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.
110 @xref{Option Index}, for an index to GCC's options.
113 * Option Summary:: Brief list of all options, without explanations.
114 * Overall Options:: Controlling the kind of output:
115 an executable, object files, assembler files,
116 or preprocessed source.
117 * Invoking G++:: Compiling C++ programs.
118 * C Dialect Options:: Controlling the variant of C language compiled.
119 * C++ Dialect Options:: Variations on C++.
120 * Objective-C Dialect Options:: Variations on Objective-C.
121 * Language Independent Options:: Controlling how diagnostics should be
123 * Warning Options:: How picky should the compiler be?
124 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
125 * Optimize Options:: How much optimization?
126 * Preprocessor Options:: Controlling header files and macro definitions.
127 Also, getting dependency information for Make.
128 * Assembler Options:: Passing options to the assembler.
129 * Link Options:: Specifying libraries and so on.
130 * Directory Options:: Where to find header files and libraries.
131 Where to find the compiler executable files.
132 * Spec Files:: How to pass switches to sub-processes.
133 * Target Options:: Running a cross-compiler, or an old version of GCC.
134 * Submodel Options:: Specifying minor hardware or convention variations,
135 such as 68010 vs 68020.
136 * Code Gen Options:: Specifying conventions for function calls, data layout
138 * Environment Variables:: Env vars that affect GCC.
139 * Running Protoize:: Automatically adding or removing function prototypes.
145 @section Option Summary
147 Here is a summary of all the options, grouped by type. Explanations are
148 in the following sections.
151 @item Overall Options
152 @xref{Overall Options,,Options Controlling the Kind of Output}.
154 -c -S -E -o @var{file} -pipe -pass-exit-codes -x @var{language} @gol
155 -v --target-help --help}
157 @item C Language Options
158 @xref{C Dialect Options,,Options Controlling C Dialect}.
160 -ansi -std=@var{standard} -fno-asm -fno-builtin @gol
161 -fhosted -ffreestanding @gol
162 -trigraphs -traditional -traditional-cpp @gol
163 -fallow-single-precision -fcond-mismatch @gol
164 -fsigned-bitfields -fsigned-char @gol
165 -funsigned-bitfields -funsigned-char @gol
166 -fwritable-strings -fshort-wchar}
168 @item C++ Language Options
169 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
171 -fno-access-control -fcheck-new -fconserve-space @gol
172 -fno-const-strings -fdollars-in-identifiers @gol
173 -fno-elide-constructors @gol
174 -fno-enforce-eh-specs -fexternal-templates @gol
175 -falt-external-templates @gol
176 -ffor-scope -fno-for-scope -fno-gnu-keywords -fhonor-std @gol
177 -fno-implicit-templates @gol
178 -fno-implicit-inline-templates @gol
179 -fno-implement-inlines -fms-extensions @gol
180 -fno-nonansi-builtins -fno-operator-names @gol
181 -fno-optional-diags -fpermissive @gol
182 -frepo -fno-rtti -fstats -ftemplate-depth-@var{n} @gol
183 -fuse-cxa-atexit -fvtable-gc -fno-weak -nostdinc++ @gol
184 -fno-default-inline -Wctor-dtor-privacy @gol
185 -Wnon-virtual-dtor -Wreorder @gol
186 -Weffc++ -Wno-deprecated @gol
187 -Wno-non-template-friend -Wold-style-cast @gol
188 -Woverloaded-virtual -Wno-pmf-conversions @gol
189 -Wsign-promo -Wsynth}
191 @item Objective-C Language Options
192 @xref{Objective-C Dialect Options,,Options Controlling Objective-C Dialect}.
194 -fconstant-string-class=@var{class name} @gol
195 -fgnu-runtime -fnext-runtime -gen-decls @gol
196 -Wno-protocol -Wselector}
198 @item Language Independent Options
199 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
201 -fmessage-length=@var{n} @gol
202 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]}}
204 @item Warning Options
205 @xref{Warning Options,,Options to Request or Suppress Warnings}.
207 -fsyntax-only -pedantic -pedantic-errors @gol
208 -w -W -Wall -Waggregate-return @gol
209 -Wcast-align -Wcast-qual -Wchar-subscripts -Wcomment @gol
210 -Wconversion -Wdisabled-optimization -Werror @gol
211 -Wfloat-equal -Wformat -Wformat=2 @gol
212 -Wformat-nonliteral -Wformat-security @gol
213 -Wimplicit -Wimplicit-int @gol
214 -Wimplicit-function-declaration @gol
215 -Werror-implicit-function-declaration @gol
216 -Wimport -Winline @gol
217 -Wlarger-than-@var{len} -Wlong-long @gol
218 -Wmain -Wmissing-braces -Wmissing-declarations @gol
219 -Wmissing-format-attribute -Wmissing-noreturn @gol
220 -Wmultichar -Wno-format-extra-args -Wno-format-y2k @gol
221 -Wno-import -Wpacked -Wpadded @gol
222 -Wparentheses -Wpointer-arith -Wredundant-decls @gol
223 -Wreturn-type -Wsequence-point -Wshadow @gol
224 -Wsign-compare -Wswitch -Wsystem-headers @gol
225 -Wtrigraphs -Wundef -Wuninitialized @gol
226 -Wunknown-pragmas -Wunreachable-code @gol
227 -Wunused -Wunused-function -Wunused-label -Wunused-parameter @gol
228 -Wunused-value -Wunused-variable -Wwrite-strings}
230 @item C-only Warning Options
232 -Wbad-function-cast -Wmissing-prototypes -Wnested-externs @gol
233 -Wstrict-prototypes -Wtraditional}
235 @item Debugging Options
236 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
238 -a -ax -d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
239 -fdump-unnumbered -fdump-translation-unit@r{[}-@var{n}@r{]} -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
240 -fdump-ast-original@r{[}-@var{n}@r{]} -fdump-ast-optimized@r{[}-@var{n}@r{]} @gol
241 -fmem-report -fpretend-float @gol
242 -fprofile-arcs -ftest-coverage -ftime-report @gol
243 -g -g@var{level} -gcoff -gdwarf -gdwarf-1 -gdwarf-1+ -gdwarf-2 @gol
244 -ggdb -gstabs -gstabs+ -gxcoff -gxcoff+ @gol
245 -p -pg -print-file-name=@var{library} -print-libgcc-file-name @gol
246 -print-prog-name=@var{program} -print-search-dirs -Q @gol
249 @item Optimization Options
250 @xref{Optimize Options,,Options that Control Optimization}.
252 -falign-functions=@var{n} -falign-jumps=@var{n} @gol
253 -falign-labels=@var{n} -falign-loops=@var{n} @gol
254 -fbranch-probabilities -fcaller-saves @gol
255 -fcse-follow-jumps -fcse-skip-blocks -fdata-sections -fdce @gol
256 -fdelayed-branch -fdelete-null-pointer-checks @gol
257 -fexpensive-optimizations -ffast-math -ffloat-store @gol
258 -fforce-addr -fforce-mem -ffunction-sections @gol
259 -fgcse -fgcse-lm -fgcse-sm @gol
260 -finline-functions -finline-limit=@var{n} -fkeep-inline-functions @gol
261 -fkeep-static-consts -fmove-all-movables @gol
262 -fno-default-inline -fno-defer-pop @gol
263 -fno-function-cse -fno-guess-branch-probability @gol
264 -fno-inline -fno-math-errno -fno-peephole @gol
265 -funsafe-math-optimizations -fno-trapping-math @gol
266 -fomit-frame-pointer -foptimize-register-move @gol
267 -foptimize-sibling-calls -freduce-all-givs @gol
268 -fregmove -frename-registers @gol
269 -frerun-cse-after-loop -frerun-loop-opt @gol
270 -fschedule-insns -fschedule-insns2 @gol
271 -fsingle-precision-constant -fssa @gol
272 -fstrength-reduce -fstrict-aliasing -fthread-jumps -ftrapv @gol
273 -funroll-all-loops -funroll-loops @gol
274 --param @var{name}=@var{value}
275 -O -O0 -O1 -O2 -O3 -Os}
277 @item Preprocessor Options
278 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
280 -$ -A@var{question}=@var{answer} -A-@var{question}@r{[}=@var{answer}@r{]} @gol
281 -C -dD -dI -dM -dN @gol
282 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
283 -idirafter @var{dir} @gol
284 -include @var{file} -imacros @var{file} @gol
285 -iprefix @var{file} -iwithprefix @var{dir} @gol
286 -iwithprefixbefore @var{dir} -isystem @var{dir} @gol
287 -M -MM -MF -MG -MP -MQ -MT -nostdinc -P -remap @gol
288 -trigraphs -undef -U@var{macro} -Wp\,@var{option}}
290 @item Assembler Option
291 @xref{Assembler Options,,Passing Options to the Assembler}.
296 @xref{Link Options,,Options for Linking}.
298 @var{object-file-name} -l@var{library} @gol
299 -nostartfiles -nodefaultlibs -nostdlib @gol
300 -s -static -static-libgcc -shared -shared-libgcc -symbolic @gol
301 -Wl\,@var{option} -Xlinker @var{option} @gol
304 @item Directory Options
305 @xref{Directory Options,,Options for Directory Search}.
307 -B@var{prefix} -I@var{dir} -I- -L@var{dir} -specs=@var{file}}
310 @c I wrote this xref this way to avoid overfull hbox. -- rms
311 @xref{Target Options}.
313 -b @var{machine} -V @var{version}}
315 @item Machine Dependent Options
316 @xref{Submodel Options,,Hardware Models and Configurations}.
317 @emph{M680x0 Options}
319 -m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
320 -m68060 -mcpu32 -m5200 -m68881 -mbitfield -mc68000 -mc68020 @gol
321 -mfpa -mnobitfield -mrtd -mshort -msoft-float -mpcrel @gol
322 -malign-int -mstrict-align}
324 @emph{M68hc1x Options}
326 -m6811 -m6812 -m68hc11 -m68hc12 @gol
327 -mauto-incdec -mshort -msoft-reg-count=@var{count}}
335 -mcpu=@var{cpu type} @gol
336 -mtune=@var{cpu type} @gol
337 -mcmodel=@var{code model} @gol
339 -mapp-regs -mbroken-saverestore -mcypress @gol
340 -mepilogue -mfaster-structs -mflat @gol
341 -mfpu -mhard-float -mhard-quad-float @gol
342 -mimpure-text -mlive-g0 -mno-app-regs @gol
343 -mno-epilogue -mno-faster-structs -mno-flat -mno-fpu @gol
344 -mno-impure-text -mno-stack-bias -mno-unaligned-doubles @gol
345 -msoft-float -msoft-quad-float -msparclite -mstack-bias @gol
346 -msupersparc -munaligned-doubles -mv8}
348 @emph{Convex Options}
350 -mc1 -mc2 -mc32 -mc34 -mc38 @gol
351 -margcount -mnoargcount @gol
352 -mlong32 -mlong64 @gol
353 -mvolatile-cache -mvolatile-nocache}
355 @emph{AMD29K Options}
357 -m29000 -m29050 -mbw -mnbw -mdw -mndw @gol
358 -mlarge -mnormal -msmall @gol
359 -mkernel-registers -mno-reuse-arg-regs @gol
360 -mno-stack-check -mno-storem-bug @gol
361 -mreuse-arg-regs -msoft-float -mstack-check @gol
362 -mstorem-bug -muser-registers}
366 -mapcs-frame -mno-apcs-frame @gol
367 -mapcs-26 -mapcs-32 @gol
368 -mapcs-stack-check -mno-apcs-stack-check @gol
369 -mapcs-float -mno-apcs-float @gol
370 -mapcs-reentrant -mno-apcs-reentrant @gol
371 -msched-prolog -mno-sched-prolog @gol
372 -mlittle-endian -mbig-endian -mwords-little-endian @gol
373 -malignment-traps -mno-alignment-traps @gol
374 -msoft-float -mhard-float -mfpe @gol
375 -mthumb-interwork -mno-thumb-interwork @gol
376 -mcpu=@var{name} -march=@var{name} -mfpe=@var{name} @gol
377 -mstructure-size-boundary=@var{n} @gol
378 -mbsd -mxopen -mno-symrename @gol
379 -mabort-on-noreturn @gol
380 -mlong-calls -mno-long-calls @gol
381 -msingle-pic-base -mno-single-pic-base @gol
382 -mpic-register=@var{reg} @gol
383 -mnop-fun-dllimport @gol
384 -mpoke-function-name @gol
386 -mtpcs-frame -mtpcs-leaf-frame @gol
387 -mcaller-super-interworking -mcallee-super-interworking }
389 @emph{MN10200 Options}
393 @emph{MN10300 Options}
401 @emph{M32R/D Options}
403 -mcode-model=@var{model type} -msdata=@var{sdata type} @gol
408 -m88000 -m88100 -m88110 -mbig-pic @gol
409 -mcheck-zero-division -mhandle-large-shift @gol
410 -midentify-revision -mno-check-zero-division @gol
411 -mno-ocs-debug-info -mno-ocs-frame-position @gol
412 -mno-optimize-arg-area -mno-serialize-volatile @gol
413 -mno-underscores -mocs-debug-info @gol
414 -mocs-frame-position -moptimize-arg-area @gol
415 -mserialize-volatile -mshort-data-@var{num} -msvr3 @gol
416 -msvr4 -mtrap-large-shift -muse-div-instruction @gol
417 -mversion-03.00 -mwarn-passed-structs}
419 @emph{RS/6000 and PowerPC Options}
421 -mcpu=@var{cpu type} @gol
422 -mtune=@var{cpu type} @gol
423 -mpower -mno-power -mpower2 -mno-power2 @gol
424 -mpowerpc -mpowerpc64 -mno-powerpc @gol
425 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
426 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
427 -mnew-mnemonics -mold-mnemonics @gol
428 -mfull-toc -mminimal-toc -mno-fop-in-toc -mno-sum-in-toc @gol
429 -m64 -m32 -mxl-call -mno-xl-call -mthreads -mpe @gol
430 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
431 -mstring -mno-string -mupdate -mno-update @gol
432 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
433 -mstrict-align -mno-strict-align -mrelocatable @gol
434 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
435 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
436 -mcall-aix -mcall-sysv -mprototype -mno-prototype @gol
437 -msim -mmvme -mads -myellowknife -memb -msdata @gol
438 -msdata=@var{opt} -mvxworks -G @var{num}}
442 -mcall-lib-mul -mfp-arg-in-fpregs -mfp-arg-in-gregs @gol
443 -mfull-fp-blocks -mhc-struct-return -min-line-mul @gol
444 -mminimum-fp-blocks -mnohc-struct-return}
448 -mabicalls -mcpu=@var{cpu type} @gol
449 -membedded-data -muninit-const-in-rodata @gol
450 -membedded-pic -mfp32 -mfp64 -mgas -mgp32 -mgp64 @gol
451 -mgpopt -mhalf-pic -mhard-float -mint64 -mips1 @gol
452 -mips2 -mips3 -mips4 -mlong64 -mlong32 -mlong-calls -mmemcpy @gol
453 -mmips-as -mmips-tfile -mno-abicalls @gol
454 -mno-embedded-data -mno-uninit-const-in-rodata @gol
455 -mno-embedded-pic -mno-gpopt -mno-long-calls @gol
456 -mno-memcpy -mno-mips-tfile -mno-rnames -mno-stats @gol
457 -mrnames -msoft-float @gol
458 -m4650 -msingle-float -mmad @gol
459 -mstats -EL -EB -G @var{num} -nocpp @gol
460 -mabi=32 -mabi=n32 -mabi=64 -mabi=eabi @gol
465 -mcpu=@var{cpu type} -march=@var{cpu type} @gol
466 -mintel-syntax -mieee-fp -mno-fancy-math-387 @gol
467 -mno-fp-ret-in-387 -msoft-float -msvr3-shlib @gol
468 -mno-wide-multiply -mrtd -malign-double @gol
469 -mpreferred-stack-boundary=@var{num} @gol
470 -mthreads -mno-align-stringops -minline-all-stringops @gol
471 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
472 -m96bit-long-double -mregparm=@var{num} -momit-leaf-frame-pointer}
476 -march=@var{architecture type} @gol
477 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
478 -mfast-indirect-calls -mgas -mjump-in-delay @gol
479 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
480 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
481 -mno-jump-in-delay -mno-long-load-store @gol
482 -mno-portable-runtime -mno-soft-float @gol
483 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
484 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
485 -mschedule=@var{cpu type} -mspace-regs}
487 @emph{Intel 960 Options}
489 -m@var{cpu type} -masm-compat -mclean-linkage @gol
490 -mcode-align -mcomplex-addr -mleaf-procedures @gol
491 -mic-compat -mic2.0-compat -mic3.0-compat @gol
492 -mintel-asm -mno-clean-linkage -mno-code-align @gol
493 -mno-complex-addr -mno-leaf-procedures @gol
494 -mno-old-align -mno-strict-align -mno-tail-call @gol
495 -mnumerics -mold-align -msoft-float -mstrict-align @gol
498 @emph{DEC Alpha Options}
500 -mfp-regs -mno-fp-regs -mno-soft-float -msoft-float @gol
501 -malpha-as -mgas @gol
502 -mieee -mieee-with-inexact -mieee-conformant @gol
503 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
504 -mtrap-precision=@var{mode} -mbuild-constants @gol
505 -mcpu=@var{cpu type} @gol
506 -mbwx -mno-bwx -mcix -mno-cix -mmax -mno-max @gol
507 -mmemory-latency=@var{time}}
509 @emph{Clipper Options}
513 @emph{H8/300 Options}
515 -mrelax -mh -ms -mint32 -malign-300}
519 -m1 -m2 -m3 -m3e @gol
520 -m4-nofpu -m4-single-only -m4-single -m4 @gol
521 -mb -ml -mdalign -mrelax @gol
522 -mbigtable -mfmovd -mhitachi -mnomacsave @gol
523 -misize -mpadstruct -mspace @gol
527 @emph{System V Options}
529 -Qy -Qn -YP\,@var{paths} -Ym\,@var{dir}}
534 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text section} @gol
535 -mdata=@var{data section} -mrodata=@var{readonly data section}}
537 @emph{TMS320C3x/C4x Options}
539 -mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm @gol
540 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload @gol
541 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned @gol
542 -mparallel-insns -mparallel-mpy -mpreserve-float}
546 -mlong-calls -mno-long-calls -mep -mno-ep @gol
547 -mprolog-function -mno-prolog-function -mspace @gol
548 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
553 -m32032 -m32332 -m32532 -m32081 -m32381 @gol
554 -mmult-add -mnomult-add -msoft-float -mrtd -mnortd @gol
555 -mregparam -mnoregparam -msb -mnosb @gol
556 -mbitfield -mnobitfield -mhimem -mnohimem}
560 -mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts @gol
561 -mcall-prologues -mno-tablejump -mtiny-stack}
565 -mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
566 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
567 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
568 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
569 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
573 -mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
574 -mvolatile-asm-stop -mb-step -mregister-names -mno-sdata @gol
575 -mconstant-gp -mauto-pic -minline-divide-min-latency @gol
576 -minline-divide-max-throughput -mno-dwarf2-asm @gol
577 -mfixed-range=@var{register range}}
579 @item Code Generation Options
580 @xref{Code Gen Options,,Options for Code Generation Conventions}.
582 -fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
583 -fexceptions -funwind-tables -ffixed-@var{reg} @gol
584 -finhibit-size-directive -finstrument-functions @gol
585 -fcheck-memory-usage -fprefix-function-name @gol
586 -fno-common -fno-ident -fno-gnu-linker @gol
587 -fpcc-struct-return -fpic -fPIC @gol
588 -freg-struct-return -fshared-data -fshort-enums @gol
589 -fshort-double -fvolatile @gol
590 -fvolatile-global -fvolatile-static @gol
591 -fverbose-asm -fpack-struct -fstack-check @gol
592 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
593 -fargument-alias -fargument-noalias @gol
594 -fargument-noalias-global -fleading-underscore}
598 * Overall Options:: Controlling the kind of output:
599 an executable, object files, assembler files,
600 or preprocessed source.
601 * C Dialect Options:: Controlling the variant of C language compiled.
602 * C++ Dialect Options:: Variations on C++.
603 * Objective-C Dialect Options:: Variations on Objective-C.
604 * Language Independent Options:: Controlling how diagnostics should be
606 * Warning Options:: How picky should the compiler be?
607 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
608 * Optimize Options:: How much optimization?
609 * Preprocessor Options:: Controlling header files and macro definitions.
610 Also, getting dependency information for Make.
611 * Assembler Options:: Passing options to the assembler.
612 * Link Options:: Specifying libraries and so on.
613 * Directory Options:: Where to find header files and libraries.
614 Where to find the compiler executable files.
615 * Spec Files:: How to pass switches to sub-processes.
616 * Target Options:: Running a cross-compiler, or an old version of GCC.
619 @node Overall Options
620 @section Options Controlling the Kind of Output
622 Compilation can involve up to four stages: preprocessing, compilation
623 proper, assembly and linking, always in that order. The first three
624 stages apply to an individual source file, and end by producing an
625 object file; linking combines all the object files (those newly
626 compiled, and those specified as input) into an executable file.
628 @cindex file name suffix
629 For any given input file, the file name suffix determines what kind of
634 C source code which must be preprocessed.
637 C source code which should not be preprocessed.
640 C++ source code which should not be preprocessed.
643 Objective-C source code. Note that you must link with the library
644 @file{libobjc.a} to make an Objective-C program work.
647 Objective-C source code which should not be preprocessed.
650 C header file (not to be compiled or linked).
654 @itemx @var{file}.cxx
655 @itemx @var{file}.cpp
656 @itemx @var{file}.c++
658 C++ source code which must be preprocessed. Note that in @samp{.cxx},
659 the last two letters must both be literally @samp{x}. Likewise,
660 @samp{.C} refers to a literal capital C.
663 @itemx @var{file}.for
664 @itemx @var{file}.FOR
665 Fortran source code which should not be preprocessed.
668 @itemx @var{file}.fpp
669 @itemx @var{file}.FPP
670 Fortran source code which must be preprocessed (with the traditional
674 Fortran source code which must be preprocessed with a RATFOR
675 preprocessor (not included with GCC).
677 @xref{Overall Options,,Options Controlling the Kind of Output, g77,
678 Using and Porting GNU Fortran}, for more details of the handling of
681 @c FIXME: Descriptions of Java file types.
687 @c GCC also knows about some suffixes for languages not yet included:
697 @itemx @var{file}.chi
698 CHILL source code (preprocessed with the traditional preprocessor).
704 Assembler code which must be preprocessed.
707 An object file to be fed straight into linking.
708 Any file name with no recognized suffix is treated this way.
712 You can specify the input language explicitly with the @samp{-x} option:
715 @item -x @var{language}
716 Specify explicitly the @var{language} for the following input files
717 (rather than letting the compiler choose a default based on the file
718 name suffix). This option applies to all following input files until
719 the next @samp{-x} option. Possible values for @var{language} are:
721 c c-header cpp-output
723 objective-c objc-cpp-output
724 assembler assembler-with-cpp
725 f77 f77-cpp-input ratfor
728 @c Also f77-version, for internal use only.
731 Turn off any specification of a language, so that subsequent files are
732 handled according to their file name suffixes (as they are if @samp{-x}
733 has not been used at all).
735 @item -pass-exit-codes
736 @opindex pass-exit-codes
737 Normally the @command{gcc} program will exit with the code of 1 if any
738 phase of the compiler returns a non-success return code. If you specify
739 @samp{-pass-exit-codes}, the @command{gcc} program will instead return with
740 numerically highest error produced by any phase that returned an error
744 If you only want some of the stages of compilation, you can use
745 @samp{-x} (or filename suffixes) to tell @command{gcc} where to start, and
746 one of the options @samp{-c}, @samp{-S}, or @samp{-E} to say where
747 @command{gcc} is to stop. Note that some combinations (for example,
748 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
753 Compile or assemble the source files, but do not link. The linking
754 stage simply is not done. The ultimate output is in the form of an
755 object file for each source file.
757 By default, the object file name for a source file is made by replacing
758 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
760 Unrecognized input files, not requiring compilation or assembly, are
765 Stop after the stage of compilation proper; do not assemble. The output
766 is in the form of an assembler code file for each non-assembler input
769 By default, the assembler file name for a source file is made by
770 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
772 Input files that don't require compilation are ignored.
776 Stop after the preprocessing stage; do not run the compiler proper. The
777 output is in the form of preprocessed source code, which is sent to the
780 Input files which don't require preprocessing are ignored.
782 @cindex output file option
785 Place output in file @var{file}. This applies regardless to whatever
786 sort of output is being produced, whether it be an executable file,
787 an object file, an assembler file or preprocessed C code.
789 Since only one output file can be specified, it does not make sense to
790 use @samp{-o} when compiling more than one input file, unless you are
791 producing an executable file as output.
793 If @samp{-o} is not specified, the default is to put an executable file
794 in @file{a.out}, the object file for @file{@var{source}.@var{suffix}} in
795 @file{@var{source}.o}, its assembler file in @file{@var{source}.s}, and
796 all preprocessed C source on standard output.@refill
800 Print (on standard error output) the commands executed to run the stages
801 of compilation. Also print the version number of the compiler driver
802 program and of the preprocessor and the compiler proper.
806 Use pipes rather than temporary files for communication between the
807 various stages of compilation. This fails to work on some systems where
808 the assembler is unable to read from a pipe; but the GNU assembler has
813 Print (on the standard output) a description of the command line options
814 understood by @command{gcc}. If the @option{-v} option is also specified
815 then @option{--help} will also be passed on to the various processes
816 invoked by @command{gcc}, so that they can display the command line options
817 they accept. If the @option{-W} option is also specified then command
818 line options which have no documentation associated with them will also
823 Print (on the standard output) a description of target specific command
824 line options for each tool.
828 @section Compiling C++ Programs
830 @cindex suffixes for C++ source
831 @cindex C++ source file suffixes
832 C++ source files conventionally use one of the suffixes @samp{.C},
833 @samp{.cc}, @samp{.cpp}, @samp{.c++}, @samp{.cp}, or @samp{.cxx};
834 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
835 files with these names and compiles them as C++ programs even if you
836 call the compiler the same way as for compiling C programs (usually with
837 the name @command{gcc}).
841 However, C++ programs often require class libraries as well as a
842 compiler that understands the C++ language---and under some
843 circumstances, you might want to compile programs from standard input,
844 or otherwise without a suffix that flags them as C++ programs.
845 @command{g++} is a program that calls GCC with the default language
846 set to C++, and automatically specifies linking against the C++
847 library. On many systems, @command{g++} is also
848 installed with the name @command{c++}.
850 @cindex invoking @command{g++}
851 When you compile C++ programs, you may specify many of the same
852 command-line options that you use for compiling programs in any
853 language; or command-line options meaningful for C and related
854 languages; or options that are meaningful only for C++ programs.
855 @xref{C Dialect Options,,Options Controlling C Dialect}, for
856 explanations of options for languages related to C.
857 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
858 explanations of options that are meaningful only for C++ programs.
860 @node C Dialect Options
861 @section Options Controlling C Dialect
862 @cindex dialect options
863 @cindex language dialect options
864 @cindex options, dialect
866 The following options control the dialect of C (or languages derived
867 from C, such as C++ and Objective C) that the compiler accepts:
874 In C mode, support all ISO C89 programs. In C++ mode,
875 remove GNU extensions that conflict with ISO C++.
877 This turns off certain features of GCC that are incompatible with ISO
878 C (when compiling C code), or of standard C++ (when compiling C++ code),
879 such as the @code{asm} and @code{typeof} keywords, and
880 predefined macros such as @code{unix} and @code{vax} that identify the
881 type of system you are using. It also enables the undesirable and
882 rarely used ISO trigraph feature. For the C compiler,
883 it disables recognition of C++ style @samp{//} comments as well as
884 the @code{inline} keyword.
886 The alternate keywords @code{__asm__}, @code{__extension__},
887 @code{__inline__} and @code{__typeof__} continue to work despite
888 @samp{-ansi}. You would not want to use them in an ISO C program, of
889 course, but it is useful to put them in header files that might be included
890 in compilations done with @samp{-ansi}. Alternate predefined macros
891 such as @code{__unix__} and @code{__vax__} are also available, with or
892 without @samp{-ansi}.
894 The @samp{-ansi} option does not cause non-ISO programs to be
895 rejected gratuitously. For that, @samp{-pedantic} is required in
896 addition to @samp{-ansi}. @xref{Warning Options}.
898 The macro @code{__STRICT_ANSI__} is predefined when the @samp{-ansi}
899 option is used. Some header files may notice this macro and refrain
900 from declaring certain functions or defining certain macros that the
901 ISO standard doesn't call for; this is to avoid interfering with any
902 programs that might use these names for other things.
904 Functions which would normally be builtin but do not have semantics
905 defined by ISO C (such as @code{alloca} and @code{ffs}) are not builtin
906 functions with @samp{-ansi} is used. @xref{Other Builtins,,Other
907 built-in functions provided by GNU CC}, for details of the functions
912 Determine the language standard. A value for this option must be provided;
917 Same as @option{-ansi}
920 ISO C as modified in amend. 1
923 ISO C99. Note that this standard is not yet fully supported; see
924 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information.
927 same as @option{-std=iso9899:1990}
930 same as @option{-std=iso9899:1999}
933 default, iso9899:1990 + gnu extensions
936 iso9899:1999 + gnu extensions
939 same as @option{-std=iso9899:1999}, deprecated
942 same as @option{-std=iso9899:1999}, deprecated
945 same as @option{-std=gnu99}, deprecated
949 Even when this option is not specified, you can still use some of the
950 features of newer standards in so far as they do not conflict with
951 previous C standards. For example, you may use @code{__restrict__} even
952 when @option{-std=c99} is not specified.
954 The @option{-std} options specifying some version of ISO C have the same
955 effects as @option{-ansi}, except that features that were not in ISO C89
956 but are in the specified version (for example, @samp{//} comments and
957 the @code{inline} keyword in ISO C99) are not disabled.
959 @xref{Standards,,Language Standards Supported by GCC}, for details of
960 these standard versions.
964 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
965 keyword, so that code can use these words as identifiers. You can use
966 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
967 instead. @samp{-ansi} implies @samp{-fno-asm}.
969 In C++, this switch only affects the @code{typeof} keyword, since
970 @code{asm} and @code{inline} are standard keywords. You may want to
971 use the @samp{-fno-gnu-keywords} flag instead, which has the same
972 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
973 switch only affects the @code{asm} and @code{typeof} keywords, since
974 @code{inline} is a standard keyword in ISO C99.
978 @cindex builtin functions
979 Don't recognize builtin functions that do not begin with
980 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
981 functions provided by GNU CC}, for details of the functions affected,
982 including those which are not builtin functions when @option{-ansi} or
983 @option{-std} options for strict ISO C conformance are used because they
984 do not have an ISO standard meaning.
986 GCC normally generates special code to handle certain builtin functions
987 more efficiently; for instance, calls to @code{alloca} may become single
988 instructions that adjust the stack directly, and calls to @code{memcpy}
989 may become inline copy loops. The resulting code is often both smaller
990 and faster, but since the function calls no longer appear as such, you
991 cannot set a breakpoint on those calls, nor can you change the behavior
992 of the functions by linking with a different library.
994 In C++, @samp{-fno-builtin} is always in effect. The @samp{-fbuiltin}
995 option has no effect. Therefore, in C++, the only way to get the
996 optimization benefits of builtin functions is to call the function
997 using the @samp{__builtin_} prefix. The GNU C++ Standard Library uses
998 builtin functions to implement many functions (like
999 @code{std::strchr}), so that you automatically get efficient code.
1003 @cindex hosted environment
1005 Assert that compilation takes place in a hosted environment. This implies
1006 @samp{-fbuiltin}. A hosted environment is one in which the
1007 entire standard library is available, and in which @code{main} has a return
1008 type of @code{int}. Examples are nearly everything except a kernel.
1009 This is equivalent to @samp{-fno-freestanding}.
1011 @item -ffreestanding
1012 @opindex ffreestanding
1013 @cindex hosted environment
1015 Assert that compilation takes place in a freestanding environment. This
1016 implies @samp{-fno-builtin}. A freestanding environment
1017 is one in which the standard library may not exist, and program startup may
1018 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1019 This is equivalent to @samp{-fno-hosted}.
1021 @xref{Standards,,Language Standards Supported by GCC}, for details of
1022 freestanding and hosted environments.
1026 Support ISO C trigraphs. You don't want to know about this
1027 brain-damage. The @option{-ansi} option (and @option{-std} options for
1028 strict ISO C conformance) implies @option{-trigraphs}.
1030 @cindex traditional C language
1031 @cindex C language, traditional
1033 @opindex traditional
1034 Attempt to support some aspects of traditional C compilers.
1039 All @code{extern} declarations take effect globally even if they
1040 are written inside of a function definition. This includes implicit
1041 declarations of functions.
1044 The newer keywords @code{typeof}, @code{inline}, @code{signed}, @code{const}
1045 and @code{volatile} are not recognized. (You can still use the
1046 alternative keywords such as @code{__typeof__}, @code{__inline__}, and
1050 Comparisons between pointers and integers are always allowed.
1053 Integer types @code{unsigned short} and @code{unsigned char} promote
1054 to @code{unsigned int}.
1057 Out-of-range floating point literals are not an error.
1060 Certain constructs which ISO regards as a single invalid preprocessing
1061 number, such as @samp{0xe-0xd}, are treated as expressions instead.
1064 String ``constants'' are not necessarily constant; they are stored in
1065 writable space, and identical looking constants are allocated
1066 separately. (This is the same as the effect of
1067 @samp{-fwritable-strings}.)
1069 @cindex @code{longjmp} and automatic variables
1071 All automatic variables not declared @code{register} are preserved by
1072 @code{longjmp}. Ordinarily, GNU C follows ISO C: automatic variables
1073 not declared @code{volatile} may be clobbered.
1078 @cindex escape sequences, traditional
1079 The character escape sequences @samp{\x} and @samp{\a} evaluate as the
1080 literal characters @samp{x} and @samp{a} respectively. Without
1081 @w{@samp{-traditional}}, @samp{\x} is a prefix for the hexadecimal
1082 representation of a character, and @samp{\a} produces a bell.
1085 You may wish to use @samp{-fno-builtin} as well as @samp{-traditional}
1086 if your program uses names that are normally GNU C builtin functions for
1087 other purposes of its own.
1089 You cannot use @samp{-traditional} if you include any header files that
1090 rely on ISO C features. Some vendors are starting to ship systems with
1091 ISO C header files and you cannot use @samp{-traditional} on such
1092 systems to compile files that include any system headers.
1094 The @samp{-traditional} option also enables @samp{-traditional-cpp},
1095 which is described next.
1097 @item -traditional-cpp
1098 @opindex traditional-cpp
1099 Attempt to support some aspects of traditional C preprocessors.
1104 Comments convert to nothing at all, rather than to a space. This allows
1105 traditional token concatenation.
1108 In a preprocessing directive, the @samp{#} symbol must appear as the first
1109 character of a line.
1112 Macro arguments are recognized within string constants in a macro
1113 definition (and their values are stringified, though without additional
1114 quote marks, when they appear in such a context). The preprocessor
1115 always considers a string constant to end at a newline.
1118 @cindex detecting @w{@samp{-traditional}}
1119 The predefined macro @code{__STDC__} is not defined when you use
1120 @samp{-traditional}, but @code{__GNUC__} is (since the GNU extensions
1121 which @code{__GNUC__} indicates are not affected by
1122 @samp{-traditional}). If you need to write header files that work
1123 differently depending on whether @samp{-traditional} is in use, by
1124 testing both of these predefined macros you can distinguish four
1125 situations: GNU C, traditional GNU C, other ISO C compilers, and other
1126 old C compilers. The predefined macro @code{__STDC_VERSION__} is also
1127 not defined when you use @samp{-traditional}. @xref{Standard
1128 Predefined,,Standard Predefined Macros,cpp.info,The C Preprocessor},
1129 for more discussion of these and other predefined macros.
1132 @cindex string constants vs newline
1133 @cindex newline vs string constants
1134 The preprocessor considers a string constant to end at a newline (unless
1135 the newline is escaped with @samp{\}). (Without @w{@samp{-traditional}},
1136 string constants can contain the newline character as typed.)
1139 @item -fcond-mismatch
1140 @opindex fcond-mismatch
1141 Allow conditional expressions with mismatched types in the second and
1142 third arguments. The value of such an expression is void. This option
1143 is not supported for C++.
1145 @item -funsigned-char
1146 @opindex funsigned-char
1147 Let the type @code{char} be unsigned, like @code{unsigned char}.
1149 Each kind of machine has a default for what @code{char} should
1150 be. It is either like @code{unsigned char} by default or like
1151 @code{signed char} by default.
1153 Ideally, a portable program should always use @code{signed char} or
1154 @code{unsigned char} when it depends on the signedness of an object.
1155 But many programs have been written to use plain @code{char} and
1156 expect it to be signed, or expect it to be unsigned, depending on the
1157 machines they were written for. This option, and its inverse, let you
1158 make such a program work with the opposite default.
1160 The type @code{char} is always a distinct type from each of
1161 @code{signed char} or @code{unsigned char}, even though its behavior
1162 is always just like one of those two.
1165 @opindex fsigned-char
1166 Let the type @code{char} be signed, like @code{signed char}.
1168 Note that this is equivalent to @samp{-fno-unsigned-char}, which is
1169 the negative form of @samp{-funsigned-char}. Likewise, the option
1170 @samp{-fno-signed-char} is equivalent to @samp{-funsigned-char}.
1172 You may wish to use @samp{-fno-builtin} as well as @samp{-traditional}
1173 if your program uses names that are normally GNU C builtin functions for
1174 other purposes of its own.
1176 You cannot use @samp{-traditional} if you include any header files that
1177 rely on ISO C features. Some vendors are starting to ship systems with
1178 ISO C header files and you cannot use @samp{-traditional} on such
1179 systems to compile files that include any system headers.
1181 @item -fsigned-bitfields
1182 @itemx -funsigned-bitfields
1183 @itemx -fno-signed-bitfields
1184 @itemx -fno-unsigned-bitfields
1185 @opindex fsigned-bitfields
1186 @opindex funsigned-bitfields
1187 @opindex fno-signed-bitfields
1188 @opindex fno-unsigned-bitfields
1189 These options control whether a bitfield is signed or unsigned, when the
1190 declaration does not use either @code{signed} or @code{unsigned}. By
1191 default, such a bitfield is signed, because this is consistent: the
1192 basic integer types such as @code{int} are signed types.
1194 However, when @samp{-traditional} is used, bitfields are all unsigned
1197 @item -fwritable-strings
1198 @opindex fwritable-strings
1199 Store string constants in the writable data segment and don't uniquize
1200 them. This is for compatibility with old programs which assume they can
1201 write into string constants. The option @samp{-traditional} also has
1204 Writing into string constants is a very bad idea; ``constants'' should
1207 @item -fallow-single-precision
1208 @opindex fallow-single-precision
1209 Do not promote single precision math operations to double precision,
1210 even when compiling with @samp{-traditional}.
1212 Traditional K&R C promotes all floating point operations to double
1213 precision, regardless of the sizes of the operands. On the
1214 architecture for which you are compiling, single precision may be faster
1215 than double precision. If you must use @samp{-traditional}, but want
1216 to use single precision operations when the operands are single
1217 precision, use this option. This option has no effect when compiling
1218 with ISO or GNU C conventions (the default).
1221 @opindex fshort-wchar
1222 Override the underlying type for @samp{wchar_t} to be @samp{short
1223 unsigned int} instead of the default for the target. This option is
1224 useful for building programs to run under WINE.
1227 @node C++ Dialect Options
1228 @section Options Controlling C++ Dialect
1230 @cindex compiler options, C++
1231 @cindex C++ options, command line
1232 @cindex options, C++
1233 This section describes the command-line options that are only meaningful
1234 for C++ programs; but you can also use most of the GNU compiler options
1235 regardless of what language your program is in. For example, you
1236 might compile a file @code{firstClass.C} like this:
1239 g++ -g -frepo -O -c firstClass.C
1243 In this example, only @samp{-frepo} is an option meant
1244 only for C++ programs; you can use the other options with any
1245 language supported by GCC.
1247 Here is a list of options that are @emph{only} for compiling C++ programs:
1250 @item -fno-access-control
1251 @opindex fno-access-control
1252 Turn off all access checking. This switch is mainly useful for working
1253 around bugs in the access control code.
1257 Check that the pointer returned by @code{operator new} is non-null
1258 before attempting to modify the storage allocated. The current Working
1259 Paper requires that @code{operator new} never return a null pointer, so
1260 this check is normally unnecessary.
1262 An alternative to using this option is to specify that your
1263 @code{operator new} does not throw any exceptions; if you declare it
1264 @samp{throw()}, g++ will check the return value. See also @samp{new
1267 @item -fconserve-space
1268 @opindex fconserve-space
1269 Put uninitialized or runtime-initialized global variables into the
1270 common segment, as C does. This saves space in the executable at the
1271 cost of not diagnosing duplicate definitions. If you compile with this
1272 flag and your program mysteriously crashes after @code{main()} has
1273 completed, you may have an object that is being destroyed twice because
1274 two definitions were merged.
1276 This option is no longer useful on most targets, now that support has
1277 been added for putting variables into BSS without making them common.
1279 @item -fno-const-strings
1280 @opindex fno-const-strings
1281 Give string constants type @code{char *} instead of type @code{const
1282 char *}. By default, G++ uses type @code{const char *} as required by
1283 the standard. Even if you use @samp{-fno-const-strings}, you cannot
1284 actually modify the value of a string constant, unless you also use
1285 @samp{-fwritable-strings}.
1287 This option might be removed in a future release of G++. For maximum
1288 portability, you should structure your code so that it works with
1289 string constants that have type @code{const char *}.
1291 @item -fdollars-in-identifiers
1292 @opindex fdollars-in-identifiers
1293 Accept @samp{$} in identifiers. You can also explicitly prohibit use of
1294 @samp{$} with the option @samp{-fno-dollars-in-identifiers}. (GNU C allows
1295 @samp{$} by default on most target systems, but there are a few exceptions.)
1296 Traditional C allowed the character @samp{$} to form part of
1297 identifiers. However, ISO C and C++ forbid @samp{$} in identifiers.
1299 @item -fno-elide-constructors
1300 @opindex fno-elide-constructors
1301 The C++ standard allows an implementation to omit creating a temporary
1302 which is only used to initialize another object of the same type.
1303 Specifying this option disables that optimization, and forces g++ to
1304 call the copy constructor in all cases.
1306 @item -fno-enforce-eh-specs
1307 @opindex fno-enforce-eh-specs
1308 Don't check for violation of exception specifications at runtime. This
1309 option violates the C++ standard, but may be useful for reducing code
1310 size in production builds, much like defining @samp{NDEBUG}. The compiler
1311 will still optimize based on the exception specifications.
1313 @item -fexternal-templates
1314 @opindex fexternal-templates
1315 Cause template instantiations to obey @samp{#pragma interface} and
1316 @samp{implementation}; template instances are emitted or not according
1317 to the location of the template definition. @xref{Template
1318 Instantiation}, for more information.
1320 This option is deprecated.
1322 @item -falt-external-templates
1323 @opindex falt-external-templates
1324 Similar to -fexternal-templates, but template instances are emitted or
1325 not according to the place where they are first instantiated.
1326 @xref{Template Instantiation}, for more information.
1328 This option is deprecated.
1331 @itemx -fno-for-scope
1333 @opindex fno-for-scope
1334 If -ffor-scope is specified, the scope of variables declared in
1335 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1336 as specified by the C++ standard.
1337 If -fno-for-scope is specified, the scope of variables declared in
1338 a @i{for-init-statement} extends to the end of the enclosing scope,
1339 as was the case in old versions of gcc, and other (traditional)
1340 implementations of C++.
1342 The default if neither flag is given to follow the standard,
1343 but to allow and give a warning for old-style code that would
1344 otherwise be invalid, or have different behavior.
1346 @item -fno-gnu-keywords
1347 @opindex fno-gnu-keywords
1348 Do not recognize @code{typeof} as a keyword, so that code can use this
1349 word as an identifier. You can use the keyword @code{__typeof__} instead.
1350 @samp{-ansi} implies @samp{-fno-gnu-keywords}.
1354 Treat the @code{namespace std} as a namespace, instead of ignoring
1355 it. For compatibility with earlier versions of g++, the compiler will,
1356 by default, ignore @code{namespace-declarations},
1357 @code{using-declarations}, @code{using-directives}, and
1358 @code{namespace-names}, if they involve @code{std}.
1360 @item -fno-implicit-templates
1361 @opindex fno-implicit-templates
1362 Never emit code for non-inline templates which are instantiated
1363 implicitly (i.e. by use); only emit code for explicit instantiations.
1364 @xref{Template Instantiation}, for more information.
1366 @item -fno-implicit-inline-templates
1367 @opindex fno-implicit-inline-templates
1368 Don't emit code for implicit instantiations of inline templates, either.
1369 The default is to handle inlines differently so that compiles with and
1370 without optimization will need the same set of explicit instantiations.
1372 @item -fno-implement-inlines
1373 @opindex fno-implement-inlines
1374 To save space, do not emit out-of-line copies of inline functions
1375 controlled by @samp{#pragma implementation}. This will cause linker
1376 errors if these functions are not inlined everywhere they are called.
1378 @item -fms-extensions
1379 @opindex fms-extensions
1380 Disable pedantic warnings about constructs used in MFC, such as implicit
1381 int and getting a pointer to member function via non-standard syntax.
1383 @item -fno-nonansi-builtins
1384 @opindex fno-nonansi-builtins
1385 Disable builtin declarations of functions that are not mandated by
1386 ANSI/ISO C. These include @code{ffs}, @code{alloca}, @code{_exit},
1387 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1389 @item -fno-operator-names
1390 @opindex fno-operator-names
1391 Do not treat the operator name keywords @code{and}, @code{bitand},
1392 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1393 synonyms as keywords.
1395 @item -fno-optional-diags
1396 @opindex fno-optional-diags
1397 Disable diagnostics that the standard says a compiler does not need to
1398 issue. Currently, the only such diagnostic issued by g++ is the one for
1399 a name having multiple meanings within a class.
1402 @opindex fpermissive
1403 Downgrade messages about nonconformant code from errors to warnings. By
1404 default, g++ effectively sets @samp{-pedantic-errors} without
1405 @samp{-pedantic}; this option reverses that. This behavior and this
1406 option are superseded by @samp{-pedantic}, which works as it does for GNU C.
1410 Enable automatic template instantiation. This option also implies
1411 @samp{-fno-implicit-templates}. @xref{Template Instantiation}, for more
1416 Disable generation of information about every class with virtual
1417 functions for use by the C++ runtime type identification features
1418 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1419 of the language, you can save some space by using this flag. Note that
1420 exception handling uses the same information, but it will generate it as
1425 Emit statistics about front-end processing at the end of the compilation.
1426 This information is generally only useful to the G++ development team.
1428 @item -ftemplate-depth-@var{n}
1429 @opindex ftemplate-depth
1430 Set the maximum instantiation depth for template classes to @var{n}.
1431 A limit on the template instantiation depth is needed to detect
1432 endless recursions during template class instantiation. ANSI/ISO C++
1433 conforming programs must not rely on a maximum depth greater than 17.
1435 @item -fuse-cxa-atexit
1436 @opindex fuse-cxa-atexit
1437 Register destructors for objects with static storage duration with the
1438 @code{__cxa_atexit} function rather than the @code{atexit} function.
1439 This option is required for fully standards-compliant handling of static
1440 destructors, but will only work if your C library supports
1441 @code{__cxa_atexit}.
1445 Emit special relocations for vtables and virtual function references
1446 so that the linker can identify unused virtual functions and zero out
1447 vtable slots that refer to them. This is most useful with
1448 @samp{-ffunction-sections} and @samp{-Wl,--gc-sections}, in order to
1449 also discard the functions themselves.
1451 This optimization requires GNU as and GNU ld. Not all systems support
1452 this option. @samp{-Wl,--gc-sections} is ignored without @samp{-static}.
1456 Do not use weak symbol support, even if it is provided by the linker.
1457 By default, G++ will use weak symbols if they are available. This
1458 option exists only for testing, and should not be used by end-users;
1459 it will result in inferior code and has no benefits. This option may
1460 be removed in a future release of G++.
1464 Do not search for header files in the standard directories specific to
1465 C++, but do still search the other standard directories. (This option
1466 is used when building the C++ library.)
1469 In addition, these optimization, warning, and code generation options
1470 have meanings only for C++ programs:
1473 @item -fno-default-inline
1474 @opindex fno-default-inline
1475 Do not assume @samp{inline} for functions defined inside a class scope.
1476 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1477 functions will have linkage like inline functions; they just won't be
1480 @item -Wctor-dtor-privacy (C++ only)
1481 @opindex Wctor-dtor-privacy
1482 Warn when a class seems unusable, because all the constructors or
1483 destructors in a class are private and the class has no friends or
1484 public static member functions.
1486 @item -Wnon-virtual-dtor (C++ only)
1487 @opindex Wnon-virtual-dtor
1488 Warn when a class declares a non-virtual destructor that should probably
1489 be virtual, because it looks like the class will be used polymorphically.
1491 @item -Wreorder (C++ only)
1493 @cindex reordering, warning
1494 @cindex warning for reordering of member initializers
1495 Warn when the order of member initializers given in the code does not
1496 match the order in which they must be executed. For instance:
1502 A(): j (0), i (1) @{ @}
1506 Here the compiler will warn that the member initializers for @samp{i}
1507 and @samp{j} will be rearranged to match the declaration order of the
1511 The following @samp{-W@dots{}} options are not affected by @samp{-Wall}.
1514 @item -Weffc++ (C++ only)
1516 Warn about violations of various style guidelines from Scott Meyers'
1517 @cite{Effective C++} books. If you use this option, you should be aware
1518 that the standard library headers do not obey all of these guidelines;
1519 you can use @samp{grep -v} to filter out those warnings.
1521 @item -Wno-deprecated (C++ only)
1522 @opindex Wno-deprecated
1523 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1525 @item -Wno-non-template-friend (C++ only)
1526 @opindex Wno-non-template-friend
1527 Disable warnings when non-templatized friend functions are declared
1528 within a template. With the advent of explicit template specification
1529 support in g++, if the name of the friend is an unqualified-id (ie,
1530 @samp{friend foo(int)}), the C++ language specification demands that the
1531 friend declare or define an ordinary, nontemplate function. (Section
1532 14.5.3). Before g++ implemented explicit specification, unqualified-ids
1533 could be interpreted as a particular specialization of a templatized
1534 function. Because this non-conforming behavior is no longer the default
1535 behavior for g++, @samp{-Wnon-template-friend} allows the compiler to
1536 check existing code for potential trouble spots, and is on by default.
1537 This new compiler behavior can be turned off with
1538 @samp{-Wno-non-template-friend} which keeps the conformant compiler code
1539 but disables the helpful warning.
1541 @item -Wold-style-cast (C++ only)
1542 @opindex Wold-style-cast
1543 Warn if an old-style (C-style) cast is used within a C++ program. The
1544 new-style casts (@samp{static_cast}, @samp{reinterpret_cast}, and
1545 @samp{const_cast}) are less vulnerable to unintended effects, and much
1548 @item -Woverloaded-virtual (C++ only)
1549 @opindex Woverloaded-virtual
1550 @cindex overloaded virtual fn, warning
1551 @cindex warning for overloaded virtual fn
1552 Warn when a function declaration hides virtual functions from a
1553 base class. For example, in:
1560 struct B: public A @{
1565 the @code{A} class version of @code{f} is hidden in @code{B}, and code
1573 will fail to compile.
1575 @item -Wno-pmf-conversions (C++ only)
1576 @opindex Wno-pmf-conversions
1577 Disable the diagnostic for converting a bound pointer to member function
1580 @item -Wsign-promo (C++ only)
1581 @opindex Wsign-promo
1582 Warn when overload resolution chooses a promotion from unsigned or
1583 enumeral type to a signed type over a conversion to an unsigned type of
1584 the same size. Previous versions of g++ would try to preserve
1585 unsignedness, but the standard mandates the current behavior.
1587 @item -Wsynth (C++ only)
1589 @cindex warning for synthesized methods
1590 @cindex synthesized methods, warning
1591 Warn when g++'s synthesis behavior does not match that of cfront. For
1597 A& operator = (int);
1607 In this example, g++ will synthesize a default @samp{A& operator =
1608 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1611 @node Objective-C Dialect Options
1612 @section Options Controlling Objective-C Dialect
1614 @cindex compiler options, Objective-C
1615 @cindex Objective-C options, command line
1616 @cindex options, Objective-C
1617 This section describes the command-line options that are only meaningful
1618 for Objective-C programs; but you can also use most of the GNU compiler
1619 options regardless of what language your program is in. For example,
1620 you might compile a file @code{some_class.m} like this:
1623 gcc -g -fgnu-runtime -O -c some_class.m
1627 In this example, only @samp{-fgnu-runtime} is an option meant only for
1628 Objective-C programs; you can use the other options with any language
1631 Here is a list of options that are @emph{only} for compiling Objective-C
1635 @item -fconstant-string-class=@var{class name}
1636 @opindex fconstant-string-class
1637 Use @var{class name} as the name of the class to instantiate for each
1638 literal string specified with the syntax @code{@@"..."}. The default
1639 class name is @code{NXConstantString}.
1642 @opindex fgnu-runtime
1643 Generate object code compatible with the standard GNU Objective-C
1644 runtime. This is the default for most types of systems.
1646 @item -fnext-runtime
1647 @opindex fnext-runtime
1648 Generate output compatible with the NeXT runtime. This is the default
1649 for NeXT-based systems, including Darwin and Mac OS X.
1653 Dump interface declarations for all classes seen in the source file to a
1654 file named @file{@var{sourcename}.decl}.
1657 @opindex Wno-protocol
1658 Do not warn if methods required by a protocol are not implemented
1659 in the class adopting it.
1663 Warn if a selector has multiple methods of different types defined.
1665 @c not documented because only avail via -Wp
1666 @c @item -print-objc-runtime-info
1670 @node Language Independent Options
1671 @section Options to Control Diagnostic Messages Formatting
1672 @cindex options to control diagnostics formatting
1673 @cindex diagnostic messages
1674 @cindex message formatting
1676 Traditionally, diagnostic messages have been formatted irrespective of
1677 the output device's aspect (e.g. its width, ...). The options described
1678 below can be used to control the diagnostic messages formatting
1679 algorithm, e.g. how many characters per line, how often source location
1680 information should be reported. Right now, only the C++ front-end can
1681 honor these options. However it is expected, in the near future, that
1682 the remaining front-ends would be able to digest them correctly.
1685 @item -fmessage-length=@var{n}
1686 @opindex fmessage-length
1687 Try to format error messages so that they fit on lines of about @var{n}
1688 characters. The default is 72 characters for g++ and 0 for the rest of
1689 the front-ends supported by GCC. If @var{n} is zero, then no
1690 line-wrapping will be done; each error message will appear on a single
1693 @opindex fdiagnostics-show-location
1694 @item -fdiagnostics-show-location=once
1695 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
1696 reporter to emit @emph{once} source location information; that is, in
1697 case the message is too long to fit on a single physical line and has to
1698 be wrapped, the source location won't be emitted (as prefix) again,
1699 over and over, in subsequent continuation lines. This is the default
1702 @item -fdiagnostics-show-location=every-line
1703 Only meaningful in line-wrapping mode. Instructs the diagnostic
1704 messages reporter to emit the same source location information (as
1705 prefix) for physical lines that result from the process of breaking a
1706 a message which is too long to fit on a single line.
1710 @node Warning Options
1711 @section Options to Request or Suppress Warnings
1712 @cindex options to control warnings
1713 @cindex warning messages
1714 @cindex messages, warning
1715 @cindex suppressing warnings
1717 Warnings are diagnostic messages that report constructions which
1718 are not inherently erroneous but which are risky or suggest there
1719 may have been an error.
1721 You can request many specific warnings with options beginning @samp{-W},
1722 for example @samp{-Wimplicit} to request warnings on implicit
1723 declarations. Each of these specific warning options also has a
1724 negative form beginning @samp{-Wno-} to turn off warnings;
1725 for example, @samp{-Wno-implicit}. This manual lists only one of the
1726 two forms, whichever is not the default.
1728 These options control the amount and kinds of warnings produced by GCC:
1731 @cindex syntax checking
1733 @opindex fsyntax-only
1734 Check the code for syntax errors, but don't do anything beyond that.
1738 Issue all the warnings demanded by strict ISO C and ISO C++;
1739 reject all programs that use forbidden extensions, and some other
1740 programs that do not follow ISO C and ISO C++. For ISO C, follows the
1741 version of the ISO C standard specified by any @samp{-std} option used.
1743 Valid ISO C and ISO C++ programs should compile properly with or without
1744 this option (though a rare few will require @option{-ansi} or a
1745 @option{-std} option specifying the required version of ISO C). However,
1746 without this option, certain GNU extensions and traditional C and C++
1747 features are supported as well. With this option, they are rejected.
1749 @samp{-pedantic} does not cause warning messages for use of the
1750 alternate keywords whose names begin and end with @samp{__}. Pedantic
1751 warnings are also disabled in the expression that follows
1752 @code{__extension__}. However, only system header files should use
1753 these escape routes; application programs should avoid them.
1754 @xref{Alternate Keywords}.
1756 Some users try to use @samp{-pedantic} to check programs for strict ISO
1757 C conformance. They soon find that it does not do quite what they want:
1758 it finds some non-ISO practices, but not all---only those for which
1759 ISO C @emph{requires} a diagnostic, and some others for which
1760 diagnostics have been added.
1762 A feature to report any failure to conform to ISO C might be useful in
1763 some instances, but would require considerable additional work and would
1764 be quite different from @samp{-pedantic}. We don't have plans to
1765 support such a feature in the near future.
1767 Where the standard specified with @option{-std} represents a GNU
1768 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
1769 corresponding @dfn{base standard}, the version of ISO C on which the GNU
1770 extended dialect is based. Warnings from @option{-pedantic} are given
1771 where they are required by the base standard. (It would not make sense
1772 for such warnings to be given only for features not in the specified GNU
1773 C dialect, since by definition the GNU dialects of C include all
1774 features the compiler supports with the given option, and there would be
1775 nothing to warn about.)
1777 @item -pedantic-errors
1778 @opindex pedantic-errors
1779 Like @samp{-pedantic}, except that errors are produced rather than
1784 Inhibit all warning messages.
1788 Inhibit warning messages about the use of @samp{#import}.
1790 @item -Wchar-subscripts
1791 @opindex Wchar-subscripts
1792 Warn if an array subscript has type @code{char}. This is a common cause
1793 of error, as programmers often forget that this type is signed on some
1798 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
1799 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
1803 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
1804 the arguments supplied have types appropriate to the format string
1805 specified, and that the conversions specified in the format string make
1806 sense. This includes standard functions, and others specified by format
1807 attributes (@pxref{Function Attributes}), in the @code{printf},
1808 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
1809 not in the C standard) families.
1811 The formats are checked against the format features supported by GNU
1812 libc version 2.2. These include all ISO C89 and C99 features, as well
1813 as features from the Single Unix Specification and some BSD and GNU
1814 extensions. Other library implementations may not support all these
1815 features; GCC does not support warning about features that go beyond a
1816 particular library's limitations. However, if @samp{-pedantic} is used
1817 with @samp{-Wformat}, warnings will be given about format features not
1818 in the selected standard version (but not for @code{strfmon} formats,
1819 since those are not in any version of the C standard). @xref{C Dialect
1820 Options,,Options Controlling C Dialect}.
1822 @samp{-Wformat} is included in @samp{-Wall}. For more control over some
1823 aspects of format checking, the options @samp{-Wno-format-y2k},
1824 @samp{-Wno-format-extra-args}, @samp{-Wformat-nonliteral},
1825 @samp{-Wformat-security} and @samp{-Wformat=2} are available, but are
1826 not included in @samp{-Wall}.
1828 @item -Wno-format-y2k
1829 @opindex Wno-format-y2k
1830 If @samp{-Wformat} is specified, do not warn about @code{strftime}
1831 formats which may yield only a two-digit year.
1833 @item -Wno-format-extra-args
1834 @opindex Wno-format-extra-args
1835 If @samp{-Wformat} is specified, do not warn about excess arguments to a
1836 @code{printf} or @code{scanf} format function. The C standard specifies
1837 that such arguments are ignored.
1839 @item -Wformat-nonliteral
1840 @opindex Wformat-nonliteral
1841 If @samp{-Wformat} is specified, also warn if the format string is not a
1842 string literal and so cannot be checked, unless the format function
1843 takes its format arguments as a @code{va_list}.
1845 @item -Wformat-security
1846 @opindex Wformat-security
1847 If @samp{-Wformat} is specified, also warn about uses of format
1848 functions that represent possible security problems. At present, this
1849 warns about calls to @code{printf} and @code{scanf} functions where the
1850 format string is not a string literal and there are no format arguments,
1851 as in @code{printf (foo);}. This may be a security hole if the format
1852 string came from untrusted input and contains @samp{%n}. (This is
1853 currently a subset of what @samp{-Wformat-nonliteral} warns about, but
1854 in future warnings may be added to @samp{-Wformat-security} that are not
1855 included in @samp{-Wformat-nonliteral}.)
1859 Enable @samp{-Wformat} plus format checks not included in
1860 @samp{-Wformat}. Currently equivalent to @samp{-Wformat
1861 -Wformat-nonliteral -Wformat-security}.
1863 @item -Wimplicit-int
1864 @opindex Wimplicit-int
1865 Warn when a declaration does not specify a type.
1867 @item -Wimplicit-function-declaration
1868 @itemx -Werror-implicit-function-declaration
1869 @opindex Wimplicit-function-declaration
1870 @opindex Werror-implicit-function-declaration
1871 Give a warning (or error) whenever a function is used before being
1876 Same as @samp{-Wimplicit-int} and @samp{-Wimplicit-function-}@*
1881 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
1882 function with external linkage, returning int, taking either zero
1883 arguments, two, or three arguments of appropriate types.
1885 @item -Wmissing-braces
1886 @opindex Wmissing-braces
1887 Warn if an aggregate or union initializer is not fully bracketed. In
1888 the following example, the initializer for @samp{a} is not fully
1889 bracketed, but that for @samp{b} is fully bracketed.
1892 int a[2][2] = @{ 0, 1, 2, 3 @};
1893 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
1898 Warn if a multicharacter constant (@samp{'FOOF'}) is used. Usually they
1899 indicate a typo in the user's code, as they have implementation-defined
1900 values, and should not be used in portable code.
1903 @opindex Wparentheses
1904 Warn if parentheses are omitted in certain contexts, such
1905 as when there is an assignment in a context where a truth value
1906 is expected, or when operators are nested whose precedence people
1907 often get confused about.
1909 Also warn about constructions where there may be confusion to which
1910 @code{if} statement an @code{else} branch belongs. Here is an example of
1923 In C, every @code{else} branch belongs to the innermost possible @code{if}
1924 statement, which in this example is @code{if (b)}. This is often not
1925 what the programmer expected, as illustrated in the above example by
1926 indentation the programmer chose. When there is the potential for this
1927 confusion, GNU C will issue a warning when this flag is specified.
1928 To eliminate the warning, add explicit braces around the innermost
1929 @code{if} statement so there is no way the @code{else} could belong to
1930 the enclosing @code{if}. The resulting code would look like this:
1944 @item -Wsequence-point
1945 @opindex Wsequence-point
1946 Warn about code that may have undefined semantics because of violations
1947 of sequence point rules in the C standard.
1949 The C standard defines the order in which expressions in a C program are
1950 evaluated in terms of @dfn{sequence points}, which represent a partial
1951 ordering between the execution of parts of the program: those executed
1952 before the sequence point, and those executed after it. These occur
1953 after the evaluation of a full expression (one which is not part of a
1954 larger expression), after the evaluation of the first operand of a
1955 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
1956 function is called (but after the evaluation of its arguments and the
1957 expression denoting the called function), and in certain other places.
1958 Other than as expressed by the sequence point rules, the order of
1959 evaluation of subexpressions of an expression is not specified. All
1960 these rules describe only a partial order rather than a total order,
1961 since, for example, if two functions are called within one expression
1962 with no sequence point between them, the order in which the functions
1963 are called is not specified. However, the standards committee have
1964 ruled that function calls do not overlap.
1966 It is not specified when between sequence points modifications to the
1967 values of objects take effect. Programs whose behavior depends on this
1968 have undefined behavior; the C standard specifies that ``Between the
1969 previous and next sequence point an object shall have its stored value
1970 modified at most once by the evaluation of an expression. Furthermore,
1971 the prior value shall be read only to determine the value to be
1972 stored.''. If a program breaks these rules, the results on any
1973 particular implementation are entirely unpredictable.
1975 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
1976 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
1977 diagnosed by this option, and it may give an occasional false positive
1978 result, but in general it has been found fairly effective at detecting
1979 this sort of problem in programs.
1981 The present implementation of this option only works for C programs. A
1982 future implementation may also work for C++ programs.
1984 There is some controversy over the precise meaning of the sequence point
1985 rules in subtle cases. Links to papers with alternative formal definitions
1986 and other related discussions may be found on our readings page
1987 @w{@uref{http://gcc.gnu.org/readings.html}}.
1990 @opindex Wreturn-type
1991 Warn whenever a function is defined with a return-type that defaults to
1992 @code{int}. Also warn about any @code{return} statement with no
1993 return-value in a function whose return-type is not @code{void}.
1995 For C++, a function without return type always produces a diagnostic
1996 message, even when @samp{-Wno-return-type} is specified. The only
1997 exceptions are @samp{main} and functions defined in system headers.
2001 Warn whenever a @code{switch} statement has an index of enumeral type
2002 and lacks a @code{case} for one or more of the named codes of that
2003 enumeration. (The presence of a @code{default} label prevents this
2004 warning.) @code{case} labels outside the enumeration range also
2005 provoke warnings when this option is used.
2009 Warn if any trigraphs are encountered that might change the meaning of
2010 the program (trigraphs within comments are not warned about).
2012 @item -Wunused-function
2013 @opindex Wunused-function
2014 Warn whenever a static function is declared but not defined or a
2015 non\-inline static function is unused.
2017 @item -Wunused-label
2018 @opindex Wunused-label
2019 Warn whenever a label is declared but not used.
2021 To suppress this warning use the @samp{unused} attribute
2022 (@pxref{Variable Attributes}).
2024 @item -Wunused-parameter
2025 @opindex Wunused-parameter
2026 Warn whenever a function parameter is unused aside from its declaration.
2028 To suppress this warning use the @samp{unused} attribute
2029 (@pxref{Variable Attributes}).
2031 @item -Wunused-variable
2032 @opindex Wunused-variable
2033 Warn whenever a local variable or non-constant static variable is unused
2034 aside from its declaration
2036 To suppress this warning use the @samp{unused} attribute
2037 (@pxref{Variable Attributes}).
2039 @item -Wunused-value
2040 @opindex Wunused-value
2041 Warn whenever a statement computes a result that is explicitly not used.
2043 To suppress this warning cast the expression to @samp{void}.
2047 All all the above @samp{-Wunused} options combined.
2049 In order to get a warning about an unused function parameter, you must
2050 either specify @samp{-W -Wunused} or separately specify
2051 @samp{-Wunused-parameter}.
2053 @item -Wuninitialized
2054 @opindex Wuninitialized
2055 Warn if an automatic variable is used without first being initialized or
2056 if a variable may be clobbered by a @code{setjmp} call.
2058 These warnings are possible only in optimizing compilation,
2059 because they require data flow information that is computed only
2060 when optimizing. If you don't specify @samp{-O}, you simply won't
2063 These warnings occur only for variables that are candidates for
2064 register allocation. Therefore, they do not occur for a variable that
2065 is declared @code{volatile}, or whose address is taken, or whose size
2066 is other than 1, 2, 4 or 8 bytes. Also, they do not occur for
2067 structures, unions or arrays, even when they are in registers.
2069 Note that there may be no warning about a variable that is used only
2070 to compute a value that itself is never used, because such
2071 computations may be deleted by data flow analysis before the warnings
2074 These warnings are made optional because GCC is not smart
2075 enough to see all the reasons why the code might be correct
2076 despite appearing to have an error. Here is one example of how
2095 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
2096 always initialized, but GCC doesn't know this. Here is
2097 another common case:
2102 if (change_y) save_y = y, y = new_y;
2104 if (change_y) y = save_y;
2109 This has no bug because @code{save_y} is used only if it is set.
2111 @cindex @code{longjmp} warnings
2112 This option also warns when a non-volatile automatic variable might be
2113 changed by a call to @code{longjmp}. These warnings as well are possible
2114 only in optimizing compilation.
2116 The compiler sees only the calls to @code{setjmp}. It cannot know
2117 where @code{longjmp} will be called; in fact, a signal handler could
2118 call it at any point in the code. As a result, you may get a warning
2119 even when there is in fact no problem because @code{longjmp} cannot
2120 in fact be called at the place which would cause a problem.
2122 Some spurious warnings can be avoided if you declare all the functions
2123 you use that never return as @code{noreturn}. @xref{Function
2126 @item -Wreorder (C++ only)
2128 @cindex reordering, warning
2129 @cindex warning for reordering of member initializers
2130 Warn when the order of member initializers given in the code does not
2131 match the order in which they must be executed. For instance:
2133 @item -Wunknown-pragmas
2134 @opindex Wunknown-pragmas
2135 @cindex warning for unknown pragmas
2136 @cindex unknown pragmas, warning
2137 @cindex pragmas, warning of unknown
2138 Warn when a #pragma directive is encountered which is not understood by
2139 GCC. If this command line option is used, warnings will even be issued
2140 for unknown pragmas in system header files. This is not the case if
2141 the warnings were only enabled by the @samp{-Wall} command line option.
2145 All of the above @samp{-W} options combined. This enables all the
2146 warnings about constructions that some users consider questionable, and
2147 that are easy to avoid (or modify to prevent the warning), even in
2148 conjunction with macros.
2150 @item -Wsystem-headers
2151 @opindex Wsystem-headers
2152 @cindex warnings from system headers
2153 @cindex system headers, warnings from
2154 Print warning messages for constructs found in system header files.
2155 Warnings from system headers are normally suppressed, on the assumption
2156 that they usually do not indicate real problems and would only make the
2157 compiler output harder to read. Using this command line option tells
2158 GCC to emit warnings from system headers as if they occurred in user
2159 code. However, note that using @samp{-Wall} in conjunction with this
2160 option will @emph{not} warn about unknown pragmas in system
2161 headers---for that, @samp{-Wunknown-pragmas} must also be used.
2164 The following @samp{-W@dots{}} options are not implied by @samp{-Wall}.
2165 Some of them warn about constructions that users generally do not
2166 consider questionable, but which occasionally you might wish to check
2167 for; others warn about constructions that are necessary or hard to avoid
2168 in some cases, and there is no simple way to modify the code to suppress
2174 Print extra warning messages for these events:
2178 A function can return either with or without a value. (Falling
2179 off the end of the function body is considered returning without
2180 a value.) For example, this function would evoke such a
2194 An expression-statement or the left-hand side of a comma expression
2195 contains no side effects.
2196 To suppress the warning, cast the unused expression to void.
2197 For example, an expression such as @samp{x[i,j]} will cause a warning,
2198 but @samp{x[(void)i,j]} will not.
2201 An unsigned value is compared against zero with @samp{<} or @samp{<=}.
2204 A comparison like @samp{x<=y<=z} appears; this is equivalent to
2205 @samp{(x<=y ? 1 : 0) <= z}, which is a different interpretation from
2206 that of ordinary mathematical notation.
2209 Storage-class specifiers like @code{static} are not the first things in
2210 a declaration. According to the C Standard, this usage is obsolescent.
2213 The return type of a function has a type qualifier such as @code{const}.
2214 Such a type qualifier has no effect, since the value returned by a
2215 function is not an lvalue. (But don't warn about the GNU extension of
2216 @code{volatile void} return types. That extension will be warned about
2217 if @samp{-pedantic} is specified.)
2220 If @samp{-Wall} or @samp{-Wunused} is also specified, warn about unused
2224 A comparison between signed and unsigned values could produce an
2225 incorrect result when the signed value is converted to unsigned.
2226 (But don't warn if @samp{-Wno-sign-compare} is also specified.)
2229 An aggregate has a partly bracketed initializer.
2230 For example, the following code would evoke such a warning,
2231 because braces are missing around the initializer for @code{x.h}:
2234 struct s @{ int f, g; @};
2235 struct t @{ struct s h; int i; @};
2236 struct t x = @{ 1, 2, 3 @};
2240 An aggregate has an initializer which does not initialize all members.
2241 For example, the following code would cause such a warning, because
2242 @code{x.h} would be implicitly initialized to zero:
2245 struct s @{ int f, g, h; @};
2246 struct s x = @{ 3, 4 @};
2251 @opindex Wfloat-equal
2252 Warn if floating point values are used in equality comparisons.
2254 The idea behind this is that sometimes it is convenient (for the
2255 programmer) to consider floating-point values as approximations to
2256 infinitely precise real numbers. If you are doing this, then you need
2257 to compute (by analysing the code, or in some other way) the maximum or
2258 likely maximum error that the computation introduces, and allow for it
2259 when performing comparisons (and when producing output, but that's a
2260 different problem). In particular, instead of testing for equality, you
2261 would check to see whether the two values have ranges that overlap; and
2262 this is done with the relational operators, so equality comparisons are
2265 @item -Wtraditional (C only)
2266 @opindex Wtraditional
2267 Warn about certain constructs that behave differently in traditional and
2268 ISO C. Also warn about ISO C constructs that have no traditional C
2269 equivalent, and/or problematic constructs which should be avoided.
2273 Macro parameters that appear within string literals in the macro body.
2274 In traditional C macro replacement takes place within string literals,
2275 but does not in ISO C.
2278 In traditional C, some preprocessor directives did not exist.
2279 Traditional preprocessors would only consider a line to be a directive
2280 if the @samp{#} appeared in column 1 on the line. Therefore
2281 @samp{-Wtraditional} warns about directives that traditional C
2282 understands but would ignore because the @samp{#} does not appear as the
2283 first character on the line. It also suggests you hide directives like
2284 @samp{#pragma} not understood by traditional C by indenting them. Some
2285 traditional implementations would not recognise @samp{#elif}, so it
2286 suggests avoiding it altogether.
2289 A function-like macro that appears without arguments.
2292 The unary plus operator.
2295 The `U' integer constant suffix, or the `F' or `L' floating point
2296 constant suffixes. (Traditonal C does support the `L' suffix on integer
2297 constants.) Note, these suffixes appear in macros defined in the system
2298 headers of most modern systems, e.g. the _MIN/_MAX macros in limits.h.
2299 Use of these macros in user code might normally lead to spurious
2300 warnings, however gcc's integrated preprocessor has enough context to
2301 avoid warning in these cases.
2304 A function declared external in one block and then used after the end of
2308 A @code{switch} statement has an operand of type @code{long}.
2311 A non-@code{static} function declaration follows a @code{static} one.
2312 This construct is not accepted by some traditional C compilers.
2315 The ISO type of an integer constant has a different width or
2316 signedness from its traditional type. This warning is only issued if
2317 the base of the constant is ten. I.e. hexadecimal or octal values, which
2318 typically represent bit patterns, are not warned about.
2321 Usage of ISO string concatenation is detected.
2324 Initialization of automatic aggregates.
2327 Identifier conflicts with labels. Traditional C lacks a separate
2328 namespace for labels.
2331 Initialization of unions. If the initializer is zero, the warning is
2332 omitted. This is done under the assumption that the zero initializer in
2333 user code appears conditioned on e.g. @code{__STDC__} to avoid missing
2334 initializer warnings and relies on default initialization to zero in the
2338 Conversions by prototypes between fixed/floating point values and vice
2339 versa. The absence of these prototypes when compiling with traditional
2340 C would cause serious problems. This is a subset of the possible
2341 conversion warnings, for the full set use @samp{-Wconversion}.
2346 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
2350 Warn whenever a local variable shadows another local variable, parameter or
2351 global variable or whenever a built-in function is shadowed.
2353 @item -Wlarger-than-@var{len}
2354 @opindex Wlarger-than
2355 Warn whenever an object of larger than @var{len} bytes is defined.
2357 @item -Wpointer-arith
2358 @opindex Wpointer-arith
2359 Warn about anything that depends on the ``size of'' a function type or
2360 of @code{void}. GNU C assigns these types a size of 1, for
2361 convenience in calculations with @code{void *} pointers and pointers
2364 @item -Wbad-function-cast (C only)
2365 @opindex Wbad-function-cast
2366 Warn whenever a function call is cast to a non-matching type.
2367 For example, warn if @code{int malloc()} is cast to @code{anything *}.
2371 Warn whenever a pointer is cast so as to remove a type qualifier from
2372 the target type. For example, warn if a @code{const char *} is cast
2373 to an ordinary @code{char *}.
2376 @opindex Wcast-align
2377 Warn whenever a pointer is cast such that the required alignment of the
2378 target is increased. For example, warn if a @code{char *} is cast to
2379 an @code{int *} on machines where integers can only be accessed at
2380 two- or four-byte boundaries.
2382 @item -Wwrite-strings
2383 @opindex Wwrite-strings
2384 Give string constants the type @code{const char[@var{length}]} so that
2385 copying the address of one into a non-@code{const} @code{char *}
2386 pointer will get a warning. These warnings will help you find at
2387 compile time code that can try to write into a string constant, but
2388 only if you have been very careful about using @code{const} in
2389 declarations and prototypes. Otherwise, it will just be a nuisance;
2390 this is why we did not make @samp{-Wall} request these warnings.
2393 @opindex Wconversion
2394 Warn if a prototype causes a type conversion that is different from what
2395 would happen to the same argument in the absence of a prototype. This
2396 includes conversions of fixed point to floating and vice versa, and
2397 conversions changing the width or signedness of a fixed point argument
2398 except when the same as the default promotion.
2400 Also, warn if a negative integer constant expression is implicitly
2401 converted to an unsigned type. For example, warn about the assignment
2402 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
2403 casts like @code{(unsigned) -1}.
2405 @item -Wsign-compare
2406 @opindex Wsign-compare
2407 @cindex warning for comparison of signed and unsigned values
2408 @cindex comparison of signed and unsigned values, warning
2409 @cindex signed and unsigned values, comparison warning
2410 Warn when a comparison between signed and unsigned values could produce
2411 an incorrect result when the signed value is converted to unsigned.
2412 This warning is also enabled by @samp{-W}; to get the other warnings
2413 of @samp{-W} without this warning, use @samp{-W -Wno-sign-compare}.
2415 @item -Waggregate-return
2416 @opindex Waggregate-return
2417 Warn if any functions that return structures or unions are defined or
2418 called. (In languages where you can return an array, this also elicits
2421 @item -Wstrict-prototypes (C only)
2422 @opindex Wstrict-prototypes
2423 Warn if a function is declared or defined without specifying the
2424 argument types. (An old-style function definition is permitted without
2425 a warning if preceded by a declaration which specifies the argument
2428 @item -Wmissing-prototypes (C only)
2429 @opindex Wmissing-prototypes
2430 Warn if a global function is defined without a previous prototype
2431 declaration. This warning is issued even if the definition itself
2432 provides a prototype. The aim is to detect global functions that fail
2433 to be declared in header files.
2435 @item -Wmissing-declarations
2436 @opindex Wmissing-declarations
2437 Warn if a global function is defined without a previous declaration.
2438 Do so even if the definition itself provides a prototype.
2439 Use this option to detect global functions that are not declared in
2442 @item -Wmissing-noreturn
2443 @opindex Wmissing-noreturn
2444 Warn about functions which might be candidates for attribute @code{noreturn}.
2445 Note these are only possible candidates, not absolute ones. Care should
2446 be taken to manually verify functions actually do not ever return before
2447 adding the @code{noreturn} attribute, otherwise subtle code generation
2448 bugs could be introduced. You will not get a warning for @code{main} in
2449 hosted C environments.
2451 @item -Wmissing-format-attribute
2452 @opindex Wmissing-format-attribute
2454 If @samp{-Wformat} is enabled, also warn about functions which might be
2455 candidates for @code{format} attributes. Note these are only possible
2456 candidates, not absolute ones. GCC will guess that @code{format}
2457 attributes might be appropriate for any function that calls a function
2458 like @code{vprintf} or @code{vscanf}, but this might not always be the
2459 case, and some functions for which @code{format} attributes are
2460 appropriate may not be detected. This option has no effect unless
2461 @samp{-Wformat} is enabled (possibly by @samp{-Wall}).
2465 Warn if a structure is given the packed attribute, but the packed
2466 attribute has no effect on the layout or size of the structure.
2467 Such structures may be mis-aligned for little benefit. For
2468 instance, in this code, the variable @code{f.x} in @code{struct bar}
2469 will be misaligned even though @code{struct bar} does not itself
2470 have the packed attribute:
2477 @} __attribute__((packed));
2487 Warn if padding is included in a structure, either to align an element
2488 of the structure or to align the whole structure. Sometimes when this
2489 happens it is possible to rearrange the fields of the structure to
2490 reduce the padding and so make the structure smaller.
2492 @item -Wredundant-decls
2493 @opindex Wredundant-decls
2494 Warn if anything is declared more than once in the same scope, even in
2495 cases where multiple declaration is valid and changes nothing.
2497 @item -Wnested-externs (C only)
2498 @opindex Wnested-externs
2499 Warn if an @code{extern} declaration is encountered within a function.
2501 @item -Wunreachable-code
2502 @opindex Wunreachable-code
2503 Warn if the compiler detects that code will never be executed.
2505 This option is intended to warn when the compiler detects that at
2506 least a whole line of source code will never be executed, because
2507 some condition is never satisfied or because it is after a
2508 procedure that never returns.
2510 It is possible for this option to produce a warning even though there
2511 are circumstances under which part of the affected line can be executed,
2512 so care should be taken when removing apparently-unreachable code.
2514 For instance, when a function is inlined, a warning may mean that the
2515 line is unreachable in only one inlined copy of the function.
2517 This option is not made part of @samp{-Wall} because in a debugging
2518 version of a program there is often substantial code which checks
2519 correct functioning of the program and is, hopefully, unreachable
2520 because the program does work. Another common use of unreachable
2521 code is to provide behaviour which is selectable at compile-time.
2525 Warn if a function can not be inlined and it was declared as inline.
2529 @opindex Wno-long-long
2530 Warn if @samp{long long} type is used. This is default. To inhibit
2531 the warning messages, use @samp{-Wno-long-long}. Flags
2532 @samp{-Wlong-long} and @samp{-Wno-long-long} are taken into account
2533 only when @samp{-pedantic} flag is used.
2535 @item -Wdisabled-optimization
2536 @opindex Wdisabled-optimization
2537 Warn if a requested optimization pass is disabled. This warning does
2538 not generally indicate that there is anything wrong with your code; it
2539 merely indicates that GCC's optimizers were unable to handle the code
2540 effectively. Often, the problem is that your code is too big or too
2541 complex; GCC will refuse to optimize programs when the optimization
2542 itself is likely to take inordinate amounts of time.
2546 Make all warnings into errors.
2549 @node Debugging Options
2550 @section Options for Debugging Your Program or GCC
2551 @cindex options, debugging
2552 @cindex debugging information options
2554 GCC has various special options that are used for debugging
2555 either your program or GCC:
2560 Produce debugging information in the operating system's native format
2561 (stabs, COFF, XCOFF, or DWARF). GDB can work with this debugging
2564 On most systems that use stabs format, @samp{-g} enables use of extra
2565 debugging information that only GDB can use; this extra information
2566 makes debugging work better in GDB but will probably make other debuggers
2568 refuse to read the program. If you want to control for certain whether
2569 to generate the extra information, use @samp{-gstabs+}, @samp{-gstabs},
2570 @samp{-gxcoff+}, @samp{-gxcoff}, @samp{-gdwarf-1+}, or @samp{-gdwarf-1}
2573 Unlike most other C compilers, GCC allows you to use @samp{-g} with
2574 @samp{-O}. The shortcuts taken by optimized code may occasionally
2575 produce surprising results: some variables you declared may not exist
2576 at all; flow of control may briefly move where you did not expect it;
2577 some statements may not be executed because they compute constant
2578 results or their values were already at hand; some statements may
2579 execute in different places because they were moved out of loops.
2581 Nevertheless it proves possible to debug optimized output. This makes
2582 it reasonable to use the optimizer for programs that might have bugs.
2584 The following options are useful when GCC is generated with the
2585 capability for more than one debugging format.
2589 Produce debugging information for use by GDB. This means to use the
2590 most expressive format available (DWARF 2, stabs, or the native format
2591 if neither of those are supported), including GDB extensions if at all
2596 Produce debugging information in stabs format (if that is supported),
2597 without GDB extensions. This is the format used by DBX on most BSD
2598 systems. On MIPS, Alpha and System V Release 4 systems this option
2599 produces stabs debugging output which is not understood by DBX or SDB.
2600 On System V Release 4 systems this option requires the GNU assembler.
2604 Produce debugging information in stabs format (if that is supported),
2605 using GNU extensions understood only by the GNU debugger (GDB). The
2606 use of these extensions is likely to make other debuggers crash or
2607 refuse to read the program.
2611 Produce debugging information in COFF format (if that is supported).
2612 This is the format used by SDB on most System V systems prior to
2617 Produce debugging information in XCOFF format (if that is supported).
2618 This is the format used by the DBX debugger on IBM RS/6000 systems.
2622 Produce debugging information in XCOFF format (if that is supported),
2623 using GNU extensions understood only by the GNU debugger (GDB). The
2624 use of these extensions is likely to make other debuggers crash or
2625 refuse to read the program, and may cause assemblers other than the GNU
2626 assembler (GAS) to fail with an error.
2630 Produce debugging information in DWARF version 1 format (if that is
2631 supported). This is the format used by SDB on most System V Release 4
2636 Produce debugging information in DWARF version 1 format (if that is
2637 supported), using GNU extensions understood only by the GNU debugger
2638 (GDB). The use of these extensions is likely to make other debuggers
2639 crash or refuse to read the program.
2643 Produce debugging information in DWARF version 2 format (if that is
2644 supported). This is the format used by DBX on IRIX 6.
2647 @itemx -ggdb@var{level}
2648 @itemx -gstabs@var{level}
2649 @itemx -gcoff@var{level}
2650 @itemx -gxcoff@var{level}
2651 @itemx -gdwarf@var{level}
2652 @itemx -gdwarf-2@var{level}
2653 Request debugging information and also use @var{level} to specify how
2654 much information. The default level is 2.
2656 Level 1 produces minimal information, enough for making backtraces in
2657 parts of the program that you don't plan to debug. This includes
2658 descriptions of functions and external variables, but no information
2659 about local variables and no line numbers.
2661 Level 3 includes extra information, such as all the macro definitions
2662 present in the program. Some debuggers support macro expansion when
2668 Generate extra code to write profile information suitable for the
2669 analysis program @code{prof}. You must use this option when compiling
2670 the source files you want data about, and you must also use it when
2673 @cindex @code{gprof}
2676 Generate extra code to write profile information suitable for the
2677 analysis program @code{gprof}. You must use this option when compiling
2678 the source files you want data about, and you must also use it when
2684 Generate extra code to write profile information for basic blocks, which will
2685 record the number of times each basic block is executed, the basic block start
2686 address, and the function name containing the basic block. If @samp{-g} is
2687 used, the line number and filename of the start of the basic block will also be
2688 recorded. If not overridden by the machine description, the default action is
2689 to append to the text file @file{bb.out}.
2691 This data could be analyzed by a program like @code{tcov}. Note,
2692 however, that the format of the data is not what @code{tcov} expects.
2693 Eventually GNU @code{gprof} should be extended to process this data.
2697 Makes the compiler print out each function name as it is compiled, and
2698 print some statistics about each pass when it finishes.
2701 @opindex ftime-report
2702 Makes the compiler print some statistics about the time consumed by each
2703 pass when it finishes.
2706 @opindex fmem-report
2707 Makes the compiler print some statistics about permanent memory
2708 allocation when it finishes.
2712 Generate extra code to profile basic blocks. Your executable will
2713 produce output that is a superset of that produced when @samp{-a} is
2714 used. Additional output is the source and target address of the basic
2715 blocks where a jump takes place, the number of times a jump is executed,
2716 and (optionally) the complete sequence of basic blocks being executed.
2717 The output is appended to file @file{bb.out}.
2719 You can examine different profiling aspects without recompilation. Your
2720 executable will read a list of function names from file @file{bb.in}.
2721 Profiling starts when a function on the list is entered and stops when
2722 that invocation is exited. To exclude a function from profiling, prefix
2723 its name with `-'. If a function name is not unique, you can
2724 disambiguate it by writing it in the form
2725 @samp{/path/filename.d:functionname}. Your executable will write the
2726 available paths and filenames in file @file{bb.out}.
2728 Several function names have a special meaning:
2731 Write source, target and frequency of jumps to file @file{bb.out}.
2732 @item __bb_hidecall__
2733 Exclude function calls from frequency count.
2734 @item __bb_showret__
2735 Include function returns in frequency count.
2737 Write the sequence of basic blocks executed to file @file{bbtrace.gz}.
2738 The file will be compressed using the program @samp{gzip}, which must
2739 exist in your @env{PATH}. On systems without the @samp{popen}
2740 function, the file will be named @file{bbtrace} and will not be
2741 compressed. @strong{Profiling for even a few seconds on these systems
2742 will produce a very large file.} Note: @code{__bb_hidecall__} and
2743 @code{__bb_showret__} will not affect the sequence written to
2747 Here's a short example using different profiling parameters
2748 in file @file{bb.in}. Assume function @code{foo} consists of basic blocks
2749 1 and 2 and is called twice from block 3 of function @code{main}. After
2750 the calls, block 3 transfers control to block 4 of @code{main}.
2752 With @code{__bb_trace__} and @code{main} contained in file @file{bb.in},
2753 the following sequence of blocks is written to file @file{bbtrace.gz}:
2754 0 3 1 2 1 2 4. The return from block 2 to block 3 is not shown, because
2755 the return is to a point inside the block and not to the top. The
2756 block address 0 always indicates, that control is transferred
2757 to the trace from somewhere outside the observed functions. With
2758 @samp{-foo} added to @file{bb.in}, the blocks of function
2759 @code{foo} are removed from the trace, so only 0 3 4 remains.
2761 With @code{__bb_jumps__} and @code{main} contained in file @file{bb.in},
2762 jump frequencies will be written to file @file{bb.out}. The
2763 frequencies are obtained by constructing a trace of blocks
2764 and incrementing a counter for every neighbouring pair of blocks
2765 in the trace. The trace 0 3 1 2 1 2 4 displays the following
2769 Jump from block 0x0 to block 0x3 executed 1 time(s)
2770 Jump from block 0x3 to block 0x1 executed 1 time(s)
2771 Jump from block 0x1 to block 0x2 executed 2 time(s)
2772 Jump from block 0x2 to block 0x1 executed 1 time(s)
2773 Jump from block 0x2 to block 0x4 executed 1 time(s)
2776 With @code{__bb_hidecall__}, control transfer due to call instructions
2777 is removed from the trace, that is the trace is cut into three parts: 0
2778 3 4, 0 1 2 and 0 1 2. With @code{__bb_showret__}, control transfer due
2779 to return instructions is added to the trace. The trace becomes: 0 3 1
2780 2 3 1 2 3 4. Note, that this trace is not the same, as the sequence
2781 written to @file{bbtrace.gz}. It is solely used for counting jump
2784 @item -fprofile-arcs
2785 @opindex fprofile-arcs
2786 Instrument @dfn{arcs} during compilation. For each function of your
2787 program, GCC creates a program flow graph, then finds a spanning tree
2788 for the graph. Only arcs that are not on the spanning tree have to be
2789 instrumented: the compiler adds code to count the number of times that these
2790 arcs are executed. When an arc is the only exit or only entrance to a
2791 block, the instrumentation code can be added to the block; otherwise, a
2792 new basic block must be created to hold the instrumentation code.
2794 Since not every arc in the program must be instrumented, programs
2795 compiled with this option run faster than programs compiled with
2796 @samp{-a}, which adds instrumentation code to every basic block in the
2797 program. The tradeoff: since @code{gcov} does not have
2798 execution counts for all branches, it must start with the execution
2799 counts for the instrumented branches, and then iterate over the program
2800 flow graph until the entire graph has been solved. Hence, @code{gcov}
2801 runs a little more slowly than a program which uses information from
2804 @samp{-fprofile-arcs} also makes it possible to estimate branch
2805 probabilities, and to calculate basic block execution counts. In
2806 general, basic block execution counts do not give enough information to
2807 estimate all branch probabilities. When the compiled program exits, it
2808 saves the arc execution counts to a file called
2809 @file{@var{sourcename}.da}. Use the compiler option
2810 @samp{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
2811 Control Optimization}) when recompiling, to optimize using estimated
2812 branch probabilities.
2815 @item -ftest-coverage
2816 @opindex ftest-coverage
2817 Create data files for the @code{gcov} code-coverage utility
2818 (@pxref{Gcov,, @code{gcov}: a GCC Test Coverage Program}).
2819 The data file names begin with the name of your source file:
2822 @item @var{sourcename}.bb
2823 A mapping from basic blocks to line numbers, which @code{gcov} uses to
2824 associate basic block execution counts with line numbers.
2826 @item @var{sourcename}.bbg
2827 A list of all arcs in the program flow graph. This allows @code{gcov}
2828 to reconstruct the program flow graph, so that it can compute all basic
2829 block and arc execution counts from the information in the
2830 @code{@var{sourcename}.da} file (this last file is the output from
2831 @samp{-fprofile-arcs}).
2834 @item -d@var{letters}
2836 Says to make debugging dumps during compilation at times specified by
2837 @var{letters}. This is used for debugging the compiler. The file names
2838 for most of the dumps are made by appending a pass number and a word to
2839 the source file name (e.g. @file{foo.c.00.rtl} or @file{foo.c.01.sibling}).
2840 Here are the possible letters for use in @var{letters}, and their meanings:
2845 Annotate the assembler output with miscellaneous debugging information.
2848 Dump after computing branch probabilities, to @file{@var{file}.11.bp}.
2851 Dump after block reordering, to @file{@var{file}.26.bbro}.
2854 Dump after instruction combination, to the file @file{@var{file}.14.combine}.
2857 Dump after the first if conversion, to the file @file{@var{file}.15.ce}.
2860 Dump after delayed branch scheduling, to @file{@var{file}.29.dbr}.
2863 Dump all macro definitions, at the end of preprocessing, in addition to
2867 Dump after SSA optimizations, to @file{@var{file}.05.ssa} and
2868 @file{@var{file}.06.ussa}.
2871 Dump after the second if conversion, to @file{@var{file}.24.ce2}.
2874 Dump after life analysis, to @file{@var{file}.13.life}.
2877 Dump after purging @code{ADDRESSOF} codes, to @file{@var{file}.04.addressof}.
2880 Dump after global register allocation, to @file{@var{file}.19.greg}.
2883 Dump after post-reload CSE and other optimizations, to @file{@var{file}.20.postreload}.
2886 Dump after GCSE, to @file{@var{file}.08.gcse}.
2889 Dump after sibling call optimizations, to @file{@var{file}.01.sibling}.
2892 Dump after the first jump optimization, to @file{@var{file}.02.jump}.
2895 Dump after the last jump optimization, to @file{@var{file}.27.jump2}.
2898 Dump after conversion from registers to stack, to @file{@var{file}.29.stack}.
2901 Dump after local register allocation, to @file{@var{file}.18.lreg}.
2904 Dump after loop optimization, to @file{@var{file}.09.loop}.
2907 Dump after performing the machine dependent reorganisation pass, to
2908 @file{@var{file}.28.mach}.
2911 Dump after register renumbering, to @file{@var{file}.23.rnreg}.
2914 Dump after the register move pass, to @file{@var{file}.16.regmove}.
2917 Dump after RTL generation, to @file{@var{file}.00.rtl}.
2920 Dump after the second instruction scheduling pass, to
2921 @file{@var{file}.25.sched2}.
2924 Dump after CSE (including the jump optimization that sometimes follows
2925 CSE), to @file{@var{file}.03.cse}.
2928 Dump after the first instruction scheduling pass, to
2929 @file{@var{file}.17.sched}.
2932 Dump after the second CSE pass (including the jump optimization that
2933 sometimes follows CSE), to @file{@var{file}.10.cse2}.
2936 Dump after the second flow pass, to @file{@var{file}.21.flow2}.
2939 Dump after dead code elimination, to @file{@var{file}.06.dce}.
2942 Dump after the peephole pass, to @file{@var{file}.22.peephole2}.
2945 Produce all the dumps listed above.
2948 Print statistics on memory usage, at the end of the run, to
2952 Annotate the assembler output with a comment indicating which
2953 pattern and alternative was used. The length of each instruction is
2957 Dump the RTL in the assembler output as a comment before each instruction.
2958 Also turns on @samp{-dp} annotation.
2961 For each of the other indicated dump files (except for
2962 @file{@var{file}.00.rtl}), dump a representation of the control flow graph
2963 suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
2966 Just generate RTL for a function instead of compiling it. Usually used
2970 Dump debugging information during parsing, to standard error.
2973 @item -fdump-unnumbered
2974 @opindex fdump-unnumbered
2975 When doing debugging dumps (see -d option above), suppress instruction
2976 numbers and line number note output. This makes it more feasible to
2977 use diff on debugging dumps for compiler invocations with different
2978 options, in particular with and without -g.
2980 @item -fdump-translation-unit (C and C++ only)
2981 @item -fdump-translation-unit-@var{number} (C and C++ only)
2982 @opindex fdump-translation-unit
2983 Dump a representation of the tree structure for the entire translation
2984 unit to a file. The file name is made by appending @file{.tu} to the
2985 source file name. If the -@var{number} form is used, @var{number}
2986 controls the details of the dump as described for the -fdump-tree options.
2988 @item -fdump-class-hierarchy (C++ only)
2989 @item -fdump-class-hierarchy-@var{number} (C++ only)
2990 @opindex fdump-class-hierarchy
2991 Dump a representation of each class's hierarchy and virtual function
2992 table layout to a file. The file name is made by appending @file{.class}
2993 to the source file name. If the -@var{number} form is used, @var{number}
2994 controls the details of the dump as described for the -fdump-tree
2997 @item -fdump-ast-@var{switch} (C++ only)
2998 @item -fdump-ast-@var{switch}-@var{number} (C++ only)
3000 Control the dumping at various stages of processing the abstract syntax
3001 tree to a file. The file name is generated by appending a switch
3002 specific suffix to the source file name. If the -@var{number} form is
3003 used, @var{number} is a bit mask which controls the details of the
3004 dump. The following bits are meaningful (these are not set symbolically,
3005 as the primary function of these dumps is for debugging gcc itself):
3009 Print the address of each node. Usually this is not meaningful as it
3010 changes according to the environment and source file.
3012 Inhibit dumping of members of a scope or body of a function, unless they
3013 are reachable by some other path.
3016 The following tree dumps are possible:
3019 Dump before any tree based optimization, to @file{@var{file}.original}.
3021 Dump after all tree based optimization, to @file{@var{file}.optimized}.
3024 @item -fpretend-float
3025 @opindex fpretend-float
3026 When running a cross-compiler, pretend that the target machine uses the
3027 same floating point format as the host machine. This causes incorrect
3028 output of the actual floating constants, but the actual instruction
3029 sequence will probably be the same as GCC would make when running on
3034 Store the usual ``temporary'' intermediate files permanently; place them
3035 in the current directory and name them based on the source file. Thus,
3036 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
3037 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
3038 preprocessed @file{foo.i} output file even though the compiler now
3039 normally uses an integrated preprocessor.
3043 Report the CPU time taken by each subprocess in the compilation
3044 sequence. For C source files, this is the compiler proper and assembler
3045 (plus the linker if linking is done). The output looks like this:
3052 The first number on each line is the ``user time,'' that is time spent
3053 executing the program itself. The second number is ``system time,''
3054 time spent executing operating system routines on behalf of the program.
3055 Both numbers are in seconds.
3057 @item -print-file-name=@var{library}
3058 @opindex print-file-name
3059 Print the full absolute name of the library file @var{library} that
3060 would be used when linking---and don't do anything else. With this
3061 option, GCC does not compile or link anything; it just prints the
3064 @item -print-prog-name=@var{program}
3065 @opindex print-prog-name
3066 Like @samp{-print-file-name}, but searches for a program such as @samp{cpp}.
3068 @item -print-libgcc-file-name
3069 @opindex print-libgcc-file-name
3070 Same as @samp{-print-file-name=libgcc.a}.
3072 This is useful when you use @samp{-nostdlib} or @samp{-nodefaultlibs}
3073 but you do want to link with @file{libgcc.a}. You can do
3076 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
3079 @item -print-search-dirs
3080 @opindex print-search-dirs
3081 Print the name of the configured installation directory and a list of
3082 program and library directories gcc will search---and don't do anything else.
3084 This is useful when gcc prints the error message
3085 @samp{installation problem, cannot exec cpp0: No such file or directory}.
3086 To resolve this you either need to put @file{cpp0} and the other compiler
3087 components where gcc expects to find them, or you can set the environment
3088 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
3089 Don't forget the trailing '/'.
3090 @xref{Environment Variables}.
3093 @opindex dumpmachine
3094 Print the compiler's target machine (for example,
3095 @samp{i686-pc-linux-gnu})---and don't do anything else.
3098 @opindex dumpversion
3099 Print the compiler version (for example, @samp{3.0})---and don't do
3104 Print the compiler's built-in specs---and don't do anything else. (This
3105 is used when GCC itself is being built.) @xref{Spec Files}.
3108 @node Optimize Options
3109 @section Options That Control Optimization
3110 @cindex optimize options
3111 @cindex options, optimization
3113 These options control various sorts of optimizations:
3120 Optimize. Optimizing compilation takes somewhat more time, and a lot
3121 more memory for a large function.
3123 Without @samp{-O}, the compiler's goal is to reduce the cost of
3124 compilation and to make debugging produce the expected results.
3125 Statements are independent: if you stop the program with a breakpoint
3126 between statements, you can then assign a new value to any variable or
3127 change the program counter to any other statement in the function and
3128 get exactly the results you would expect from the source code.
3130 Without @samp{-O}, the compiler only allocates variables declared
3131 @code{register} in registers. The resulting compiled code is a little
3132 worse than produced by PCC without @samp{-O}.
3134 With @samp{-O}, the compiler tries to reduce code size and execution
3137 When you specify @samp{-O}, the compiler turns on @samp{-fthread-jumps}
3138 and @samp{-fdefer-pop} on all machines. The compiler turns on
3139 @samp{-fdelayed-branch} on machines that have delay slots, and
3140 @samp{-fomit-frame-pointer} on machines that can support debugging even
3141 without a frame pointer. On some machines the compiler also turns
3142 on other flags.@refill
3146 Optimize even more. GCC performs nearly all supported optimizations
3147 that do not involve a space-speed tradeoff. The compiler does not
3148 perform loop unrolling or function inlining when you specify @samp{-O2}.
3149 As compared to @samp{-O}, this option increases both compilation time
3150 and the performance of the generated code.
3152 @samp{-O2} turns on all optional optimizations except for loop unrolling,
3153 function inlining, and register renaming. It also turns on the
3154 @samp{-fforce-mem} option on all machines and frame pointer elimination
3155 on machines where doing so does not interfere with debugging.
3159 Optimize yet more. @samp{-O3} turns on all optimizations specified by
3160 @samp{-O2} and also turns on the @samp{-finline-functions} and
3161 @samp{-frename-registers} options.
3169 Optimize for size. @samp{-Os} enables all @samp{-O2} optimizations that
3170 do not typically increase code size. It also performs further
3171 optimizations designed to reduce code size.
3173 If you use multiple @samp{-O} options, with or without level numbers,
3174 the last such option is the one that is effective.
3177 Options of the form @samp{-f@var{flag}} specify machine-independent
3178 flags. Most flags have both positive and negative forms; the negative
3179 form of @samp{-ffoo} would be @samp{-fno-foo}. In the table below,
3180 only one of the forms is listed---the one which is not the default.
3181 You can figure out the other form by either removing @samp{no-} or
3186 @opindex ffloat-store
3187 Do not store floating point variables in registers, and inhibit other
3188 options that might change whether a floating point value is taken from a
3191 @cindex floating point precision
3192 This option prevents undesirable excess precision on machines such as
3193 the 68000 where the floating registers (of the 68881) keep more
3194 precision than a @code{double} is supposed to have. Similarly for the
3195 x86 architecture. For most programs, the excess precision does only
3196 good, but a few programs rely on the precise definition of IEEE floating
3197 point. Use @samp{-ffloat-store} for such programs, after modifying
3198 them to store all pertinent intermediate computations into variables.
3200 @item -fno-default-inline
3201 @opindex fno-default-inline
3202 Do not make member functions inline by default merely because they are
3203 defined inside the class scope (C++ only). Otherwise, when you specify
3204 @w{@samp{-O}}, member functions defined inside class scope are compiled
3205 inline by default; i.e., you don't need to add @samp{inline} in front of
3206 the member function name.
3208 @item -fno-defer-pop
3209 @opindex fno-defer-pop
3210 Always pop the arguments to each function call as soon as that function
3211 returns. For machines which must pop arguments after a function call,
3212 the compiler normally lets arguments accumulate on the stack for several
3213 function calls and pops them all at once.
3217 Force memory operands to be copied into registers before doing
3218 arithmetic on them. This produces better code by making all memory
3219 references potential common subexpressions. When they are not common
3220 subexpressions, instruction combination should eliminate the separate
3221 register-load. The @samp{-O2} option turns on this option.
3224 @opindex fforce-addr
3225 Force memory address constants to be copied into registers before
3226 doing arithmetic on them. This may produce better code just as
3227 @samp{-fforce-mem} may.
3229 @item -fomit-frame-pointer
3230 @opindex fomit-frame-pointer
3231 Don't keep the frame pointer in a register for functions that
3232 don't need one. This avoids the instructions to save, set up and
3233 restore frame pointers; it also makes an extra register available
3234 in many functions. @strong{It also makes debugging impossible on
3238 On some machines, such as the Vax, this flag has no effect, because
3239 the standard calling sequence automatically handles the frame pointer
3240 and nothing is saved by pretending it doesn't exist. The
3241 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
3242 whether a target machine supports this flag. @xref{Registers}.@refill
3245 On some machines, such as the Vax, this flag has no effect, because
3246 the standard calling sequence automatically handles the frame pointer
3247 and nothing is saved by pretending it doesn't exist. The
3248 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
3249 whether a target machine supports this flag. @xref{Registers,,Register
3250 Usage, gcc.info, Using and Porting GCC}.@refill
3253 @item -foptimize-sibling-calls
3254 @opindex foptimize-sibling-calls
3255 Optimize sibling and tail recursive calls.
3259 This option generates traps for signed overflow on addition, subtraction,
3260 multiplication operations.
3264 Don't pay attention to the @code{inline} keyword. Normally this option
3265 is used to keep the compiler from expanding any functions inline.
3266 Note that if you are not optimizing, no functions can be expanded inline.
3268 @item -finline-functions
3269 @opindex finline-functions
3270 Integrate all simple functions into their callers. The compiler
3271 heuristically decides which functions are simple enough to be worth
3272 integrating in this way.
3274 If all calls to a given function are integrated, and the function is
3275 declared @code{static}, then the function is normally not output as
3276 assembler code in its own right.
3278 @item -finline-limit=@var{n}
3279 @opindex finline-limit
3280 By default, gcc limits the size of functions that can be inlined. This flag
3281 allows the control of this limit for functions that are explicitly marked as
3282 inline (ie marked with the inline keyword or defined within the class
3283 definition in c++). @var{n} is the size of functions that can be inlined in
3284 number of pseudo instructions (not counting parameter handling). The default
3285 value of n is 10000. Increasing this value can result in more inlined code at
3286 the cost of compilation time and memory consumption. Decreasing usually makes
3287 the compilation faster and less code will be inlined (which presumably
3288 means slower programs). This option is particularly useful for programs that
3289 use inlining heavily such as those based on recursive templates with c++.
3291 @emph{Note:} pseudo instruction represents, in this particular context, an
3292 abstract measurement of function's size. In no way, it represents a count
3293 of assembly instructions and as such its exact meaning might change from one
3294 release to an another.
3296 @item -fkeep-inline-functions
3297 @opindex fkeep-inline-functions
3298 Even if all calls to a given function are integrated, and the function
3299 is declared @code{static}, nevertheless output a separate run-time
3300 callable version of the function. This switch does not affect
3301 @code{extern inline} functions.
3303 @item -fkeep-static-consts
3304 @opindex fkeep-static-consts
3305 Emit variables declared @code{static const} when optimization isn't turned
3306 on, even if the variables aren't referenced.
3308 GCC enables this option by default. If you want to force the compiler to
3309 check if the variable was referenced, regardless of whether or not
3310 optimization is turned on, use the @samp{-fno-keep-static-consts} option.
3312 @item -fno-function-cse
3313 @opindex fno-function-cse
3314 Do not put function addresses in registers; make each instruction that
3315 calls a constant function contain the function's address explicitly.
3317 This option results in less efficient code, but some strange hacks
3318 that alter the assembler output may be confused by the optimizations
3319 performed when this option is not used.
3323 Sets @samp{-fno-math-errno}, @samp{-funsafe-math-optimizations},
3324 and @samp{-fno-trapping-math}.
3326 This option causes the preprocessor macro __FAST_MATH__ to be defined.
3328 This option should never be turned on by any @samp{-O} option since
3329 it can result in incorrect output for programs which depend on
3330 an exact implementation of IEEE or ISO rules/specifications for
3333 @item -fno-math-errno
3334 @opindex fno-math-errno
3335 Do not set ERRNO after calling math functions that are executed
3336 with a single instruction, e.g., sqrt. A program that relies on
3337 IEEE exceptions for math error handling may want to use this flag
3338 for speed while maintaining IEEE arithmetic compatibility.
3340 This option should never be turned on by any @samp{-O} option since
3341 it can result in incorrect output for programs which depend on
3342 an exact implementation of IEEE or ISO rules/specifications for
3345 The default is @samp{-fmath-errno}. The @samp{-ffast-math} option
3346 sets @samp{-fno-math-errno}.
3348 @item -funsafe-math-optimizations
3349 @opindex funsafe-math-optimizations
3350 Allow optimizations for floating-point arithmetic that (a) assume
3351 that arguments and results are valid and (b) may violate IEEE or
3354 This option should never be turned on by any @samp{-O} option since
3355 it can result in incorrect output for programs which depend on
3356 an exact implementation of IEEE or ISO rules/specifications for
3359 The default is @samp{-fno-unsafe-math-optimizations}. The
3360 @samp{-ffast-math} option sets @samp{-funsafe-math-optimizations}.
3362 @item -fno-trapping-math
3363 @opindex fno-trapping-math
3364 Compile code assuming that floating-point operations cannot generate
3365 user-visible traps. Setting this option may allow faster code
3366 if one relies on ``non-stop'' IEEE arithmetic, for example.
3368 This option should never be turned on by any @samp{-O} option since
3369 it can result in incorrect output for programs which depend on
3370 an exact implementation of IEEE or ISO rules/specifications for
3373 The default is @samp{-ftrapping-math}. The @samp{-ffast-math}
3374 option sets @samp{-fno-trapping-math}.
3377 @c following causes underfulls.. they don't look great, but we deal.
3379 The following options control specific optimizations. The @samp{-O2}
3380 option turns on all of these optimizations except @samp{-funroll-loops}
3381 and @samp{-funroll-all-loops}. On most machines, the @samp{-O} option
3382 turns on the @samp{-fthread-jumps} and @samp{-fdelayed-branch} options,
3383 but specific machines may handle it differently.
3385 You can use the following flags in the rare cases when ``fine-tuning''
3386 of optimizations to be performed is desired.
3389 @item -fstrength-reduce
3390 @opindex fstrength-reduce
3391 Perform the optimizations of loop strength reduction and
3392 elimination of iteration variables.
3394 @item -fthread-jumps
3395 @opindex fthread-jumps
3396 Perform optimizations where we check to see if a jump branches to a
3397 location where another comparison subsumed by the first is found. If
3398 so, the first branch is redirected to either the destination of the
3399 second branch or a point immediately following it, depending on whether
3400 the condition is known to be true or false.
3402 @item -fcse-follow-jumps
3403 @opindex fcse-follow-jumps
3404 In common subexpression elimination, scan through jump instructions
3405 when the target of the jump is not reached by any other path. For
3406 example, when CSE encounters an @code{if} statement with an
3407 @code{else} clause, CSE will follow the jump when the condition
3410 @item -fcse-skip-blocks
3411 @opindex fcse-skip-blocks
3412 This is similar to @samp{-fcse-follow-jumps}, but causes CSE to
3413 follow jumps which conditionally skip over blocks. When CSE
3414 encounters a simple @code{if} statement with no else clause,
3415 @samp{-fcse-skip-blocks} causes CSE to follow the jump around the
3416 body of the @code{if}.
3418 @item -frerun-cse-after-loop
3419 @opindex frerun-cse-after-loop
3420 Re-run common subexpression elimination after loop optimizations has been
3423 @item -frerun-loop-opt
3424 @opindex frerun-loop-opt
3425 Run the loop optimizer twice.
3429 Perform a global common subexpression elimination pass.
3430 This pass also performs global constant and copy propagation.
3434 When -fgcse-lm is enabled, global common subexpression elimination will
3435 attempt to move loads which are only killed by stores into themselves. This
3436 allows a loop containing a load/store sequence to be changed to a load outside
3437 the loop, and a copy/store within the loop.
3441 When -fgcse-sm is enabled, A store motion pass is run after global common
3442 subexpression elimination. This pass will attempt to move stores out of loops.
3443 When used in conjunction with -fgcse-lm, loops containing a load/store sequence
3444 can be changed to a load before the loop and a store after the loop.
3446 @item -fdelete-null-pointer-checks
3447 @opindex fdelete-null-pointer-checks
3448 Use global dataflow analysis to identify and eliminate useless null
3449 pointer checks. Programs which rely on NULL pointer dereferences @emph{not}
3450 halting the program may not work properly with this option. Use
3451 -fno-delete-null-pointer-checks to disable this optimizing for programs
3452 which depend on that behavior.
3454 @item -fexpensive-optimizations
3455 @opindex fexpensive-optimizations
3456 Perform a number of minor optimizations that are relatively expensive.
3458 @item -foptimize-register-move
3460 @opindex foptimize-register-move
3462 Attempt to reassign register numbers in move instructions and as
3463 operands of other simple instructions in order to maximize the amount of
3464 register tying. This is especially helpful on machines with two-operand
3465 instructions. GCC enables this optimization by default with @samp{-O2}
3468 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
3471 @item -fdelayed-branch
3472 @opindex fdelayed-branch
3473 If supported for the target machine, attempt to reorder instructions
3474 to exploit instruction slots available after delayed branch
3477 @item -fschedule-insns
3478 @opindex fschedule-insns
3479 If supported for the target machine, attempt to reorder instructions to
3480 eliminate execution stalls due to required data being unavailable. This
3481 helps machines that have slow floating point or memory load instructions
3482 by allowing other instructions to be issued until the result of the load
3483 or floating point instruction is required.
3485 @item -fschedule-insns2
3486 @opindex fschedule-insns2
3487 Similar to @samp{-fschedule-insns}, but requests an additional pass of
3488 instruction scheduling after register allocation has been done. This is
3489 especially useful on machines with a relatively small number of
3490 registers and where memory load instructions take more than one cycle.
3492 @item -ffunction-sections
3493 @itemx -fdata-sections
3494 @opindex ffunction-sections
3495 @opindex fdata-sections
3496 Place each function or data item into its own section in the output
3497 file if the target supports arbitrary sections. The name of the
3498 function or the name of the data item determines the section's name
3501 Use these options on systems where the linker can perform optimizations
3502 to improve locality of reference in the instruction space. HPPA
3503 processors running HP-UX and Sparc processors running Solaris 2 have
3504 linkers with such optimizations. Other systems using the ELF object format
3505 as well as AIX may have these optimizations in the future.
3507 Only use these options when there are significant benefits from doing
3508 so. When you specify these options, the assembler and linker will
3509 create larger object and executable files and will also be slower.
3510 You will not be able to use @code{gprof} on all systems if you
3511 specify this option and you may have problems with debugging if
3512 you specify both this option and @samp{-g}.
3514 @item -fcaller-saves
3515 @opindex fcaller-saves
3516 Enable values to be allocated in registers that will be clobbered by
3517 function calls, by emitting extra instructions to save and restore the
3518 registers around such calls. Such allocation is done only when it
3519 seems to result in better code than would otherwise be produced.
3521 This option is always enabled by default on certain machines, usually
3522 those which have no call-preserved registers to use instead.
3524 For all machines, optimization level 2 and higher enables this flag by
3527 @item -funroll-loops
3528 @opindex funroll-loops
3529 Perform the optimization of loop unrolling. This is only done for loops
3530 whose number of iterations can be determined at compile time or run time.
3531 @samp{-funroll-loops} implies both @samp{-fstrength-reduce} and
3532 @samp{-frerun-cse-after-loop}.
3534 @item -funroll-all-loops
3535 @opindex funroll-all-loops
3536 Perform the optimization of loop unrolling. This is done for all loops
3537 and usually makes programs run more slowly. @samp{-funroll-all-loops}
3538 implies @samp{-fstrength-reduce} as well as @samp{-frerun-cse-after-loop}.
3540 @item -fmove-all-movables
3541 @opindex fmove-all-movables
3542 Forces all invariant computations in loops to be moved
3545 @item -freduce-all-givs
3546 @opindex freduce-all-givs
3547 Forces all general-induction variables in loops to be
3550 @emph{Note:} When compiling programs written in Fortran,
3551 @samp{-fmove-all-movables} and @samp{-freduce-all-givs} are enabled
3552 by default when you use the optimizer.
3554 These options may generate better or worse code; results are highly
3555 dependent on the structure of loops within the source code.
3557 These two options are intended to be removed someday, once
3558 they have helped determine the efficacy of various
3559 approaches to improving loop optimizations.
3561 Please let us (@w{@email{gcc@@gcc.gnu.org}} and @w{@email{fortran@@gnu.org}})
3562 know how use of these options affects
3563 the performance of your production code.
3564 We're very interested in code that runs @emph{slower}
3565 when these options are @emph{enabled}.
3568 @opindex fno-peephole
3569 Disable any machine-specific peephole optimizations.
3571 @item -fbranch-probabilities
3572 @opindex fbranch-probabilities
3573 After running a program compiled with @samp{-fprofile-arcs}
3574 (@pxref{Debugging Options,, Options for Debugging Your Program or
3575 @command{gcc}}), you can compile it a second time using
3576 @samp{-fbranch-probabilities}, to improve optimizations based on
3577 guessing the path a branch might take.
3580 With @samp{-fbranch-probabilities}, GCC puts a @samp{REG_EXEC_COUNT}
3581 note on the first instruction of each basic block, and a
3582 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
3583 These can be used to improve optimization. Currently, they are only
3584 used in one place: in @file{reorg.c}, instead of guessing which path a
3585 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
3586 exactly determine which path is taken more often.
3589 @item -fno-guess-branch-probability
3590 @opindex fno-guess-branch-probability
3591 Sometimes gcc will opt to guess branch probabilities when none are
3592 available from either profile directed feedback (@samp{-fprofile-arcs})
3593 or @samp{__builtin_expect}. In a hard real-time system, people don't
3594 want different runs of the compiler to produce code that has different
3595 behavior; minimizing non-determinism is of paramount import. This
3596 switch allows users to reduce non-determinism, possibly at the expense
3597 of inferior optimization.
3599 @item -fstrict-aliasing
3600 @opindex fstrict-aliasing
3601 Allows the compiler to assume the strictest aliasing rules applicable to
3602 the language being compiled. For C (and C++), this activates
3603 optimizations based on the type of expressions. In particular, an
3604 object of one type is assumed never to reside at the same address as an
3605 object of a different type, unless the types are almost the same. For
3606 example, an @code{unsigned int} can alias an @code{int}, but not a
3607 @code{void*} or a @code{double}. A character type may alias any other
3610 Pay special attention to code like this:
3623 The practice of reading from a different union member than the one most
3624 recently written to (called ``type-punning'') is common. Even with
3625 @samp{-fstrict-aliasing}, type-punning is allowed, provided the memory
3626 is accessed through the union type. So, the code above will work as
3627 expected. However, this code might not:
3639 Every language that wishes to perform language-specific alias analysis
3640 should define a function that computes, given an @code{tree}
3641 node, an alias set for the node. Nodes in different alias sets are not
3642 allowed to alias. For an example, see the C front-end function
3643 @code{c_get_alias_set}.
3646 @item -falign-functions
3647 @itemx -falign-functions=@var{n}
3648 @opindex falign-functions
3649 Align the start of functions to the next power-of-two greater than
3650 @var{n}, skipping up to @var{n} bytes. For instance,
3651 @samp{-falign-functions=32} aligns functions to the next 32-byte
3652 boundary, but @samp{-falign-functions=24} would align to the next
3653 32-byte boundary only if this can be done by skipping 23 bytes or less.
3655 @samp{-fno-align-functions} and @samp{-falign-functions=1} are
3656 equivalent and mean that functions will not be aligned.
3658 Some assemblers only support this flag when @var{n} is a power of two;
3659 in that case, it is rounded up.
3661 If @var{n} is not specified, use a machine-dependent default.
3663 @item -falign-labels
3664 @itemx -falign-labels=@var{n}
3665 @opindex falign-labels
3666 Align all branch targets to a power-of-two boundary, skipping up to
3667 @var{n} bytes like @samp{-falign-functions}. This option can easily
3668 make code slower, because it must insert dummy operations for when the
3669 branch target is reached in the usual flow of the code.
3671 If @samp{-falign-loops} or @samp{-falign-jumps} are applicable and
3672 are greater than this value, then their values are used instead.
3674 If @var{n} is not specified, use a machine-dependent default which is
3675 very likely to be @samp{1}, meaning no alignment.
3678 @itemx -falign-loops=@var{n}
3679 @opindex falign-loops
3680 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
3681 like @samp{-falign-functions}. The hope is that the loop will be
3682 executed many times, which will make up for any execution of the dummy
3685 If @var{n} is not specified, use a machine-dependent default.
3688 @itemx -falign-jumps=@var{n}
3689 @opindex falign-jumps
3690 Align branch targets to a power-of-two boundary, for branch targets
3691 where the targets can only be reached by jumping, skipping up to @var{n}
3692 bytes like @samp{-falign-functions}. In this case, no dummy operations
3695 If @var{n} is not specified, use a machine-dependent default.
3699 Perform optimizations in static single assignment form. Each function's
3700 flow graph is translated into SSA form, optimizations are performed, and
3701 the flow graph is translated back from SSA form. Users should not
3702 specify this option, since it is not yet ready for production use.
3706 Perform dead-code elimination in SSA form. Requires @samp{-fssa}. Like
3707 @samp{-fssa}, this is an experimental feature.
3709 @item -fsingle-precision-constant
3710 @opindex fsingle-precision-constant
3711 Treat floating point constant as single precision constant instead of
3712 implicitly converting it to double precision constant.
3714 @item -frename-registers
3715 @opindex frename-registers
3716 Attempt to avoid false dependancies in scheduled code by making use
3717 of registers left over after register allocation. This optimization
3718 will most benefit processors with lots of registers. It can, however,
3719 make debugging impossible, since variables will no longer stay in
3720 a ``home register''.
3722 @item --param @var{name}=@var{value}
3724 In some places, GCC uses various constants to control the amount of
3725 optimization that is done. For example, GCC will not inline functions
3726 that contain more that a certain number of instructions. You can
3727 control some of these constants on the command-line using the
3728 @samp{--param} option.
3730 In each case, the @var{value} is a integer. The allowable choices for
3731 @var{name} are given in the following table:
3734 @item max-delay-slot-insn-search
3735 The maximum number of instructions to consider when looking for an
3736 instruction to fill a delay slot. If more than this arbitrary number of
3737 instructions is searched, the time savings from filling the delay slot
3738 will be minimal so stop searching. Increasing values mean more
3739 aggressive optimization, making the compile time increase with probably
3740 small improvement in executable run time.
3742 @item max-delay-slot-live-search
3743 When trying to fill delay slots, the maximum number of instructions to
3744 consider when searching for a block with valid live register
3745 information. Increasing this arbitrarily chosen value means more
3746 aggressive optimization, increasing the compile time. This parameter
3747 should be removed when the delay slot code is rewritten to maintain the
3750 @item max-gcse-memory
3751 The approximate maximum amount of memory that will be allocated in
3752 order to perform the global common subexpression elimination
3753 optimization. If more memory than specified is required, the
3754 optimization will not be done.
3756 @item max-inline-insns
3757 If an function contains more than this many instructions, it
3758 will not be inlined. This option is precisely equivalent to
3759 @samp{-finline-limit}.
3764 @node Preprocessor Options
3765 @section Options Controlling the Preprocessor
3766 @cindex preprocessor options
3767 @cindex options, preprocessor
3769 These options control the C preprocessor, which is run on each C source
3770 file before actual compilation.
3772 If you use the @samp{-E} option, nothing is done except preprocessing.
3773 Some of these options make sense only together with @samp{-E} because
3774 they cause the preprocessor output to be unsuitable for actual
3778 @item -include @var{file}
3780 Process @var{file} as input before processing the regular input file.
3781 In effect, the contents of @var{file} are compiled first. Any @samp{-D}
3782 and @samp{-U} options on the command line are always processed before
3783 @samp{-include @var{file}}, regardless of the order in which they are
3784 written. All the @samp{-include} and @samp{-imacros} options are
3785 processed in the order in which they are written.
3787 @item -imacros @var{file}
3789 Process @var{file} as input, discarding the resulting output, before
3790 processing the regular input file. Because the output generated from
3791 @var{file} is discarded, the only effect of @samp{-imacros @var{file}}
3792 is to make the macros defined in @var{file} available for use in the
3793 main input. All the @samp{-include} and @samp{-imacros} options are
3794 processed in the order in which they are written.
3796 @item -idirafter @var{dir}
3798 @cindex second include path
3799 Add the directory @var{dir} to the second include path. The directories
3800 on the second include path are searched when a header file is not found
3801 in any of the directories in the main include path (the one that
3804 @item -iprefix @var{prefix}
3806 Specify @var{prefix} as the prefix for subsequent @samp{-iwithprefix}
3809 @item -iwithprefix @var{dir}
3810 @opindex iwithprefix
3811 Add a directory to the second include path. The directory's name is
3812 made by concatenating @var{prefix} and @var{dir}, where @var{prefix} was
3813 specified previously with @samp{-iprefix}. If you have not specified a
3814 prefix yet, the directory containing the installed passes of the
3815 compiler is used as the default.
3817 @item -iwithprefixbefore @var{dir}
3818 @opindex iwithprefixbefore
3819 Add a directory to the main include path. The directory's name is made
3820 by concatenating @var{prefix} and @var{dir}, as in the case of
3821 @samp{-iwithprefix}.
3823 @item -isystem @var{dir}
3825 Add a directory to the beginning of the second include path, marking it
3826 as a system directory, so that it gets the same special treatment as
3827 is applied to the standard system directories.
3831 Do not search the standard system directories for header files. Only
3832 the directories you have specified with @samp{-I} options (and the
3833 current directory, if appropriate) are searched. @xref{Directory
3834 Options}, for information on @samp{-I}.
3836 By using both @samp{-nostdinc} and @samp{-I-}, you can limit the include-file
3837 search path to only those directories you specify explicitly.
3841 When searching for a header file in a directory, remap file names if a
3842 file named @file{header.gcc} exists in that directory. This can be used
3843 to work around limitations of file systems with file name restrictions.
3844 The @file{header.gcc} file should contain a series of lines with two
3845 tokens on each line: the first token is the name to map, and the second
3846 token is the actual name to use.
3850 Do not predefine any nonstandard macros. (Including architecture flags).
3854 Run only the C preprocessor. Preprocess all the C source files
3855 specified and output the results to standard output or to the
3856 specified output file.
3860 Tell the preprocessor not to discard comments. Used with the
3865 Tell the preprocessor not to generate @samp{#line} directives.
3866 Used with the @samp{-E} option.
3869 @cindex dependencies, make
3872 Instead of outputting the result of preprocessing, output a rule
3873 suitable for @code{make} describing the dependencies of the main source
3874 file. The preprocessor outputs one @code{make} rule containing the
3875 object file name for that source file, a colon, and the names of all the
3876 included files. Unless overridden explicitly, the object file name
3877 consists of the basename of the source file with any suffix replaced with
3878 object file suffix. If there are many included files then the
3879 rule is split into several lines using @samp{\}-newline.
3881 @samp{-M} implies @samp{-E}.
3885 Like @samp{-M}, but mention only the files included with @samp{#include
3886 "@var{file}"}. System header files included with @samp{#include
3887 <@var{file}>} are omitted.
3891 Like @samp{-M} but the dependency information is written to a file
3892 rather than stdout. @code{gcc} will use the same file name and
3893 directory as the object file, but with the suffix ".d" instead.
3895 This is in addition to compiling the main file as specified ---
3896 @samp{-MD} does not inhibit ordinary compilation the way @samp{-M} does,
3897 unless you also specify @samp{-MG}.
3899 With Mach, you can use the utility @code{md} to merge multiple
3900 dependency files into a single dependency file suitable for using with
3901 the @samp{make} command.
3905 Like @samp{-MD} except mention only user header files, not system
3908 @item -MF @var{file}
3910 When used with @samp{-M} or @samp{-MM}, specifies a file to write the
3911 dependencies to. This allows the preprocessor to write the preprocessed
3912 file to stdout normally. If no @samp{-MF} switch is given, CPP sends
3913 the rules to stdout and suppresses normal preprocessed output.
3915 Another way to specify output of a @code{make} rule is by setting
3916 the environment variable @env{DEPENDENCIES_OUTPUT} (@pxref{Environment
3921 When used with @samp{-M} or @samp{-MM}, @samp{-MG} says to treat missing
3922 header files as generated files and assume they live in the same
3923 directory as the source file. It suppresses preprocessed output, as a
3924 missing header file is ordinarily an error.
3926 This feature is used in automatic updating of makefiles.
3930 This option instructs CPP to add a phony target for each dependency
3931 other than the main file, causing each to depend on nothing. These
3932 dummy rules work around errors @code{make} gives if you remove header
3933 files without updating the @code{Makefile} to match.
3935 This is typical output:-
3938 /tmp/test.o: /tmp/test.c /tmp/test.h
3943 @item -MQ @var{target}
3944 @item -MT @var{target}
3947 By default CPP uses the main file name, including any path, and appends
3948 the object suffix, normally ``.o'', to it to obtain the name of the
3949 target for dependency generation. With @samp{-MT} you can specify a
3950 target yourself, overriding the default one.
3952 If you want multiple targets, you can specify them as a single argument
3953 to @samp{-MT}, or use multiple @samp{-MT} options.
3955 The targets you specify are output in the order they appear on the
3956 command line. @samp{-MQ} is identical to @samp{-MT}, except that the
3957 target name is quoted for Make, but with @samp{-MT} it isn't. For
3958 example, -MT '$(objpfx)foo.o' gives
3961 $(objpfx)foo.o: /tmp/foo.c
3964 but -MQ '$(objpfx)foo.o' gives
3967 $$(objpfx)foo.o: /tmp/foo.c
3970 The default target is automatically quoted, as if it were given with
3975 Print the name of each header file used, in addition to other normal
3978 @item -A@var{question}(@var{answer})
3980 Assert the answer @var{answer} for @var{question}, in case it is tested
3981 with a preprocessing conditional such as @samp{#if
3982 #@var{question}(@var{answer})}. @samp{-A-} disables the standard
3983 assertions that normally describe the target machine.
3987 Define macro @var{macro} with the string @samp{1} as its definition.
3989 @item -D@var{macro}=@var{defn}
3990 Define macro @var{macro} as @var{defn}. All instances of @samp{-D} on
3991 the command line are processed before any @samp{-U} options.
3993 Any @samp{-D} and @samp{-U} options on the command line are processed in
3994 order, and always before @samp{-imacros @var{file}}, regardless of the
3995 order in which they are written.
3999 Undefine macro @var{macro}. @samp{-U} options are evaluated after all
4000 @samp{-D} options, but before any @samp{-include} and @samp{-imacros}
4003 Any @samp{-D} and @samp{-U} options on the command line are processed in
4004 order, and always before @samp{-imacros @var{file}}, regardless of the
4005 order in which they are written.
4009 Tell the preprocessor to output only a list of the macro definitions
4010 that are in effect at the end of preprocessing. Used with the @samp{-E}
4015 Tell the preprocessing to pass all macro definitions into the output, in
4016 their proper sequence in the rest of the output.
4020 Like @samp{-dD} except that the macro arguments and contents are omitted.
4021 Only @samp{#define @var{name}} is included in the output.
4025 Output @samp{#include} directives in addition to the result of
4030 Process ISO standard trigraph sequences. These are three-character
4031 sequences, all starting with @samp{??}, that are defined by ISO C to
4032 stand for single characters. For example, @samp{??/} stands for
4033 @samp{\}, so @samp{'??/n'} is a character constant for a newline. By
4034 default, GCC ignores trigraphs, but in standard-conforming modes it
4035 converts them. See the @samp{-std} and @samp{-ansi} options.
4037 The nine trigraph sequences are
4068 Trigraph support is not popular, so many compilers do not implement it
4069 properly. Portable code should not rely on trigraphs being either
4070 converted or ignored.
4072 @item -Wp\,@var{option}
4074 Pass @var{option} as an option to the preprocessor. If @var{option}
4075 contains commas, it is split into multiple options at the commas.
4078 @node Assembler Options
4079 @section Passing Options to the Assembler
4081 @c prevent bad page break with this line
4082 You can pass options to the assembler.
4085 @item -Wa\,@var{option}
4087 Pass @var{option} as an option to the assembler. If @var{option}
4088 contains commas, it is split into multiple options at the commas.
4092 @section Options for Linking
4093 @cindex link options
4094 @cindex options, linking
4096 These options come into play when the compiler links object files into
4097 an executable output file. They are meaningless if the compiler is
4098 not doing a link step.
4102 @item @var{object-file-name}
4103 A file name that does not end in a special recognized suffix is
4104 considered to name an object file or library. (Object files are
4105 distinguished from libraries by the linker according to the file
4106 contents.) If linking is done, these object files are used as input
4115 If any of these options is used, then the linker is not run, and
4116 object file names should not be used as arguments. @xref{Overall
4120 @item -l@var{library}
4121 @itemx -l @var{library}
4123 Search the library named @var{library} when linking. (The second
4124 alternative with the library as a separate argument is only for
4125 POSIX compliance and is not recommended.)
4127 It makes a difference where in the command you write this option; the
4128 linker searches and processes libraries and object files in the order they
4129 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
4130 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
4131 to functions in @samp{z}, those functions may not be loaded.
4133 The linker searches a standard list of directories for the library,
4134 which is actually a file named @file{lib@var{library}.a}. The linker
4135 then uses this file as if it had been specified precisely by name.
4137 The directories searched include several standard system directories
4138 plus any that you specify with @samp{-L}.
4140 Normally the files found this way are library files---archive files
4141 whose members are object files. The linker handles an archive file by
4142 scanning through it for members which define symbols that have so far
4143 been referenced but not defined. But if the file that is found is an
4144 ordinary object file, it is linked in the usual fashion. The only
4145 difference between using an @samp{-l} option and specifying a file name
4146 is that @samp{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
4147 and searches several directories.
4151 You need this special case of the @samp{-l} option in order to
4152 link an Objective C program.
4155 @opindex nostartfiles
4156 Do not use the standard system startup files when linking.
4157 The standard system libraries are used normally, unless @option{-nostdlib}
4158 or @option{-nodefaultlibs} is used.
4160 @item -nodefaultlibs
4161 @opindex nodefaultlibs
4162 Do not use the standard system libraries when linking.
4163 Only the libraries you specify will be passed to the linker.
4164 The standard startup files are used normally, unless @option{-nostartfiles}
4165 is used. The compiler may generate calls to memcmp, memset, and memcpy
4166 for System V (and ISO C) environments or to bcopy and bzero for
4167 BSD environments. These entries are usually resolved by entries in
4168 libc. These entry points should be supplied through some other
4169 mechanism when this option is specified.
4173 Do not use the standard system startup files or libraries when linking.
4174 No startup files and only the libraries you specify will be passed to
4175 the linker. The compiler may generate calls to memcmp, memset, and memcpy
4176 for System V (and ISO C) environments or to bcopy and bzero for
4177 BSD environments. These entries are usually resolved by entries in
4178 libc. These entry points should be supplied through some other
4179 mechanism when this option is specified.
4181 @cindex @code{-lgcc}, use with @code{-nostdlib}
4182 @cindex @code{-nostdlib} and unresolved references
4183 @cindex unresolved references and @code{-nostdlib}
4184 @cindex @code{-lgcc}, use with @code{-nodefaultlibs}
4185 @cindex @code{-nodefaultlibs} and unresolved references
4186 @cindex unresolved references and @code{-nodefaultlibs}
4187 One of the standard libraries bypassed by @samp{-nostdlib} and
4188 @samp{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
4189 that GCC uses to overcome shortcomings of particular machines, or special
4190 needs for some languages.
4192 (@xref{Interface,,Interfacing to GCC Output}, for more discussion of
4196 (@xref{Interface,,Interfacing to GCC Output,gcc.info,Porting GCC},
4197 for more discussion of @file{libgcc.a}.)
4199 In most cases, you need @file{libgcc.a} even when you want to avoid
4200 other standard libraries. In other words, when you specify @samp{-nostdlib}
4201 or @samp{-nodefaultlibs} you should usually specify @samp{-lgcc} as well.
4202 This ensures that you have no unresolved references to internal GCC
4203 library subroutines. (For example, @samp{__main}, used to ensure C++
4204 constructors will be called; @pxref{Collect2,,@command{collect2}}.)
4208 Remove all symbol table and relocation information from the executable.
4212 On systems that support dynamic linking, this prevents linking with the shared
4213 libraries. On other systems, this option has no effect.
4217 Produce a shared object which can then be linked with other objects to
4218 form an executable. Not all systems support this option. For predictable
4219 results, you must also specify the same set of options that were used to
4220 generate code (@samp{-fpic}, @samp{-fPIC}, or model suboptions)
4221 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
4222 needs to build supplementary stub code for constructors to work. On
4223 multi-libbed systems, @samp{gcc -shared} must select the correct support
4224 libraries to link against. Failing to supply the correct flags may lead
4225 to subtle defects. Supplying them in cases where they are not necessary
4228 @item -shared-libgcc
4229 @itemx -static-libgcc
4230 @opindex shared-libgcc
4231 @opindex static-libgcc
4232 On systems that provide @file{libgcc} as a shared library, these options
4233 force the use of either the shared or static version respectively.
4234 If no shared version of @file{libgcc} was built when the compiler was
4235 configured, these options have no effect.
4237 There are several situations in which an application should use the
4238 shared @file{libgcc} instead of the static version. The most common
4239 of these is when the application wishes to throw and catch exceptions
4240 across different shared libraries. In that case, each of the libraries
4241 as well as the application itself should use the shared @file{libgcc}.
4243 Therefore, whenever you specify the @samp{-shared} option, the GCC
4244 driver automatically adds @samp{-shared-libgcc}, unless you explicitly
4245 specify @samp{-static-libgcc}. The G++ driver automatically adds
4246 @samp{-shared-libgcc} when you build a main executable as well because
4247 for C++ programs that is typically the right thing to do.
4248 (Exception-handling will not work reliably otherwise.)
4250 However, when linking a main executable written in C, you must
4251 explicitly say @samp{-shared-libgcc} if you want to use the shared
4256 Bind references to global symbols when building a shared object. Warn
4257 about any unresolved references (unless overridden by the link editor
4258 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
4261 @item -Xlinker @var{option}
4263 Pass @var{option} as an option to the linker. You can use this to
4264 supply system-specific linker options which GCC does not know how to
4267 If you want to pass an option that takes an argument, you must use
4268 @samp{-Xlinker} twice, once for the option and once for the argument.
4269 For example, to pass @samp{-assert definitions}, you must write
4270 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
4271 @samp{-Xlinker "-assert definitions"}, because this passes the entire
4272 string as a single argument, which is not what the linker expects.
4274 @item -Wl\,@var{option}
4276 Pass @var{option} as an option to the linker. If @var{option} contains
4277 commas, it is split into multiple options at the commas.
4279 @item -u @var{symbol}
4281 Pretend the symbol @var{symbol} is undefined, to force linking of
4282 library modules to define it. You can use @samp{-u} multiple times with
4283 different symbols to force loading of additional library modules.
4286 @node Directory Options
4287 @section Options for Directory Search
4288 @cindex directory options
4289 @cindex options, directory search
4292 These options specify directories to search for header files, for
4293 libraries and for parts of the compiler:
4298 Add the directory @var{dir} to the head of the list of directories to be
4299 searched for header files. This can be used to override a system header
4300 file, substituting your own version, since these directories are
4301 searched before the system header file directories. However, you should
4302 not use this option to add directories that contain vendor-supplied
4303 system header files (use @samp{-isystem} for that). If you use more than
4304 one @samp{-I} option, the directories are scanned in left-to-right
4305 order; the standard system directories come after.
4309 Any directories you specify with @samp{-I} options before the @samp{-I-}
4310 option are searched only for the case of @samp{#include "@var{file}"};
4311 they are not searched for @samp{#include <@var{file}>}.
4313 If additional directories are specified with @samp{-I} options after
4314 the @samp{-I-}, these directories are searched for all @samp{#include}
4315 directives. (Ordinarily @emph{all} @samp{-I} directories are used
4318 In addition, the @samp{-I-} option inhibits the use of the current
4319 directory (where the current input file came from) as the first search
4320 directory for @samp{#include "@var{file}"}. There is no way to
4321 override this effect of @samp{-I-}. With @samp{-I.} you can specify
4322 searching the directory which was current when the compiler was
4323 invoked. That is not exactly the same as what the preprocessor does
4324 by default, but it is often satisfactory.
4326 @samp{-I-} does not inhibit the use of the standard system directories
4327 for header files. Thus, @samp{-I-} and @samp{-nostdinc} are
4332 Add directory @var{dir} to the list of directories to be searched
4335 @item -B@var{prefix}
4337 This option specifies where to find the executables, libraries,
4338 include files, and data files of the compiler itself.
4340 The compiler driver program runs one or more of the subprograms
4341 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
4342 @var{prefix} as a prefix for each program it tries to run, both with and
4343 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
4345 For each subprogram to be run, the compiler driver first tries the
4346 @samp{-B} prefix, if any. If that name is not found, or if @samp{-B}
4347 was not specified, the driver tries two standard prefixes, which are
4348 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc-lib/}. If neither of
4349 those results in a file name that is found, the unmodified program
4350 name is searched for using the directories specified in your
4351 @env{PATH} environment variable.
4353 @samp{-B} prefixes that effectively specify directory names also apply
4354 to libraries in the linker, because the compiler translates these
4355 options into @samp{-L} options for the linker. They also apply to
4356 includes files in the preprocessor, because the compiler translates these
4357 options into @samp{-isystem} options for the preprocessor. In this case,
4358 the compiler appends @samp{include} to the prefix.
4360 The run-time support file @file{libgcc.a} can also be searched for using
4361 the @samp{-B} prefix, if needed. If it is not found there, the two
4362 standard prefixes above are tried, and that is all. The file is left
4363 out of the link if it is not found by those means.
4365 Another way to specify a prefix much like the @samp{-B} prefix is to use
4366 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
4369 @item -specs=@var{file}
4371 Process @var{file} after the compiler reads in the standard @file{specs}
4372 file, in order to override the defaults that the @file{gcc} driver
4373 program uses when determining what switches to pass to @file{cc1},
4374 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
4375 @samp{-specs=}@var{file} can be specified on the command line, and they
4376 are processed in order, from left to right.
4382 @section Specifying subprocesses and the switches to pass to them
4384 @command{gcc} is a driver program. It performs its job by invoking a
4385 sequence of other programs to do the work of compiling, assembling and
4386 linking. GCC interprets its command-line parameters and uses these to
4387 deduce which programs it should invoke, and which command-line options
4388 it ought to place on their command lines. This behaviour is controlled
4389 by @dfn{spec strings}. In most cases there is one spec string for each
4390 program that GCC can invoke, but a few programs have multiple spec
4391 strings to control their behaviour. The spec strings built into GCC can
4392 be overridden by using the @samp{-specs=} command-line switch to specify
4395 @dfn{Spec files} are plaintext files that are used to construct spec
4396 strings. They consist of a sequence of directives separated by blank
4397 lines. The type of directive is determined by the first non-whitespace
4398 character on the line and it can be one of the following:
4401 @item %@var{command}
4402 Issues a @var{command} to the spec file processor. The commands that can
4406 @item %include <@var{file}>
4408 Search for @var{file} and insert its text at the current point in the
4411 @item %include_noerr <@var{file}>
4412 @cindex %include_noerr
4413 Just like @samp{%include}, but do not generate an error message if the include
4414 file cannot be found.
4416 @item %rename @var{old_name} @var{new_name}
4418 Rename the spec string @var{old_name} to @var{new_name}.
4422 @item *[@var{spec_name}]:
4423 This tells the compiler to create, override or delete the named spec
4424 string. All lines after this directive up to the next directive or
4425 blank line are considered to be the text for the spec string. If this
4426 results in an empty string then the spec will be deleted. (Or, if the
4427 spec did not exist, then nothing will happened.) Otherwise, if the spec
4428 does not currently exist a new spec will be created. If the spec does
4429 exist then its contents will be overridden by the text of this
4430 directive, unless the first character of that text is the @samp{+}
4431 character, in which case the text will be appended to the spec.
4433 @item [@var{suffix}]:
4434 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
4435 and up to the next directive or blank line are considered to make up the
4436 spec string for the indicated suffix. When the compiler encounters an
4437 input file with the named suffix, it will processes the spec string in
4438 order to work out how to compile that file. For example:
4445 This says that any input file whose name ends in @samp{.ZZ} should be
4446 passed to the program @samp{z-compile}, which should be invoked with the
4447 command-line switch @samp{-input} and with the result of performing the
4448 @samp{%i} substitution. (See below.)
4450 As an alternative to providing a spec string, the text that follows a
4451 suffix directive can be one of the following:
4454 @item @@@var{language}
4455 This says that the suffix is an alias for a known @var{language}. This is
4456 similar to using the @option{-x} command-line switch to GCC to specify a
4457 language explicitly. For example:
4464 Says that .ZZ files are, in fact, C++ source files.
4467 This causes an error messages saying:
4470 @var{name} compiler not installed on this system.
4474 GCC already has an extensive list of suffixes built into it.
4475 This directive will add an entry to the end of the list of suffixes, but
4476 since the list is searched from the end backwards, it is effectively
4477 possible to override earlier entries using this technique.
4481 GCC has the following spec strings built into it. Spec files can
4482 override these strings or create their own. Note that individual
4483 targets can also add their own spec strings to this list.
4486 asm Options to pass to the assembler
4487 asm_final Options to pass to the assembler post-processor
4488 cpp Options to pass to the C preprocessor
4489 cc1 Options to pass to the C compiler
4490 cc1plus Options to pass to the C++ compiler
4491 endfile Object files to include at the end of the link
4492 link Options to pass to the linker
4493 lib Libraries to include on the command line to the linker
4494 libgcc Decides which GCC support library to pass to the linker
4495 linker Sets the name of the linker
4496 predefines Defines to be passed to the C preprocessor
4497 signed_char Defines to pass to CPP to say whether @code{char} is signed
4499 startfile Object files to include at the start of the link
4502 Here is a small example of a spec file:
4508 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
4511 This example renames the spec called @samp{lib} to @samp{old_lib} and
4512 then overrides the previous definition of @samp{lib} with a new one.
4513 The new definition adds in some extra command-line options before
4514 including the text of the old definition.
4516 @dfn{Spec strings} are a list of command-line options to be passed to their
4517 corresponding program. In addition, the spec strings can contain
4518 @samp{%}-prefixed sequences to substitute variable text or to
4519 conditionally insert text into the command line. Using these constructs
4520 it is possible to generate quite complex command lines.
4522 Here is a table of all defined @samp{%}-sequences for spec
4523 strings. Note that spaces are not generated automatically around the
4524 results of expanding these sequences. Therefore you can concatenate them
4525 together or combine them with constant text in a single argument.
4529 Substitute one @samp{%} into the program name or argument.
4532 Substitute the name of the input file being processed.
4535 Substitute the basename of the input file being processed.
4536 This is the substring up to (and not including) the last period
4537 and not including the directory.
4540 This is the same as @samp{%b}, but include the file suffix (text after
4544 Marks the argument containing or following the @samp{%d} as a
4545 temporary file name, so that that file will be deleted if GCC exits
4546 successfully. Unlike @samp{%g}, this contributes no text to the
4549 @item %g@var{suffix}
4550 Substitute a file name that has suffix @var{suffix} and is chosen
4551 once per compilation, and mark the argument in the same way as
4552 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
4553 name is now chosen in a way that is hard to predict even when previously
4554 chosen file names are known. For example, @samp{%g.s ... %g.o ... %g.s}
4555 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
4556 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
4557 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
4558 was simply substituted with a file name chosen once per compilation,
4559 without regard to any appended suffix (which was therefore treated
4560 just like ordinary text), making such attacks more likely to succeed.
4562 @item %u@var{suffix}
4563 Like @samp{%g}, but generates a new temporary file name even if
4564 @samp{%u@var{suffix}} was already seen.
4566 @item %U@var{suffix}
4567 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
4568 new one if there is no such last file name. In the absence of any
4569 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
4570 the same suffix @emph{space}, so @samp{%g.s ... %U.s ... %g.s ... %U.s}
4571 would involve the generation of two distinct file names, one
4572 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
4573 simply substituted with a file name chosen for the previous @samp{%u},
4574 without regard to any appended suffix.
4576 @item %j@var{SUFFIX}
4577 Substitutes the name of the HOST_BIT_BUCKET, if any, and if it is
4578 writable, and if save-temps is off; otherwise, substitute the name
4579 of a temporary file, just like @samp{%u}. This temporary file is not
4580 meant for communication between processes, but rather as a junk
4583 @item %.@var{SUFFIX}
4584 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
4585 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
4586 terminated by the next space or %.
4589 Marks the argument containing or following the @samp{%w} as the
4590 designated output file of this compilation. This puts the argument
4591 into the sequence of arguments that @samp{%o} will substitute later.
4594 Substitutes the names of all the output files, with spaces
4595 automatically placed around them. You should write spaces
4596 around the @samp{%o} as well or the results are undefined.
4597 @samp{%o} is for use in the specs for running the linker.
4598 Input files whose names have no recognized suffix are not compiled
4599 at all, but they are included among the output files, so they will
4603 Substitutes the suffix for object files. Note that this is
4604 handled specially when it immediately follows @samp{%g, %u, or %U},
4605 because of the need for those to form complete file names. The
4606 handling is such that @samp{%O} is treated exactly as if it had already
4607 been substituted, except that @samp{%g, %u, and %U} do not currently
4608 support additional @var{suffix} characters following @samp{%O} as they would
4609 following, for example, @samp{.o}.
4612 Substitutes the standard macro predefinitions for the
4613 current target machine. Use this when running @code{cpp}.
4616 Like @samp{%p}, but puts @samp{__} before and after the name of each
4617 predefined macro, except for macros that start with @samp{__} or with
4618 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
4622 Substitute a @samp{-iprefix} option made from GCC_EXEC_PREFIX.
4625 Current argument is the name of a library or startup file of some sort.
4626 Search for that file in a standard list of directories and substitute
4627 the full name found.
4630 Print @var{str} as an error message. @var{str} is terminated by a newline.
4631 Use this when inconsistent options are detected.
4634 Output @samp{-} if the input for the current command is coming from a pipe.
4637 Substitute the contents of spec string @var{name} at this point.
4640 Like @samp{%(...)} but put @samp{__} around @samp{-D} arguments.
4642 @item %x@{@var{option}@}
4643 Accumulate an option for @samp{%X}.
4646 Output the accumulated linker options specified by @samp{-Wl} or a @samp{%x}
4650 Output the accumulated assembler options specified by @samp{-Wa}.
4653 Output the accumulated preprocessor options specified by @samp{-Wp}.
4656 Substitute the major version number of GCC.
4657 (For version 2.9.5, this is 2.)
4660 Substitute the minor version number of GCC.
4661 (For version 2.9.5, this is 9.)
4664 Substitute the patch level number of GCC.
4665 (For version 2.9.5, this is 5.)
4668 Process the @code{asm} spec. This is used to compute the
4669 switches to be passed to the assembler.
4672 Process the @code{asm_final} spec. This is a spec string for
4673 passing switches to an assembler post-processor, if such a program is
4677 Process the @code{link} spec. This is the spec for computing the
4678 command line passed to the linker. Typically it will make use of the
4679 @samp{%L %G %S %D and %E} sequences.
4682 Dump out a @samp{-L} option for each directory that GCC believes might
4683 contain startup files. If the target supports multilibs then the
4684 current multilib directory will be prepended to each of these paths.
4687 Output the multilib directory with directory seperators replaced with
4688 "_". If multilib directories are not set, or the multilib directory is
4689 "." then this option emits nothing.
4692 Process the @code{lib} spec. This is a spec string for deciding which
4693 libraries should be included on the command line to the linker.
4696 Process the @code{libgcc} spec. This is a spec string for deciding
4697 which GCC support library should be included on the command line to the linker.
4700 Process the @code{startfile} spec. This is a spec for deciding which
4701 object files should be the first ones passed to the linker. Typically
4702 this might be a file named @file{crt0.o}.
4705 Process the @code{endfile} spec. This is a spec string that specifies
4706 the last object files that will be passed to the linker.
4709 Process the @code{cpp} spec. This is used to construct the arguments
4710 to be passed to the C preprocessor.
4713 Process the @code{signed_char} spec. This is intended to be used
4714 to tell cpp whether a char is signed. It typically has the definition:
4716 %@{funsigned-char:-D__CHAR_UNSIGNED__@}
4720 Process the @code{cc1} spec. This is used to construct the options to be
4721 passed to the actual C compiler (@samp{cc1}).
4724 Process the @code{cc1plus} spec. This is used to construct the options to be
4725 passed to the actual C++ compiler (@samp{cc1plus}).
4728 Substitute the variable part of a matched option. See below.
4729 Note that each comma in the substituted string is replaced by
4733 Substitutes the @code{-S} switch, if that switch was given to GCC.
4734 If that switch was not specified, this substitutes nothing. Note that
4735 the leading dash is omitted when specifying this option, and it is
4736 automatically inserted if the substitution is performed. Thus the spec
4737 string @samp{%@{foo@}} would match the command-line option @samp{-foo}
4738 and would output the command line option @samp{-foo}.
4740 @item %W@{@code{S}@}
4741 Like %@{@code{S}@} but mark last argument supplied within as a file to be
4744 @item %@{@code{S}*@}
4745 Substitutes all the switches specified to GCC whose names start
4746 with @code{-S}, but which also take an argument. This is used for
4747 switches like @samp{-o, -D, -I}, etc. GCC considers @samp{-o foo} as being
4748 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
4749 text, including the space. Thus two arguments would be generated.
4751 @item %@{^@code{S}*@}
4752 Like %@{@code{S}*@}, but don't put a blank between a switch and its
4753 argument. Thus %@{^o*@} would only generate one argument, not two.
4755 @item %@{@code{S}*&@code{T}*@}
4756 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
4757 (the order of @code{S} and @code{T} in the spec is not significant).
4758 There can be any number of ampersand-separated variables; for each the
4759 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
4761 @item %@{<@code{S}@}
4762 Remove all occurrences of @code{-S} from the command line. Note - this
4763 command is position dependent. @samp{%} commands in the spec string
4764 before this option will see @code{-S}, @samp{%} commands in the spec
4765 string after this option will not.
4767 @item %@{@code{S}*:@code{X}@}
4768 Substitutes @code{X} if one or more switches whose names start with
4769 @code{-S} are specified to GCC. Note that the tail part of the
4770 @code{-S} option (i.e. the part matched by the @samp{*}) will be substituted
4771 for each occurrence of @samp{%*} within @code{X}.
4773 @item %@{@code{S}:@code{X}@}
4774 Substitutes @code{X}, but only if the @samp{-S} switch was given to GCC.
4776 @item %@{!@code{S}:@code{X}@}
4777 Substitutes @code{X}, but only if the @samp{-S} switch was @emph{not} given to GCC.
4779 @item %@{|@code{S}:@code{X}@}
4780 Like %@{@code{S}:@code{X}@}, but if no @code{S} switch, substitute @samp{-}.
4782 @item %@{|!@code{S}:@code{X}@}
4783 Like %@{!@code{S}:@code{X}@}, but if there is an @code{S} switch, substitute @samp{-}.
4785 @item %@{.@code{S}:@code{X}@}
4786 Substitutes @code{X}, but only if processing a file with suffix @code{S}.
4788 @item %@{!.@code{S}:@code{X}@}
4789 Substitutes @code{X}, but only if @emph{not} processing a file with suffix @code{S}.
4791 @item %@{@code{S}|@code{P}:@code{X}@}
4792 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC. This may be
4793 combined with @samp{!} and @samp{.} sequences as well, although they
4794 have a stronger binding than the @samp{|}. For example a spec string
4798 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
4801 will output the following command-line options from the following input
4802 command-line options:
4807 -d fred.c -foo -baz -boggle
4808 -d jim.d -bar -baz -boggle
4813 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or
4814 %@{!@code{S}:@code{X}@} construct may contain other nested @samp{%} constructs
4815 or spaces, or even newlines. They are processed as usual, as described
4818 The @samp{-O, -f, -m, and -W} switches are handled specifically in these
4819 constructs. If another value of @samp{-O} or the negated form of a @samp{-f, -m, or
4820 -W} switch is found later in the command line, the earlier switch
4821 value is ignored, except with @{@code{S}*@} where @code{S} is just one
4822 letter, which passes all matching options.
4824 The character @samp{|} at the beginning of the predicate text is used to indicate
4825 that a command should be piped to the following command, but only if @samp{-pipe}
4828 It is built into GCC which switches take arguments and which do not.
4829 (You might think it would be useful to generalize this to allow each
4830 compiler's spec to say which switches take arguments. But this cannot
4831 be done in a consistent fashion. GCC cannot even decide which input
4832 files have been specified without knowing which switches take arguments,
4833 and it must know which input files to compile in order to tell which
4836 GCC also knows implicitly that arguments starting in @samp{-l} are to be
4837 treated as compiler output files, and passed to the linker in their
4838 proper position among the other output files.
4840 @c man begin OPTIONS
4842 @node Target Options
4843 @section Specifying Target Machine and Compiler Version
4844 @cindex target options
4845 @cindex cross compiling
4846 @cindex specifying machine version
4847 @cindex specifying compiler version and target machine
4848 @cindex compiler version, specifying
4849 @cindex target machine, specifying
4851 By default, GCC compiles code for the same type of machine that you
4852 are using. However, it can also be installed as a cross-compiler, to
4853 compile for some other type of machine. In fact, several different
4854 configurations of GCC, for different target machines, can be
4855 installed side by side. Then you specify which one to use with the
4858 In addition, older and newer versions of GCC can be installed side
4859 by side. One of them (probably the newest) will be the default, but
4860 you may sometimes wish to use another.
4863 @item -b @var{machine}
4865 The argument @var{machine} specifies the target machine for compilation.
4866 This is useful when you have installed GCC as a cross-compiler.
4868 The value to use for @var{machine} is the same as was specified as the
4869 machine type when configuring GCC as a cross-compiler. For
4870 example, if a cross-compiler was configured with @samp{configure
4871 i386v}, meaning to compile for an 80386 running System V, then you
4872 would specify @samp{-b i386v} to run that cross compiler.
4874 When you do not specify @samp{-b}, it normally means to compile for
4875 the same type of machine that you are using.
4877 @item -V @var{version}
4879 The argument @var{version} specifies which version of GCC to run.
4880 This is useful when multiple versions are installed. For example,
4881 @var{version} might be @samp{2.0}, meaning to run GCC version 2.0.
4883 The default version, when you do not specify @samp{-V}, is the last
4884 version of GCC that you installed.
4887 The @samp{-b} and @samp{-V} options actually work by controlling part of
4888 the file name used for the executable files and libraries used for
4889 compilation. A given version of GCC, for a given target machine, is
4890 normally kept in the directory @file{/usr/local/lib/gcc-lib/@var{machine}/@var{version}}.@refill
4892 Thus, sites can customize the effect of @samp{-b} or @samp{-V} either by
4893 changing the names of these directories or adding alternate names (or
4894 symbolic links). If in directory @file{/usr/local/lib/gcc-lib/} the
4895 file @file{80386} is a link to the file @file{i386v}, then @samp{-b
4896 80386} becomes an alias for @samp{-b i386v}.
4898 In one respect, the @samp{-b} or @samp{-V} do not completely change
4899 to a different compiler: the top-level driver program @command{gcc}
4900 that you originally invoked continues to run and invoke the other
4901 executables (preprocessor, compiler per se, assembler and linker)
4902 that do the real work. However, since no real work is done in the
4903 driver program, it usually does not matter that the driver program
4904 in use is not the one for the specified target. It is common for the
4905 interface to the other executables to change incompatibly between
4906 compiler versions, so unless the version specified is very close to that
4907 of the driver (for example, @samp{-V 3.0} with a driver program from GCC
4908 version 3.0.1), use of @samp{-V} may not work; for example, using
4909 @samp{-V 2.95.2} will not work with a driver program from GCC 3.0.
4911 The only way that the driver program depends on the target machine is
4912 in the parsing and handling of special machine-specific options.
4913 However, this is controlled by a file which is found, along with the
4914 other executables, in the directory for the specified version and
4915 target machine. As a result, a single installed driver program adapts
4916 to any specified target machine, and sufficiently similar compiler
4919 The driver program executable does control one significant thing,
4920 however: the default version and target machine. Therefore, you can
4921 install different instances of the driver program, compiled for
4922 different targets or versions, under different names.
4924 For example, if the driver for version 2.0 is installed as @command{ogcc}
4925 and that for version 2.1 is installed as @command{gcc}, then the command
4926 @command{gcc} will use version 2.1 by default, while @command{ogcc} will use
4927 2.0 by default. However, you can choose either version with either
4928 command with the @samp{-V} option.
4930 @node Submodel Options
4931 @section Hardware Models and Configurations
4932 @cindex submodel options
4933 @cindex specifying hardware config
4934 @cindex hardware models and configurations, specifying
4935 @cindex machine dependent options
4937 Earlier we discussed the standard option @samp{-b} which chooses among
4938 different installed compilers for completely different target
4939 machines, such as Vax vs. 68000 vs. 80386.
4941 In addition, each of these target machine types can have its own
4942 special options, starting with @samp{-m}, to choose among various
4943 hardware models or configurations---for example, 68010 vs 68020,
4944 floating coprocessor or none. A single installed version of the
4945 compiler can compile for any model or configuration, according to the
4948 Some configurations of the compiler also support additional special
4949 options, usually for compatibility with other compilers on the same
4953 These options are defined by the macro @code{TARGET_SWITCHES} in the
4954 machine description. The default for the options is also defined by
4955 that macro, which enables you to change the defaults.
4970 * RS/6000 and PowerPC Options::
4975 * Intel 960 Options::
4976 * DEC Alpha Options::
4980 * System V Options::
4981 * TMS320C3x/C4x Options::
4991 @node M680x0 Options
4992 @subsection M680x0 Options
4993 @cindex M680x0 options
4995 These are the @samp{-m} options defined for the 68000 series. The default
4996 values for these options depends on which style of 68000 was selected when
4997 the compiler was configured; the defaults for the most common choices are
5005 Generate output for a 68000. This is the default
5006 when the compiler is configured for 68000-based systems.
5008 Use this option for microcontrollers with a 68000 or EC000 core,
5009 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
5015 Generate output for a 68020. This is the default
5016 when the compiler is configured for 68020-based systems.
5020 Generate output containing 68881 instructions for floating point.
5021 This is the default for most 68020 systems unless @samp{-nfp} was
5022 specified when the compiler was configured.
5026 Generate output for a 68030. This is the default when the compiler is
5027 configured for 68030-based systems.
5031 Generate output for a 68040. This is the default when the compiler is
5032 configured for 68040-based systems.
5034 This option inhibits the use of 68881/68882 instructions that have to be
5035 emulated by software on the 68040. Use this option if your 68040 does not
5036 have code to emulate those instructions.
5040 Generate output for a 68060. This is the default when the compiler is
5041 configured for 68060-based systems.
5043 This option inhibits the use of 68020 and 68881/68882 instructions that
5044 have to be emulated by software on the 68060. Use this option if your 68060
5045 does not have code to emulate those instructions.
5049 Generate output for a CPU32. This is the default
5050 when the compiler is configured for CPU32-based systems.
5052 Use this option for microcontrollers with a
5053 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
5054 68336, 68340, 68341, 68349 and 68360.
5058 Generate output for a 520X "coldfire" family cpu. This is the default
5059 when the compiler is configured for 520X-based systems.
5061 Use this option for microcontroller with a 5200 core, including
5062 the MCF5202, MCF5203, MCF5204 and MCF5202.
5067 Generate output for a 68040, without using any of the new instructions.
5068 This results in code which can run relatively efficiently on either a
5069 68020/68881 or a 68030 or a 68040. The generated code does use the
5070 68881 instructions that are emulated on the 68040.
5074 Generate output for a 68060, without using any of the new instructions.
5075 This results in code which can run relatively efficiently on either a
5076 68020/68881 or a 68030 or a 68040. The generated code does use the
5077 68881 instructions that are emulated on the 68060.
5081 Generate output containing Sun FPA instructions for floating point.
5084 @opindex msoft-float
5085 Generate output containing library calls for floating point.
5086 @strong{Warning:} the requisite libraries are not available for all m68k
5087 targets. Normally the facilities of the machine's usual C compiler are
5088 used, but this can't be done directly in cross-compilation. You must
5089 make your own arrangements to provide suitable library functions for
5090 cross-compilation. The embedded targets @samp{m68k-*-aout} and
5091 @samp{m68k-*-coff} do provide software floating point support.
5095 Consider type @code{int} to be 16 bits wide, like @code{short int}.
5098 @opindex mnobitfield
5099 Do not use the bit-field instructions. The @samp{-m68000}, @samp{-mcpu32}
5100 and @samp{-m5200} options imply @w{@samp{-mnobitfield}}.
5104 Do use the bit-field instructions. The @samp{-m68020} option implies
5105 @samp{-mbitfield}. This is the default if you use a configuration
5106 designed for a 68020.
5110 Use a different function-calling convention, in which functions
5111 that take a fixed number of arguments return with the @code{rtd}
5112 instruction, which pops their arguments while returning. This
5113 saves one instruction in the caller since there is no need to pop
5114 the arguments there.
5116 This calling convention is incompatible with the one normally
5117 used on Unix, so you cannot use it if you need to call libraries
5118 compiled with the Unix compiler.
5120 Also, you must provide function prototypes for all functions that
5121 take variable numbers of arguments (including @code{printf});
5122 otherwise incorrect code will be generated for calls to those
5125 In addition, seriously incorrect code will result if you call a
5126 function with too many arguments. (Normally, extra arguments are
5127 harmlessly ignored.)
5129 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
5130 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
5133 @itemx -mno-align-int
5135 @opindex mno-align-int
5136 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
5137 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
5138 boundary (@samp{-malign-int}) or a 16-bit boundary (@samp{-mno-align-int}).
5139 Aligning variables on 32-bit boundaries produces code that runs somewhat
5140 faster on processors with 32-bit busses at the expense of more memory.
5142 @strong{Warning:} if you use the @samp{-malign-int} switch, GCC will
5143 align structures containing the above types differently than
5144 most published application binary interface specifications for the m68k.
5148 Use the pc-relative addressing mode of the 68000 directly, instead of
5149 using a global offset table. At present, this option implies -fpic,
5150 allowing at most a 16-bit offset for pc-relative addressing. -fPIC is
5151 not presently supported with -mpcrel, though this could be supported for
5152 68020 and higher processors.
5154 @item -mno-strict-align
5155 @itemx -mstrict-align
5156 @opindex mno-strict-align
5157 @opindex mstrict-align
5158 Do not (do) assume that unaligned memory references will be handled by
5163 @node M68hc1x Options
5164 @subsection M68hc1x Options
5165 @cindex M68hc1x options
5167 These are the @samp{-m} options defined for the 68hc11 and 68hc12
5168 microcontrollers. The default values for these options depends on
5169 which style of microcontroller was selected when the compiler was configured;
5170 the defaults for the most common choices are given below.
5177 Generate output for a 68HC11. This is the default
5178 when the compiler is configured for 68HC11-based systems.
5184 Generate output for a 68HC12. This is the default
5185 when the compiler is configured for 68HC12-based systems.
5188 @opindex mauto-incdec
5189 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
5194 Consider type @code{int} to be 16 bits wide, like @code{short int}.
5196 @item -msoft-reg-count=@var{count}
5197 @opindex msoft-reg-count
5198 Specify the number of pseudo-soft registers which are used for the
5199 code generation. The maximum number is 32. Using more pseudo-soft
5200 register may or may not result in better code depending on the program.
5201 The default is 4 for 68HC11 and 2 for 68HC12.
5206 @subsection VAX Options
5209 These @samp{-m} options are defined for the Vax:
5214 Do not output certain jump instructions (@code{aobleq} and so on)
5215 that the Unix assembler for the Vax cannot handle across long
5220 Do output those jump instructions, on the assumption that you
5221 will assemble with the GNU assembler.
5225 Output code for g-format floating point numbers instead of d-format.
5229 @subsection SPARC Options
5230 @cindex SPARC options
5232 These @samp{-m} switches are supported on the SPARC:
5237 @opindex mno-app-regs
5239 Specify @samp{-mapp-regs} to generate output using the global registers
5240 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
5243 To be fully SVR4 ABI compliant at the cost of some performance loss,
5244 specify @samp{-mno-app-regs}. You should compile libraries and system
5245 software with this option.
5250 @opindex mhard-float
5251 Generate output containing floating point instructions. This is the
5257 @opindex msoft-float
5258 Generate output containing library calls for floating point.
5259 @strong{Warning:} the requisite libraries are not available for all SPARC
5260 targets. Normally the facilities of the machine's usual C compiler are
5261 used, but this cannot be done directly in cross-compilation. You must make
5262 your own arrangements to provide suitable library functions for
5263 cross-compilation. The embedded targets @samp{sparc-*-aout} and
5264 @samp{sparclite-*-*} do provide software floating point support.
5266 @samp{-msoft-float} changes the calling convention in the output file;
5267 therefore, it is only useful if you compile @emph{all} of a program with
5268 this option. In particular, you need to compile @file{libgcc.a}, the
5269 library that comes with GCC, with @samp{-msoft-float} in order for
5272 @item -mhard-quad-float
5273 @opindex mhard-quad-float
5274 Generate output containing quad-word (long double) floating point
5277 @item -msoft-quad-float
5278 @opindex msoft-quad-float
5279 Generate output containing library calls for quad-word (long double)
5280 floating point instructions. The functions called are those specified
5281 in the SPARC ABI. This is the default.
5283 As of this writing, there are no sparc implementations that have hardware
5284 support for the quad-word floating point instructions. They all invoke
5285 a trap handler for one of these instructions, and then the trap handler
5286 emulates the effect of the instruction. Because of the trap handler overhead,
5287 this is much slower than calling the ABI library routines. Thus the
5288 @samp{-msoft-quad-float} option is the default.
5292 @opindex mno-epilogue
5294 With @samp{-mepilogue} (the default), the compiler always emits code for
5295 function exit at the end of each function. Any function exit in
5296 the middle of the function (such as a return statement in C) will
5297 generate a jump to the exit code at the end of the function.
5299 With @samp{-mno-epilogue}, the compiler tries to emit exit code inline
5300 at every function exit.
5306 With @samp{-mflat}, the compiler does not generate save/restore instructions
5307 and will use a "flat" or single register window calling convention.
5308 This model uses %i7 as the frame pointer and is compatible with the normal
5309 register window model. Code from either may be intermixed.
5310 The local registers and the input registers (0-5) are still treated as
5311 "call saved" registers and will be saved on the stack as necessary.
5313 With @samp{-mno-flat} (the default), the compiler emits save/restore
5314 instructions (except for leaf functions) and is the normal mode of operation.
5316 @item -mno-unaligned-doubles
5317 @itemx -munaligned-doubles
5318 @opindex mno-unaligned-doubles
5319 @opindex munaligned-doubles
5320 Assume that doubles have 8 byte alignment. This is the default.
5322 With @samp{-munaligned-doubles}, GCC assumes that doubles have 8 byte
5323 alignment only if they are contained in another type, or if they have an
5324 absolute address. Otherwise, it assumes they have 4 byte alignment.
5325 Specifying this option avoids some rare compatibility problems with code
5326 generated by other compilers. It is not the default because it results
5327 in a performance loss, especially for floating point code.
5329 @item -mno-faster-structs
5330 @itemx -mfaster-structs
5331 @opindex mno-faster-structs
5332 @opindex mfaster-structs
5333 With @samp{-mfaster-structs}, the compiler assumes that structures
5334 should have 8 byte alignment. This enables the use of pairs of
5335 @code{ldd} and @code{std} instructions for copies in structure
5336 assignment, in place of twice as many @code{ld} and @code{st} pairs.
5337 However, the use of this changed alignment directly violates the Sparc
5338 ABI. Thus, it's intended only for use on targets where the developer
5339 acknowledges that their resulting code will not be directly in line with
5340 the rules of the ABI.
5346 These two options select variations on the SPARC architecture.
5348 By default (unless specifically configured for the Fujitsu SPARClite),
5349 GCC generates code for the v7 variant of the SPARC architecture.
5351 @samp{-mv8} will give you SPARC v8 code. The only difference from v7
5352 code is that the compiler emits the integer multiply and integer
5353 divide instructions which exist in SPARC v8 but not in SPARC v7.
5355 @samp{-msparclite} will give you SPARClite code. This adds the integer
5356 multiply, integer divide step and scan (@code{ffs}) instructions which
5357 exist in SPARClite but not in SPARC v7.
5359 These options are deprecated and will be deleted in a future GCC release.
5360 They have been replaced with @samp{-mcpu=xxx}.
5365 @opindex msupersparc
5366 These two options select the processor for which the code is optimised.
5368 With @samp{-mcypress} (the default), the compiler optimizes code for the
5369 Cypress CY7C602 chip, as used in the SparcStation/SparcServer 3xx series.
5370 This is also appropriate for the older SparcStation 1, 2, IPX etc.
5372 With @samp{-msupersparc} the compiler optimizes code for the SuperSparc cpu, as
5373 used in the SparcStation 10, 1000 and 2000 series. This flag also enables use
5374 of the full SPARC v8 instruction set.
5376 These options are deprecated and will be deleted in a future GCC release.
5377 They have been replaced with @samp{-mcpu=xxx}.
5379 @item -mcpu=@var{cpu_type}
5381 Set the instruction set, register set, and instruction scheduling parameters
5382 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
5383 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
5384 @samp{hypersparc}, @samp{sparclite86x}, @samp{f930}, @samp{f934},
5385 @samp{sparclet}, @samp{tsc701}, @samp{v9}, and @samp{ultrasparc}.
5387 Default instruction scheduling parameters are used for values that select
5388 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
5389 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
5391 Here is a list of each supported architecture and their supported
5396 v8: supersparc, hypersparc
5397 sparclite: f930, f934, sparclite86x
5402 @item -mtune=@var{cpu_type}
5404 Set the instruction scheduling parameters for machine type
5405 @var{cpu_type}, but do not set the instruction set or register set that the
5406 option @samp{-mcpu=}@var{cpu_type} would.
5408 The same values for @samp{-mcpu=}@var{cpu_type} are used for
5409 @samp{-mtune=}@*@var{cpu_type}, though the only useful values are those that
5410 select a particular cpu implementation: @samp{cypress}, @samp{supersparc},
5411 @samp{hypersparc}, @samp{f930}, @samp{f934}, @samp{sparclite86x},
5412 @samp{tsc701}, @samp{ultrasparc}.
5416 These @samp{-m} switches are supported in addition to the above
5417 on the SPARCLET processor.
5420 @item -mlittle-endian
5421 @opindex mlittle-endian
5422 Generate code for a processor running in little-endian mode.
5426 Treat register @code{%g0} as a normal register.
5427 GCC will continue to clobber it as necessary but will not assume
5428 it always reads as 0.
5430 @item -mbroken-saverestore
5431 @opindex mbroken-saverestore
5432 Generate code that does not use non-trivial forms of the @code{save} and
5433 @code{restore} instructions. Early versions of the SPARCLET processor do
5434 not correctly handle @code{save} and @code{restore} instructions used with
5435 arguments. They correctly handle them used without arguments. A @code{save}
5436 instruction used without arguments increments the current window pointer
5437 but does not allocate a new stack frame. It is assumed that the window
5438 overflow trap handler will properly handle this case as will interrupt
5442 These @samp{-m} switches are supported in addition to the above
5443 on SPARC V9 processors in 64-bit environments.
5446 @item -mlittle-endian
5447 @opindex mlittle-endian
5448 Generate code for a processor running in little-endian mode.
5454 Generate code for a 32-bit or 64-bit environment.
5455 The 32-bit environment sets int, long and pointer to 32 bits.
5456 The 64-bit environment sets int to 32 bits and long and pointer
5459 @item -mcmodel=medlow
5460 @opindex mcmodel=medlow
5461 Generate code for the Medium/Low code model: the program must be linked
5462 in the low 32 bits of the address space. Pointers are 64 bits.
5463 Programs can be statically or dynamically linked.
5465 @item -mcmodel=medmid
5466 @opindex mcmodel=medmid
5467 Generate code for the Medium/Middle code model: the program must be linked
5468 in the low 44 bits of the address space, the text segment must be less than
5469 2G bytes, and data segment must be within 2G of the text segment.
5470 Pointers are 64 bits.
5472 @item -mcmodel=medany
5473 @opindex mcmodel=medany
5474 Generate code for the Medium/Anywhere code model: the program may be linked
5475 anywhere in the address space, the text segment must be less than
5476 2G bytes, and data segment must be within 2G of the text segment.
5477 Pointers are 64 bits.
5479 @item -mcmodel=embmedany
5480 @opindex mcmodel=embmedany
5481 Generate code for the Medium/Anywhere code model for embedded systems:
5482 assume a 32-bit text and a 32-bit data segment, both starting anywhere
5483 (determined at link time). Register %g4 points to the base of the
5484 data segment. Pointers are still 64 bits.
5485 Programs are statically linked, PIC is not supported.
5488 @itemx -mno-stack-bias
5489 @opindex mstack-bias
5490 @opindex mno-stack-bias
5491 With @samp{-mstack-bias}, GCC assumes that the stack pointer, and
5492 frame pointer if present, are offset by -2047 which must be added back
5493 when making stack frame references.
5494 Otherwise, assume no such offset is present.
5497 @node Convex Options
5498 @subsection Convex Options
5499 @cindex Convex options
5501 These @samp{-m} options are defined for Convex:
5506 Generate output for C1. The code will run on any Convex machine.
5507 The preprocessor symbol @code{__convex__c1__} is defined.
5511 Generate output for C2. Uses instructions not available on C1.
5512 Scheduling and other optimizations are chosen for max performance on C2.
5513 The preprocessor symbol @code{__convex_c2__} is defined.
5517 Generate output for C32xx. Uses instructions not available on C1.
5518 Scheduling and other optimizations are chosen for max performance on C32.
5519 The preprocessor symbol @code{__convex_c32__} is defined.
5523 Generate output for C34xx. Uses instructions not available on C1.
5524 Scheduling and other optimizations are chosen for max performance on C34.
5525 The preprocessor symbol @code{__convex_c34__} is defined.
5529 Generate output for C38xx. Uses instructions not available on C1.
5530 Scheduling and other optimizations are chosen for max performance on C38.
5531 The preprocessor symbol @code{__convex_c38__} is defined.
5535 Generate code which puts an argument count in the word preceding each
5536 argument list. This is compatible with regular CC, and a few programs
5537 may need the argument count word. GDB and other source-level debuggers
5538 do not need it; this info is in the symbol table.
5541 @opindex mnoargcount
5542 Omit the argument count word. This is the default.
5544 @item -mvolatile-cache
5545 @opindex mvolatile-cache
5546 Allow volatile references to be cached. This is the default.
5548 @item -mvolatile-nocache
5549 @opindex mvolatile-nocache
5550 Volatile references bypass the data cache, going all the way to memory.
5551 This is only needed for multi-processor code that does not use standard
5552 synchronization instructions. Making non-volatile references to volatile
5553 locations will not necessarily work.
5557 Type long is 32 bits, the same as type int. This is the default.
5561 Type long is 64 bits, the same as type long long. This option is useless,
5562 because no library support exists for it.
5565 @node AMD29K Options
5566 @subsection AMD29K Options
5567 @cindex AMD29K options
5569 These @samp{-m} options are defined for the AMD Am29000:
5574 @cindex DW bit (29k)
5575 Generate code that assumes the @code{DW} bit is set, i.e., that byte and
5576 halfword operations are directly supported by the hardware. This is the
5581 Generate code that assumes the @code{DW} bit is not set.
5585 @cindex byte writes (29k)
5586 Generate code that assumes the system supports byte and halfword write
5587 operations. This is the default.
5591 Generate code that assumes the systems does not support byte and
5592 halfword write operations. @samp{-mnbw} implies @samp{-mndw}.
5596 @cindex memory model (29k)
5597 Use a small memory model that assumes that all function addresses are
5598 either within a single 256 KB segment or at an absolute address of less
5599 than 256k. This allows the @code{call} instruction to be used instead
5600 of a @code{const}, @code{consth}, @code{calli} sequence.
5604 Use the normal memory model: Generate @code{call} instructions only when
5605 calling functions in the same file and @code{calli} instructions
5606 otherwise. This works if each file occupies less than 256 KB but allows
5607 the entire executable to be larger than 256 KB. This is the default.
5611 Always use @code{calli} instructions. Specify this option if you expect
5612 a single file to compile into more than 256 KB of code.
5616 @cindex processor selection (29k)
5617 Generate code for the Am29050.
5621 Generate code for the Am29000. This is the default.
5623 @item -mkernel-registers
5624 @opindex mkernel-registers
5625 @cindex kernel and user registers (29k)
5626 Generate references to registers @code{gr64-gr95} instead of to
5627 registers @code{gr96-gr127}. This option can be used when compiling
5628 kernel code that wants a set of global registers disjoint from that used
5631 Note that when this option is used, register names in @samp{-f} flags
5632 must use the normal, user-mode, names.
5634 @item -muser-registers
5635 @opindex muser-registers
5636 Use the normal set of global registers, @code{gr96-gr127}. This is the
5640 @itemx -mno-stack-check
5641 @opindex mstack-check
5642 @opindex mno-stack-check
5643 @cindex stack checks (29k)
5644 Insert (or do not insert) a call to @code{__msp_check} after each stack
5645 adjustment. This is often used for kernel code.
5648 @itemx -mno-storem-bug
5649 @opindex mstorem-bug
5650 @opindex mno-storem-bug
5651 @cindex storem bug (29k)
5652 @samp{-mstorem-bug} handles 29k processors which cannot handle the
5653 separation of a mtsrim insn and a storem instruction (most 29000 chips
5654 to date, but not the 29050).
5656 @item -mno-reuse-arg-regs
5657 @itemx -mreuse-arg-regs
5658 @opindex mno-reuse-arg-regs
5659 @opindex mreuse-arg-regs
5660 @samp{-mno-reuse-arg-regs} tells the compiler to only use incoming argument
5661 registers for copying out arguments. This helps detect calling a function
5662 with fewer arguments than it was declared with.
5664 @item -mno-impure-text
5665 @itemx -mimpure-text
5666 @opindex mno-impure-text
5667 @opindex mimpure-text
5668 @samp{-mimpure-text}, used in addition to @samp{-shared}, tells the compiler to
5669 not pass @samp{-assert pure-text} to the linker when linking a shared object.
5672 @opindex msoft-float
5673 Generate output containing library calls for floating point.
5674 @strong{Warning:} the requisite libraries are not part of GCC.
5675 Normally the facilities of the machine's usual C compiler are used, but
5676 this can't be done directly in cross-compilation. You must make your
5677 own arrangements to provide suitable library functions for
5682 Do not generate multm or multmu instructions. This is useful for some embedded
5683 systems which do not have trap handlers for these instructions.
5687 @subsection ARM Options
5690 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
5695 @opindex mapcs-frame
5696 Generate a stack frame that is compliant with the ARM Procedure Call
5697 Standard for all functions, even if this is not strictly necessary for
5698 correct execution of the code. Specifying @samp{-fomit-frame-pointer}
5699 with this option will cause the stack frames not to be generated for
5700 leaf functions. The default is @samp{-mno-apcs-frame}.
5704 This is a synonym for @samp{-mapcs-frame}.
5708 Generate code for a processor running with a 26-bit program counter,
5709 and conforming to the function calling standards for the APCS 26-bit
5710 option. This option replaces the @samp{-m2} and @samp{-m3} options
5711 of previous releases of the compiler.
5715 Generate code for a processor running with a 32-bit program counter,
5716 and conforming to the function calling standards for the APCS 32-bit
5717 option. This option replaces the @samp{-m6} option of previous releases
5721 @c not currently implemented
5722 @item -mapcs-stack-check
5723 @opindex mapcs-stack-check
5724 Generate code to check the amount of stack space available upon entry to
5725 every function (that actually uses some stack space). If there is
5726 insufficient space available then either the function
5727 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
5728 called, depending upon the amount of stack space required. The run time
5729 system is required to provide these functions. The default is
5730 @samp{-mno-apcs-stack-check}, since this produces smaller code.
5732 @c not currently implemented
5734 @opindex mapcs-float
5735 Pass floating point arguments using the float point registers. This is
5736 one of the variants of the APCS. This option is recommended if the
5737 target hardware has a floating point unit or if a lot of floating point
5738 arithmetic is going to be performed by the code. The default is
5739 @samp{-mno-apcs-float}, since integer only code is slightly increased in
5740 size if @samp{-mapcs-float} is used.
5742 @c not currently implemented
5743 @item -mapcs-reentrant
5744 @opindex mapcs-reentrant
5745 Generate reentrant, position independent code. The default is
5746 @samp{-mno-apcs-reentrant}.
5749 @item -mthumb-interwork
5750 @opindex mthumb-interwork
5751 Generate code which supports calling between the ARM and Thumb
5752 instruction sets. Without this option the two instruction sets cannot
5753 be reliably used inside one program. The default is
5754 @samp{-mno-thumb-interwork}, since slightly larger code is generated
5755 when @samp{-mthumb-interwork} is specified.
5757 @item -mno-sched-prolog
5758 @opindex mno-sched-prolog
5759 Prevent the reordering of instructions in the function prolog, or the
5760 merging of those instruction with the instructions in the function's
5761 body. This means that all functions will start with a recognizable set
5762 of instructions (or in fact one of a choice from a small set of
5763 different function prologues), and this information can be used to
5764 locate the start if functions inside an executable piece of code. The
5765 default is @samp{-msched-prolog}.
5768 @opindex mhard-float
5769 Generate output containing floating point instructions. This is the
5773 @opindex msoft-float
5774 Generate output containing library calls for floating point.
5775 @strong{Warning:} the requisite libraries are not available for all ARM
5776 targets. Normally the facilities of the machine's usual C compiler are
5777 used, but this cannot be done directly in cross-compilation. You must make
5778 your own arrangements to provide suitable library functions for
5781 @samp{-msoft-float} changes the calling convention in the output file;
5782 therefore, it is only useful if you compile @emph{all} of a program with
5783 this option. In particular, you need to compile @file{libgcc.a}, the
5784 library that comes with GCC, with @samp{-msoft-float} in order for
5787 @item -mlittle-endian
5788 @opindex mlittle-endian
5789 Generate code for a processor running in little-endian mode. This is
5790 the default for all standard configurations.
5793 @opindex mbig-endian
5794 Generate code for a processor running in big-endian mode; the default is
5795 to compile code for a little-endian processor.
5797 @item -mwords-little-endian
5798 @opindex mwords-little-endian
5799 This option only applies when generating code for big-endian processors.
5800 Generate code for a little-endian word order but a big-endian byte
5801 order. That is, a byte order of the form @samp{32107654}. Note: this
5802 option should only be used if you require compatibility with code for
5803 big-endian ARM processors generated by versions of the compiler prior to
5806 @item -malignment-traps
5807 @opindex malignment-traps
5808 Generate code that will not trap if the MMU has alignment traps enabled.
5809 On ARM architectures prior to ARMv4, there were no instructions to
5810 access half-word objects stored in memory. However, when reading from
5811 memory a feature of the ARM architecture allows a word load to be used,
5812 even if the address is unaligned, and the processor core will rotate the
5813 data as it is being loaded. This option tells the compiler that such
5814 misaligned accesses will cause a MMU trap and that it should instead
5815 synthesise the access as a series of byte accesses. The compiler can
5816 still use word accesses to load half-word data if it knows that the
5817 address is aligned to a word boundary.
5819 This option is ignored when compiling for ARM architecture 4 or later,
5820 since these processors have instructions to directly access half-word
5823 @item -mno-alignment-traps
5824 @opindex mno-alignment-traps
5825 Generate code that assumes that the MMU will not trap unaligned
5826 accesses. This produces better code when the target instruction set
5827 does not have half-word memory operations (i.e. implementations prior to
5830 Note that you cannot use this option to access unaligned word objects,
5831 since the processor will only fetch one 32-bit aligned object from
5834 The default setting for most targets is -mno-alignment-traps, since
5835 this produces better code when there are no half-word memory
5836 instructions available.
5838 @item -mshort-load-bytes
5839 @itemx -mno-short-load-words
5840 @opindex mshort-load-bytes
5841 @opindex mno-short-load-words
5842 These are deprecated aliases for @samp{-malignment-traps}.
5844 @item -mno-short-load-bytes
5845 @itemx -mshort-load-words
5846 @opindex mno-short-load-bytes
5847 @opindex mshort-load-words
5848 This are deprecated aliases for @samp{-mno-alignment-traps}.
5852 This option only applies to RISC iX. Emulate the native BSD-mode
5853 compiler. This is the default if @samp{-ansi} is not specified.
5857 This option only applies to RISC iX. Emulate the native X/Open-mode
5860 @item -mno-symrename
5861 @opindex mno-symrename
5862 This option only applies to RISC iX. Do not run the assembler
5863 post-processor, @samp{symrename}, after code has been assembled.
5864 Normally it is necessary to modify some of the standard symbols in
5865 preparation for linking with the RISC iX C library; this option
5866 suppresses this pass. The post-processor is never run when the
5867 compiler is built for cross-compilation.
5869 @item -mcpu=@var{name}
5871 This specifies the name of the target ARM processor. GCC uses this name
5872 to determine what kind of instructions it can emit when generating
5873 assembly code. Permissible names are: arm2, arm250, arm3, arm6, arm60,
5874 arm600, arm610, arm620, arm7, arm7m, arm7d, arm7dm, arm7di, arm7dmi,
5875 arm70, arm700, arm700i, arm710, arm710c, arm7100, arm7500, arm7500fe,
5876 arm7tdmi, arm8, strongarm, strongarm110, strongarm1100, arm8, arm810,
5877 arm9, arm9e, arm920, arm920t, arm940t, arm9tdmi, arm10tdmi, arm1020t,
5880 @itemx -mtune=@var{name}
5882 This option is very similar to the @samp{-mcpu=} option, except that
5883 instead of specifying the actual target processor type, and hence
5884 restricting which instructions can be used, it specifies that GCC should
5885 tune the performance of the code as if the target were of the type
5886 specified in this option, but still choosing the instructions that it
5887 will generate based on the cpu specified by a @samp{-mcpu=} option.
5888 For some ARM implementations better performance can be obtained by using
5891 @item -march=@var{name}
5893 This specifies the name of the target ARM architecture. GCC uses this
5894 name to determine what kind of instructions it can emit when generating
5895 assembly code. This option can be used in conjunction with or instead
5896 of the @samp{-mcpu=} option. Permissible names are: armv2, armv2a,
5897 armv3, armv3m, armv4, armv4t, armv5, armv5t, armv5te.
5899 @item -mfpe=@var{number}
5900 @itemx -mfp=@var{number}
5903 This specifies the version of the floating point emulation available on
5904 the target. Permissible values are 2 and 3. @samp{-mfp=} is a synonym
5905 for @samp{-mfpe=}, for compatibility with older versions of GCC.
5907 @item -mstructure-size-boundary=@var{n}
5908 @opindex mstructure-size-boundary
5909 The size of all structures and unions will be rounded up to a multiple
5910 of the number of bits set by this option. Permissible values are 8 and
5911 32. The default value varies for different toolchains. For the COFF
5912 targeted toolchain the default value is 8. Specifying the larger number
5913 can produce faster, more efficient code, but can also increase the size
5914 of the program. The two values are potentially incompatible. Code
5915 compiled with one value cannot necessarily expect to work with code or
5916 libraries compiled with the other value, if they exchange information
5917 using structures or unions.
5919 @item -mabort-on-noreturn
5920 @opindex mabort-on-noreturn
5921 Generate a call to the function @code{abort} at the end of a
5922 @code{noreturn} function. It will be executed if the function tries to
5926 @itemx -mno-long-calls
5927 @opindex mlong-calls
5928 @opindex mno-long-calls
5929 Tells the compiler to perform function calls by first loading the
5930 address of the function into a register and then performing a subroutine
5931 call on this register. This switch is needed if the target function
5932 will lie outside of the 64 megabyte addressing range of the offset based
5933 version of subroutine call instruction.
5935 Even if this switch is enabled, not all function calls will be turned
5936 into long calls. The heuristic is that static functions, functions
5937 which have the @samp{short-call} attribute, functions that are inside
5938 the scope of a @samp{#pragma no_long_calls} directive and functions whose
5939 definitions have already been compiled within the current compilation
5940 unit, will not be turned into long calls. The exception to this rule is
5941 that weak function definitions, functions with the @samp{long-call}
5942 attribute or the @samp{section} attribute, and functions that are within
5943 the scope of a @samp{#pragma long_calls} directive, will always be
5944 turned into long calls.
5946 This feature is not enabled by default. Specifying
5947 @samp{-mno-long-calls} will restore the default behaviour, as will
5948 placing the function calls within the scope of a @samp{#pragma
5949 long_calls_off} directive. Note these switches have no effect on how
5950 the compiler generates code to handle function calls via function
5953 @item -mnop-fun-dllimport
5954 @opindex mnop-fun-dllimport
5955 Disable support for the @emph{dllimport} attribute.
5957 @item -msingle-pic-base
5958 @opindex msingle-pic-base
5959 Treat the register used for PIC addressing as read-only, rather than
5960 loading it in the prologue for each function. The run-time system is
5961 responsible for initialising this register with an appropriate value
5962 before execution begins.
5964 @item -mpic-register=@var{reg}
5965 @opindex mpic-register
5966 Specify the register to be used for PIC addressing. The default is R10
5967 unless stack-checking is enabled, when R9 is used.
5969 @item -mpoke-function-name
5970 @opindex mpoke-function-name
5971 Write the name of each function into the text section, directly
5972 preceding the function prologue. The generated code is similar to this:
5976 .ascii "arm_poke_function_name", 0
5979 .word 0xff000000 + (t1 - t0)
5980 arm_poke_function_name
5982 stmfd sp!, @{fp, ip, lr, pc@}
5986 When performing a stack backtrace, code can inspect the value of
5987 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
5988 location @code{pc - 12} and the top 8 bits are set, then we know that
5989 there is a function name embedded immediately preceding this location
5990 and has length @code{((pc[-3]) & 0xff000000)}.
5994 Generate code for the 16-bit Thumb instruction set. The default is to
5995 use the 32-bit ARM instruction set.
5998 @opindex mtpcs-frame
5999 Generate a stack frame that is compliant with the Thumb Procedure Call
6000 Standard for all non-leaf functions. (A leaf function is one that does
6001 not call any other functions.) The default is @samp{-mno-tpcs-frame}.
6003 @item -mtpcs-leaf-frame
6004 @opindex mtpcs-leaf-frame
6005 Generate a stack frame that is compliant with the Thumb Procedure Call
6006 Standard for all leaf functions. (A leaf function is one that does
6007 not call any other functions.) The default is @samp{-mno-apcs-leaf-frame}.
6009 @item -mcallee-super-interworking
6010 @opindex mcallee-super-interworking
6011 Gives all externally visible functions in the file being compiled an ARM
6012 instruction set header which switches to Thumb mode before executing the
6013 rest of the function. This allows these functions to be called from
6014 non-interworking code.
6016 @item -mcaller-super-interworking
6017 @opindex mcaller-super-interworking
6018 Allows calls via function pointers (including virtual functions) to
6019 execute correctly regardless of whether the target code has been
6020 compiled for interworking or not. There is a small overhead in the cost
6021 of executing a function pointer if this option is enabled.
6025 @node MN10200 Options
6026 @subsection MN10200 Options
6027 @cindex MN10200 options
6028 These @samp{-m} options are defined for Matsushita MN10200 architectures:
6033 Indicate to the linker that it should perform a relaxation optimization pass
6034 to shorten branches, calls and absolute memory addresses. This option only
6035 has an effect when used on the command line for the final link step.
6037 This option makes symbolic debugging impossible.
6040 @node MN10300 Options
6041 @subsection MN10300 Options
6042 @cindex MN10300 options
6043 These @samp{-m} options are defined for Matsushita MN10300 architectures:
6048 Generate code to avoid bugs in the multiply instructions for the MN10300
6049 processors. This is the default.
6052 @opindex mno-mult-bug
6053 Do not generate code to avoid bugs in the multiply instructions for the
6058 Generate code which uses features specific to the AM33 processor.
6062 Do not generate code which uses features specific to the AM33 processor. This
6067 Indicate to the linker that it should perform a relaxation optimization pass
6068 to shorten branches, calls and absolute memory addresses. This option only
6069 has an effect when used on the command line for the final link step.
6071 This option makes symbolic debugging impossible.
6075 @node M32R/D Options
6076 @subsection M32R/D Options
6077 @cindex M32R/D options
6079 These @samp{-m} options are defined for Mitsubishi M32R/D architectures:
6082 @item -mcode-model=small
6083 @opindex mcode-model=small
6084 Assume all objects live in the lower 16MB of memory (so that their addresses
6085 can be loaded with the @code{ld24} instruction), and assume all subroutines
6086 are reachable with the @code{bl} instruction.
6087 This is the default.
6089 The addressability of a particular object can be set with the
6090 @code{model} attribute.
6092 @item -mcode-model=medium
6093 @opindex mcode-model=medium
6094 Assume objects may be anywhere in the 32-bit address space (the compiler
6095 will generate @code{seth/add3} instructions to load their addresses), and
6096 assume all subroutines are reachable with the @code{bl} instruction.
6098 @item -mcode-model=large
6099 @opindex mcode-model=large
6100 Assume objects may be anywhere in the 32-bit address space (the compiler
6101 will generate @code{seth/add3} instructions to load their addresses), and
6102 assume subroutines may not be reachable with the @code{bl} instruction
6103 (the compiler will generate the much slower @code{seth/add3/jl}
6104 instruction sequence).
6107 @opindex msdata=none
6108 Disable use of the small data area. Variables will be put into
6109 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
6110 @code{section} attribute has been specified).
6111 This is the default.
6113 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
6114 Objects may be explicitly put in the small data area with the
6115 @code{section} attribute using one of these sections.
6118 @opindex msdata=sdata
6119 Put small global and static data in the small data area, but do not
6120 generate special code to reference them.
6124 Put small global and static data in the small data area, and generate
6125 special instructions to reference them.
6129 @cindex smaller data references
6130 Put global and static objects less than or equal to @var{num} bytes
6131 into the small data or bss sections instead of the normal data or bss
6132 sections. The default value of @var{num} is 8.
6133 The @samp{-msdata} option must be set to one of @samp{sdata} or @samp{use}
6134 for this option to have any effect.
6136 All modules should be compiled with the same @samp{-G @var{num}} value.
6137 Compiling with different values of @var{num} may or may not work; if it
6138 doesn't the linker will give an error message - incorrect code will not be
6144 @subsection M88K Options
6145 @cindex M88k options
6147 These @samp{-m} options are defined for Motorola 88k architectures:
6152 Generate code that works well on both the m88100 and the
6157 Generate code that works best for the m88100, but that also
6162 Generate code that works best for the m88110, and may not run
6167 Obsolete option to be removed from the next revision.
6170 @item -midentify-revision
6171 @opindex midentify-revision
6172 @cindex identifying source, compiler (88k)
6173 Include an @code{ident} directive in the assembler output recording the
6174 source file name, compiler name and version, timestamp, and compilation
6177 @item -mno-underscores
6178 @opindex mno-underscores
6179 @cindex underscores, avoiding (88k)
6180 In assembler output, emit symbol names without adding an underscore
6181 character at the beginning of each name. The default is to use an
6182 underscore as prefix on each name.
6184 @item -mocs-debug-info
6185 @itemx -mno-ocs-debug-info
6186 @opindex mocs-debug-info
6187 @opindex mno-ocs-debug-info
6189 @cindex debugging, 88k OCS
6190 Include (or omit) additional debugging information (about registers used
6191 in each stack frame) as specified in the 88open Object Compatibility
6192 Standard, ``OCS''. This extra information allows debugging of code that
6193 has had the frame pointer eliminated. The default for DG/UX, SVr4, and
6194 Delta 88 SVr3.2 is to include this information; other 88k configurations
6195 omit this information by default.
6197 @item -mocs-frame-position
6198 @opindex mocs-frame-position
6199 @cindex register positions in frame (88k)
6200 When emitting COFF debugging information for automatic variables and
6201 parameters stored on the stack, use the offset from the canonical frame
6202 address, which is the stack pointer (register 31) on entry to the
6203 function. The DG/UX, SVr4, Delta88 SVr3.2, and BCS configurations use
6204 @samp{-mocs-frame-position}; other 88k configurations have the default
6205 @samp{-mno-ocs-frame-position}.
6207 @item -mno-ocs-frame-position
6208 @opindex mno-ocs-frame-position
6209 @cindex register positions in frame (88k)
6210 When emitting COFF debugging information for automatic variables and
6211 parameters stored on the stack, use the offset from the frame pointer
6212 register (register 30). When this option is in effect, the frame
6213 pointer is not eliminated when debugging information is selected by the
6216 @item -moptimize-arg-area
6217 @itemx -mno-optimize-arg-area
6218 @opindex moptimize-arg-area
6219 @opindex mno-optimize-arg-area
6220 @cindex arguments in frame (88k)
6221 Control how function arguments are stored in stack frames.
6222 @samp{-moptimize-arg-area} saves space by optimizing them, but this
6223 conflicts with the 88open specifications. The opposite alternative,
6224 @samp{-mno-optimize-arg-area}, agrees with 88open standards. By default
6225 GCC does not optimize the argument area.
6227 @item -mshort-data-@var{num}
6228 @opindex mshort-data
6229 @cindex smaller data references (88k)
6230 @cindex r0-relative references (88k)
6231 Generate smaller data references by making them relative to @code{r0},
6232 which allows loading a value using a single instruction (rather than the
6233 usual two). You control which data references are affected by
6234 specifying @var{num} with this option. For example, if you specify
6235 @samp{-mshort-data-512}, then the data references affected are those
6236 involving displacements of less than 512 bytes.
6237 @samp{-mshort-data-@var{num}} is not effective for @var{num} greater
6240 @item -mserialize-volatile
6241 @opindex mserialize-volatile
6242 @itemx -mno-serialize-volatile
6243 @opindex mno-serialize-volatile
6244 @cindex sequential consistency on 88k
6245 Do, or don't, generate code to guarantee sequential consistency
6246 of volatile memory references. By default, consistency is
6249 The order of memory references made by the MC88110 processor does
6250 not always match the order of the instructions requesting those
6251 references. In particular, a load instruction may execute before
6252 a preceding store instruction. Such reordering violates
6253 sequential consistency of volatile memory references, when there
6254 are multiple processors. When consistency must be guaranteed,
6255 GNU C generates special instructions, as needed, to force
6256 execution in the proper order.
6258 The MC88100 processor does not reorder memory references and so
6259 always provides sequential consistency. However, by default, GNU
6260 C generates the special instructions to guarantee consistency
6261 even when you use @samp{-m88100}, so that the code may be run on an
6262 MC88110 processor. If you intend to run your code only on the
6263 MC88100 processor, you may use @samp{-mno-serialize-volatile}.
6265 The extra code generated to guarantee consistency may affect the
6266 performance of your application. If you know that you can safely
6267 forgo this guarantee, you may use @samp{-mno-serialize-volatile}.
6273 @cindex assembler syntax, 88k
6275 Turn on (@samp{-msvr4}) or off (@samp{-msvr3}) compiler extensions
6276 related to System V release 4 (SVr4). This controls the following:
6280 Which variant of the assembler syntax to emit.
6282 @samp{-msvr4} makes the C preprocessor recognize @samp{#pragma weak}
6283 that is used on System V release 4.
6285 @samp{-msvr4} makes GCC issue additional declaration directives used in
6289 @samp{-msvr4} is the default for the m88k-motorola-sysv4 and
6290 m88k-dg-dgux m88k configurations. @samp{-msvr3} is the default for all
6291 other m88k configurations.
6293 @item -mversion-03.00
6294 @opindex mversion-03.00
6295 This option is obsolete, and is ignored.
6296 @c ??? which asm syntax better for GAS? option there too?
6298 @item -mno-check-zero-division
6299 @itemx -mcheck-zero-division
6300 @opindex mno-check-zero-division
6301 @opindex mcheck-zero-division
6302 @cindex zero division on 88k
6303 Do, or don't, generate code to guarantee that integer division by
6304 zero will be detected. By default, detection is guaranteed.
6306 Some models of the MC88100 processor fail to trap upon integer
6307 division by zero under certain conditions. By default, when
6308 compiling code that might be run on such a processor, GNU C
6309 generates code that explicitly checks for zero-valued divisors
6310 and traps with exception number 503 when one is detected. Use of
6311 mno-check-zero-division suppresses such checking for code
6312 generated to run on an MC88100 processor.
6314 GNU C assumes that the MC88110 processor correctly detects all
6315 instances of integer division by zero. When @samp{-m88110} is
6316 specified, both @samp{-mcheck-zero-division} and
6317 @samp{-mno-check-zero-division} are ignored, and no explicit checks for
6318 zero-valued divisors are generated.
6320 @item -muse-div-instruction
6321 @opindex muse-div-instruction
6322 @cindex divide instruction, 88k
6323 Use the div instruction for signed integer division on the
6324 MC88100 processor. By default, the div instruction is not used.
6326 On the MC88100 processor the signed integer division instruction
6327 div) traps to the operating system on a negative operand. The
6328 operating system transparently completes the operation, but at a
6329 large cost in execution time. By default, when compiling code
6330 that might be run on an MC88100 processor, GNU C emulates signed
6331 integer division using the unsigned integer division instruction
6332 divu), thereby avoiding the large penalty of a trap to the
6333 operating system. Such emulation has its own, smaller, execution
6334 cost in both time and space. To the extent that your code's
6335 important signed integer division operations are performed on two
6336 nonnegative operands, it may be desirable to use the div
6337 instruction directly.
6339 On the MC88110 processor the div instruction (also known as the
6340 divs instruction) processes negative operands without trapping to
6341 the operating system. When @samp{-m88110} is specified,
6342 @samp{-muse-div-instruction} is ignored, and the div instruction is used
6343 for signed integer division.
6345 Note that the result of dividing INT_MIN by -1 is undefined. In
6346 particular, the behavior of such a division with and without
6347 @samp{-muse-div-instruction} may differ.
6349 @item -mtrap-large-shift
6350 @itemx -mhandle-large-shift
6351 @opindex mtrap-large-shift
6352 @opindex mhandle-large-shift
6353 @cindex bit shift overflow (88k)
6354 @cindex large bit shifts (88k)
6355 Include code to detect bit-shifts of more than 31 bits; respectively,
6356 trap such shifts or emit code to handle them properly. By default GCC
6357 makes no special provision for large bit shifts.
6359 @item -mwarn-passed-structs
6360 @opindex mwarn-passed-structs
6361 @cindex structure passing (88k)
6362 Warn when a function passes a struct as an argument or result.
6363 Structure-passing conventions have changed during the evolution of the C
6364 language, and are often the source of portability problems. By default,
6365 GCC issues no such warning.
6368 @node RS/6000 and PowerPC Options
6369 @subsection IBM RS/6000 and PowerPC Options
6370 @cindex RS/6000 and PowerPC Options
6371 @cindex IBM RS/6000 and PowerPC Options
6373 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
6381 @itemx -mpowerpc-gpopt
6382 @itemx -mno-powerpc-gpopt
6383 @itemx -mpowerpc-gfxopt
6384 @itemx -mno-powerpc-gfxopt
6386 @itemx -mno-powerpc64
6392 @opindex mno-powerpc
6393 @opindex mpowerpc-gpopt
6394 @opindex mno-powerpc-gpopt
6395 @opindex mpowerpc-gfxopt
6396 @opindex mno-powerpc-gfxopt
6398 @opindex mno-powerpc64
6399 GCC supports two related instruction set architectures for the
6400 RS/6000 and PowerPC. The @dfn{POWER} instruction set are those
6401 instructions supported by the @samp{rios} chip set used in the original
6402 RS/6000 systems and the @dfn{PowerPC} instruction set is the
6403 architecture of the Motorola MPC5xx, MPC6xx, MPC8xx microprocessors, and
6404 the IBM 4xx microprocessors.
6406 Neither architecture is a subset of the other. However there is a
6407 large common subset of instructions supported by both. An MQ
6408 register is included in processors supporting the POWER architecture.
6410 You use these options to specify which instructions are available on the
6411 processor you are using. The default value of these options is
6412 determined when configuring GCC. Specifying the
6413 @samp{-mcpu=@var{cpu_type}} overrides the specification of these
6414 options. We recommend you use the @samp{-mcpu=@var{cpu_type}} option
6415 rather than the options listed above.
6417 The @samp{-mpower} option allows GCC to generate instructions that
6418 are found only in the POWER architecture and to use the MQ register.
6419 Specifying @samp{-mpower2} implies @samp{-power} and also allows GCC
6420 to generate instructions that are present in the POWER2 architecture but
6421 not the original POWER architecture.
6423 The @samp{-mpowerpc} option allows GCC to generate instructions that
6424 are found only in the 32-bit subset of the PowerPC architecture.
6425 Specifying @samp{-mpowerpc-gpopt} implies @samp{-mpowerpc} and also allows
6426 GCC to use the optional PowerPC architecture instructions in the
6427 General Purpose group, including floating-point square root. Specifying
6428 @samp{-mpowerpc-gfxopt} implies @samp{-mpowerpc} and also allows GCC to
6429 use the optional PowerPC architecture instructions in the Graphics
6430 group, including floating-point select.
6432 The @samp{-mpowerpc64} option allows GCC to generate the additional
6433 64-bit instructions that are found in the full PowerPC64 architecture
6434 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
6435 @samp{-mno-powerpc64}.
6437 If you specify both @samp{-mno-power} and @samp{-mno-powerpc}, GCC
6438 will use only the instructions in the common subset of both
6439 architectures plus some special AIX common-mode calls, and will not use
6440 the MQ register. Specifying both @samp{-mpower} and @samp{-mpowerpc}
6441 permits GCC to use any instruction from either architecture and to
6442 allow use of the MQ register; specify this for the Motorola MPC601.
6444 @item -mnew-mnemonics
6445 @itemx -mold-mnemonics
6446 @opindex mnew-mnemonics
6447 @opindex mold-mnemonics
6448 Select which mnemonics to use in the generated assembler code.
6449 @samp{-mnew-mnemonics} requests output that uses the assembler mnemonics
6450 defined for the PowerPC architecture, while @samp{-mold-mnemonics}
6451 requests the assembler mnemonics defined for the POWER architecture.
6452 Instructions defined in only one architecture have only one mnemonic;
6453 GCC uses that mnemonic irrespective of which of these options is
6456 GCC defaults to the mnemonics appropriate for the architecture in
6457 use. Specifying @samp{-mcpu=@var{cpu_type}} sometimes overrides the
6458 value of these option. Unless you are building a cross-compiler, you
6459 should normally not specify either @samp{-mnew-mnemonics} or
6460 @samp{-mold-mnemonics}, but should instead accept the default.
6462 @item -mcpu=@var{cpu_type}
6464 Set architecture type, register usage, choice of mnemonics, and
6465 instruction scheduling parameters for machine type @var{cpu_type}.
6466 Supported values for @var{cpu_type} are @samp{rios}, @samp{rios1},
6467 @samp{rsc}, @samp{rios2}, @samp{rs64a}, @samp{601}, @samp{602},
6468 @samp{603}, @samp{603e}, @samp{604}, @samp{604e}, @samp{620},
6469 @samp{630}, @samp{740}, @samp{750}, @samp{power}, @samp{power2},
6470 @samp{powerpc}, @samp{403}, @samp{505}, @samp{801}, @samp{821},
6471 @samp{823}, and @samp{860} and @samp{common}. @samp{-mcpu=power},
6472 @samp{-mcpu=power2}, @samp{-mcpu=powerpc}, and @samp{-mcpu=powerpc64}
6473 specify generic POWER, POWER2, pure 32-bit PowerPC (i.e., not MPC601),
6474 and 64-bit PowerPC architecture machine types, with an appropriate,
6475 generic processor model assumed for scheduling purposes.@refill
6477 Specifying any of the following options:
6478 @samp{-mcpu=rios1}, @samp{-mcpu=rios2}, @samp{-mcpu=rsc},
6479 @samp{-mcpu=power}, or @samp{-mcpu=power2}
6480 enables the @samp{-mpower} option and disables the @samp{-mpowerpc} option;
6481 @samp{-mcpu=601} enables both the @samp{-mpower} and @samp{-mpowerpc} options.
6482 All of @samp{-mcpu=rs64a}, @samp{-mcpu=602}, @samp{-mcpu=603},
6483 @samp{-mcpu=603e}, @samp{-mcpu=604}, @samp{-mcpu=620}, @samp{-mcpu=630},
6484 @samp{-mcpu=740}, and @samp{-mcpu=750}
6485 enable the @samp{-mpowerpc} option and disable the @samp{-mpower} option.
6486 Exactly similarly, all of @samp{-mcpu=403},
6487 @samp{-mcpu=505}, @samp{-mcpu=821}, @samp{-mcpu=860} and @samp{-mcpu=powerpc}
6488 enable the @samp{-mpowerpc} option and disable the @samp{-mpower} option.
6489 @samp{-mcpu=common} disables both the
6490 @samp{-mpower} and @samp{-mpowerpc} options.@refill
6492 AIX versions 4 or greater selects @samp{-mcpu=common} by default, so
6493 that code will operate on all members of the RS/6000 POWER and PowerPC
6494 families. In that case, GCC will use only the instructions in the
6495 common subset of both architectures plus some special AIX common-mode
6496 calls, and will not use the MQ register. GCC assumes a generic
6497 processor model for scheduling purposes.
6499 Specifying any of the options @samp{-mcpu=rios1}, @samp{-mcpu=rios2},
6500 @samp{-mcpu=rsc}, @samp{-mcpu=power}, or @samp{-mcpu=power2} also
6501 disables the @samp{new-mnemonics} option. Specifying @samp{-mcpu=601},
6502 @samp{-mcpu=602}, @samp{-mcpu=603}, @samp{-mcpu=603e}, @samp{-mcpu=604},
6503 @samp{-mcpu=620}, @samp{-mcpu=630}, @samp{-mcpu=403}, @samp{-mcpu=505},
6504 @samp{-mcpu=821}, @samp{-mcpu=860} or @samp{-mcpu=powerpc} also enables
6505 the @samp{new-mnemonics} option.@refill
6507 Specifying @samp{-mcpu=403}, @samp{-mcpu=821}, or @samp{-mcpu=860} also
6508 enables the @samp{-msoft-float} option.
6510 @item -mtune=@var{cpu_type}
6512 Set the instruction scheduling parameters for machine type
6513 @var{cpu_type}, but do not set the architecture type, register usage,
6514 choice of mnemonics like @samp{-mcpu=}@var{cpu_type} would. The same
6515 values for @var{cpu_type} are used for @samp{-mtune=}@var{cpu_type} as
6516 for @samp{-mcpu=}@var{cpu_type}. The @samp{-mtune=}@var{cpu_type}
6517 option overrides the @samp{-mcpu=}@var{cpu_type} option in terms of
6518 instruction scheduling parameters.
6521 @itemx -mno-fp-in-toc
6522 @itemx -mno-sum-in-toc
6523 @itemx -mminimal-toc
6525 @opindex mno-fp-in-toc
6526 @opindex mno-sum-in-toc
6527 @opindex mminimal-toc
6528 Modify generation of the TOC (Table Of Contents), which is created for
6529 every executable file. The @samp{-mfull-toc} option is selected by
6530 default. In that case, GCC will allocate at least one TOC entry for
6531 each unique non-automatic variable reference in your program. GCC
6532 will also place floating-point constants in the TOC. However, only
6533 16,384 entries are available in the TOC.
6535 If you receive a linker error message that saying you have overflowed
6536 the available TOC space, you can reduce the amount of TOC space used
6537 with the @samp{-mno-fp-in-toc} and @samp{-mno-sum-in-toc} options.
6538 @samp{-mno-fp-in-toc} prevents GCC from putting floating-point
6539 constants in the TOC and @samp{-mno-sum-in-toc} forces GCC to
6540 generate code to calculate the sum of an address and a constant at
6541 run-time instead of putting that sum into the TOC. You may specify one
6542 or both of these options. Each causes GCC to produce very slightly
6543 slower and larger code at the expense of conserving TOC space.
6545 If you still run out of space in the TOC even when you specify both of
6546 these options, specify @samp{-mminimal-toc} instead. This option causes
6547 GCC to make only one TOC entry for every file. When you specify this
6548 option, GCC will produce code that is slower and larger but which
6549 uses extremely little TOC space. You may wish to use this option
6550 only on files that contain less frequently executed code. @refill
6556 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
6557 @code{long} type, and the infrastructure needed to support them.
6558 Specifying @samp{-maix64} implies @samp{-mpowerpc64} and
6559 @samp{-mpowerpc}, while @samp{-maix32} disables the 64-bit ABI and
6560 implies @samp{-mno-powerpc64}. GCC defaults to @samp{-maix32}.
6565 @opindex mno-xl-call
6566 On AIX, pass floating-point arguments to prototyped functions beyond the
6567 register save area (RSA) on the stack in addition to argument FPRs. The
6568 AIX calling convention was extended but not initially documented to
6569 handle an obscure K&R C case of calling a function that takes the
6570 address of its arguments with fewer arguments than declared. AIX XL
6571 compilers access floating point arguments which do not fit in the
6572 RSA from the stack when a subroutine is compiled without
6573 optimization. Because always storing floating-point arguments on the
6574 stack is inefficient and rarely needed, this option is not enabled by
6575 default and only is necessary when calling subroutines compiled by AIX
6576 XL compilers without optimization.
6580 Support @dfn{AIX Threads}. Link an application written to use
6581 @dfn{pthreads} with special libraries and startup code to enable the
6586 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE). Link an
6587 application written to use message passing with special startup code to
6588 enable the application to run. The system must have PE installed in the
6589 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
6590 must be overridden with the @samp{-specs=} option to specify the
6591 appropriate directory location. The Parallel Environment does not
6592 support threads, so the @samp{-mpe} option and the @samp{-mthreads}
6593 option are incompatible.
6597 @opindex msoft-float
6598 @opindex mhard-float
6599 Generate code that does not use (uses) the floating-point register set.
6600 Software floating point emulation is provided if you use the
6601 @samp{-msoft-float} option, and pass the option to GCC when linking.
6604 @itemx -mno-multiple
6606 @opindex mno-multiple
6607 Generate code that uses (does not use) the load multiple word
6608 instructions and the store multiple word instructions. These
6609 instructions are generated by default on POWER systems, and not
6610 generated on PowerPC systems. Do not use @samp{-mmultiple} on little
6611 endian PowerPC systems, since those instructions do not work when the
6612 processor is in little endian mode. The exceptions are PPC740 and
6613 PPC750 which permit the instructions usage in little endian mode.
6619 Generate code that uses (does not use) the load string instructions
6620 and the store string word instructions to save multiple registers and
6621 do small block moves. These instructions are generated by default on
6622 POWER systems, and not generated on PowerPC systems. Do not use
6623 @samp{-mstring} on little endian PowerPC systems, since those
6624 instructions do not work when the processor is in little endian mode.
6625 The exceptions are PPC740 and PPC750 which permit the instructions
6626 usage in little endian mode.
6632 Generate code that uses (does not use) the load or store instructions
6633 that update the base register to the address of the calculated memory
6634 location. These instructions are generated by default. If you use
6635 @samp{-mno-update}, there is a small window between the time that the
6636 stack pointer is updated and the address of the previous frame is
6637 stored, which means code that walks the stack frame across interrupts or
6638 signals may get corrupted data.
6641 @itemx -mno-fused-madd
6642 @opindex mfused-madd
6643 @opindex mno-fused-madd
6644 Generate code that uses (does not use) the floating point multiply and
6645 accumulate instructions. These instructions are generated by default if
6646 hardware floating is used.
6648 @item -mno-bit-align
6650 @opindex mno-bit-align
6652 On System V.4 and embedded PowerPC systems do not (do) force structures
6653 and unions that contain bit fields to be aligned to the base type of the
6656 For example, by default a structure containing nothing but 8
6657 @code{unsigned} bitfields of length 1 would be aligned to a 4 byte
6658 boundary and have a size of 4 bytes. By using @samp{-mno-bit-align},
6659 the structure would be aligned to a 1 byte boundary and be one byte in
6662 @item -mno-strict-align
6663 @itemx -mstrict-align
6664 @opindex mno-strict-align
6665 @opindex mstrict-align
6666 On System V.4 and embedded PowerPC systems do not (do) assume that
6667 unaligned memory references will be handled by the system.
6670 @itemx -mno-relocatable
6671 @opindex mrelocatable
6672 @opindex mno-relocatable
6673 On embedded PowerPC systems generate code that allows (does not allow)
6674 the program to be relocated to a different address at runtime. If you
6675 use @samp{-mrelocatable} on any module, all objects linked together must
6676 be compiled with @samp{-mrelocatable} or @samp{-mrelocatable-lib}.
6678 @item -mrelocatable-lib
6679 @itemx -mno-relocatable-lib
6680 @opindex mrelocatable-lib
6681 @opindex mno-relocatable-lib
6682 On embedded PowerPC systems generate code that allows (does not allow)
6683 the program to be relocated to a different address at runtime. Modules
6684 compiled with @samp{-mrelocatable-lib} can be linked with either modules
6685 compiled without @samp{-mrelocatable} and @samp{-mrelocatable-lib} or
6686 with modules compiled with the @samp{-mrelocatable} options.
6692 On System V.4 and embedded PowerPC systems do not (do) assume that
6693 register 2 contains a pointer to a global area pointing to the addresses
6694 used in the program.
6697 @itemx -mlittle-endian
6699 @opindex mlittle-endian
6700 On System V.4 and embedded PowerPC systems compile code for the
6701 processor in little endian mode. The @samp{-mlittle-endian} option is
6702 the same as @samp{-mlittle}.
6707 @opindex mbig-endian
6708 On System V.4 and embedded PowerPC systems compile code for the
6709 processor in big endian mode. The @samp{-mbig-endian} option is
6710 the same as @samp{-mbig}.
6714 On System V.4 and embedded PowerPC systems compile code using calling
6715 conventions that adheres to the March 1995 draft of the System V
6716 Application Binary Interface, PowerPC processor supplement. This is the
6717 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
6719 @item -mcall-sysv-eabi
6720 @opindex mcall-sysv-eabi
6721 Specify both @samp{-mcall-sysv} and @samp{-meabi} options.
6723 @item -mcall-sysv-noeabi
6724 @opindex mcall-sysv-noeabi
6725 Specify both @samp{-mcall-sysv} and @samp{-mno-eabi} options.
6729 On System V.4 and embedded PowerPC systems compile code using calling
6730 conventions that are similar to those used on AIX. This is the
6731 default if you configured GCC using @samp{powerpc-*-eabiaix}.
6733 @item -mcall-solaris
6734 @opindex mcall-solaris
6735 On System V.4 and embedded PowerPC systems compile code for the Solaris
6739 @opindex mcall-linux
6740 On System V.4 and embedded PowerPC systems compile code for the
6741 Linux-based GNU system.
6744 @itemx -mno-prototype
6746 @opindex mno-prototype
6747 On System V.4 and embedded PowerPC systems assume that all calls to
6748 variable argument functions are properly prototyped. Otherwise, the
6749 compiler must insert an instruction before every non prototyped call to
6750 set or clear bit 6 of the condition code register (@var{CR}) to
6751 indicate whether floating point values were passed in the floating point
6752 registers in case the function takes a variable arguments. With
6753 @samp{-mprototype}, only calls to prototyped variable argument functions
6754 will set or clear the bit.
6758 On embedded PowerPC systems, assume that the startup module is called
6759 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
6760 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
6765 On embedded PowerPC systems, assume that the startup module is called
6766 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
6771 On embedded PowerPC systems, assume that the startup module is called
6772 @file{crt0.o} and the standard C libraries are @file{libads.a} and
6776 @opindex myellowknife
6777 On embedded PowerPC systems, assume that the startup module is called
6778 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
6783 On System V.4 and embedded PowerPC systems, specify that you are
6784 compiling for a VxWorks system.
6788 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
6789 header to indicate that @samp{eabi} extended relocations are used.
6795 On System V.4 and embedded PowerPC systems do (do not) adhere to the
6796 Embedded Applications Binary Interface (eabi) which is a set of
6797 modifications to the System V.4 specifications. Selecting @option{-meabi}
6798 means that the stack is aligned to an 8 byte boundary, a function
6799 @code{__eabi} is called to from @code{main} to set up the eabi
6800 environment, and the @samp{-msdata} option can use both @code{r2} and
6801 @code{r13} to point to two separate small data areas. Selecting
6802 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
6803 do not call an initialization function from @code{main}, and the
6804 @samp{-msdata} option will only use @code{r13} to point to a single
6805 small data area. The @samp{-meabi} option is on by default if you
6806 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
6809 @opindex msdata=eabi
6810 On System V.4 and embedded PowerPC systems, put small initialized
6811 @code{const} global and static data in the @samp{.sdata2} section, which
6812 is pointed to by register @code{r2}. Put small initialized
6813 non-@code{const} global and static data in the @samp{.sdata} section,
6814 which is pointed to by register @code{r13}. Put small uninitialized
6815 global and static data in the @samp{.sbss} section, which is adjacent to
6816 the @samp{.sdata} section. The @samp{-msdata=eabi} option is
6817 incompatible with the @samp{-mrelocatable} option. The
6818 @samp{-msdata=eabi} option also sets the @samp{-memb} option.
6821 @opindex msdata=sysv
6822 On System V.4 and embedded PowerPC systems, put small global and static
6823 data in the @samp{.sdata} section, which is pointed to by register
6824 @code{r13}. Put small uninitialized global and static data in the
6825 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
6826 The @samp{-msdata=sysv} option is incompatible with the
6827 @samp{-mrelocatable} option.
6829 @item -msdata=default
6831 @opindex msdata=default
6833 On System V.4 and embedded PowerPC systems, if @samp{-meabi} is used,
6834 compile code the same as @samp{-msdata=eabi}, otherwise compile code the
6835 same as @samp{-msdata=sysv}.
6838 @opindex msdata-data
6839 On System V.4 and embedded PowerPC systems, put small global and static
6840 data in the @samp{.sdata} section. Put small uninitialized global and
6841 static data in the @samp{.sbss} section. Do not use register @code{r13}
6842 to address small data however. This is the default behavior unless
6843 other @samp{-msdata} options are used.
6847 @opindex msdata=none
6849 On embedded PowerPC systems, put all initialized global and static data
6850 in the @samp{.data} section, and all uninitialized data in the
6851 @samp{.bss} section.
6855 @cindex smaller data references (PowerPC)
6856 @cindex .sdata/.sdata2 references (PowerPC)
6857 On embedded PowerPC systems, put global and static items less than or
6858 equal to @var{num} bytes into the small data or bss sections instead of
6859 the normal data or bss section. By default, @var{num} is 8. The
6860 @samp{-G @var{num}} switch is also passed to the linker.
6861 All modules should be compiled with the same @samp{-G @var{num}} value.
6864 @itemx -mno-regnames
6866 @opindex mno-regnames
6867 On System V.4 and embedded PowerPC systems do (do not) emit register
6868 names in the assembly language output using symbolic forms.
6873 @subsection IBM RT Options
6875 @cindex IBM RT options
6877 These @samp{-m} options are defined for the IBM RT PC:
6881 @opindex min-line-mul
6882 Use an in-line code sequence for integer multiplies. This is the
6885 @item -mcall-lib-mul
6886 @opindex mcall-lib-mul
6887 Call @code{lmul$$} for integer multiples.
6889 @item -mfull-fp-blocks
6890 @opindex mfull-fp-blocks
6891 Generate full-size floating point data blocks, including the minimum
6892 amount of scratch space recommended by IBM. This is the default.
6894 @item -mminimum-fp-blocks
6895 @opindex mminimum-fp-blocks
6896 Do not include extra scratch space in floating point data blocks. This
6897 results in smaller code, but slower execution, since scratch space must
6898 be allocated dynamically.
6900 @cindex @file{varargs.h} and RT PC
6901 @cindex @file{stdarg.h} and RT PC
6902 @item -mfp-arg-in-fpregs
6903 @opindex mfp-arg-in-fpregs
6904 Use a calling sequence incompatible with the IBM calling convention in
6905 which floating point arguments are passed in floating point registers.
6906 Note that @code{varargs.h} and @code{stdargs.h} will not work with
6907 floating point operands if this option is specified.
6909 @item -mfp-arg-in-gregs
6910 @opindex mfp-arg-in-gregs
6911 Use the normal calling convention for floating point arguments. This is
6914 @item -mhc-struct-return
6915 @opindex mhc-struct-return
6916 Return structures of more than one word in memory, rather than in a
6917 register. This provides compatibility with the MetaWare HighC (hc)
6918 compiler. Use the option @samp{-fpcc-struct-return} for compatibility
6919 with the Portable C Compiler (pcc).
6921 @item -mnohc-struct-return
6922 @opindex mnohc-struct-return
6923 Return some structures of more than one word in registers, when
6924 convenient. This is the default. For compatibility with the
6925 IBM-supplied compilers, use the option @samp{-fpcc-struct-return} or the
6926 option @samp{-mhc-struct-return}.
6930 @subsection MIPS Options
6931 @cindex MIPS options
6933 These @samp{-m} options are defined for the MIPS family of computers:
6936 @item -mcpu=@var{cpu type}
6938 Assume the defaults for the machine type @var{cpu type} when scheduling
6939 instructions. The choices for @var{cpu type} are @samp{r2000}, @samp{r3000},
6940 @samp{r3900}, @samp{r4000}, @samp{r4100}, @samp{r4300}, @samp{r4400},
6941 @samp{r4600}, @samp{r4650}, @samp{r5000}, @samp{r6000}, @samp{r8000},
6942 and @samp{orion}. Additionally, the @samp{r2000}, @samp{r3000},
6943 @samp{r4000}, @samp{r5000}, and @samp{r6000} can be abbreviated as
6944 @samp{r2k} (or @samp{r2K}), @samp{r3k}, etc. While picking a specific
6945 @var{cpu type} will schedule things appropriately for that particular
6946 chip, the compiler will not generate any code that does not meet level 1
6947 of the MIPS ISA (instruction set architecture) without a @samp{-mipsX}
6948 or @samp{-mabi} switch being used.
6952 Issue instructions from level 1 of the MIPS ISA. This is the default.
6953 @samp{r3000} is the default @var{cpu type} at this ISA level.
6957 Issue instructions from level 2 of the MIPS ISA (branch likely, square
6958 root instructions). @samp{r6000} is the default @var{cpu type} at this
6963 Issue instructions from level 3 of the MIPS ISA (64-bit instructions).
6964 @samp{r4000} is the default @var{cpu type} at this ISA level.
6968 Issue instructions from level 4 of the MIPS ISA (conditional move,
6969 prefetch, enhanced FPU instructions). @samp{r8000} is the default
6970 @var{cpu type} at this ISA level.
6974 Assume that 32 32-bit floating point registers are available. This is
6979 Assume that 32 64-bit floating point registers are available. This is
6980 the default when the @samp{-mips3} option is used.
6984 Assume that 32 32-bit general purpose registers are available. This is
6989 Assume that 32 64-bit general purpose registers are available. This is
6990 the default when the @samp{-mips3} option is used.
6994 Force int and long types to be 64 bits wide. See @samp{-mlong32} for an
6995 explanation of the default, and the width of pointers.
6999 Force long types to be 64 bits wide. See @samp{-mlong32} for an
7000 explanation of the default, and the width of pointers.
7004 Force long, int, and pointer types to be 32 bits wide.
7006 If none of @samp{-mlong32}, @samp{-mlong64}, or @samp{-mint64} are set,
7007 the size of ints, longs, and pointers depends on the ABI and ISA chosen.
7008 For @samp{-mabi=32}, and @samp{-mabi=n32}, ints and longs are 32 bits
7009 wide. For @samp{-mabi=64}, ints are 32 bits, and longs are 64 bits wide.
7010 For @samp{-mabi=eabi} and either @samp{-mips1} or @samp{-mips2}, ints
7011 and longs are 32 bits wide. For @samp{-mabi=eabi} and higher ISAs, ints
7012 are 32 bits, and longs are 64 bits wide. The width of pointer types is
7013 the smaller of the width of longs or the width of general purpose
7014 registers (which in turn depends on the ISA).
7026 Generate code for the indicated ABI. The default instruction level is
7027 @samp{-mips1} for @samp{32}, @samp{-mips3} for @samp{n32}, and
7028 @samp{-mips4} otherwise. Conversely, with @samp{-mips1} or
7029 @samp{-mips2}, the default ABI is @samp{32}; otherwise, the default ABI
7034 Generate code for the MIPS assembler, and invoke @file{mips-tfile} to
7035 add normal debug information. This is the default for all
7036 platforms except for the OSF/1 reference platform, using the OSF/rose
7037 object format. If the either of the @samp{-gstabs} or @samp{-gstabs+}
7038 switches are used, the @file{mips-tfile} program will encapsulate the
7039 stabs within MIPS ECOFF.
7043 Generate code for the GNU assembler. This is the default on the OSF/1
7044 reference platform, using the OSF/rose object format. Also, this is
7045 the default if the configure option @samp{--with-gnu-as} is used.
7047 @item -msplit-addresses
7048 @itemx -mno-split-addresses
7049 @opindex msplit-addresses
7050 @opindex mno-split-addresses
7051 Generate code to load the high and low parts of address constants separately.
7052 This allows @code{gcc} to optimize away redundant loads of the high order
7053 bits of addresses. This optimization requires GNU as and GNU ld.
7054 This optimization is enabled by default for some embedded targets where
7055 GNU as and GNU ld are standard.
7061 The @samp{-mrnames} switch says to output code using the MIPS software
7062 names for the registers, instead of the hardware names (ie, @var{a0}
7063 instead of @var{$4}). The only known assembler that supports this option
7064 is the Algorithmics assembler.
7070 The @samp{-mgpopt} switch says to write all of the data declarations
7071 before the instructions in the text section, this allows the MIPS
7072 assembler to generate one word memory references instead of using two
7073 words for short global or static data items. This is on by default if
7074 optimization is selected.
7080 For each non-inline function processed, the @samp{-mstats} switch
7081 causes the compiler to emit one line to the standard error file to
7082 print statistics about the program (number of registers saved, stack
7089 The @samp{-mmemcpy} switch makes all block moves call the appropriate
7090 string function (@samp{memcpy} or @samp{bcopy}) instead of possibly
7091 generating inline code.
7094 @itemx -mno-mips-tfile
7095 @opindex mmips-tfile
7096 @opindex mno-mips-tfile
7097 The @samp{-mno-mips-tfile} switch causes the compiler not
7098 postprocess the object file with the @file{mips-tfile} program,
7099 after the MIPS assembler has generated it to add debug support. If
7100 @file{mips-tfile} is not run, then no local variables will be
7101 available to the debugger. In addition, @file{stage2} and
7102 @file{stage3} objects will have the temporary file names passed to the
7103 assembler embedded in the object file, which means the objects will
7104 not compare the same. The @samp{-mno-mips-tfile} switch should only
7105 be used when there are bugs in the @file{mips-tfile} program that
7106 prevents compilation.
7109 @opindex msoft-float
7110 Generate output containing library calls for floating point.
7111 @strong{Warning:} the requisite libraries are not part of GCC.
7112 Normally the facilities of the machine's usual C compiler are used, but
7113 this can't be done directly in cross-compilation. You must make your
7114 own arrangements to provide suitable library functions for
7118 @opindex mhard-float
7119 Generate output containing floating point instructions. This is the
7120 default if you use the unmodified sources.
7123 @itemx -mno-abicalls
7125 @opindex mno-abicalls
7126 Emit (or do not emit) the pseudo operations @samp{.abicalls},
7127 @samp{.cpload}, and @samp{.cprestore} that some System V.4 ports use for
7128 position independent code.
7131 @itemx -mno-long-calls
7132 @opindex mlong-calls
7133 @opindex mno-long-calls
7134 Do all calls with the @samp{JALR} instruction, which requires
7135 loading up a function's address into a register before the call.
7136 You need to use this switch, if you call outside of the current
7137 512 megabyte segment to functions that are not through pointers.
7140 @itemx -mno-half-pic
7142 @opindex mno-half-pic
7143 Put pointers to extern references into the data section and load them
7144 up, rather than put the references in the text section.
7146 @item -membedded-pic
7147 @itemx -mno-embedded-pic
7148 @opindex membedded-pic
7149 @opindex mno-embedded-pic
7150 Generate PIC code suitable for some embedded systems. All calls are
7151 made using PC relative address, and all data is addressed using the $gp
7152 register. No more than 65536 bytes of global data may be used. This
7153 requires GNU as and GNU ld which do most of the work. This currently
7154 only works on targets which use ECOFF; it does not work with ELF.
7156 @item -membedded-data
7157 @itemx -mno-embedded-data
7158 @opindex membedded-data
7159 @opindex mno-embedded-data
7160 Allocate variables to the read-only data section first if possible, then
7161 next in the small data section if possible, otherwise in data. This gives
7162 slightly slower code than the default, but reduces the amount of RAM required
7163 when executing, and thus may be preferred for some embedded systems.
7165 @item -muninit-const-in-rodata
7166 @itemx -mno-uninit-const-in-rodata
7167 @opindex muninit-const-in-rodata
7168 @opindex mno-uninit-const-in-rodata
7169 When used together with -membedded-data, it will always store uninitialized
7170 const variables in the read-only data section.
7172 @item -msingle-float
7173 @itemx -mdouble-float
7174 @opindex msingle-float
7175 @opindex mdouble-float
7176 The @samp{-msingle-float} switch tells gcc to assume that the floating
7177 point coprocessor only supports single precision operations, as on the
7178 @samp{r4650} chip. The @samp{-mdouble-float} switch permits gcc to use
7179 double precision operations. This is the default.
7185 Permit use of the @samp{mad}, @samp{madu} and @samp{mul} instructions,
7186 as on the @samp{r4650} chip.
7190 Turns on @samp{-msingle-float}, @samp{-mmad}, and, at least for now,
7197 Enable 16-bit instructions.
7201 Use the entry and exit pseudo ops. This option can only be used with
7206 Compile code for the processor in little endian mode.
7207 The requisite libraries are assumed to exist.
7211 Compile code for the processor in big endian mode.
7212 The requisite libraries are assumed to exist.
7216 @cindex smaller data references (MIPS)
7217 @cindex gp-relative references (MIPS)
7218 Put global and static items less than or equal to @var{num} bytes into
7219 the small data or bss sections instead of the normal data or bss
7220 section. This allows the assembler to emit one word memory reference
7221 instructions based on the global pointer (@var{gp} or @var{$28}),
7222 instead of the normal two words used. By default, @var{num} is 8 when
7223 the MIPS assembler is used, and 0 when the GNU assembler is used. The
7224 @samp{-G @var{num}} switch is also passed to the assembler and linker.
7225 All modules should be compiled with the same @samp{-G @var{num}}
7230 Tell the MIPS assembler to not run its preprocessor over user
7231 assembler files (with a @samp{.s} suffix) when assembling them.
7235 Pass an option to gas which will cause nops to be inserted if
7236 the read of the destination register of an mfhi or mflo instruction
7237 occurs in the following two instructions.
7241 Do not include the default crt0.
7245 These options are defined by the macro
7246 @code{TARGET_SWITCHES} in the machine description. The default for the
7247 options is also defined by that macro, which enables you to change the
7252 @subsection Intel 386 Options
7253 @cindex i386 Options
7254 @cindex Intel 386 Options
7256 These @samp{-m} options are defined for the i386 family of computers:
7259 @item -mcpu=@var{cpu type}
7261 Assume the defaults for the machine type @var{cpu type} when scheduling
7262 instructions. The choices for @var{cpu type} are @samp{i386},
7263 @samp{i486}, @samp{i586}, @samp{i686}, @samp{pentium},
7264 @samp{pentiumpro}, @samp{pentium4}, @samp{k6}, and @samp{athlon}
7266 While picking a specific @var{cpu type} will schedule things appropriately
7267 for that particular chip, the compiler will not generate any code that
7268 does not run on the i386 without the @samp{-march=@var{cpu type}} option
7269 being used. @samp{i586} is equivalent to @samp{pentium} and @samp{i686}
7270 is equivalent to @samp{pentiumpro}. @samp{k6} and @samp{athlon} are the
7271 AMD chips as opposed to the Intel ones.
7273 @item -march=@var{cpu type}
7275 Generate instructions for the machine type @var{cpu type}. The choices
7276 for @var{cpu type} are the same as for @samp{-mcpu}. Moreover,
7277 specifying @samp{-march=@var{cpu type}} implies @samp{-mcpu=@var{cpu type}}.
7286 @opindex mpentiumpro
7287 Synonyms for -mcpu=i386, -mcpu=i486, -mcpu=pentium, and -mcpu=pentiumpro
7288 respectively. These synonyms are deprecated.
7290 @item -mintel-syntax
7291 @opindex mintel-syntax
7292 Emit assembly using Intel syntax opcodes instead of AT&T syntax.
7297 @opindex mno-ieee-fp
7298 Control whether or not the compiler uses IEEE floating point
7299 comparisons. These handle correctly the case where the result of a
7300 comparison is unordered.
7303 @opindex msoft-float
7304 Generate output containing library calls for floating point.
7305 @strong{Warning:} the requisite libraries are not part of GCC.
7306 Normally the facilities of the machine's usual C compiler are used, but
7307 this can't be done directly in cross-compilation. You must make your
7308 own arrangements to provide suitable library functions for
7311 On machines where a function returns floating point results in the 80387
7312 register stack, some floating point opcodes may be emitted even if
7313 @samp{-msoft-float} is used.
7315 @item -mno-fp-ret-in-387
7316 @opindex mno-fp-ret-in-387
7317 Do not use the FPU registers for return values of functions.
7319 The usual calling convention has functions return values of types
7320 @code{float} and @code{double} in an FPU register, even if there
7321 is no FPU. The idea is that the operating system should emulate
7324 The option @samp{-mno-fp-ret-in-387} causes such values to be returned
7325 in ordinary CPU registers instead.
7327 @item -mno-fancy-math-387
7328 @opindex mno-fancy-math-387
7329 Some 387 emulators do not support the @code{sin}, @code{cos} and
7330 @code{sqrt} instructions for the 387. Specify this option to avoid
7331 generating those instructions. This option is the default on FreeBSD.
7332 As of revision 2.6.1, these instructions are not generated unless you
7333 also use the @samp{-funsafe-math-optimizations} switch.
7335 @item -malign-double
7336 @itemx -mno-align-double
7337 @opindex malign-double
7338 @opindex mno-align-double
7339 Control whether GCC aligns @code{double}, @code{long double}, and
7340 @code{long long} variables on a two word boundary or a one word
7341 boundary. Aligning @code{double} variables on a two word boundary will
7342 produce code that runs somewhat faster on a @samp{Pentium} at the
7343 expense of more memory.
7345 @item -m128bit-long-double
7346 @itemx -m128bit-long-double
7347 @opindex m128bit-long-double
7348 @opindex m128bit-long-double
7349 Control the size of @code{long double} type. i386 application binary interface
7350 specify the size to be 12 bytes, while modern architectures (Pentium and newer)
7351 preffer @code{long double} aligned to 8 or 16 byte boundary. This is
7352 impossible to reach with 12 byte long doubles in the array accesses.
7354 @strong{Warning:} if you use the @samp{-m128bit-long-double} switch, the
7355 structures and arrays containing @code{long double} will change their size as
7356 well as function calling convention for function taking @code{long double}
7359 @item -m96bit-long-double
7360 @itemx -m96bit-long-double
7361 @opindex m96bit-long-double
7362 @opindex m96bit-long-double
7363 Set the size of @code{long double} to 96 bits as required by the i386
7364 application binary interface. This is the default.
7367 @itemx -mno-svr3-shlib
7368 @opindex msvr3-shlib
7369 @opindex mno-svr3-shlib
7370 Control whether GCC places uninitialized locals into @code{bss} or
7371 @code{data}. @samp{-msvr3-shlib} places these locals into @code{bss}.
7372 These options are meaningful only on System V Release 3.
7374 @item -mno-wide-multiply
7375 @itemx -mwide-multiply
7376 @opindex mno-wide-multiply
7377 @opindex mwide-multiply
7378 Control whether GCC uses the @code{mul} and @code{imul} that produce
7379 64-bit results in @code{eax:edx} from 32-bit operands to do @code{long
7380 long} multiplies and 32-bit division by constants.
7384 Use a different function-calling convention, in which functions that
7385 take a fixed number of arguments return with the @code{ret} @var{num}
7386 instruction, which pops their arguments while returning. This saves one
7387 instruction in the caller since there is no need to pop the arguments
7390 You can specify that an individual function is called with this calling
7391 sequence with the function attribute @samp{stdcall}. You can also
7392 override the @samp{-mrtd} option by using the function attribute
7393 @samp{cdecl}. @xref{Function Attributes}.
7395 @strong{Warning:} this calling convention is incompatible with the one
7396 normally used on Unix, so you cannot use it if you need to call
7397 libraries compiled with the Unix compiler.
7399 Also, you must provide function prototypes for all functions that
7400 take variable numbers of arguments (including @code{printf});
7401 otherwise incorrect code will be generated for calls to those
7404 In addition, seriously incorrect code will result if you call a
7405 function with too many arguments. (Normally, extra arguments are
7406 harmlessly ignored.)
7408 @item -mregparm=@var{num}
7410 Control how many registers are used to pass integer arguments. By
7411 default, no registers are used to pass arguments, and at most 3
7412 registers can be used. You can control this behavior for a specific
7413 function by using the function attribute @samp{regparm}.
7414 @xref{Function Attributes}.
7416 @strong{Warning:} if you use this switch, and
7417 @var{num} is nonzero, then you must build all modules with the same
7418 value, including any libraries. This includes the system libraries and
7421 @item -mpreferred-stack-boundary=@var{num}
7422 @opindex mpreferred-stack-boundary
7423 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
7424 byte boundary. If @samp{-mpreferred-stack-boundary} is not specified,
7425 the default is 4 (16 bytes or 128 bits).
7427 The stack is required to be aligned on a 4 byte boundary. On Pentium
7428 and PentiumPro, @code{double} and @code{long double} values should be
7429 aligned to an 8 byte boundary (see @samp{-malign-double}) or suffer
7430 significant run time performance penalties. On Pentium III, the
7431 Streaming SIMD Extension (SSE) data type @code{__m128} suffers similar
7432 penalties if it is not 16 byte aligned.
7434 To ensure proper alignment of this values on the stack, the stack boundary
7435 must be as aligned as that required by any value stored on the stack.
7436 Further, every function must be generated such that it keeps the stack
7437 aligned. Thus calling a function compiled with a higher preferred
7438 stack boundary from a function compiled with a lower preferred stack
7439 boundary will most likely misalign the stack. It is recommended that
7440 libraries that use callbacks always use the default setting.
7442 This extra alignment does consume extra stack space. Code that is sensitive
7443 to stack space usage, such as embedded systems and operating system kernels,
7444 may want to reduce the preferred alignment to
7445 @samp{-mpreferred-stack-boundary=2}.
7449 Use PUSH operations to store outgoing parameters. This method is shorter
7450 and usually equally fast as method using SUB/MOV operations and is enabled
7451 by default. In some cases disabling it may improve performance because of
7452 improved scheduling and reduced dependencies.
7454 @item -maccumulate-outgoing-args
7455 @opindex maccumulate-outgoing-args
7456 If enabled, the maximum amount of space required for outgoing arguments will be
7457 computed in the function prologue. This in faster on most modern CPUs
7458 because of reduced dependencies, improved scheduling and reduced stack usage
7459 when preferred stack boundary is not equal to 2. The drawback is a notable
7460 increase in code size. This switch implies -mno-push-args.
7464 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
7465 on thread-safe exception handling must compile and link all code with the
7466 @samp{-mthreads} option. When compiling, @samp{-mthreads} defines
7467 @samp{-D_MT}; when linking, it links in a special thread helper library
7468 @samp{-lmingwthrd} which cleans up per thread exception handling data.
7470 @item -mno-align-stringops
7471 @opindex mno-align-stringops
7472 Do not align destination of inlined string operations. This switch reduces
7473 code size and improves performance in case the destination is already aligned,
7474 but gcc don't know about it.
7476 @item -minline-all-stringops
7477 @opindex minline-all-stringops
7478 By default GCC inlines string operations only when destination is known to be
7479 aligned at least to 4 byte boundary. This enables more inlining, increase code
7480 size, but may improve performance of code that depends on fast memcpy, strlen
7481 and memset for short lengths.
7483 @item -momit-leaf-frame-pointer
7484 @opindex momit-leaf-frame-pointer
7485 Don't keep the frame pointer in a register for leaf functions. This
7486 avoids the instructions to save, set up and restore frame pointers and
7487 makes an extra register available in leaf functions. The option
7488 @samp{-fomit-frame-pointer} removes the frame pointer for all functions
7489 which might make debugging harder.
7493 @subsection HPPA Options
7494 @cindex HPPA Options
7496 These @samp{-m} options are defined for the HPPA family of computers:
7499 @item -march=@var{architecture type}
7501 Generate code for the specified architecture. The choices for
7502 @var{architecture type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
7503 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
7504 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
7505 architecture option for your machine. Code compiled for lower numbered
7506 architectures will run on higher numbered architectures, but not the
7509 PA 2.0 support currently requires gas snapshot 19990413 or later. The
7510 next release of binutils (current is 2.9.1) will probably contain PA 2.0
7514 @itemx -mpa-risc-1-1
7515 @itemx -mpa-risc-2-0
7516 @opindex mpa-risc-1-0
7517 @opindex mpa-risc-1-1
7518 @opindex mpa-risc-2-0
7519 Synonyms for -march=1.0, -march=1.1, and -march=2.0 respectively.
7522 @opindex mbig-switch
7523 Generate code suitable for big switch tables. Use this option only if
7524 the assembler/linker complain about out of range branches within a switch
7527 @item -mjump-in-delay
7528 @opindex mjump-in-delay
7529 Fill delay slots of function calls with unconditional jump instructions
7530 by modifying the return pointer for the function call to be the target
7531 of the conditional jump.
7533 @item -mdisable-fpregs
7534 @opindex mdisable-fpregs
7535 Prevent floating point registers from being used in any manner. This is
7536 necessary for compiling kernels which perform lazy context switching of
7537 floating point registers. If you use this option and attempt to perform
7538 floating point operations, the compiler will abort.
7540 @item -mdisable-indexing
7541 @opindex mdisable-indexing
7542 Prevent the compiler from using indexing address modes. This avoids some
7543 rather obscure problems when compiling MIG generated code under MACH.
7545 @item -mno-space-regs
7546 @opindex mno-space-regs
7547 Generate code that assumes the target has no space registers. This allows
7548 GCC to generate faster indirect calls and use unscaled index address modes.
7550 Such code is suitable for level 0 PA systems and kernels.
7552 @item -mfast-indirect-calls
7553 @opindex mfast-indirect-calls
7554 Generate code that assumes calls never cross space boundaries. This
7555 allows GCC to emit code which performs faster indirect calls.
7557 This option will not work in the presence of shared libraries or nested
7560 @item -mlong-load-store
7561 @opindex mlong-load-store
7562 Generate 3-instruction load and store sequences as sometimes required by
7563 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
7566 @item -mportable-runtime
7567 @opindex mportable-runtime
7568 Use the portable calling conventions proposed by HP for ELF systems.
7572 Enable the use of assembler directives only GAS understands.
7574 @item -mschedule=@var{cpu type}
7576 Schedule code according to the constraints for the machine type
7577 @var{cpu type}. The choices for @var{cpu type} are @samp{700}
7578 @samp{7100}, @samp{7100LC}, @samp{7200}, and @samp{8000}. Refer to
7579 @file{/usr/lib/sched.models} on an HP-UX system to determine the
7580 proper scheduling option for your machine.
7583 @opindex mlinker-opt
7584 Enable the optimization pass in the HPUX linker. Note this makes symbolic
7585 debugging impossible. It also triggers a bug in the HPUX 8 and HPUX 9 linkers
7586 in which they give bogus error messages when linking some programs.
7589 @opindex msoft-float
7590 Generate output containing library calls for floating point.
7591 @strong{Warning:} the requisite libraries are not available for all HPPA
7592 targets. Normally the facilities of the machine's usual C compiler are
7593 used, but this cannot be done directly in cross-compilation. You must make
7594 your own arrangements to provide suitable library functions for
7595 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
7596 does provide software floating point support.
7598 @samp{-msoft-float} changes the calling convention in the output file;
7599 therefore, it is only useful if you compile @emph{all} of a program with
7600 this option. In particular, you need to compile @file{libgcc.a}, the
7601 library that comes with GCC, with @samp{-msoft-float} in order for
7605 @node Intel 960 Options
7606 @subsection Intel 960 Options
7608 These @samp{-m} options are defined for the Intel 960 implementations:
7611 @item -m@var{cpu-type}
7619 Assume the defaults for the machine type @var{cpu-type} for some of
7620 the other options, including instruction scheduling, floating point
7621 support, and addressing modes. The choices for @var{cpu-type} are
7622 @samp{ka}, @samp{kb}, @samp{mc}, @samp{ca}, @samp{cf},
7623 @samp{sa}, and @samp{sb}.
7630 @opindex msoft-float
7631 The @samp{-mnumerics} option indicates that the processor does support
7632 floating-point instructions. The @samp{-msoft-float} option indicates
7633 that floating-point support should not be assumed.
7635 @item -mleaf-procedures
7636 @itemx -mno-leaf-procedures
7637 @opindex mleaf-procedures
7638 @opindex mno-leaf-procedures
7639 Do (or do not) attempt to alter leaf procedures to be callable with the
7640 @code{bal} instruction as well as @code{call}. This will result in more
7641 efficient code for explicit calls when the @code{bal} instruction can be
7642 substituted by the assembler or linker, but less efficient code in other
7643 cases, such as calls via function pointers, or using a linker that doesn't
7644 support this optimization.
7647 @itemx -mno-tail-call
7649 @opindex mno-tail-call
7650 Do (or do not) make additional attempts (beyond those of the
7651 machine-independent portions of the compiler) to optimize tail-recursive
7652 calls into branches. You may not want to do this because the detection of
7653 cases where this is not valid is not totally complete. The default is
7654 @samp{-mno-tail-call}.
7656 @item -mcomplex-addr
7657 @itemx -mno-complex-addr
7658 @opindex mcomplex-addr
7659 @opindex mno-complex-addr
7660 Assume (or do not assume) that the use of a complex addressing mode is a
7661 win on this implementation of the i960. Complex addressing modes may not
7662 be worthwhile on the K-series, but they definitely are on the C-series.
7663 The default is currently @samp{-mcomplex-addr} for all processors except
7667 @itemx -mno-code-align
7668 @opindex mcode-align
7669 @opindex mno-code-align
7670 Align code to 8-byte boundaries for faster fetching (or don't bother).
7671 Currently turned on by default for C-series implementations only.
7674 @item -mclean-linkage
7675 @itemx -mno-clean-linkage
7676 @opindex mclean-linkage
7677 @opindex mno-clean-linkage
7678 These options are not fully implemented.
7682 @itemx -mic2.0-compat
7683 @itemx -mic3.0-compat
7685 @opindex mic2.0-compat
7686 @opindex mic3.0-compat
7687 Enable compatibility with iC960 v2.0 or v3.0.
7691 @opindex masm-compat
7693 Enable compatibility with the iC960 assembler.
7695 @item -mstrict-align
7696 @itemx -mno-strict-align
7697 @opindex mstrict-align
7698 @opindex mno-strict-align
7699 Do not permit (do permit) unaligned accesses.
7703 Enable structure-alignment compatibility with Intel's gcc release version
7704 1.3 (based on gcc 1.37). This option implies @samp{-mstrict-align}.
7706 @item -mlong-double-64
7707 @opindex mlong-double-64
7708 Implement type @samp{long double} as 64-bit floating point numbers.
7709 Without the option @samp{long double} is implemented by 80-bit
7710 floating point numbers. The only reason we have it because there is
7711 no 128-bit @samp{long double} support in @samp{fp-bit.c} yet. So it
7712 is only useful for people using soft-float targets. Otherwise, we
7713 should recommend against use of it.
7717 @node DEC Alpha Options
7718 @subsection DEC Alpha Options
7720 These @samp{-m} options are defined for the DEC Alpha implementations:
7723 @item -mno-soft-float
7725 @opindex mno-soft-float
7726 @opindex msoft-float
7727 Use (do not use) the hardware floating-point instructions for
7728 floating-point operations. When @option{-msoft-float} is specified,
7729 functions in @file{libgcc.a} will be used to perform floating-point
7730 operations. Unless they are replaced by routines that emulate the
7731 floating-point operations, or compiled in such a way as to call such
7732 emulations routines, these routines will issue floating-point
7733 operations. If you are compiling for an Alpha without floating-point
7734 operations, you must ensure that the library is built so as not to call
7737 Note that Alpha implementations without floating-point operations are
7738 required to have floating-point registers.
7743 @opindex mno-fp-regs
7744 Generate code that uses (does not use) the floating-point register set.
7745 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
7746 register set is not used, floating point operands are passed in integer
7747 registers as if they were integers and floating-point results are passed
7748 in $0 instead of $f0. This is a non-standard calling sequence, so any
7749 function with a floating-point argument or return value called by code
7750 compiled with @option{-mno-fp-regs} must also be compiled with that
7753 A typical use of this option is building a kernel that does not use,
7754 and hence need not save and restore, any floating-point registers.
7758 The Alpha architecture implements floating-point hardware optimized for
7759 maximum performance. It is mostly compliant with the IEEE floating
7760 point standard. However, for full compliance, software assistance is
7761 required. This option generates code fully IEEE compliant code
7762 @emph{except} that the @var{inexact flag} is not maintained (see below).
7763 If this option is turned on, the CPP macro @code{_IEEE_FP} is defined
7764 during compilation. The option is a shorthand for: @samp{-D_IEEE_FP
7765 -mfp-trap-mode=su -mtrap-precision=i -mieee-conformant}. The resulting
7766 code is less efficient but is able to correctly support denormalized
7767 numbers and exceptional IEEE values such as not-a-number and plus/minus
7768 infinity. Other Alpha compilers call this option
7769 @option{-ieee_with_no_inexact}.
7771 @item -mieee-with-inexact
7772 @opindex mieee-with-inexact
7773 @c overfull hbox here --bob 22 jul96
7774 @c original text between ignore ... end ignore
7776 This is like @samp{-mieee} except the generated code also maintains the
7777 IEEE @var{inexact flag}. Turning on this option causes the generated
7778 code to implement fully-compliant IEEE math. The option is a shorthand
7779 for @samp{-D_IEEE_FP -D_IEEE_FP_INEXACT} plus @samp{-mieee-conformant},
7780 @samp{-mfp-trap-mode=sui}, and @samp{-mtrap-precision=i}. On some Alpha
7781 implementations the resulting code may execute significantly slower than
7782 the code generated by default. Since there is very little code that
7783 depends on the @var{inexact flag}, you should normally not specify this
7784 option. Other Alpha compilers call this option
7785 @samp{-ieee_with_inexact}.
7787 @c changed paragraph
7788 This is like @samp{-mieee} except the generated code also maintains the
7789 IEEE @var{inexact flag}. Turning on this option causes the generated
7790 code to implement fully-compliant IEEE math. The option is a shorthand
7791 for @samp{-D_IEEE_FP -D_IEEE_FP_INEXACT} plus the three following:
7792 @samp{-mieee-conformant},
7793 @samp{-mfp-trap-mode=sui},
7794 and @samp{-mtrap-precision=i}.
7795 On some Alpha implementations the resulting code may execute
7796 significantly slower than the code generated by default. Since there
7797 is very little code that depends on the @var{inexact flag}, you should
7798 normally not specify this option. Other Alpha compilers call this
7799 option @samp{-ieee_with_inexact}.
7800 @c end changes to prevent overfull hboxes
7802 @item -mfp-trap-mode=@var{trap mode}
7803 @opindex mfp-trap-mode
7804 This option controls what floating-point related traps are enabled.
7805 Other Alpha compilers call this option @samp{-fptm }@var{trap mode}.
7806 The trap mode can be set to one of four values:
7810 This is the default (normal) setting. The only traps that are enabled
7811 are the ones that cannot be disabled in software (e.g., division by zero
7815 In addition to the traps enabled by @samp{n}, underflow traps are enabled
7819 Like @samp{su}, but the instructions are marked to be safe for software
7820 completion (see Alpha architecture manual for details).
7823 Like @samp{su}, but inexact traps are enabled as well.
7826 @item -mfp-rounding-mode=@var{rounding mode}
7827 @opindex mfp-rounding-mode
7828 Selects the IEEE rounding mode. Other Alpha compilers call this option
7829 @samp{-fprm }@var{rounding mode}. The @var{rounding mode} can be one
7834 Normal IEEE rounding mode. Floating point numbers are rounded towards
7835 the nearest machine number or towards the even machine number in case
7839 Round towards minus infinity.
7842 Chopped rounding mode. Floating point numbers are rounded towards zero.
7845 Dynamic rounding mode. A field in the floating point control register
7846 (@var{fpcr}, see Alpha architecture reference manual) controls the
7847 rounding mode in effect. The C library initializes this register for
7848 rounding towards plus infinity. Thus, unless your program modifies the
7849 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
7852 @item -mtrap-precision=@var{trap precision}
7853 @opindex mtrap-precision
7854 In the Alpha architecture, floating point traps are imprecise. This
7855 means without software assistance it is impossible to recover from a
7856 floating trap and program execution normally needs to be terminated.
7857 GCC can generate code that can assist operating system trap handlers
7858 in determining the exact location that caused a floating point trap.
7859 Depending on the requirements of an application, different levels of
7860 precisions can be selected:
7864 Program precision. This option is the default and means a trap handler
7865 can only identify which program caused a floating point exception.
7868 Function precision. The trap handler can determine the function that
7869 caused a floating point exception.
7872 Instruction precision. The trap handler can determine the exact
7873 instruction that caused a floating point exception.
7876 Other Alpha compilers provide the equivalent options called
7877 @samp{-scope_safe} and @samp{-resumption_safe}.
7879 @item -mieee-conformant
7880 @opindex mieee-conformant
7881 This option marks the generated code as IEEE conformant. You must not
7882 use this option unless you also specify @samp{-mtrap-precision=i} and either
7883 @samp{-mfp-trap-mode=su} or @samp{-mfp-trap-mode=sui}. Its only effect
7884 is to emit the line @samp{.eflag 48} in the function prologue of the
7885 generated assembly file. Under DEC Unix, this has the effect that
7886 IEEE-conformant math library routines will be linked in.
7888 @item -mbuild-constants
7889 @opindex mbuild-constants
7890 Normally GCC examines a 32- or 64-bit integer constant to
7891 see if it can construct it from smaller constants in two or three
7892 instructions. If it cannot, it will output the constant as a literal and
7893 generate code to load it from the data segment at runtime.
7895 Use this option to require GCC to construct @emph{all} integer constants
7896 using code, even if it takes more instructions (the maximum is six).
7898 You would typically use this option to build a shared library dynamic
7899 loader. Itself a shared library, it must relocate itself in memory
7900 before it can find the variables and constants in its own data segment.
7906 Select whether to generate code to be assembled by the vendor-supplied
7907 assembler (@samp{-malpha-as}) or by the GNU assembler @samp{-mgas}.
7921 Indicate whether GCC should generate code to use the optional BWX,
7922 CIX, and MAX instruction sets. The default is to use the instruction sets
7923 supported by the CPU type specified via @samp{-mcpu=} option or that
7924 of the CPU on which GCC was built if none was specified.
7926 @item -mcpu=@var{cpu_type}
7928 Set the instruction set, register set, and instruction scheduling
7929 parameters for machine type @var{cpu_type}. You can specify either the
7930 @samp{EV} style name or the corresponding chip number. GCC
7931 supports scheduling parameters for the EV4 and EV5 family of processors
7932 and will choose the default values for the instruction set from
7933 the processor you specify. If you do not specify a processor type,
7934 GCC will default to the processor on which the compiler was built.
7936 Supported values for @var{cpu_type} are
7941 Schedules as an EV4 and has no instruction set extensions.
7945 Schedules as an EV5 and has no instruction set extensions.
7949 Schedules as an EV5 and supports the BWX extension.
7954 Schedules as an EV5 and supports the BWX and MAX extensions.
7958 Schedules as an EV5 (until Digital releases the scheduling parameters
7959 for the EV6) and supports the BWX, CIX, and MAX extensions.
7962 @item -mmemory-latency=@var{time}
7963 @opindex mmemory-latency
7964 Sets the latency the scheduler should assume for typical memory
7965 references as seen by the application. This number is highly
7966 dependent on the memory access patterns used by the application
7967 and the size of the external cache on the machine.
7969 Valid options for @var{time} are
7973 A decimal number representing clock cycles.
7979 The compiler contains estimates of the number of clock cycles for
7980 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
7981 (also called Dcache, Scache, and Bcache), as well as to main memory.
7982 Note that L3 is only valid for EV5.
7987 @node Clipper Options
7988 @subsection Clipper Options
7990 These @samp{-m} options are defined for the Clipper implementations:
7995 Produce code for a C300 Clipper processor. This is the default.
7999 Produce code for a C400 Clipper processor i.e. use floating point
8003 @node H8/300 Options
8004 @subsection H8/300 Options
8006 These @samp{-m} options are defined for the H8/300 implementations:
8011 Shorten some address references at link time, when possible; uses the
8012 linker option @samp{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
8013 ld.info, Using ld}, for a fuller description.
8017 Generate code for the H8/300H.
8021 Generate code for the H8/S.
8025 Generate code for the H8/S2600. This switch must be used with -ms.
8029 Make @code{int} data 32 bits by default.
8033 On the H8/300H and H8/S, use the same alignment rules as for the H8/300.
8034 The default for the H8/300H and H8/S is to align longs and floats on 4
8036 @samp{-malign-300} causes them to be aligned on 2 byte boundaries.
8037 This option has no effect on the H8/300.
8041 @subsection SH Options
8043 These @samp{-m} options are defined for the SH implementations:
8048 Generate code for the SH1.
8052 Generate code for the SH2.
8056 Generate code for the SH3.
8060 Generate code for the SH3e.
8064 Generate code for the SH4 without a floating-point unit.
8066 @item -m4-single-only
8067 @opindex m4-single-only
8068 Generate code for the SH4 with a floating-point unit that only
8069 supports single-precision arithmentic.
8073 Generate code for the SH4 assuming the floating-point unit is in
8074 single-precision mode by default.
8078 Generate code for the SH4.
8082 Compile code for the processor in big endian mode.
8086 Compile code for the processor in little endian mode.
8090 Align doubles at 64-bit boundaries. Note that this changes the calling
8091 conventions, and thus some functions from the standard C library will
8092 not work unless you recompile it first with -mdalign.
8096 Shorten some address references at link time, when possible; uses the
8097 linker option @samp{-relax}.
8101 Use 32-bit offsets in @code{switch} tables. The default is to use
8106 Enable the use of the instruction @code{fmovd}.
8110 Comply with the calling conventions defined by Hitachi.
8114 Mark the @code{MAC} register as call-clobbered, even if
8115 @option{-mhitachi} is given.
8119 Dump instruction size and location in the assembly code.
8123 This option is deprecated. It pads structures to multiple of 4 bytes,
8124 which is incompatible with the SH ABI.
8128 Optimize for space instead of speed. Implied by @option{-Os}.
8132 When generating position-independent code, emit function calls using
8133 the Global Offset Table instead of the Procedure Linkage Table.
8137 Generate a library function call to invalidate instruction cache
8138 entries, after fixing up a trampoline. This library function call
8139 doesn't assume it can write to the whole memory address space. This
8140 is the default when the target is @code{sh-*-linux*}.
8143 @node System V Options
8144 @subsection Options for System V
8146 These additional options are available on System V Release 4 for
8147 compatibility with other compilers on those systems:
8152 Create a shared object.
8153 It is recommended that @samp{-symbolic} or @samp{-shared} be used instead.
8157 Identify the versions of each tool used by the compiler, in a
8158 @code{.ident} assembler directive in the output.
8162 Refrain from adding @code{.ident} directives to the output file (this is
8165 @item -YP\,@var{dirs}
8167 Search the directories @var{dirs}, and no others, for libraries
8168 specified with @samp{-l}.
8170 @item -Ym\,@var{dir}
8172 Look in the directory @var{dir} to find the M4 preprocessor.
8173 The assembler uses this option.
8174 @c This is supposed to go with a -Yd for predefined M4 macro files, but
8175 @c the generic assembler that comes with Solaris takes just -Ym.
8178 @node TMS320C3x/C4x Options
8179 @subsection TMS320C3x/C4x Options
8180 @cindex TMS320C3x/C4x Options
8182 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
8186 @item -mcpu=@var{cpu_type}
8188 Set the instruction set, register set, and instruction scheduling
8189 parameters for machine type @var{cpu_type}. Supported values for
8190 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
8191 @samp{c44}. The default is @samp{c40} to generate code for the
8196 @itemx -msmall-memory
8198 @opindex mbig-memory
8200 @opindex msmall-memory
8202 Generates code for the big or small memory model. The small memory
8203 model assumed that all data fits into one 64K word page. At run-time
8204 the data page (DP) register must be set to point to the 64K page
8205 containing the .bss and .data program sections. The big memory model is
8206 the default and requires reloading of the DP register for every direct
8213 Allow (disallow) allocation of general integer operands into the block
8220 Enable (disable) generation of code using decrement and branch,
8221 DBcond(D), instructions. This is enabled by default for the C4x. To be
8222 on the safe side, this is disabled for the C3x, since the maximum
8223 iteration count on the C3x is 2^23 + 1 (but who iterates loops more than
8224 2^23 times on the C3x?). Note that GCC will try to reverse a loop so
8225 that it can utilise the decrement and branch instruction, but will give
8226 up if there is more than one memory reference in the loop. Thus a loop
8227 where the loop counter is decremented can generate slightly more
8228 efficient code, in cases where the RPTB instruction cannot be utilised.
8230 @item -mdp-isr-reload
8232 @opindex mdp-isr-reload
8234 Force the DP register to be saved on entry to an interrupt service
8235 routine (ISR), reloaded to point to the data section, and restored on
8236 exit from the ISR. This should not be required unless someone has
8237 violated the small memory model by modifying the DP register, say within
8244 For the C3x use the 24-bit MPYI instruction for integer multiplies
8245 instead of a library call to guarantee 32-bit results. Note that if one
8246 of the operands is a constant, then the multiplication will be performed
8247 using shifts and adds. If the -mmpyi option is not specified for the C3x,
8248 then squaring operations are performed inline instead of a library call.
8251 @itemx -mno-fast-fix
8253 @opindex mno-fast-fix
8254 The C3x/C4x FIX instruction to convert a floating point value to an
8255 integer value chooses the nearest integer less than or equal to the
8256 floating point value rather than to the nearest integer. Thus if the
8257 floating point number is negative, the result will be incorrectly
8258 truncated an additional code is necessary to detect and correct this
8259 case. This option can be used to disable generation of the additional
8260 code required to correct the result.
8266 Enable (disable) generation of repeat block sequences using the RPTB
8267 instruction for zero overhead looping. The RPTB construct is only used
8268 for innermost loops that do not call functions or jump across the loop
8269 boundaries. There is no advantage having nested RPTB loops due to the
8270 overhead required to save and restore the RC, RS, and RE registers.
8271 This is enabled by default with -O2.
8273 @item -mrpts=@var{count}
8277 Enable (disable) the use of the single instruction repeat instruction
8278 RPTS. If a repeat block contains a single instruction, and the loop
8279 count can be guaranteed to be less than the value @var{count}, GCC will
8280 emit a RPTS instruction instead of a RPTB. If no value is specified,
8281 then a RPTS will be emitted even if the loop count cannot be determined
8282 at compile time. Note that the repeated instruction following RPTS does
8283 not have to be reloaded from memory each iteration, thus freeing up the
8284 CPU buses for operands. However, since interrupts are blocked by this
8285 instruction, it is disabled by default.
8287 @item -mloop-unsigned
8288 @itemx -mno-loop-unsigned
8289 @opindex mloop-unsigned
8290 @opindex mno-loop-unsigned
8291 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
8292 is 2^31 + 1 since these instructions test if the iteration count is
8293 negative to terminate the loop. If the iteration count is unsigned
8294 there is a possibility than the 2^31 + 1 maximum iteration count may be
8295 exceeded. This switch allows an unsigned iteration count.
8299 Try to emit an assembler syntax that the TI assembler (asm30) is happy
8300 with. This also enforces compatibility with the API employed by the TI
8301 C3x C compiler. For example, long doubles are passed as structures
8302 rather than in floating point registers.
8308 Generate code that uses registers (stack) for passing arguments to functions.
8309 By default, arguments are passed in registers where possible rather
8310 than by pushing arguments on to the stack.
8312 @item -mparallel-insns
8313 @itemx -mno-parallel-insns
8314 @opindex mparallel-insns
8315 @opindex mno-parallel-insns
8316 Allow the generation of parallel instructions. This is enabled by
8319 @item -mparallel-mpy
8320 @itemx -mno-parallel-mpy
8321 @opindex mparallel-mpy
8322 @opindex mno-parallel-mpy
8323 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
8324 provided -mparallel-insns is also specified. These instructions have
8325 tight register constraints which can pessimize the code generation
8331 @subsection V850 Options
8332 @cindex V850 Options
8334 These @samp{-m} options are defined for V850 implementations:
8338 @itemx -mno-long-calls
8339 @opindex mlong-calls
8340 @opindex mno-long-calls
8341 Treat all calls as being far away (near). If calls are assumed to be
8342 far away, the compiler will always load the functions address up into a
8343 register, and call indirect through the pointer.
8349 Do not optimize (do optimize) basic blocks that use the same index
8350 pointer 4 or more times to copy pointer into the @code{ep} register, and
8351 use the shorter @code{sld} and @code{sst} instructions. The @samp{-mep}
8352 option is on by default if you optimize.
8354 @item -mno-prolog-function
8355 @itemx -mprolog-function
8356 @opindex mno-prolog-function
8357 @opindex mprolog-function
8358 Do not use (do use) external functions to save and restore registers at
8359 the prolog and epilog of a function. The external functions are slower,
8360 but use less code space if more than one function saves the same number
8361 of registers. The @samp{-mprolog-function} option is on by default if
8366 Try to make the code as small as possible. At present, this just turns
8367 on the @samp{-mep} and @samp{-mprolog-function} options.
8371 Put static or global variables whose size is @var{n} bytes or less into
8372 the tiny data area that register @code{ep} points to. The tiny data
8373 area can hold up to 256 bytes in total (128 bytes for byte references).
8377 Put static or global variables whose size is @var{n} bytes or less into
8378 the small data area that register @code{gp} points to. The small data
8379 area can hold up to 64 kilobytes.
8383 Put static or global variables whose size is @var{n} bytes or less into
8384 the first 32 kilobytes of memory.
8388 Specify that the target processor is the V850.
8391 @opindex mbig-switch
8392 Generate code suitable for big switch tables. Use this option only if
8393 the assembler/linker complain about out of range branches within a switch
8398 @subsection ARC Options
8401 These options are defined for ARC implementations:
8406 Compile code for little endian mode. This is the default.
8410 Compile code for big endian mode.
8413 @opindex mmangle-cpu
8414 Prepend the name of the cpu to all public symbol names.
8415 In multiple-processor systems, there are many ARC variants with different
8416 instruction and register set characteristics. This flag prevents code
8417 compiled for one cpu to be linked with code compiled for another.
8418 No facility exists for handling variants that are "almost identical".
8419 This is an all or nothing option.
8421 @item -mcpu=@var{cpu}
8423 Compile code for ARC variant @var{cpu}.
8424 Which variants are supported depend on the configuration.
8425 All variants support @samp{-mcpu=base}, this is the default.
8427 @item -mtext=@var{text section}
8428 @itemx -mdata=@var{data section}
8429 @itemx -mrodata=@var{readonly data section}
8433 Put functions, data, and readonly data in @var{text section},
8434 @var{data section}, and @var{readonly data section} respectively
8435 by default. This can be overridden with the @code{section} attribute.
8436 @xref{Variable Attributes}.
8441 @subsection NS32K Options
8442 @cindex NS32K options
8444 These are the @samp{-m} options defined for the 32000 series. The default
8445 values for these options depends on which style of 32000 was selected when
8446 the compiler was configured; the defaults for the most common choices are
8454 Generate output for a 32032. This is the default
8455 when the compiler is configured for 32032 and 32016 based systems.
8461 Generate output for a 32332. This is the default
8462 when the compiler is configured for 32332-based systems.
8468 Generate output for a 32532. This is the default
8469 when the compiler is configured for 32532-based systems.
8473 Generate output containing 32081 instructions for floating point.
8474 This is the default for all systems.
8478 Generate output containing 32381 instructions for floating point. This
8479 also implies @samp{-m32081}. The 32381 is only compatible with the 32332
8480 and 32532 cpus. This is the default for the pc532-netbsd configuration.
8484 Try and generate multiply-add floating point instructions @code{polyF}
8485 and @code{dotF}. This option is only available if the @samp{-m32381}
8486 option is in effect. Using these instructions requires changes to to
8487 register allocation which generally has a negative impact on
8488 performance. This option should only be enabled when compiling code
8489 particularly likely to make heavy use of multiply-add instructions.
8492 @opindex mnomulti-add
8493 Do not try and generate multiply-add floating point instructions
8494 @code{polyF} and @code{dotF}. This is the default on all platforms.
8497 @opindex msoft-float
8498 Generate output containing library calls for floating point.
8499 @strong{Warning:} the requisite libraries may not be available.
8502 @opindex mnobitfield
8503 Do not use the bit-field instructions. On some machines it is faster to
8504 use shifting and masking operations. This is the default for the pc532.
8508 Do use the bit-field instructions. This is the default for all platforms
8513 Use a different function-calling convention, in which functions
8514 that take a fixed number of arguments return pop their
8515 arguments on return with the @code{ret} instruction.
8517 This calling convention is incompatible with the one normally
8518 used on Unix, so you cannot use it if you need to call libraries
8519 compiled with the Unix compiler.
8521 Also, you must provide function prototypes for all functions that
8522 take variable numbers of arguments (including @code{printf});
8523 otherwise incorrect code will be generated for calls to those
8526 In addition, seriously incorrect code will result if you call a
8527 function with too many arguments. (Normally, extra arguments are
8528 harmlessly ignored.)
8530 This option takes its name from the 680x0 @code{rtd} instruction.
8535 Use a different function-calling convention where the first two arguments
8536 are passed in registers.
8538 This calling convention is incompatible with the one normally
8539 used on Unix, so you cannot use it if you need to call libraries
8540 compiled with the Unix compiler.
8543 @opindex mnoregparam
8544 Do not pass any arguments in registers. This is the default for all
8549 It is OK to use the sb as an index register which is always loaded with
8550 zero. This is the default for the pc532-netbsd target.
8554 The sb register is not available for use or has not been initialized to
8555 zero by the run time system. This is the default for all targets except
8556 the pc532-netbsd. It is also implied whenever @samp{-mhimem} or
8557 @samp{-fpic} is set.
8561 Many ns32000 series addressing modes use displacements of up to 512MB.
8562 If an address is above 512MB then displacements from zero can not be used.
8563 This option causes code to be generated which can be loaded above 512MB.
8564 This may be useful for operating systems or ROM code.
8568 Assume code will be loaded in the first 512MB of virtual address space.
8569 This is the default for all platforms.
8575 @subsection AVR Options
8578 These options are defined for AVR implementations:
8581 @item -mmcu=@var{mcu}
8583 Specify ATMEL AVR instruction set or MCU type.
8585 Instruction set avr1 is for the minimal AVR core, not supported by the C
8586 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
8587 attiny11, attiny12, attiny15, attiny28).
8589 Instruction set avr2 (default) is for the classic AVR core with up to
8590 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
8591 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
8592 at90c8534, at90s8535).
8594 Instruction set avr3 is for the classic AVR core with up to 128K program
8595 memory space (MCU types: atmega103, atmega603).
8597 Instruction set avr4 is for the enhanced AVR core with up to 8K program
8598 memory space (MCU types: atmega83, atmega85).
8600 Instruction set avr5 is for the enhanced AVR core with up to 128K program
8601 memory space (MCU types: atmega161, atmega163, atmega32, at94k).
8605 Output instruction sizes to the asm file.
8607 @item -minit-stack=@var{N}
8608 @opindex minit-stack
8609 Specify the initial stack address, which may be a symbol or numeric value,
8610 __stack is the default.
8612 @item -mno-interrupts
8613 @opindex mno-interrupts
8614 Generated code is not compatible with hardware interrupts.
8615 Code size will be smaller.
8617 @item -mcall-prologues
8618 @opindex mcall-prologues
8619 Functions prologues/epilogues expanded as call to appropriate
8620 subroutines. Code size will be smaller.
8622 @item -mno-tablejump
8623 @opindex mno-tablejump
8624 Do not generate tablejump insns which sometimes increase code size.
8627 @opindex mtiny-stack
8628 Change only the low 8 bits of the stack pointer.
8632 @subsection MCore Options
8633 @cindex MCore options
8635 These are the @samp{-m} options defined for the Motorola M*Core
8645 @opindex mno-hardlit
8646 Inline constants into the code stream if it can be done in two
8647 instructions or less.
8655 Use the divide instruction. (Enabled by default).
8657 @item -mrelax-immediate
8658 @itemx -mrelax-immediate
8659 @itemx -mno-relax-immediate
8660 @opindex mrelax-immediate
8661 @opindex mrelax-immediate
8662 @opindex mno-relax-immediate
8663 Allow arbitrary sized immediates in bit operations.
8665 @item -mwide-bitfields
8666 @itemx -mwide-bitfields
8667 @itemx -mno-wide-bitfields
8668 @opindex mwide-bitfields
8669 @opindex mwide-bitfields
8670 @opindex mno-wide-bitfields
8671 Always treat bitfields as int-sized.
8673 @item -m4byte-functions
8674 @itemx -m4byte-functions
8675 @itemx -mno-4byte-functions
8676 @opindex m4byte-functions
8677 @opindex m4byte-functions
8678 @opindex mno-4byte-functions
8679 Force all functions to be aligned to a four byte boundary.
8681 @item -mcallgraph-data
8682 @itemx -mcallgraph-data
8683 @itemx -mno-callgraph-data
8684 @opindex mcallgraph-data
8685 @opindex mcallgraph-data
8686 @opindex mno-callgraph-data
8687 Emit callgraph information.
8691 @itemx -mno-slow-bytes
8692 @opindex mslow-bytes
8693 @opindex mslow-bytes
8694 @opindex mno-slow-bytes
8695 Prefer word access when reading byte quantities.
8697 @item -mlittle-endian
8698 @itemx -mlittle-endian
8700 @opindex mlittle-endian
8701 @opindex mlittle-endian
8702 @opindex mbig-endian
8703 Generate code for a little endian target.
8711 Generate code for the 210 processor.
8715 @subsection IA-64 Options
8716 @cindex IA-64 Options
8718 These are the @samp{-m} options defined for the Intel IA-64 architecture.
8722 @opindex mbig-endian
8723 Generate code for a big endian target. This is the default for HPUX.
8725 @item -mlittle-endian
8726 @opindex mlittle-endian
8727 Generate code for a little endian target. This is the default for AIX5
8734 Generate (or don't) code for the GNU assembler. This is the default.
8735 @c Also, this is the default if the configure option @samp{--with-gnu-as}
8742 Generate (or don't) code for the GNU linker. This is the default.
8743 @c Also, this is the default if the configure option @samp{--with-gnu-ld}
8748 Generate code that does not use a global pointer register. The result
8749 is not position independent code, and violates the IA-64 ABI.
8751 @item -mvolatile-asm-stop
8752 @itemx -mno-volatile-asm-stop
8753 @opindex mvolatile-asm-stop
8754 @opindex mno-volatile-asm-stop
8755 Generate (or don't) a stop bit immediately before and after volatile asm
8760 Generate code that works around Itanium B step errata.
8762 @item -mregister-names
8763 @itemx -mno-register-names
8764 @opindex mregister-names
8765 @opindex mno-register-names
8766 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
8767 the stacked registers. This may make assembler output more readable.
8773 Disable (or enable) optimizations that use the small data section. This may
8774 be useful for working around optimizer bugs.
8777 @opindex mconstant-gp
8778 Generate code that uses a single constant global pointer value. This is
8779 useful when compiling kernel code.
8783 Generate code that is self-relocatable. This implies @samp{-mconstant-gp}.
8784 This is useful when compiling firmware code.
8786 @item -minline-divide-min-latency
8787 @opindex minline-divide-min-latency
8788 Generate code for inline divides using the minimum latency algorithm.
8790 @item -minline-divide-max-throughput
8791 @opindex minline-divide-max-throughput
8792 Generate code for inline divides using the maximum throughput algorithm.
8794 @item -mno-dwarf2-asm
8796 @opindex mno-dwarf2-asm
8797 @opindex mdwarf2-asm
8798 Don't (or do) generate assembler code for the DWARF2 line number debugging
8799 info. This may be useful when not using the GNU assembler.
8801 @item -mfixed-range=@var{register range}
8802 @opindex mfixed-range
8803 Generate code treating the given register range as fixed registers.
8804 A fixed register is one that the register allocator can not use. This is
8805 useful when compiling kernel code. A register range is specified as
8806 two registers separated by a dash. Multiple register ranges can be
8807 specified separated by a comma.
8811 @subsection D30V Options
8812 @cindex D30V Options
8814 These @samp{-m} options are defined for D30V implementations:
8819 Link the @samp{.text}, @samp{.data}, @samp{.bss}, @samp{.strings},
8820 @samp{.rodata}, @samp{.rodata1}, @samp{.data1} sections into external
8821 memory, which starts at location @code{0x80000000}.
8825 Same as the @samp{-mextmem} switch.
8829 Link the @samp{.text} section into onchip text memory, which starts at
8830 location @code{0x0}. Also link @samp{.data}, @samp{.bss},
8831 @samp{.strings}, @samp{.rodata}, @samp{.rodata1}, @samp{.data1} sections
8832 into onchip data memory, which starts at location @code{0x20000000}.
8834 @item -mno-asm-optimize
8835 @itemx -masm-optimize
8836 @opindex mno-asm-optimize
8837 @opindex masm-optimize
8838 Disable (enable) passing @samp{-O} to the assembler when optimizing.
8839 The assembler uses the @samp{-O} option to automatically parallelize
8840 adjacent short instructions where possible.
8842 @item -mbranch-cost=@var{n}
8843 @opindex mbranch-cost
8844 Increase the internal costs of branches to @var{n}. Higher costs means
8845 that the compiler will issue more instructions to avoid doing a branch.
8848 @item -mcond-exec=@var{n}
8850 Specify the maximum number of conditionally executed instructions that
8851 replace a branch. The default is 4.
8854 @node Code Gen Options
8855 @section Options for Code Generation Conventions
8856 @cindex code generation conventions
8857 @cindex options, code generation
8858 @cindex run-time options
8860 These machine-independent options control the interface conventions
8861 used in code generation.
8863 Most of them have both positive and negative forms; the negative form
8864 of @samp{-ffoo} would be @samp{-fno-foo}. In the table below, only
8865 one of the forms is listed---the one which is not the default. You
8866 can figure out the other form by either removing @samp{no-} or adding
8871 @opindex fexceptions
8872 Enable exception handling. Generates extra code needed to propagate
8873 exceptions. For some targets, this implies GNU CC will generate frame
8874 unwind information for all functions, which can produce significant data
8875 size overhead, although it does not affect execution. If you do not
8876 specify this option, GNU CC will enable it by default for languages like
8877 C++ which normally require exception handling, and disable it for
8878 languages like C that do not normally require it. However, you may need
8879 to enable this option when compiling C code that needs to interoperate
8880 properly with exception handlers written in C++. You may also wish to
8881 disable this option if you are compiling older C++ programs that don't
8882 use exception handling.
8884 @item -funwind-tables
8885 @opindex funwind-tables
8886 Similar to @option{-fexceptions}, except that it will just generate any needed
8887 static data, but will not affect the generated code in any other way.
8888 You will normally not enable this option; instead, a language processor
8889 that needs this handling would enable it on your behalf.
8891 @item -fpcc-struct-return
8892 @opindex fpcc-struct-return
8893 Return ``short'' @code{struct} and @code{union} values in memory like
8894 longer ones, rather than in registers. This convention is less
8895 efficient, but it has the advantage of allowing intercallability between
8896 GCC-compiled files and files compiled with other compilers.
8898 The precise convention for returning structures in memory depends
8899 on the target configuration macros.
8901 Short structures and unions are those whose size and alignment match
8902 that of some integer type.
8904 @item -freg-struct-return
8905 @opindex freg-struct-return
8906 Use the convention that @code{struct} and @code{union} values are
8907 returned in registers when possible. This is more efficient for small
8908 structures than @samp{-fpcc-struct-return}.
8910 If you specify neither @samp{-fpcc-struct-return} nor its contrary
8911 @samp{-freg-struct-return}, GCC defaults to whichever convention is
8912 standard for the target. If there is no standard convention, GCC
8913 defaults to @samp{-fpcc-struct-return}, except on targets where GCC
8914 is the principal compiler. In those cases, we can choose the standard,
8915 and we chose the more efficient register return alternative.
8918 @opindex fshort-enums
8919 Allocate to an @code{enum} type only as many bytes as it needs for the
8920 declared range of possible values. Specifically, the @code{enum} type
8921 will be equivalent to the smallest integer type which has enough room.
8923 @item -fshort-double
8924 @opindex fshort-double
8925 Use the same size for @code{double} as for @code{float}.
8928 @opindex fshared-data
8929 Requests that the data and non-@code{const} variables of this
8930 compilation be shared data rather than private data. The distinction
8931 makes sense only on certain operating systems, where shared data is
8932 shared between processes running the same program, while private data
8933 exists in one copy per process.
8937 In C, allocate even uninitialized global variables in the data section of the
8938 object file, rather than generating them as common blocks. This has the
8939 effect that if the same variable is declared (without @code{extern}) in
8940 two different compilations, you will get an error when you link them.
8941 The only reason this might be useful is if you wish to verify that the
8942 program will work on other systems which always work this way.
8946 Ignore the @samp{#ident} directive.
8948 @item -fno-gnu-linker
8949 @opindex fno-gnu-linker
8950 Do not output global initializations (such as C++ constructors and
8951 destructors) in the form used by the GNU linker (on systems where the GNU
8952 linker is the standard method of handling them). Use this option when
8953 you want to use a non-GNU linker, which also requires using the
8954 @command{collect2} program to make sure the system linker includes
8955 constructors and destructors. (@command{collect2} is included in the GCC
8956 distribution.) For systems which @emph{must} use @command{collect2}, the
8957 compiler driver @command{gcc} is configured to do this automatically.
8959 @item -finhibit-size-directive
8960 @opindex finhibit-size-directive
8961 Don't output a @code{.size} assembler directive, or anything else that
8962 would cause trouble if the function is split in the middle, and the
8963 two halves are placed at locations far apart in memory. This option is
8964 used when compiling @file{crtstuff.c}; you should not need to use it
8968 @opindex fverbose-asm
8969 Put extra commentary information in the generated assembly code to
8970 make it more readable. This option is generally only of use to those
8971 who actually need to read the generated assembly code (perhaps while
8972 debugging the compiler itself).
8974 @samp{-fno-verbose-asm}, the default, causes the
8975 extra information to be omitted and is useful when comparing two assembler
8980 Consider all memory references through pointers to be volatile.
8982 @item -fvolatile-global
8983 @opindex fvolatile-global
8984 Consider all memory references to extern and global data items to
8985 be volatile. GCC does not consider static data items to be volatile
8986 because of this switch.
8988 @item -fvolatile-static
8989 @opindex fvolatile-static
8990 Consider all memory references to static data to be volatile.
8994 @cindex global offset table
8996 Generate position-independent code (PIC) suitable for use in a shared
8997 library, if supported for the target machine. Such code accesses all
8998 constant addresses through a global offset table (GOT). The dynamic
8999 loader resolves the GOT entries when the program starts (the dynamic
9000 loader is not part of GCC; it is part of the operating system). If
9001 the GOT size for the linked executable exceeds a machine-specific
9002 maximum size, you get an error message from the linker indicating that
9003 @samp{-fpic} does not work; in that case, recompile with @samp{-fPIC}
9004 instead. (These maximums are 16k on the m88k, 8k on the Sparc, and 32k
9005 on the m68k and RS/6000. The 386 has no such limit.)
9007 Position-independent code requires special support, and therefore works
9008 only on certain machines. For the 386, GCC supports PIC for System V
9009 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
9010 position-independent.
9014 If supported for the target machine, emit position-independent code,
9015 suitable for dynamic linking and avoiding any limit on the size of the
9016 global offset table. This option makes a difference on the m68k, m88k,
9019 Position-independent code requires special support, and therefore works
9020 only on certain machines.
9022 @item -ffixed-@var{reg}
9024 Treat the register named @var{reg} as a fixed register; generated code
9025 should never refer to it (except perhaps as a stack pointer, frame
9026 pointer or in some other fixed role).
9028 @var{reg} must be the name of a register. The register names accepted
9029 are machine-specific and are defined in the @code{REGISTER_NAMES}
9030 macro in the machine description macro file.
9032 This flag does not have a negative form, because it specifies a
9035 @item -fcall-used-@var{reg}
9037 Treat the register named @var{reg} as an allocable register that is
9038 clobbered by function calls. It may be allocated for temporaries or
9039 variables that do not live across a call. Functions compiled this way
9040 will not save and restore the register @var{reg}.
9042 It is an error to used this flag with the frame pointer or stack pointer.
9043 Use of this flag for other registers that have fixed pervasive roles in
9044 the machine's execution model will produce disastrous results.
9046 This flag does not have a negative form, because it specifies a
9049 @item -fcall-saved-@var{reg}
9050 @opindex fcall-saved
9051 Treat the register named @var{reg} as an allocable register saved by
9052 functions. It may be allocated even for temporaries or variables that
9053 live across a call. Functions compiled this way will save and restore
9054 the register @var{reg} if they use it.
9056 It is an error to used this flag with the frame pointer or stack pointer.
9057 Use of this flag for other registers that have fixed pervasive roles in
9058 the machine's execution model will produce disastrous results.
9060 A different sort of disaster will result from the use of this flag for
9061 a register in which function values may be returned.
9063 This flag does not have a negative form, because it specifies a
9067 @opindex fpack-struct
9068 Pack all structure members together without holes. Usually you would
9069 not want to use this option, since it makes the code suboptimal, and
9070 the offsets of structure members won't agree with system libraries.
9072 @item -fcheck-memory-usage
9073 @opindex fcheck-memory-usage
9074 Generate extra code to check each memory access. GCC will generate
9075 code that is suitable for a detector of bad memory accesses such as
9078 Normally, you should compile all, or none, of your code with this option.
9080 If you do mix code compiled with and without this option,
9081 you must ensure that all code that has side effects
9082 and that is called by code compiled with this option
9083 is, itself, compiled with this option.
9084 If you do not, you might get erroneous messages from the detector.
9086 If you use functions from a library that have side-effects (such as
9087 @code{read}), you might not be able to recompile the library and
9088 specify this option. In that case, you can enable the
9089 @samp{-fprefix-function-name} option, which requests GCC to encapsulate
9090 your code and make other functions look as if they were compiled with
9091 @samp{-fcheck-memory-usage}. This is done by calling ``stubs'',
9092 which are provided by the detector. If you cannot find or build
9093 stubs for every function you call, you might have to specify
9094 @samp{-fcheck-memory-usage} without @samp{-fprefix-function-name}.
9096 If you specify this option, you can not use the @code{asm} or
9097 @code{__asm__} keywords in functions with memory checking enabled. GNU
9098 CC cannot understand what the @code{asm} statement may do, and therefore
9099 cannot generate the appropriate code, so it will reject it. However, if
9100 you specify the function attribute @code{no_check_memory_usage}
9101 (@pxref{Function Attributes}), GNU CC will disable memory checking within a
9102 function; you may use @code{asm} statements inside such functions. You
9103 may have an inline expansion of a non-checked function within a checked
9104 function; in that case GNU CC will not generate checks for the inlined
9105 function's memory accesses.
9107 If you move your @code{asm} statements to non-checked inline functions
9108 and they do access memory, you can add calls to the support code in your
9109 inline function, to indicate any reads, writes, or copies being done.
9110 These calls would be similar to those done in the stubs described above.
9112 @item -fprefix-function-name
9113 @opindex fprefix-function-name
9114 Request GCC to add a prefix to the symbols generated for function names.
9115 GCC adds a prefix to the names of functions defined as well as
9116 functions called. Code compiled with this option and code compiled
9117 without the option can't be linked together, unless stubs are used.
9119 If you compile the following code with @samp{-fprefix-function-name}
9121 extern void bar (int);
9130 GCC will compile the code as if it was written:
9132 extern void prefix_bar (int);
9136 return prefix_bar (a + 5);
9139 This option is designed to be used with @samp{-fcheck-memory-usage}.
9141 @item -finstrument-functions
9142 @opindex finstrument-functions
9143 Generate instrumentation calls for entry and exit to functions. Just
9144 after function entry and just before function exit, the following
9145 profiling functions will be called with the address of the current
9146 function and its call site. (On some platforms,
9147 @code{__builtin_return_address} does not work beyond the current
9148 function, so the call site information may not be available to the
9149 profiling functions otherwise.)
9152 void __cyg_profile_func_enter (void *this_fn,
9154 void __cyg_profile_func_exit (void *this_fn,
9158 The first argument is the address of the start of the current function,
9159 which may be looked up exactly in the symbol table.
9161 This instrumentation is also done for functions expanded inline in other
9162 functions. The profiling calls will indicate where, conceptually, the
9163 inline function is entered and exited. This means that addressable
9164 versions of such functions must be available. If all your uses of a
9165 function are expanded inline, this may mean an additional expansion of
9166 code size. If you use @samp{extern inline} in your C code, an
9167 addressable version of such functions must be provided. (This is
9168 normally the case anyways, but if you get lucky and the optimizer always
9169 expands the functions inline, you might have gotten away without
9170 providing static copies.)
9172 A function may be given the attribute @code{no_instrument_function}, in
9173 which case this instrumentation will not be done. This can be used, for
9174 example, for the profiling functions listed above, high-priority
9175 interrupt routines, and any functions from which the profiling functions
9176 cannot safely be called (perhaps signal handlers, if the profiling
9177 routines generate output or allocate memory).
9180 @opindex fstack-check
9181 Generate code to verify that you do not go beyond the boundary of the
9182 stack. You should specify this flag if you are running in an
9183 environment with multiple threads, but only rarely need to specify it in
9184 a single-threaded environment since stack overflow is automatically
9185 detected on nearly all systems if there is only one stack.
9187 Note that this switch does not actually cause checking to be done; the
9188 operating system must do that. The switch causes generation of code
9189 to ensure that the operating system sees the stack being extended.
9191 @item -fstack-limit-register=@var{reg}
9192 @itemx -fstack-limit-symbol=@var{sym}
9193 @itemx -fno-stack-limit
9194 @opindex fstack-limit-register
9195 @opindex fstack-limit-symbol
9196 @opindex fno-stack-limit
9197 Generate code to ensure that the stack does not grow beyond a certain value,
9198 either the value of a register or the address of a symbol. If the stack
9199 would grow beyond the value, a signal is raised. For most targets,
9200 the signal is raised before the stack overruns the boundary, so
9201 it is possible to catch the signal without taking special precautions.
9203 For instance, if the stack starts at address @samp{0x80000000} and grows
9204 downwards you can use the flags
9205 @samp{-fstack-limit-symbol=__stack_limit}
9206 @samp{-Wl,--defsym,__stack_limit=0x7ffe0000} which will enforce a stack
9209 @cindex aliasing of parameters
9210 @cindex parameters, aliased
9211 @item -fargument-alias
9212 @itemx -fargument-noalias
9213 @itemx -fargument-noalias-global
9214 @opindex fargument-alias
9215 @opindex fargument-noalias
9216 @opindex fargument-noalias-global
9217 Specify the possible relationships among parameters and between
9218 parameters and global data.
9220 @samp{-fargument-alias} specifies that arguments (parameters) may
9221 alias each other and may alias global storage.
9222 @samp{-fargument-noalias} specifies that arguments do not alias
9223 each other, but may alias global storage.
9224 @samp{-fargument-noalias-global} specifies that arguments do not
9225 alias each other and do not alias global storage.
9227 Each language will automatically use whatever option is required by
9228 the language standard. You should not need to use these options yourself.
9230 @item -fleading-underscore
9231 @opindex fleading-underscore
9232 This option and its counterpart, -fno-leading-underscore, forcibly
9233 change the way C symbols are represented in the object file. One use
9234 is to help link with legacy assembly code.
9236 Be warned that you should know what you are doing when invoking this
9237 option, and that not all targets provide complete support for it.
9242 @node Environment Variables
9243 @section Environment Variables Affecting GCC
9244 @cindex environment variables
9246 @c man begin ENVIRONMENT
9248 This section describes several environment variables that affect how GCC
9249 operates. Some of them work by specifying directories or prefixes to use
9250 when searching for various kinds of files. Some are used to specify other
9251 aspects of the compilation environment.
9254 Note that you can also specify places to search using options such as
9255 @samp{-B}, @samp{-I} and @samp{-L} (@pxref{Directory Options}). These
9256 take precedence over places specified using environment variables, which
9257 in turn take precedence over those specified by the configuration of GCC.
9261 Note that you can also specify places to search using options such as
9262 @samp{-B}, @samp{-I} and @samp{-L} (@pxref{Directory Options}). These
9263 take precedence over places specified using environment variables, which
9264 in turn take precedence over those specified by the configuration of GCC.
9271 @c @itemx LC_COLLATE
9273 @c @itemx LC_MONETARY
9274 @c @itemx LC_NUMERIC
9279 @c @findex LC_COLLATE
9281 @c @findex LC_MONETARY
9282 @c @findex LC_NUMERIC
9286 These environment variables control the way that GCC uses
9287 localization information that allow GCC to work with different
9288 national conventions. GCC inspects the locale categories
9289 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
9290 so. These locale categories can be set to any value supported by your
9291 installation. A typical value is @samp{en_UK} for English in the United
9294 The @env{LC_CTYPE} environment variable specifies character
9295 classification. GCC uses it to determine the character boundaries in
9296 a string; this is needed for some multibyte encodings that contain quote
9297 and escape characters that would otherwise be interpreted as a string
9300 The @env{LC_MESSAGES} environment variable specifies the language to
9301 use in diagnostic messages.
9303 If the @env{LC_ALL} environment variable is set, it overrides the value
9304 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
9305 and @env{LC_MESSAGES} default to the value of the @env{LANG}
9306 environment variable. If none of these variables are set, GCC
9307 defaults to traditional C English behavior.
9311 If @env{TMPDIR} is set, it specifies the directory to use for temporary
9312 files. GCC uses temporary files to hold the output of one stage of
9313 compilation which is to be used as input to the next stage: for example,
9314 the output of the preprocessor, which is the input to the compiler
9317 @item GCC_EXEC_PREFIX
9318 @findex GCC_EXEC_PREFIX
9319 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
9320 names of the subprograms executed by the compiler. No slash is added
9321 when this prefix is combined with the name of a subprogram, but you can
9322 specify a prefix that ends with a slash if you wish.
9324 If @env{GCC_EXEC_PREFIX} is not set, GNU CC will attempt to figure out
9325 an appropriate prefix to use based on the pathname it was invoked with.
9327 If GCC cannot find the subprogram using the specified prefix, it
9328 tries looking in the usual places for the subprogram.
9330 The default value of @env{GCC_EXEC_PREFIX} is
9331 @file{@var{prefix}/lib/gcc-lib/} where @var{prefix} is the value
9332 of @code{prefix} when you ran the @file{configure} script.
9334 Other prefixes specified with @samp{-B} take precedence over this prefix.
9336 This prefix is also used for finding files such as @file{crt0.o} that are
9339 In addition, the prefix is used in an unusual way in finding the
9340 directories to search for header files. For each of the standard
9341 directories whose name normally begins with @samp{/usr/local/lib/gcc-lib}
9342 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
9343 replacing that beginning with the specified prefix to produce an
9344 alternate directory name. Thus, with @samp{-Bfoo/}, GCC will search
9345 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
9346 These alternate directories are searched first; the standard directories
9350 @findex COMPILER_PATH
9351 The value of @env{COMPILER_PATH} is a colon-separated list of
9352 directories, much like @env{PATH}. GCC tries the directories thus
9353 specified when searching for subprograms, if it can't find the
9354 subprograms using @env{GCC_EXEC_PREFIX}.
9357 @findex LIBRARY_PATH
9358 The value of @env{LIBRARY_PATH} is a colon-separated list of
9359 directories, much like @env{PATH}. When configured as a native compiler,
9360 GCC tries the directories thus specified when searching for special
9361 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
9362 using GCC also uses these directories when searching for ordinary
9363 libraries for the @samp{-l} option (but directories specified with
9364 @samp{-L} come first).
9366 @item C_INCLUDE_PATH
9367 @itemx CPLUS_INCLUDE_PATH
9368 @itemx OBJC_INCLUDE_PATH
9369 @findex C_INCLUDE_PATH
9370 @findex CPLUS_INCLUDE_PATH
9371 @findex OBJC_INCLUDE_PATH
9372 @c @itemx OBJCPLUS_INCLUDE_PATH
9373 These environment variables pertain to particular languages. Each
9374 variable's value is a colon-separated list of directories, much like
9375 @env{PATH}. When GCC searches for header files, it tries the
9376 directories listed in the variable for the language you are using, after
9377 the directories specified with @samp{-I} but before the standard header
9380 @item DEPENDENCIES_OUTPUT
9381 @findex DEPENDENCIES_OUTPUT
9382 @cindex dependencies for make as output
9383 If this variable is set, its value specifies how to output dependencies
9384 for Make based on the header files processed by the compiler. This
9385 output looks much like the output from the @samp{-M} option
9386 (@pxref{Preprocessor Options}), but it goes to a separate file, and is
9387 in addition to the usual results of compilation.
9389 The value of @env{DEPENDENCIES_OUTPUT} can be just a file name, in
9390 which case the Make rules are written to that file, guessing the target
9391 name from the source file name. Or the value can have the form
9392 @samp{@var{file} @var{target}}, in which case the rules are written to
9393 file @var{file} using @var{target} as the target name.
9397 @cindex locale definition
9398 This variable is used to pass locale information to the compiler. One way in
9399 which this information is used is to determine the character set to be used
9400 when character literals, string literals and comments are parsed in C and C++.
9401 When the compiler is configured to allow multibyte characters,
9402 the following values for @env{LANG} are recognized:
9406 Recognize JIS characters.
9408 Recognize SJIS characters.
9410 Recognize EUCJP characters.
9413 If @env{LANG} is not defined, or if it has some other value, then the
9414 compiler will use mblen and mbtowc as defined by the default locale to
9415 recognize and translate multibyte characters.
9420 @node Running Protoize
9421 @section Running Protoize
9423 The program @code{protoize} is an optional part of GNU C. You can use
9424 it to add prototypes to a program, thus converting the program to ISO
9425 C in one respect. The companion program @code{unprotoize} does the
9426 reverse: it removes argument types from any prototypes that are found.
9428 When you run these programs, you must specify a set of source files as
9429 command line arguments. The conversion programs start out by compiling
9430 these files to see what functions they define. The information gathered
9431 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
9433 After scanning comes actual conversion. The specified files are all
9434 eligible to be converted; any files they include (whether sources or
9435 just headers) are eligible as well.
9437 But not all the eligible files are converted. By default,
9438 @code{protoize} and @code{unprotoize} convert only source and header
9439 files in the current directory. You can specify additional directories
9440 whose files should be converted with the @samp{-d @var{directory}}
9441 option. You can also specify particular files to exclude with the
9442 @samp{-x @var{file}} option. A file is converted if it is eligible, its
9443 directory name matches one of the specified directory names, and its
9444 name within the directory has not been excluded.
9446 Basic conversion with @code{protoize} consists of rewriting most
9447 function definitions and function declarations to specify the types of
9448 the arguments. The only ones not rewritten are those for varargs
9451 @code{protoize} optionally inserts prototype declarations at the
9452 beginning of the source file, to make them available for any calls that
9453 precede the function's definition. Or it can insert prototype
9454 declarations with block scope in the blocks where undeclared functions
9457 Basic conversion with @code{unprotoize} consists of rewriting most
9458 function declarations to remove any argument types, and rewriting
9459 function definitions to the old-style pre-ISO form.
9461 Both conversion programs print a warning for any function declaration or
9462 definition that they can't convert. You can suppress these warnings
9465 The output from @code{protoize} or @code{unprotoize} replaces the
9466 original source file. The original file is renamed to a name ending
9467 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
9468 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
9469 for DOS) file already exists, then the source file is simply discarded.
9471 @code{protoize} and @code{unprotoize} both depend on GCC itself to
9472 scan the program and collect information about the functions it uses.
9473 So neither of these programs will work until GCC is installed.
9475 Here is a table of the options you can use with @code{protoize} and
9476 @code{unprotoize}. Each option works with both programs unless
9480 @item -B @var{directory}
9481 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
9482 usual directory (normally @file{/usr/local/lib}). This file contains
9483 prototype information about standard system functions. This option
9484 applies only to @code{protoize}.
9486 @item -c @var{compilation-options}
9487 Use @var{compilation-options} as the options when running @code{gcc} to
9488 produce the @samp{.X} files. The special option @samp{-aux-info} is
9489 always passed in addition, to tell @code{gcc} to write a @samp{.X} file.
9491 Note that the compilation options must be given as a single argument to
9492 @code{protoize} or @code{unprotoize}. If you want to specify several
9493 @code{gcc} options, you must quote the entire set of compilation options
9494 to make them a single word in the shell.
9496 There are certain @code{gcc} arguments that you cannot use, because they
9497 would produce the wrong kind of output. These include @samp{-g},
9498 @samp{-O}, @samp{-c}, @samp{-S}, and @samp{-o} If you include these in
9499 the @var{compilation-options}, they are ignored.
9502 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
9503 systems) instead of @samp{.c}. This is convenient if you are converting
9504 a C program to C++. This option applies only to @code{protoize}.
9507 Add explicit global declarations. This means inserting explicit
9508 declarations at the beginning of each source file for each function
9509 that is called in the file and was not declared. These declarations
9510 precede the first function definition that contains a call to an
9511 undeclared function. This option applies only to @code{protoize}.
9513 @item -i @var{string}
9514 Indent old-style parameter declarations with the string @var{string}.
9515 This option applies only to @code{protoize}.
9517 @code{unprotoize} converts prototyped function definitions to old-style
9518 function definitions, where the arguments are declared between the
9519 argument list and the initial @samp{@{}. By default, @code{unprotoize}
9520 uses five spaces as the indentation. If you want to indent with just
9521 one space instead, use @samp{-i " "}.
9524 Keep the @samp{.X} files. Normally, they are deleted after conversion
9528 Add explicit local declarations. @code{protoize} with @samp{-l} inserts
9529 a prototype declaration for each function in each block which calls the
9530 function without any declaration. This option applies only to
9534 Make no real changes. This mode just prints information about the conversions
9535 that would have been done without @samp{-n}.
9538 Make no @samp{.save} files. The original files are simply deleted.
9539 Use this option with caution.
9541 @item -p @var{program}
9542 Use the program @var{program} as the compiler. Normally, the name
9546 Work quietly. Most warnings are suppressed.
9549 Print the version number, just like @samp{-v} for @code{gcc}.
9552 If you need special compiler options to compile one of your program's
9553 source files, then you should generate that file's @samp{.X} file
9554 specially, by running @code{gcc} on that source file with the
9555 appropriate options and the option @samp{-aux-info}. Then run
9556 @code{protoize} on the entire set of files. @code{protoize} will use
9557 the existing @samp{.X} file because it is newer than the source file.
9561 gcc -Dfoo=bar file1.c -aux-info
9566 You need to include the special files along with the rest in the
9567 @code{protoize} command, even though their @samp{.X} files already
9568 exist, because otherwise they won't get converted.
9570 @xref{Protoize Caveats}, for more information on how to use
9571 @code{protoize} successfully.