 You use these options to specify which instructions are available on the
5849 processor you are using. The default value of these options is
5850 determined when configuring GCC. Specifying the
5851 @samp{-mcpu=@var{cpu_type}} overrides the specification of these
5852 options. We recommend you use the @samp{-mcpu=@var{cpu_type}} option
5853 rather than the options listed above.
5855 The @samp{-mpower} option allows GCC to generate instructions that
5856 are found only in the POWER architecture and to use the MQ register.
5857 Specifying @samp{-mpower2} implies @samp{-power} and also allows GCC
5858 to generate instructions that are present in the POWER2 architecture but
5859 not the original POWER architecture.
5861 The @samp{-mpowerpc} option allows GCC to generate instructions that
5862 are found only in the 32-bit subset of the PowerPC architecture.
5863 Specifying @samp{-mpowerpc-gpopt} implies @samp{-mpowerpc} and also allows
5864 GCC to use the optional PowerPC architecture instructions in the
5865 General Purpose group, including floating-point square root. Specifying
5866 @samp{-mpowerpc-gfxopt} implies @samp{-mpowerpc} and also allows GCC to
5867 use the optional PowerPC architecture instructions in the Graphics
5868 group, including floating-point select.
5870 The @samp{-mpowerpc64} option allows GCC to generate the additional
5871 64-bit instructions that are found in the full PowerPC64 architecture
5872 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
5873 @samp{-mno-powerpc64}.
5875 If you specify both @samp{-mno-power} and @samp{-mno-powerpc}, GCC
5876 will use only the instructions in the common subset of both
5877 architectures plus some special AIX common-mode calls, and will not use
5878 the MQ register. Specifying both @samp{-mpower} and @samp{-mpowerpc}
5879 permits GCC to use any instruction from either architecture and to
5880 allow use of the MQ register; specify this for the Motorola MPC601.
5882 @item -mnew-mnemonics
5883 @itemx -mold-mnemonics
5884 @kindex -mnew-mnemonics
5885 @kindex -mold-mnemonics
5886 Select which mnemonics to use in the generated assembler code.
5887 @samp{-mnew-mnemonics} requests output that uses the assembler mnemonics
5888 defined for the PowerPC architecture, while @samp{-mold-mnemonics}
5889 requests the assembler mnemonics defined for the POWER architecture.
5890 Instructions defined in only one architecture have only one mnemonic;
5891 GCC uses that mnemonic irrespective of which of these options is
5894 GCC defaults to the mnemonics appropriate for the architecture in
5895 use. Specifying @samp{-mcpu=@var{cpu_type}} sometimes overrides the
5896 value of these option. Unless you are building a cross-compiler, you
5897 should normally not specify either @samp{-mnew-mnemonics} or
5898 @samp{-mold-mnemonics}, but should instead accept the default.
5900 @item -mcpu=@var{cpu_type}
5902 Set architecture type, register usage, choice of mnemonics, and
5903 instruction scheduling parameters for machine type @var{cpu_type}.
5904 Supported values for @var{cpu_type} are @samp{rios}, @samp{rios1},
5905 @samp{rsc}, @samp{rios2}, @samp{rs64a}, @samp{601}, @samp{602},
5906 @samp{603}, @samp{603e}, @samp{604}, @samp{604e}, @samp{620},
5907 @samp{630}, @samp{740}, @samp{750}, @samp{power}, @samp{power2},
5908 @samp{powerpc}, @samp{403}, @samp{505}, @samp{801}, @samp{821},
5909 @samp{823}, and @samp{860} and @samp{common}. @samp{-mcpu=power},
5910 @samp{-mcpu=power2}, @samp{-mcpu=powerpc}, and @samp{-mcpu=powerpc64}
5911 specify generic POWER, POWER2, pure 32-bit PowerPC (i.e., not MPC601),
5912 and 64-bit PowerPC architecture machine types, with an appropriate,
5913 generic processor model assumed for scheduling purposes.@refill
5915 Specifying any of the following options:
5916 @samp{-mcpu=rios1}, @samp{-mcpu=rios2}, @samp{-mcpu=rsc},
5917 @samp{-mcpu=power}, or @samp{-mcpu=power2}
5918 enables the @samp{-mpower} option and disables the @samp{-mpowerpc} option;
5919 @samp{-mcpu=601} enables both the @samp{-mpower} and @samp{-mpowerpc} options.
5920 All of @samp{-mcpu=rs64a}, @samp{-mcpu=602}, @samp{-mcpu=603},
5921 @samp{-mcpu=603e}, @samp{-mcpu=604}, @samp{-mcpu=620}, @samp{-mcpu=630},
5922 @samp{-mcpu=740}, and @samp{-mcpu=750}
5923 enable the @samp{-mpowerpc} option and disable the @samp{-mpower} option.
5924 Exactly similarly, all of @samp{-mcpu=403},
5925 @samp{-mcpu=505}, @samp{-mcpu=821}, @samp{-mcpu=860} and @samp{-mcpu=powerpc}
5926 enable the @samp{-mpowerpc} option and disable the @samp{-mpower} option.
5927 @samp{-mcpu=common} disables both the
5928 @samp{-mpower} and @samp{-mpowerpc} options.@refill
5930 AIX versions 4 or greater selects @samp{-mcpu=common} by default, so
5931 that code will operate on all members of the RS/6000 POWER and PowerPC
5932 families. In that case, GCC will use only the instructions in the
5933 common subset of both architectures plus some special AIX common-mode
5934 calls, and will not use the MQ register. GCC assumes a generic
5935 processor model for scheduling purposes.
5937 Specifying any of the options @samp{-mcpu=rios1}, @samp{-mcpu=rios2},
5938 @samp{-mcpu=rsc}, @samp{-mcpu=power}, or @samp{-mcpu=power2} also
5939 disables the @samp{new-mnemonics} option. Specifying @samp{-mcpu=601},
5940 @samp{-mcpu=602}, @samp{-mcpu=603}, @samp{-mcpu=603e}, @samp{-mcpu=604},
5941 @samp{-mcpu=620}, @samp{-mcpu=630}, @samp{-mcpu=403}, @samp{-mcpu=505},
5942 @samp{-mcpu=821}, @samp{-mcpu=860} or @samp{-mcpu=powerpc} also enables
5943 the @samp{new-mnemonics} option.@refill
5945 Specifying @samp{-mcpu=403}, @samp{-mcpu=821}, or @samp{-mcpu=860} also
5946 enables the @samp{-msoft-float} option.
5948 @item -mtune=@var{cpu_type}
5949 Set the instruction scheduling parameters for machine type
5950 @var{cpu_type}, but do not set the architecture type, register usage,
5951 choice of mnemonics like @samp{-mcpu=}@var{cpu_type} would. The same
5952 values for @var{cpu_type} are used for @samp{-mtune=}@var{cpu_type} as
5953 for @samp{-mcpu=}@var{cpu_type}. The @samp{-mtune=}@var{cpu_type}
5954 option overrides the @samp{-mcpu=}@var{cpu_type} option in terms of
5955 instruction scheduling parameters.
5958 @itemx -mno-fp-in-toc
5959 @itemx -mno-sum-in-toc
5960 @itemx -mminimal-toc
5961 @kindex -mminimal-toc
5962 Modify generation of the TOC (Table Of Contents), which is created for
5963 every executable file. The @samp{-mfull-toc} option is selected by
5964 default. In that case, GCC will allocate at least one TOC entry for
5965 each unique non-automatic variable reference in your program. GCC
5966 will also place floating-point constants in the TOC. However, only
5967 16,384 entries are available in the TOC.
5969 If you receive a linker error message that saying you have overflowed
5970 the available TOC space, you can reduce the amount of TOC space used
5971 with the @samp{-mno-fp-in-toc} and @samp{-mno-sum-in-toc} options.
5972 @samp{-mno-fp-in-toc} prevents GCC from putting floating-point
5973 constants in the TOC and @samp{-mno-sum-in-toc} forces GCC to
5974 generate code to calculate the sum of an address and a constant at
5975 run-time instead of putting that sum into the TOC. You may specify one
5976 or both of these options. Each causes GCC to produce very slightly
5977 slower and larger code at the expense of conserving TOC space.
5979 If you still run out of space in the TOC even when you specify both of
5980 these options, specify @samp{-mminimal-toc} instead. This option causes
5981 GCC to make only one TOC entry for every file. When you specify this
5982 option, GCC will produce code that is slower and larger but which
5983 uses extremely little TOC space. You may wish to use this option
5984 only on files that contain less frequently executed code. @refill
5990 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
5991 @code{long} type, and the infrastructure needed to support them.
5992 Specifying @samp{-maix64} implies @samp{-mpowerpc64} and
5993 @samp{-mpowerpc}, while @samp{-maix32} disables the 64-bit ABI and
5994 implies @samp{-mno-powerpc64}. GCC defaults to @samp{-maix32}.
5999 On AIX, pass floating-point arguments to prototyped functions beyond the
6000 register save area (RSA) on the stack in addition to argument FPRs. The
6001 AIX calling convention was extended but not initially documented to
6002 handle an obscure K&R C case of calling a function that takes the
6003 address of its arguments with fewer arguments than declared. AIX XL
6004 compilers access floating point arguments which do not fit in the
6005 RSA from the stack when a subroutine is compiled without
6006 optimization. Because always storing floating-point arguments on the
6007 stack is inefficient and rarely needed, this option is not enabled by
6008 default and only is necessary when calling subroutines compiled by AIX
6009 XL compilers without optimization.
6013 Support @dfn{AIX Threads}. Link an application written to use
6014 @dfn{pthreads} with special libraries and startup code to enable the
6019 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE). Link an
6020 application written to use message passing with special startup code to
6021 enable the application to run. The system must have PE installed in the
6022 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
6023 must be overridden with the @samp{-specs=} option to specify the
6024 appropriate directory location. The Parallel Environment does not
6025 support threads, so the @samp{-mpe} option and the @samp{-mthreads}
6026 option are incompatible.
6030 @kindex -msoft-float
6031 Generate code that does not use (uses) the floating-point register set.
6032 Software floating point emulation is provided if you use the
6033 @samp{-msoft-float} option, and pass the option to GCC when linking.
6036 @itemx -mno-multiple
6037 Generate code that uses (does not use) the load multiple word
6038 instructions and the store multiple word instructions. These
6039 instructions are generated by default on POWER systems, and not
6040 generated on PowerPC systems. Do not use @samp{-mmultiple} on little
6041 endian PowerPC systems, since those instructions do not work when the
6042 processor is in little endian mode. The exceptions are PPC740 and
6043 PPC750 which permit the instructions usage in little endian mode.
6048 Generate code that uses (does not use) the load string instructions
6049 and the store string word instructions to save multiple registers and
6050 do small block moves. These instructions are generated by default on
6051 POWER systems, and not generated on PowerPC systems. Do not use
6052 @samp{-mstring} on little endian PowerPC systems, since those
6053 instructions do not work when the processor is in little endian mode.
6054 The exceptions are PPC740 and PPC750 which permit the instructions
6055 usage in little endian mode.
6060 Generate code that uses (does not use) the load or store instructions
6061 that update the base register to the address of the calculated memory
6062 location. These instructions are generated by default. If you use
6063 @samp{-mno-update}, there is a small window between the time that the
6064 stack pointer is updated and the address of the previous frame is
6065 stored, which means code that walks the stack frame across interrupts or
6066 signals may get corrupted data.
6069 @itemx -mno-fused-madd
6070 @kindex -mfused-madd
6071 Generate code that uses (does not use) the floating point multiply and
6072 accumulate instructions. These instructions are generated by default if
6073 hardware floating is used.
6075 @item -mno-bit-align
6078 On System V.4 and embedded PowerPC systems do not (do) force structures
6079 and unions that contain bit fields to be aligned to the base type of the
6082 For example, by default a structure containing nothing but 8
6083 @code{unsigned} bitfields of length 1 would be aligned to a 4 byte
6084 boundary and have a size of 4 bytes. By using @samp{-mno-bit-align},
6085 the structure would be aligned to a 1 byte boundary and be one byte in
6088 @item -mno-strict-align
6089 @itemx -mstrict-align
6090 @kindex -mstrict-align
6091 On System V.4 and embedded PowerPC systems do not (do) assume that
6092 unaligned memory references will be handled by the system.
6095 @itemx -mno-relocatable
6096 @kindex -mrelocatable
6097 On embedded PowerPC systems generate code that allows (does not allow)
6098 the program to be relocated to a different address at runtime. If you
6099 use @samp{-mrelocatable} on any module, all objects linked together must
6100 be compiled with @samp{-mrelocatable} or @samp{-mrelocatable-lib}.
6102 @item -mrelocatable-lib
6103 @itemx -mno-relocatable-lib
6104 On embedded PowerPC systems generate code that allows (does not allow)
6105 the program to be relocated to a different address at runtime. Modules
6106 compiled with @samp{-mrelocatable-lib} can be linked with either modules
6107 compiled without @samp{-mrelocatable} and @samp{-mrelocatable-lib} or
6108 with modules compiled with the @samp{-mrelocatable} options.
6112 On System V.4 and embedded PowerPC systems do not (do) assume that
6113 register 2 contains a pointer to a global area pointing to the addresses
6114 used in the program.
6117 @itemx -mlittle-endian
6118 On System V.4 and embedded PowerPC systems compile code for the
6119 processor in little endian mode. The @samp{-mlittle-endian} option is
6120 the same as @samp{-mlittle}.
6124 On System V.4 and embedded PowerPC systems compile code for the
6125 processor in big endian mode. The @samp{-mbig-endian} option is
6126 the same as @samp{-mbig}.
6129 On System V.4 and embedded PowerPC systems compile code using calling
6130 conventions that adheres to the March 1995 draft of the System V
6131 Application Binary Interface, PowerPC processor supplement. This is the
6132 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
6134 @item -mcall-sysv-eabi
6135 Specify both @samp{-mcall-sysv} and @samp{-meabi} options.
6137 @item -mcall-sysv-noeabi
6138 Specify both @samp{-mcall-sysv} and @samp{-mno-eabi} options.
6141 On System V.4 and embedded PowerPC systems compile code using calling
6142 conventions that are similar to those used on AIX. This is the
6143 default if you configured GCC using @samp{powerpc-*-eabiaix}.
6145 @item -mcall-solaris
6146 On System V.4 and embedded PowerPC systems compile code for the Solaris
6150 On System V.4 and embedded PowerPC systems compile code for the
6151 Linux-based GNU system.
6154 @itemx -mno-prototype
6155 On System V.4 and embedded PowerPC systems assume that all calls to
6156 variable argument functions are properly prototyped. Otherwise, the
6157 compiler must insert an instruction before every non prototyped call to
6158 set or clear bit 6 of the condition code register (@var{CR}) to
6159 indicate whether floating point values were passed in the floating point
6160 registers in case the function takes a variable arguments. With
6161 @samp{-mprototype}, only calls to prototyped variable argument functions
6162 will set or clear the bit.
6165 On embedded PowerPC systems, assume that the startup module is called
6166 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
6167 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
6171 On embedded PowerPC systems, assume that the startup module is called
6172 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
6176 On embedded PowerPC systems, assume that the startup module is called
6177 @file{crt0.o} and the standard C libraries are @file{libads.a} and
6181 On embedded PowerPC systems, assume that the startup module is called
6182 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
6186 On System V.4 and embedded PowerPC systems, specify that you are
6187 compiling for a VxWorks system.
6190 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
6191 header to indicate that @samp{eabi} extended relocations are used.
6195 On System V.4 and embedded PowerPC systems do (do not) adhere to the
6196 Embedded Applications Binary Interface (eabi) which is a set of
6197 modifications to the System V.4 specifications. Selecting @option{-meabi}
6198 means that the stack is aligned to an 8 byte boundary, a function
6199 @code{__eabi} is called to from @code{main} to set up the eabi
6200 environment, and the @samp{-msdata} option can use both @code{r2} and
6201 @code{r13} to point to two separate small data areas. Selecting
6202 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
6203 do not call an initialization function from @code{main}, and the
6204 @samp{-msdata} option will only use @code{r13} to point to a single
6205 small data area. The @samp{-meabi} option is on by default if you
6206 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
6209 On System V.4 and embedded PowerPC systems, put small initialized
6210 @code{const} global and static data in the @samp{.sdata2} section, which
6211 is pointed to by register @code{r2}. Put small initialized
6212 non-@code{const} global and static data in the @samp{.sdata} section,
6213 which is pointed to by register @code{r13}. Put small uninitialized
6214 global and static data in the @samp{.sbss} section, which is adjacent to
6215 the @samp{.sdata} section. The @samp{-msdata=eabi} option is
6216 incompatible with the @samp{-mrelocatable} option. The
6217 @samp{-msdata=eabi} option also sets the @samp{-memb} option.
6220 On System V.4 and embedded PowerPC systems, put small global and static
6221 data in the @samp{.sdata} section, which is pointed to by register
6222 @code{r13}. Put small uninitialized global and static data in the
6223 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
6224 The @samp{-msdata=sysv} option is incompatible with the
6225 @samp{-mrelocatable} option.
6227 @item -msdata=default
6229 On System V.4 and embedded PowerPC systems, if @samp{-meabi} is used,
6230 compile code the same as @samp{-msdata=eabi}, otherwise compile code the
6231 same as @samp{-msdata=sysv}.
6234 On System V.4 and embedded PowerPC systems, put small global and static
6235 data in the @samp{.sdata} section. Put small uninitialized global and
6236 static data in the @samp{.sbss} section. Do not use register @code{r13}
6237 to address small data however. This is the default behavior unless
6238 other @samp{-msdata} options are used.
6242 On embedded PowerPC systems, put all initialized global and static data
6243 in the @samp{.data} section, and all uninitialized data in the
6244 @samp{.bss} section.
6247 @cindex smaller data references (PowerPC)
6248 @cindex .sdata/.sdata2 references (PowerPC)
6249 On embedded PowerPC systems, put global and static items less than or
6250 equal to @var{num} bytes into the small data or bss sections instead of
6251 the normal data or bss section. By default, @var{num} is 8. The
6252 @samp{-G @var{num}} switch is also passed to the linker.
6253 All modules should be compiled with the same @samp{-G @var{num}} value.
6256 @itemx -mno-regnames
6257 On System V.4 and embedded PowerPC systems do (do not) emit register
6258 names in the assembly language output using symbolic forms.
6263 @subsection IBM RT Options
6265 @cindex IBM RT options
6267 These @samp{-m} options are defined for the IBM RT PC:
6271 Use an in-line code sequence for integer multiplies. This is the
6274 @item -mcall-lib-mul
6275 Call @code{lmul$$} for integer multiples.
6277 @item -mfull-fp-blocks
6278 Generate full-size floating point data blocks, including the minimum
6279 amount of scratch space recommended by IBM. This is the default.
6281 @item -mminimum-fp-blocks
6282 Do not include extra scratch space in floating point data blocks. This
6283 results in smaller code, but slower execution, since scratch space must
6284 be allocated dynamically.
6286 @cindex @file{varargs.h} and RT PC
6287 @cindex @file{stdarg.h} and RT PC
6288 @item -mfp-arg-in-fpregs
6289 Use a calling sequence incompatible with the IBM calling convention in
6290 which floating point arguments are passed in floating point registers.
6291 Note that @code{varargs.h} and @code{stdargs.h} will not work with
6292 floating point operands if this option is specified.
6294 @item -mfp-arg-in-gregs
6295 Use the normal calling convention for floating point arguments. This is
6298 @item -mhc-struct-return
6299 Return structures of more than one word in memory, rather than in a
6300 register. This provides compatibility with the MetaWare HighC (hc)
6301 compiler. Use the option @samp{-fpcc-struct-return} for compatibility
6302 with the Portable C Compiler (pcc).
6304 @item -mnohc-struct-return
6305 Return some structures of more than one word in registers, when
6306 convenient. This is the default. For compatibility with the
6307 IBM-supplied compilers, use the option @samp{-fpcc-struct-return} or the
6308 option @samp{-mhc-struct-return}.
6312 @subsection MIPS Options
6313 @cindex MIPS options
6315 These @samp{-m} options are defined for the MIPS family of computers:
6318 @item -mcpu=@var{cpu type}
6319 Assume the defaults for the machine type @var{cpu type} when scheduling
6320 instructions. The choices for @var{cpu type} are @samp{r2000}, @samp{r3000},
6321 @samp{r3900}, @samp{r4000}, @samp{r4100}, @samp{r4300}, @samp{r4400},
6322 @samp{r4600}, @samp{r4650}, @samp{r5000}, @samp{r6000}, @samp{r8000},
6323 and @samp{orion}. Additionally, the @samp{r2000}, @samp{r3000},
6324 @samp{r4000}, @samp{r5000}, and @samp{r6000} can be abbreviated as
6325 @samp{r2k} (or @samp{r2K}), @samp{r3k}, etc. While picking a specific
6326 @var{cpu type} will schedule things appropriately for that particular
6327 chip, the compiler will not generate any code that does not meet level 1
6328 of the MIPS ISA (instruction set architecture) without a @samp{-mipsX}
6329 or @samp{-mabi} switch being used.
6332 Issue instructions from level 1 of the MIPS ISA. This is the default.
6333 @samp{r3000} is the default @var{cpu type} at this ISA level.
6336 Issue instructions from level 2 of the MIPS ISA (branch likely, square
6337 root instructions). @samp{r6000} is the default @var{cpu type} at this
6341 Issue instructions from level 3 of the MIPS ISA (64 bit instructions).
6342 @samp{r4000} is the default @var{cpu type} at this ISA level.
6345 Issue instructions from level 4 of the MIPS ISA (conditional move,
6346 prefetch, enhanced FPU instructions). @samp{r8000} is the default
6347 @var{cpu type} at this ISA level.
6350 Assume that 32 32-bit floating point registers are available. This is
6354 Assume that 32 64-bit floating point registers are available. This is
6355 the default when the @samp{-mips3} option is used.
6358 Assume that 32 32-bit general purpose registers are available. This is
6362 Assume that 32 64-bit general purpose registers are available. This is
6363 the default when the @samp{-mips3} option is used.
6366 Force int and long types to be 64 bits wide. See @samp{-mlong32} for an
6367 explanation of the default, and the width of pointers.
6370 Force long types to be 64 bits wide. See @samp{-mlong32} for an
6371 explanation of the default, and the width of pointers.
6374 Force long, int, and pointer types to be 32 bits wide.
6376 If none of @samp{-mlong32}, @samp{-mlong64}, or @samp{-mint64} are set,
6377 the size of ints, longs, and pointers depends on the ABI and ISA chosen.
6378 For @samp{-mabi=32}, and @samp{-mabi=n32}, ints and longs are 32 bits
6379 wide. For @samp{-mabi=64}, ints are 32 bits, and longs are 64 bits wide.
6380 For @samp{-mabi=eabi} and either @samp{-mips1} or @samp{-mips2}, ints
6381 and longs are 32 bits wide. For @samp{-mabi=eabi} and higher ISAs, ints
6382 are 32 bits, and longs are 64 bits wide. The width of pointer types is
6383 the smaller of the width of longs or the width of general purpose
6384 registers (which in turn depends on the ISA).
6391 Generate code for the indicated ABI. The default instruction level is
6392 @samp{-mips1} for @samp{32}, @samp{-mips3} for @samp{n32}, and
6393 @samp{-mips4} otherwise. Conversely, with @samp{-mips1} or
6394 @samp{-mips2}, the default ABI is @samp{32}; otherwise, the default ABI
6398 Generate code for the MIPS assembler, and invoke @file{mips-tfile} to
6399 add normal debug information. This is the default for all
6400 platforms except for the OSF/1 reference platform, using the OSF/rose
6401 object format. If the either of the @samp{-gstabs} or @samp{-gstabs+}
6402 switches are used, the @file{mips-tfile} program will encapsulate the
6403 stabs within MIPS ECOFF.
6406 Generate code for the GNU assembler. This is the default on the OSF/1
6407 reference platform, using the OSF/rose object format. Also, this is
6408 the default if the configure option @samp{--with-gnu-as} is used.
6410 @item -msplit-addresses
6411 @itemx -mno-split-addresses
6412 Generate code to load the high and low parts of address constants separately.
6413 This allows @code{gcc} to optimize away redundant loads of the high order
6414 bits of addresses. This optimization requires GNU as and GNU ld.
6415 This optimization is enabled by default for some embedded targets where
6416 GNU as and GNU ld are standard.
6420 The @samp{-mrnames} switch says to output code using the MIPS software
6421 names for the registers, instead of the hardware names (ie, @var{a0}
6422 instead of @var{$4}). The only known assembler that supports this option
6423 is the Algorithmics assembler.
6427 The @samp{-mgpopt} switch says to write all of the data declarations
6428 before the instructions in the text section, this allows the MIPS
6429 assembler to generate one word memory references instead of using two
6430 words for short global or static data items. This is on by default if
6431 optimization is selected.
6435 For each non-inline function processed, the @samp{-mstats} switch
6436 causes the compiler to emit one line to the standard error file to
6437 print statistics about the program (number of registers saved, stack
6442 The @samp{-mmemcpy} switch makes all block moves call the appropriate
6443 string function (@samp{memcpy} or @samp{bcopy}) instead of possibly
6444 generating inline code.
6447 @itemx -mno-mips-tfile
6448 The @samp{-mno-mips-tfile} switch causes the compiler not
6449 postprocess the object file with the @file{mips-tfile} program,
6450 after the MIPS assembler has generated it to add debug support. If
6451 @file{mips-tfile} is not run, then no local variables will be
6452 available to the debugger. In addition, @file{stage2} and
6453 @file{stage3} objects will have the temporary file names passed to the
6454 assembler embedded in the object file, which means the objects will
6455 not compare the same. The @samp{-mno-mips-tfile} switch should only
6456 be used when there are bugs in the @file{mips-tfile} program that
6457 prevents compilation.
6460 Generate output containing library calls for floating point.
6461 @strong{Warning:} the requisite libraries are not part of GCC.
6462 Normally the facilities of the machine's usual C compiler are used, but
6463 this can't be done directly in cross-compilation. You must make your
6464 own arrangements to provide suitable library functions for
6468 Generate output containing floating point instructions. This is the
6469 default if you use the unmodified sources.
6472 @itemx -mno-abicalls
6473 Emit (or do not emit) the pseudo operations @samp{.abicalls},
6474 @samp{.cpload}, and @samp{.cprestore} that some System V.4 ports use for
6475 position independent code.
6478 @itemx -mno-long-calls
6479 Do all calls with the @samp{JALR} instruction, which requires
6480 loading up a function's address into a register before the call.
6481 You need to use this switch, if you call outside of the current
6482 512 megabyte segment to functions that are not through pointers.
6485 @itemx -mno-half-pic
6486 Put pointers to extern references into the data section and load them
6487 up, rather than put the references in the text section.
6489 @item -membedded-pic
6490 @itemx -mno-embedded-pic
6491 Generate PIC code suitable for some embedded systems. All calls are
6492 made using PC relative address, and all data is addressed using the $gp
6493 register. No more than 65536 bytes of global data may be used. This
6494 requires GNU as and GNU ld which do most of the work. This currently
6495 only works on targets which use ECOFF; it does not work with ELF.
6497 @item -membedded-data
6498 @itemx -mno-embedded-data
6499 Allocate variables to the read-only data section first if possible, then
6500 next in the small data section if possible, otherwise in data. This gives
6501 slightly slower code than the default, but reduces the amount of RAM required
6502 when executing, and thus may be preferred for some embedded systems.
6504 @item -muninit-const-in-rodata
6505 @itemx -mno-uninit-const-in-rodata
6506 When used together with -membedded-data, it will always store uninitialized
6507 const variables in the read-only data section.
6509 @item -msingle-float
6510 @itemx -mdouble-float
6511 The @samp{-msingle-float} switch tells gcc to assume that the floating
6512 point coprocessor only supports single precision operations, as on the
6513 @samp{r4650} chip. The @samp{-mdouble-float} switch permits gcc to use
6514 double precision operations. This is the default.
6518 Permit use of the @samp{mad}, @samp{madu} and @samp{mul} instructions,
6519 as on the @samp{r4650} chip.
6522 Turns on @samp{-msingle-float}, @samp{-mmad}, and, at least for now,
6527 Enable 16-bit instructions.
6530 Use the entry and exit pseudo ops. This option can only be used with
6534 Compile code for the processor in little endian mode.
6535 The requisite libraries are assumed to exist.
6538 Compile code for the processor in big endian mode.
6539 The requisite libraries are assumed to exist.
6542 @cindex smaller data references (MIPS)
6543 @cindex gp-relative references (MIPS)
6544 Put global and static items less than or equal to @var{num} bytes into
6545 the small data or bss sections instead of the normal data or bss
6546 section. This allows the assembler to emit one word memory reference
6547 instructions based on the global pointer (@var{gp} or @var{$28}),
6548 instead of the normal two words used. By default, @var{num} is 8 when
6549 the MIPS assembler is used, and 0 when the GNU assembler is used. The
6550 @samp{-G @var{num}} switch is also passed to the assembler and linker.
6551 All modules should be compiled with the same @samp{-G @var{num}}
6555 Tell the MIPS assembler to not run its preprocessor over user
6556 assembler files (with a @samp{.s} suffix) when assembling them.
6559 Pass an option to gas which will cause nops to be inserted if
6560 the read of the destination register of an mfhi or mflo instruction
6561 occurs in the following two instructions.
6564 Do not include the default crt0.
6568 These options are defined by the macro
6569 @code{TARGET_SWITCHES} in the machine description. The default for the
6570 options is also defined by that macro, which enables you to change the
6575 @subsection Intel 386 Options
6576 @cindex i386 Options
6577 @cindex Intel 386 Options
6579 These @samp{-m} options are defined for the i386 family of computers:
6582 @item -mcpu=@var{cpu type}
6583 Assume the defaults for the machine type @var{cpu type} when scheduling
6584 instructions. The choices for @var{cpu type} are @samp{i386},
6585 @samp{i486}, @samp{i586}, @samp{i686}, @samp{pentium},
6586 @samp{pentiumpro}, @samp{pentium4}, @samp{k6}, and @samp{athlon}
6588 While picking a specific @var{cpu type} will schedule things appropriately
6589 for that particular chip, the compiler will not generate any code that
6590 does not run on the i386 without the @samp{-march=@var{cpu type}} option
6591 being used. @samp{i586} is equivalent to @samp{pentium} and @samp{i686}
6592 is equivalent to @samp{pentiumpro}. @samp{k6} and @samp{athlon} are the
6593 AMD chips as opposed to the Intel ones.
6595 @item -march=@var{cpu type}
6596 Generate instructions for the machine type @var{cpu type}. The choices
6597 for @var{cpu type} are the same as for @samp{-mcpu}. Moreover,
6598 specifying @samp{-march=@var{cpu type}} implies @samp{-mcpu=@var{cpu type}}.
6604 Synonyms for -mcpu=i386, -mcpu=i486, -mcpu=pentium, and -mcpu=pentiumpro
6605 respectively. These synonyms are deprecated.
6607 @item -mintel-syntax
6608 Emit assembly using Intel syntax opcodes instead of AT&T syntax.
6612 Control whether or not the compiler uses IEEE floating point
6613 comparisons. These handle correctly the case where the result of a
6614 comparison is unordered.
6617 Generate output containing library calls for floating point.
6618 @strong{Warning:} the requisite libraries are not part of GCC.
6619 Normally the facilities of the machine's usual C compiler are used, but
6620 this can't be done directly in cross-compilation. You must make your
6621 own arrangements to provide suitable library functions for
6624 On machines where a function returns floating point results in the 80387
6625 register stack, some floating point opcodes may be emitted even if
6626 @samp{-msoft-float} is used.
6628 @item -mno-fp-ret-in-387
6629 Do not use the FPU registers for return values of functions.
6631 The usual calling convention has functions return values of types
6632 @code{float} and @code{double} in an FPU register, even if there
6633 is no FPU. The idea is that the operating system should emulate
6636 The option @samp{-mno-fp-ret-in-387} causes such values to be returned
6637 in ordinary CPU registers instead.
6639 @item -mno-fancy-math-387
6640 Some 387 emulators do not support the @code{sin}, @code{cos} and
6641 @code{sqrt} instructions for the 387. Specify this option to avoid
6642 generating those instructions. This option is the default on FreeBSD.
6643 As of revision 2.6.1, these instructions are not generated unless you
6644 also use the @samp{-funsafe-math-optimizations} switch.
6646 @item -malign-double
6647 @itemx -mno-align-double
6648 Control whether GCC aligns @code{double}, @code{long double}, and
6649 @code{long long} variables on a two word boundary or a one word
6650 boundary. Aligning @code{double} variables on a two word boundary will
6651 produce code that runs somewhat faster on a @samp{Pentium} at the
6652 expense of more memory.
6654 @item -m128bit-long-double
6655 @itemx -m128bit-long-double
6656 Control the size of @code{long double} type. i386 application binary interface
6657 specify the size to be 12 bytes, while modern architectures (Pentium and newer)
6658 preffer @code{long double} aligned to 8 or 16 byte boundary. This is
6659 impossible to reach with 12 byte long doubles in the array accesses.
6661 @strong{Warning:} if you use the @samp{-m128bit-long-double} switch, the
6662 structures and arrays containing @code{long double} will change their size as
6663 well as function calling convention for function taking @code{long double}
6666 @item -m96bit-long-double
6667 @itemx -m96bit-long-double
6668 Set the size of @code{long double} to 96 bits as required by the i386
6669 application binary interface. This is the default.
6672 @itemx -mno-svr3-shlib
6673 Control whether GCC places uninitialized locals into @code{bss} or
6674 @code{data}. @samp{-msvr3-shlib} places these locals into @code{bss}.
6675 These options are meaningful only on System V Release 3.
6677 @item -mno-wide-multiply
6678 @itemx -mwide-multiply
6679 Control whether GCC uses the @code{mul} and @code{imul} that produce
6680 64 bit results in @code{eax:edx} from 32 bit operands to do @code{long
6681 long} multiplies and 32-bit division by constants.
6684 Use a different function-calling convention, in which functions that
6685 take a fixed number of arguments return with the @code{ret} @var{num}
6686 instruction, which pops their arguments while returning. This saves one
6687 instruction in the caller since there is no need to pop the arguments
6690 You can specify that an individual function is called with this calling
6691 sequence with the function attribute @samp{stdcall}. You can also
6692 override the @samp{-mrtd} option by using the function attribute
6693 @samp{cdecl}. @xref{Function Attributes}.
6695 @strong{Warning:} this calling convention is incompatible with the one
6696 normally used on Unix, so you cannot use it if you need to call
6697 libraries compiled with the Unix compiler.
6699 Also, you must provide function prototypes for all functions that
6700 take variable numbers of arguments (including @code{printf});
6701 otherwise incorrect code will be generated for calls to those
6704 In addition, seriously incorrect code will result if you call a
6705 function with too many arguments. (Normally, extra arguments are
6706 harmlessly ignored.)
6708 @item -mregparm=@var{num}
6709 Control how many registers are used to pass integer arguments. By
6710 default, no registers are used to pass arguments, and at most 3
6711 registers can be used. You can control this behavior for a specific
6712 function by using the function attribute @samp{regparm}.
6713 @xref{Function Attributes}.
6715 @strong{Warning:} if you use this switch, and
6716 @var{num} is nonzero, then you must build all modules with the same
6717 value, including any libraries. This includes the system libraries and
6720 @item -malign-loops=@var{num}
6721 Align loops to a 2 raised to a @var{num} byte boundary. If
6722 @samp{-malign-loops} is not specified, the default is 2 unless
6723 gas 2.8 (or later) is being used in which case the default is
6724 to align the loop on a 16 byte boundary if it is less than 8
6727 @item -malign-jumps=@var{num}
6728 Align instructions that are only jumped to to a 2 raised to a @var{num}
6729 byte boundary. If @samp{-malign-jumps} is not specified, the default is
6730 2 if optimizing for a 386, and 4 if optimizing for a 486 unless
6731 gas 2.8 (or later) is being used in which case the default is
6732 to align the instruction on a 16 byte boundary if it is less
6735 @item -malign-functions=@var{num}
6736 Align the start of functions to a 2 raised to @var{num} byte boundary.
6737 If @samp{-malign-functions} is not specified, the default is 2 if optimizing
6738 for a 386, and 4 if optimizing for a 486.
6740 @item -mpreferred-stack-boundary=@var{num}
6741 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
6742 byte boundary. If @samp{-mpreferred-stack-boundary} is not specified,
6743 the default is 4 (16 bytes or 128 bits).
6745 The stack is required to be aligned on a 4 byte boundary. On Pentium
6746 and PentiumPro, @code{double} and @code{long double} values should be
6747 aligned to an 8 byte boundary (see @samp{-malign-double}) or suffer
6748 significant run time performance penalties. On Pentium III, the
6749 Streaming SIMD Extension (SSE) data type @code{__m128} suffers similar
6750 penalties if it is not 16 byte aligned.
6752 To ensure proper alignment of this values on the stack, the stack boundary
6753 must be as aligned as that required by any value stored on the stack.
6754 Further, every function must be generated such that it keeps the stack
6755 aligned. Thus calling a function compiled with a higher preferred
6756 stack boundary from a function compiled with a lower preferred stack
6757 boundary will most likely misalign the stack. It is recommended that
6758 libraries that use callbacks always use the default setting.
6760 This extra alignment does consume extra stack space. Code that is sensitive
6761 to stack space usage, such as embedded systems and operating system kernels,
6762 may want to reduce the preferred alignment to
6763 @samp{-mpreferred-stack-boundary=2}.
6767 Use PUSH operations to store outgoing parameters. This method is shorter
6768 and usually equally fast as method using SUB/MOV operations and is enabled
6769 by default. In some cases disabling it may improve performance because of
6770 improved scheduling and reduced dependencies.
6772 @item -maccumulate-outgoing-args
6773 @kindex -maccumulate-outgoing-args
6774 If enabled, the maximum amount of space required for outgoing arguments will be
6775 computed in the function prologue. This in faster on most modern CPUs
6776 because of reduced dependencies, improved scheduling and reduced stack usage
6777 when preferred stack boundary is not equal to 2. The drawback is a notable
6778 increase in code size. This switch implies -mno-push-args.
6782 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
6783 on thread-safe exception handling must compile and link all code with the
6784 @samp{-mthreads} option. When compiling, @samp{-mthreads} defines
6785 @samp{-D_MT}; when linking, it links in a special thread helper library
6786 @samp{-lmingwthrd} which cleans up per thread exception handling data.
6788 @item -mno-align-stringops
6789 @kindex -mno-align-stringops
6790 Do not align destination of inlined string operations. This switch reduces
6791 code size and improves performance in case the destination is already aligned,
6792 but gcc don't know about it.
6794 @item -minline-all-stringops
6795 @kindex -minline-all-stringops
6796 By default GCC inlines string operations only when destination is known to be
6797 aligned at least to 4 byte boundary. This enables more inlining, increase code
6798 size, but may improve performance of code that depends on fast memcpy, strlen
6799 and memset for short lengths.
6803 @subsection HPPA Options
6804 @cindex HPPA Options
6806 These @samp{-m} options are defined for the HPPA family of computers:
6809 @item -march=@var{architecture type}
6810 Generate code for the specified architecture. The choices for
6811 @var{architecture type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
6812 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
6813 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
6814 architecture option for your machine. Code compiled for lower numbered
6815 architectures will run on higher numbered architectures, but not the
6818 PA 2.0 support currently requires gas snapshot 19990413 or later. The
6819 next release of binutils (current is 2.9.1) will probably contain PA 2.0
6823 @itemx -mpa-risc-1-1
6824 @itemx -mpa-risc-2-0
6825 Synonyms for -march=1.0, -march=1.1, and -march=2.0 respectively.
6828 Generate code suitable for big switch tables. Use this option only if
6829 the assembler/linker complain about out of range branches within a switch
6832 @item -mjump-in-delay
6833 Fill delay slots of function calls with unconditional jump instructions
6834 by modifying the return pointer for the function call to be the target
6835 of the conditional jump.
6837 @item -mdisable-fpregs
6838 Prevent floating point registers from being used in any manner. This is
6839 necessary for compiling kernels which perform lazy context switching of
6840 floating point registers. If you use this option and attempt to perform
6841 floating point operations, the compiler will abort.
6843 @item -mdisable-indexing
6844 Prevent the compiler from using indexing address modes. This avoids some
6845 rather obscure problems when compiling MIG generated code under MACH.
6847 @item -mno-space-regs
6848 Generate code that assumes the target has no space registers. This allows
6849 GCC to generate faster indirect calls and use unscaled index address modes.
6851 Such code is suitable for level 0 PA systems and kernels.
6853 @item -mfast-indirect-calls
6854 Generate code that assumes calls never cross space boundaries. This
6855 allows GCC to emit code which performs faster indirect calls.
6857 This option will not work in the presence of shared libraries or nested
6860 @item -mlong-load-store
6861 Generate 3-instruction load and store sequences as sometimes required by
6862 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
6865 @item -mportable-runtime
6866 Use the portable calling conventions proposed by HP for ELF systems.
6869 Enable the use of assembler directives only GAS understands.
6871 @item -mschedule=@var{cpu type}
6872 Schedule code according to the constraints for the machine type
6873 @var{cpu type}. The choices for @var{cpu type} are @samp{700}
6874 @samp{7100}, @samp{7100LC}, @samp{7200}, and @samp{8000}. Refer to
6875 @file{/usr/lib/sched.models} on an HP-UX system to determine the
6876 proper scheduling option for your machine.
6879 Enable the optimization pass in the HPUX linker. Note this makes symbolic
6880 debugging impossible. It also triggers a bug in the HPUX 8 and HPUX 9 linkers
6881 in which they give bogus error messages when linking some programs.
6884 Generate output containing library calls for floating point.
6885 @strong{Warning:} the requisite libraries are not available for all HPPA
6886 targets. Normally the facilities of the machine's usual C compiler are
6887 used, but this cannot be done directly in cross-compilation. You must make
6888 your own arrangements to provide suitable library functions for
6889 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
6890 does provide software floating point support.
6892 @samp{-msoft-float} changes the calling convention in the output file;
6893 therefore, it is only useful if you compile @emph{all} of a program with
6894 this option. In particular, you need to compile @file{libgcc.a}, the
6895 library that comes with GCC, with @samp{-msoft-float} in order for
6899 @node Intel 960 Options
6900 @subsection Intel 960 Options
6902 These @samp{-m} options are defined for the Intel 960 implementations:
6905 @item -m@var{cpu type}
6906 Assume the defaults for the machine type @var{cpu type} for some of
6907 the other options, including instruction scheduling, floating point
6908 support, and addressing modes. The choices for @var{cpu type} are
6909 @samp{ka}, @samp{kb}, @samp{mc}, @samp{ca}, @samp{cf},
6910 @samp{sa}, and @samp{sb}.
6916 The @samp{-mnumerics} option indicates that the processor does support
6917 floating-point instructions. The @samp{-msoft-float} option indicates
6918 that floating-point support should not be assumed.
6920 @item -mleaf-procedures
6921 @itemx -mno-leaf-procedures
6922 Do (or do not) attempt to alter leaf procedures to be callable with the
6923 @code{bal} instruction as well as @code{call}. This will result in more
6924 efficient code for explicit calls when the @code{bal} instruction can be
6925 substituted by the assembler or linker, but less efficient code in other
6926 cases, such as calls via function pointers, or using a linker that doesn't
6927 support this optimization.
6930 @itemx -mno-tail-call
6931 Do (or do not) make additional attempts (beyond those of the
6932 machine-independent portions of the compiler) to optimize tail-recursive
6933 calls into branches. You may not want to do this because the detection of
6934 cases where this is not valid is not totally complete. The default is
6935 @samp{-mno-tail-call}.
6937 @item -mcomplex-addr
6938 @itemx -mno-complex-addr
6939 Assume (or do not assume) that the use of a complex addressing mode is a
6940 win on this implementation of the i960. Complex addressing modes may not
6941 be worthwhile on the K-series, but they definitely are on the C-series.
6942 The default is currently @samp{-mcomplex-addr} for all processors except
6946 @itemx -mno-code-align
6947 Align code to 8-byte boundaries for faster fetching (or don't bother).
6948 Currently turned on by default for C-series implementations only.
6951 @item -mclean-linkage
6952 @itemx -mno-clean-linkage
6953 These options are not fully implemented.
6957 @itemx -mic2.0-compat
6958 @itemx -mic3.0-compat
6959 Enable compatibility with iC960 v2.0 or v3.0.
6963 Enable compatibility with the iC960 assembler.
6965 @item -mstrict-align
6966 @itemx -mno-strict-align
6967 Do not permit (do permit) unaligned accesses.
6970 Enable structure-alignment compatibility with Intel's gcc release version
6971 1.3 (based on gcc 1.37). This option implies @samp{-mstrict-align}.
6973 @item -mlong-double-64
6974 Implement type @samp{long double} as 64-bit floating point numbers.
6975 Without the option @samp{long double} is implemented by 80-bit
6976 floating point numbers. The only reason we have it because there is
6977 no 128-bit @samp{long double} support in @samp{fp-bit.c} yet. So it
6978 is only useful for people using soft-float targets. Otherwise, we
6979 should recommend against use of it.
6983 @node DEC Alpha Options
6984 @subsection DEC Alpha Options
6986 These @samp{-m} options are defined for the DEC Alpha implementations:
6989 @item -mno-soft-float
6991 Use (do not use) the hardware floating-point instructions for
6992 floating-point operations. When @option{-msoft-float} is specified,
6993 functions in @file{libgcc1.c} will be used to perform floating-point
6994 operations. Unless they are replaced by routines that emulate the
6995 floating-point operations, or compiled in such a way as to call such
6996 emulations routines, these routines will issue floating-point
6997 operations. If you are compiling for an Alpha without floating-point
6998 operations, you must ensure that the library is built so as not to call
7001 Note that Alpha implementations without floating-point operations are
7002 required to have floating-point registers.
7006 Generate code that uses (does not use) the floating-point register set.
7007 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
7008 register set is not used, floating point operands are passed in integer
7009 registers as if they were integers and floating-point results are passed
7010 in $0 instead of $f0. This is a non-standard calling sequence, so any
7011 function with a floating-point argument or return value called by code
7012 compiled with @option{-mno-fp-regs} must also be compiled with that
7015 A typical use of this option is building a kernel that does not use,
7016 and hence need not save and restore, any floating-point registers.
7019 The Alpha architecture implements floating-point hardware optimized for
7020 maximum performance. It is mostly compliant with the IEEE floating
7021 point standard. However, for full compliance, software assistance is
7022 required. This option generates code fully IEEE compliant code
7023 @emph{except} that the @var{inexact flag} is not maintained (see below).
7024 If this option is turned on, the CPP macro @code{_IEEE_FP} is defined
7025 during compilation. The option is a shorthand for: @samp{-D_IEEE_FP
7026 -mfp-trap-mode=su -mtrap-precision=i -mieee-conformant}. The resulting
7027 code is less efficient but is able to correctly support denormalized
7028 numbers and exceptional IEEE values such as not-a-number and plus/minus
7029 infinity. Other Alpha compilers call this option
7030 @option{-ieee_with_no_inexact}.
7032 @item -mieee-with-inexact
7033 @c overfull hbox here --bob 22 jul96
7034 @c original text between ignore ... end ignore
7036 This is like @samp{-mieee} except the generated code also maintains the
7037 IEEE @var{inexact flag}. Turning on this option causes the generated
7038 code to implement fully-compliant IEEE math. The option is a shorthand
7039 for @samp{-D_IEEE_FP -D_IEEE_FP_INEXACT} plus @samp{-mieee-conformant},
7040 @samp{-mfp-trap-mode=sui}, and @samp{-mtrap-precision=i}. On some Alpha
7041 implementations the resulting code may execute significantly slower than
7042 the code generated by default. Since there is very little code that
7043 depends on the @var{inexact flag}, you should normally not specify this
7044 option. Other Alpha compilers call this option
7045 @samp{-ieee_with_inexact}.
7047 @c changed paragraph
7048 This is like @samp{-mieee} except the generated code also maintains the
7049 IEEE @var{inexact flag}. Turning on this option causes the generated
7050 code to implement fully-compliant IEEE math. The option is a shorthand
7051 for @samp{-D_IEEE_FP -D_IEEE_FP_INEXACT} plus the three following:
7052 @samp{-mieee-conformant},
7053 @samp{-mfp-trap-mode=sui},
7054 and @samp{-mtrap-precision=i}.
7055 On some Alpha implementations the resulting code may execute
7056 significantly slower than the code generated by default. Since there
7057 is very little code that depends on the @var{inexact flag}, you should
7058 normally not specify this option. Other Alpha compilers call this
7059 option @samp{-ieee_with_inexact}.
7060 @c end changes to prevent overfull hboxes
7062 @item -mfp-trap-mode=@var{trap mode}
7063 This option controls what floating-point related traps are enabled.
7064 Other Alpha compilers call this option @samp{-fptm }@var{trap mode}.
7065 The trap mode can be set to one of four values:
7069 This is the default (normal) setting. The only traps that are enabled
7070 are the ones that cannot be disabled in software (e.g., division by zero
7074 In addition to the traps enabled by @samp{n}, underflow traps are enabled
7078 Like @samp{su}, but the instructions are marked to be safe for software
7079 completion (see Alpha architecture manual for details).
7082 Like @samp{su}, but inexact traps are enabled as well.
7085 @item -mfp-rounding-mode=@var{rounding mode}
7086 Selects the IEEE rounding mode. Other Alpha compilers call this option
7087 @samp{-fprm }@var{rounding mode}. The @var{rounding mode} can be one
7092 Normal IEEE rounding mode. Floating point numbers are rounded towards
7093 the nearest machine number or towards the even machine number in case
7097 Round towards minus infinity.
7100 Chopped rounding mode. Floating point numbers are rounded towards zero.
7103 Dynamic rounding mode. A field in the floating point control register
7104 (@var{fpcr}, see Alpha architecture reference manual) controls the
7105 rounding mode in effect. The C library initializes this register for
7106 rounding towards plus infinity. Thus, unless your program modifies the
7107 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
7110 @item -mtrap-precision=@var{trap precision}
7111 In the Alpha architecture, floating point traps are imprecise. This
7112 means without software assistance it is impossible to recover from a
7113 floating trap and program execution normally needs to be terminated.
7114 GCC can generate code that can assist operating system trap handlers
7115 in determining the exact location that caused a floating point trap.
7116 Depending on the requirements of an application, different levels of
7117 precisions can be selected:
7121 Program precision. This option is the default and means a trap handler
7122 can only identify which program caused a floating point exception.
7125 Function precision. The trap handler can determine the function that
7126 caused a floating point exception.
7129 Instruction precision. The trap handler can determine the exact
7130 instruction that caused a floating point exception.
7133 Other Alpha compilers provide the equivalent options called
7134 @samp{-scope_safe} and @samp{-resumption_safe}.
7136 @item -mieee-conformant
7137 This option marks the generated code as IEEE conformant. You must not
7138 use this option unless you also specify @samp{-mtrap-precision=i} and either
7139 @samp{-mfp-trap-mode=su} or @samp{-mfp-trap-mode=sui}. Its only effect
7140 is to emit the line @samp{.eflag 48} in the function prologue of the
7141 generated assembly file. Under DEC Unix, this has the effect that
7142 IEEE-conformant math library routines will be linked in.
7144 @item -mbuild-constants
7145 Normally GCC examines a 32- or 64-bit integer constant to
7146 see if it can construct it from smaller constants in two or three
7147 instructions. If it cannot, it will output the constant as a literal and
7148 generate code to load it from the data segment at runtime.
7150 Use this option to require GCC to construct @emph{all} integer constants
7151 using code, even if it takes more instructions (the maximum is six).
7153 You would typically use this option to build a shared library dynamic
7154 loader. Itself a shared library, it must relocate itself in memory
7155 before it can find the variables and constants in its own data segment.
7159 Select whether to generate code to be assembled by the vendor-supplied
7160 assembler (@samp{-malpha-as}) or by the GNU assembler @samp{-mgas}.
7168 Indicate whether GCC should generate code to use the optional BWX,
7169 CIX, and MAX instruction sets. The default is to use the instruction sets
7170 supported by the CPU type specified via @samp{-mcpu=} option or that
7171 of the CPU on which GCC was built if none was specified.
7173 @item -mcpu=@var{cpu_type}
7174 Set the instruction set, register set, and instruction scheduling
7175 parameters for machine type @var{cpu_type}. You can specify either the
7176 @samp{EV} style name or the corresponding chip number. GCC
7177 supports scheduling parameters for the EV4 and EV5 family of processors
7178 and will choose the default values for the instruction set from
7179 the processor you specify. If you do not specify a processor type,
7180 GCC will default to the processor on which the compiler was built.
7182 Supported values for @var{cpu_type} are
7187 Schedules as an EV4 and has no instruction set extensions.
7191 Schedules as an EV5 and has no instruction set extensions.
7195 Schedules as an EV5 and supports the BWX extension.
7200 Schedules as an EV5 and supports the BWX and MAX extensions.
7204 Schedules as an EV5 (until Digital releases the scheduling parameters
7205 for the EV6) and supports the BWX, CIX, and MAX extensions.
7208 @item -mmemory-latency=@var{time}
7209 Sets the latency the scheduler should assume for typical memory
7210 references as seen by the application. This number is highly
7211 dependent on the memory access patterns used by the application
7212 and the size of the external cache on the machine.
7214 Valid options for @var{time} are
7218 A decimal number representing clock cycles.
7224 The compiler contains estimates of the number of clock cycles for
7225 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
7226 (also called Dcache, Scache, and Bcache), as well as to main memory.
7227 Note that L3 is only valid for EV5.
7232 @node Clipper Options
7233 @subsection Clipper Options
7235 These @samp{-m} options are defined for the Clipper implementations:
7239 Produce code for a C300 Clipper processor. This is the default.
7242 Produce code for a C400 Clipper processor i.e. use floating point
7246 @node H8/300 Options
7247 @subsection H8/300 Options
7249 These @samp{-m} options are defined for the H8/300 implementations:
7253 Shorten some address references at link time, when possible; uses the
7254 linker option @samp{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
7255 ld.info, Using ld}, for a fuller description.
7258 Generate code for the H8/300H.
7261 Generate code for the H8/S.
7264 Generate code for the H8/S2600. This switch must be used with -ms.
7267 Make @code{int} data 32 bits by default.
7270 On the H8/300H and H8/S, use the same alignment rules as for the H8/300.
7271 The default for the H8/300H and H8/S is to align longs and floats on 4
7273 @samp{-malign-300} causes them to be aligned on 2 byte boundaries.
7274 This option has no effect on the H8/300.
7278 @subsection SH Options
7280 These @samp{-m} options are defined for the SH implementations:
7284 Generate code for the SH1.
7287 Generate code for the SH2.
7290 Generate code for the SH3.
7293 Generate code for the SH3e.
7296 Generate code for the SH4 without a floating-point unit.
7298 @item -m4-single-only
7299 Generate code for the SH4 with a floating-point unit that only
7300 supports single-precision arithmentic.
7303 Generate code for the SH4 assuming the floating-point unit is in
7304 single-precision mode by default.
7307 Generate code for the SH4.
7310 Compile code for the processor in big endian mode.
7313 Compile code for the processor in little endian mode.
7316 Align doubles at 64 bit boundaries. Note that this changes the calling
7317 conventions, and thus some functions from the standard C library will
7318 not work unless you recompile it first with -mdalign.
7321 Shorten some address references at link time, when possible; uses the
7322 linker option @samp{-relax}.
7325 Use 32-bit offsets in @code{switch} tables. The default is to use
7329 Enable the use of the instruction @code{fmovd}.
7332 Comply with the calling conventions defined by Hitachi.
7335 Mark the @code{MAC} register as call-clobbered, even if
7336 @option{-mhitachi} is given.
7339 Dump instruction size and location in the assembly code.
7342 This option is deprecated. It pads structures to multiple of 4 bytes,
7343 which is incompatible with the SH ABI.
7346 Optimize for space instead of speed. Implied by @option{-Os}.
7349 When generating position-independent code, emit function calls using
7350 the Global Offset Table instead of the Procedure Linkage Table.
7353 Generate a library function call to invalidate instruction cache
7354 entries, after fixing up a trampoline. This library function call
7355 doesn't assume it can write to the whole memory address space. This
7356 is the default when the target is @code{sh-*-linux*}.
7359 @node System V Options
7360 @subsection Options for System V
7362 These additional options are available on System V Release 4 for
7363 compatibility with other compilers on those systems:
7367 Create a shared object.
7368 It is recommended that @samp{-symbolic} or @samp{-shared} be used instead.
7371 Identify the versions of each tool used by the compiler, in a
7372 @code{.ident} assembler directive in the output.
7375 Refrain from adding @code{.ident} directives to the output file (this is
7378 @item -YP\,@var{dirs}
7379 Search the directories @var{dirs}, and no others, for libraries
7380 specified with @samp{-l}.
7382 @item -Ym\,@var{dir}
7383 Look in the directory @var{dir} to find the M4 preprocessor.
7384 The assembler uses this option.
7385 @c This is supposed to go with a -Yd for predefined M4 macro files, but
7386 @c the generic assembler that comes with Solaris takes just -Ym.
7389 @node TMS320C3x/C4x Options
7390 @subsection TMS320C3x/C4x Options
7391 @cindex TMS320C3x/C4x Options
7393 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
7397 @item -mcpu=@var{cpu_type}
7398 Set the instruction set, register set, and instruction scheduling
7399 parameters for machine type @var{cpu_type}. Supported values for
7400 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
7401 @samp{c44}. The default is @samp{c40} to generate code for the
7406 @itemx -msmall-memory
7408 Generates code for the big or small memory model. The small memory
7409 model assumed that all data fits into one 64K word page. At run-time
7410 the data page (DP) register must be set to point to the 64K page
7411 containing the .bss and .data program sections. The big memory model is
7412 the default and requires reloading of the DP register for every direct
7417 Allow (disallow) allocation of general integer operands into the block
7422 Enable (disable) generation of code using decrement and branch,
7423 DBcond(D), instructions. This is enabled by default for the C4x. To be
7424 on the safe side, this is disabled for the C3x, since the maximum
7425 iteration count on the C3x is 2^23 + 1 (but who iterates loops more than
7426 2^23 times on the C3x?). Note that GCC will try to reverse a loop so
7427 that it can utilise the decrement and branch instruction, but will give
7428 up if there is more than one memory reference in the loop. Thus a loop
7429 where the loop counter is decremented can generate slightly more
7430 efficient code, in cases where the RPTB instruction cannot be utilised.
7432 @item -mdp-isr-reload
7434 Force the DP register to be saved on entry to an interrupt service
7435 routine (ISR), reloaded to point to the data section, and restored on
7436 exit from the ISR. This should not be required unless someone has
7437 violated the small memory model by modifying the DP register, say within
7442 For the C3x use the 24-bit MPYI instruction for integer multiplies
7443 instead of a library call to guarantee 32-bit results. Note that if one
7444 of the operands is a constant, then the multiplication will be performed
7445 using shifts and adds. If the -mmpyi option is not specified for the C3x,
7446 then squaring operations are performed inline instead of a library call.
7449 @itemx -mno-fast-fix
7450 The C3x/C4x FIX instruction to convert a floating point value to an
7451 integer value chooses the nearest integer less than or equal to the
7452 floating point value rather than to the nearest integer. Thus if the
7453 floating point number is negative, the result will be incorrectly
7454 truncated an additional code is necessary to detect and correct this
7455 case. This option can be used to disable generation of the additional
7456 code required to correct the result.
7460 Enable (disable) generation of repeat block sequences using the RPTB
7461 instruction for zero overhead looping. The RPTB construct is only used
7462 for innermost loops that do not call functions or jump across the loop
7463 boundaries. There is no advantage having nested RPTB loops due to the
7464 overhead required to save and restore the RC, RS, and RE registers.
7465 This is enabled by default with -O2.
7467 @item -mrpts=@var{count}
7469 Enable (disable) the use of the single instruction repeat instruction
7470 RPTS. If a repeat block contains a single instruction, and the loop
7471 count can be guaranteed to be less than the value @var{count}, GCC will
7472 emit a RPTS instruction instead of a RPTB. If no value is specified,
7473 then a RPTS will be emitted even if the loop count cannot be determined
7474 at compile time. Note that the repeated instruction following RPTS does
7475 not have to be reloaded from memory each iteration, thus freeing up the
7476 CPU buses for operands. However, since interrupts are blocked by this
7477 instruction, it is disabled by default.
7479 @item -mloop-unsigned
7480 @itemx -mno-loop-unsigned
7481 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
7482 is 2^31 + 1 since these instructions test if the iteration count is
7483 negative to terminate the loop. If the iteration count is unsigned
7484 there is a possibility than the 2^31 + 1 maximum iteration count may be
7485 exceeded. This switch allows an unsigned iteration count.
7488 Try to emit an assembler syntax that the TI assembler (asm30) is happy
7489 with. This also enforces compatibility with the API employed by the TI
7490 C3x C compiler. For example, long doubles are passed as structures
7491 rather than in floating point registers.
7495 Generate code that uses registers (stack) for passing arguments to functions.
7496 By default, arguments are passed in registers where possible rather
7497 than by pushing arguments on to the stack.
7499 @item -mparallel-insns
7500 @itemx -mno-parallel-insns
7501 Allow the generation of parallel instructions. This is enabled by
7504 @item -mparallel-mpy
7505 @itemx -mno-parallel-mpy
7506 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
7507 provided -mparallel-insns is also specified. These instructions have
7508 tight register constraints which can pessimize the code generation
7514 @subsection V850 Options
7515 @cindex V850 Options
7517 These @samp{-m} options are defined for V850 implementations:
7521 @itemx -mno-long-calls
7522 Treat all calls as being far away (near). If calls are assumed to be
7523 far away, the compiler will always load the functions address up into a
7524 register, and call indirect through the pointer.
7528 Do not optimize (do optimize) basic blocks that use the same index
7529 pointer 4 or more times to copy pointer into the @code{ep} register, and
7530 use the shorter @code{sld} and @code{sst} instructions. The @samp{-mep}
7531 option is on by default if you optimize.
7533 @item -mno-prolog-function
7534 @itemx -mprolog-function
7535 Do not use (do use) external functions to save and restore registers at
7536 the prolog and epilog of a function. The external functions are slower,
7537 but use less code space if more than one function saves the same number
7538 of registers. The @samp{-mprolog-function} option is on by default if
7542 Try to make the code as small as possible. At present, this just turns
7543 on the @samp{-mep} and @samp{-mprolog-function} options.
7546 Put static or global variables whose size is @var{n} bytes or less into
7547 the tiny data area that register @code{ep} points to. The tiny data
7548 area can hold up to 256 bytes in total (128 bytes for byte references).
7551 Put static or global variables whose size is @var{n} bytes or less into
7552 the small data area that register @code{gp} points to. The small data
7553 area can hold up to 64 kilobytes.
7556 Put static or global variables whose size is @var{n} bytes or less into
7557 the first 32 kilobytes of memory.
7560 Specify that the target processor is the V850.
7563 Generate code suitable for big switch tables. Use this option only if
7564 the assembler/linker complain about out of range branches within a switch
7569 @subsection ARC Options
7572 These options are defined for ARC implementations:
7576 Compile code for little endian mode. This is the default.
7579 Compile code for big endian mode.
7582 Prepend the name of the cpu to all public symbol names.
7583 In multiple-processor systems, there are many ARC variants with different
7584 instruction and register set characteristics. This flag prevents code
7585 compiled for one cpu to be linked with code compiled for another.
7586 No facility exists for handling variants that are "almost identical".
7587 This is an all or nothing option.
7589 @item -mcpu=@var{cpu}
7590 Compile code for ARC variant @var{cpu}.
7591 Which variants are supported depend on the configuration.
7592 All variants support @samp{-mcpu=base}, this is the default.
7594 @item -mtext=@var{text section}
7595 @itemx -mdata=@var{data section}
7596 @itemx -mrodata=@var{readonly data section}
7597 Put functions, data, and readonly data in @var{text section},
7598 @var{data section}, and @var{readonly data section} respectively
7599 by default. This can be overridden with the @code{section} attribute.
7600 @xref{Variable Attributes}.
7605 @subsection NS32K Options
7606 @cindex NS32K options
7608 These are the @samp{-m} options defined for the 32000 series. The default
7609 values for these options depends on which style of 32000 was selected when
7610 the compiler was configured; the defaults for the most common choices are
7616 Generate output for a 32032. This is the default
7617 when the compiler is configured for 32032 and 32016 based systems.
7621 Generate output for a 32332. This is the default
7622 when the compiler is configured for 32332-based systems.
7626 Generate output for a 32532. This is the default
7627 when the compiler is configured for 32532-based systems.
7630 Generate output containing 32081 instructions for floating point.
7631 This is the default for all systems.
7634 Generate output containing 32381 instructions for floating point. This
7635 also implies @samp{-m32081}. The 32381 is only compatible with the 32332
7636 and 32532 cpus. This is the default for the pc532-netbsd configuration.
7639 Try and generate multiply-add floating point instructions @code{polyF}
7640 and @code{dotF}. This option is only available if the @samp{-m32381}
7641 option is in effect. Using these instructions requires changes to to
7642 register allocation which generally has a negative impact on
7643 performance. This option should only be enabled when compiling code
7644 particularly likely to make heavy use of multiply-add instructions.
7647 Do not try and generate multiply-add floating point instructions
7648 @code{polyF} and @code{dotF}. This is the default on all platforms.
7651 Generate output containing library calls for floating point.
7652 @strong{Warning:} the requisite libraries may not be available.
7655 Do not use the bit-field instructions. On some machines it is faster to
7656 use shifting and masking operations. This is the default for the pc532.
7659 Do use the bit-field instructions. This is the default for all platforms
7663 Use a different function-calling convention, in which functions
7664 that take a fixed number of arguments return pop their
7665 arguments on return with the @code{ret} instruction.
7667 This calling convention is incompatible with the one normally
7668 used on Unix, so you cannot use it if you need to call libraries
7669 compiled with the Unix compiler.
7671 Also, you must provide function prototypes for all functions that
7672 take variable numbers of arguments (including @code{printf});
7673 otherwise incorrect code will be generated for calls to those
7676 In addition, seriously incorrect code will result if you call a
7677 function with too many arguments. (Normally, extra arguments are
7678 harmlessly ignored.)
7680 This option takes its name from the 680x0 @code{rtd} instruction.
7684 Use a different function-calling convention where the first two arguments
7685 are passed in registers.
7687 This calling convention is incompatible with the one normally
7688 used on Unix, so you cannot use it if you need to call libraries
7689 compiled with the Unix compiler.
7692 Do not pass any arguments in registers. This is the default for all
7696 It is OK to use the sb as an index register which is always loaded with
7697 zero. This is the default for the pc532-netbsd target.
7700 The sb register is not available for use or has not been initialized to
7701 zero by the run time system. This is the default for all targets except
7702 the pc532-netbsd. It is also implied whenever @samp{-mhimem} or
7703 @samp{-fpic} is set.
7706 Many ns32000 series addressing modes use displacements of up to 512MB.
7707 If an address is above 512MB then displacements from zero can not be used.
7708 This option causes code to be generated which can be loaded above 512MB.
7709 This may be useful for operating systems or ROM code.
7712 Assume code will be loaded in the first 512MB of virtual address space.
7713 This is the default for all platforms.
7719 @subsection AVR Options
7722 These options are defined for AVR implementations:
7725 @item -mmcu=@var{mcu}
7726 Specify ATMEL AVR instruction set or MCU type.
7728 Instruction set avr1 is for the minimal AVR core, not supported by the C
7729 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
7730 attiny11, attiny12, attiny15, attiny28).
7732 Instruction set avr2 (default) is for the classic AVR core with up to
7733 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
7734 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
7735 at90c8534, at90s8535).
7737 Instruction set avr3 is for the classic AVR core with up to 128K program
7738 memory space (MCU types: atmega103, atmega603).
7740 Instruction set avr4 is for the enhanced AVR core with up to 8K program
7741 memory space (MCU types: atmega83, atmega85).
7743 Instruction set avr5 is for the enhanced AVR core with up to 128K program
7744 memory space (MCU types: atmega161, atmega163, atmega32, at94k).
7747 Output instruction sizes to the asm file.
7749 @item -minit-stack=@var{N}
7750 Specify the initial stack address, which may be a symbol or numeric value,
7751 __stack is the default.
7753 @item -mno-interrupts
7754 Generated code is not compatible with hardware interrupts.
7755 Code size will be smaller.
7757 @item -mcall-prologues
7758 Functions prologues/epilogues expanded as call to appropriate
7759 subroutines. Code size will be smaller.
7761 @item -mno-tablejump
7762 Do not generate tablejump insns which sometimes increase code size.
7765 Change only the low 8 bits of the stack pointer.
7769 @subsection MCore Options
7770 @cindex MCore options
7772 These are the @samp{-m} options defined for the Motorola M*Core
7780 Inline constants into the code stream if it can be done in two
7781 instructions or less.
7786 Use the divide instruction. (Enabled by default).
7788 @item -mrelax-immediate
7789 @itemx -mrelax-immediate
7790 @itemx -mno-relax-immediate
7791 Allow arbitrary sized immediates in bit operations.
7793 @item -mwide-bitfields
7794 @itemx -mwide-bitfields
7795 @itemx -mno-wide-bitfields
7796 Always treat bitfields as int-sized.
7798 @item -m4byte-functions
7799 @itemx -m4byte-functions
7800 @itemx -mno-4byte-functions
7801 Force all functions to be aligned to a four byte boundary.
7803 @item -mcallgraph-data
7804 @itemx -mcallgraph-data
7805 @itemx -mno-callgraph-data
7806 Emit callgraph information.
7810 @itemx -mno-slow-bytes
7811 Prefer word access when reading byte quantities.
7813 @item -mlittle-endian
7814 @itemx -mlittle-endian
7816 Generate code for a little endian target.
7821 Generate code for the 210 processor.
7825 @subsection IA-64 Options
7826 @cindex IA-64 Options
7828 These are the @samp{-m} options defined for the Intel IA-64 architecture.
7832 Generate code for a big endian target. This is the default for HPUX.
7834 @item -mlittle-endian
7835 Generate code for a little endian target. This is the default for AIX5
7840 Generate (or don't) code for the GNU assembler. This is the default.
7841 @c Also, this is the default if the configure option @samp{--with-gnu-as}
7846 Generate (or don't) code for the GNU linker. This is the default.
7847 @c Also, this is the default if the configure option @samp{--with-gnu-ld}
7851 Generate code that does not use a global pointer register. The result
7852 is not position independent code, and violates the IA-64 ABI.
7854 @item -mvolatile-asm-stop
7855 @itemx -mno-volatile-asm-stop
7856 Generate (or don't) a stop bit immediately before and after volatile asm
7860 Generate code that works around Itanium B step errata.
7862 @item -mregister-names
7863 @itemx -mno-register-names
7864 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
7865 the stacked registers. This may make assembler output more readable.
7869 Disable (or enable) optimizations that use the small data section. This may
7870 be useful for working around optimizer bugs.
7873 Generate code that uses a single constant global pointer value. This is
7874 useful when compiling kernel code.
7877 Generate code that is self-relocatable. This implies @samp{-mconstant-gp}.
7878 This is useful when compiling firmware code.
7880 @item -minline-divide-min-latency
7881 Generate code for inline divides using the minimum latency algorithm.
7883 @item -minline-divide-max-throughput
7884 Generate code for inline divides using the maximum throughput algorithm.
7886 @item -mno-dwarf2-asm
7888 Don't (or do) generate assembler code for the DWARF2 line number debugging
7889 info. This may be useful when not using the GNU assembler.
7891 @item -mfixed-range=@var{register range}
7892 Generate code treating the given register range as fixed registers.
7893 A fixed register is one that the register allocator can not use. This is
7894 useful when compiling kernel code. A register range is specified as
7895 two registers separated by a dash. Multiple register ranges can be
7896 specified separated by a comma.
7900 @subsection D30V Options
7901 @cindex D30V Options
7903 These @samp{-m} options are defined for D30V implementations:
7907 Link the @samp{.text}, @samp{.data}, @samp{.bss}, @samp{.strings},
7908 @samp{.rodata}, @samp{.rodata1}, @samp{.data1} sections into external
7909 memory, which starts at location @code{0x80000000}.
7912 Same as the @samp{-mextmem} switch.
7915 Link the @samp{.text} section into onchip text memory, which starts at
7916 location @code{0x0}. Also link @samp{.data}, @samp{.bss},
7917 @samp{.strings}, @samp{.rodata}, @samp{.rodata1}, @samp{.data1} sections
7918 into onchip data memory, which starts at location @code{0x20000000}.
7920 @item -mno-asm-optimize
7921 @itemx -masm-optimize
7922 Disable (enable) passing @samp{-O} to the assembler when optimizing.
7923 The assembler uses the @samp{-O} option to automatically parallelize
7924 adjacent short instructions where possible.
7926 @item -mbranch-cost=@var{n}
7927 Increase the internal costs of branches to @var{n}. Higher costs means
7928 that the compiler will issue more instructions to avoid doing a branch.
7931 @item -mcond-exec=@var{n}
7932 Specify the maximum number of conditionally executed instructions that
7933 replace a branch. The default is 4.
7936 @node Code Gen Options
7937 @section Options for Code Generation Conventions
7938 @cindex code generation conventions
7939 @cindex options, code generation
7940 @cindex run-time options
7942 These machine-independent options control the interface conventions
7943 used in code generation.
7945 Most of them have both positive and negative forms; the negative form
7946 of @samp{-ffoo} would be @samp{-fno-foo}. In the table below, only
7947 one of the forms is listed---the one which is not the default. You
7948 can figure out the other form by either removing @samp{no-} or adding
7953 Enable exception handling. Generates extra code needed to propagate
7954 exceptions. For some targets, this implies GNU CC will generate frame
7955 unwind information for all functions, which can produce significant data
7956 size overhead, although it does not affect execution. If you do not
7957 specify this option, GNU CC will enable it by default for languages like
7958 C++ which normally require exception handling, and disable itfor
7959 languages like C that do not normally require it. However, you may need
7960 to enable this option when compiling C code that needs to interoperate
7961 properly with exception handlers written in C++. You may also wish to
7962 disable this option if you are compiling older C++ programs that don't
7963 use exception handling.
7965 @item -funwind-tables
7966 Similar to @option{-fexceptions}, except that it will just generate any needed
7967 static data, but will not affect the generated code in any other way.
7968 You will normally not enable this option; instead, a language processor
7969 that needs this handling would enable it on your behalf.
7971 @item -fpcc-struct-return
7972 Return ``short'' @code{struct} and @code{union} values in memory like
7973 longer ones, rather than in registers. This convention is less
7974 efficient, but it has the advantage of allowing intercallability between
7975 GCC-compiled files and files compiled with other compilers.
7977 The precise convention for returning structures in memory depends
7978 on the target configuration macros.
7980 Short structures and unions are those whose size and alignment match
7981 that of some integer type.
7983 @item -freg-struct-return
7984 Use the convention that @code{struct} and @code{union} values are
7985 returned in registers when possible. This is more efficient for small
7986 structures than @samp{-fpcc-struct-return}.
7988 If you specify neither @samp{-fpcc-struct-return} nor its contrary
7989 @samp{-freg-struct-return}, GCC defaults to whichever convention is
7990 standard for the target. If there is no standard convention, GCC
7991 defaults to @samp{-fpcc-struct-return}, except on targets where GCC
7992 is the principal compiler. In those cases, we can choose the standard,
7993 and we chose the more efficient register return alternative.
7996 Allocate to an @code{enum} type only as many bytes as it needs for the
7997 declared range of possible values. Specifically, the @code{enum} type
7998 will be equivalent to the smallest integer type which has enough room.
8000 @item -fshort-double
8001 Use the same size for @code{double} as for @code{float}.
8004 Requests that the data and non-@code{const} variables of this
8005 compilation be shared data rather than private data. The distinction
8006 makes sense only on certain operating systems, where shared data is
8007 shared between processes running the same program, while private data
8008 exists in one copy per process.
8011 Allocate even uninitialized global variables in the data section of the
8012 object file, rather than generating them as common blocks. This has the
8013 effect that if the same variable is declared (without @code{extern}) in
8014 two different compilations, you will get an error when you link them.
8015 The only reason this might be useful is if you wish to verify that the
8016 program will work on other systems which always work this way.
8019 Ignore the @samp{#ident} directive.
8021 @item -fno-gnu-linker
8022 Do not output global initializations (such as C++ constructors and
8023 destructors) in the form used by the GNU linker (on systems where the GNU
8024 linker is the standard method of handling them). Use this option when
8025 you want to use a non-GNU linker, which also requires using the
8026 @command{collect2} program to make sure the system linker includes
8027 constructors and destructors. (@command{collect2} is included in the GCC
8028 distribution.) For systems which @emph{must} use @command{collect2}, the
8029 compiler driver @command{gcc} is configured to do this automatically.
8031 @item -finhibit-size-directive
8032 Don't output a @code{.size} assembler directive, or anything else that
8033 would cause trouble if the function is split in the middle, and the
8034 two halves are placed at locations far apart in memory. This option is
8035 used when compiling @file{crtstuff.c}; you should not need to use it
8039 Put extra commentary information in the generated assembly code to
8040 make it more readable. This option is generally only of use to those
8041 who actually need to read the generated assembly code (perhaps while
8042 debugging the compiler itself).
8044 @samp{-fno-verbose-asm}, the default, causes the
8045 extra information to be omitted and is useful when comparing two assembler
8049 Consider all memory references through pointers to be volatile.
8051 @item -fvolatile-global
8052 Consider all memory references to extern and global data items to
8053 be volatile. GCC does not consider static data items to be volatile
8054 because of this switch.
8056 @item -fvolatile-static
8057 Consider all memory references to static data to be volatile.
8060 @cindex global offset table
8062 Generate position-independent code (PIC) suitable for use in a shared
8063 library, if supported for the target machine. Such code accesses all
8064 constant addresses through a global offset table (GOT). The dynamic
8065 loader resolves the GOT entries when the program starts (the dynamic
8066 loader is not part of GCC; it is part of the operating system). If
8067 the GOT size for the linked executable exceeds a machine-specific
8068 maximum size, you get an error message from the linker indicating that
8069 @samp{-fpic} does not work; in that case, recompile with @samp{-fPIC}
8070 instead. (These maximums are 16k on the m88k, 8k on the Sparc, and 32k
8071 on the m68k and RS/6000. The 386 has no such limit.)
8073 Position-independent code requires special support, and therefore works
8074 only on certain machines. For the 386, GCC supports PIC for System V
8075 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
8076 position-independent.
8079 If supported for the target machine, emit position-independent code,
8080 suitable for dynamic linking and avoiding any limit on the size of the
8081 global offset table. This option makes a difference on the m68k, m88k,
8084 Position-independent code requires special support, and therefore works
8085 only on certain machines.
8087 @item -ffixed-@var{reg}
8088 Treat the register named @var{reg} as a fixed register; generated code
8089 should never refer to it (except perhaps as a stack pointer, frame
8090 pointer or in some other fixed role).
8092 @var{reg} must be the name of a register. The register names accepted
8093 are machine-specific and are defined in the @code{REGISTER_NAMES}
8094 macro in the machine description macro file.
8096 This flag does not have a negative form, because it specifies a
8099 @item -fcall-used-@var{reg}
8100 Treat the register named @var{reg} as an allocable register that is
8101 clobbered by function calls. It may be allocated for temporaries or
8102 variables that do not live across a call. Functions compiled this way
8103 will not save and restore the register @var{reg}.
8105 It is an error to used this flag with the frame pointer or stack pointer.
8106 Use of this flag for other registers that have fixed pervasive roles in
8107 the machine's execution model will produce disastrous results.
8109 This flag does not have a negative form, because it specifies a
8112 @item -fcall-saved-@var{reg}
8113 Treat the register named @var{reg} as an allocable register saved by
8114 functions. It may be allocated even for temporaries or variables that
8115 live across a call. Functions compiled this way will save and restore
8116 the register @var{reg} if they use it.
8118 It is an error to used this flag with the frame pointer or stack pointer.
8119 Use of this flag for other registers that have fixed pervasive roles in
8120 the machine's execution model will produce disastrous results.
8122 A different sort of disaster will result from the use of this flag for
8123 a register in which function values may be returned.
8125 This flag does not have a negative form, because it specifies a
8129 Pack all structure members together without holes. Usually you would
8130 not want to use this option, since it makes the code suboptimal, and
8131 the offsets of structure members won't agree with system libraries.
8133 @item -fcheck-memory-usage
8134 Generate extra code to check each memory access. GCC will generate
8135 code that is suitable for a detector of bad memory accesses such as
8138 Normally, you should compile all, or none, of your code with this option.
8140 If you do mix code compiled with and without this option,
8141 you must ensure that all code that has side effects
8142 and that is called by code compiled with this option
8143 is, itself, compiled with this option.
8144 If you do not, you might get erroneous messages from the detector.
8146 If you use functions from a library that have side-effects (such as
8147 @code{read}), you might not be able to recompile the library and
8148 specify this option. In that case, you can enable the
8149 @samp{-fprefix-function-name} option, which requests GCC to encapsulate
8150 your code and make other functions look as if they were compiled with
8151 @samp{-fcheck-memory-usage}. This is done by calling ``stubs'',
8152 which are provided by the detector. If you cannot find or build
8153 stubs for every function you call, you might have to specify
8154 @samp{-fcheck-memory-usage} without @samp{-fprefix-function-name}.
8156 If you specify this option, you can not use the @code{asm} or
8157 @code{__asm__} keywords in functions with memory checking enabled. GNU
8158 CC cannot understand what the @code{asm} statement may do, and therefore
8159 cannot generate the appropriate code, so it will reject it. However, if
8160 you specify the function attribute @code{no_check_memory_usage}
8161 (@pxref{Function Attributes}), GNU CC will disable memory checking within a
8162 function; you may use @code{asm} statements inside such functions. You
8163 may have an inline expansion of a non-checked function within a checked
8164 function; in that case GNU CC will not generate checks for the inlined
8165 function's memory accesses.
8167 If you move your @code{asm} statements to non-checked inline functions
8168 and they do access memory, you can add calls to the support code in your
8169 inline function, to indicate any reads, writes, or copies being done.
8170 These calls would be similar to those done in the stubs described above.
8172 @item -fprefix-function-name
8173 Request GCC to add a prefix to the symbols generated for function names.
8174 GCC adds a prefix to the names of functions defined as well as
8175 functions called. Code compiled with this option and code compiled
8176 without the option can't be linked together, unless stubs are used.
8178 If you compile the following code with @samp{-fprefix-function-name}
8180 extern void bar (int);
8189 GCC will compile the code as if it was written:
8191 extern void prefix_bar (int);
8195 return prefix_bar (a + 5);
8198 This option is designed to be used with @samp{-fcheck-memory-usage}.
8200 @item -finstrument-functions
8201 Generate instrumentation calls for entry and exit to functions. Just
8202 after function entry and just before function exit, the following
8203 profiling functions will be called with the address of the current
8204 function and its call site. (On some platforms,
8205 @code{__builtin_return_address} does not work beyond the current
8206 function, so the call site information may not be available to the
8207 profiling functions otherwise.)
8210 void __cyg_profile_func_enter (void *this_fn, void *call_site);
8211 void __cyg_profile_func_exit (void *this_fn, void *call_site);
8214 The first argument is the address of the start of the current function,
8215 which may be looked up exactly in the symbol table.
8217 This instrumentation is also done for functions expanded inline in other
8218 functions. The profiling calls will indicate where, conceptually, the
8219 inline function is entered and exited. This means that addressable
8220 versions of such functions must be available. If all your uses of a
8221 function are expanded inline, this may mean an additional expansion of
8222 code size. If you use @samp{extern inline} in your C code, an
8223 addressable version of such functions must be provided. (This is
8224 normally the case anyways, but if you get lucky and the optimizer always
8225 expands the functions inline, you might have gotten away without
8226 providing static copies.)
8228 A function may be given the attribute @code{no_instrument_function}, in
8229 which case this instrumentation will not be done. This can be used, for
8230 example, for the profiling functions listed above, high-priority
8231 interrupt routines, and any functions from which the profiling functions
8232 cannot safely be called (perhaps signal handlers, if the profiling
8233 routines generate output or allocate memory).
8236 Generate code to verify that you do not go beyond the boundary of the
8237 stack. You should specify this flag if you are running in an
8238 environment with multiple threads, but only rarely need to specify it in
8239 a single-threaded environment since stack overflow is automatically
8240 detected on nearly all systems if there is only one stack.
8242 Note that this switch does not actually cause checking to be done; the
8243 operating system must do that. The switch causes generation of code
8244 to ensure that the operating system sees the stack being extended.
8246 @item -fstack-limit-register=@var{reg}
8247 @itemx -fstack-limit-symbol=@var{sym}
8248 @itemx -fno-stack-limit
8249 Generate code to ensure that the stack does not grow beyond a certain value,
8250 either the value of a register or the address of a symbol. If the stack
8251 would grow beyond the value, a signal is raised. For most targets,
8252 the signal is raised before the stack overruns the boundary, so
8253 it is possible to catch the signal without taking special precautions.
8255 For instance, if the stack starts at address @samp{0x80000000} and grows
8256 downwards you can use the flags
8257 @samp{-fstack-limit-symbol=__stack_limit}
8258 @samp{-Wl,--defsym,__stack_limit=0x7ffe0000} which will enforce a stack
8261 @cindex aliasing of parameters
8262 @cindex parameters, aliased
8263 @item -fargument-alias
8264 @itemx -fargument-noalias
8265 @itemx -fargument-noalias-global
8266 Specify the possible relationships among parameters and between
8267 parameters and global data.
8269 @samp{-fargument-alias} specifies that arguments (parameters) may
8270 alias each other and may alias global storage.
8271 @samp{-fargument-noalias} specifies that arguments do not alias
8272 each other, but may alias global storage.
8273 @samp{-fargument-noalias-global} specifies that arguments do not
8274 alias each other and do not alias global storage.
8276 Each language will automatically use whatever option is required by
8277 the language standard. You should not need to use these options yourself.
8279 @item -fleading-underscore
8280 This option and its counterpart, -fno-leading-underscore, forcibly
8281 change the way C symbols are represented in the object file. One use
8282 is to help link with legacy assembly code.
8284 Be warned that you should know what you are doing when invoking this
8285 option, and that not all targets provide complete support for it.
8290 @node Environment Variables
8291 @section Environment Variables Affecting GCC
8292 @cindex environment variables
8294 @c man begin ENVIRONMENT
8296 This section describes several environment variables that affect how GCC
8297 operates. Some of them work by specifying directories or prefixes to use
8298 when searching for various kinds of files. Some are used to specify other
8299 aspects of the compilation environment.
8302 Note that you can also specify places to search using options such as
8303 @samp{-B}, @samp{-I} and @samp{-L} (@pxref{Directory Options}). These
8304 take precedence over places specified using environment variables, which
8305 in turn take precedence over those specified by the configuration of GCC.
8309 Note that you can also specify places to search using options such as
8310 @samp{-B}, @samp{-I} and @samp{-L} (@pxref{Directory Options}). These
8311 take precedence over places specified using environment variables, which
8312 in turn take precedence over those specified by the configuration of GCC.
8319 @c @itemx LC_COLLATE
8321 @c @itemx LC_MONETARY
8322 @c @itemx LC_NUMERIC
8327 @c @findex LC_COLLATE
8329 @c @findex LC_MONETARY
8330 @c @findex LC_NUMERIC
8334 These environment variables control the way that GCC uses
8335 localization information that allow GCC to work with different
8336 national conventions. GCC inspects the locale categories
8337 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
8338 so. These locale categories can be set to any value supported by your
8339 installation. A typical value is @samp{en_UK} for English in the United
8342 The @env{LC_CTYPE} environment variable specifies character
8343 classification. GCC uses it to determine the character boundaries in
8344 a string; this is needed for some multibyte encodings that contain quote
8345 and escape characters that would otherwise be interpreted as a string
8348 The @env{LC_MESSAGES} environment variable specifies the language to
8349 use in diagnostic messages.
8351 If the @env{LC_ALL} environment variable is set, it overrides the value
8352 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
8353 and @env{LC_MESSAGES} default to the value of the @env{LANG}
8354 environment variable. If none of these variables are set, GCC
8355 defaults to traditional C English behavior.
8359 If @env{TMPDIR} is set, it specifies the directory to use for temporary
8360 files. GCC uses temporary files to hold the output of one stage of
8361 compilation which is to be used as input to the next stage: for example,
8362 the output of the preprocessor, which is the input to the compiler
8365 @item GCC_EXEC_PREFIX
8366 @findex GCC_EXEC_PREFIX
8367 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
8368 names of the subprograms executed by the compiler. No slash is added
8369 when this prefix is combined with the name of a subprogram, but you can
8370 specify a prefix that ends with a slash if you wish.
8372 If @env{GCC_EXEC_PREFIX} is not set, GNU CC will attempt to figure out
8373 an appropriate prefix to use based on the pathname it was invoked with.
8375 If GCC cannot find the subprogram using the specified prefix, it
8376 tries looking in the usual places for the subprogram.
8378 The default value of @env{GCC_EXEC_PREFIX} is
8379 @file{@var{prefix}/lib/gcc-lib/} where @var{prefix} is the value
8380 of @code{prefix} when you ran the @file{configure} script.
8382 Other prefixes specified with @samp{-B} take precedence over this prefix.
8384 This prefix is also used for finding files such as @file{crt0.o} that are
8387 In addition, the prefix is used in an unusual way in finding the
8388 directories to search for header files. For each of the standard
8389 directories whose name normally begins with @samp{/usr/local/lib/gcc-lib}
8390 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
8391 replacing that beginning with the specified prefix to produce an
8392 alternate directory name. Thus, with @samp{-Bfoo/}, GCC will search
8393 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
8394 These alternate directories are searched first; the standard directories
8398 @findex COMPILER_PATH
8399 The value of @env{COMPILER_PATH} is a colon-separated list of
8400 directories, much like @env{PATH}. GCC tries the directories thus
8401 specified when searching for subprograms, if it can't find the
8402 subprograms using @env{GCC_EXEC_PREFIX}.
8405 @findex LIBRARY_PATH
8406 The value of @env{LIBRARY_PATH} is a colon-separated list of
8407 directories, much like @env{PATH}. When configured as a native compiler,
8408 GCC tries the directories thus specified when searching for special
8409 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
8410 using GCC also uses these directories when searching for ordinary
8411 libraries for the @samp{-l} option (but directories specified with
8412 @samp{-L} come first).
8414 @item C_INCLUDE_PATH
8415 @itemx CPLUS_INCLUDE_PATH
8416 @itemx OBJC_INCLUDE_PATH
8417 @findex C_INCLUDE_PATH
8418 @findex CPLUS_INCLUDE_PATH
8419 @findex OBJC_INCLUDE_PATH
8420 @c @itemx OBJCPLUS_INCLUDE_PATH
8421 These environment variables pertain to particular languages. Each
8422 variable's value is a colon-separated list of directories, much like
8423 @env{PATH}. When GCC searches for header files, it tries the
8424 directories listed in the variable for the language you are using, after
8425 the directories specified with @samp{-I} but before the standard header
8428 @item DEPENDENCIES_OUTPUT
8429 @findex DEPENDENCIES_OUTPUT
8430 @cindex dependencies for make as output
8431 If this variable is set, its value specifies how to output dependencies
8432 for Make based on the header files processed by the compiler. This
8433 output looks much like the output from the @samp{-M} option
8434 (@pxref{Preprocessor Options}), but it goes to a separate file, and is
8435 in addition to the usual results of compilation.
8437 The value of @env{DEPENDENCIES_OUTPUT} can be just a file name, in
8438 which case the Make rules are written to that file, guessing the target
8439 name from the source file name. Or the value can have the form
8440 @samp{@var{file} @var{target}}, in which case the rules are written to
8441 file @var{file} using @var{target} as the target name.
8445 @cindex locale definition
8446 This variable is used to pass locale information to the compiler. One way in
8447 which this information is used is to determine the character set to be used
8448 when character literals, string literals and comments are parsed in C and C++.
8449 When the compiler is configured to allow multibyte characters,
8450 the following values for @env{LANG} are recognized:
8454 Recognize JIS characters.
8456 Recognize SJIS characters.
8458 Recognize EUCJP characters.
8461 If @env{LANG} is not defined, or if it has some other value, then the
8462 compiler will use mblen and mbtowc as defined by the default locale to
8463 recognize and translate multibyte characters.
8468 @node Running Protoize
8469 @section Running Protoize
8471 The program @code{protoize} is an optional part of GNU C. You can use
8472 it to add prototypes to a program, thus converting the program to ISO
8473 C in one respect. The companion program @code{unprotoize} does the
8474 reverse: it removes argument types from any prototypes that are found.
8476 When you run these programs, you must specify a set of source files as
8477 command line arguments. The conversion programs start out by compiling
8478 these files to see what functions they define. The information gathered
8479 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
8481 After scanning comes actual conversion. The specified files are all
8482 eligible to be converted; any files they include (whether sources or
8483 just headers) are eligible as well.
8485 But not all the eligible files are converted. By default,
8486 @code{protoize} and @code{unprotoize} convert only source and header
8487 files in the current directory. You can specify additional directories
8488 whose files should be converted with the @samp{-d @var{directory}}
8489 option. You can also specify particular files to exclude with the
8490 @samp{-x @var{file}} option. A file is converted if it is eligible, its
8491 directory name matches one of the specified directory names, and its
8492 name within the directory has not been excluded.
8494 Basic conversion with @code{protoize} consists of rewriting most
8495 function definitions and function declarations to specify the types of
8496 the arguments. The only ones not rewritten are those for varargs
8499 @code{protoize} optionally inserts prototype declarations at the
8500 beginning of the source file, to make them available for any calls that
8501 precede the function's definition. Or it can insert prototype
8502 declarations with block scope in the blocks where undeclared functions
8505 Basic conversion with @code{unprotoize} consists of rewriting most
8506 function declarations to remove any argument types, and rewriting
8507 function definitions to the old-style pre-ISO form.
8509 Both conversion programs print a warning for any function declaration or
8510 definition that they can't convert. You can suppress these warnings
8513 The output from @code{protoize} or @code{unprotoize} replaces the
8514 original source file. The original file is renamed to a name ending
8515 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
8516 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
8517 for DOS) file already exists, then the source file is simply discarded.
8519 @code{protoize} and @code{unprotoize} both depend on GCC itself to
8520 scan the program and collect information about the functions it uses.
8521 So neither of these programs will work until GCC is installed.
8523 Here is a table of the options you can use with @code{protoize} and
8524 @code{unprotoize}. Each option works with both programs unless
8528 @item -B @var{directory}
8529 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
8530 usual directory (normally @file{/usr/local/lib}). This file contains
8531 prototype information about standard system functions. This option
8532 applies only to @code{protoize}.
8534 @item -c @var{compilation-options}
8535 Use @var{compilation-options} as the options when running @code{gcc} to
8536 produce the @samp{.X} files. The special option @samp{-aux-info} is
8537 always passed in addition, to tell @code{gcc} to write a @samp{.X} file.
8539 Note that the compilation options must be given as a single argument to
8540 @code{protoize} or @code{unprotoize}. If you want to specify several
8541 @code{gcc} options, you must quote the entire set of compilation options
8542 to make them a single word in the shell.
8544 There are certain @code{gcc} arguments that you cannot use, because they
8545 would produce the wrong kind of output. These include @samp{-g},
8546 @samp{-O}, @samp{-c}, @samp{-S}, and @samp{-o} If you include these in
8547 the @var{compilation-options}, they are ignored.
8550 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
8551 systems) instead of @samp{.c}. This is convenient if you are converting
8552 a C program to C++. This option applies only to @code{protoize}.
8555 Add explicit global declarations. This means inserting explicit
8556 declarations at the beginning of each source file for each function
8557 that is called in the file and was not declared. These declarations
8558 precede the first function definition that contains a call to an
8559 undeclared function. This option applies only to @code{protoize}.
8561 @item -i @var{string}
8562 Indent old-style parameter declarations with the string @var{string}.
8563 This option applies only to @code{protoize}.
8565 @code{unprotoize} converts prototyped function definitions to old-style
8566 function definitions, where the arguments are declared between the
8567 argument list and the initial @samp{@{}. By default, @code{unprotoize}
8568 uses five spaces as the indentation. If you want to indent with just
8569 one space instead, use @samp{-i " "}.
8572 Keep the @samp{.X} files. Normally, they are deleted after conversion
8576 Add explicit local declarations. @code{protoize} with @samp{-l} inserts
8577 a prototype declaration for each function in each block which calls the
8578 function without any declaration. This option applies only to
8582 Make no real changes. This mode just prints information about the conversions
8583 that would have been done without @samp{-n}.
8586 Make no @samp{.save} files. The original files are simply deleted.
8587 Use this option with caution.
8589 @item -p @var{program}
8590 Use the program @var{program} as the compiler. Normally, the name
8594 Work quietly. Most warnings are suppressed.
8597 Print the version number, just like @samp{-v} for @code{gcc}.
8600 If you need special compiler options to compile one of your program's
8601 source files, then you should generate that file's @samp{.X} file
8602 specially, by running @code{gcc} on that source file with the
8603 appropriate options and the option @samp{-aux-info}. Then run
8604 @code{protoize} on the entire set of files. @code{protoize} will use
8605 the existing @samp{.X} file because it is newer than the source file.
8609 gcc -Dfoo=bar file1.c -aux-info
8614 You need to include the special files along with the rest in the
8615 @code{protoize} command, even though their @samp{.X} files already
8616 exist, because otherwise they won't get converted.
8618 @xref{Protoize Caveats}, for more information on how to use
8619 @code{protoize} successfully.