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-braces -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} -dumpspecs -dumpmachine -dumpversion @gol
228 -fdump-unnumbered -fdump-translation-unit=@var{file} @gol
229 -fdump-class-layout=@var{file} -fmem-report -fpretend-float @gol
230 -fprofile-arcs -ftest-coverage -ftime-report @gol
231 -g -g@var{level} -gcoff -gdwarf -gdwarf-1 -gdwarf-1+ -gdwarf-2 @gol
232 -ggdb -gstabs -gstabs+ -gxcoff -gxcoff+ @gol
233 -p -pg -print-file-name=@var{library} -print-libgcc-file-name @gol
234 -print-prog-name=@var{program} -print-search-dirs -Q @gol
237 @item Optimization Options
238 @xref{Optimize Options,,Options that Control Optimization}.
240 -falign-functions=@var{n} -falign-jumps=@var{n} @gol
241 -falign-labels=@var{n} -falign-loops=@var{n} @gol
242 -fbranch-probabilities -fcaller-saves @gol
243 -fcse-follow-jumps -fcse-skip-blocks -fdata-sections -fdce @gol
244 -fdelayed-branch -fdelete-null-pointer-checks @gol
245 -fexpensive-optimizations -ffast-math -ffloat-store @gol
246 -fforce-addr -fforce-mem -ffunction-sections -fgcse -fgcse-lm -fgcse-sm @gol
247 -finline-functions -finline-limit=@var{n} -fkeep-inline-functions @gol
248 -fkeep-static-consts -fmove-all-movables @gol
249 -fno-default-inline -fno-defer-pop @gol
250 -fno-function-cse -fno-guess-branch-probability @gol
251 -fno-inline -fno-math-errno -fno-peephole @gol
252 -funsafe-math-optimizations -fno-trapping-math @gol
253 -fomit-frame-pointer -foptimize-register-move @gol
254 -foptimize-sibling-calls -freduce-all-givs @gol
255 -fregmove -frename-registers @gol
256 -frerun-cse-after-loop -frerun-loop-opt @gol
257 -fschedule-insns -fschedule-insns2 @gol
258 -fsingle-precision-constant -fssa @gol
259 -fstrength-reduce -fstrict-aliasing -fthread-jumps -ftrapv @gol
260 -funroll-all-loops -funroll-loops @gol
261 --param @var{name}=@var{value}
262 -O -O0 -O1 -O2 -O3 -Os}
264 @item Preprocessor Options
265 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
267 -$ -A@var{question}=@var{answer} -A-@var{question}@r{[}=@var{answer}@r{]} @gol
268 -C -dD -dI -dM -dN @gol
269 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
270 -idirafter @var{dir} @gol
271 -include @var{file} -imacros @var{file} @gol
272 -iprefix @var{file} -iwithprefix @var{dir} @gol
273 -iwithprefixbefore @var{dir} -isystem @var{dir} -isystem-c++ @var{dir} @gol
274 -M -MM -MF -MG -MP -MQ -MT -nostdinc -P -remap @gol
275 -trigraphs -undef -U@var{macro} -Wp\,@var{option}}
277 @item Assembler Option
278 @xref{Assembler Options,,Passing Options to the Assembler}.
283 @xref{Link Options,,Options for Linking}.
285 @var{object-file-name} -l@var{library} @gol
286 -nostartfiles -nodefaultlibs -nostdlib @gol
287 -s -static -static-libgcc -shared -shared-libgcc -symbolic @gol
288 -Wl\,@var{option} -Xlinker @var{option} @gol
291 @item Directory Options
292 @xref{Directory Options,,Options for Directory Search}.
294 -B@var{prefix} -I@var{dir} -I- -L@var{dir} -specs=@var{file}}
297 @c I wrote this xref this way to avoid overfull hbox. -- rms
298 @xref{Target Options}.
300 -b @var{machine} -V @var{version}}
302 @item Machine Dependent Options
303 @xref{Submodel Options,,Hardware Models and Configurations}.
304 @emph{M680x0 Options}
306 -m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
307 -m68060 -mcpu32 -m5200 -m68881 -mbitfield -mc68000 -mc68020 @gol
308 -mfpa -mnobitfield -mrtd -mshort -msoft-float -mpcrel @gol
309 -malign-int -mstrict-align}
311 @emph{M68hc1x Options}
313 -m6811 -m6812 -m68hc11 -m68hc12 @gol
314 -mauto-incdec -mshort -msoft-reg-count=@var{count}}
322 -mcpu=@var{cpu type} @gol
323 -mtune=@var{cpu type} @gol
324 -mcmodel=@var{code model} @gol
326 -mapp-regs -mbroken-saverestore -mcypress @gol
327 -mepilogue -mfaster-structs -mflat @gol
328 -mfpu -mhard-float -mhard-quad-float @gol
329 -mimpure-text -mlive-g0 -mno-app-regs @gol
330 -mno-epilogue -mno-faster-structs -mno-flat -mno-fpu @gol
331 -mno-impure-text -mno-stack-bias -mno-unaligned-doubles @gol
332 -msoft-float -msoft-quad-float -msparclite -mstack-bias @gol
333 -msupersparc -munaligned-doubles -mv8}
335 @emph{Convex Options}
337 -mc1 -mc2 -mc32 -mc34 -mc38 @gol
338 -margcount -mnoargcount @gol
339 -mlong32 -mlong64 @gol
340 -mvolatile-cache -mvolatile-nocache}
342 @emph{AMD29K Options}
344 -m29000 -m29050 -mbw -mnbw -mdw -mndw @gol
345 -mlarge -mnormal -msmall @gol
346 -mkernel-registers -mno-reuse-arg-regs @gol
347 -mno-stack-check -mno-storem-bug @gol
348 -mreuse-arg-regs -msoft-float -mstack-check @gol
349 -mstorem-bug -muser-registers}
353 -mapcs-frame -mno-apcs-frame @gol
354 -mapcs-26 -mapcs-32 @gol
355 -mapcs-stack-check -mno-apcs-stack-check @gol
356 -mapcs-float -mno-apcs-float @gol
357 -mapcs-reentrant -mno-apcs-reentrant @gol
358 -msched-prolog -mno-sched-prolog @gol
359 -mlittle-endian -mbig-endian -mwords-little-endian @gol
360 -malignment-traps -mno-alignment-traps @gol
361 -msoft-float -mhard-float -mfpe @gol
362 -mthumb-interwork -mno-thumb-interwork @gol
363 -mcpu= -march= -mfpe= @gol
364 -mstructure-size-boundary= @gol
365 -mbsd -mxopen -mno-symrename @gol
366 -mabort-on-noreturn @gol
367 -mlong-calls -mno-long-calls @gol
368 -mnop-fun-dllimport -mno-nop-fun-dllimport @gol
369 -msingle-pic-base -mno-single-pic-base @gol
374 -mtpcs-frame -mno-tpcs-frame @gol
375 -mtpcs-leaf-frame -mno-tpcs-leaf-frame @gol
376 -mlittle-endian -mbig-endian @gol
377 -mthumb-interwork -mno-thumb-interwork @gol
378 -mstructure-size-boundary= @gol
379 -mnop-fun-dllimport -mno-nop-fun-dllimport @gol
380 -mcallee-super-interworking -mno-callee-super-interworking @gol
381 -mcaller-super-interworking -mno-caller-super-interworking @gol
382 -msingle-pic-base -mno-single-pic-base @gol
385 @emph{MN10200 Options}
389 @emph{MN10300 Options}
397 @emph{M32R/D Options}
399 -mcode-model=@var{model type} -msdata=@var{sdata type} @gol
404 -m88000 -m88100 -m88110 -mbig-pic @gol
405 -mcheck-zero-division -mhandle-large-shift @gol
406 -midentify-revision -mno-check-zero-division @gol
407 -mno-ocs-debug-info -mno-ocs-frame-position @gol
408 -mno-optimize-arg-area -mno-serialize-volatile @gol
409 -mno-underscores -mocs-debug-info @gol
410 -mocs-frame-position -moptimize-arg-area @gol
411 -mserialize-volatile -mshort-data-@var{num} -msvr3 @gol
412 -msvr4 -mtrap-large-shift -muse-div-instruction @gol
413 -mversion-03.00 -mwarn-passed-structs}
415 @emph{RS/6000 and PowerPC Options}
417 -mcpu=@var{cpu type} @gol
418 -mtune=@var{cpu type} @gol
419 -mpower -mno-power -mpower2 -mno-power2 @gol
420 -mpowerpc -mpowerpc64 -mno-powerpc @gol
421 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
422 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
423 -mnew-mnemonics -mold-mnemonics @gol
424 -mfull-toc -mminimal-toc -mno-fop-in-toc -mno-sum-in-toc @gol
425 -m64 -m32 -mxl-call -mno-xl-call -mthreads -mpe @gol
426 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
427 -mstring -mno-string -mupdate -mno-update @gol
428 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
429 -mstrict-align -mno-strict-align -mrelocatable @gol
430 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
431 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
432 -mcall-aix -mcall-sysv -mprototype -mno-prototype @gol
433 -msim -mmvme -mads -myellowknife -memb -msdata @gol
434 -msdata=@var{opt} -mvxworks -G @var{num}}
438 -mcall-lib-mul -mfp-arg-in-fpregs -mfp-arg-in-gregs @gol
439 -mfull-fp-blocks -mhc-struct-return -min-line-mul @gol
440 -mminimum-fp-blocks -mnohc-struct-return}
444 -mabicalls -mcpu=@var{cpu type}
445 -membedded-data -muninit-const-in-rodata @gol
446 -membedded-pic -mfp32 -mfp64 -mgas -mgp32 -mgp64 @gol
447 -mgpopt -mhalf-pic -mhard-float -mint64 -mips1 @gol
448 -mips2 -mips3 -mips4 -mlong64 -mlong32 -mlong-calls -mmemcpy @gol
449 -mmips-as -mmips-tfile -mno-abicalls @gol
450 -mno-embedded-data -mno-uninit-const-in-rodata -mno-embedded-pic @gol
451 -mno-gpopt -mno-long-calls @gol
452 -mno-memcpy -mno-mips-tfile -mno-rnames -mno-stats @gol
453 -mrnames -msoft-float @gol
454 -m4650 -msingle-float -mmad @gol
455 -mstats -EL -EB -G @var{num} -nocpp @gol
456 -mabi=32 -mabi=n32 -mabi=64 -mabi=eabi @gol
461 -mcpu=@var{cpu type} -march=@var{cpu type} @gol
462 -mintel-syntax -mieee-fp -mno-fancy-math-387 @gol
463 -mno-fp-ret-in-387 -msoft-float -msvr3-shlib @gol
464 -mno-wide-multiply -mrtd -malign-double @gol
465 -malign-jumps=@var{num} -malign-loops=@var{num} @gol
466 -malign-functions=@var{num} -mpreferred-stack-boundary=@var{num} @gol
467 -mthreads -mno-align-stringops -minline-all-stringops @gol
468 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
469 -m96bit-long-double -mregparm=@var{num}}
473 -march=@var{architecture type} @gol
474 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
475 -mfast-indirect-calls -mgas -mjump-in-delay @gol
476 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
477 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
478 -mno-jump-in-delay -mno-long-load-store @gol
479 -mno-portable-runtime -mno-soft-float @gol
480 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
481 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
482 -mschedule=@var{cpu type} -mspace-regs}
484 @emph{Intel 960 Options}
486 -m@var{cpu type} -masm-compat -mclean-linkage @gol
487 -mcode-align -mcomplex-addr -mleaf-procedures @gol
488 -mic-compat -mic2.0-compat -mic3.0-compat @gol
489 -mintel-asm -mno-clean-linkage -mno-code-align @gol
490 -mno-complex-addr -mno-leaf-procedures @gol
491 -mno-old-align -mno-strict-align -mno-tail-call @gol
492 -mnumerics -mold-align -msoft-float -mstrict-align @gol
495 @emph{DEC Alpha Options}
497 -mfp-regs -mno-fp-regs -mno-soft-float -msoft-float @gol
498 -malpha-as -mgas @gol
499 -mieee -mieee-with-inexact -mieee-conformant @gol
500 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
501 -mtrap-precision=@var{mode} -mbuild-constants @gol
502 -mcpu=@var{cpu type} @gol
503 -mbwx -mno-bwx -mcix -mno-cix -mmax -mno-max @gol
504 -mmemory-latency=@var{time}}
506 @emph{Clipper Options}
510 @emph{H8/300 Options}
512 -mrelax -mh -ms -mint32 -malign-300}
516 -m1 -m2 -m3 -m3e @gol
517 -m4-nofpu -m4-single-only -m4-single -m4 @gol
518 -mb -ml -mdalign -mrelax @gol
519 -mbigtable -mfmovd -mhitachi -mnomacsave @gol
520 -misize -mpadstruct -mspace @gol
524 @emph{System V Options}
526 -Qy -Qn -YP\,@var{paths} -Ym\,@var{dir}}
531 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text section} @gol
532 -mdata=@var{data section} -mrodata=@var{readonly data section}}
534 @emph{TMS320C3x/C4x Options}
536 -mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm @gol
537 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload @gol
538 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned @gol
539 -mparallel-insns -mparallel-mpy -mpreserve-float}
543 -mlong-calls -mno-long-calls -mep -mno-ep @gol
544 -mprolog-function -mno-prolog-function -mspace @gol
545 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
550 -m32032 -m32332 -m32532 -m32081 -m32381 -mmult-add -mnomult-add @gol
551 -msoft-float -mrtd -mnortd -mregparam -mnoregparam -msb -mnosb @gol
552 -mbitfield -mnobitfield -mhimem -mnohimem}
556 -mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts @gol
557 -mcall-prologues -mno-tablejump -mtiny-stack}
561 -mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
562 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
563 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
564 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
565 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
569 -mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
570 -mvolatile-asm-stop -mb-step -mregister-names -mno-sdata @gol
571 -mconstant-gp -mauto-pic -minline-divide-min-latency @gol
572 -minline-divide-max-throughput -mno-dwarf2-asm @gol
573 -mfixed-range=@var{register range}}
575 @item Code Generation Options
576 @xref{Code Gen Options,,Options for Code Generation Conventions}.
578 -fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
579 -fexceptions -funwind-tables -ffixed-@var{reg} @gol
580 -finhibit-size-directive -finstrument-functions @gol
581 -fcheck-memory-usage -fprefix-function-name @gol
582 -fno-common -fno-ident -fno-gnu-linker @gol
583 -fpcc-struct-return -fpic -fPIC @gol
584 -freg-struct-return -fshared-data -fshort-enums @gol
585 -fshort-double -fvolatile -fvolatile-global -fvolatile-static @gol
586 -fverbose-asm -fpack-struct -fstack-check @gol
587 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
588 -fargument-alias -fargument-noalias @gol
589 -fargument-noalias-global @gol
590 -fleading-underscore}
594 * Overall Options:: Controlling the kind of output:
595 an executable, object files, assembler files,
596 or preprocessed source.
597 * C Dialect Options:: Controlling the variant of C language compiled.
598 * C++ Dialect Options:: Variations on C++.
599 * Language Independent Options:: Controlling how diagnostics should be
601 * Warning Options:: How picky should the compiler be?
602 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
603 * Optimize Options:: How much optimization?
604 * Preprocessor Options:: Controlling header files and macro definitions.
605 Also, getting dependency information for Make.
606 * Assembler Options:: Passing options to the assembler.
607 * Link Options:: Specifying libraries and so on.
608 * Directory Options:: Where to find header files and libraries.
609 Where to find the compiler executable files.
610 * Spec Files:: How to pass switches to sub-processes.
611 * Target Options:: Running a cross-compiler, or an old version of GCC.
614 @node Overall Options
615 @section Options Controlling the Kind of Output
617 Compilation can involve up to four stages: preprocessing, compilation
618 proper, assembly and linking, always in that order. The first three
619 stages apply to an individual source file, and end by producing an
620 object file; linking combines all the object files (those newly
621 compiled, and those specified as input) into an executable file.
623 @cindex file name suffix
624 For any given input file, the file name suffix determines what kind of
629 C source code which must be preprocessed.
632 C source code which should not be preprocessed.
635 C++ source code which should not be preprocessed.
638 Objective-C source code. Note that you must link with the library
639 @file{libobjc.a} to make an Objective-C program work.
642 Objective-C source code which should not be preprocessed.
645 C header file (not to be compiled or linked).
649 @itemx @var{file}.cxx
650 @itemx @var{file}.cpp
651 @itemx @var{file}.c++
653 C++ source code which must be preprocessed. Note that in @samp{.cxx},
654 the last two letters must both be literally @samp{x}. Likewise,
655 @samp{.C} refers to a literal capital C.
658 @itemx @var{file}.for
659 @itemx @var{file}.FOR
660 Fortran source code which should not be preprocessed.
663 @itemx @var{file}.fpp
664 @itemx @var{file}.FPP
665 Fortran source code which must be preprocessed (with the traditional
669 Fortran source code which must be preprocessed with a RATFOR
670 preprocessor (not included with GCC).
672 @xref{Overall Options,,Options Controlling the Kind of Output, g77,
673 Using and Porting GNU Fortran}, for more details of the handling of
676 @c FIXME: Descriptions of Java file types.
682 @c GCC also knows about some suffixes for languages not yet included:
692 @itemx @var{file}.chi
693 CHILL source code (preprocessed with the traditional preprocessor).
699 Assembler code which must be preprocessed.
702 An object file to be fed straight into linking.
703 Any file name with no recognized suffix is treated this way.
706 You can specify the input language explicitly with the @samp{-x} option:
709 @item -x @var{language}
710 Specify explicitly the @var{language} for the following input files
711 (rather than letting the compiler choose a default based on the file
712 name suffix). This option applies to all following input files until
713 the next @samp{-x} option. Possible values for @var{language} are:
715 c c-header cpp-output
717 objective-c objc-cpp-output
718 assembler assembler-with-cpp
719 f77 f77-cpp-input ratfor
722 @c Also f77-version, for internal use only.
725 Turn off any specification of a language, so that subsequent files are
726 handled according to their file name suffixes (as they are if @samp{-x}
727 has not been used at all).
729 @item -pass-exit-codes
730 Normally the @command{gcc} program will exit with the code of 1 if any
731 phase of the compiler returns a non-success return code. If you specify
732 @samp{-pass-exit-codes}, the @command{gcc} program will instead return with
733 numerically highest error produced by any phase that returned an error
737 If you only want some of the stages of compilation, you can use
738 @samp{-x} (or filename suffixes) to tell @command{gcc} where to start, and
739 one of the options @samp{-c}, @samp{-S}, or @samp{-E} to say where
740 @command{gcc} is to stop. Note that some combinations (for example,
741 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
745 Compile or assemble the source files, but do not link. The linking
746 stage simply is not done. The ultimate output is in the form of an
747 object file for each source file.
749 By default, the object file name for a source file is made by replacing
750 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
752 Unrecognized input files, not requiring compilation or assembly, are
756 Stop after the stage of compilation proper; do not assemble. The output
757 is in the form of an assembler code file for each non-assembler input
760 By default, the assembler file name for a source file is made by
761 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
763 Input files that don't require compilation are ignored.
766 Stop after the preprocessing stage; do not run the compiler proper. The
767 output is in the form of preprocessed source code, which is sent to the
770 Input files which don't require preprocessing are ignored.
772 @cindex output file option
774 Place output in file @var{file}. This applies regardless to whatever
775 sort of output is being produced, whether it be an executable file,
776 an object file, an assembler file or preprocessed C code.
778 Since only one output file can be specified, it does not make sense to
779 use @samp{-o} when compiling more than one input file, unless you are
780 producing an executable file as output.
782 If @samp{-o} is not specified, the default is to put an executable file
783 in @file{a.out}, the object file for @file{@var{source}.@var{suffix}} in
784 @file{@var{source}.o}, its assembler file in @file{@var{source}.s}, and
785 all preprocessed C source on standard output.@refill
788 Print (on standard error output) the commands executed to run the stages
789 of compilation. Also print the version number of the compiler driver
790 program and of the preprocessor and the compiler proper.
793 Use pipes rather than temporary files for communication between the
794 various stages of compilation. This fails to work on some systems where
795 the assembler is unable to read from a pipe; but the GNU assembler has
799 Print (on the standard output) a description of the command line options
800 understood by @command{gcc}. If the @option{-v} option is also specified
801 then @option{--help} will also be passed on to the various processes
802 invoked by @command{gcc}, so that they can display the command line options
803 they accept. If the @option{-W} option is also specified then command
804 line options which have no documentation associated with them will also
808 Print (on the standard output) a description of target specific command
809 line options for each tool.
813 @section Compiling C++ Programs
815 @cindex suffixes for C++ source
816 @cindex C++ source file suffixes
817 C++ source files conventionally use one of the suffixes @samp{.C},
818 @samp{.cc}, @samp{.cpp}, @samp{.c++}, @samp{.cp}, or @samp{.cxx};
819 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
820 files with these names and compiles them as C++ programs even if you
821 call the compiler the same way as for compiling C programs (usually with
822 the name @command{gcc}).
826 However, C++ programs often require class libraries as well as a
827 compiler that understands the C++ language---and under some
828 circumstances, you might want to compile programs from standard input,
829 or otherwise without a suffix that flags them as C++ programs.
830 @command{g++} is a program that calls GCC with the default language
831 set to C++, and automatically specifies linking against the C++
832 library. On many systems, @command{g++} is also
833 installed with the name @command{c++}.
835 @cindex invoking @command{g++}
836 When you compile C++ programs, you may specify many of the same
837 command-line options that you use for compiling programs in any
838 language; or command-line options meaningful for C and related
839 languages; or options that are meaningful only for C++ programs.
840 @xref{C Dialect Options,,Options Controlling C Dialect}, for
841 explanations of options for languages related to C.
842 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
843 explanations of options that are meaningful only for C++ programs.
845 @node C Dialect Options
846 @section Options Controlling C Dialect
847 @cindex dialect options
848 @cindex language dialect options
849 @cindex options, dialect
851 The following options control the dialect of C (or languages derived
852 from C, such as C++ and Objective C) that the compiler accepts:
858 In C mode, support all ISO C89 programs. In C++ mode,
859 remove GNU extensions that conflict with ISO C++.
861 This turns off certain features of GCC that are incompatible with ISO
862 C (when compiling C code), or of standard C++ (when compiling C++ code),
863 such as the @code{asm} and @code{typeof} keywords, and
864 predefined macros such as @code{unix} and @code{vax} that identify the
865 type of system you are using. It also enables the undesirable and
866 rarely used ISO trigraph feature. For the C compiler,
867 it disables recognition of C++ style @samp{//} comments as well as
868 the @code{inline} keyword.
870 The alternate keywords @code{__asm__}, @code{__extension__},
871 @code{__inline__} and @code{__typeof__} continue to work despite
872 @samp{-ansi}. You would not want to use them in an ISO C program, of
873 course, but it is useful to put them in header files that might be included
874 in compilations done with @samp{-ansi}. Alternate predefined macros
875 such as @code{__unix__} and @code{__vax__} are also available, with or
876 without @samp{-ansi}.
878 The @samp{-ansi} option does not cause non-ISO programs to be
879 rejected gratuitously. For that, @samp{-pedantic} is required in
880 addition to @samp{-ansi}. @xref{Warning Options}.
882 The macro @code{__STRICT_ANSI__} is predefined when the @samp{-ansi}
883 option is used. Some header files may notice this macro and refrain
884 from declaring certain functions or defining certain macros that the
885 ISO standard doesn't call for; this is to avoid interfering with any
886 programs that might use these names for other things.
888 Functions which would normally be builtin but do not have semantics
889 defined by ISO C (such as @code{alloca} and @code{ffs}) are not builtin
890 functions with @samp{-ansi} is used. @xref{Other Builtins,,Other
891 built-in functions provided by GNU CC}, for details of the functions
895 Determine the language standard. A value for this option must be provided;
900 Same as @option{-ansi}
903 ISO C as modified in amend. 1
906 ISO C99. Note that this standard is not yet fully supported; see
907 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information.
910 same as @option{-std=iso9899:1990}
913 same as @option{-std=iso9899:1999}
916 default, iso9899:1990 + gnu extensions
919 iso9899:1999 + gnu extensions
922 same as @option{-std=iso9899:1999}, deprecated
925 same as @option{-std=iso9899:1999}, deprecated
928 same as @option{-std=gnu99}, deprecated
932 Even when this option is not specified, you can still use some of the
933 features of newer standards in so far as they do not conflict with
934 previous C standards. For example, you may use @code{__restrict__} even
935 when @option{-std=c99} is not specified.
937 The @option{-std} options specifying some version of ISO C have the same
938 effects as @option{-ansi}, except that features that were not in ISO C89
939 but are in the specified version (for example, @samp{//} comments and
940 the @code{inline} keyword in ISO C99) are not disabled.
942 @xref{Standards,,Language Standards Supported by GCC}, for details of
943 these standard versions.
946 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
947 keyword, so that code can use these words as identifiers. You can use
948 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
949 instead. @samp{-ansi} implies @samp{-fno-asm}.
951 In C++, this switch only affects the @code{typeof} keyword, since
952 @code{asm} and @code{inline} are standard keywords. You may want to
953 use the @samp{-fno-gnu-keywords} flag instead, which has the same
954 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
955 switch only affects the @code{asm} and @code{typeof} keywords, since
956 @code{inline} is a standard keyword in ISO C99.
959 @cindex builtin functions
960 Don't recognize builtin functions that do not begin with
961 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
962 functions provided by GNU CC}, for details of the functions affected,
963 including those which are not builtin functions when @option{-ansi} or
964 @option{-std} options for strict ISO C conformance are used because they
965 do not have an ISO standard meaning.
967 GCC normally generates special code to handle certain builtin functions
968 more efficiently; for instance, calls to @code{alloca} may become single
969 instructions that adjust the stack directly, and calls to @code{memcpy}
970 may become inline copy loops. The resulting code is often both smaller
971 and faster, but since the function calls no longer appear as such, you
972 cannot set a breakpoint on those calls, nor can you change the behavior
973 of the functions by linking with a different library.
976 @cindex hosted environment
978 Assert that compilation takes place in a hosted environment. This implies
979 @samp{-fbuiltin}. A hosted environment is one in which the
980 entire standard library is available, and in which @code{main} has a return
981 type of @code{int}. Examples are nearly everything except a kernel.
982 This is equivalent to @samp{-fno-freestanding}.
985 @cindex hosted environment
987 Assert that compilation takes place in a freestanding environment. This
988 implies @samp{-fno-builtin}. A freestanding environment
989 is one in which the standard library may not exist, and program startup may
990 not necessarily be at @code{main}. The most obvious example is an OS kernel.
991 This is equivalent to @samp{-fno-hosted}.
993 @xref{Standards,,Language Standards Supported by GCC}, for details of
994 freestanding and hosted environments.
997 Support ISO C trigraphs. You don't want to know about this
998 brain-damage. The @option{-ansi} option (and @option{-std} options for
999 strict ISO C conformance) implies @option{-trigraphs}.
1001 @cindex traditional C language
1002 @cindex C language, traditional
1004 Attempt to support some aspects of traditional C compilers.
1009 All @code{extern} declarations take effect globally even if they
1010 are written inside of a function definition. This includes implicit
1011 declarations of functions.
1014 The newer keywords @code{typeof}, @code{inline}, @code{signed}, @code{const}
1015 and @code{volatile} are not recognized. (You can still use the
1016 alternative keywords such as @code{__typeof__}, @code{__inline__}, and
1020 Comparisons between pointers and integers are always allowed.
1023 Integer types @code{unsigned short} and @code{unsigned char} promote
1024 to @code{unsigned int}.
1027 Out-of-range floating point literals are not an error.
1030 Certain constructs which ISO regards as a single invalid preprocessing
1031 number, such as @samp{0xe-0xd}, are treated as expressions instead.
1034 String ``constants'' are not necessarily constant; they are stored in
1035 writable space, and identical looking constants are allocated
1036 separately. (This is the same as the effect of
1037 @samp{-fwritable-strings}.)
1039 @cindex @code{longjmp} and automatic variables
1041 All automatic variables not declared @code{register} are preserved by
1042 @code{longjmp}. Ordinarily, GNU C follows ISO C: automatic variables
1043 not declared @code{volatile} may be clobbered.
1048 @cindex escape sequences, traditional
1049 The character escape sequences @samp{\x} and @samp{\a} evaluate as the
1050 literal characters @samp{x} and @samp{a} respectively. Without
1051 @w{@samp{-traditional}}, @samp{\x} is a prefix for the hexadecimal
1052 representation of a character, and @samp{\a} produces a bell.
1055 You may wish to use @samp{-fno-builtin} as well as @samp{-traditional}
1056 if your program uses names that are normally GNU C builtin functions for
1057 other purposes of its own.
1059 You cannot use @samp{-traditional} if you include any header files that
1060 rely on ISO C features. Some vendors are starting to ship systems with
1061 ISO C header files and you cannot use @samp{-traditional} on such
1062 systems to compile files that include any system headers.
1064 The @samp{-traditional} option also enables @samp{-traditional-cpp},
1065 which is described next.
1067 @item -traditional-cpp
1068 Attempt to support some aspects of traditional C preprocessors.
1073 Comments convert to nothing at all, rather than to a space. This allows
1074 traditional token concatenation.
1077 In a preprocessing directive, the @samp{#} symbol must appear as the first
1078 character of a line.
1081 Macro arguments are recognized within string constants in a macro
1082 definition (and their values are stringified, though without additional
1083 quote marks, when they appear in such a context). The preprocessor
1084 always considers a string constant to end at a newline.
1087 @cindex detecting @w{@samp{-traditional}}
1088 The predefined macro @code{__STDC__} is not defined when you use
1089 @samp{-traditional}, but @code{__GNUC__} is (since the GNU extensions
1090 which @code{__GNUC__} indicates are not affected by
1091 @samp{-traditional}). If you need to write header files that work
1092 differently depending on whether @samp{-traditional} is in use, by
1093 testing both of these predefined macros you can distinguish four
1094 situations: GNU C, traditional GNU C, other ISO C compilers, and other
1095 old C compilers. The predefined macro @code{__STDC_VERSION__} is also
1096 not defined when you use @samp{-traditional}. @xref{Standard
1097 Predefined,,Standard Predefined Macros,cpp.info,The C Preprocessor},
1098 for more discussion of these and other predefined macros.
1101 @cindex string constants vs newline
1102 @cindex newline vs string constants
1103 The preprocessor considers a string constant to end at a newline (unless
1104 the newline is escaped with @samp{\}). (Without @w{@samp{-traditional}},
1105 string constants can contain the newline character as typed.)
1108 @item -fcond-mismatch
1109 Allow conditional expressions with mismatched types in the second and
1110 third arguments. The value of such an expression is void. This option
1111 is not supported for C++.
1113 @item -funsigned-char
1114 Let the type @code{char} be unsigned, like @code{unsigned char}.
1116 Each kind of machine has a default for what @code{char} should
1117 be. It is either like @code{unsigned char} by default or like
1118 @code{signed char} by default.
1120 Ideally, a portable program should always use @code{signed char} or
1121 @code{unsigned char} when it depends on the signedness of an object.
1122 But many programs have been written to use plain @code{char} and
1123 expect it to be signed, or expect it to be unsigned, depending on the
1124 machines they were written for. This option, and its inverse, let you
1125 make such a program work with the opposite default.
1127 The type @code{char} is always a distinct type from each of
1128 @code{signed char} or @code{unsigned char}, even though its behavior
1129 is always just like one of those two.
1132 Let the type @code{char} be signed, like @code{signed char}.
1134 Note that this is equivalent to @samp{-fno-unsigned-char}, which is
1135 the negative form of @samp{-funsigned-char}. Likewise, the option
1136 @samp{-fno-signed-char} is equivalent to @samp{-funsigned-char}.
1138 You may wish to use @samp{-fno-builtin} as well as @samp{-traditional}
1139 if your program uses names that are normally GNU C builtin functions for
1140 other purposes of its own.
1142 You cannot use @samp{-traditional} if you include any header files that
1143 rely on ISO C features. Some vendors are starting to ship systems with
1144 ISO C header files and you cannot use @samp{-traditional} on such
1145 systems to compile files that include any system headers.
1147 @item -fsigned-bitfields
1148 @itemx -funsigned-bitfields
1149 @itemx -fno-signed-bitfields
1150 @itemx -fno-unsigned-bitfields
1151 These options control whether a bitfield is signed or unsigned, when the
1152 declaration does not use either @code{signed} or @code{unsigned}. By
1153 default, such a bitfield is signed, because this is consistent: the
1154 basic integer types such as @code{int} are signed types.
1156 However, when @samp{-traditional} is used, bitfields are all unsigned
1159 @item -fwritable-strings
1160 Store string constants in the writable data segment and don't uniquize
1161 them. This is for compatibility with old programs which assume they can
1162 write into string constants. The option @samp{-traditional} also has
1165 Writing into string constants is a very bad idea; ``constants'' should
1168 @item -fallow-single-precision
1169 Do not promote single precision math operations to double precision,
1170 even when compiling with @samp{-traditional}.
1172 Traditional K&R C promotes all floating point operations to double
1173 precision, regardless of the sizes of the operands. On the
1174 architecture for which you are compiling, single precision may be faster
1175 than double precision. If you must use @samp{-traditional}, but want
1176 to use single precision operations when the operands are single
1177 precision, use this option. This option has no effect when compiling
1178 with ISO or GNU C conventions (the default).
1181 Override the underlying type for @samp{wchar_t} to be @samp{short
1182 unsigned int} instead of the default for the target. This option is
1183 useful for building programs to run under WINE.
1186 @node C++ Dialect Options
1187 @section Options Controlling C++ Dialect
1189 @cindex compiler options, C++
1190 @cindex C++ options, command line
1191 @cindex options, C++
1192 This section describes the command-line options that are only meaningful
1193 for C++ programs; but you can also use most of the GNU compiler options
1194 regardless of what language your program is in. For example, you
1195 might compile a file @code{firstClass.C} like this:
1198 g++ -g -frepo -O -c firstClass.C
1202 In this example, only @samp{-frepo} is an option meant
1203 only for C++ programs; you can use the other options with any
1204 language supported by GCC.
1206 Here is a list of options that are @emph{only} for compiling C++ programs:
1209 @item -fno-access-control
1210 Turn off all access checking. This switch is mainly useful for working
1211 around bugs in the access control code.
1214 Check that the pointer returned by @code{operator new} is non-null
1215 before attempting to modify the storage allocated. The current Working
1216 Paper requires that @code{operator new} never return a null pointer, so
1217 this check is normally unnecessary.
1219 An alternative to using this option is to specify that your
1220 @code{operator new} does not throw any exceptions; if you declare it
1221 @samp{throw()}, g++ will check the return value. See also @samp{new
1224 @item -fconserve-space
1225 Put uninitialized or runtime-initialized global variables into the
1226 common segment, as C does. This saves space in the executable at the
1227 cost of not diagnosing duplicate definitions. If you compile with this
1228 flag and your program mysteriously crashes after @code{main()} has
1229 completed, you may have an object that is being destroyed twice because
1230 two definitions were merged.
1232 This option is no longer useful on most targets, now that support has
1233 been added for putting variables into BSS without making them common.
1235 @item -fno-const-strings
1236 Give string constants type @code{char *} instead of type @code{const
1237 char *}. By default, G++ uses type @code{const char *} as required by
1238 the standard. Even if you use @samp{-fno-const-strings}, you cannot
1239 actually modify the value of a string constant, unless you also use
1240 @samp{-fwritable-strings}.
1242 This option might be removed in a future release of G++. For maximum
1243 portability, you should structure your code so that it works with
1244 string constants that have type @code{const char *}.
1246 @item -fdollars-in-identifiers
1247 Accept @samp{$} in identifiers. You can also explicitly prohibit use of
1248 @samp{$} with the option @samp{-fno-dollars-in-identifiers}. (GNU C allows
1249 @samp{$} by default on most target systems, but there are a few exceptions.)
1250 Traditional C allowed the character @samp{$} to form part of
1251 identifiers. However, ISO C and C++ forbid @samp{$} in identifiers.
1253 @item -fno-elide-constructors
1254 The C++ standard allows an implementation to omit creating a temporary
1255 which is only used to initialize another object of the same type.
1256 Specifying this option disables that optimization, and forces g++ to
1257 call the copy constructor in all cases.
1259 @item -fno-enforce-eh-specs
1260 Don't check for violation of exception specifications at runtime. This
1261 option violates the C++ standard, but may be useful for reducing code
1262 size in production builds, much like defining @samp{NDEBUG}. The compiler
1263 will still optimize based on the exception specifications.
1265 @item -fexternal-templates
1266 Cause template instantiations to obey @samp{#pragma interface} and
1267 @samp{implementation}; template instances are emitted or not according
1268 to the location of the template definition. @xref{Template
1269 Instantiation}, for more information.
1271 This option is deprecated.
1273 @item -falt-external-templates
1274 Similar to -fexternal-templates, but template instances are emitted or
1275 not according to the place where they are first instantiated.
1276 @xref{Template Instantiation}, for more information.
1278 This option is deprecated.
1281 @itemx -fno-for-scope
1282 If -ffor-scope is specified, the scope of variables declared in
1283 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1284 as specified by the C++ standard.
1285 If -fno-for-scope is specified, the scope of variables declared in
1286 a @i{for-init-statement} extends to the end of the enclosing scope,
1287 as was the case in old versions of gcc, and other (traditional)
1288 implementations of C++.
1290 The default if neither flag is given to follow the standard,
1291 but to allow and give a warning for old-style code that would
1292 otherwise be invalid, or have different behavior.
1294 @item -fno-gnu-keywords
1295 Do not recognize @code{typeof} as a keyword, so that code can use this
1296 word as an identifier. You can use the keyword @code{__typeof__} instead.
1297 @samp{-ansi} implies @samp{-fno-gnu-keywords}.
1300 Treat the @code{namespace std} as a namespace, instead of ignoring
1301 it. For compatibility with earlier versions of g++, the compiler will,
1302 by default, ignore @code{namespace-declarations},
1303 @code{using-declarations}, @code{using-directives}, and
1304 @code{namespace-names}, if they involve @code{std}.
1306 @item -fhuge-objects
1307 Support virtual function calls for objects that exceed the size
1308 representable by a @samp{short int}. Users should not use this flag by
1309 default; if you need to use it, the compiler will tell you so.
1311 This flag is not useful when compiling with -fvtable-thunks.
1313 Like all options that change the ABI, all C++ code, @emph{including
1314 libgcc} must be built with the same setting of this option.
1316 @item -fno-implicit-templates
1317 Never emit code for non-inline templates which are instantiated
1318 implicitly (i.e. by use); only emit code for explicit instantiations.
1319 @xref{Template Instantiation}, for more information.
1321 @item -fno-implicit-inline-templates
1322 Don't emit code for implicit instantiations of inline templates, either.
1323 The default is to handle inlines differently so that compiles with and
1324 without optimization will need the same set of explicit instantiations.
1326 @item -fno-implement-inlines
1327 To save space, do not emit out-of-line copies of inline functions
1328 controlled by @samp{#pragma implementation}. This will cause linker
1329 errors if these functions are not inlined everywhere they are called.
1331 @item -fms-extensions
1332 Disable pedantic warnings about constructs used in MFC, such as implicit
1333 int and getting a pointer to member function via non-standard syntax.
1335 @item -fno-nonansi-builtins
1336 Disable builtin declarations of functions that are not mandated by
1337 ANSI/ISO C. These include @code{ffs}, @code{alloca}, @code{_exit},
1338 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1340 @item -fno-operator-names
1341 Do not treat the operator name keywords @code{and}, @code{bitand},
1342 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1343 synonyms as keywords.
1345 @item -fno-optional-diags
1346 Disable diagnostics that the standard says a compiler does not need to
1347 issue. Currently, the only such diagnostic issued by g++ is the one for
1348 a name having multiple meanings within a class.
1351 Downgrade messages about nonconformant code from errors to warnings. By
1352 default, g++ effectively sets @samp{-pedantic-errors} without
1353 @samp{-pedantic}; this option reverses that. This behavior and this
1354 option are superseded by @samp{-pedantic}, which works as it does for GNU C.
1357 Enable automatic template instantiation. This option also implies
1358 @samp{-fno-implicit-templates}. @xref{Template Instantiation}, for more
1362 Disable generation of information about every class with virtual
1363 functions for use by the C++ runtime type identification features
1364 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1365 of the language, you can save some space by using this flag. Note that
1366 exception handling uses the same information, but it will generate it as
1370 Emit statistics about front-end processing at the end of the compilation.
1371 This information is generally only useful to the G++ development team.
1373 @item -ftemplate-depth-@var{n}
1374 Set the maximum instantiation depth for template classes to @var{n}.
1375 A limit on the template instantiation depth is needed to detect
1376 endless recursions during template class instantiation. ANSI/ISO C++
1377 conforming programs must not rely on a maximum depth greater than 17.
1379 @item -fuse-cxa-atexit
1380 Register destructors for objects with static storage duration with the
1381 @code{__cxa_atexit} function rather than the @code{atexit} function.
1382 This option is required for fully standards-compliant handling of static
1383 destructors, but will only work if your C library supports
1384 @code{__cxa_atexit}.
1386 @item -fvtable-thunks
1387 Use @samp{thunks} to implement the virtual function dispatch table
1388 (@samp{vtable}). The traditional (cfront-style) approach to
1389 implementing vtables was to store a pointer to the function and two
1390 offsets for adjusting the @samp{this} pointer at the call site. Newer
1391 implementations store a single pointer to a @samp{thunk} function which
1392 does any necessary adjustment and then calls the target function.
1394 This option also enables a heuristic for controlling emission of
1395 vtables; if a class has any non-inline virtual functions, the vtable
1396 will be emitted in the translation unit containing the first one of
1399 Like all options that change the ABI, all C++ code, @emph{including
1400 libgcc.a} must be built with the same setting of this option.
1403 Do not use weak symbol support, even if it is provied by the linker.
1404 By default, G++ will use weak symbols if they are available. This
1405 option exists only for testing, and should not be used by end-users;
1406 it will result in inferior code and has no benefits. This option may
1407 be removed in a future release of G++.
1410 Do not search for header files in the standard directories specific to
1411 C++, but do still search the other standard directories. (This option
1412 is used when building the C++ library.)
1415 In addition, these optimization, warning, and code generation options
1416 have meanings only for C++ programs:
1419 @item -fno-default-inline
1420 Do not assume @samp{inline} for functions defined inside a class scope.
1421 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1422 functions will have linkage like inline functions; they just won't be
1425 @item -Wctor-dtor-privacy (C++ only)
1426 Warn when a class seems unusable, because all the constructors or
1427 destructors in a class are private and the class has no friends or
1428 public static member functions.
1430 @item -Wnon-virtual-dtor (C++ only)
1431 Warn when a class declares a non-virtual destructor that should probably
1432 be virtual, because it looks like the class will be used polymorphically.
1434 @item -Wreorder (C++ only)
1435 @cindex reordering, warning
1436 @cindex warning for reordering of member initializers
1437 Warn when the order of member initializers given in the code does not
1438 match the order in which they must be executed. For instance:
1444 A(): j (0), i (1) @{ @}
1448 Here the compiler will warn that the member initializers for @samp{i}
1449 and @samp{j} will be rearranged to match the declaration order of the
1453 The following @samp{-W@dots{}} options are not affected by @samp{-Wall}.
1456 @item -Weffc++ (C++ only)
1457 Warn about violations of various style guidelines from Scott Meyers'
1458 @cite{Effective C++} books. If you use this option, you should be aware
1459 that the standard library headers do not obey all of these guidelines;
1460 you can use @samp{grep -v} to filter out those warnings.
1462 @item -Wno-deprecated (C++ only)
1463 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1465 @item -Wno-non-template-friend (C++ only)
1466 Disable warnings when non-templatized friend functions are declared
1467 within a template. With the advent of explicit template specification
1468 support in g++, if the name of the friend is an unqualified-id (ie,
1469 @samp{friend foo(int)}), the C++ language specification demands that the
1470 friend declare or define an ordinary, nontemplate function. (Section
1471 14.5.3). Before g++ implemented explicit specification, unqualified-ids
1472 could be interpreted as a particular specialization of a templatized
1473 function. Because this non-conforming behavior is no longer the default
1474 behavior for g++, @samp{-Wnon-template-friend} allows the compiler to
1475 check existing code for potential trouble spots, and is on by default.
1476 This new compiler behavior can be turned off with
1477 @samp{-Wno-non-template-friend} which keeps the conformant compiler code
1478 but disables the helpful warning.
1480 @item -Wold-style-cast (C++ only)
1481 Warn if an old-style (C-style) cast is used within a C++ program. The
1482 new-style casts (@samp{static_cast}, @samp{reinterpret_cast}, and
1483 @samp{const_cast}) are less vulnerable to unintended effects.
1485 @item -Woverloaded-virtual (C++ only)
1486 @cindex overloaded virtual fn, warning
1487 @cindex warning for overloaded virtual fn
1488 Warn when a function declaration hides virtual functions from a
1489 base class. For example, in:
1496 struct B: public A @{
1501 the @code{A} class version of @code{f} is hidden in @code{B}, and code
1509 will fail to compile.
1511 @item -Wno-pmf-conversions (C++ only)
1512 Disable the diagnostic for converting a bound pointer to member function
1515 @item -Wsign-promo (C++ only)
1516 Warn when overload resolution chooses a promotion from unsigned or
1517 enumeral type to a signed type over a conversion to an unsigned type of
1518 the same size. Previous versions of g++ would try to preserve
1519 unsignedness, but the standard mandates the current behavior.
1521 @item -Wsynth (C++ only)
1522 @cindex warning for synthesized methods
1523 @cindex synthesized methods, warning
1524 Warn when g++'s synthesis behavior does not match that of cfront. For
1530 A& operator = (int);
1540 In this example, g++ will synthesize a default @samp{A& operator =
1541 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1544 @node Language Independent Options
1545 @section Options to Control Diagnostic Messages Formatting
1546 @cindex options to control diagnostics formatting
1547 @cindex diagnostic messages
1548 @cindex message formatting
1550 Traditionally, diagnostic messages have been formatted irrespective of
1551 the output device's aspect (e.g. its width, ...). The options described
1552 below can be used to control the diagnostic messages formatting
1553 algorithm, e.g. how many characters per line, how often source location
1554 information should be reported. Right now, only the C++ front-end can
1555 honor these options. However it is expected, in the near future, that
1556 the remaining front-ends would be able to digest them correctly.
1559 @item -fmessage-length=@var{n}
1560 Try to format error messages so that they fit on lines of about @var{n}
1561 characters. The default is 72 characters for g++ and 0 for the rest of
1562 the front-ends supported by GCC. If @var{n} is zero, then no
1563 line-wrapping will be done; each error message will appear on a single
1566 @item -fdiagnostics-show-location=once
1567 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
1568 reporter to emit @emph{once} source location information; that is, in
1569 case the message is too long to fit on a single physical line and has to
1570 be wrapped, the source location won't be emitted (as prefix) again,
1571 over and over, in subsequent continuation lines. This is the default
1574 @item -fdiagnostics-show-location=every-line
1575 Only meaningful in line-wrapping mode. Instructs the diagnostic
1576 messages reporter to emit the same source location information (as
1577 prefix) for physical lines that result from the process of breaking a
1578 a message which is too long to fit on a single line.
1582 @node Warning Options
1583 @section Options to Request or Suppress Warnings
1584 @cindex options to control warnings
1585 @cindex warning messages
1586 @cindex messages, warning
1587 @cindex suppressing warnings
1589 Warnings are diagnostic messages that report constructions which
1590 are not inherently erroneous but which are risky or suggest there
1591 may have been an error.
1593 You can request many specific warnings with options beginning @samp{-W},
1594 for example @samp{-Wimplicit} to request warnings on implicit
1595 declarations. Each of these specific warning options also has a
1596 negative form beginning @samp{-Wno-} to turn off warnings;
1597 for example, @samp{-Wno-implicit}. This manual lists only one of the
1598 two forms, whichever is not the default.
1600 These options control the amount and kinds of warnings produced by GCC:
1603 @cindex syntax checking
1605 Check the code for syntax errors, but don't do anything beyond that.
1608 Issue all the warnings demanded by strict ISO C and ISO C++;
1609 reject all programs that use forbidden extensions, and some other
1610 programs that do not follow ISO C and ISO C++. For ISO C, follows the
1611 version of the ISO C standard specified by any @samp{-std} option used.
1613 Valid ISO C and ISO C++ programs should compile properly with or without
1614 this option (though a rare few will require @option{-ansi} or a
1615 @option{-std} option specifying the required version of ISO C). However,
1616 without this option, certain GNU extensions and traditional C and C++
1617 features are supported as well. With this option, they are rejected.
1619 @samp{-pedantic} does not cause warning messages for use of the
1620 alternate keywords whose names begin and end with @samp{__}. Pedantic
1621 warnings are also disabled in the expression that follows
1622 @code{__extension__}. However, only system header files should use
1623 these escape routes; application programs should avoid them.
1624 @xref{Alternate Keywords}.
1626 Some users try to use @samp{-pedantic} to check programs for strict ISO
1627 C conformance. They soon find that it does not do quite what they want:
1628 it finds some non-ISO practices, but not all---only those for which
1629 ISO C @emph{requires} a diagnostic, and some others for which
1630 diagnostics have been added.
1632 A feature to report any failure to conform to ISO C might be useful in
1633 some instances, but would require considerable additional work and would
1634 be quite different from @samp{-pedantic}. We don't have plans to
1635 support such a feature in the near future.
1637 @item -pedantic-errors
1638 Like @samp{-pedantic}, except that errors are produced rather than
1642 Inhibit all warning messages.
1645 Inhibit warning messages about the use of @samp{#import}.
1647 @item -Wchar-subscripts
1648 Warn if an array subscript has type @code{char}. This is a common cause
1649 of error, as programmers often forget that this type is signed on some
1653 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
1654 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
1657 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
1658 the arguments supplied have types appropriate to the format string
1659 specified, and that the conversions specified in the format string make
1660 sense. This includes standard functions, and others specified by format
1661 attributes (@pxref{Function Attributes}), in the @code{printf},
1662 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
1663 not in the C standard) families.
1665 The formats are checked against the format features supported by GNU
1666 libc version 2.2. These include all ISO C89 and C99 features, as well
1667 as features from the Single Unix Specification and some BSD and GNU
1668 extensions. Other library implementations may not support all these
1669 features; GCC does not support warning about features that go beyond a
1670 particular library's limitations. However, if @samp{-pedantic} is used
1671 with @samp{-Wformat}, warnings will be given about format features not
1672 in the selected standard version (but not for @code{strfmon} formats,
1673 since those are not in any version of the C standard). @xref{C Dialect
1674 Options,,Options Controlling C Dialect}.
1676 @samp{-Wformat} is included in @samp{-Wall}. For more control over some
1677 aspects of format checking, the options @samp{-Wno-format-y2k},
1678 @samp{-Wno-format-extra-args}, @samp{-Wformat-nonliteral},
1679 @samp{-Wformat-security} and @samp{-Wformat=2} are available, but are
1680 not included in @samp{-Wall}.
1682 @item -Wno-format-y2k
1683 If @samp{-Wformat} is specified, do not warn about @code{strftime}
1684 formats which may yield only a two-digit year.
1686 @item -Wno-format-extra-args
1687 If @samp{-Wformat} is specified, do not warn about excess arguments to a
1688 @code{printf} or @code{scanf} format function. The C standard specifies
1689 that such arguments are ignored.
1691 @item -Wformat-nonliteral
1692 If @samp{-Wformat} is specified, also warn if the format string is not a
1693 string literal and so cannot be checked, unless the format function
1694 takes its format arguments as a @code{va_list}.
1696 @item -Wformat-security
1697 If @samp{-Wformat} is specified, also warn about uses of format
1698 functions that represent possible security problems. At present, this
1699 warns about calls to @code{printf} and @code{scanf} functions where the
1700 format string is not a string literal and there are no format arguments,
1701 as in @code{printf (foo);}. This may be a security hole if the format
1702 string came from untrusted input and contains @samp{%n}. (This is
1703 currently a subset of what @samp{-Wformat-nonliteral} warns about, but
1704 in future warnings may be added to @samp{-Wformat-security} that are not
1705 included in @samp{-Wformat-nonliteral}.)
1708 Enable @samp{-Wformat} plus format checks not included in
1709 @samp{-Wformat}. Currently equivalent to @samp{-Wformat
1710 -Wformat-nonliteral -Wformat-security}.
1712 @item -Wimplicit-int
1713 Warn when a declaration does not specify a type.
1715 @item -Wimplicit-function-declaration
1716 @itemx -Werror-implicit-function-declaration
1717 Give a warning (or error) whenever a function is used before being
1721 Same as @samp{-Wimplicit-int} and @samp{-Wimplicit-function-}@*
1725 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
1726 function with external linkage, returning int, taking either zero
1727 arguments, two, or three arguments of appropriate types.
1729 @item -Wmissing-braces
1730 Warn if an aggregate or union initializer is not fully bracketed. In
1731 the following example, the initializer for @samp{a} is not fully
1732 bracketed, but that for @samp{b} is fully bracketed.
1735 int a[2][2] = @{ 0, 1, 2, 3 @};
1736 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
1740 Warn if a multicharacter constant (@samp{'FOOF'}) is used. Usually they
1741 indicate a typo in the user's code, as they have implementation-defined
1742 values, and should not be used in portable code.
1745 Warn if parentheses are omitted in certain contexts, such
1746 as when there is an assignment in a context where a truth value
1747 is expected, or when operators are nested whose precedence people
1748 often get confused about.
1750 Also warn about constructions where there may be confusion to which
1751 @code{if} statement an @code{else} branch belongs. Here is an example of
1764 In C, every @code{else} branch belongs to the innermost possible @code{if}
1765 statement, which in this example is @code{if (b)}. This is often not
1766 what the programmer expected, as illustrated in the above example by
1767 indentation the programmer chose. When there is the potential for this
1768 confusion, GNU C will issue a warning when this flag is specified.
1769 To eliminate the warning, add explicit braces around the innermost
1770 @code{if} statement so there is no way the @code{else} could belong to
1771 the enclosing @code{if}. The resulting code would look like this:
1785 @item -Wsequence-point
1786 Warn about code that may have undefined semantics because of violations
1787 of sequence point rules in the C standard.
1789 The C standard defines the order in which expressions in a C program are
1790 evaluated in terms of @dfn{sequence points}, which represent a partial
1791 ordering between the execution of parts of the program: those executed
1792 before the sequence point, and those executed after it. These occur
1793 after the evaluation of a full expression (one which is not part of a
1794 larger expression), after the evaluation of the first operand of a
1795 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
1796 function is called (but after the evaluation of its arguments and the
1797 expression denoting the called function), and in certain other places.
1798 Other than as expressed by the sequence point rules, the order of
1799 evaluation of subexpressions of an expression is not specified. All
1800 these rules describe only a partial order rather than a total order,
1801 since, for example, if two functions are called within one expression
1802 with no sequence point between them, the order in which the functions
1803 are called is not specified. However, the standards committee have
1804 ruled that function calls do not overlap.
1806 It is not specified when between sequence points modifications to the
1807 values of objects take effect. Programs whose behavior depends on this
1808 have undefined behavior; the C standard specifies that ``Between the
1809 previous and next sequence point an object shall have its stored value
1810 modified at most once by the evaluation of an expression. Furthermore,
1811 the prior value shall be read only to determine the value to be
1812 stored.''. If a program breaks these rules, the results on any
1813 particular implementation are entirely unpredictable.
1815 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
1816 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
1817 diagnosed by this option, and it may give an occasional false positive
1818 result, but in general it has been found fairly effective at detecting
1819 this sort of problem in programs.
1821 The present implementation of this option only works for C programs. A
1822 future implementation may also work for C++ programs.
1824 There is some controversy over the precise meaning of the sequence point
1825 rules in subtle cases. Alternative formal definitions may be found in
1826 Clive Feather's ``Annex S''
1827 @w{@uref{http://wwwold.dkuug.dk/JTC1/SC22/WG14/www/docs/n925.htm}} and in
1828 Michael Norrish's thesis
1829 @w{@uref{http://www.cl.cam.ac.uk/users/mn200/PhD/thesis-report.ps.gz}}.
1830 Other discussions are by Raymond Mak
1831 @w{@uref{http://wwwold.dkuug.dk/JTC1/SC22/WG14/www/docs/n926.htm}} and
1833 @w{@uref{http://wwwold.dkuug.dk/JTC1/SC22/WG14/www/docs/n927.htm}}.
1836 Warn whenever a function is defined with a return-type that defaults to
1837 @code{int}. Also warn about any @code{return} statement with no
1838 return-value in a function whose return-type is not @code{void}.
1840 For C++, a function without return type always produces a diagnostic
1841 message, even when @samp{-Wno-return-type} is specified. The only
1842 exceptions are @samp{main} and functions defined in system headers.
1845 Warn whenever a @code{switch} statement has an index of enumeral type
1846 and lacks a @code{case} for one or more of the named codes of that
1847 enumeration. (The presence of a @code{default} label prevents this
1848 warning.) @code{case} labels outside the enumeration range also
1849 provoke warnings when this option is used.
1852 Warn if any trigraphs are encountered that might change the meaning of
1853 the program (trigraphs within comments are not warned about).
1855 @item -Wunused-function
1856 Warn whenever a static function is declared but not defined or a
1857 non\-inline static function is unused.
1859 @item -Wunused-label
1860 Warn whenever a label is declared but not used.
1862 To suppress this warning use the @samp{unused} attribute
1863 (@pxref{Variable Attributes}).
1865 @item -Wunused-parameter
1866 Warn whenever a function parameter is unused aside from its declaration.
1868 To suppress this warning use the @samp{unused} attribute
1869 (@pxref{Variable Attributes}).
1871 @item -Wunused-variable
1872 Warn whenever a local variable or non-constant static variable is unused
1873 aside from its declaration
1875 To suppress this warning use the @samp{unused} attribute
1876 (@pxref{Variable Attributes}).
1878 @item -Wunused-value
1879 Warn whenever a statement computes a result that is explicitly not used.
1881 To suppress this warning cast the expression to @samp{void}.
1884 All all the above @samp{-Wunused} options combined.
1886 In order to get a warning about an unused function parameter, you must
1887 either specify @samp{-W -Wunused} or separately specify
1888 @samp{-Wunused-parameter}.
1890 @item -Wuninitialized
1891 Warn if an automatic variable is used without first being initialized or
1892 if a variable may be clobbered by a @code{setjmp} call.
1894 These warnings are possible only in optimizing compilation,
1895 because they require data flow information that is computed only
1896 when optimizing. If you don't specify @samp{-O}, you simply won't
1899 These warnings occur only for variables that are candidates for
1900 register allocation. Therefore, they do not occur for a variable that
1901 is declared @code{volatile}, or whose address is taken, or whose size
1902 is other than 1, 2, 4 or 8 bytes. Also, they do not occur for
1903 structures, unions or arrays, even when they are in registers.
1905 Note that there may be no warning about a variable that is used only
1906 to compute a value that itself is never used, because such
1907 computations may be deleted by data flow analysis before the warnings
1910 These warnings are made optional because GCC is not smart
1911 enough to see all the reasons why the code might be correct
1912 despite appearing to have an error. Here is one example of how
1931 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
1932 always initialized, but GCC doesn't know this. Here is
1933 another common case:
1938 if (change_y) save_y = y, y = new_y;
1940 if (change_y) y = save_y;
1945 This has no bug because @code{save_y} is used only if it is set.
1947 @cindex @code{longjmp} warnings
1948 This option also warns when a non-volatile automatic variable might be
1949 changed by a call to @code{longjmp}. These warnings as well are possible
1950 only in optimizing compilation.
1952 The compiler sees only the calls to @code{setjmp}. It cannot know
1953 where @code{longjmp} will be called; in fact, a signal handler could
1954 call it at any point in the code. As a result, you may get a warning
1955 even when there is in fact no problem because @code{longjmp} cannot
1956 in fact be called at the place which would cause a problem.
1958 Some spurious warnings can be avoided if you declare all the functions
1959 you use that never return as @code{noreturn}. @xref{Function
1962 @item -Wreorder (C++ only)
1963 @cindex reordering, warning
1964 @cindex warning for reordering of member initializers
1965 Warn when the order of member initializers given in the code does not
1966 match the order in which they must be executed. For instance:
1968 @item -Wunknown-pragmas
1969 @cindex warning for unknown pragmas
1970 @cindex unknown pragmas, warning
1971 @cindex pragmas, warning of unknown
1972 Warn when a #pragma directive is encountered which is not understood by
1973 GCC. If this command line option is used, warnings will even be issued
1974 for unknown pragmas in system header files. This is not the case if
1975 the warnings were only enabled by the @samp{-Wall} command line option.
1978 All of the above @samp{-W} options combined. This enables all the
1979 warnings about constructions that some users consider questionable, and
1980 that are easy to avoid (or modify to prevent the warning), even in
1981 conjunction with macros.
1983 @item -Wsystem-headers
1984 @cindex warnings from system headers
1985 @cindex system headers, warnings from
1986 Print warning messages for constructs found in system header files.
1987 Warnings from system headers are normally suppressed, on the assumption
1988 that they usually do not indicate real problems and would only make the
1989 compiler output harder to read. Using this command line option tells
1990 GCC to emit warnings from system headers as if they occurred in user
1991 code. However, note that using @samp{-Wall} in conjunction with this
1992 option will @emph{not} warn about unknown pragmas in system
1993 headers---for that, @samp{-Wunknown-pragmas} must also be used.
1996 The following @samp{-W@dots{}} options are not implied by @samp{-Wall}.
1997 Some of them warn about constructions that users generally do not
1998 consider questionable, but which occasionally you might wish to check
1999 for; others warn about constructions that are necessary or hard to avoid
2000 in some cases, and there is no simple way to modify the code to suppress
2005 Print extra warning messages for these events:
2009 A function can return either with or without a value. (Falling
2010 off the end of the function body is considered returning without
2011 a value.) For example, this function would evoke such a
2025 An expression-statement or the left-hand side of a comma expression
2026 contains no side effects.
2027 To suppress the warning, cast the unused expression to void.
2028 For example, an expression such as @samp{x[i,j]} will cause a warning,
2029 but @samp{x[(void)i,j]} will not.
2032 An unsigned value is compared against zero with @samp{<} or @samp{<=}.
2035 A comparison like @samp{x<=y<=z} appears; this is equivalent to
2036 @samp{(x<=y ? 1 : 0) <= z}, which is a different interpretation from
2037 that of ordinary mathematical notation.
2040 Storage-class specifiers like @code{static} are not the first things in
2041 a declaration. According to the C Standard, this usage is obsolescent.
2044 The return type of a function has a type qualifier such as @code{const}.
2045 Such a type qualifier has no effect, since the value returned by a
2046 function is not an lvalue. (But don't warn about the GNU extension of
2047 @code{volatile void} return types. That extension will be warned about
2048 if @samp{-pedantic} is specified.)
2051 If @samp{-Wall} or @samp{-Wunused} is also specified, warn about unused
2055 A comparison between signed and unsigned values could produce an
2056 incorrect result when the signed value is converted to unsigned.
2057 (But don't warn if @samp{-Wno-sign-compare} is also specified.)
2060 An aggregate has a partly bracketed initializer.
2061 For example, the following code would evoke such a warning,
2062 because braces are missing around the initializer for @code{x.h}:
2065 struct s @{ int f, g; @};
2066 struct t @{ struct s h; int i; @};
2067 struct t x = @{ 1, 2, 3 @};
2071 An aggregate has an initializer which does not initialize all members.
2072 For example, the following code would cause such a warning, because
2073 @code{x.h} would be implicitly initialized to zero:
2076 struct s @{ int f, g, h; @};
2077 struct s x = @{ 3, 4 @};
2082 Warn if floating point values are used in equality comparisons.
2084 The idea behind this is that sometimes it is convenient (for the
2085 programmer) to consider floating-point values as approximations to
2086 infinitely precise real numbers. If you are doing this, then you need
2087 to compute (by analysing the code, or in some other way) the maximum or
2088 likely maximum error that the computation introduces, and allow for it
2089 when performing comparisons (and when producing output, but that's a
2090 different problem). In particular, instead of testing for equality, you
2091 would check to see whether the two values have ranges that overlap; and
2092 this is done with the relational operators, so equality comparisons are
2095 @item -Wtraditional (C only)
2096 Warn about certain constructs that behave differently in traditional and
2097 ISO C. Also warn about ISO C constructs that have no traditional C
2098 equivalent, and/or problematic constructs which should be avoided.
2102 Macro parameters that appear within string literals in the macro body.
2103 In traditional C macro replacement takes place within string literals,
2104 but does not in ISO C.
2107 In traditional C, some preprocessor directives did not exist.
2108 Traditional preprocessors would only consider a line to be a directive
2109 if the @samp{#} appeared in column 1 on the line. Therefore
2110 @samp{-Wtraditional} warns about directives that traditional C
2111 understands but would ignore because the @samp{#} does not appear as the
2112 first character on the line. It also suggests you hide directives like
2113 @samp{#pragma} not understood by traditional C by indenting them. Some
2114 traditional implementations would not recognise @samp{#elif}, so it
2115 suggests avoiding it altogether.
2118 A function-like macro that appears without arguments.
2121 The unary plus operator.
2124 The `U' integer constant suffix, or the `F' or `L' floating point
2125 constant suffixes. (Traditonal C does support the `L' suffix on integer
2126 constants.) Note, these suffixes appear in macros defined in the system
2127 headers of most modern systems, e.g. the _MIN/_MAX macros in limits.h.
2128 Use of these macros in user code might normally lead to spurious
2129 warnings, however gcc's integrated preprocessor has enough context to
2130 avoid warning in these cases.
2133 A function declared external in one block and then used after the end of
2137 A @code{switch} statement has an operand of type @code{long}.
2140 A non-@code{static} function declaration follows a @code{static} one.
2141 This construct is not accepted by some traditional C compilers.
2144 The ISO type of an integer constant has a different width or
2145 signedness from its traditional type. This warning is only issued if
2146 the base of the constant is ten. I.e. hexadecimal or octal values, which
2147 typically represent bit patterns, are not warned about.
2150 Usage of ISO string concatenation is detected.
2153 Initialization of automatic aggregates.
2156 Identifier conflicts with labels. Traditional C lacks a separate
2157 namespace for labels.
2160 Initialization of unions. If the initializer is zero, the warning is
2161 omitted. This is done under the assumption that the zero initializer in
2162 user code appears conditioned on e.g. @code{__STDC__} to avoid missing
2163 initializer warnings and relies on default initialization to zero in the
2167 Conversions by prototypes between fixed/floating point values and vice
2168 versa. The absence of these prototypes when compiling with traditional
2169 C would cause serious problems. This is a subset of the possible
2170 conversion warnings, for the full set use @samp{-Wconversion}.
2174 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
2177 Warn whenever a local variable shadows another local variable, parameter or
2178 global variable or whenever a built-in function is shadowed.
2180 @item -Wid-clash-@var{len}
2181 Warn whenever two distinct identifiers match in the first @var{len}
2182 characters. This may help you prepare a program that will compile
2183 with certain obsolete, brain-damaged compilers.
2185 @item -Wlarger-than-@var{len}
2186 Warn whenever an object of larger than @var{len} bytes is defined.
2188 @item -Wpointer-arith
2189 Warn about anything that depends on the ``size of'' a function type or
2190 of @code{void}. GNU C assigns these types a size of 1, for
2191 convenience in calculations with @code{void *} pointers and pointers
2194 @item -Wbad-function-cast (C only)
2195 Warn whenever a function call is cast to a non-matching type.
2196 For example, warn if @code{int malloc()} is cast to @code{anything *}.
2199 Warn whenever a pointer is cast so as to remove a type qualifier from
2200 the target type. For example, warn if a @code{const char *} is cast
2201 to an ordinary @code{char *}.
2204 Warn whenever a pointer is cast such that the required alignment of the
2205 target is increased. For example, warn if a @code{char *} is cast to
2206 an @code{int *} on machines where integers can only be accessed at
2207 two- or four-byte boundaries.
2209 @item -Wwrite-strings
2210 Give string constants the type @code{const char[@var{length}]} so that
2211 copying the address of one into a non-@code{const} @code{char *}
2212 pointer will get a warning. These warnings will help you find at
2213 compile time code that can try to write into a string constant, but
2214 only if you have been very careful about using @code{const} in
2215 declarations and prototypes. Otherwise, it will just be a nuisance;
2216 this is why we did not make @samp{-Wall} request these warnings.
2219 Warn if a prototype causes a type conversion that is different from what
2220 would happen to the same argument in the absence of a prototype. This
2221 includes conversions of fixed point to floating and vice versa, and
2222 conversions changing the width or signedness of a fixed point argument
2223 except when the same as the default promotion.
2225 Also, warn if a negative integer constant expression is implicitly
2226 converted to an unsigned type. For example, warn about the assignment
2227 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
2228 casts like @code{(unsigned) -1}.
2230 @item -Wsign-compare
2231 @cindex warning for comparison of signed and unsigned values
2232 @cindex comparison of signed and unsigned values, warning
2233 @cindex signed and unsigned values, comparison warning
2234 Warn when a comparison between signed and unsigned values could produce
2235 an incorrect result when the signed value is converted to unsigned.
2236 This warning is also enabled by @samp{-W}; to get the other warnings
2237 of @samp{-W} without this warning, use @samp{-W -Wno-sign-compare}.
2239 @item -Waggregate-return
2240 Warn if any functions that return structures or unions are defined or
2241 called. (In languages where you can return an array, this also elicits
2244 @item -Wstrict-prototypes (C only)
2245 Warn if a function is declared or defined without specifying the
2246 argument types. (An old-style function definition is permitted without
2247 a warning if preceded by a declaration which specifies the argument
2250 @item -Wmissing-prototypes (C only)
2251 Warn if a global function is defined without a previous prototype
2252 declaration. This warning is issued even if the definition itself
2253 provides a prototype. The aim is to detect global functions that fail
2254 to be declared in header files.
2256 @item -Wmissing-declarations
2257 Warn if a global function is defined without a previous declaration.
2258 Do so even if the definition itself provides a prototype.
2259 Use this option to detect global functions that are not declared in
2262 @item -Wmissing-noreturn
2263 Warn about functions which might be candidates for attribute @code{noreturn}.
2264 Note these are only possible candidates, not absolute ones. Care should
2265 be taken to manually verify functions actually do not ever return before
2266 adding the @code{noreturn} attribute, otherwise subtle code generation
2267 bugs could be introduced. You will not get a warning for @code{main} in
2268 hosted C environments.
2270 @item -Wmissing-format-attribute
2271 If @samp{-Wformat} is enabled, also warn about functions which might be
2272 candidates for @code{format} attributes. Note these are only possible
2273 candidates, not absolute ones. GCC will guess that @code{format}
2274 attributes might be appropriate for any function that calls a function
2275 like @code{vprintf} or @code{vscanf}, but this might not always be the
2276 case, and some functions for which @code{format} attributes are
2277 appropriate may not be detected. This option has no effect unless
2278 @samp{-Wformat} is enabled (possibly by @samp{-Wall}).
2281 Warn if a structure is given the packed attribute, but the packed
2282 attribute has no effect on the layout or size of the structure.
2283 Such structures may be mis-aligned for little benefit. For
2284 instance, in this code, the variable @code{f.x} in @code{struct bar}
2285 will be misaligned even though @code{struct bar} does not itself
2286 have the packed attribute:
2293 @} __attribute__((packed));
2302 Warn if padding is included in a structure, either to align an element
2303 of the structure or to align the whole structure. Sometimes when this
2304 happens it is possible to rearrange the fields of the structure to
2305 reduce the padding and so make the structure smaller.
2307 @item -Wredundant-decls
2308 Warn if anything is declared more than once in the same scope, even in
2309 cases where multiple declaration is valid and changes nothing.
2311 @item -Wnested-externs (C only)
2312 Warn if an @code{extern} declaration is encountered within a function.
2314 @item -Wunreachable-code
2315 Warn if the compiler detects that code will never be executed.
2317 This option is intended to warn when the compiler detects that at
2318 least a whole line of source code will never be executed, because
2319 some condition is never satisfied or because it is after a
2320 procedure that never returns.
2322 It is possible for this option to produce a warning even though there
2323 are circumstances under which part of the affected line can be executed,
2324 so care should be taken when removing apparently-unreachable code.
2326 For instance, when a function is inlined, a warning may mean that the
2327 line is unreachable in only one inlined copy of the function.
2329 This option is not made part of @samp{-Wall} because in a debugging
2330 version of a program there is often substantial code which checks
2331 correct functioning of the program and is, hopefully, unreachable
2332 because the program does work. Another common use of unreachable
2333 code is to provide behaviour which is selectable at compile-time.
2336 Warn if a function can not be inlined and it was declared as inline.
2339 Warn if @samp{long long} type is used. This is default. To inhibit
2340 the warning messages, use @samp{-Wno-long-long}. Flags
2341 @samp{-Wlong-long} and @samp{-Wno-long-long} are taken into account
2342 only when @samp{-pedantic} flag is used.
2344 @item -Wdisabled-optimization
2345 Warn if a requested optimization pass is disabled. This warning does
2346 not generally indicate that there is anything wrong with your code; it
2347 merely indicates that GCC's optimizers were unable to handle the code
2348 effectively. Often, the problem is that your code is too big or too
2349 complex; GCC will refuse to optimize programs when the optimization
2350 itself is likely to take inordinate amounts of time.
2353 Make all warnings into errors.
2356 @node Debugging Options
2357 @section Options for Debugging Your Program or GCC
2358 @cindex options, debugging
2359 @cindex debugging information options
2361 GCC has various special options that are used for debugging
2362 either your program or GCC:
2366 Produce debugging information in the operating system's native format
2367 (stabs, COFF, XCOFF, or DWARF). GDB can work with this debugging
2370 On most systems that use stabs format, @samp{-g} enables use of extra
2371 debugging information that only GDB can use; this extra information
2372 makes debugging work better in GDB but will probably make other debuggers
2374 refuse to read the program. If you want to control for certain whether
2375 to generate the extra information, use @samp{-gstabs+}, @samp{-gstabs},
2376 @samp{-gxcoff+}, @samp{-gxcoff}, @samp{-gdwarf-1+}, or @samp{-gdwarf-1}
2379 Unlike most other C compilers, GCC allows you to use @samp{-g} with
2380 @samp{-O}. The shortcuts taken by optimized code may occasionally
2381 produce surprising results: some variables you declared may not exist
2382 at all; flow of control may briefly move where you did not expect it;
2383 some statements may not be executed because they compute constant
2384 results or their values were already at hand; some statements may
2385 execute in different places because they were moved out of loops.
2387 Nevertheless it proves possible to debug optimized output. This makes
2388 it reasonable to use the optimizer for programs that might have bugs.
2390 The following options are useful when GCC is generated with the
2391 capability for more than one debugging format.
2394 Produce debugging information for use by GDB. This means to use the
2395 most expressive format available (DWARF 2, stabs, or the native format
2396 if neither of those are supported), including GDB extensions if at all
2400 Produce debugging information in stabs format (if that is supported),
2401 without GDB extensions. This is the format used by DBX on most BSD
2402 systems. On MIPS, Alpha and System V Release 4 systems this option
2403 produces stabs debugging output which is not understood by DBX or SDB.
2404 On System V Release 4 systems this option requires the GNU assembler.
2407 Produce debugging information in stabs format (if that is supported),
2408 using GNU extensions understood only by the GNU debugger (GDB). The
2409 use of these extensions is likely to make other debuggers crash or
2410 refuse to read the program.
2413 Produce debugging information in COFF format (if that is supported).
2414 This is the format used by SDB on most System V systems prior to
2418 Produce debugging information in XCOFF format (if that is supported).
2419 This is the format used by the DBX debugger on IBM RS/6000 systems.
2422 Produce debugging information in XCOFF format (if that is supported),
2423 using GNU extensions understood only by the GNU debugger (GDB). The
2424 use of these extensions is likely to make other debuggers crash or
2425 refuse to read the program, and may cause assemblers other than the GNU
2426 assembler (GAS) to fail with an error.
2429 Produce debugging information in DWARF version 1 format (if that is
2430 supported). This is the format used by SDB on most System V Release 4
2434 Produce debugging information in DWARF version 1 format (if that is
2435 supported), using GNU extensions understood only by the GNU debugger
2436 (GDB). The use of these extensions is likely to make other debuggers
2437 crash or refuse to read the program.
2440 Produce debugging information in DWARF version 2 format (if that is
2441 supported). This is the format used by DBX on IRIX 6.
2444 @itemx -ggdb@var{level}
2445 @itemx -gstabs@var{level}
2446 @itemx -gcoff@var{level}
2447 @itemx -gxcoff@var{level}
2448 @itemx -gdwarf@var{level}
2449 @itemx -gdwarf-2@var{level}
2450 Request debugging information and also use @var{level} to specify how
2451 much information. The default level is 2.
2453 Level 1 produces minimal information, enough for making backtraces in
2454 parts of the program that you don't plan to debug. This includes
2455 descriptions of functions and external variables, but no information
2456 about local variables and no line numbers.
2458 Level 3 includes extra information, such as all the macro definitions
2459 present in the program. Some debuggers support macro expansion when
2464 Generate extra code to write profile information suitable for the
2465 analysis program @code{prof}. You must use this option when compiling
2466 the source files you want data about, and you must also use it when
2469 @cindex @code{gprof}
2471 Generate extra code to write profile information suitable for the
2472 analysis program @code{gprof}. You must use this option when compiling
2473 the source files you want data about, and you must also use it when
2478 Generate extra code to write profile information for basic blocks, which will
2479 record the number of times each basic block is executed, the basic block start
2480 address, and the function name containing the basic block. If @samp{-g} is
2481 used, the line number and filename of the start of the basic block will also be
2482 recorded. If not overridden by the machine description, the default action is
2483 to append to the text file @file{bb.out}.
2485 This data could be analyzed by a program like @code{tcov}. Note,
2486 however, that the format of the data is not what @code{tcov} expects.
2487 Eventually GNU @code{gprof} should be extended to process this data.
2490 Makes the compiler print out each function name as it is compiled, and
2491 print some statistics about each pass when it finishes.
2494 Makes the compiler print some statistics about the time consumed by each
2495 pass when it finishes.
2498 Makes the compiler print some statistics about permanent memory
2499 allocation when it finishes.
2502 Generate extra code to profile basic blocks. Your executable will
2503 produce output that is a superset of that produced when @samp{-a} is
2504 used. Additional output is the source and target address of the basic
2505 blocks where a jump takes place, the number of times a jump is executed,
2506 and (optionally) the complete sequence of basic blocks being executed.
2507 The output is appended to file @file{bb.out}.
2509 You can examine different profiling aspects without recompilation. Your
2510 executable will read a list of function names from file @file{bb.in}.
2511 Profiling starts when a function on the list is entered and stops when
2512 that invocation is exited. To exclude a function from profiling, prefix
2513 its name with `-'. If a function name is not unique, you can
2514 disambiguate it by writing it in the form
2515 @samp{/path/filename.d:functionname}. Your executable will write the
2516 available paths and filenames in file @file{bb.out}.
2518 Several function names have a special meaning:
2521 Write source, target and frequency of jumps to file @file{bb.out}.
2522 @item __bb_hidecall__
2523 Exclude function calls from frequency count.
2524 @item __bb_showret__
2525 Include function returns in frequency count.
2527 Write the sequence of basic blocks executed to file @file{bbtrace.gz}.
2528 The file will be compressed using the program @samp{gzip}, which must
2529 exist in your @env{PATH}. On systems without the @samp{popen}
2530 function, the file will be named @file{bbtrace} and will not be
2531 compressed. @strong{Profiling for even a few seconds on these systems
2532 will produce a very large file.} Note: @code{__bb_hidecall__} and
2533 @code{__bb_showret__} will not affect the sequence written to
2537 Here's a short example using different profiling parameters
2538 in file @file{bb.in}. Assume function @code{foo} consists of basic blocks
2539 1 and 2 and is called twice from block 3 of function @code{main}. After
2540 the calls, block 3 transfers control to block 4 of @code{main}.
2542 With @code{__bb_trace__} and @code{main} contained in file @file{bb.in},
2543 the following sequence of blocks is written to file @file{bbtrace.gz}:
2544 0 3 1 2 1 2 4. The return from block 2 to block 3 is not shown, because
2545 the return is to a point inside the block and not to the top. The
2546 block address 0 always indicates, that control is transferred
2547 to the trace from somewhere outside the observed functions. With
2548 @samp{-foo} added to @file{bb.in}, the blocks of function
2549 @code{foo} are removed from the trace, so only 0 3 4 remains.
2551 With @code{__bb_jumps__} and @code{main} contained in file @file{bb.in},
2552 jump frequencies will be written to file @file{bb.out}. The
2553 frequencies are obtained by constructing a trace of blocks
2554 and incrementing a counter for every neighbouring pair of blocks
2555 in the trace. The trace 0 3 1 2 1 2 4 displays the following
2559 Jump from block 0x0 to block 0x3 executed 1 time(s)
2560 Jump from block 0x3 to block 0x1 executed 1 time(s)
2561 Jump from block 0x1 to block 0x2 executed 2 time(s)
2562 Jump from block 0x2 to block 0x1 executed 1 time(s)
2563 Jump from block 0x2 to block 0x4 executed 1 time(s)
2566 With @code{__bb_hidecall__}, control transfer due to call instructions
2567 is removed from the trace, that is the trace is cut into three parts: 0
2568 3 4, 0 1 2 and 0 1 2. With @code{__bb_showret__}, control transfer due
2569 to return instructions is added to the trace. The trace becomes: 0 3 1
2570 2 3 1 2 3 4. Note, that this trace is not the same, as the sequence
2571 written to @file{bbtrace.gz}. It is solely used for counting jump
2574 @item -fprofile-arcs
2575 Instrument @dfn{arcs} during compilation. For each function of your
2576 program, GCC creates a program flow graph, then finds a spanning tree
2577 for the graph. Only arcs that are not on the spanning tree have to be
2578 instrumented: the compiler adds code to count the number of times that these
2579 arcs are executed. When an arc is the only exit or only entrance to a
2580 block, the instrumentation code can be added to the block; otherwise, a
2581 new basic block must be created to hold the instrumentation code.
2583 Since not every arc in the program must be instrumented, programs
2584 compiled with this option run faster than programs compiled with
2585 @samp{-a}, which adds instrumentation code to every basic block in the
2586 program. The tradeoff: since @code{gcov} does not have
2587 execution counts for all branches, it must start with the execution
2588 counts for the instrumented branches, and then iterate over the program
2589 flow graph until the entire graph has been solved. Hence, @code{gcov}
2590 runs a little more slowly than a program which uses information from
2593 @samp{-fprofile-arcs} also makes it possible to estimate branch
2594 probabilities, and to calculate basic block execution counts. In
2595 general, basic block execution counts do not give enough information to
2596 estimate all branch probabilities. When the compiled program exits, it
2597 saves the arc execution counts to a file called
2598 @file{@var{sourcename}.da}. Use the compiler option
2599 @samp{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
2600 Control Optimization}) when recompiling, to optimize using estimated
2601 branch probabilities.
2604 @item -ftest-coverage
2605 Create data files for the @code{gcov} code-coverage utility
2606 (@pxref{Gcov,, @code{gcov}: a GCC Test Coverage Program}).
2607 The data file names begin with the name of your source file:
2610 @item @var{sourcename}.bb
2611 A mapping from basic blocks to line numbers, which @code{gcov} uses to
2612 associate basic block execution counts with line numbers.
2614 @item @var{sourcename}.bbg
2615 A list of all arcs in the program flow graph. This allows @code{gcov}
2616 to reconstruct the program flow graph, so that it can compute all basic
2617 block and arc execution counts from the information in the
2618 @code{@var{sourcename}.da} file (this last file is the output from
2619 @samp{-fprofile-arcs}).
2622 @item -d@var{letters}
2623 Says to make debugging dumps during compilation at times specified by
2624 @var{letters}. This is used for debugging the compiler. The file names
2625 for most of the dumps are made by appending a pass number and a word to
2626 the source file name (e.g. @file{foo.c.00.rtl} or @file{foo.c.01.sibling}).
2627 Here are the possible letters for use in @var{letters}, and their meanings:
2631 Annotate the assembler output with miscellaneous debugging information.
2633 Dump after computing branch probabilities, to @file{@var{file}.11.bp}.
2635 Dump after block reordering, to @file{@var{file}.26.bbro}.
2637 Dump after instruction combination, to the file @file{@var{file}.14.combine}.
2639 Dump after the first if conversion, to the file @file{@var{file}.15.ce}.
2641 Dump after delayed branch scheduling, to @file{@var{file}.29.dbr}.
2643 Dump all macro definitions, at the end of preprocessing, in addition to
2646 Dump after SSA optimizations, to @file{@var{file}.05.ssa} and
2647 @file{@var{file}.06.ussa}.
2649 Dump after the second if conversion, to @file{@var{file}.24.ce2}.
2651 Dump after life analysis, to @file{@var{file}.13.life}.
2653 Dump after purging @code{ADDRESSOF} codes, to @file{@var{file}.04.addressof}.
2655 Dump after global register allocation, to @file{@var{file}.19.greg}.
2657 Dump after post-reload CSE and other optimizations, to @file{@var{file}.20.postreload}.
2659 Dump after GCSE, to @file{@var{file}.08.gcse}.
2661 Dump after sibling call optimizations, to @file{@var{file}.01.sibling}.
2663 Dump after the first jump optimization, to @file{@var{file}.02.jump}.
2665 Dump after the last jump optimization, to @file{@var{file}.27.jump2}.
2667 Dump after conversion from registers to stack, to @file{@var{file}.29.stack}.
2669 Dump after local register allocation, to @file{@var{file}.18.lreg}.
2671 Dump after loop optimization, to @file{@var{file}.09.loop}.
2673 Dump after performing the machine dependent reorganisation pass, to
2674 @file{@var{file}.28.mach}.
2676 Dump after register renumbering, to @file{@var{file}.23.rnreg}.
2678 Dump after the register move pass, to @file{@var{file}.16.regmove}.
2680 Dump after RTL generation, to @file{@var{file}.00.rtl}.
2682 Dump after the second instruction scheduling pass, to
2683 @file{@var{file}.25.sched2}.
2685 Dump after CSE (including the jump optimization that sometimes follows
2686 CSE), to @file{@var{file}.03.cse}.
2688 Dump after the first instruction scheduling pass, to
2689 @file{@var{file}.17.sched}.
2691 Dump after the second CSE pass (including the jump optimization that
2692 sometimes follows CSE), to @file{@var{file}.10.cse2}.
2694 Dump after the second flow pass, to @file{@var{file}.21.flow2}.
2696 Dump after dead code elimination, to @file{@var{file}.06.dce}.
2698 Dump after the peephole pass, to @file{@var{file}.22.peephole2}.
2700 Produce all the dumps listed above.
2702 Print statistics on memory usage, at the end of the run, to
2705 Annotate the assembler output with a comment indicating which
2706 pattern and alternative was used. The length of each instruction is
2709 Dump the RTL in the assembler output as a comment before each instruction.
2710 Also turns on @samp{-dp} annotation.
2712 For each of the other indicated dump files (except for
2713 @file{@var{file}.00.rtl}), dump a representation of the control flow graph
2714 suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
2716 Just generate RTL for a function instead of compiling it. Usually used
2719 Dump debugging information during parsing, to standard error.
2722 @item -fdump-unnumbered
2723 When doing debugging dumps (see -d option above), suppress instruction
2724 numbers and line number note output. This makes it more feasible to
2725 use diff on debugging dumps for compiler invocations with different
2726 options, in particular with and without -g.
2728 @item -fdump-translation-unit=@var{file} (C and C++ only)
2729 Dump a representation of the tree structure for the entire translation
2732 @item -fdump-class_layout=@var{file} (C++ only)
2733 @item -fdump-class_layout (C++ only)
2734 Dump a representation of each class's heirarchy to @var{file}, or
2735 @code{stderr} if not specified.
2737 @item -fpretend-float
2738 When running a cross-compiler, pretend that the target machine uses the
2739 same floating point format as the host machine. This causes incorrect
2740 output of the actual floating constants, but the actual instruction
2741 sequence will probably be the same as GCC would make when running on
2745 Store the usual ``temporary'' intermediate files permanently; place them
2746 in the current directory and name them based on the source file. Thus,
2747 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
2748 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
2749 preprocessed @file{foo.i} output file even though the compiler now
2750 normally uses an integrated preprocessor.
2753 Report the CPU time taken by each subprocess in the compilation
2754 sequence. For C source files, this is the compiler proper and assembler
2755 (plus the linker if linking is done). The output looks like this:
2762 The first number on each line is the ``user time,'' that is time spent
2763 executing the program itself. The second number is ``system time,''
2764 time spent executing operating system routines on behalf of the program.
2765 Both numbers are in seconds.
2767 @item -print-file-name=@var{library}
2768 Print the full absolute name of the library file @var{library} that
2769 would be used when linking---and don't do anything else. With this
2770 option, GCC does not compile or link anything; it just prints the
2773 @item -print-prog-name=@var{program}
2774 Like @samp{-print-file-name}, but searches for a program such as @samp{cpp}.
2776 @item -print-libgcc-file-name
2777 Same as @samp{-print-file-name=libgcc.a}.
2779 This is useful when you use @samp{-nostdlib} or @samp{-nodefaultlibs}
2780 but you do want to link with @file{libgcc.a}. You can do
2783 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
2786 @item -print-search-dirs
2787 Print the name of the configured installation directory and a list of
2788 program and library directories gcc will search---and don't do anything else.
2790 This is useful when gcc prints the error message
2791 @samp{installation problem, cannot exec cpp0: No such file or directory}.
2792 To resolve this you either need to put @file{cpp0} and the other compiler
2793 components where gcc expects to find them, or you can set the environment
2794 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
2795 Don't forget the trailing '/'.
2796 @xref{Environment Variables}.
2799 Print the compiler's target machine (for example,
2800 @samp{i686-pc-linux-gnu})---and don't do anything else.
2803 Print the compiler version (for example, @samp{3.0})---and don't do
2807 Print the compiler's built-in specs---and don't do anything else. (This
2808 is used when GCC itself is being built.) @xref{Spec Files}.
2811 @node Optimize Options
2812 @section Options That Control Optimization
2813 @cindex optimize options
2814 @cindex options, optimization
2816 These options control various sorts of optimizations:
2821 Optimize. Optimizing compilation takes somewhat more time, and a lot
2822 more memory for a large function.
2824 Without @samp{-O}, the compiler's goal is to reduce the cost of
2825 compilation and to make debugging produce the expected results.
2826 Statements are independent: if you stop the program with a breakpoint
2827 between statements, you can then assign a new value to any variable or
2828 change the program counter to any other statement in the function and
2829 get exactly the results you would expect from the source code.
2831 Without @samp{-O}, the compiler only allocates variables declared
2832 @code{register} in registers. The resulting compiled code is a little
2833 worse than produced by PCC without @samp{-O}.
2835 With @samp{-O}, the compiler tries to reduce code size and execution
2838 When you specify @samp{-O}, the compiler turns on @samp{-fthread-jumps}
2839 and @samp{-fdefer-pop} on all machines. The compiler turns on
2840 @samp{-fdelayed-branch} on machines that have delay slots, and
2841 @samp{-fomit-frame-pointer} on machines that can support debugging even
2842 without a frame pointer. On some machines the compiler also turns
2843 on other flags.@refill
2846 Optimize even more. GCC performs nearly all supported optimizations
2847 that do not involve a space-speed tradeoff. The compiler does not
2848 perform loop unrolling or function inlining when you specify @samp{-O2}.
2849 As compared to @samp{-O}, this option increases both compilation time
2850 and the performance of the generated code.
2852 @samp{-O2} turns on all optional optimizations except for loop unrolling,
2853 function inlining, and register renaming. It also turns on the
2854 @samp{-fforce-mem} option on all machines and frame pointer elimination
2855 on machines where doing so does not interfere with debugging.
2858 Optimize yet more. @samp{-O3} turns on all optimizations specified by
2859 @samp{-O2} and also turns on the @samp{-finline-functions} and
2860 @samp{-frename-registers} options.
2866 Optimize for size. @samp{-Os} enables all @samp{-O2} optimizations that
2867 do not typically increase code size. It also performs further
2868 optimizations designed to reduce code size.
2870 If you use multiple @samp{-O} options, with or without level numbers,
2871 the last such option is the one that is effective.
2874 Options of the form @samp{-f@var{flag}} specify machine-independent
2875 flags. Most flags have both positive and negative forms; the negative
2876 form of @samp{-ffoo} would be @samp{-fno-foo}. In the table below,
2877 only one of the forms is listed---the one which is not the default.
2878 You can figure out the other form by either removing @samp{no-} or
2883 Do not store floating point variables in registers, and inhibit other
2884 options that might change whether a floating point value is taken from a
2887 @cindex floating point precision
2888 This option prevents undesirable excess precision on machines such as
2889 the 68000 where the floating registers (of the 68881) keep more
2890 precision than a @code{double} is supposed to have. Similarly for the
2891 x86 architecture. For most programs, the excess precision does only
2892 good, but a few programs rely on the precise definition of IEEE floating
2893 point. Use @samp{-ffloat-store} for such programs, after modifying
2894 them to store all pertinent intermediate computations into variables.
2896 @item -fno-default-inline
2897 Do not make member functions inline by default merely because they are
2898 defined inside the class scope (C++ only). Otherwise, when you specify
2899 @w{@samp{-O}}, member functions defined inside class scope are compiled
2900 inline by default; i.e., you don't need to add @samp{inline} in front of
2901 the member function name.
2903 @item -fno-defer-pop
2904 Always pop the arguments to each function call as soon as that function
2905 returns. For machines which must pop arguments after a function call,
2906 the compiler normally lets arguments accumulate on the stack for several
2907 function calls and pops them all at once.
2910 Force memory operands to be copied into registers before doing
2911 arithmetic on them. This produces better code by making all memory
2912 references potential common subexpressions. When they are not common
2913 subexpressions, instruction combination should eliminate the separate
2914 register-load. The @samp{-O2} option turns on this option.
2917 Force memory address constants to be copied into registers before
2918 doing arithmetic on them. This may produce better code just as
2919 @samp{-fforce-mem} may.
2921 @item -fomit-frame-pointer
2922 Don't keep the frame pointer in a register for functions that
2923 don't need one. This avoids the instructions to save, set up and
2924 restore frame pointers; it also makes an extra register available
2925 in many functions. @strong{It also makes debugging impossible on
2929 On some machines, such as the Vax, this flag has no effect, because
2930 the standard calling sequence automatically handles the frame pointer
2931 and nothing is saved by pretending it doesn't exist. The
2932 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
2933 whether a target machine supports this flag. @xref{Registers}.@refill
2936 On some machines, such as the Vax, this flag has no effect, because
2937 the standard calling sequence automatically handles the frame pointer
2938 and nothing is saved by pretending it doesn't exist. The
2939 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
2940 whether a target machine supports this flag. @xref{Registers,,Register
2941 Usage, gcc.info, Using and Porting GCC}.@refill
2944 @item -foptimize-sibling-calls
2945 Optimize sibling and tail recursive calls.
2948 This option generates traps for signed overflow on addition, subtraction,
2949 multiplication operations.
2952 Don't pay attention to the @code{inline} keyword. Normally this option
2953 is used to keep the compiler from expanding any functions inline.
2954 Note that if you are not optimizing, no functions can be expanded inline.
2956 @item -finline-functions
2957 Integrate all simple functions into their callers. The compiler
2958 heuristically decides which functions are simple enough to be worth
2959 integrating in this way.
2961 If all calls to a given function are integrated, and the function is
2962 declared @code{static}, then the function is normally not output as
2963 assembler code in its own right.
2965 @item -finline-limit=@var{n}
2966 By default, gcc limits the size of functions that can be inlined. This flag
2967 allows the control of this limit for functions that are explicitly marked as
2968 inline (ie marked with the inline keyword or defined within the class
2969 definition in c++). @var{n} is the size of functions that can be inlined in
2970 number of pseudo instructions (not counting parameter handling). The default
2971 value of n is 10000. Increasing this value can result in more inlined code at
2972 the cost of compilation time and memory consumption. Decreasing usually makes
2973 the compilation faster and less code will be inlined (which presumably
2974 means slower programs). This option is particularly useful for programs that
2975 use inlining heavily such as those based on recursive templates with c++.
2977 @emph{Note:} pseudo instruction represents, in this particular context, an
2978 abstract measurement of function's size. In no way, it represents a count
2979 of assembly instructions and as such its exact meaning might change from one
2980 release to an another.
2982 @item -fkeep-inline-functions
2983 Even if all calls to a given function are integrated, and the function
2984 is declared @code{static}, nevertheless output a separate run-time
2985 callable version of the function. This switch does not affect
2986 @code{extern inline} functions.
2988 @item -fkeep-static-consts
2989 Emit variables declared @code{static const} when optimization isn't turned
2990 on, even if the variables aren't referenced.
2992 GCC enables this option by default. If you want to force the compiler to
2993 check if the variable was referenced, regardless of whether or not
2994 optimization is turned on, use the @samp{-fno-keep-static-consts} option.
2996 @item -fno-function-cse
2997 Do not put function addresses in registers; make each instruction that
2998 calls a constant function contain the function's address explicitly.
3000 This option results in less efficient code, but some strange hacks
3001 that alter the assembler output may be confused by the optimizations
3002 performed when this option is not used.
3005 Sets @samp{-fno-math-errno}, @samp{-funsafe-math-optimizations},
3006 and @samp{-fno-trapping-math}.
3008 This option causes the preprocessor macro __FAST_MATH__ to be defined.
3010 This option should never be turned on by any @samp{-O} option since
3011 it can result in incorrect output for programs which depend on
3012 an exact implementation of IEEE or ISO rules/specifications for
3015 @item -fno-math-errno
3016 Do not set ERRNO after calling math functions that are executed
3017 with a single instruction, e.g., sqrt. A program that relies on
3018 IEEE exceptions for math error handling may want to use this flag
3019 for speed while maintaining IEEE arithmetic compatibility.
3021 This option should never be turned on by any @samp{-O} option since
3022 it can result in incorrect output for programs which depend on
3023 an exact implementation of IEEE or ISO rules/specifications for
3026 The default is @samp{-fmath-errno}. The @samp{-ffast-math} option
3027 sets @samp{-fno-math-errno}.
3029 @item -funsafe-math-optimizations
3030 Allow optimizations for floating-point arithmetic that (a) assume
3031 that arguments and results are valid and (b) may violate IEEE or
3034 This option should never be turned on by any @samp{-O} option since
3035 it can result in incorrect output for programs which depend on
3036 an exact implementation of IEEE or ISO rules/specifications for
3039 The default is @samp{-fno-unsafe-math-optimizations}. The
3040 @samp{-ffast-math} option sets @samp{-funsafe-math-optimizations}.
3042 @item -fno-trapping-math
3043 Compile code assuming that floating-point operations cannot generate
3044 user-visible traps. Setting this option may allow faster code
3045 if one relies on ``non-stop'' IEEE arithmetic, for example.
3047 This option should never be turned on by any @samp{-O} option since
3048 it can result in incorrect output for programs which depend on
3049 an exact implementation of IEEE or ISO rules/specifications for
3052 The default is @samp{-ftrapping-math}. The @samp{-ffast-math}
3053 option sets @samp{-fno-trapping-math}.
3056 @c following causes underfulls.. they don't look great, but we deal.
3058 The following options control specific optimizations. The @samp{-O2}
3059 option turns on all of these optimizations except @samp{-funroll-loops}
3060 and @samp{-funroll-all-loops}. On most machines, the @samp{-O} option
3061 turns on the @samp{-fthread-jumps} and @samp{-fdelayed-branch} options,
3062 but specific machines may handle it differently.
3064 You can use the following flags in the rare cases when ``fine-tuning''
3065 of optimizations to be performed is desired.
3068 @item -fstrength-reduce
3069 Perform the optimizations of loop strength reduction and
3070 elimination of iteration variables.
3072 @item -fthread-jumps
3073 Perform optimizations where we check to see if a jump branches to a
3074 location where another comparison subsumed by the first is found. If
3075 so, the first branch is redirected to either the destination of the
3076 second branch or a point immediately following it, depending on whether
3077 the condition is known to be true or false.
3079 @item -fcse-follow-jumps
3080 In common subexpression elimination, scan through jump instructions
3081 when the target of the jump is not reached by any other path. For
3082 example, when CSE encounters an @code{if} statement with an
3083 @code{else} clause, CSE will follow the jump when the condition
3086 @item -fcse-skip-blocks
3087 This is similar to @samp{-fcse-follow-jumps}, but causes CSE to
3088 follow jumps which conditionally skip over blocks. When CSE
3089 encounters a simple @code{if} statement with no else clause,
3090 @samp{-fcse-skip-blocks} causes CSE to follow the jump around the
3091 body of the @code{if}.
3093 @item -frerun-cse-after-loop
3094 Re-run common subexpression elimination after loop optimizations has been
3097 @item -frerun-loop-opt
3098 Run the loop optimizer twice.
3101 Perform a global common subexpression elimination pass.
3102 This pass also performs global constant and copy propagation.
3105 When -fgcse-lm is enabled, global common subexpression elimination will
3106 attempt to move loads which are only killed by stores into themselves. This
3107 allows a loop containing a load/store sequence to be changed to a load outside
3108 the loop, and a copy/store within the loop.
3111 When -fgcse-sm is enabled, A store motion pass is run after global common
3112 subexpression elimination. This pass will attempt to move stores out of loops.
3113 When used in conjunction with -fgcse-lm, loops containing a load/store sequence
3114 can be changed to a load before the loop and a store after the loop.
3116 @item -fdelete-null-pointer-checks
3117 Use global dataflow analysis to identify and eliminate useless null
3118 pointer checks. Programs which rely on NULL pointer dereferences @emph{not}
3119 halting the program may not work properly with this option. Use
3120 -fno-delete-null-pointer-checks to disable this optimizing for programs
3121 which depend on that behavior.
3123 @item -fexpensive-optimizations
3124 Perform a number of minor optimizations that are relatively expensive.
3126 @item -foptimize-register-move
3128 Attempt to reassign register numbers in move instructions and as
3129 operands of other simple instructions in order to maximize the amount of
3130 register tying. This is especially helpful on machines with two-operand
3131 instructions. GCC enables this optimization by default with @samp{-O2}
3134 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
3137 @item -fdelayed-branch
3138 If supported for the target machine, attempt to reorder instructions
3139 to exploit instruction slots available after delayed branch
3142 @item -fschedule-insns
3143 If supported for the target machine, attempt to reorder instructions to
3144 eliminate execution stalls due to required data being unavailable. This
3145 helps machines that have slow floating point or memory load instructions
3146 by allowing other instructions to be issued until the result of the load
3147 or floating point instruction is required.
3149 @item -fschedule-insns2
3150 Similar to @samp{-fschedule-insns}, but requests an additional pass of
3151 instruction scheduling after register allocation has been done. This is
3152 especially useful on machines with a relatively small number of
3153 registers and where memory load instructions take more than one cycle.
3155 @item -ffunction-sections
3156 @itemx -fdata-sections
3157 Place each function or data item into its own section in the output
3158 file if the target supports arbitrary sections. The name of the
3159 function or the name of the data item determines the section's name
3162 Use these options on systems where the linker can perform optimizations
3163 to improve locality of reference in the instruction space. HPPA
3164 processors running HP-UX and Sparc processors running Solaris 2 have
3165 linkers with such optimizations. Other systems using the ELF object format
3166 as well as AIX may have these optimizations in the future.
3168 Only use these options when there are significant benefits from doing
3169 so. When you specify these options, the assembler and linker will
3170 create larger object and executable files and will also be slower.
3171 You will not be able to use @code{gprof} on all systems if you
3172 specify this option and you may have problems with debugging if
3173 you specify both this option and @samp{-g}.
3175 @item -fcaller-saves
3176 Enable values to be allocated in registers that will be clobbered by
3177 function calls, by emitting extra instructions to save and restore the
3178 registers around such calls. Such allocation is done only when it
3179 seems to result in better code than would otherwise be produced.
3181 This option is always enabled by default on certain machines, usually
3182 those which have no call-preserved registers to use instead.
3184 For all machines, optimization level 2 and higher enables this flag by
3187 @item -funroll-loops
3188 Perform the optimization of loop unrolling. This is only done for loops
3189 whose number of iterations can be determined at compile time or run time.
3190 @samp{-funroll-loops} implies both @samp{-fstrength-reduce} and
3191 @samp{-frerun-cse-after-loop}.
3193 @item -funroll-all-loops
3194 Perform the optimization of loop unrolling. This is done for all loops
3195 and usually makes programs run more slowly. @samp{-funroll-all-loops}
3196 implies @samp{-fstrength-reduce} as well as @samp{-frerun-cse-after-loop}.
3198 @item -fmove-all-movables
3199 Forces all invariant computations in loops to be moved
3202 @item -freduce-all-givs
3203 Forces all general-induction variables in loops to be
3206 @emph{Note:} When compiling programs written in Fortran,
3207 @samp{-fmove-all-movables} and @samp{-freduce-all-givs} are enabled
3208 by default when you use the optimizer.
3210 These options may generate better or worse code; results are highly
3211 dependent on the structure of loops within the source code.
3213 These two options are intended to be removed someday, once
3214 they have helped determine the efficacy of various
3215 approaches to improving loop optimizations.
3217 Please let us (@w{@email{gcc@@gcc.gnu.org}} and @w{@email{fortran@@gnu.org}})
3218 know how use of these options affects
3219 the performance of your production code.
3220 We're very interested in code that runs @emph{slower}
3221 when these options are @emph{enabled}.
3224 Disable any machine-specific peephole optimizations.
3226 @item -fbranch-probabilities
3227 After running a program compiled with @samp{-fprofile-arcs}
3228 (@pxref{Debugging Options,, Options for Debugging Your Program or
3229 @command{gcc}}), you can compile it a second time using
3230 @samp{-fbranch-probabilities}, to improve optimizations based on
3231 guessing the path a branch might take.
3234 With @samp{-fbranch-probabilities}, GCC puts a @samp{REG_EXEC_COUNT}
3235 note on the first instruction of each basic block, and a
3236 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
3237 These can be used to improve optimization. Currently, they are only
3238 used in one place: in @file{reorg.c}, instead of guessing which path a
3239 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
3240 exactly determine which path is taken more often.
3243 @item -fno-guess-branch-probability
3244 Sometimes gcc will opt to guess branch probabilities when none are
3245 available from either profile directed feedback (@samp{-fprofile-arcs})
3246 or @samp{__builtin_expect}. In a hard real-time system, people don't
3247 want different runs of the compiler to produce code that has different
3248 behavior; minimizing non-determinism is of paramount import. This
3249 switch allows users to reduce non-determinism, possibly at the expense
3250 of inferior optimization.
3252 @item -fstrict-aliasing
3253 Allows the compiler to assume the strictest aliasing rules applicable to
3254 the language being compiled. For C (and C++), this activates
3255 optimizations based on the type of expressions. In particular, an
3256 object of one type is assumed never to reside at the same address as an
3257 object of a different type, unless the types are almost the same. For
3258 example, an @code{unsigned int} can alias an @code{int}, but not a
3259 @code{void*} or a @code{double}. A character type may alias any other
3262 Pay special attention to code like this:
3275 The practice of reading from a different union member than the one most
3276 recently written to (called ``type-punning'') is common. Even with
3277 @samp{-fstrict-aliasing}, type-punning is allowed, provided the memory
3278 is accessed through the union type. So, the code above will work as
3279 expected. However, this code might not:
3291 Every language that wishes to perform language-specific alias analysis
3292 should define a function that computes, given an @code{tree}
3293 node, an alias set for the node. Nodes in different alias sets are not
3294 allowed to alias. For an example, see the C front-end function
3295 @code{c_get_alias_set}.
3298 @item -falign-functions
3299 @itemx -falign-functions=@var{n}
3300 Align the start of functions to the next power-of-two greater than
3301 @var{n}, skipping up to @var{n} bytes. For instance,
3302 @samp{-falign-functions=32} aligns functions to the next 32-byte
3303 boundary, but @samp{-falign-functions=24} would align to the next
3304 32-byte boundary only if this can be done by skipping 23 bytes or less.
3306 @samp{-fno-align-functions} and @samp{-falign-functions=1} are
3307 equivalent and mean that functions will not be aligned.
3309 Some assemblers only support this flag when @var{n} is a power of two;
3310 in that case, it is rounded up.
3312 If @var{n} is not specified, use a machine-dependent default.
3314 @item -falign-labels
3315 @itemx -falign-labels=@var{n}
3316 Align all branch targets to a power-of-two boundary, skipping up to
3317 @var{n} bytes like @samp{-falign-functions}. This option can easily
3318 make code slower, because it must insert dummy operations for when the
3319 branch target is reached in the usual flow of the code.
3321 If @samp{-falign-loops} or @samp{-falign-jumps} are applicable and
3322 are greater than this value, then their values are used instead.
3324 If @var{n} is not specified, use a machine-dependent default which is
3325 very likely to be @samp{1}, meaning no alignment.
3328 @itemx -falign-loops=@var{n}
3329 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
3330 like @samp{-falign-functions}. The hope is that the loop will be
3331 executed many times, which will make up for any execution of the dummy
3334 If @var{n} is not specified, use a machine-dependent default.
3337 @itemx -falign-jumps=@var{n}
3338 Align branch targets to a power-of-two boundary, for branch targets
3339 where the targets can only be reached by jumping, skipping up to @var{n}
3340 bytes like @samp{-falign-functions}. In this case, no dummy operations
3343 If @var{n} is not specified, use a machine-dependent default.
3346 Perform optimizations in static single assignment form. Each function's
3347 flow graph is translated into SSA form, optimizations are performed, and
3348 the flow graph is translated back from SSA form. User's should not
3349 specify this option, since it is not yet ready for production use.
3352 Perform dead-code elimination in SSA form. Requires @samp{-fssa}. Like
3353 @samp{-fssa}, this is an experimental feature.
3355 @item -fsingle-precision-constant
3356 Treat floating point constant as single precision constant instead of
3357 implicitly converting it to double precision constant.
3359 @item -frename-registers
3360 Attempt to avoid false dependancies in scheduled code by making use
3361 of registers left over after register allocation. This optimization
3362 will most benefit processors with lots of registers. It can, however,
3363 make debugging impossible, since variables will no longer stay in
3364 a ``home register''.
3366 @item --param @var{name}=@var{value}
3367 In some places, GCC uses various constants to control the amount of
3368 optimization that is done. For example, GCC will not inline functions
3369 that contain more that a certain number of instructions. You can
3370 control some of these constants on the command-line using the
3371 @samp{--param} option.
3373 In each case, the @var{value} is a integer. The allowable choices for
3374 @var{name} are given in the following table:
3377 @item max-inline-insns
3378 If an function contains more than this many instructions, it
3379 will not be inlined. This option is precisely equivalent to
3380 @samp{-finline-limit}.
3382 @item max-gcse-memory
3383 The approximate maximum amount of memory that will be allocated in
3384 order to perform the global common subexpression elimination
3385 optimization. If more memory than specified is required, the
3386 optimization will not be done.
3391 @node Preprocessor Options
3392 @section Options Controlling the Preprocessor
3393 @cindex preprocessor options
3394 @cindex options, preprocessor
3396 These options control the C preprocessor, which is run on each C source
3397 file before actual compilation.
3399 If you use the @samp{-E} option, nothing is done except preprocessing.
3400 Some of these options make sense only together with @samp{-E} because
3401 they cause the preprocessor output to be unsuitable for actual
3405 @item -include @var{file}
3406 Process @var{file} as input before processing the regular input file.
3407 In effect, the contents of @var{file} are compiled first. Any @samp{-D}
3408 and @samp{-U} options on the command line are always processed before
3409 @samp{-include @var{file}}, regardless of the order in which they are
3410 written. All the @samp{-include} and @samp{-imacros} options are
3411 processed in the order in which they are written.
3413 @item -imacros @var{file}
3414 Process @var{file} as input, discarding the resulting output, before
3415 processing the regular input file. Because the output generated from
3416 @var{file} is discarded, the only effect of @samp{-imacros @var{file}}
3417 is to make the macros defined in @var{file} available for use in the
3418 main input. All the @samp{-include} and @samp{-imacros} options are
3419 processed in the order in which they are written.
3421 @item -idirafter @var{dir}
3422 @cindex second include path
3423 Add the directory @var{dir} to the second include path. The directories
3424 on the second include path are searched when a header file is not found
3425 in any of the directories in the main include path (the one that
3428 @item -iprefix @var{prefix}
3429 Specify @var{prefix} as the prefix for subsequent @samp{-iwithprefix}
3432 @item -iwithprefix @var{dir}
3433 Add a directory to the second include path. The directory's name is
3434 made by concatenating @var{prefix} and @var{dir}, where @var{prefix} was
3435 specified previously with @samp{-iprefix}. If you have not specified a
3436 prefix yet, the directory containing the installed passes of the
3437 compiler is used as the default.
3439 @item -iwithprefixbefore @var{dir}
3440 Add a directory to the main include path. The directory's name is made
3441 by concatenating @var{prefix} and @var{dir}, as in the case of
3442 @samp{-iwithprefix}.
3444 @item -isystem @var{dir}
3445 Add a directory to the beginning of the second include path, marking it
3446 as a system directory, so that it gets the same special treatment as
3447 is applied to the standard system directories.
3450 Do not search the standard system directories for header files. Only
3451 the directories you have specified with @samp{-I} options (and the
3452 current directory, if appropriate) are searched. @xref{Directory
3453 Options}, for information on @samp{-I}.
3455 By using both @samp{-nostdinc} and @samp{-I-}, you can limit the include-file
3456 search path to only those directories you specify explicitly.
3460 When searching for a header file in a directory, remap file names if a
3461 file named @file{header.gcc} exists in that directory. This can be used
3462 to work around limitations of file systems with file name restrictions.
3463 The @file{header.gcc} file should contain a series of lines with two
3464 tokens on each line: the first token is the name to map, and the second
3465 token is the actual name to use.
3468 Do not predefine any nonstandard macros. (Including architecture flags).
3471 Run only the C preprocessor. Preprocess all the C source files
3472 specified and output the results to standard output or to the
3473 specified output file.
3476 Tell the preprocessor not to discard comments. Used with the
3480 Tell the preprocessor not to generate @samp{#line} directives.
3481 Used with the @samp{-E} option.
3484 @cindex dependencies, make
3487 Instead of outputting the result of preprocessing, output a rule
3488 suitable for @code{make} describing the dependencies of the main source
3489 file. The preprocessor outputs one @code{make} rule containing the
3490 object file name for that source file, a colon, and the names of all the
3491 included files. Unless overridden explicitly, the object file name
3492 consists of the basename of the source file with any suffix replaced with
3493 object file suffix. If there are many included files then the
3494 rule is split into several lines using @samp{\}-newline.
3496 @samp{-M} implies @samp{-E}.
3500 Like @samp{-M}, but mention only the files included with @samp{#include
3501 "@var{file}"}. System header files included with @samp{#include
3502 <@var{file}>} are omitted.
3506 Like @samp{-M} but the dependency information is written to a file
3507 rather than stdout. @code{gcc} will use the same file name and
3508 directory as the object file, but with the suffix ".d" instead.
3510 This is in addition to compiling the main file as specified ---
3511 @samp{-MD} does not inhibit ordinary compilation the way @samp{-M} does,
3512 unless you also specify @samp{-MG}.
3514 With Mach, you can use the utility @code{md} to merge multiple
3515 dependency files into a single dependency file suitable for using with
3516 the @samp{make} command.
3520 Like @samp{-MD} except mention only user header files, not system
3523 @item -MF @var{file}
3525 When used with @samp{-M} or @samp{-MM}, specifies a file to write the
3526 dependencies to. This allows the preprocessor to write the preprocessed
3527 file to stdout normally. If no @samp{-MF} switch is given, CPP sends
3528 the rules to stdout and suppresses normal preprocessed output.
3530 Another way to specify output of a @code{make} rule is by setting
3531 the environment variable @env{DEPENDENCIES_OUTPUT} (@pxref{Environment
3536 When used with @samp{-M} or @samp{-MM}, @samp{-MG} says to treat missing
3537 header files as generated files and assume they live in the same
3538 directory as the source file. It suppresses preprocessed output, as a
3539 missing header file is ordinarily an error.
3541 This feature is used in automatic updating of makefiles.
3545 This option instructs CPP to add a phony target for each dependency
3546 other than the main file, causing each to depend on nothing. These
3547 dummy rules work around errors @code{make} gives if you remove header
3548 files without updating the @code{Makefile} to match.
3550 This is typical output:-
3553 /tmp/test.o: /tmp/test.c /tmp/test.h
3558 @item -MQ @var{target}
3559 @item -MT @var{target}
3562 By default CPP uses the main file name, including any path, and appends
3563 the object suffix, normally ``.o'', to it to obtain the name of the
3564 target for dependency generation. With @samp{-MT} you can specify a
3565 target yourself, overriding the default one.
3567 If you want multiple targets, you can specify them as a single argument
3568 to @samp{-MT}, or use multiple @samp{-MT} options.
3570 The targets you specify are output in the order they appear on the
3571 command line. @samp{-MQ} is identical to @samp{-MT}, except that the
3572 target name is quoted for Make, but with @samp{-MT} it isn't. For
3573 example, -MT '$(objpfx)foo.o' gives
3576 $(objpfx)foo.o: /tmp/foo.c
3579 but -MQ '$(objpfx)foo.o' gives
3582 $$(objpfx)foo.o: /tmp/foo.c
3585 The default target is automatically quoted, as if it were given with
3589 Print the name of each header file used, in addition to other normal
3592 @item -A@var{question}(@var{answer})
3593 Assert the answer @var{answer} for @var{question}, in case it is tested
3594 with a preprocessing conditional such as @samp{#if
3595 #@var{question}(@var{answer})}. @samp{-A-} disables the standard
3596 assertions that normally describe the target machine.
3599 Define macro @var{macro} with the string @samp{1} as its definition.
3601 @item -D@var{macro}=@var{defn}
3602 Define macro @var{macro} as @var{defn}. All instances of @samp{-D} on
3603 the command line are processed before any @samp{-U} options.
3605 Any @samp{-D} and @samp{-U} options on the command line are processed in
3606 order, and always before @samp{-imacros @var{file}}, regardless of the
3607 order in which they are written.
3610 Undefine macro @var{macro}. @samp{-U} options are evaluated after all
3611 @samp{-D} options, but before any @samp{-include} and @samp{-imacros}
3614 Any @samp{-D} and @samp{-U} options on the command line are processed in
3615 order, and always before @samp{-imacros @var{file}}, regardless of the
3616 order in which they are written.
3619 Tell the preprocessor to output only a list of the macro definitions
3620 that are in effect at the end of preprocessing. Used with the @samp{-E}
3624 Tell the preprocessing to pass all macro definitions into the output, in
3625 their proper sequence in the rest of the output.
3628 Like @samp{-dD} except that the macro arguments and contents are omitted.
3629 Only @samp{#define @var{name}} is included in the output.
3633 Output @samp{#include} directives in addition to the result of
3638 Process ISO standard trigraph sequences. These are three-character
3639 sequences, all starting with @samp{??}, that are defined by ISO C to
3640 stand for single characters. For example, @samp{??/} stands for
3641 @samp{\}, so @samp{'??/n'} is a character constant for a newline. By
3642 default, GCC ignores trigraphs, but in standard-conforming modes it
3643 converts them. See the @samp{-std} and @samp{-ansi} options.
3645 The nine trigraph sequences are
3676 Trigraph support is not popular, so many compilers do not implement it
3677 properly. Portable code should not rely on trigraphs being either
3678 converted or ignored.
3680 @item -Wp\,@var{option}
3681 Pass @var{option} as an option to the preprocessor. If @var{option}
3682 contains commas, it is split into multiple options at the commas.
3685 @node Assembler Options
3686 @section Passing Options to the Assembler
3688 @c prevent bad page break with this line
3689 You can pass options to the assembler.
3692 @item -Wa\,@var{option}
3693 Pass @var{option} as an option to the assembler. If @var{option}
3694 contains commas, it is split into multiple options at the commas.
3698 @section Options for Linking
3699 @cindex link options
3700 @cindex options, linking
3702 These options come into play when the compiler links object files into
3703 an executable output file. They are meaningless if the compiler is
3704 not doing a link step.
3708 @item @var{object-file-name}
3709 A file name that does not end in a special recognized suffix is
3710 considered to name an object file or library. (Object files are
3711 distinguished from libraries by the linker according to the file
3712 contents.) If linking is done, these object files are used as input
3718 If any of these options is used, then the linker is not run, and
3719 object file names should not be used as arguments. @xref{Overall
3723 @item -l@var{library}
3724 @itemx -l @var{library}
3725 Search the library named @var{library} when linking. (The second
3726 alternative with the library as a separate argument is only for
3727 POSIX compliance and is not recommended.)
3729 It makes a difference where in the command you write this option; the
3730 linker searches and processes libraries and object files in the order they
3731 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
3732 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
3733 to functions in @samp{z}, those functions may not be loaded.
3735 The linker searches a standard list of directories for the library,
3736 which is actually a file named @file{lib@var{library}.a}. The linker
3737 then uses this file as if it had been specified precisely by name.
3739 The directories searched include several standard system directories
3740 plus any that you specify with @samp{-L}.
3742 Normally the files found this way are library files---archive files
3743 whose members are object files. The linker handles an archive file by
3744 scanning through it for members which define symbols that have so far
3745 been referenced but not defined. But if the file that is found is an
3746 ordinary object file, it is linked in the usual fashion. The only
3747 difference between using an @samp{-l} option and specifying a file name
3748 is that @samp{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
3749 and searches several directories.
3752 You need this special case of the @samp{-l} option in order to
3753 link an Objective C program.
3756 Do not use the standard system startup files when linking.
3757 The standard system libraries are used normally, unless @option{-nostdlib}
3758 or @option{-nodefaultlibs} is used.
3760 @item -nodefaultlibs
3761 Do not use the standard system libraries when linking.
3762 Only the libraries you specify will be passed to the linker.
3763 The standard startup files are used normally, unless @option{-nostartfiles}
3764 is used. The compiler may generate calls to memcmp, memset, and memcpy
3765 for System V (and ISO C) environments or to bcopy and bzero for
3766 BSD environments. These entries are usually resolved by entries in
3767 libc. These entry points should be supplied through some other
3768 mechanism when this option is specified.
3771 Do not use the standard system startup files or libraries when linking.
3772 No startup files and only the libraries you specify will be passed to
3773 the linker. The compiler may generate calls to memcmp, memset, and memcpy
3774 for System V (and ISO C) environments or to bcopy and bzero for
3775 BSD environments. These entries are usually resolved by entries in
3776 libc. These entry points should be supplied through some other
3777 mechanism when this option is specified.
3779 @cindex @code{-lgcc}, use with @code{-nostdlib}
3780 @cindex @code{-nostdlib} and unresolved references
3781 @cindex unresolved references and @code{-nostdlib}
3782 @cindex @code{-lgcc}, use with @code{-nodefaultlibs}
3783 @cindex @code{-nodefaultlibs} and unresolved references
3784 @cindex unresolved references and @code{-nodefaultlibs}
3785 One of the standard libraries bypassed by @samp{-nostdlib} and
3786 @samp{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
3787 that GCC uses to overcome shortcomings of particular machines, or special
3788 needs for some languages.
3790 (@xref{Interface,,Interfacing to GCC Output}, for more discussion of
3794 (@xref{Interface,,Interfacing to GCC Output,gcc.info,Porting GCC},
3795 for more discussion of @file{libgcc.a}.)
3797 In most cases, you need @file{libgcc.a} even when you want to avoid
3798 other standard libraries. In other words, when you specify @samp{-nostdlib}
3799 or @samp{-nodefaultlibs} you should usually specify @samp{-lgcc} as well.
3800 This ensures that you have no unresolved references to internal GCC
3801 library subroutines. (For example, @samp{__main}, used to ensure C++
3802 constructors will be called; @pxref{Collect2,,@command{collect2}}.)
3805 Remove all symbol table and relocation information from the executable.
3808 On systems that support dynamic linking, this prevents linking with the shared
3809 libraries. On other systems, this option has no effect.
3812 Produce a shared object which can then be linked with other objects to
3813 form an executable. Not all systems support this option. For predictable
3814 results, you must also specify the same set of options that were used to
3815 generate code (@samp{-fpic}, @samp{-fPIC}, or model suboptions)
3816 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
3817 needs to build supplementary stub code for constructors to work. On
3818 multi-libbed systems, @samp{gcc -shared} must select the correct support
3819 libraries to link against. Failing to supply the correct flags may lead
3820 to subtle defects. Supplying them in cases where they are not necessary
3823 @item -shared-libgcc
3824 @itemx -static-libgcc
3825 On systems that provide @file{libgcc} as a shared library, these options
3826 force the use of either the shared or static version respectively.
3827 If no shared version of @file{libgcc} was built when the compiler was
3828 configured, these options have no effect.
3830 There are several situations in which an application should use the
3831 shared @file{libgcc} instead of the static version. The most common
3832 of these is when the application wishes to throw and catch exceptions
3833 across different shared libraries. In that case, each of the libraries
3834 as well as the application itself should use the shared @file{libgcc}.
3836 Therefore, whenever you specify the @samp{-shared} option, the GCC
3837 driver automatically adds @samp{-shared-libgcc}, unless you explicitly
3838 specify @samp{-static-libgcc}. The G++ driver automatically adds
3839 @samp{-shared-libgcc} when you build a main executable as well because
3840 for C++ programs that is typically the right thing to do.
3841 (Exception-handling will not work reliably otherwise.)
3843 However, when linking a main executable written in C, you must
3844 explicitly say @samp{-shared-libgcc} if you want to use the shared
3848 Bind references to global symbols when building a shared object. Warn
3849 about any unresolved references (unless overridden by the link editor
3850 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
3853 @item -Xlinker @var{option}
3854 Pass @var{option} as an option to the linker. You can use this to
3855 supply system-specific linker options which GCC does not know how to
3858 If you want to pass an option that takes an argument, you must use
3859 @samp{-Xlinker} twice, once for the option and once for the argument.
3860 For example, to pass @samp{-assert definitions}, you must write
3861 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
3862 @samp{-Xlinker "-assert definitions"}, because this passes the entire
3863 string as a single argument, which is not what the linker expects.
3865 @item -Wl\,@var{option}
3866 Pass @var{option} as an option to the linker. If @var{option} contains
3867 commas, it is split into multiple options at the commas.
3869 @item -u @var{symbol}
3870 Pretend the symbol @var{symbol} is undefined, to force linking of
3871 library modules to define it. You can use @samp{-u} multiple times with
3872 different symbols to force loading of additional library modules.
3875 @node Directory Options
3876 @section Options for Directory Search
3877 @cindex directory options
3878 @cindex options, directory search
3881 These options specify directories to search for header files, for
3882 libraries and for parts of the compiler:
3886 Add the directory @var{dir} to the head of the list of directories to be
3887 searched for header files. This can be used to override a system header
3888 file, substituting your own version, since these directories are
3889 searched before the system header file directories. However, you should
3890 not use this option to add directories that contain vendor-supplied
3891 system header files (use @samp{-isystem} for that). If you use more than
3892 one @samp{-I} option, the directories are scanned in left-to-right
3893 order; the standard system directories come after.
3896 Any directories you specify with @samp{-I} options before the @samp{-I-}
3897 option are searched only for the case of @samp{#include "@var{file}"};
3898 they are not searched for @samp{#include <@var{file}>}.
3900 If additional directories are specified with @samp{-I} options after
3901 the @samp{-I-}, these directories are searched for all @samp{#include}
3902 directives. (Ordinarily @emph{all} @samp{-I} directories are used
3905 In addition, the @samp{-I-} option inhibits the use of the current
3906 directory (where the current input file came from) as the first search
3907 directory for @samp{#include "@var{file}"}. There is no way to
3908 override this effect of @samp{-I-}. With @samp{-I.} you can specify
3909 searching the directory which was current when the compiler was
3910 invoked. That is not exactly the same as what the preprocessor does
3911 by default, but it is often satisfactory.
3913 @samp{-I-} does not inhibit the use of the standard system directories
3914 for header files. Thus, @samp{-I-} and @samp{-nostdinc} are
3918 Add directory @var{dir} to the list of directories to be searched
3921 @item -B@var{prefix}
3922 This option specifies where to find the executables, libraries,
3923 include files, and data files of the compiler itself.
3925 The compiler driver program runs one or more of the subprograms
3926 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
3927 @var{prefix} as a prefix for each program it tries to run, both with and
3928 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
3930 For each subprogram to be run, the compiler driver first tries the
3931 @samp{-B} prefix, if any. If that name is not found, or if @samp{-B}
3932 was not specified, the driver tries two standard prefixes, which are
3933 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc-lib/}. If neither of
3934 those results in a file name that is found, the unmodified program
3935 name is searched for using the directories specified in your
3936 @env{PATH} environment variable.
3938 @samp{-B} prefixes that effectively specify directory names also apply
3939 to libraries in the linker, because the compiler translates these
3940 options into @samp{-L} options for the linker. They also apply to
3941 includes files in the preprocessor, because the compiler translates these
3942 options into @samp{-isystem} options for the preprocessor. In this case,
3943 the compiler appends @samp{include} to the prefix.
3945 The run-time support file @file{libgcc.a} can also be searched for using
3946 the @samp{-B} prefix, if needed. If it is not found there, the two
3947 standard prefixes above are tried, and that is all. The file is left
3948 out of the link if it is not found by those means.
3950 Another way to specify a prefix much like the @samp{-B} prefix is to use
3951 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
3954 @item -specs=@var{file}
3955 Process @var{file} after the compiler reads in the standard @file{specs}
3956 file, in order to override the defaults that the @file{gcc} driver
3957 program uses when determining what switches to pass to @file{cc1},
3958 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
3959 @samp{-specs=}@var{file} can be specified on the command line, and they
3960 are processed in order, from left to right.
3966 @section Specifying subprocesses and the switches to pass to them
3968 @command{gcc} is a driver program. It performs its job by invoking a
3969 sequence of other programs to do the work of compiling, assembling and
3970 linking. GCC interprets its command-line parameters and uses these to
3971 deduce which programs it should invoke, and which command-line options
3972 it ought to place on their command lines. This behaviour is controlled
3973 by @dfn{spec strings}. In most cases there is one spec string for each
3974 program that GCC can invoke, but a few programs have multiple spec
3975 strings to control their behaviour. The spec strings built into GCC can
3976 be overridden by using the @samp{-specs=} command-line switch to specify
3979 @dfn{Spec files} are plaintext files that are used to construct spec
3980 strings. They consist of a sequence of directives separated by blank
3981 lines. The type of directive is determined by the first non-whitespace
3982 character on the line and it can be one of the following:
3985 @item %@var{command}
3986 Issues a @var{command} to the spec file processor. The commands that can
3990 @item %include <@var{file}>
3992 Search for @var{file} and insert its text at the current point in the
3995 @item %include_noerr <@var{file}>
3996 @cindex %include_noerr
3997 Just like @samp{%include}, but do not generate an error message if the include
3998 file cannot be found.
4000 @item %rename @var{old_name} @var{new_name}
4002 Rename the spec string @var{old_name} to @var{new_name}.
4006 @item *[@var{spec_name}]:
4007 This tells the compiler to create, override or delete the named spec
4008 string. All lines after this directive up to the next directive or
4009 blank line are considered to be the text for the spec string. If this
4010 results in an empty string then the spec will be deleted. (Or, if the
4011 spec did not exist, then nothing will happened.) Otherwise, if the spec
4012 does not currently exist a new spec will be created. If the spec does
4013 exist then its contents will be overridden by the text of this
4014 directive, unless the first character of that text is the @samp{+}
4015 character, in which case the text will be appended to the spec.
4017 @item [@var{suffix}]:
4018 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
4019 and up to the next directive or blank line are considered to make up the
4020 spec string for the indicated suffix. When the compiler encounters an
4021 input file with the named suffix, it will processes the spec string in
4022 order to work out how to compile that file. For example:
4029 This says that any input file whose name ends in @samp{.ZZ} should be
4030 passed to the program @samp{z-compile}, which should be invoked with the
4031 command-line switch @samp{-input} and with the result of performing the
4032 @samp{%i} substitution. (See below.)
4034 As an alternative to providing a spec string, the text that follows a
4035 suffix directive can be one of the following:
4038 @item @@@var{language}
4039 This says that the suffix is an alias for a known @var{language}. This is
4040 similar to using the @option{-x} command-line switch to GCC to specify a
4041 language explicitly. For example:
4048 Says that .ZZ files are, in fact, C++ source files.
4051 This causes an error messages saying:
4054 @var{name} compiler not installed on this system.
4058 GCC already has an extensive list of suffixes built into it.
4059 This directive will add an entry to the end of the list of suffixes, but
4060 since the list is searched from the end backwards, it is effectively
4061 possible to override earlier entries using this technique.
4065 GCC has the following spec strings built into it. Spec files can
4066 override these strings or create their own. Note that individual
4067 targets can also add their own spec strings to this list.
4070 asm Options to pass to the assembler
4071 asm_final Options to pass to the assembler post-processor
4072 cpp Options to pass to the C preprocessor
4073 cc1 Options to pass to the C compiler
4074 cc1plus Options to pass to the C++ compiler
4075 endfile Object files to include at the end of the link
4076 link Options to pass to the linker
4077 lib Libraries to include on the command line to the linker
4078 libgcc Decides which GCC support library to pass to the linker
4079 linker Sets the name of the linker
4080 predefines Defines to be passed to the C preprocessor
4081 signed_char Defines to pass to CPP to say whether @code{char} is signed by default
4082 startfile Object files to include at the start of the link
4085 Here is a small example of a spec file:
4091 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
4094 This example renames the spec called @samp{lib} to @samp{old_lib} and
4095 then overrides the previous definition of @samp{lib} with a new one.
4096 The new definition adds in some extra command-line options before
4097 including the text of the old definition.
4099 @dfn{Spec strings} are a list of command-line options to be passed to their
4100 corresponding program. In addition, the spec strings can contain
4101 @samp{%}-prefixed sequences to substitute variable text or to
4102 conditionally insert text into the command line. Using these constructs
4103 it is possible to generate quite complex command lines.
4105 Here is a table of all defined @samp{%}-sequences for spec
4106 strings. Note that spaces are not generated automatically around the
4107 results of expanding these sequences. Therefore you can concatenate them
4108 together or combine them with constant text in a single argument.
4112 Substitute one @samp{%} into the program name or argument.
4115 Substitute the name of the input file being processed.
4118 Substitute the basename of the input file being processed.
4119 This is the substring up to (and not including) the last period
4120 and not including the directory.
4123 This is the same as @samp{%b}, but include the file suffix (text after
4127 Marks the argument containing or following the @samp{%d} as a
4128 temporary file name, so that that file will be deleted if GCC exits
4129 successfully. Unlike @samp{%g}, this contributes no text to the
4132 @item %g@var{suffix}
4133 Substitute a file name that has suffix @var{suffix} and is chosen
4134 once per compilation, and mark the argument in the same way as
4135 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
4136 name is now chosen in a way that is hard to predict even when previously
4137 chosen file names are known. For example, @samp{%g.s ... %g.o ... %g.s}
4138 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
4139 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
4140 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
4141 was simply substituted with a file name chosen once per compilation,
4142 without regard to any appended suffix (which was therefore treated
4143 just like ordinary text), making such attacks more likely to succeed.
4145 @item %u@var{suffix}
4146 Like @samp{%g}, but generates a new temporary file name even if
4147 @samp{%u@var{suffix}} was already seen.
4149 @item %U@var{suffix}
4150 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
4151 new one if there is no such last file name. In the absence of any
4152 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
4153 the same suffix @emph{space}, so @samp{%g.s ... %U.s ... %g.s ... %U.s}
4154 would involve the generation of two distinct file names, one
4155 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
4156 simply substituted with a file name chosen for the previous @samp{%u},
4157 without regard to any appended suffix.
4159 @item %j@var{SUFFIX}
4160 Substitutes the name of the HOST_BIT_BUCKET, if any, and if it is
4161 writable, and if save-temps is off; otherwise, substitute the name
4162 of a temporary file, just like @samp{%u}. This temporary file is not
4163 meant for communication between processes, but rather as a junk
4166 @item %.@var{SUFFIX}
4167 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
4168 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
4169 terminated by the next space or %.
4172 Marks the argument containing or following the @samp{%w} as the
4173 designated output file of this compilation. This puts the argument
4174 into the sequence of arguments that @samp{%o} will substitute later.
4177 Substitutes the names of all the output files, with spaces
4178 automatically placed around them. You should write spaces
4179 around the @samp{%o} as well or the results are undefined.
4180 @samp{%o} is for use in the specs for running the linker.
4181 Input files whose names have no recognized suffix are not compiled
4182 at all, but they are included among the output files, so they will
4186 Substitutes the suffix for object files. Note that this is
4187 handled specially when it immediately follows @samp{%g, %u, or %U},
4188 because of the need for those to form complete file names. The
4189 handling is such that @samp{%O} is treated exactly as if it had already
4190 been substituted, except that @samp{%g, %u, and %U} do not currently
4191 support additional @var{suffix} characters following @samp{%O} as they would
4192 following, for example, @samp{.o}.
4195 Substitutes the standard macro predefinitions for the
4196 current target machine. Use this when running @code{cpp}.
4199 Like @samp{%p}, but puts @samp{__} before and after the name of each
4200 predefined macro, except for macros that start with @samp{__} or with
4201 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
4205 Substitute a @samp{-iprefix} option made from GCC_EXEC_PREFIX.
4208 Current argument is the name of a library or startup file of some sort.
4209 Search for that file in a standard list of directories and substitute
4210 the full name found.
4213 Print @var{str} as an error message. @var{str} is terminated by a newline.
4214 Use this when inconsistent options are detected.
4217 Output @samp{-} if the input for the current command is coming from a pipe.
4220 Substitute the contents of spec string @var{name} at this point.
4223 Like @samp{%(...)} but put @samp{__} around @samp{-D} arguments.
4225 @item %x@{@var{option}@}
4226 Accumulate an option for @samp{%X}.
4229 Output the accumulated linker options specified by @samp{-Wl} or a @samp{%x}
4233 Output the accumulated assembler options specified by @samp{-Wa}.
4236 Output the accumulated preprocessor options specified by @samp{-Wp}.
4239 Substitute the major version number of GCC.
4240 (For version 2.9.5, this is 2.)
4243 Substitute the minor version number of GCC.
4244 (For version 2.9.5, this is 9.)
4247 Substitute the patch level number of GCC.
4248 (For version 2.9.5, this is 5.)
4251 Process the @code{asm} spec. This is used to compute the
4252 switches to be passed to the assembler.
4255 Process the @code{asm_final} spec. This is a spec string for
4256 passing switches to an assembler post-processor, if such a program is
4260 Process the @code{link} spec. This is the spec for computing the
4261 command line passed to the linker. Typically it will make use of the
4262 @samp{%L %G %S %D and %E} sequences.
4265 Dump out a @samp{-L} option for each directory that GCC believes might
4266 contain startup files. If the target supports multilibs then the
4267 current multilib directory will be prepended to each of these paths.
4270 Output the multilib directory with directory seperators replaced with
4271 "_". If multilib directories are not set, or the multilib directory is
4272 "." then this option emits nothing.
4275 Process the @code{lib} spec. This is a spec string for deciding which
4276 libraries should be included on the command line to the linker.
4279 Process the @code{libgcc} spec. This is a spec string for deciding
4280 which GCC support library should be included on the command line to the linker.
4283 Process the @code{startfile} spec. This is a spec for deciding which
4284 object files should be the first ones passed to the linker. Typically
4285 this might be a file named @file{crt0.o}.
4288 Process the @code{endfile} spec. This is a spec string that specifies
4289 the last object files that will be passed to the linker.
4292 Process the @code{cpp} spec. This is used to construct the arguments
4293 to be passed to the C preprocessor.
4296 Process the @code{signed_char} spec. This is intended to be used
4297 to tell cpp whether a char is signed. It typically has the definition:
4299 %@{funsigned-char:-D__CHAR_UNSIGNED__@}
4303 Process the @code{cc1} spec. This is used to construct the options to be
4304 passed to the actual C compiler (@samp{cc1}).
4307 Process the @code{cc1plus} spec. This is used to construct the options to be
4308 passed to the actual C++ compiler (@samp{cc1plus}).
4311 Substitute the variable part of a matched option. See below.
4312 Note that each comma in the substituted string is replaced by
4316 Substitutes the @code{-S} switch, if that switch was given to GCC.
4317 If that switch was not specified, this substitutes nothing. Note that
4318 the leading dash is omitted when specifying this option, and it is
4319 automatically inserted if the substitution is performed. Thus the spec
4320 string @samp{%@{foo@}} would match the command-line option @samp{-foo}
4321 and would output the command line option @samp{-foo}.
4323 @item %W@{@code{S}@}
4324 Like %@{@code{S}@} but mark last argument supplied within as a file to be
4327 @item %@{@code{S}*@}
4328 Substitutes all the switches specified to GCC whose names start
4329 with @code{-S}, but which also take an argument. This is used for
4330 switches like @samp{-o, -D, -I}, etc. GCC considers @samp{-o foo} as being
4331 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
4332 text, including the space. Thus two arguments would be generated.
4334 @item %@{^@code{S}*@}
4335 Like %@{@code{S}*@}, but don't put a blank between a switch and its
4336 argument. Thus %@{^o*@} would only generate one argument, not two.
4338 @item %@{@code{S}*&@code{T}*@}
4339 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
4340 (the order of @code{S} and @code{T} in the spec is not significant).
4341 There can be any number of ampersand-separated variables; for each the
4342 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
4344 @item %@{<@code{S}@}
4345 Remove all occurrences of @code{-S} from the command line. Note - this
4346 command is position dependent. @samp{%} commands in the spec string
4347 before this option will see @code{-S}, @samp{%} commands in the spec
4348 string after this option will not.
4350 @item %@{@code{S}*:@code{X}@}
4351 Substitutes @code{X} if one or more switches whose names start with
4352 @code{-S} are specified to GCC. Note that the tail part of the
4353 @code{-S} option (i.e. the part matched by the @samp{*}) will be substituted
4354 for each occurrence of @samp{%*} within @code{X}.
4356 @item %@{@code{S}:@code{X}@}
4357 Substitutes @code{X}, but only if the @samp{-S} switch was given to GCC.
4359 @item %@{!@code{S}:@code{X}@}
4360 Substitutes @code{X}, but only if the @samp{-S} switch was @emph{not} given to GCC.
4362 @item %@{|@code{S}:@code{X}@}
4363 Like %@{@code{S}:@code{X}@}, but if no @code{S} switch, substitute @samp{-}.
4365 @item %@{|!@code{S}:@code{X}@}
4366 Like %@{!@code{S}:@code{X}@}, but if there is an @code{S} switch, substitute @samp{-}.
4368 @item %@{.@code{S}:@code{X}@}
4369 Substitutes @code{X}, but only if processing a file with suffix @code{S}.
4371 @item %@{!.@code{S}:@code{X}@}
4372 Substitutes @code{X}, but only if @emph{not} processing a file with suffix @code{S}.
4374 @item %@{@code{S}|@code{P}:@code{X}@}
4375 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC. This may be
4376 combined with @samp{!} and @samp{.} sequences as well, although they
4377 have a stronger binding than the @samp{|}. For example a spec string
4381 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
4384 will output the following command-line options from the following input
4385 command-line options:
4390 -d fred.c -foo -baz -boggle
4391 -d jim.d -bar -baz -boggle
4396 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or
4397 %@{!@code{S}:@code{X}@} construct may contain other nested @samp{%} constructs
4398 or spaces, or even newlines. They are processed as usual, as described
4401 The @samp{-O, -f, -m, and -W} switches are handled specifically in these
4402 constructs. If another value of @samp{-O} or the negated form of a @samp{-f, -m, or
4403 -W} switch is found later in the command line, the earlier switch
4404 value is ignored, except with @{@code{S}*@} where @code{S} is just one
4405 letter, which passes all matching options.
4407 The character @samp{|} at the beginning of the predicate text is used to indicate
4408 that a command should be piped to the following command, but only if @samp{-pipe}
4411 It is built into GCC which switches take arguments and which do not.
4412 (You might think it would be useful to generalize this to allow each
4413 compiler's spec to say which switches take arguments. But this cannot
4414 be done in a consistent fashion. GCC cannot even decide which input
4415 files have been specified without knowing which switches take arguments,
4416 and it must know which input files to compile in order to tell which
4419 GCC also knows implicitly that arguments starting in @samp{-l} are to be
4420 treated as compiler output files, and passed to the linker in their
4421 proper position among the other output files.
4423 @c man begin OPTIONS
4425 @node Target Options
4426 @section Specifying Target Machine and Compiler Version
4427 @cindex target options
4428 @cindex cross compiling
4429 @cindex specifying machine version
4430 @cindex specifying compiler version and target machine
4431 @cindex compiler version, specifying
4432 @cindex target machine, specifying
4434 By default, GCC compiles code for the same type of machine that you
4435 are using. However, it can also be installed as a cross-compiler, to
4436 compile for some other type of machine. In fact, several different
4437 configurations of GCC, for different target machines, can be
4438 installed side by side. Then you specify which one to use with the
4441 In addition, older and newer versions of GCC can be installed side
4442 by side. One of them (probably the newest) will be the default, but
4443 you may sometimes wish to use another.
4446 @item -b @var{machine}
4447 The argument @var{machine} specifies the target machine for compilation.
4448 This is useful when you have installed GCC as a cross-compiler.
4450 The value to use for @var{machine} is the same as was specified as the
4451 machine type when configuring GCC as a cross-compiler. For
4452 example, if a cross-compiler was configured with @samp{configure
4453 i386v}, meaning to compile for an 80386 running System V, then you
4454 would specify @samp{-b i386v} to run that cross compiler.
4456 When you do not specify @samp{-b}, it normally means to compile for
4457 the same type of machine that you are using.
4459 @item -V @var{version}
4460 The argument @var{version} specifies which version of GCC to run.
4461 This is useful when multiple versions are installed. For example,
4462 @var{version} might be @samp{2.0}, meaning to run GCC version 2.0.
4464 The default version, when you do not specify @samp{-V}, is the last
4465 version of GCC that you installed.
4468 The @samp{-b} and @samp{-V} options actually work by controlling part of
4469 the file name used for the executable files and libraries used for
4470 compilation. A given version of GCC, for a given target machine, is
4471 normally kept in the directory @file{/usr/local/lib/gcc-lib/@var{machine}/@var{version}}.@refill
4473 Thus, sites can customize the effect of @samp{-b} or @samp{-V} either by
4474 changing the names of these directories or adding alternate names (or
4475 symbolic links). If in directory @file{/usr/local/lib/gcc-lib/} the
4476 file @file{80386} is a link to the file @file{i386v}, then @samp{-b
4477 80386} becomes an alias for @samp{-b i386v}.
4479 In one respect, the @samp{-b} or @samp{-V} do not completely change
4480 to a different compiler: the top-level driver program @command{gcc}
4481 that you originally invoked continues to run and invoke the other
4482 executables (preprocessor, compiler per se, assembler and linker)
4483 that do the real work. However, since no real work is done in the
4484 driver program, it usually does not matter that the driver program
4485 in use is not the one for the specified target. It is common for the
4486 interface to the other executables to change incompatibly between
4487 compiler versions, so unless the version specified is very close to that
4488 of the driver (for example, @samp{-V 3.0} with a driver program from GCC
4489 version 3.0.1), use of @samp{-V} may not work; for example, using
4490 @samp{-V 2.95.2} will not work with a driver program from GCC 3.0.
4492 The only way that the driver program depends on the target machine is
4493 in the parsing and handling of special machine-specific options.
4494 However, this is controlled by a file which is found, along with the
4495 other executables, in the directory for the specified version and
4496 target machine. As a result, a single installed driver program adapts
4497 to any specified target machine, and sufficiently similar compiler
4500 The driver program executable does control one significant thing,
4501 however: the default version and target machine. Therefore, you can
4502 install different instances of the driver program, compiled for
4503 different targets or versions, under different names.
4505 For example, if the driver for version 2.0 is installed as @command{ogcc}
4506 and that for version 2.1 is installed as @command{gcc}, then the command
4507 @command{gcc} will use version 2.1 by default, while @command{ogcc} will use
4508 2.0 by default. However, you can choose either version with either
4509 command with the @samp{-V} option.
4511 @node Submodel Options
4512 @section Hardware Models and Configurations
4513 @cindex submodel options
4514 @cindex specifying hardware config
4515 @cindex hardware models and configurations, specifying
4516 @cindex machine dependent options
4518 Earlier we discussed the standard option @samp{-b} which chooses among
4519 different installed compilers for completely different target
4520 machines, such as Vax vs. 68000 vs. 80386.
4522 In addition, each of these target machine types can have its own
4523 special options, starting with @samp{-m}, to choose among various
4524 hardware models or configurations---for example, 68010 vs 68020,
4525 floating coprocessor or none. A single installed version of the
4526 compiler can compile for any model or configuration, according to the
4529 Some configurations of the compiler also support additional special
4530 options, usually for compatibility with other compilers on the same
4534 These options are defined by the macro @code{TARGET_SWITCHES} in the
4535 machine description. The default for the options is also defined by
4536 that macro, which enables you to change the defaults.
4552 * RS/6000 and PowerPC Options::
4557 * Intel 960 Options::
4558 * DEC Alpha Options::
4562 * System V Options::
4563 * TMS320C3x/C4x Options::
4573 @node M680x0 Options
4574 @subsection M680x0 Options
4575 @cindex M680x0 options
4577 These are the @samp{-m} options defined for the 68000 series. The default
4578 values for these options depends on which style of 68000 was selected when
4579 the compiler was configured; the defaults for the most common choices are
4585 Generate output for a 68000. This is the default
4586 when the compiler is configured for 68000-based systems.
4588 Use this option for microcontrollers with a 68000 or EC000 core,
4589 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
4593 Generate output for a 68020. This is the default
4594 when the compiler is configured for 68020-based systems.
4597 Generate output containing 68881 instructions for floating point.
4598 This is the default for most 68020 systems unless @samp{-nfp} was
4599 specified when the compiler was configured.
4602 Generate output for a 68030. This is the default when the compiler is
4603 configured for 68030-based systems.
4606 Generate output for a 68040. This is the default when the compiler is
4607 configured for 68040-based systems.
4609 This option inhibits the use of 68881/68882 instructions that have to be
4610 emulated by software on the 68040. Use this option if your 68040 does not
4611 have code to emulate those instructions.
4614 Generate output for a 68060. This is the default when the compiler is
4615 configured for 68060-based systems.
4617 This option inhibits the use of 68020 and 68881/68882 instructions that
4618 have to be emulated by software on the 68060. Use this option if your 68060
4619 does not have code to emulate those instructions.
4622 Generate output for a CPU32. This is the default
4623 when the compiler is configured for CPU32-based systems.
4625 Use this option for microcontrollers with a
4626 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
4627 68336, 68340, 68341, 68349 and 68360.
4630 Generate output for a 520X "coldfire" family cpu. This is the default
4631 when the compiler is configured for 520X-based systems.
4633 Use this option for microcontroller with a 5200 core, including
4634 the MCF5202, MCF5203, MCF5204 and MCF5202.
4638 Generate output for a 68040, without using any of the new instructions.
4639 This results in code which can run relatively efficiently on either a
4640 68020/68881 or a 68030 or a 68040. The generated code does use the
4641 68881 instructions that are emulated on the 68040.
4644 Generate output for a 68060, without using any of the new instructions.
4645 This results in code which can run relatively efficiently on either a
4646 68020/68881 or a 68030 or a 68040. The generated code does use the
4647 68881 instructions that are emulated on the 68060.
4650 Generate output containing Sun FPA instructions for floating point.
4653 Generate output containing library calls for floating point.
4654 @strong{Warning:} the requisite libraries are not available for all m68k
4655 targets. Normally the facilities of the machine's usual C compiler are
4656 used, but this can't be done directly in cross-compilation. You must
4657 make your own arrangements to provide suitable library functions for
4658 cross-compilation. The embedded targets @samp{m68k-*-aout} and
4659 @samp{m68k-*-coff} do provide software floating point support.
4662 Consider type @code{int} to be 16 bits wide, like @code{short int}.
4665 Do not use the bit-field instructions. The @samp{-m68000}, @samp{-mcpu32}
4666 and @samp{-m5200} options imply @w{@samp{-mnobitfield}}.
4669 Do use the bit-field instructions. The @samp{-m68020} option implies
4670 @samp{-mbitfield}. This is the default if you use a configuration
4671 designed for a 68020.
4674 Use a different function-calling convention, in which functions
4675 that take a fixed number of arguments return with the @code{rtd}
4676 instruction, which pops their arguments while returning. This
4677 saves one instruction in the caller since there is no need to pop
4678 the arguments there.
4680 This calling convention is incompatible with the one normally
4681 used on Unix, so you cannot use it if you need to call libraries
4682 compiled with the Unix compiler.
4684 Also, you must provide function prototypes for all functions that
4685 take variable numbers of arguments (including @code{printf});
4686 otherwise incorrect code will be generated for calls to those
4689 In addition, seriously incorrect code will result if you call a
4690 function with too many arguments. (Normally, extra arguments are
4691 harmlessly ignored.)
4693 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
4694 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
4697 @itemx -mno-align-int
4698 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
4699 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
4700 boundary (@samp{-malign-int}) or a 16-bit boundary (@samp{-mno-align-int}).
4701 Aligning variables on 32-bit boundaries produces code that runs somewhat
4702 faster on processors with 32-bit busses at the expense of more memory.
4704 @strong{Warning:} if you use the @samp{-malign-int} switch, GCC will
4705 align structures containing the above types differently than
4706 most published application binary interface specifications for the m68k.
4709 Use the pc-relative addressing mode of the 68000 directly, instead of
4710 using a global offset table. At present, this option implies -fpic,
4711 allowing at most a 16-bit offset for pc-relative addressing. -fPIC is
4712 not presently supported with -mpcrel, though this could be supported for
4713 68020 and higher processors.
4715 @item -mno-strict-align
4716 @itemx -mstrict-align
4717 @kindex -mstrict-align
4718 Do not (do) assume that unaligned memory references will be handled by
4723 @node M68hc1x Options
4724 @subsection M68hc1x Options
4725 @cindex M68hc1x options
4727 These are the @samp{-m} options defined for the 68hc11 and 68hc12
4728 microcontrollers. The default values for these options depends on
4729 which style of microcontroller was selected when the compiler was configured;
4730 the defaults for the most common choices are given below.
4735 Generate output for a 68HC11. This is the default
4736 when the compiler is configured for 68HC11-based systems.
4740 Generate output for a 68HC12. This is the default
4741 when the compiler is configured for 68HC12-based systems.
4744 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
4748 Consider type @code{int} to be 16 bits wide, like @code{short int}.
4750 @item -msoft-reg-count=@var{count}
4751 Specify the number of pseudo-soft registers which are used for the
4752 code generation. The maximum number is 32. Using more pseudo-soft
4753 register may or may not result in better code depending on the program.
4754 The default is 4 for 68HC11 and 2 for 68HC12.
4759 @subsection VAX Options
4762 These @samp{-m} options are defined for the Vax:
4766 Do not output certain jump instructions (@code{aobleq} and so on)
4767 that the Unix assembler for the Vax cannot handle across long
4771 Do output those jump instructions, on the assumption that you
4772 will assemble with the GNU assembler.
4775 Output code for g-format floating point numbers instead of d-format.
4779 @subsection SPARC Options
4780 @cindex SPARC options
4782 These @samp{-m} switches are supported on the SPARC:
4787 Specify @samp{-mapp-regs} to generate output using the global registers
4788 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
4791 To be fully SVR4 ABI compliant at the cost of some performance loss,
4792 specify @samp{-mno-app-regs}. You should compile libraries and system
4793 software with this option.
4797 Generate output containing floating point instructions. This is the
4802 Generate output containing library calls for floating point.
4803 @strong{Warning:} the requisite libraries are not available for all SPARC
4804 targets. Normally the facilities of the machine's usual C compiler are
4805 used, but this cannot be done directly in cross-compilation. You must make
4806 your own arrangements to provide suitable library functions for
4807 cross-compilation. The embedded targets @samp{sparc-*-aout} and
4808 @samp{sparclite-*-*} do provide software floating point support.
4810 @samp{-msoft-float} changes the calling convention in the output file;
4811 therefore, it is only useful if you compile @emph{all} of a program with
4812 this option. In particular, you need to compile @file{libgcc.a}, the
4813 library that comes with GCC, with @samp{-msoft-float} in order for
4816 @item -mhard-quad-float
4817 Generate output containing quad-word (long double) floating point
4820 @item -msoft-quad-float
4821 Generate output containing library calls for quad-word (long double)
4822 floating point instructions. The functions called are those specified
4823 in the SPARC ABI. This is the default.
4825 As of this writing, there are no sparc implementations that have hardware
4826 support for the quad-word floating point instructions. They all invoke
4827 a trap handler for one of these instructions, and then the trap handler
4828 emulates the effect of the instruction. Because of the trap handler overhead,
4829 this is much slower than calling the ABI library routines. Thus the
4830 @samp{-msoft-quad-float} option is the default.
4834 With @samp{-mepilogue} (the default), the compiler always emits code for
4835 function exit at the end of each function. Any function exit in
4836 the middle of the function (such as a return statement in C) will
4837 generate a jump to the exit code at the end of the function.
4839 With @samp{-mno-epilogue}, the compiler tries to emit exit code inline
4840 at every function exit.
4844 With @samp{-mflat}, the compiler does not generate save/restore instructions
4845 and will use a "flat" or single register window calling convention.
4846 This model uses %i7 as the frame pointer and is compatible with the normal
4847 register window model. Code from either may be intermixed.
4848 The local registers and the input registers (0-5) are still treated as
4849 "call saved" registers and will be saved on the stack as necessary.
4851 With @samp{-mno-flat} (the default), the compiler emits save/restore
4852 instructions (except for leaf functions) and is the normal mode of operation.
4854 @item -mno-unaligned-doubles
4855 @itemx -munaligned-doubles
4856 Assume that doubles have 8 byte alignment. This is the default.
4858 With @samp{-munaligned-doubles}, GCC assumes that doubles have 8 byte
4859 alignment only if they are contained in another type, or if they have an
4860 absolute address. Otherwise, it assumes they have 4 byte alignment.
4861 Specifying this option avoids some rare compatibility problems with code
4862 generated by other compilers. It is not the default because it results
4863 in a performance loss, especially for floating point code.
4865 @item -mno-faster-structs
4866 @itemx -mfaster-structs
4867 With @samp{-mfaster-structs}, the compiler assumes that structures
4868 should have 8 byte alignment. This enables the use of pairs of
4869 @code{ldd} and @code{std} instructions for copies in structure
4870 assignment, in place of twice as many @code{ld} and @code{st} pairs.
4871 However, the use of this changed alignment directly violates the Sparc
4872 ABI. Thus, it's intended only for use on targets where the developer
4873 acknowledges that their resulting code will not be directly in line with
4874 the rules of the ABI.
4878 These two options select variations on the SPARC architecture.
4880 By default (unless specifically configured for the Fujitsu SPARClite),
4881 GCC generates code for the v7 variant of the SPARC architecture.
4883 @samp{-mv8} will give you SPARC v8 code. The only difference from v7
4884 code is that the compiler emits the integer multiply and integer
4885 divide instructions which exist in SPARC v8 but not in SPARC v7.
4887 @samp{-msparclite} will give you SPARClite code. This adds the integer
4888 multiply, integer divide step and scan (@code{ffs}) instructions which
4889 exist in SPARClite but not in SPARC v7.
4891 These options are deprecated and will be deleted in a future GCC release.
4892 They have been replaced with @samp{-mcpu=xxx}.
4896 These two options select the processor for which the code is optimised.
4898 With @samp{-mcypress} (the default), the compiler optimizes code for the
4899 Cypress CY7C602 chip, as used in the SparcStation/SparcServer 3xx series.
4900 This is also appropriate for the older SparcStation 1, 2, IPX etc.
4902 With @samp{-msupersparc} the compiler optimizes code for the SuperSparc cpu, as
4903 used in the SparcStation 10, 1000 and 2000 series. This flag also enables use
4904 of the full SPARC v8 instruction set.
4906 These options are deprecated and will be deleted in a future GCC release.
4907 They have been replaced with @samp{-mcpu=xxx}.
4909 @item -mcpu=@var{cpu_type}
4910 Set the instruction set, register set, and instruction scheduling parameters
4911 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
4912 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
4913 @samp{hypersparc}, @samp{sparclite86x}, @samp{f930}, @samp{f934},
4914 @samp{sparclet}, @samp{tsc701}, @samp{v9}, and @samp{ultrasparc}.
4916 Default instruction scheduling parameters are used for values that select
4917 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
4918 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
4920 Here is a list of each supported architecture and their supported
4925 v8: supersparc, hypersparc
4926 sparclite: f930, f934, sparclite86x
4931 @item -mtune=@var{cpu_type}
4932 Set the instruction scheduling parameters for machine type
4933 @var{cpu_type}, but do not set the instruction set or register set that the
4934 option @samp{-mcpu=}@var{cpu_type} would.
4936 The same values for @samp{-mcpu=}@var{cpu_type} are used for
4937 @samp{-mtune=}@*@var{cpu_type}, though the only useful values are those that
4938 select a particular cpu implementation: @samp{cypress}, @samp{supersparc},
4939 @samp{hypersparc}, @samp{f930}, @samp{f934}, @samp{sparclite86x},
4940 @samp{tsc701}, @samp{ultrasparc}.
4944 These @samp{-m} switches are supported in addition to the above
4945 on the SPARCLET processor.
4948 @item -mlittle-endian
4949 Generate code for a processor running in little-endian mode.
4952 Treat register @code{%g0} as a normal register.
4953 GCC will continue to clobber it as necessary but will not assume
4954 it always reads as 0.
4956 @item -mbroken-saverestore
4957 Generate code that does not use non-trivial forms of the @code{save} and
4958 @code{restore} instructions. Early versions of the SPARCLET processor do
4959 not correctly handle @code{save} and @code{restore} instructions used with
4960 arguments. They correctly handle them used without arguments. A @code{save}
4961 instruction used without arguments increments the current window pointer
4962 but does not allocate a new stack frame. It is assumed that the window
4963 overflow trap handler will properly handle this case as will interrupt
4967 These @samp{-m} switches are supported in addition to the above
4968 on SPARC V9 processors in 64 bit environments.
4971 @item -mlittle-endian
4972 Generate code for a processor running in little-endian mode.
4976 Generate code for a 32 bit or 64 bit environment.
4977 The 32 bit environment sets int, long and pointer to 32 bits.
4978 The 64 bit environment sets int to 32 bits and long and pointer
4981 @item -mcmodel=medlow
4982 Generate code for the Medium/Low code model: the program must be linked
4983 in the low 32 bits of the address space. Pointers are 64 bits.
4984 Programs can be statically or dynamically linked.
4986 @item -mcmodel=medmid
4987 Generate code for the Medium/Middle code model: the program must be linked
4988 in the low 44 bits of the address space, the text segment must be less than
4989 2G bytes, and data segment must be within 2G of the text segment.
4990 Pointers are 64 bits.
4992 @item -mcmodel=medany
4993 Generate code for the Medium/Anywhere code model: the program may be linked
4994 anywhere in the address space, the text segment must be less than
4995 2G bytes, and data segment must be within 2G of the text segment.
4996 Pointers are 64 bits.
4998 @item -mcmodel=embmedany
4999 Generate code for the Medium/Anywhere code model for embedded systems:
5000 assume a 32 bit text and a 32 bit data segment, both starting anywhere
5001 (determined at link time). Register %g4 points to the base of the
5002 data segment. Pointers still 64 bits.
5003 Programs are statically linked, PIC is not supported.
5006 @itemx -mno-stack-bias
5007 With @samp{-mstack-bias}, GCC assumes that the stack pointer, and
5008 frame pointer if present, are offset by -2047 which must be added back
5009 when making stack frame references.
5010 Otherwise, assume no such offset is present.
5013 @node Convex Options
5014 @subsection Convex Options
5015 @cindex Convex options
5017 These @samp{-m} options are defined for Convex:
5021 Generate output for C1. The code will run on any Convex machine.
5022 The preprocessor symbol @code{__convex__c1__} is defined.
5025 Generate output for C2. Uses instructions not available on C1.
5026 Scheduling and other optimizations are chosen for max performance on C2.
5027 The preprocessor symbol @code{__convex_c2__} is defined.
5030 Generate output for C32xx. Uses instructions not available on C1.
5031 Scheduling and other optimizations are chosen for max performance on C32.
5032 The preprocessor symbol @code{__convex_c32__} is defined.
5035 Generate output for C34xx. Uses instructions not available on C1.
5036 Scheduling and other optimizations are chosen for max performance on C34.
5037 The preprocessor symbol @code{__convex_c34__} is defined.
5040 Generate output for C38xx. Uses instructions not available on C1.
5041 Scheduling and other optimizations are chosen for max performance on C38.
5042 The preprocessor symbol @code{__convex_c38__} is defined.
5045 Generate code which puts an argument count in the word preceding each
5046 argument list. This is compatible with regular CC, and a few programs
5047 may need the argument count word. GDB and other source-level debuggers
5048 do not need it; this info is in the symbol table.
5051 Omit the argument count word. This is the default.
5053 @item -mvolatile-cache
5054 Allow volatile references to be cached. This is the default.
5056 @item -mvolatile-nocache
5057 Volatile references bypass the data cache, going all the way to memory.
5058 This is only needed for multi-processor code that does not use standard
5059 synchronization instructions. Making non-volatile references to volatile
5060 locations will not necessarily work.
5063 Type long is 32 bits, the same as type int. This is the default.
5066 Type long is 64 bits, the same as type long long. This option is useless,
5067 because no library support exists for it.
5070 @node AMD29K Options
5071 @subsection AMD29K Options
5072 @cindex AMD29K options
5074 These @samp{-m} options are defined for the AMD Am29000:
5079 @cindex DW bit (29k)
5080 Generate code that assumes the @code{DW} bit is set, i.e., that byte and
5081 halfword operations are directly supported by the hardware. This is the
5086 Generate code that assumes the @code{DW} bit is not set.
5090 @cindex byte writes (29k)
5091 Generate code that assumes the system supports byte and halfword write
5092 operations. This is the default.
5096 Generate code that assumes the systems does not support byte and
5097 halfword write operations. @samp{-mnbw} implies @samp{-mndw}.
5101 @cindex memory model (29k)
5102 Use a small memory model that assumes that all function addresses are
5103 either within a single 256 KB segment or at an absolute address of less
5104 than 256k. This allows the @code{call} instruction to be used instead
5105 of a @code{const}, @code{consth}, @code{calli} sequence.
5109 Use the normal memory model: Generate @code{call} instructions only when
5110 calling functions in the same file and @code{calli} instructions
5111 otherwise. This works if each file occupies less than 256 KB but allows
5112 the entire executable to be larger than 256 KB. This is the default.
5115 Always use @code{calli} instructions. Specify this option if you expect
5116 a single file to compile into more than 256 KB of code.
5120 @cindex processor selection (29k)
5121 Generate code for the Am29050.
5125 Generate code for the Am29000. This is the default.
5127 @item -mkernel-registers
5128 @kindex -mkernel-registers
5129 @cindex kernel and user registers (29k)
5130 Generate references to registers @code{gr64-gr95} instead of to
5131 registers @code{gr96-gr127}. This option can be used when compiling
5132 kernel code that wants a set of global registers disjoint from that used
5135 Note that when this option is used, register names in @samp{-f} flags
5136 must use the normal, user-mode, names.
5138 @item -muser-registers
5139 @kindex -muser-registers
5140 Use the normal set of global registers, @code{gr96-gr127}. This is the
5144 @itemx -mno-stack-check
5145 @kindex -mstack-check
5146 @cindex stack checks (29k)
5147 Insert (or do not insert) a call to @code{__msp_check} after each stack
5148 adjustment. This is often used for kernel code.
5151 @itemx -mno-storem-bug
5152 @kindex -mstorem-bug
5153 @cindex storem bug (29k)
5154 @samp{-mstorem-bug} handles 29k processors which cannot handle the
5155 separation of a mtsrim insn and a storem instruction (most 29000 chips
5156 to date, but not the 29050).
5158 @item -mno-reuse-arg-regs
5159 @itemx -mreuse-arg-regs
5160 @kindex -mreuse-arg-regs
5161 @samp{-mno-reuse-arg-regs} tells the compiler to only use incoming argument
5162 registers for copying out arguments. This helps detect calling a function
5163 with fewer arguments than it was declared with.
5165 @item -mno-impure-text
5166 @itemx -mimpure-text
5167 @kindex -mimpure-text
5168 @samp{-mimpure-text}, used in addition to @samp{-shared}, tells the compiler to
5169 not pass @samp{-assert pure-text} to the linker when linking a shared object.
5172 @kindex -msoft-float
5173 Generate output containing library calls for floating point.
5174 @strong{Warning:} the requisite libraries are not part of GCC.
5175 Normally the facilities of the machine's usual C compiler are used, but
5176 this can't be done directly in cross-compilation. You must make your
5177 own arrangements to provide suitable library functions for
5182 Do not generate multm or multmu instructions. This is useful for some embedded
5183 systems which do not have trap handlers for these instructions.
5187 @subsection ARM Options
5190 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
5195 @kindex -mapcs-frame
5196 Generate a stack frame that is compliant with the ARM Procedure Call
5197 Standard for all functions, even if this is not strictly necessary for
5198 correct execution of the code. Specifying @samp{-fomit-frame-pointer}
5199 with this option will cause the stack frames not to be generated for
5200 leaf functions. The default is @samp{-mno-apcs-frame}.
5204 This is a synonym for @samp{-mapcs-frame}.
5208 Generate code for a processor running with a 26-bit program counter,
5209 and conforming to the function calling standards for the APCS 26-bit
5210 option. This option replaces the @samp{-m2} and @samp{-m3} options
5211 of previous releases of the compiler.
5215 Generate code for a processor running with a 32-bit program counter,
5216 and conforming to the function calling standards for the APCS 32-bit
5217 option. This option replaces the @samp{-m6} option of previous releases
5220 @item -mapcs-stack-check
5221 @kindex -mapcs-stack-check
5222 @kindex -mno-apcs-stack-check
5223 Generate code to check the amount of stack space available upon entry to
5224 every function (that actually uses some stack space). If there is
5225 insufficient space available then either the function
5226 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
5227 called, depending upon the amount of stack space required. The run time
5228 system is required to provide these functions. The default is
5229 @samp{-mno-apcs-stack-check}, since this produces smaller code.
5232 @kindex -mapcs-float
5233 @kindex -mno-apcs-float
5234 Pass floating point arguments using the float point registers. This is
5235 one of the variants of the APCS. This option is recommended if the
5236 target hardware has a floating point unit or if a lot of floating point
5237 arithmetic is going to be performed by the code. The default is
5238 @samp{-mno-apcs-float}, since integer only code is slightly increased in
5239 size if @samp{-mapcs-float} is used.
5241 @item -mapcs-reentrant
5242 @kindex -mapcs-reentrant
5243 @kindex -mno-apcs-reentrant
5244 Generate reentrant, position independent code. This is the equivalent
5245 to specifying the @samp{-fpic} option. The default is
5246 @samp{-mno-apcs-reentrant}.
5248 @item -mthumb-interwork
5249 @kindex -mthumb-interwork
5250 @kindex -mno-thumb-interwork
5251 Generate code which supports calling between the ARM and THUMB
5252 instruction sets. Without this option the two instruction sets cannot
5253 be reliably used inside one program. The default is
5254 @samp{-mno-thumb-interwork}, since slightly larger code is generated
5255 when @samp{-mthumb-interwork} is specified.
5257 @item -mno-sched-prolog
5258 @kindex -mno-sched-prolog
5259 @kindex -msched-prolog
5260 Prevent the reordering of instructions in the function prolog, or the
5261 merging of those instruction with the instructions in the function's
5262 body. This means that all functions will start with a recognizable set
5263 of instructions (or in fact one of a choice from a small set of
5264 different function prologues), and this information can be used to
5265 locate the start if functions inside an executable piece of code. The
5266 default is @samp{-msched-prolog}.
5269 Generate output containing floating point instructions. This is the
5273 Generate output containing library calls for floating point.
5274 @strong{Warning:} the requisite libraries are not available for all ARM
5275 targets. Normally the facilities of the machine's usual C compiler are
5276 used, but this cannot be done directly in cross-compilation. You must make
5277 your own arrangements to provide suitable library functions for
5280 @samp{-msoft-float} changes the calling convention in the output file;
5281 therefore, it is only useful if you compile @emph{all} of a program with
5282 this option. In particular, you need to compile @file{libgcc.a}, the
5283 library that comes with GCC, with @samp{-msoft-float} in order for
5286 @item -mlittle-endian
5287 Generate code for a processor running in little-endian mode. This is
5288 the default for all standard configurations.
5291 Generate code for a processor running in big-endian mode; the default is
5292 to compile code for a little-endian processor.
5294 @item -mwords-little-endian
5295 This option only applies when generating code for big-endian processors.
5296 Generate code for a little-endian word order but a big-endian byte
5297 order. That is, a byte order of the form @samp{32107654}. Note: this
5298 option should only be used if you require compatibility with code for
5299 big-endian ARM processors generated by versions of the compiler prior to
5302 @item -malignment-traps
5303 @kindex -malignment-traps
5304 Generate code that will not trap if the MMU has alignment traps enabled.
5305 On ARM architectures prior to ARMv4, there were no instructions to
5306 access half-word objects stored in memory. However, when reading from
5307 memory a feature of the ARM architecture allows a word load to be used,
5308 even if the address is unaligned, and the processor core will rotate the
5309 data as it is being loaded. This option tells the compiler that such
5310 misaligned accesses will cause a MMU trap and that it should instead
5311 synthesise the access as a series of byte accesses. The compiler can
5312 still use word accesses to load half-word data if it knows that the
5313 address is aligned to a word boundary.
5315 This option is ignored when compiling for ARM architecture 4 or later,
5316 since these processors have instructions to directly access half-word
5319 @item -mno-alignment-traps
5320 @kindex -mno-alignment-traps
5321 Generate code that assumes that the MMU will not trap unaligned
5322 accesses. This produces better code when the target instruction set
5323 does not have half-word memory operations (implementations prior to
5326 Note that you cannot use this option to access unaligned word objects,
5327 since the processor will only fetch one 32-bit aligned object from
5330 The default setting for most targets is -mno-alignment-traps, since
5331 this produces better code when there are no half-word memory
5332 instructions available.
5334 @item -mshort-load-bytes
5335 @kindex -mshort-load-bytes
5336 This is a deprecated alias for @samp{-malignment-traps}.
5338 @item -mno-short-load-bytes
5339 @kindex -mno-short-load-bytes
5340 This is a deprecated alias for @samp{-mno-alignment-traps}.
5342 @item -mshort-load-words
5343 @kindex -mshort-load-words
5344 This is a deprecated alias for @samp{-mno-alignment-traps}.
5346 @item -mno-short-load-words
5347 @kindex -mno-short-load-words
5348 This is a deprecated alias for @samp{-malignment-traps}.
5352 This option only applies to RISC iX. Emulate the native BSD-mode
5353 compiler. This is the default if @samp{-ansi} is not specified.
5357 This option only applies to RISC iX. Emulate the native X/Open-mode
5360 @item -mno-symrename
5361 @kindex -mno-symrename
5362 This option only applies to RISC iX. Do not run the assembler
5363 post-processor, @samp{symrename}, after code has been assembled.
5364 Normally it is necessary to modify some of the standard symbols in
5365 preparation for linking with the RISC iX C library; this option
5366 suppresses this pass. The post-processor is never run when the
5367 compiler is built for cross-compilation.
5371 This specifies the name of the target ARM processor. GCC uses this name
5372 to determine what kind of instructions it can use when generating
5373 assembly code. Permissible names are: arm2, arm250, arm3, arm6, arm60,
5374 arm600, arm610, arm620, arm7, arm7m, arm7d, arm7dm, arm7di, arm7dmi,
5375 arm70, arm700, arm700i, arm710, arm710c, arm7100, arm7500, arm7500fe,
5376 arm7tdmi, arm8, strongarm, strongarm110, strongarm1100, arm8, arm810,
5377 arm9, arm920, arm920t, arm9tdmi.
5379 @itemx -mtune=<name>
5381 This option is very similar to the @samp{-mcpu=} option, except that
5382 instead of specifying the actual target processor type, and hence
5383 restricting which instructions can be used, it specifies that GCC should
5384 tune the performance of the code as if the target were of the type
5385 specified in this option, but still choosing the instructions that it
5386 will generate based on the cpu specified by a @samp{-mcpu=} option.
5387 For some arm implementations better performance can be obtained by using
5392 This specifies the name of the target ARM architecture. GCC uses this
5393 name to determine what kind of instructions it can use when generating
5394 assembly code. This option can be used in conjunction with or instead
5395 of the @samp{-mcpu=} option. Permissible names are: armv2, armv2a,
5396 armv3, armv3m, armv4, armv4t, armv5.
5398 @item -mfpe=<number>
5399 @itemx -mfp=<number>
5402 This specifies the version of the floating point emulation available on
5403 the target. Permissible values are 2 and 3. @samp{-mfp=} is a synonym
5404 for @samp{-mfpe=} to support older versions of GCC.
5406 @item -mstructure-size-boundary=<n>
5407 @kindex -mstructure-size-boundary
5408 The size of all structures and unions will be rounded up to a multiple
5409 of the number of bits set by this option. Permissible values are 8 and
5410 32. The default value varies for different toolchains. For the COFF
5411 targeted toolchain the default value is 8. Specifying the larger number
5412 can produce faster, more efficient code, but can also increase the size
5413 of the program. The two values are potentially incompatible. Code
5414 compiled with one value cannot necessarily expect to work with code or
5415 libraries compiled with the other value, if they exchange information
5416 using structures or unions. Programmers are encouraged to use the 32
5417 value as future versions of the toolchain may default to this value.
5419 @item -mabort-on-noreturn
5420 @kindex -mabort-on-noreturn
5421 @kindex -mnoabort-on-noreturn
5422 Generate a call to the function abort at the end of a noreturn function.
5423 It will be executed if the function tries to return.
5426 @itemx -mno-long-calls
5427 Tells the compiler to perform function calls by first loading the
5428 address of the function into a register and then performing a subroutine
5429 call on this register. This switch is needed if the target function
5430 will lie outside of the 64 megabyte addressing range of the offset based
5431 version of subroutine call instruction.
5433 Even if this switch is enabled, not all function calls will be turned
5434 into long calls. The heuristic is that static functions, functions
5435 which have the @samp{short-call} attribute, functions that are inside
5436 the scope of a @samp{#pragma no_long_calls} directive and functions whose
5437 definitions have already been compiled within the current compilation
5438 unit, will not be turned into long calls. The exception to this rule is
5439 that weak function definitions, functions with the @samp{long-call}
5440 attribute or the @samp{section} attribute, and functions that are within
5441 the scope of a @samp{#pragma long_calls} directive, will always be
5442 turned into long calls.
5444 This feature is not enabled by default. Specifying
5445 @samp{--no-long-calls} will restore the default behaviour, as will
5446 placing the function calls within the scope of a @samp{#pragma
5447 long_calls_off} directive. Note these switches have no effect on how
5448 the compiler generates code to handle function calls via function
5451 @item -mnop-fun-dllimport
5452 @kindex -mnop-fun-dllimport
5453 Disable the support for the @emph{dllimport} attribute.
5455 @item -msingle-pic-base
5456 @kindex -msingle-pic-base
5457 Treat the register used for PIC addressing as read-only, rather than
5458 loading it in the prologue for each function. The run-time system is
5459 responsible for initialising this register with an appropriate value
5460 before execution begins.
5462 @item -mpic-register=<reg>
5463 @kindex -mpic-register=
5464 Specify the register to be used for PIC addressing. The default is R10
5465 unless stack-checking is enabled, when R9 is used.
5470 @subsection Thumb Options
5471 @cindex Thumb Options
5475 @item -mthumb-interwork
5476 @kindex -mthumb-interwork
5477 @kindex -mno-thumb-interwork
5478 Generate code which supports calling between the THUMB and ARM
5479 instruction sets. Without this option the two instruction sets cannot
5480 be reliably used inside one program. The default is
5481 @samp{-mno-thumb-interwork}, since slightly smaller code is generated
5485 @kindex -mtpcs-frame
5486 @kindex -mno-tpcs-frame
5487 Generate a stack frame that is compliant with the Thumb Procedure Call
5488 Standard for all non-leaf functions. (A leaf function is one that does
5489 not call any other functions). The default is @samp{-mno-apcs-frame}.
5491 @item -mtpcs-leaf-frame
5492 @kindex -mtpcs-leaf-frame
5493 @kindex -mno-tpcs-leaf-frame
5494 Generate a stack frame that is compliant with the Thumb Procedure Call
5495 Standard for all leaf functions. (A leaf function is one that does
5496 not call any other functions). The default is @samp{-mno-apcs-leaf-frame}.
5498 @item -mlittle-endian
5499 @kindex -mlittle-endian
5500 Generate code for a processor running in little-endian mode. This is
5501 the default for all standard configurations.
5504 @kindex -mbig-endian
5505 Generate code for a processor running in big-endian mode.
5507 @item -mstructure-size-boundary=<n>
5508 @kindex -mstructure-size-boundary
5509 The size of all structures and unions will be rounded up to a multiple
5510 of the number of bits set by this option. Permissible values are 8 and
5511 32. The default value varies for different toolchains. For the COFF
5512 targeted toolchain the default value is 8. Specifying the larger number
5513 can produced faster, more efficient code, but can also increase the size
5514 of the program. The two values are potentially incompatible. Code
5515 compiled with one value cannot necessarily expect to work with code or
5516 libraries compiled with the other value, if they exchange information
5517 using structures or unions. Programmers are encouraged to use the 32
5518 value as future versions of the toolchain may default to this value.
5520 @item -mnop-fun-dllimport
5521 @kindex -mnop-fun-dllimport
5522 Disable the support for the @emph{dllimport} attribute.
5524 @item -mcallee-super-interworking
5525 @kindex -mcallee-super-interworking
5526 Gives all externally visible functions in the file being compiled an ARM
5527 instruction set header which switches to Thumb mode before executing the
5528 rest of the function. This allows these functions to be called from
5529 non-interworking code.
5531 @item -mcaller-super-interworking
5532 @kindex -mcaller-super-interworking
5533 Allows calls via function pointers (including virtual functions) to
5534 execute correctly regardless of whether the target code has been
5535 compiled for interworking or not. There is a small overhead in the cost
5536 of executing a function pointer if this option is enabled.
5538 @item -msingle-pic-base
5539 @kindex -msingle-pic-base
5540 Treat the register used for PIC addressing as read-only, rather than
5541 loading it in the prologue for each function. The run-time system is
5542 responsible for initialising this register with an appropriate value
5543 before execution begins.
5545 @item -mpic-register=<reg>
5546 @kindex -mpic-register=
5547 Specify the register to be used for PIC addressing. The default is R10.
5551 @node MN10200 Options
5552 @subsection MN10200 Options
5553 @cindex MN10200 options
5554 These @samp{-m} options are defined for Matsushita MN10200 architectures:
5558 Indicate to the linker that it should perform a relaxation optimization pass
5559 to shorten branches, calls and absolute memory addresses. This option only
5560 has an effect when used on the command line for the final link step.
5562 This option makes symbolic debugging impossible.
5565 @node MN10300 Options
5566 @subsection MN10300 Options
5567 @cindex MN10300 options
5568 These @samp{-m} options are defined for Matsushita MN10300 architectures:
5572 Generate code to avoid bugs in the multiply instructions for the MN10300
5573 processors. This is the default.
5576 Do not generate code to avoid bugs in the multiply instructions for the
5580 Generate code which uses features specific to the AM33 processor.
5583 Do not generate code which uses features specific to the AM33 processor. This
5587 Indicate to the linker that it should perform a relaxation optimization pass
5588 to shorten branches, calls and absolute memory addresses. This option only
5589 has an effect when used on the command line for the final link step.
5591 This option makes symbolic debugging impossible.
5595 @node M32R/D Options
5596 @subsection M32R/D Options
5597 @cindex M32R/D options
5599 These @samp{-m} options are defined for Mitsubishi M32R/D architectures:
5602 @item -mcode-model=small
5603 Assume all objects live in the lower 16MB of memory (so that their addresses
5604 can be loaded with the @code{ld24} instruction), and assume all subroutines
5605 are reachable with the @code{bl} instruction.
5606 This is the default.
5608 The addressability of a particular object can be set with the
5609 @code{model} attribute.
5611 @item -mcode-model=medium
5612 Assume objects may be anywhere in the 32 bit address space (the compiler
5613 will generate @code{seth/add3} instructions to load their addresses), and
5614 assume all subroutines are reachable with the @code{bl} instruction.
5616 @item -mcode-model=large
5617 Assume objects may be anywhere in the 32 bit address space (the compiler
5618 will generate @code{seth/add3} instructions to load their addresses), and
5619 assume subroutines may not be reachable with the @code{bl} instruction
5620 (the compiler will generate the much slower @code{seth/add3/jl}
5621 instruction sequence).
5624 Disable use of the small data area. Variables will be put into
5625 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
5626 @code{section} attribute has been specified).
5627 This is the default.
5629 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
5630 Objects may be explicitly put in the small data area with the
5631 @code{section} attribute using one of these sections.
5634 Put small global and static data in the small data area, but do not
5635 generate special code to reference them.
5638 Put small global and static data in the small data area, and generate
5639 special instructions to reference them.
5642 @cindex smaller data references
5643 Put global and static objects less than or equal to @var{num} bytes
5644 into the small data or bss sections instead of the normal data or bss
5645 sections. The default value of @var{num} is 8.
5646 The @samp{-msdata} option must be set to one of @samp{sdata} or @samp{use}
5647 for this option to have any effect.
5649 All modules should be compiled with the same @samp{-G @var{num}} value.
5650 Compiling with different values of @var{num} may or may not work; if it
5651 doesn't the linker will give an error message - incorrect code will not be
5657 @subsection M88K Options
5658 @cindex M88k options
5660 These @samp{-m} options are defined for Motorola 88k architectures:
5665 Generate code that works well on both the m88100 and the
5670 Generate code that works best for the m88100, but that also
5675 Generate code that works best for the m88110, and may not run
5680 Obsolete option to be removed from the next revision.
5683 @item -midentify-revision
5684 @kindex -midentify-revision
5686 @cindex identifying source, compiler (88k)
5687 Include an @code{ident} directive in the assembler output recording the
5688 source file name, compiler name and version, timestamp, and compilation
5691 @item -mno-underscores
5692 @kindex -mno-underscores
5693 @cindex underscores, avoiding (88k)
5694 In assembler output, emit symbol names without adding an underscore
5695 character at the beginning of each name. The default is to use an
5696 underscore as prefix on each name.
5698 @item -mocs-debug-info
5699 @itemx -mno-ocs-debug-info
5700 @kindex -mocs-debug-info
5701 @kindex -mno-ocs-debug-info
5703 @cindex debugging, 88k OCS
5704 Include (or omit) additional debugging information (about registers used
5705 in each stack frame) as specified in the 88open Object Compatibility
5706 Standard, ``OCS''. This extra information allows debugging of code that
5707 has had the frame pointer eliminated. The default for DG/UX, SVr4, and
5708 Delta 88 SVr3.2 is to include this information; other 88k configurations
5709 omit this information by default.
5711 @item -mocs-frame-position
5712 @kindex -mocs-frame-position
5713 @cindex register positions in frame (88k)
5714 When emitting COFF debugging information for automatic variables and
5715 parameters stored on the stack, use the offset from the canonical frame
5716 address, which is the stack pointer (register 31) on entry to the
5717 function. The DG/UX, SVr4, Delta88 SVr3.2, and BCS configurations use
5718 @samp{-mocs-frame-position}; other 88k configurations have the default
5719 @samp{-mno-ocs-frame-position}.
5721 @item -mno-ocs-frame-position
5722 @kindex -mno-ocs-frame-position
5723 @cindex register positions in frame (88k)
5724 When emitting COFF debugging information for automatic variables and
5725 parameters stored on the stack, use the offset from the frame pointer
5726 register (register 30). When this option is in effect, the frame
5727 pointer is not eliminated when debugging information is selected by the
5730 @item -moptimize-arg-area
5731 @itemx -mno-optimize-arg-area
5732 @kindex -moptimize-arg-area
5733 @kindex -mno-optimize-arg-area
5734 @cindex arguments in frame (88k)
5735 Control how function arguments are stored in stack frames.
5736 @samp{-moptimize-arg-area} saves space by optimizing them, but this
5737 conflicts with the 88open specifications. The opposite alternative,
5738 @samp{-mno-optimize-arg-area}, agrees with 88open standards. By default
5739 GCC does not optimize the argument area.
5741 @item -mshort-data-@var{num}
5742 @kindex -mshort-data-@var{num}
5743 @cindex smaller data references (88k)
5744 @cindex r0-relative references (88k)
5745 Generate smaller data references by making them relative to @code{r0},
5746 which allows loading a value using a single instruction (rather than the
5747 usual two). You control which data references are affected by
5748 specifying @var{num} with this option. For example, if you specify
5749 @samp{-mshort-data-512}, then the data references affected are those
5750 involving displacements of less than 512 bytes.
5751 @samp{-mshort-data-@var{num}} is not effective for @var{num} greater
5754 @item -mserialize-volatile
5755 @kindex -mserialize-volatile
5756 @itemx -mno-serialize-volatile
5757 @kindex -mno-serialize-volatile
5758 @cindex sequential consistency on 88k
5759 Do, or don't, generate code to guarantee sequential consistency
5760 of volatile memory references. By default, consistency is
5763 The order of memory references made by the MC88110 processor does
5764 not always match the order of the instructions requesting those
5765 references. In particular, a load instruction may execute before
5766 a preceding store instruction. Such reordering violates
5767 sequential consistency of volatile memory references, when there
5768 are multiple processors. When consistency must be guaranteed,
5769 GNU C generates special instructions, as needed, to force
5770 execution in the proper order.
5772 The MC88100 processor does not reorder memory references and so
5773 always provides sequential consistency. However, by default, GNU
5774 C generates the special instructions to guarantee consistency
5775 even when you use @samp{-m88100}, so that the code may be run on an
5776 MC88110 processor. If you intend to run your code only on the
5777 MC88100 processor, you may use @samp{-mno-serialize-volatile}.
5779 The extra code generated to guarantee consistency may affect the
5780 performance of your application. If you know that you can safely
5781 forgo this guarantee, you may use @samp{-mno-serialize-volatile}.
5787 @cindex assembler syntax, 88k
5789 Turn on (@samp{-msvr4}) or off (@samp{-msvr3}) compiler extensions
5790 related to System V release 4 (SVr4). This controls the following:
5794 Which variant of the assembler syntax to emit.
5796 @samp{-msvr4} makes the C preprocessor recognize @samp{#pragma weak}
5797 that is used on System V release 4.
5799 @samp{-msvr4} makes GCC issue additional declaration directives used in
5803 @samp{-msvr4} is the default for the m88k-motorola-sysv4 and
5804 m88k-dg-dgux m88k configurations. @samp{-msvr3} is the default for all
5805 other m88k configurations.
5807 @item -mversion-03.00
5808 @kindex -mversion-03.00
5809 This option is obsolete, and is ignored.
5810 @c ??? which asm syntax better for GAS? option there too?
5812 @item -mno-check-zero-division
5813 @itemx -mcheck-zero-division
5814 @kindex -mno-check-zero-division
5815 @kindex -mcheck-zero-division
5816 @cindex zero division on 88k
5817 Do, or don't, generate code to guarantee that integer division by
5818 zero will be detected. By default, detection is guaranteed.
5820 Some models of the MC88100 processor fail to trap upon integer
5821 division by zero under certain conditions. By default, when
5822 compiling code that might be run on such a processor, GNU C
5823 generates code that explicitly checks for zero-valued divisors
5824 and traps with exception number 503 when one is detected. Use of
5825 mno-check-zero-division suppresses such checking for code
5826 generated to run on an MC88100 processor.
5828 GNU C assumes that the MC88110 processor correctly detects all
5829 instances of integer division by zero. When @samp{-m88110} is
5830 specified, both @samp{-mcheck-zero-division} and
5831 @samp{-mno-check-zero-division} are ignored, and no explicit checks for
5832 zero-valued divisors are generated.
5834 @item -muse-div-instruction
5835 @kindex -muse-div-instruction
5836 @cindex divide instruction, 88k
5837 Use the div instruction for signed integer division on the
5838 MC88100 processor. By default, the div instruction is not used.
5840 On the MC88100 processor the signed integer division instruction
5841 div) traps to the operating system on a negative operand. The
5842 operating system transparently completes the operation, but at a
5843 large cost in execution time. By default, when compiling code
5844 that might be run on an MC88100 processor, GNU C emulates signed
5845 integer division using the unsigned integer division instruction
5846 divu), thereby avoiding the large penalty of a trap to the
5847 operating system. Such emulation has its own, smaller, execution
5848 cost in both time and space. To the extent that your code's
5849 important signed integer division operations are performed on two
5850 nonnegative operands, it may be desirable to use the div
5851 instruction directly.
5853 On the MC88110 processor the div instruction (also known as the
5854 divs instruction) processes negative operands without trapping to
5855 the operating system. When @samp{-m88110} is specified,
5856 @samp{-muse-div-instruction} is ignored, and the div instruction is used
5857 for signed integer division.
5859 Note that the result of dividing INT_MIN by -1 is undefined. In
5860 particular, the behavior of such a division with and without
5861 @samp{-muse-div-instruction} may differ.
5863 @item -mtrap-large-shift
5864 @itemx -mhandle-large-shift
5865 @kindex -mtrap-large-shift
5866 @kindex -mhandle-large-shift
5867 @cindex bit shift overflow (88k)
5868 @cindex large bit shifts (88k)
5869 Include code to detect bit-shifts of more than 31 bits; respectively,
5870 trap such shifts or emit code to handle them properly. By default GCC
5871 makes no special provision for large bit shifts.
5873 @item -mwarn-passed-structs
5874 @kindex -mwarn-passed-structs
5875 @cindex structure passing (88k)
5876 Warn when a function passes a struct as an argument or result.
5877 Structure-passing conventions have changed during the evolution of the C
5878 language, and are often the source of portability problems. By default,
5879 GCC issues no such warning.
5882 @node RS/6000 and PowerPC Options
5883 @subsection IBM RS/6000 and PowerPC Options
5884 @cindex RS/6000 and PowerPC Options
5885 @cindex IBM RS/6000 and PowerPC Options
5887 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
5895 @itemx -mpowerpc-gpopt
5896 @itemx -mno-powerpc-gpopt
5897 @itemx -mpowerpc-gfxopt
5898 @itemx -mno-powerpc-gfxopt
5900 @itemx -mno-powerpc64
5904 @kindex -mpowerpc-gpopt
5905 @kindex -mpowerpc-gfxopt
5907 GCC supports two related instruction set architectures for the
5908 RS/6000 and PowerPC. The @dfn{POWER} instruction set are those
5909 instructions supported by the @samp{rios} chip set used in the original
5910 RS/6000 systems and the @dfn{PowerPC} instruction set is the
5911 architecture of the Motorola MPC5xx, MPC6xx, MPC8xx microprocessors, and
5912 the IBM 4xx microprocessors.
5914 Neither architecture is a subset of the other. However there is a
5915 large common subset of instructions supported by both. An MQ
5916 register is included in processors supporting the POWER architecture.
5918 You use these options to specify which instructions are available on the
5919 processor you are using. The default value of these options is
5920 determined when configuring GCC. Specifying the
5921 @samp{-mcpu=@var{cpu_type}} overrides the specification of these
5922 options. We recommend you use the @samp{-mcpu=@var{cpu_type}} option
5923 rather than the options listed above.
5925 The @samp{-mpower} option allows GCC to generate instructions that
5926 are found only in the POWER architecture and to use the MQ register.
5927 Specifying @samp{-mpower2} implies @samp{-power} and also allows GCC
5928 to generate instructions that are present in the POWER2 architecture but
5929 not the original POWER architecture.
5931 The @samp{-mpowerpc} option allows GCC to generate instructions that
5932 are found only in the 32-bit subset of the PowerPC architecture.
5933 Specifying @samp{-mpowerpc-gpopt} implies @samp{-mpowerpc} and also allows
5934 GCC to use the optional PowerPC architecture instructions in the
5935 General Purpose group, including floating-point square root. Specifying
5936 @samp{-mpowerpc-gfxopt} implies @samp{-mpowerpc} and also allows GCC to
5937 use the optional PowerPC architecture instructions in the Graphics
5938 group, including floating-point select.
5940 The @samp{-mpowerpc64} option allows GCC to generate the additional
5941 64-bit instructions that are found in the full PowerPC64 architecture
5942 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
5943 @samp{-mno-powerpc64}.
5945 If you specify both @samp{-mno-power} and @samp{-mno-powerpc}, GCC
5946 will use only the instructions in the common subset of both
5947 architectures plus some special AIX common-mode calls, and will not use
5948 the MQ register. Specifying both @samp{-mpower} and @samp{-mpowerpc}
5949 permits GCC to use any instruction from either architecture and to
5950 allow use of the MQ register; specify this for the Motorola MPC601.
5952 @item -mnew-mnemonics
5953 @itemx -mold-mnemonics
5954 @kindex -mnew-mnemonics
5955 @kindex -mold-mnemonics
5956 Select which mnemonics to use in the generated assembler code.
5957 @samp{-mnew-mnemonics} requests output that uses the assembler mnemonics
5958 defined for the PowerPC architecture, while @samp{-mold-mnemonics}
5959 requests the assembler mnemonics defined for the POWER architecture.
5960 Instructions defined in only one architecture have only one mnemonic;
5961 GCC uses that mnemonic irrespective of which of these options is
5964 GCC defaults to the mnemonics appropriate for the architecture in
5965 use. Specifying @samp{-mcpu=@var{cpu_type}} sometimes overrides the
5966 value of these option. Unless you are building a cross-compiler, you
5967 should normally not specify either @samp{-mnew-mnemonics} or
5968 @samp{-mold-mnemonics}, but should instead accept the default.
5970 @item -mcpu=@var{cpu_type}
5972 Set architecture type, register usage, choice of mnemonics, and
5973 instruction scheduling parameters for machine type @var{cpu_type}.
5974 Supported values for @var{cpu_type} are @samp{rios}, @samp{rios1},
5975 @samp{rsc}, @samp{rios2}, @samp{rs64a}, @samp{601}, @samp{602},
5976 @samp{603}, @samp{603e}, @samp{604}, @samp{604e}, @samp{620},
5977 @samp{630}, @samp{740}, @samp{750}, @samp{power}, @samp{power2},
5978 @samp{powerpc}, @samp{403}, @samp{505}, @samp{801}, @samp{821},
5979 @samp{823}, and @samp{860} and @samp{common}. @samp{-mcpu=power},
5980 @samp{-mcpu=power2}, @samp{-mcpu=powerpc}, and @samp{-mcpu=powerpc64}
5981 specify generic POWER, POWER2, pure 32-bit PowerPC (i.e., not MPC601),
5982 and 64-bit PowerPC architecture machine types, with an appropriate,
5983 generic processor model assumed for scheduling purposes.@refill
5985 Specifying any of the following options:
5986 @samp{-mcpu=rios1}, @samp{-mcpu=rios2}, @samp{-mcpu=rsc},
5987 @samp{-mcpu=power}, or @samp{-mcpu=power2}
5988 enables the @samp{-mpower} option and disables the @samp{-mpowerpc} option;
5989 @samp{-mcpu=601} enables both the @samp{-mpower} and @samp{-mpowerpc} options.
5990 All of @samp{-mcpu=rs64a}, @samp{-mcpu=602}, @samp{-mcpu=603},
5991 @samp{-mcpu=603e}, @samp{-mcpu=604}, @samp{-mcpu=620}, @samp{-mcpu=630},
5992 @samp{-mcpu=740}, and @samp{-mcpu=750}
5993 enable the @samp{-mpowerpc} option and disable the @samp{-mpower} option.
5994 Exactly similarly, all of @samp{-mcpu=403},
5995 @samp{-mcpu=505}, @samp{-mcpu=821}, @samp{-mcpu=860} and @samp{-mcpu=powerpc}
5996 enable the @samp{-mpowerpc} option and disable the @samp{-mpower} option.
5997 @samp{-mcpu=common} disables both the
5998 @samp{-mpower} and @samp{-mpowerpc} options.@refill
6000 AIX versions 4 or greater selects @samp{-mcpu=common} by default, so
6001 that code will operate on all members of the RS/6000 POWER and PowerPC
6002 families. In that case, GCC will use only the instructions in the
6003 common subset of both architectures plus some special AIX common-mode
6004 calls, and will not use the MQ register. GCC assumes a generic
6005 processor model for scheduling purposes.
6007 Specifying any of the options @samp{-mcpu=rios1}, @samp{-mcpu=rios2},
6008 @samp{-mcpu=rsc}, @samp{-mcpu=power}, or @samp{-mcpu=power2} also
6009 disables the @samp{new-mnemonics} option. Specifying @samp{-mcpu=601},
6010 @samp{-mcpu=602}, @samp{-mcpu=603}, @samp{-mcpu=603e}, @samp{-mcpu=604},
6011 @samp{-mcpu=620}, @samp{-mcpu=630}, @samp{-mcpu=403}, @samp{-mcpu=505},
6012 @samp{-mcpu=821}, @samp{-mcpu=860} or @samp{-mcpu=powerpc} also enables
6013 the @samp{new-mnemonics} option.@refill
6015 Specifying @samp{-mcpu=403}, @samp{-mcpu=821}, or @samp{-mcpu=860} also
6016 enables the @samp{-msoft-float} option.
6018 @item -mtune=@var{cpu_type}
6019 Set the instruction scheduling parameters for machine type
6020 @var{cpu_type}, but do not set the architecture type, register usage,
6021 choice of mnemonics like @samp{-mcpu=}@var{cpu_type} would. The same
6022 values for @var{cpu_type} are used for @samp{-mtune=}@var{cpu_type} as
6023 for @samp{-mcpu=}@var{cpu_type}. The @samp{-mtune=}@var{cpu_type}
6024 option overrides the @samp{-mcpu=}@var{cpu_type} option in terms of
6025 instruction scheduling parameters.
6028 @itemx -mno-fp-in-toc
6029 @itemx -mno-sum-in-toc
6030 @itemx -mminimal-toc
6031 @kindex -mminimal-toc
6032 Modify generation of the TOC (Table Of Contents), which is created for
6033 every executable file. The @samp{-mfull-toc} option is selected by
6034 default. In that case, GCC will allocate at least one TOC entry for
6035 each unique non-automatic variable reference in your program. GCC
6036 will also place floating-point constants in the TOC. However, only
6037 16,384 entries are available in the TOC.
6039 If you receive a linker error message that saying you have overflowed
6040 the available TOC space, you can reduce the amount of TOC space used
6041 with the @samp{-mno-fp-in-toc} and @samp{-mno-sum-in-toc} options.
6042 @samp{-mno-fp-in-toc} prevents GCC from putting floating-point
6043 constants in the TOC and @samp{-mno-sum-in-toc} forces GCC to
6044 generate code to calculate the sum of an address and a constant at
6045 run-time instead of putting that sum into the TOC. You may specify one
6046 or both of these options. Each causes GCC to produce very slightly
6047 slower and larger code at the expense of conserving TOC space.
6049 If you still run out of space in the TOC even when you specify both of
6050 these options, specify @samp{-mminimal-toc} instead. This option causes
6051 GCC to make only one TOC entry for every file. When you specify this
6052 option, GCC will produce code that is slower and larger but which
6053 uses extremely little TOC space. You may wish to use this option
6054 only on files that contain less frequently executed code. @refill
6060 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
6061 @code{long} type, and the infrastructure needed to support them.
6062 Specifying @samp{-maix64} implies @samp{-mpowerpc64} and
6063 @samp{-mpowerpc}, while @samp{-maix32} disables the 64-bit ABI and
6064 implies @samp{-mno-powerpc64}. GCC defaults to @samp{-maix32}.
6069 On AIX, pass floating-point arguments to prototyped functions beyond the
6070 register save area (RSA) on the stack in addition to argument FPRs. The
6071 AIX calling convention was extended but not initially documented to
6072 handle an obscure K&R C case of calling a function that takes the
6073 address of its arguments with fewer arguments than declared. AIX XL
6074 compilers access floating point arguments which do not fit in the
6075 RSA from the stack when a subroutine is compiled without
6076 optimization. Because always storing floating-point arguments on the
6077 stack is inefficient and rarely needed, this option is not enabled by
6078 default and only is necessary when calling subroutines compiled by AIX
6079 XL compilers without optimization.
6083 Support @dfn{AIX Threads}. Link an application written to use
6084 @dfn{pthreads} with special libraries and startup code to enable the
6089 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE). Link an
6090 application written to use message passing with special startup code to
6091 enable the application to run. The system must have PE installed in the
6092 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
6093 must be overridden with the @samp{-specs=} option to specify the
6094 appropriate directory location. The Parallel Environment does not
6095 support threads, so the @samp{-mpe} option and the @samp{-mthreads}
6096 option are incompatible.
6100 @kindex -msoft-float
6101 Generate code that does not use (uses) the floating-point register set.
6102 Software floating point emulation is provided if you use the
6103 @samp{-msoft-float} option, and pass the option to GCC when linking.
6106 @itemx -mno-multiple
6107 Generate code that uses (does not use) the load multiple word
6108 instructions and the store multiple word instructions. These
6109 instructions are generated by default on POWER systems, and not
6110 generated on PowerPC systems. Do not use @samp{-mmultiple} on little
6111 endian PowerPC systems, since those instructions do not work when the
6112 processor is in little endian mode. The exceptions are PPC740 and
6113 PPC750 which permit the instructions usage in little endian mode.
6118 Generate code that uses (does not use) the load string instructions
6119 and the store string word instructions to save multiple registers and
6120 do small block moves. These instructions are generated by default on
6121 POWER systems, and not generated on PowerPC systems. Do not use
6122 @samp{-mstring} on little endian PowerPC systems, since those
6123 instructions do not work when the processor is in little endian mode.
6124 The exceptions are PPC740 and PPC750 which permit the instructions
6125 usage in little endian mode.
6130 Generate code that uses (does not use) the load or store instructions
6131 that update the base register to the address of the calculated memory
6132 location. These instructions are generated by default. If you use
6133 @samp{-mno-update}, there is a small window between the time that the
6134 stack pointer is updated and the address of the previous frame is
6135 stored, which means code that walks the stack frame across interrupts or
6136 signals may get corrupted data.
6139 @itemx -mno-fused-madd
6140 @kindex -mfused-madd
6141 Generate code that uses (does not use) the floating point multiply and
6142 accumulate instructions. These instructions are generated by default if
6143 hardware floating is used.
6145 @item -mno-bit-align
6148 On System V.4 and embedded PowerPC systems do not (do) force structures
6149 and unions that contain bit fields to be aligned to the base type of the
6152 For example, by default a structure containing nothing but 8
6153 @code{unsigned} bitfields of length 1 would be aligned to a 4 byte
6154 boundary and have a size of 4 bytes. By using @samp{-mno-bit-align},
6155 the structure would be aligned to a 1 byte boundary and be one byte in
6158 @item -mno-strict-align
6159 @itemx -mstrict-align
6160 @kindex -mstrict-align
6161 On System V.4 and embedded PowerPC systems do not (do) assume that
6162 unaligned memory references will be handled by the system.
6165 @itemx -mno-relocatable
6166 @kindex -mrelocatable
6167 On embedded PowerPC systems generate code that allows (does not allow)
6168 the program to be relocated to a different address at runtime. If you
6169 use @samp{-mrelocatable} on any module, all objects linked together must
6170 be compiled with @samp{-mrelocatable} or @samp{-mrelocatable-lib}.
6172 @item -mrelocatable-lib
6173 @itemx -mno-relocatable-lib
6174 On embedded PowerPC systems generate code that allows (does not allow)
6175 the program to be relocated to a different address at runtime. Modules
6176 compiled with @samp{-mrelocatable-lib} can be linked with either modules
6177 compiled without @samp{-mrelocatable} and @samp{-mrelocatable-lib} or
6178 with modules compiled with the @samp{-mrelocatable} options.
6182 On System V.4 and embedded PowerPC systems do not (do) assume that
6183 register 2 contains a pointer to a global area pointing to the addresses
6184 used in the program.
6187 @itemx -mlittle-endian
6188 On System V.4 and embedded PowerPC systems compile code for the
6189 processor in little endian mode. The @samp{-mlittle-endian} option is
6190 the same as @samp{-mlittle}.
6194 On System V.4 and embedded PowerPC systems compile code for the
6195 processor in big endian mode. The @samp{-mbig-endian} option is
6196 the same as @samp{-mbig}.
6199 On System V.4 and embedded PowerPC systems compile code using calling
6200 conventions that adheres to the March 1995 draft of the System V
6201 Application Binary Interface, PowerPC processor supplement. This is the
6202 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
6204 @item -mcall-sysv-eabi
6205 Specify both @samp{-mcall-sysv} and @samp{-meabi} options.
6207 @item -mcall-sysv-noeabi
6208 Specify both @samp{-mcall-sysv} and @samp{-mno-eabi} options.
6211 On System V.4 and embedded PowerPC systems compile code using calling
6212 conventions that are similar to those used on AIX. This is the
6213 default if you configured GCC using @samp{powerpc-*-eabiaix}.
6215 @item -mcall-solaris
6216 On System V.4 and embedded PowerPC systems compile code for the Solaris
6220 On System V.4 and embedded PowerPC systems compile code for the
6221 Linux-based GNU system.
6224 @itemx -mno-prototype
6225 On System V.4 and embedded PowerPC systems assume that all calls to
6226 variable argument functions are properly prototyped. Otherwise, the
6227 compiler must insert an instruction before every non prototyped call to
6228 set or clear bit 6 of the condition code register (@var{CR}) to
6229 indicate whether floating point values were passed in the floating point
6230 registers in case the function takes a variable arguments. With
6231 @samp{-mprototype}, only calls to prototyped variable argument functions
6232 will set or clear the bit.
6235 On embedded PowerPC systems, assume that the startup module is called
6236 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
6237 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
6241 On embedded PowerPC systems, assume that the startup module is called
6242 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
6246 On embedded PowerPC systems, assume that the startup module is called
6247 @file{crt0.o} and the standard C libraries are @file{libads.a} and
6251 On embedded PowerPC systems, assume that the startup module is called
6252 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
6256 On System V.4 and embedded PowerPC systems, specify that you are
6257 compiling for a VxWorks system.
6260 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
6261 header to indicate that @samp{eabi} extended relocations are used.
6265 On System V.4 and embedded PowerPC systems do (do not) adhere to the
6266 Embedded Applications Binary Interface (eabi) which is a set of
6267 modifications to the System V.4 specifications. Selecting @option{-meabi}
6268 means that the stack is aligned to an 8 byte boundary, a function
6269 @code{__eabi} is called to from @code{main} to set up the eabi
6270 environment, and the @samp{-msdata} option can use both @code{r2} and
6271 @code{r13} to point to two separate small data areas. Selecting
6272 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
6273 do not call an initialization function from @code{main}, and the
6274 @samp{-msdata} option will only use @code{r13} to point to a single
6275 small data area. The @samp{-meabi} option is on by default if you
6276 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
6279 On System V.4 and embedded PowerPC systems, put small initialized
6280 @code{const} global and static data in the @samp{.sdata2} section, which
6281 is pointed to by register @code{r2}. Put small initialized
6282 non-@code{const} global and static data in the @samp{.sdata} section,
6283 which is pointed to by register @code{r13}. Put small uninitialized
6284 global and static data in the @samp{.sbss} section, which is adjacent to
6285 the @samp{.sdata} section. The @samp{-msdata=eabi} option is
6286 incompatible with the @samp{-mrelocatable} option. The
6287 @samp{-msdata=eabi} option also sets the @samp{-memb} option.
6290 On System V.4 and embedded PowerPC systems, put small global and static
6291 data in the @samp{.sdata} section, which is pointed to by register
6292 @code{r13}. Put small uninitialized global and static data in the
6293 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
6294 The @samp{-msdata=sysv} option is incompatible with the
6295 @samp{-mrelocatable} option.
6297 @item -msdata=default
6299 On System V.4 and embedded PowerPC systems, if @samp{-meabi} is used,
6300 compile code the same as @samp{-msdata=eabi}, otherwise compile code the
6301 same as @samp{-msdata=sysv}.
6304 On System V.4 and embedded PowerPC systems, put small global and static
6305 data in the @samp{.sdata} section. Put small uninitialized global and
6306 static data in the @samp{.sbss} section. Do not use register @code{r13}
6307 to address small data however. This is the default behavior unless
6308 other @samp{-msdata} options are used.
6312 On embedded PowerPC systems, put all initialized global and static data
6313 in the @samp{.data} section, and all uninitialized data in the
6314 @samp{.bss} section.
6317 @cindex smaller data references (PowerPC)
6318 @cindex .sdata/.sdata2 references (PowerPC)
6319 On embedded PowerPC systems, put global and static items less than or
6320 equal to @var{num} bytes into the small data or bss sections instead of
6321 the normal data or bss section. By default, @var{num} is 8. The
6322 @samp{-G @var{num}} switch is also passed to the linker.
6323 All modules should be compiled with the same @samp{-G @var{num}} value.
6326 @itemx -mno-regnames
6327 On System V.4 and embedded PowerPC systems do (do not) emit register
6328 names in the assembly language output using symbolic forms.
6333 @subsection IBM RT Options
6335 @cindex IBM RT options
6337 These @samp{-m} options are defined for the IBM RT PC:
6341 Use an in-line code sequence for integer multiplies. This is the
6344 @item -mcall-lib-mul
6345 Call @code{lmul$$} for integer multiples.
6347 @item -mfull-fp-blocks
6348 Generate full-size floating point data blocks, including the minimum
6349 amount of scratch space recommended by IBM. This is the default.
6351 @item -mminimum-fp-blocks
6352 Do not include extra scratch space in floating point data blocks. This
6353 results in smaller code, but slower execution, since scratch space must
6354 be allocated dynamically.
6356 @cindex @file{varargs.h} and RT PC
6357 @cindex @file{stdarg.h} and RT PC
6358 @item -mfp-arg-in-fpregs
6359 Use a calling sequence incompatible with the IBM calling convention in
6360 which floating point arguments are passed in floating point registers.
6361 Note that @code{varargs.h} and @code{stdargs.h} will not work with
6362 floating point operands if this option is specified.
6364 @item -mfp-arg-in-gregs
6365 Use the normal calling convention for floating point arguments. This is
6368 @item -mhc-struct-return
6369 Return structures of more than one word in memory, rather than in a
6370 register. This provides compatibility with the MetaWare HighC (hc)
6371 compiler. Use the option @samp{-fpcc-struct-return} for compatibility
6372 with the Portable C Compiler (pcc).
6374 @item -mnohc-struct-return
6375 Return some structures of more than one word in registers, when
6376 convenient. This is the default. For compatibility with the
6377 IBM-supplied compilers, use the option @samp{-fpcc-struct-return} or the
6378 option @samp{-mhc-struct-return}.
6382 @subsection MIPS Options
6383 @cindex MIPS options
6385 These @samp{-m} options are defined for the MIPS family of computers:
6388 @item -mcpu=@var{cpu type}
6389 Assume the defaults for the machine type @var{cpu type} when scheduling
6390 instructions. The choices for @var{cpu type} are @samp{r2000}, @samp{r3000},
6391 @samp{r3900}, @samp{r4000}, @samp{r4100}, @samp{r4300}, @samp{r4400},
6392 @samp{r4600}, @samp{r4650}, @samp{r5000}, @samp{r6000}, @samp{r8000},
6393 and @samp{orion}. Additionally, the @samp{r2000}, @samp{r3000},
6394 @samp{r4000}, @samp{r5000}, and @samp{r6000} can be abbreviated as
6395 @samp{r2k} (or @samp{r2K}), @samp{r3k}, etc. While picking a specific
6396 @var{cpu type} will schedule things appropriately for that particular
6397 chip, the compiler will not generate any code that does not meet level 1
6398 of the MIPS ISA (instruction set architecture) without a @samp{-mipsX}
6399 or @samp{-mabi} switch being used.
6402 Issue instructions from level 1 of the MIPS ISA. This is the default.
6403 @samp{r3000} is the default @var{cpu type} at this ISA level.
6406 Issue instructions from level 2 of the MIPS ISA (branch likely, square
6407 root instructions). @samp{r6000} is the default @var{cpu type} at this
6411 Issue instructions from level 3 of the MIPS ISA (64 bit instructions).
6412 @samp{r4000} is the default @var{cpu type} at this ISA level.
6415 Issue instructions from level 4 of the MIPS ISA (conditional move,
6416 prefetch, enhanced FPU instructions). @samp{r8000} is the default
6417 @var{cpu type} at this ISA level.
6420 Assume that 32 32-bit floating point registers are available. This is
6424 Assume that 32 64-bit floating point registers are available. This is
6425 the default when the @samp{-mips3} option is used.
6428 Assume that 32 32-bit general purpose registers are available. This is
6432 Assume that 32 64-bit general purpose registers are available. This is
6433 the default when the @samp{-mips3} option is used.
6436 Force int and long types to be 64 bits wide. See @samp{-mlong32} for an
6437 explanation of the default, and the width of pointers.
6440 Force long types to be 64 bits wide. See @samp{-mlong32} for an
6441 explanation of the default, and the width of pointers.
6444 Force long, int, and pointer types to be 32 bits wide.
6446 If none of @samp{-mlong32}, @samp{-mlong64}, or @samp{-mint64} are set,
6447 the size of ints, longs, and pointers depends on the ABI and ISA chosen.
6448 For @samp{-mabi=32}, and @samp{-mabi=n32}, ints and longs are 32 bits
6449 wide. For @samp{-mabi=64}, ints are 32 bits, and longs are 64 bits wide.
6450 For @samp{-mabi=eabi} and either @samp{-mips1} or @samp{-mips2}, ints
6451 and longs are 32 bits wide. For @samp{-mabi=eabi} and higher ISAs, ints
6452 are 32 bits, and longs are 64 bits wide. The width of pointer types is
6453 the smaller of the width of longs or the width of general purpose
6454 registers (which in turn depends on the ISA).
6461 Generate code for the indicated ABI. The default instruction level is
6462 @samp{-mips1} for @samp{32}, @samp{-mips3} for @samp{n32}, and
6463 @samp{-mips4} otherwise. Conversely, with @samp{-mips1} or
6464 @samp{-mips2}, the default ABI is @samp{32}; otherwise, the default ABI
6468 Generate code for the MIPS assembler, and invoke @file{mips-tfile} to
6469 add normal debug information. This is the default for all
6470 platforms except for the OSF/1 reference platform, using the OSF/rose
6471 object format. If the either of the @samp{-gstabs} or @samp{-gstabs+}
6472 switches are used, the @file{mips-tfile} program will encapsulate the
6473 stabs within MIPS ECOFF.
6476 Generate code for the GNU assembler. This is the default on the OSF/1
6477 reference platform, using the OSF/rose object format. Also, this is
6478 the default if the configure option @samp{--with-gnu-as} is used.
6480 @item -msplit-addresses
6481 @itemx -mno-split-addresses
6482 Generate code to load the high and low parts of address constants separately.
6483 This allows @code{gcc} to optimize away redundant loads of the high order
6484 bits of addresses. This optimization requires GNU as and GNU ld.
6485 This optimization is enabled by default for some embedded targets where
6486 GNU as and GNU ld are standard.
6490 The @samp{-mrnames} switch says to output code using the MIPS software
6491 names for the registers, instead of the hardware names (ie, @var{a0}
6492 instead of @var{$4}). The only known assembler that supports this option
6493 is the Algorithmics assembler.
6497 The @samp{-mgpopt} switch says to write all of the data declarations
6498 before the instructions in the text section, this allows the MIPS
6499 assembler to generate one word memory references instead of using two
6500 words for short global or static data items. This is on by default if
6501 optimization is selected.
6505 For each non-inline function processed, the @samp{-mstats} switch
6506 causes the compiler to emit one line to the standard error file to
6507 print statistics about the program (number of registers saved, stack
6512 The @samp{-mmemcpy} switch makes all block moves call the appropriate
6513 string function (@samp{memcpy} or @samp{bcopy}) instead of possibly
6514 generating inline code.
6517 @itemx -mno-mips-tfile
6518 The @samp{-mno-mips-tfile} switch causes the compiler not
6519 postprocess the object file with the @file{mips-tfile} program,
6520 after the MIPS assembler has generated it to add debug support. If
6521 @file{mips-tfile} is not run, then no local variables will be
6522 available to the debugger. In addition, @file{stage2} and
6523 @file{stage3} objects will have the temporary file names passed to the
6524 assembler embedded in the object file, which means the objects will
6525 not compare the same. The @samp{-mno-mips-tfile} switch should only
6526 be used when there are bugs in the @file{mips-tfile} program that
6527 prevents compilation.
6530 Generate output containing library calls for floating point.
6531 @strong{Warning:} the requisite libraries are not part of GCC.
6532 Normally the facilities of the machine's usual C compiler are used, but
6533 this can't be done directly in cross-compilation. You must make your
6534 own arrangements to provide suitable library functions for
6538 Generate output containing floating point instructions. This is the
6539 default if you use the unmodified sources.
6542 @itemx -mno-abicalls
6543 Emit (or do not emit) the pseudo operations @samp{.abicalls},
6544 @samp{.cpload}, and @samp{.cprestore} that some System V.4 ports use for
6545 position independent code.
6548 @itemx -mno-long-calls
6549 Do all calls with the @samp{JALR} instruction, which requires
6550 loading up a function's address into a register before the call.
6551 You need to use this switch, if you call outside of the current
6552 512 megabyte segment to functions that are not through pointers.
6555 @itemx -mno-half-pic
6556 Put pointers to extern references into the data section and load them
6557 up, rather than put the references in the text section.
6559 @item -membedded-pic
6560 @itemx -mno-embedded-pic
6561 Generate PIC code suitable for some embedded systems. All calls are
6562 made using PC relative address, and all data is addressed using the $gp
6563 register. No more than 65536 bytes of global data may be used. This
6564 requires GNU as and GNU ld which do most of the work. This currently
6565 only works on targets which use ECOFF; it does not work with ELF.
6567 @item -membedded-data
6568 @itemx -mno-embedded-data
6569 Allocate variables to the read-only data section first if possible, then
6570 next in the small data section if possible, otherwise in data. This gives
6571 slightly slower code than the default, but reduces the amount of RAM required
6572 when executing, and thus may be preferred for some embedded systems.
6574 @item -muninit-const-in-rodata
6575 @itemx -mno-uninit-const-in-rodata
6576 When used together with -membedded-data, it will always store uninitialized
6577 const variables in the read-only data section.
6579 @item -msingle-float
6580 @itemx -mdouble-float
6581 The @samp{-msingle-float} switch tells gcc to assume that the floating
6582 point coprocessor only supports single precision operations, as on the
6583 @samp{r4650} chip. The @samp{-mdouble-float} switch permits gcc to use
6584 double precision operations. This is the default.
6588 Permit use of the @samp{mad}, @samp{madu} and @samp{mul} instructions,
6589 as on the @samp{r4650} chip.
6592 Turns on @samp{-msingle-float}, @samp{-mmad}, and, at least for now,
6597 Enable 16-bit instructions.
6600 Use the entry and exit pseudo ops. This option can only be used with
6604 Compile code for the processor in little endian mode.
6605 The requisite libraries are assumed to exist.
6608 Compile code for the processor in big endian mode.
6609 The requisite libraries are assumed to exist.
6612 @cindex smaller data references (MIPS)
6613 @cindex gp-relative references (MIPS)
6614 Put global and static items less than or equal to @var{num} bytes into
6615 the small data or bss sections instead of the normal data or bss
6616 section. This allows the assembler to emit one word memory reference
6617 instructions based on the global pointer (@var{gp} or @var{$28}),
6618 instead of the normal two words used. By default, @var{num} is 8 when
6619 the MIPS assembler is used, and 0 when the GNU assembler is used. The
6620 @samp{-G @var{num}} switch is also passed to the assembler and linker.
6621 All modules should be compiled with the same @samp{-G @var{num}}
6625 Tell the MIPS assembler to not run its preprocessor over user
6626 assembler files (with a @samp{.s} suffix) when assembling them.
6629 Pass an option to gas which will cause nops to be inserted if
6630 the read of the destination register of an mfhi or mflo instruction
6631 occurs in the following two instructions.
6634 Do not include the default crt0.
6638 These options are defined by the macro
6639 @code{TARGET_SWITCHES} in the machine description. The default for the
6640 options is also defined by that macro, which enables you to change the
6645 @subsection Intel 386 Options
6646 @cindex i386 Options
6647 @cindex Intel 386 Options
6649 These @samp{-m} options are defined for the i386 family of computers:
6652 @item -mcpu=@var{cpu type}
6653 Assume the defaults for the machine type @var{cpu type} when scheduling
6654 instructions. The choices for @var{cpu type} are @samp{i386},
6655 @samp{i486}, @samp{i586}, @samp{i686}, @samp{pentium},
6656 @samp{pentiumpro}, @samp{pentium4}, @samp{k6}, and @samp{athlon}
6658 While picking a specific @var{cpu type} will schedule things appropriately
6659 for that particular chip, the compiler will not generate any code that
6660 does not run on the i386 without the @samp{-march=@var{cpu type}} option
6661 being used. @samp{i586} is equivalent to @samp{pentium} and @samp{i686}
6662 is equivalent to @samp{pentiumpro}. @samp{k6} and @samp{athlon} are the
6663 AMD chips as opposed to the Intel ones.
6665 @item -march=@var{cpu type}
6666 Generate instructions for the machine type @var{cpu type}. The choices
6667 for @var{cpu type} are the same as for @samp{-mcpu}. Moreover,
6668 specifying @samp{-march=@var{cpu type}} implies @samp{-mcpu=@var{cpu type}}.
6674 Synonyms for -mcpu=i386, -mcpu=i486, -mcpu=pentium, and -mcpu=pentiumpro
6675 respectively. These synonyms are deprecated.
6677 @item -mintel-syntax
6678 Emit assembly using Intel syntax opcodes instead of AT&T syntax.
6682 Control whether or not the compiler uses IEEE floating point
6683 comparisons. These handle correctly the case where the result of a
6684 comparison is unordered.
6687 Generate output containing library calls for floating point.
6688 @strong{Warning:} the requisite libraries are not part of GCC.
6689 Normally the facilities of the machine's usual C compiler are used, but
6690 this can't be done directly in cross-compilation. You must make your
6691 own arrangements to provide suitable library functions for
6694 On machines where a function returns floating point results in the 80387
6695 register stack, some floating point opcodes may be emitted even if
6696 @samp{-msoft-float} is used.
6698 @item -mno-fp-ret-in-387
6699 Do not use the FPU registers for return values of functions.
6701 The usual calling convention has functions return values of types
6702 @code{float} and @code{double} in an FPU register, even if there
6703 is no FPU. The idea is that the operating system should emulate
6706 The option @samp{-mno-fp-ret-in-387} causes such values to be returned
6707 in ordinary CPU registers instead.
6709 @item -mno-fancy-math-387
6710 Some 387 emulators do not support the @code{sin}, @code{cos} and
6711 @code{sqrt} instructions for the 387. Specify this option to avoid
6712 generating those instructions. This option is the default on FreeBSD.
6713 As of revision 2.6.1, these instructions are not generated unless you
6714 also use the @samp{-funsafe-math-optimizations} switch.
6716 @item -malign-double
6717 @itemx -mno-align-double
6718 Control whether GCC aligns @code{double}, @code{long double}, and
6719 @code{long long} variables on a two word boundary or a one word
6720 boundary. Aligning @code{double} variables on a two word boundary will
6721 produce code that runs somewhat faster on a @samp{Pentium} at the
6722 expense of more memory.
6724 @item -m128bit-long-double
6725 @itemx -m128bit-long-double
6726 Control the size of @code{long double} type. i386 application binary interface
6727 specify the size to be 12 bytes, while modern architectures (Pentium and newer)
6728 preffer @code{long double} aligned to 8 or 16 byte boundary. This is
6729 impossible to reach with 12 byte long doubles in the array accesses.
6731 @strong{Warning:} if you use the @samp{-m128bit-long-double} switch, the
6732 structures and arrays containing @code{long double} will change their size as
6733 well as function calling convention for function taking @code{long double}
6736 @item -m96bit-long-double
6737 @itemx -m96bit-long-double
6738 Set the size of @code{long double} to 96 bits as required by the i386
6739 application binary interface. This is the default.
6742 @itemx -mno-svr3-shlib
6743 Control whether GCC places uninitialized locals into @code{bss} or
6744 @code{data}. @samp{-msvr3-shlib} places these locals into @code{bss}.
6745 These options are meaningful only on System V Release 3.
6747 @item -mno-wide-multiply
6748 @itemx -mwide-multiply
6749 Control whether GCC uses the @code{mul} and @code{imul} that produce
6750 64 bit results in @code{eax:edx} from 32 bit operands to do @code{long
6751 long} multiplies and 32-bit division by constants.
6754 Use a different function-calling convention, in which functions that
6755 take a fixed number of arguments return with the @code{ret} @var{num}
6756 instruction, which pops their arguments while returning. This saves one
6757 instruction in the caller since there is no need to pop the arguments
6760 You can specify that an individual function is called with this calling
6761 sequence with the function attribute @samp{stdcall}. You can also
6762 override the @samp{-mrtd} option by using the function attribute
6763 @samp{cdecl}. @xref{Function Attributes}.
6765 @strong{Warning:} this calling convention is incompatible with the one
6766 normally used on Unix, so you cannot use it if you need to call
6767 libraries compiled with the Unix compiler.
6769 Also, you must provide function prototypes for all functions that
6770 take variable numbers of arguments (including @code{printf});
6771 otherwise incorrect code will be generated for calls to those
6774 In addition, seriously incorrect code will result if you call a
6775 function with too many arguments. (Normally, extra arguments are
6776 harmlessly ignored.)
6778 @item -mregparm=@var{num}
6779 Control how many registers are used to pass integer arguments. By
6780 default, no registers are used to pass arguments, and at most 3
6781 registers can be used. You can control this behavior for a specific
6782 function by using the function attribute @samp{regparm}.
6783 @xref{Function Attributes}.
6785 @strong{Warning:} if you use this switch, and
6786 @var{num} is nonzero, then you must build all modules with the same
6787 value, including any libraries. This includes the system libraries and
6790 @item -malign-loops=@var{num}
6791 Align loops to a 2 raised to a @var{num} byte boundary. If
6792 @samp{-malign-loops} is not specified, the default is 2 unless
6793 gas 2.8 (or later) is being used in which case the default is
6794 to align the loop on a 16 byte boundary if it is less than 8
6797 @item -malign-jumps=@var{num}
6798 Align instructions that are only jumped to to a 2 raised to a @var{num}
6799 byte boundary. If @samp{-malign-jumps} is not specified, the default is
6800 2 if optimizing for a 386, and 4 if optimizing for a 486 unless
6801 gas 2.8 (or later) is being used in which case the default is
6802 to align the instruction on a 16 byte boundary if it is less
6805 @item -malign-functions=@var{num}
6806 Align the start of functions to a 2 raised to @var{num} byte boundary.
6807 If @samp{-malign-functions} is not specified, the default is 2 if optimizing
6808 for a 386, and 4 if optimizing for a 486.
6810 @item -mpreferred-stack-boundary=@var{num}
6811 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
6812 byte boundary. If @samp{-mpreferred-stack-boundary} is not specified,
6813 the default is 4 (16 bytes or 128 bits).
6815 The stack is required to be aligned on a 4 byte boundary. On Pentium
6816 and PentiumPro, @code{double} and @code{long double} values should be
6817 aligned to an 8 byte boundary (see @samp{-malign-double}) or suffer
6818 significant run time performance penalties. On Pentium III, the
6819 Streaming SIMD Extension (SSE) data type @code{__m128} suffers similar
6820 penalties if it is not 16 byte aligned.
6822 To ensure proper alignment of this values on the stack, the stack boundary
6823 must be as aligned as that required by any value stored on the stack.
6824 Further, every function must be generated such that it keeps the stack
6825 aligned. Thus calling a function compiled with a higher preferred
6826 stack boundary from a function compiled with a lower preferred stack
6827 boundary will most likely misalign the stack. It is recommended that
6828 libraries that use callbacks always use the default setting.
6830 This extra alignment does consume extra stack space. Code that is sensitive
6831 to stack space usage, such as embedded systems and operating system kernels,
6832 may want to reduce the preferred alignment to
6833 @samp{-mpreferred-stack-boundary=2}.
6837 Use PUSH operations to store outgoing parameters. This method is shorter
6838 and usually equally fast as method using SUB/MOV operations and is enabled
6839 by default. In some cases disabling it may improve performance because of
6840 improved scheduling and reduced dependencies.
6842 @item -maccumulate-outgoing-args
6843 @kindex -maccumulate-outgoing-args
6844 If enabled, the maximum amount of space required for outgoing arguments will be
6845 computed in the function prologue. This in faster on most modern CPUs
6846 because of reduced dependencies, improved scheduling and reduced stack usage
6847 when preferred stack boundary is not equal to 2. The drawback is a notable
6848 increase in code size. This switch implies -mno-push-args.
6852 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
6853 on thread-safe exception handling must compile and link all code with the
6854 @samp{-mthreads} option. When compiling, @samp{-mthreads} defines
6855 @samp{-D_MT}; when linking, it links in a special thread helper library
6856 @samp{-lmingwthrd} which cleans up per thread exception handling data.
6858 @item -mno-align-stringops
6859 @kindex -mno-align-stringops
6860 Do not align destination of inlined string operations. This switch reduces
6861 code size and improves performance in case the destination is already aligned,
6862 but gcc don't know about it.
6864 @item -minline-all-stringops
6865 @kindex -minline-all-stringops
6866 By default GCC inlines string operations only when destination is known to be
6867 aligned at least to 4 byte boundary. This enables more inlining, increase code
6868 size, but may improve performance of code that depends on fast memcpy, strlen
6869 and memset for short lengths.
6873 @subsection HPPA Options
6874 @cindex HPPA Options
6876 These @samp{-m} options are defined for the HPPA family of computers:
6879 @item -march=@var{architecture type}
6880 Generate code for the specified architecture. The choices for
6881 @var{architecture type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
6882 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
6883 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
6884 architecture option for your machine. Code compiled for lower numbered
6885 architectures will run on higher numbered architectures, but not the
6888 PA 2.0 support currently requires gas snapshot 19990413 or later. The
6889 next release of binutils (current is 2.9.1) will probably contain PA 2.0
6893 @itemx -mpa-risc-1-1
6894 @itemx -mpa-risc-2-0
6895 Synonyms for -march=1.0, -march=1.1, and -march=2.0 respectively.
6898 Generate code suitable for big switch tables. Use this option only if
6899 the assembler/linker complain about out of range branches within a switch
6902 @item -mjump-in-delay
6903 Fill delay slots of function calls with unconditional jump instructions
6904 by modifying the return pointer for the function call to be the target
6905 of the conditional jump.
6907 @item -mdisable-fpregs
6908 Prevent floating point registers from being used in any manner. This is
6909 necessary for compiling kernels which perform lazy context switching of
6910 floating point registers. If you use this option and attempt to perform
6911 floating point operations, the compiler will abort.
6913 @item -mdisable-indexing
6914 Prevent the compiler from using indexing address modes. This avoids some
6915 rather obscure problems when compiling MIG generated code under MACH.
6917 @item -mno-space-regs
6918 Generate code that assumes the target has no space registers. This allows
6919 GCC to generate faster indirect calls and use unscaled index address modes.
6921 Such code is suitable for level 0 PA systems and kernels.
6923 @item -mfast-indirect-calls
6924 Generate code that assumes calls never cross space boundaries. This
6925 allows GCC to emit code which performs faster indirect calls.
6927 This option will not work in the presence of shared libraries or nested
6930 @item -mlong-load-store
6931 Generate 3-instruction load and store sequences as sometimes required by
6932 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
6935 @item -mportable-runtime
6936 Use the portable calling conventions proposed by HP for ELF systems.
6939 Enable the use of assembler directives only GAS understands.
6941 @item -mschedule=@var{cpu type}
6942 Schedule code according to the constraints for the machine type
6943 @var{cpu type}. The choices for @var{cpu type} are @samp{700}
6944 @samp{7100}, @samp{7100LC}, @samp{7200}, and @samp{8000}. Refer to
6945 @file{/usr/lib/sched.models} on an HP-UX system to determine the
6946 proper scheduling option for your machine.
6949 Enable the optimization pass in the HPUX linker. Note this makes symbolic
6950 debugging impossible. It also triggers a bug in the HPUX 8 and HPUX 9 linkers
6951 in which they give bogus error messages when linking some programs.
6954 Generate output containing library calls for floating point.
6955 @strong{Warning:} the requisite libraries are not available for all HPPA
6956 targets. Normally the facilities of the machine's usual C compiler are
6957 used, but this cannot be done directly in cross-compilation. You must make
6958 your own arrangements to provide suitable library functions for
6959 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
6960 does provide software floating point support.
6962 @samp{-msoft-float} changes the calling convention in the output file;
6963 therefore, it is only useful if you compile @emph{all} of a program with
6964 this option. In particular, you need to compile @file{libgcc.a}, the
6965 library that comes with GCC, with @samp{-msoft-float} in order for
6969 @node Intel 960 Options
6970 @subsection Intel 960 Options
6972 These @samp{-m} options are defined for the Intel 960 implementations:
6975 @item -m@var{cpu type}
6976 Assume the defaults for the machine type @var{cpu type} for some of
6977 the other options, including instruction scheduling, floating point
6978 support, and addressing modes. The choices for @var{cpu type} are
6979 @samp{ka}, @samp{kb}, @samp{mc}, @samp{ca}, @samp{cf},
6980 @samp{sa}, and @samp{sb}.
6986 The @samp{-mnumerics} option indicates that the processor does support
6987 floating-point instructions. The @samp{-msoft-float} option indicates
6988 that floating-point support should not be assumed.
6990 @item -mleaf-procedures
6991 @itemx -mno-leaf-procedures
6992 Do (or do not) attempt to alter leaf procedures to be callable with the
6993 @code{bal} instruction as well as @code{call}. This will result in more
6994 efficient code for explicit calls when the @code{bal} instruction can be
6995 substituted by the assembler or linker, but less efficient code in other
6996 cases, such as calls via function pointers, or using a linker that doesn't
6997 support this optimization.
7000 @itemx -mno-tail-call
7001 Do (or do not) make additional attempts (beyond those of the
7002 machine-independent portions of the compiler) to optimize tail-recursive
7003 calls into branches. You may not want to do this because the detection of
7004 cases where this is not valid is not totally complete. The default is
7005 @samp{-mno-tail-call}.
7007 @item -mcomplex-addr
7008 @itemx -mno-complex-addr
7009 Assume (or do not assume) that the use of a complex addressing mode is a
7010 win on this implementation of the i960. Complex addressing modes may not
7011 be worthwhile on the K-series, but they definitely are on the C-series.
7012 The default is currently @samp{-mcomplex-addr} for all processors except
7016 @itemx -mno-code-align
7017 Align code to 8-byte boundaries for faster fetching (or don't bother).
7018 Currently turned on by default for C-series implementations only.
7021 @item -mclean-linkage
7022 @itemx -mno-clean-linkage
7023 These options are not fully implemented.
7027 @itemx -mic2.0-compat
7028 @itemx -mic3.0-compat
7029 Enable compatibility with iC960 v2.0 or v3.0.
7033 Enable compatibility with the iC960 assembler.
7035 @item -mstrict-align
7036 @itemx -mno-strict-align
7037 Do not permit (do permit) unaligned accesses.
7040 Enable structure-alignment compatibility with Intel's gcc release version
7041 1.3 (based on gcc 1.37). This option implies @samp{-mstrict-align}.
7043 @item -mlong-double-64
7044 Implement type @samp{long double} as 64-bit floating point numbers.
7045 Without the option @samp{long double} is implemented by 80-bit
7046 floating point numbers. The only reason we have it because there is
7047 no 128-bit @samp{long double} support in @samp{fp-bit.c} yet. So it
7048 is only useful for people using soft-float targets. Otherwise, we
7049 should recommend against use of it.
7053 @node DEC Alpha Options
7054 @subsection DEC Alpha Options
7056 These @samp{-m} options are defined for the DEC Alpha implementations:
7059 @item -mno-soft-float
7061 Use (do not use) the hardware floating-point instructions for
7062 floating-point operations. When @option{-msoft-float} is specified,
7063 functions in @file{libgcc1.c} will be used to perform floating-point
7064 operations. Unless they are replaced by routines that emulate the
7065 floating-point operations, or compiled in such a way as to call such
7066 emulations routines, these routines will issue floating-point
7067 operations. If you are compiling for an Alpha without floating-point
7068 operations, you must ensure that the library is built so as not to call
7071 Note that Alpha implementations without floating-point operations are
7072 required to have floating-point registers.
7076 Generate code that uses (does not use) the floating-point register set.
7077 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
7078 register set is not used, floating point operands are passed in integer
7079 registers as if they were integers and floating-point results are passed
7080 in $0 instead of $f0. This is a non-standard calling sequence, so any
7081 function with a floating-point argument or return value called by code
7082 compiled with @option{-mno-fp-regs} must also be compiled with that
7085 A typical use of this option is building a kernel that does not use,
7086 and hence need not save and restore, any floating-point registers.
7089 The Alpha architecture implements floating-point hardware optimized for
7090 maximum performance. It is mostly compliant with the IEEE floating
7091 point standard. However, for full compliance, software assistance is
7092 required. This option generates code fully IEEE compliant code
7093 @emph{except} that the @var{inexact flag} is not maintained (see below).
7094 If this option is turned on, the CPP macro @code{_IEEE_FP} is defined
7095 during compilation. The option is a shorthand for: @samp{-D_IEEE_FP
7096 -mfp-trap-mode=su -mtrap-precision=i -mieee-conformant}. The resulting
7097 code is less efficient but is able to correctly support denormalized
7098 numbers and exceptional IEEE values such as not-a-number and plus/minus
7099 infinity. Other Alpha compilers call this option
7100 @option{-ieee_with_no_inexact}.
7102 @item -mieee-with-inexact
7103 @c overfull hbox here --bob 22 jul96
7104 @c original text between ignore ... end ignore
7106 This is like @samp{-mieee} except the generated code also maintains the
7107 IEEE @var{inexact flag}. Turning on this option causes the generated
7108 code to implement fully-compliant IEEE math. The option is a shorthand
7109 for @samp{-D_IEEE_FP -D_IEEE_FP_INEXACT} plus @samp{-mieee-conformant},
7110 @samp{-mfp-trap-mode=sui}, and @samp{-mtrap-precision=i}. On some Alpha
7111 implementations the resulting code may execute significantly slower than
7112 the code generated by default. Since there is very little code that
7113 depends on the @var{inexact flag}, you should normally not specify this
7114 option. Other Alpha compilers call this option
7115 @samp{-ieee_with_inexact}.
7117 @c changed paragraph
7118 This is like @samp{-mieee} except the generated code also maintains the
7119 IEEE @var{inexact flag}. Turning on this option causes the generated
7120 code to implement fully-compliant IEEE math. The option is a shorthand
7121 for @samp{-D_IEEE_FP -D_IEEE_FP_INEXACT} plus the three following:
7122 @samp{-mieee-conformant},
7123 @samp{-mfp-trap-mode=sui},
7124 and @samp{-mtrap-precision=i}.
7125 On some Alpha implementations the resulting code may execute
7126 significantly slower than the code generated by default. Since there
7127 is very little code that depends on the @var{inexact flag}, you should
7128 normally not specify this option. Other Alpha compilers call this
7129 option @samp{-ieee_with_inexact}.
7130 @c end changes to prevent overfull hboxes
7132 @item -mfp-trap-mode=@var{trap mode}
7133 This option controls what floating-point related traps are enabled.
7134 Other Alpha compilers call this option @samp{-fptm }@var{trap mode}.
7135 The trap mode can be set to one of four values:
7139 This is the default (normal) setting. The only traps that are enabled
7140 are the ones that cannot be disabled in software (e.g., division by zero
7144 In addition to the traps enabled by @samp{n}, underflow traps are enabled
7148 Like @samp{su}, but the instructions are marked to be safe for software
7149 completion (see Alpha architecture manual for details).
7152 Like @samp{su}, but inexact traps are enabled as well.
7155 @item -mfp-rounding-mode=@var{rounding mode}
7156 Selects the IEEE rounding mode. Other Alpha compilers call this option
7157 @samp{-fprm }@var{rounding mode}. The @var{rounding mode} can be one
7162 Normal IEEE rounding mode. Floating point numbers are rounded towards
7163 the nearest machine number or towards the even machine number in case
7167 Round towards minus infinity.
7170 Chopped rounding mode. Floating point numbers are rounded towards zero.
7173 Dynamic rounding mode. A field in the floating point control register
7174 (@var{fpcr}, see Alpha architecture reference manual) controls the
7175 rounding mode in effect. The C library initializes this register for
7176 rounding towards plus infinity. Thus, unless your program modifies the
7177 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
7180 @item -mtrap-precision=@var{trap precision}
7181 In the Alpha architecture, floating point traps are imprecise. This
7182 means without software assistance it is impossible to recover from a
7183 floating trap and program execution normally needs to be terminated.
7184 GCC can generate code that can assist operating system trap handlers
7185 in determining the exact location that caused a floating point trap.
7186 Depending on the requirements of an application, different levels of
7187 precisions can be selected:
7191 Program precision. This option is the default and means a trap handler
7192 can only identify which program caused a floating point exception.
7195 Function precision. The trap handler can determine the function that
7196 caused a floating point exception.
7199 Instruction precision. The trap handler can determine the exact
7200 instruction that caused a floating point exception.
7203 Other Alpha compilers provide the equivalent options called
7204 @samp{-scope_safe} and @samp{-resumption_safe}.
7206 @item -mieee-conformant
7207 This option marks the generated code as IEEE conformant. You must not
7208 use this option unless you also specify @samp{-mtrap-precision=i} and either
7209 @samp{-mfp-trap-mode=su} or @samp{-mfp-trap-mode=sui}. Its only effect
7210 is to emit the line @samp{.eflag 48} in the function prologue of the
7211 generated assembly file. Under DEC Unix, this has the effect that
7212 IEEE-conformant math library routines will be linked in.
7214 @item -mbuild-constants
7215 Normally GCC examines a 32- or 64-bit integer constant to
7216 see if it can construct it from smaller constants in two or three
7217 instructions. If it cannot, it will output the constant as a literal and
7218 generate code to load it from the data segment at runtime.
7220 Use this option to require GCC to construct @emph{all} integer constants
7221 using code, even if it takes more instructions (the maximum is six).
7223 You would typically use this option to build a shared library dynamic
7224 loader. Itself a shared library, it must relocate itself in memory
7225 before it can find the variables and constants in its own data segment.
7229 Select whether to generate code to be assembled by the vendor-supplied
7230 assembler (@samp{-malpha-as}) or by the GNU assembler @samp{-mgas}.
7238 Indicate whether GCC should generate code to use the optional BWX,
7239 CIX, and MAX instruction sets. The default is to use the instruction sets
7240 supported by the CPU type specified via @samp{-mcpu=} option or that
7241 of the CPU on which GCC was built if none was specified.
7243 @item -mcpu=@var{cpu_type}
7244 Set the instruction set, register set, and instruction scheduling
7245 parameters for machine type @var{cpu_type}. You can specify either the
7246 @samp{EV} style name or the corresponding chip number. GCC
7247 supports scheduling parameters for the EV4 and EV5 family of processors
7248 and will choose the default values for the instruction set from
7249 the processor you specify. If you do not specify a processor type,
7250 GCC will default to the processor on which the compiler was built.
7252 Supported values for @var{cpu_type} are
7257 Schedules as an EV4 and has no instruction set extensions.
7261 Schedules as an EV5 and has no instruction set extensions.
7265 Schedules as an EV5 and supports the BWX extension.
7270 Schedules as an EV5 and supports the BWX and MAX extensions.
7274 Schedules as an EV5 (until Digital releases the scheduling parameters
7275 for the EV6) and supports the BWX, CIX, and MAX extensions.
7278 @item -mmemory-latency=@var{time}
7279 Sets the latency the scheduler should assume for typical memory
7280 references as seen by the application. This number is highly
7281 dependent on the memory access patterns used by the application
7282 and the size of the external cache on the machine.
7284 Valid options for @var{time} are
7288 A decimal number representing clock cycles.
7294 The compiler contains estimates of the number of clock cycles for
7295 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
7296 (also called Dcache, Scache, and Bcache), as well as to main memory.
7297 Note that L3 is only valid for EV5.
7302 @node Clipper Options
7303 @subsection Clipper Options
7305 These @samp{-m} options are defined for the Clipper implementations:
7309 Produce code for a C300 Clipper processor. This is the default.
7312 Produce code for a C400 Clipper processor i.e. use floating point
7316 @node H8/300 Options
7317 @subsection H8/300 Options
7319 These @samp{-m} options are defined for the H8/300 implementations:
7323 Shorten some address references at link time, when possible; uses the
7324 linker option @samp{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
7325 ld.info, Using ld}, for a fuller description.
7328 Generate code for the H8/300H.
7331 Generate code for the H8/S.
7334 Generate code for the H8/S2600. This switch must be used with -ms.
7337 Make @code{int} data 32 bits by default.
7340 On the H8/300H and H8/S, use the same alignment rules as for the H8/300.
7341 The default for the H8/300H and H8/S is to align longs and floats on 4
7343 @samp{-malign-300} causes them to be aligned on 2 byte boundaries.
7344 This option has no effect on the H8/300.
7348 @subsection SH Options
7350 These @samp{-m} options are defined for the SH implementations:
7354 Generate code for the SH1.
7357 Generate code for the SH2.
7360 Generate code for the SH3.
7363 Generate code for the SH3e.
7366 Generate code for the SH4 without a floating-point unit.
7368 @item -m4-single-only
7369 Generate code for the SH4 with a floating-point unit that only
7370 supports single-precision arithmentic.
7373 Generate code for the SH4 assuming the floating-point unit is in
7374 single-precision mode by default.
7377 Generate code for the SH4.
7380 Compile code for the processor in big endian mode.
7383 Compile code for the processor in little endian mode.
7386 Align doubles at 64 bit boundaries. Note that this changes the calling
7387 conventions, and thus some functions from the standard C library will
7388 not work unless you recompile it first with -mdalign.
7391 Shorten some address references at link time, when possible; uses the
7392 linker option @samp{-relax}.
7395 Use 32-bit offsets in @code{switch} tables. The default is to use
7399 Enable the use of the instruction @code{fmovd}.
7402 Comply with the calling conventions defined by Hitachi.
7405 Mark the @code{MAC} register as call-clobbered, even if
7406 @option{-mhitachi} is given.
7409 Dump instruction size and location in the assembly code.
7412 This option is deprecated. It pads structures to multiple of 4 bytes,
7413 which is incompatible with the SH ABI.
7416 Optimize for space instead of speed. Implied by @option{-Os}.
7419 When generating position-independent code, emit function calls using
7420 the Global Offset Table instead of the Procedure Linkage Table.
7423 Generate a library function call to invalidate instruction cache
7424 entries, after fixing up a trampoline. This library function call
7425 doesn't assume it can write to the whole memory address space. This
7426 is the default when the target is @code{sh-*-linux*}.
7429 @node System V Options
7430 @subsection Options for System V
7432 These additional options are available on System V Release 4 for
7433 compatibility with other compilers on those systems:
7437 Create a shared object.
7438 It is recommended that @samp{-symbolic} or @samp{-shared} be used instead.
7441 Identify the versions of each tool used by the compiler, in a
7442 @code{.ident} assembler directive in the output.
7445 Refrain from adding @code{.ident} directives to the output file (this is
7448 @item -YP\,@var{dirs}
7449 Search the directories @var{dirs}, and no others, for libraries
7450 specified with @samp{-l}.
7452 @item -Ym\,@var{dir}
7453 Look in the directory @var{dir} to find the M4 preprocessor.
7454 The assembler uses this option.
7455 @c This is supposed to go with a -Yd for predefined M4 macro files, but
7456 @c the generic assembler that comes with Solaris takes just -Ym.
7459 @node TMS320C3x/C4x Options
7460 @subsection TMS320C3x/C4x Options
7461 @cindex TMS320C3x/C4x Options
7463 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
7467 @item -mcpu=@var{cpu_type}
7468 Set the instruction set, register set, and instruction scheduling
7469 parameters for machine type @var{cpu_type}. Supported values for
7470 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
7471 @samp{c44}. The default is @samp{c40} to generate code for the
7476 @itemx -msmall-memory
7478 Generates code for the big or small memory model. The small memory
7479 model assumed that all data fits into one 64K word page. At run-time
7480 the data page (DP) register must be set to point to the 64K page
7481 containing the .bss and .data program sections. The big memory model is
7482 the default and requires reloading of the DP register for every direct
7487 Allow (disallow) allocation of general integer operands into the block
7492 Enable (disable) generation of code using decrement and branch,
7493 DBcond(D), instructions. This is enabled by default for the C4x. To be
7494 on the safe side, this is disabled for the C3x, since the maximum
7495 iteration count on the C3x is 2^23 + 1 (but who iterates loops more than
7496 2^23 times on the C3x?). Note that GCC will try to reverse a loop so
7497 that it can utilise the decrement and branch instruction, but will give
7498 up if there is more than one memory reference in the loop. Thus a loop
7499 where the loop counter is decremented can generate slightly more
7500 efficient code, in cases where the RPTB instruction cannot be utilised.
7502 @item -mdp-isr-reload
7504 Force the DP register to be saved on entry to an interrupt service
7505 routine (ISR), reloaded to point to the data section, and restored on
7506 exit from the ISR. This should not be required unless someone has
7507 violated the small memory model by modifying the DP register, say within
7512 For the C3x use the 24-bit MPYI instruction for integer multiplies
7513 instead of a library call to guarantee 32-bit results. Note that if one
7514 of the operands is a constant, then the multiplication will be performed
7515 using shifts and adds. If the -mmpyi option is not specified for the C3x,
7516 then squaring operations are performed inline instead of a library call.
7519 @itemx -mno-fast-fix
7520 The C3x/C4x FIX instruction to convert a floating point value to an
7521 integer value chooses the nearest integer less than or equal to the
7522 floating point value rather than to the nearest integer. Thus if the
7523 floating point number is negative, the result will be incorrectly
7524 truncated an additional code is necessary to detect and correct this
7525 case. This option can be used to disable generation of the additional
7526 code required to correct the result.
7530 Enable (disable) generation of repeat block sequences using the RPTB
7531 instruction for zero overhead looping. The RPTB construct is only used
7532 for innermost loops that do not call functions or jump across the loop
7533 boundaries. There is no advantage having nested RPTB loops due to the
7534 overhead required to save and restore the RC, RS, and RE registers.
7535 This is enabled by default with -O2.
7537 @item -mrpts=@var{count}
7539 Enable (disable) the use of the single instruction repeat instruction
7540 RPTS. If a repeat block contains a single instruction, and the loop
7541 count can be guaranteed to be less than the value @var{count}, GCC will
7542 emit a RPTS instruction instead of a RPTB. If no value is specified,
7543 then a RPTS will be emitted even if the loop count cannot be determined
7544 at compile time. Note that the repeated instruction following RPTS does
7545 not have to be reloaded from memory each iteration, thus freeing up the
7546 CPU buses for operands. However, since interrupts are blocked by this
7547 instruction, it is disabled by default.
7549 @item -mloop-unsigned
7550 @itemx -mno-loop-unsigned
7551 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
7552 is 2^31 + 1 since these instructions test if the iteration count is
7553 negative to terminate the loop. If the iteration count is unsigned
7554 there is a possibility than the 2^31 + 1 maximum iteration count may be
7555 exceeded. This switch allows an unsigned iteration count.
7558 Try to emit an assembler syntax that the TI assembler (asm30) is happy
7559 with. This also enforces compatibility with the API employed by the TI
7560 C3x C compiler. For example, long doubles are passed as structures
7561 rather than in floating point registers.
7565 Generate code that uses registers (stack) for passing arguments to functions.
7566 By default, arguments are passed in registers where possible rather
7567 than by pushing arguments on to the stack.
7569 @item -mparallel-insns
7570 @itemx -mno-parallel-insns
7571 Allow the generation of parallel instructions. This is enabled by
7574 @item -mparallel-mpy
7575 @itemx -mno-parallel-mpy
7576 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
7577 provided -mparallel-insns is also specified. These instructions have
7578 tight register constraints which can pessimize the code generation
7584 @subsection V850 Options
7585 @cindex V850 Options
7587 These @samp{-m} options are defined for V850 implementations:
7591 @itemx -mno-long-calls
7592 Treat all calls as being far away (near). If calls are assumed to be
7593 far away, the compiler will always load the functions address up into a
7594 register, and call indirect through the pointer.
7598 Do not optimize (do optimize) basic blocks that use the same index
7599 pointer 4 or more times to copy pointer into the @code{ep} register, and
7600 use the shorter @code{sld} and @code{sst} instructions. The @samp{-mep}
7601 option is on by default if you optimize.
7603 @item -mno-prolog-function
7604 @itemx -mprolog-function
7605 Do not use (do use) external functions to save and restore registers at
7606 the prolog and epilog of a function. The external functions are slower,
7607 but use less code space if more than one function saves the same number
7608 of registers. The @samp{-mprolog-function} option is on by default if
7612 Try to make the code as small as possible. At present, this just turns
7613 on the @samp{-mep} and @samp{-mprolog-function} options.
7616 Put static or global variables whose size is @var{n} bytes or less into
7617 the tiny data area that register @code{ep} points to. The tiny data
7618 area can hold up to 256 bytes in total (128 bytes for byte references).
7621 Put static or global variables whose size is @var{n} bytes or less into
7622 the small data area that register @code{gp} points to. The small data
7623 area can hold up to 64 kilobytes.
7626 Put static or global variables whose size is @var{n} bytes or less into
7627 the first 32 kilobytes of memory.
7630 Specify that the target processor is the V850.
7633 Generate code suitable for big switch tables. Use this option only if
7634 the assembler/linker complain about out of range branches within a switch
7639 @subsection ARC Options
7642 These options are defined for ARC implementations:
7646 Compile code for little endian mode. This is the default.
7649 Compile code for big endian mode.
7652 Prepend the name of the cpu to all public symbol names.
7653 In multiple-processor systems, there are many ARC variants with different
7654 instruction and register set characteristics. This flag prevents code
7655 compiled for one cpu to be linked with code compiled for another.
7656 No facility exists for handling variants that are "almost identical".
7657 This is an all or nothing option.
7659 @item -mcpu=@var{cpu}
7660 Compile code for ARC variant @var{cpu}.
7661 Which variants are supported depend on the configuration.
7662 All variants support @samp{-mcpu=base}, this is the default.
7664 @item -mtext=@var{text section}
7665 @itemx -mdata=@var{data section}
7666 @itemx -mrodata=@var{readonly data section}
7667 Put functions, data, and readonly data in @var{text section},
7668 @var{data section}, and @var{readonly data section} respectively
7669 by default. This can be overridden with the @code{section} attribute.
7670 @xref{Variable Attributes}.
7675 @subsection NS32K Options
7676 @cindex NS32K options
7678 These are the @samp{-m} options defined for the 32000 series. The default
7679 values for these options depends on which style of 32000 was selected when
7680 the compiler was configured; the defaults for the most common choices are
7686 Generate output for a 32032. This is the default
7687 when the compiler is configured for 32032 and 32016 based systems.
7691 Generate output for a 32332. This is the default
7692 when the compiler is configured for 32332-based systems.
7696 Generate output for a 32532. This is the default
7697 when the compiler is configured for 32532-based systems.
7700 Generate output containing 32081 instructions for floating point.
7701 This is the default for all systems.
7704 Generate output containing 32381 instructions for floating point. This
7705 also implies @samp{-m32081}. The 32381 is only compatible with the 32332
7706 and 32532 cpus. This is the default for the pc532-netbsd configuration.
7709 Try and generate multiply-add floating point instructions @code{polyF}
7710 and @code{dotF}. This option is only available if the @samp{-m32381}
7711 option is in effect. Using these instructions requires changes to to
7712 register allocation which generally has a negative impact on
7713 performance. This option should only be enabled when compiling code
7714 particularly likely to make heavy use of multiply-add instructions.
7717 Do not try and generate multiply-add floating point instructions
7718 @code{polyF} and @code{dotF}. This is the default on all platforms.
7721 Generate output containing library calls for floating point.
7722 @strong{Warning:} the requisite libraries may not be available.
7725 Do not use the bit-field instructions. On some machines it is faster to
7726 use shifting and masking operations. This is the default for the pc532.
7729 Do use the bit-field instructions. This is the default for all platforms
7733 Use a different function-calling convention, in which functions
7734 that take a fixed number of arguments return pop their
7735 arguments on return with the @code{ret} instruction.
7737 This calling convention is incompatible with the one normally
7738 used on Unix, so you cannot use it if you need to call libraries
7739 compiled with the Unix compiler.
7741 Also, you must provide function prototypes for all functions that
7742 take variable numbers of arguments (including @code{printf});
7743 otherwise incorrect code will be generated for calls to those
7746 In addition, seriously incorrect code will result if you call a
7747 function with too many arguments. (Normally, extra arguments are
7748 harmlessly ignored.)
7750 This option takes its name from the 680x0 @code{rtd} instruction.
7754 Use a different function-calling convention where the first two arguments
7755 are passed in registers.
7757 This calling convention is incompatible with the one normally
7758 used on Unix, so you cannot use it if you need to call libraries
7759 compiled with the Unix compiler.
7762 Do not pass any arguments in registers. This is the default for all
7766 It is OK to use the sb as an index register which is always loaded with
7767 zero. This is the default for the pc532-netbsd target.
7770 The sb register is not available for use or has not been initialized to
7771 zero by the run time system. This is the default for all targets except
7772 the pc532-netbsd. It is also implied whenever @samp{-mhimem} or
7773 @samp{-fpic} is set.
7776 Many ns32000 series addressing modes use displacements of up to 512MB.
7777 If an address is above 512MB then displacements from zero can not be used.
7778 This option causes code to be generated which can be loaded above 512MB.
7779 This may be useful for operating systems or ROM code.
7782 Assume code will be loaded in the first 512MB of virtual address space.
7783 This is the default for all platforms.
7789 @subsection AVR Options
7792 These options are defined for AVR implementations:
7795 @item -mmcu=@var{mcu}
7796 Specify ATMEL AVR instruction set or MCU type.
7798 Instruction set avr1 is for the minimal AVR core, not supported by the C
7799 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
7800 attiny11, attiny12, attiny15, attiny28).
7802 Instruction set avr2 (default) is for the classic AVR core with up to
7803 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
7804 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
7805 at90c8534, at90s8535).
7807 Instruction set avr3 is for the classic AVR core with up to 128K program
7808 memory space (MCU types: atmega103, atmega603).
7810 Instruction set avr4 is for the enhanced AVR core with up to 8K program
7811 memory space (MCU types: atmega83, atmega85).
7813 Instruction set avr5 is for the enhanced AVR core with up to 128K program
7814 memory space (MCU types: atmega161, atmega163, atmega32, at94k).
7817 Output instruction sizes to the asm file.
7819 @item -minit-stack=@var{N}
7820 Specify the initial stack address, which may be a symbol or numeric value,
7821 __stack is the default.
7823 @item -mno-interrupts
7824 Generated code is not compatible with hardware interrupts.
7825 Code size will be smaller.
7827 @item -mcall-prologues
7828 Functions prologues/epilogues expanded as call to appropriate
7829 subroutines. Code size will be smaller.
7831 @item -mno-tablejump
7832 Do not generate tablejump insns which sometimes increase code size.
7835 Change only the low 8 bits of the stack pointer.
7839 @subsection MCore Options
7840 @cindex MCore options
7842 These are the @samp{-m} options defined for the Motorola M*Core
7850 Inline constants into the code stream if it can be done in two
7851 instructions or less.
7856 Use the divide instruction. (Enabled by default).
7858 @item -mrelax-immediate
7859 @itemx -mrelax-immediate
7860 @itemx -mno-relax-immediate
7861 Allow arbitrary sized immediates in bit operations.
7863 @item -mwide-bitfields
7864 @itemx -mwide-bitfields
7865 @itemx -mno-wide-bitfields
7866 Always treat bitfields as int-sized.
7868 @item -m4byte-functions
7869 @itemx -m4byte-functions
7870 @itemx -mno-4byte-functions
7871 Force all functions to be aligned to a four byte boundary.
7873 @item -mcallgraph-data
7874 @itemx -mcallgraph-data
7875 @itemx -mno-callgraph-data
7876 Emit callgraph information.
7880 @itemx -mno-slow-bytes
7881 Prefer word access when reading byte quantities.
7883 @item -mlittle-endian
7884 @itemx -mlittle-endian
7886 Generate code for a little endian target.
7891 Generate code for the 210 processor.
7895 @subsection IA-64 Options
7896 @cindex IA-64 Options
7898 These are the @samp{-m} options defined for the Intel IA-64 architecture.
7902 Generate code for a big endian target. This is the default for HPUX.
7904 @item -mlittle-endian
7905 Generate code for a little endian target. This is the default for AIX5
7910 Generate (or don't) code for the GNU assembler. This is the default.
7911 @c Also, this is the default if the configure option @samp{--with-gnu-as}
7916 Generate (or don't) code for the GNU linker. This is the default.
7917 @c Also, this is the default if the configure option @samp{--with-gnu-ld}
7921 Generate code that does not use a global pointer register. The result
7922 is not position independent code, and violates the IA-64 ABI.
7924 @item -mvolatile-asm-stop
7925 @itemx -mno-volatile-asm-stop
7926 Generate (or don't) a stop bit immediately before and after volatile asm
7930 Generate code that works around Itanium B step errata.
7932 @item -mregister-names
7933 @itemx -mno-register-names
7934 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
7935 the stacked registers. This may make assembler output more readable.
7939 Disable (or enable) optimizations that use the small data section. This may
7940 be useful for working around optimizer bugs.
7943 Generate code that uses a single constant global pointer value. This is
7944 useful when compiling kernel code.
7947 Generate code that is self-relocatable. This implies @samp{-mconstant-gp}.
7948 This is useful when compiling firmware code.
7950 @item -minline-divide-min-latency
7951 Generate code for inline divides using the minimum latency algorithm.
7953 @item -minline-divide-max-throughput
7954 Generate code for inline divides using the maximum throughput algorithm.
7956 @item -mno-dwarf2-asm
7958 Don't (or do) generate assembler code for the DWARF2 line number debugging
7959 info. This may be useful when not using the GNU assembler.
7961 @item -mfixed-range=@var{register range}
7962 Generate code treating the given register range as fixed registers.
7963 A fixed register is one that the register allocator can not use. This is
7964 useful when compiling kernel code. A register range is specified as
7965 two registers separated by a dash. Multiple register ranges can be
7966 specified separated by a comma.
7970 @subsection D30V Options
7971 @cindex D30V Options
7973 These @samp{-m} options are defined for D30V implementations:
7977 Link the @samp{.text}, @samp{.data}, @samp{.bss}, @samp{.strings},
7978 @samp{.rodata}, @samp{.rodata1}, @samp{.data1} sections into external
7979 memory, which starts at location @code{0x80000000}.
7982 Same as the @samp{-mextmem} switch.
7985 Link the @samp{.text} section into onchip text memory, which starts at
7986 location @code{0x0}. Also link @samp{.data}, @samp{.bss},
7987 @samp{.strings}, @samp{.rodata}, @samp{.rodata1}, @samp{.data1} sections
7988 into onchip data memory, which starts at location @code{0x20000000}.
7990 @item -mno-asm-optimize
7991 @itemx -masm-optimize
7992 Disable (enable) passing @samp{-O} to the assembler when optimizing.
7993 The assembler uses the @samp{-O} option to automatically parallelize
7994 adjacent short instructions where possible.
7996 @item -mbranch-cost=@var{n}
7997 Increase the internal costs of branches to @var{n}. Higher costs means
7998 that the compiler will issue more instructions to avoid doing a branch.
8001 @item -mcond-exec=@var{n}
8002 Specify the maximum number of conditionally executed instructions that
8003 replace a branch. The default is 4.
8006 @node Code Gen Options
8007 @section Options for Code Generation Conventions
8008 @cindex code generation conventions
8009 @cindex options, code generation
8010 @cindex run-time options
8012 These machine-independent options control the interface conventions
8013 used in code generation.
8015 Most of them have both positive and negative forms; the negative form
8016 of @samp{-ffoo} would be @samp{-fno-foo}. In the table below, only
8017 one of the forms is listed---the one which is not the default. You
8018 can figure out the other form by either removing @samp{no-} or adding
8023 Enable exception handling. Generates extra code needed to propagate
8024 exceptions. For some targets, this implies GNU CC will generate frame
8025 unwind information for all functions, which can produce significant data
8026 size overhead, although it does not affect execution. If you do not
8027 specify this option, GNU CC will enable it by default for languages like
8028 C++ which normally require exception handling, and disable itfor
8029 languages like C that do not normally require it. However, you may need
8030 to enable this option when compiling C code that needs to interoperate
8031 properly with exception handlers written in C++. You may also wish to
8032 disable this option if you are compiling older C++ programs that don't
8033 use exception handling.
8035 @item -funwind-tables
8036 Similar to @option{-fexceptions}, except that it will just generate any needed
8037 static data, but will not affect the generated code in any other way.
8038 You will normally not enable this option; instead, a language processor
8039 that needs this handling would enable it on your behalf.
8041 @item -fpcc-struct-return
8042 Return ``short'' @code{struct} and @code{union} values in memory like
8043 longer ones, rather than in registers. This convention is less
8044 efficient, but it has the advantage of allowing intercallability between
8045 GCC-compiled files and files compiled with other compilers.
8047 The precise convention for returning structures in memory depends
8048 on the target configuration macros.
8050 Short structures and unions are those whose size and alignment match
8051 that of some integer type.
8053 @item -freg-struct-return
8054 Use the convention that @code{struct} and @code{union} values are
8055 returned in registers when possible. This is more efficient for small
8056 structures than @samp{-fpcc-struct-return}.
8058 If you specify neither @samp{-fpcc-struct-return} nor its contrary
8059 @samp{-freg-struct-return}, GCC defaults to whichever convention is
8060 standard for the target. If there is no standard convention, GCC
8061 defaults to @samp{-fpcc-struct-return}, except on targets where GCC
8062 is the principal compiler. In those cases, we can choose the standard,
8063 and we chose the more efficient register return alternative.
8066 Allocate to an @code{enum} type only as many bytes as it needs for the
8067 declared range of possible values. Specifically, the @code{enum} type
8068 will be equivalent to the smallest integer type which has enough room.
8070 @item -fshort-double
8071 Use the same size for @code{double} as for @code{float}.
8074 Requests that the data and non-@code{const} variables of this
8075 compilation be shared data rather than private data. The distinction
8076 makes sense only on certain operating systems, where shared data is
8077 shared between processes running the same program, while private data
8078 exists in one copy per process.
8081 Allocate even uninitialized global variables in the data section of the
8082 object file, rather than generating them as common blocks. This has the
8083 effect that if the same variable is declared (without @code{extern}) in
8084 two different compilations, you will get an error when you link them.
8085 The only reason this might be useful is if you wish to verify that the
8086 program will work on other systems which always work this way.
8089 Ignore the @samp{#ident} directive.
8091 @item -fno-gnu-linker
8092 Do not output global initializations (such as C++ constructors and
8093 destructors) in the form used by the GNU linker (on systems where the GNU
8094 linker is the standard method of handling them). Use this option when
8095 you want to use a non-GNU linker, which also requires using the
8096 @command{collect2} program to make sure the system linker includes
8097 constructors and destructors. (@command{collect2} is included in the GCC
8098 distribution.) For systems which @emph{must} use @command{collect2}, the
8099 compiler driver @command{gcc} is configured to do this automatically.
8101 @item -finhibit-size-directive
8102 Don't output a @code{.size} assembler directive, or anything else that
8103 would cause trouble if the function is split in the middle, and the
8104 two halves are placed at locations far apart in memory. This option is
8105 used when compiling @file{crtstuff.c}; you should not need to use it
8109 Put extra commentary information in the generated assembly code to
8110 make it more readable. This option is generally only of use to those
8111 who actually need to read the generated assembly code (perhaps while
8112 debugging the compiler itself).
8114 @samp{-fno-verbose-asm}, the default, causes the
8115 extra information to be omitted and is useful when comparing two assembler
8119 Consider all memory references through pointers to be volatile.
8121 @item -fvolatile-global
8122 Consider all memory references to extern and global data items to
8123 be volatile. GCC does not consider static data items to be volatile
8124 because of this switch.
8126 @item -fvolatile-static
8127 Consider all memory references to static data to be volatile.
8130 @cindex global offset table
8132 Generate position-independent code (PIC) suitable for use in a shared
8133 library, if supported for the target machine. Such code accesses all
8134 constant addresses through a global offset table (GOT). The dynamic
8135 loader resolves the GOT entries when the program starts (the dynamic
8136 loader is not part of GCC; it is part of the operating system). If
8137 the GOT size for the linked executable exceeds a machine-specific
8138 maximum size, you get an error message from the linker indicating that
8139 @samp{-fpic} does not work; in that case, recompile with @samp{-fPIC}
8140 instead. (These maximums are 16k on the m88k, 8k on the Sparc, and 32k
8141 on the m68k and RS/6000. The 386 has no such limit.)
8143 Position-independent code requires special support, and therefore works
8144 only on certain machines. For the 386, GCC supports PIC for System V
8145 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
8146 position-independent.
8149 If supported for the target machine, emit position-independent code,
8150 suitable for dynamic linking and avoiding any limit on the size of the
8151 global offset table. This option makes a difference on the m68k, m88k,
8154 Position-independent code requires special support, and therefore works
8155 only on certain machines.
8157 @item -ffixed-@var{reg}
8158 Treat the register named @var{reg} as a fixed register; generated code
8159 should never refer to it (except perhaps as a stack pointer, frame
8160 pointer or in some other fixed role).
8162 @var{reg} must be the name of a register. The register names accepted
8163 are machine-specific and are defined in the @code{REGISTER_NAMES}
8164 macro in the machine description macro file.
8166 This flag does not have a negative form, because it specifies a
8169 @item -fcall-used-@var{reg}
8170 Treat the register named @var{reg} as an allocable register that is
8171 clobbered by function calls. It may be allocated for temporaries or
8172 variables that do not live across a call. Functions compiled this way
8173 will not save and restore the register @var{reg}.
8175 It is an error to used this flag with the frame pointer or stack pointer.
8176 Use of this flag for other registers that have fixed pervasive roles in
8177 the machine's execution model will produce disastrous results.
8179 This flag does not have a negative form, because it specifies a
8182 @item -fcall-saved-@var{reg}
8183 Treat the register named @var{reg} as an allocable register saved by
8184 functions. It may be allocated even for temporaries or variables that
8185 live across a call. Functions compiled this way will save and restore
8186 the register @var{reg} if they use it.
8188 It is an error to used this flag with the frame pointer or stack pointer.
8189 Use of this flag for other registers that have fixed pervasive roles in
8190 the machine's execution model will produce disastrous results.
8192 A different sort of disaster will result from the use of this flag for
8193 a register in which function values may be returned.
8195 This flag does not have a negative form, because it specifies a
8199 Pack all structure members together without holes. Usually you would
8200 not want to use this option, since it makes the code suboptimal, and
8201 the offsets of structure members won't agree with system libraries.
8203 @item -fcheck-memory-usage
8204 Generate extra code to check each memory access. GCC will generate
8205 code that is suitable for a detector of bad memory accesses such as
8208 Normally, you should compile all, or none, of your code with this option.
8210 If you do mix code compiled with and without this option,
8211 you must ensure that all code that has side effects
8212 and that is called by code compiled with this option
8213 is, itself, compiled with this option.
8214 If you do not, you might get erroneous messages from the detector.
8216 If you use functions from a library that have side-effects (such as
8217 @code{read}), you might not be able to recompile the library and
8218 specify this option. In that case, you can enable the
8219 @samp{-fprefix-function-name} option, which requests GCC to encapsulate
8220 your code and make other functions look as if they were compiled with
8221 @samp{-fcheck-memory-usage}. This is done by calling ``stubs'',
8222 which are provided by the detector. If you cannot find or build
8223 stubs for every function you call, you might have to specify
8224 @samp{-fcheck-memory-usage} without @samp{-fprefix-function-name}.
8226 If you specify this option, you can not use the @code{asm} or
8227 @code{__asm__} keywords in functions with memory checking enabled. GNU
8228 CC cannot understand what the @code{asm} statement may do, and therefore
8229 cannot generate the appropriate code, so it will reject it. However, if
8230 you specify the function attribute @code{no_check_memory_usage}
8231 (@pxref{Function Attributes}), GNU CC will disable memory checking within a
8232 function; you may use @code{asm} statements inside such functions. You
8233 may have an inline expansion of a non-checked function within a checked
8234 function; in that case GNU CC will not generate checks for the inlined
8235 function's memory accesses.
8237 If you move your @code{asm} statements to non-checked inline functions
8238 and they do access memory, you can add calls to the support code in your
8239 inline function, to indicate any reads, writes, or copies being done.
8240 These calls would be similar to those done in the stubs described above.
8242 @item -fprefix-function-name
8243 Request GCC to add a prefix to the symbols generated for function names.
8244 GCC adds a prefix to the names of functions defined as well as
8245 functions called. Code compiled with this option and code compiled
8246 without the option can't be linked together, unless stubs are used.
8248 If you compile the following code with @samp{-fprefix-function-name}
8250 extern void bar (int);
8259 GCC will compile the code as if it was written:
8261 extern void prefix_bar (int);
8265 return prefix_bar (a + 5);
8268 This option is designed to be used with @samp{-fcheck-memory-usage}.
8270 @item -finstrument-functions
8271 Generate instrumentation calls for entry and exit to functions. Just
8272 after function entry and just before function exit, the following
8273 profiling functions will be called with the address of the current
8274 function and its call site. (On some platforms,
8275 @code{__builtin_return_address} does not work beyond the current
8276 function, so the call site information may not be available to the
8277 profiling functions otherwise.)
8280 void __cyg_profile_func_enter (void *this_fn, void *call_site);
8281 void __cyg_profile_func_exit (void *this_fn, void *call_site);
8284 The first argument is the address of the start of the current function,
8285 which may be looked up exactly in the symbol table.
8287 This instrumentation is also done for functions expanded inline in other
8288 functions. The profiling calls will indicate where, conceptually, the
8289 inline function is entered and exited. This means that addressable
8290 versions of such functions must be available. If all your uses of a
8291 function are expanded inline, this may mean an additional expansion of
8292 code size. If you use @samp{extern inline} in your C code, an
8293 addressable version of such functions must be provided. (This is
8294 normally the case anyways, but if you get lucky and the optimizer always
8295 expands the functions inline, you might have gotten away without
8296 providing static copies.)
8298 A function may be given the attribute @code{no_instrument_function}, in
8299 which case this instrumentation will not be done. This can be used, for
8300 example, for the profiling functions listed above, high-priority
8301 interrupt routines, and any functions from which the profiling functions
8302 cannot safely be called (perhaps signal handlers, if the profiling
8303 routines generate output or allocate memory).
8306 Generate code to verify that you do not go beyond the boundary of the
8307 stack. You should specify this flag if you are running in an
8308 environment with multiple threads, but only rarely need to specify it in
8309 a single-threaded environment since stack overflow is automatically
8310 detected on nearly all systems if there is only one stack.
8312 Note that this switch does not actually cause checking to be done; the
8313 operating system must do that. The switch causes generation of code
8314 to ensure that the operating system sees the stack being extended.
8316 @item -fstack-limit-register=@var{reg}
8317 @itemx -fstack-limit-symbol=@var{sym}
8318 @itemx -fno-stack-limit
8319 Generate code to ensure that the stack does not grow beyond a certain value,
8320 either the value of a register or the address of a symbol. If the stack
8321 would grow beyond the value, a signal is raised. For most targets,
8322 the signal is raised before the stack overruns the boundary, so
8323 it is possible to catch the signal without taking special precautions.
8325 For instance, if the stack starts at address @samp{0x80000000} and grows
8326 downwards you can use the flags
8327 @samp{-fstack-limit-symbol=__stack_limit}
8328 @samp{-Wl,--defsym,__stack_limit=0x7ffe0000} which will enforce a stack
8331 @cindex aliasing of parameters
8332 @cindex parameters, aliased
8333 @item -fargument-alias
8334 @itemx -fargument-noalias
8335 @itemx -fargument-noalias-global
8336 Specify the possible relationships among parameters and between
8337 parameters and global data.
8339 @samp{-fargument-alias} specifies that arguments (parameters) may
8340 alias each other and may alias global storage.
8341 @samp{-fargument-noalias} specifies that arguments do not alias
8342 each other, but may alias global storage.
8343 @samp{-fargument-noalias-global} specifies that arguments do not
8344 alias each other and do not alias global storage.
8346 Each language will automatically use whatever option is required by
8347 the language standard. You should not need to use these options yourself.
8349 @item -fleading-underscore
8350 This option and its counterpart, -fno-leading-underscore, forcibly
8351 change the way C symbols are represented in the object file. One use
8352 is to help link with legacy assembly code.
8354 Be warned that you should know what you are doing when invoking this
8355 option, and that not all targets provide complete support for it.
8360 @node Environment Variables
8361 @section Environment Variables Affecting GCC
8362 @cindex environment variables
8364 @c man begin ENVIRONMENT
8366 This section describes several environment variables that affect how GCC
8367 operates. Some of them work by specifying directories or prefixes to use
8368 when searching for various kinds of files. Some are used to specify other
8369 aspects of the compilation environment.
8372 Note that you can also specify places to search using options such as
8373 @samp{-B}, @samp{-I} and @samp{-L} (@pxref{Directory Options}). These
8374 take precedence over places specified using environment variables, which
8375 in turn take precedence over those specified by the configuration of GCC.
8379 Note that you can also specify places to search using options such as
8380 @samp{-B}, @samp{-I} and @samp{-L} (@pxref{Directory Options}). These
8381 take precedence over places specified using environment variables, which
8382 in turn take precedence over those specified by the configuration of GCC.
8389 @c @itemx LC_COLLATE
8391 @c @itemx LC_MONETARY
8392 @c @itemx LC_NUMERIC
8397 @c @findex LC_COLLATE
8399 @c @findex LC_MONETARY
8400 @c @findex LC_NUMERIC
8404 These environment variables control the way that GCC uses
8405 localization information that allow GCC to work with different
8406 national conventions. GCC inspects the locale categories
8407 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
8408 so. These locale categories can be set to any value supported by your
8409 installation. A typical value is @samp{en_UK} for English in the United
8412 The @env{LC_CTYPE} environment variable specifies character
8413 classification. GCC uses it to determine the character boundaries in
8414 a string; this is needed for some multibyte encodings that contain quote
8415 and escape characters that would otherwise be interpreted as a string
8418 The @env{LC_MESSAGES} environment variable specifies the language to
8419 use in diagnostic messages.
8421 If the @env{LC_ALL} environment variable is set, it overrides the value
8422 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
8423 and @env{LC_MESSAGES} default to the value of the @env{LANG}
8424 environment variable. If none of these variables are set, GCC
8425 defaults to traditional C English behavior.
8429 If @env{TMPDIR} is set, it specifies the directory to use for temporary
8430 files. GCC uses temporary files to hold the output of one stage of
8431 compilation which is to be used as input to the next stage: for example,
8432 the output of the preprocessor, which is the input to the compiler
8435 @item GCC_EXEC_PREFIX
8436 @findex GCC_EXEC_PREFIX
8437 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
8438 names of the subprograms executed by the compiler. No slash is added
8439 when this prefix is combined with the name of a subprogram, but you can
8440 specify a prefix that ends with a slash if you wish.
8442 If @env{GCC_EXEC_PREFIX} is not set, GNU CC will attempt to figure out
8443 an appropriate prefix to use based on the pathname it was invoked with.
8445 If GCC cannot find the subprogram using the specified prefix, it
8446 tries looking in the usual places for the subprogram.
8448 The default value of @env{GCC_EXEC_PREFIX} is
8449 @file{@var{prefix}/lib/gcc-lib/} where @var{prefix} is the value
8450 of @code{prefix} when you ran the @file{configure} script.
8452 Other prefixes specified with @samp{-B} take precedence over this prefix.
8454 This prefix is also used for finding files such as @file{crt0.o} that are
8457 In addition, the prefix is used in an unusual way in finding the
8458 directories to search for header files. For each of the standard
8459 directories whose name normally begins with @samp{/usr/local/lib/gcc-lib}
8460 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
8461 replacing that beginning with the specified prefix to produce an
8462 alternate directory name. Thus, with @samp{-Bfoo/}, GCC will search
8463 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
8464 These alternate directories are searched first; the standard directories
8468 @findex COMPILER_PATH
8469 The value of @env{COMPILER_PATH} is a colon-separated list of
8470 directories, much like @env{PATH}. GCC tries the directories thus
8471 specified when searching for subprograms, if it can't find the
8472 subprograms using @env{GCC_EXEC_PREFIX}.
8475 @findex LIBRARY_PATH
8476 The value of @env{LIBRARY_PATH} is a colon-separated list of
8477 directories, much like @env{PATH}. When configured as a native compiler,
8478 GCC tries the directories thus specified when searching for special
8479 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
8480 using GCC also uses these directories when searching for ordinary
8481 libraries for the @samp{-l} option (but directories specified with
8482 @samp{-L} come first).
8484 @item C_INCLUDE_PATH
8485 @itemx CPLUS_INCLUDE_PATH
8486 @itemx OBJC_INCLUDE_PATH
8487 @findex C_INCLUDE_PATH
8488 @findex CPLUS_INCLUDE_PATH
8489 @findex OBJC_INCLUDE_PATH
8490 @c @itemx OBJCPLUS_INCLUDE_PATH
8491 These environment variables pertain to particular languages. Each
8492 variable's value is a colon-separated list of directories, much like
8493 @env{PATH}. When GCC searches for header files, it tries the
8494 directories listed in the variable for the language you are using, after
8495 the directories specified with @samp{-I} but before the standard header
8498 @item DEPENDENCIES_OUTPUT
8499 @findex DEPENDENCIES_OUTPUT
8500 @cindex dependencies for make as output
8501 If this variable is set, its value specifies how to output dependencies
8502 for Make based on the header files processed by the compiler. This
8503 output looks much like the output from the @samp{-M} option
8504 (@pxref{Preprocessor Options}), but it goes to a separate file, and is
8505 in addition to the usual results of compilation.
8507 The value of @env{DEPENDENCIES_OUTPUT} can be just a file name, in
8508 which case the Make rules are written to that file, guessing the target
8509 name from the source file name. Or the value can have the form
8510 @samp{@var{file} @var{target}}, in which case the rules are written to
8511 file @var{file} using @var{target} as the target name.
8515 @cindex locale definition
8516 This variable is used to pass locale information to the compiler. One way in
8517 which this information is used is to determine the character set to be used
8518 when character literals, string literals and comments are parsed in C and C++.
8519 When the compiler is configured to allow multibyte characters,
8520 the following values for @env{LANG} are recognized:
8524 Recognize JIS characters.
8526 Recognize SJIS characters.
8528 Recognize EUCJP characters.
8531 If @env{LANG} is not defined, or if it has some other value, then the
8532 compiler will use mblen and mbtowc as defined by the default locale to
8533 recognize and translate multibyte characters.
8538 @node Running Protoize
8539 @section Running Protoize
8541 The program @code{protoize} is an optional part of GNU C. You can use
8542 it to add prototypes to a program, thus converting the program to ISO
8543 C in one respect. The companion program @code{unprotoize} does the
8544 reverse: it removes argument types from any prototypes that are found.
8546 When you run these programs, you must specify a set of source files as
8547 command line arguments. The conversion programs start out by compiling
8548 these files to see what functions they define. The information gathered
8549 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
8551 After scanning comes actual conversion. The specified files are all
8552 eligible to be converted; any files they include (whether sources or
8553 just headers) are eligible as well.
8555 But not all the eligible files are converted. By default,
8556 @code{protoize} and @code{unprotoize} convert only source and header
8557 files in the current directory. You can specify additional directories
8558 whose files should be converted with the @samp{-d @var{directory}}
8559 option. You can also specify particular files to exclude with the
8560 @samp{-x @var{file}} option. A file is converted if it is eligible, its
8561 directory name matches one of the specified directory names, and its
8562 name within the directory has not been excluded.
8564 Basic conversion with @code{protoize} consists of rewriting most
8565 function definitions and function declarations to specify the types of
8566 the arguments. The only ones not rewritten are those for varargs
8569 @code{protoize} optionally inserts prototype declarations at the
8570 beginning of the source file, to make them available for any calls that
8571 precede the function's definition. Or it can insert prototype
8572 declarations with block scope in the blocks where undeclared functions
8575 Basic conversion with @code{unprotoize} consists of rewriting most
8576 function declarations to remove any argument types, and rewriting
8577 function definitions to the old-style pre-ISO form.
8579 Both conversion programs print a warning for any function declaration or
8580 definition that they can't convert. You can suppress these warnings
8583 The output from @code{protoize} or @code{unprotoize} replaces the
8584 original source file. The original file is renamed to a name ending
8585 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
8586 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
8587 for DOS) file already exists, then the source file is simply discarded.
8589 @code{protoize} and @code{unprotoize} both depend on GCC itself to
8590 scan the program and collect information about the functions it uses.
8591 So neither of these programs will work until GCC is installed.
8593 Here is a table of the options you can use with @code{protoize} and
8594 @code{unprotoize}. Each option works with both programs unless
8598 @item -B @var{directory}
8599 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
8600 usual directory (normally @file{/usr/local/lib}). This file contains
8601 prototype information about standard system functions. This option
8602 applies only to @code{protoize}.
8604 @item -c @var{compilation-options}
8605 Use @var{compilation-options} as the options when running @code{gcc} to
8606 produce the @samp{.X} files. The special option @samp{-aux-info} is
8607 always passed in addition, to tell @code{gcc} to write a @samp{.X} file.
8609 Note that the compilation options must be given as a single argument to
8610 @code{protoize} or @code{unprotoize}. If you want to specify several
8611 @code{gcc} options, you must quote the entire set of compilation options
8612 to make them a single word in the shell.
8614 There are certain @code{gcc} arguments that you cannot use, because they
8615 would produce the wrong kind of output. These include @samp{-g},
8616 @samp{-O}, @samp{-c}, @samp{-S}, and @samp{-o} If you include these in
8617 the @var{compilation-options}, they are ignored.
8620 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
8621 systems) instead of @samp{.c}. This is convenient if you are converting
8622 a C program to C++. This option applies only to @code{protoize}.
8625 Add explicit global declarations. This means inserting explicit
8626 declarations at the beginning of each source file for each function
8627 that is called in the file and was not declared. These declarations
8628 precede the first function definition that contains a call to an
8629 undeclared function. This option applies only to @code{protoize}.
8631 @item -i @var{string}
8632 Indent old-style parameter declarations with the string @var{string}.
8633 This option applies only to @code{protoize}.
8635 @code{unprotoize} converts prototyped function definitions to old-style
8636 function definitions, where the arguments are declared between the
8637 argument list and the initial @samp{@{}. By default, @code{unprotoize}
8638 uses five spaces as the indentation. If you want to indent with just
8639 one space instead, use @samp{-i " "}.
8642 Keep the @samp{.X} files. Normally, they are deleted after conversion
8646 Add explicit local declarations. @code{protoize} with @samp{-l} inserts
8647 a prototype declaration for each function in each block which calls the
8648 function without any declaration. This option applies only to
8652 Make no real changes. This mode just prints information about the conversions
8653 that would have been done without @samp{-n}.
8656 Make no @samp{.save} files. The original files are simply deleted.
8657 Use this option with caution.
8659 @item -p @var{program}
8660 Use the program @var{program} as the compiler. Normally, the name
8664 Work quietly. Most warnings are suppressed.
8667 Print the version number, just like @samp{-v} for @code{gcc}.
8670 If you need special compiler options to compile one of your program's
8671 source files, then you should generate that file's @samp{.X} file
8672 specially, by running @code{gcc} on that source file with the
8673 appropriate options and the option @samp{-aux-info}. Then run
8674 @code{protoize} on the entire set of files. @code{protoize} will use
8675 the existing @samp{.X} file because it is newer than the source file.
8679 gcc -Dfoo=bar file1.c -aux-info
8684 You need to include the special files along with the rest in the
8685 @code{protoize} command, even though their @samp{.X} files already
8686 exist, because otherwise they won't get converted.
8688 @xref{Protoize Caveats}, for more information on how to use
8689 @code{protoize} successfully.