1 @c Copyright (C) 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
2 @c 2000, 2001 Free Software Foundation, Inc.
3 @c This is part of the GCC manual.
4 @c For copying conditions, see the file gcc.texi.
8 Copyright @copyright{} 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997,
9 1998, 1999, 2000, 2001 Free Software Foundation, Inc.
11 Permission is granted to make and distribute verbatim copies of this
12 manual provided the copyright notice and this permission notice are
13 preserved on all copies.
15 Permission is granted to copy and distribute modified versions of this
16 manual under the conditions for verbatim copying, provided also that the
17 entire resulting derived work is distributed under the terms of a
18 permission notice identical to this one.
20 Permission is granted to copy and distribute translations of this manual
21 into another language, under the above conditions for modified versions,
22 except that this permission notice may be included in translations
23 approved by the Free Software Foundation instead of in the original
26 @c Set file name and title for the man page.
28 @settitle GNU project C and C++ compiler
30 gcc [@option{-c}|@option{-S}|@option{-E}] [@option{-std=}@var{standard}]
31 [@option{-g}] [@option{-pg}] [@option{-O}@var{level}]
32 [@option{-W}@var{warn}@dots{}] [@option{-pedantic}]
33 [@option{-I}@var{dir}@dots{}] [@option{-L}@var{dir}@dots{}]
34 [@option{-D}@var{macro}[=@var{defn}]@dots{}] [@option{-U}@var{macro}]
35 [@option{-f}@var{option}@dots{}] [@option{-m}@var{machine-option}@dots{}]
36 [@option{-o} @var{outfile}] @var{infile}@dots{}
38 Only the most useful options are listed here; see below for the
39 remainder. @samp{g++} accepts mostly the same options as @samp{gcc}.
42 cpp(1), gcov(1), g77(1), as(1), ld(1), gdb(1), adb(1), dbx(1), sdb(1)
43 and the Info entries for @file{gcc}, @file{cpp}, @file{g77}, @file{as},
44 @file{ld}, @file{binutils} and @file{gdb}.
47 For instructions on reporting bugs, see
48 @w{@uref{http://gcc.gnu.org/bugs.html}}. Use of the @command{gccbug}
49 script to report bugs is recommended.
52 See the Info entry for @file{gcc}, or
53 @w{@uref{http://gcc.gnu.org/thanks.html}}, for contributors to GCC@.
58 @chapter GCC Command Options
59 @cindex GCC command options
60 @cindex command options
61 @cindex options, GCC command
63 @c man begin DESCRIPTION
65 When you invoke GCC, it normally does preprocessing, compilation,
66 assembly and linking. The ``overall options'' allow you to stop this
67 process at an intermediate stage. For example, the @option{-c} option
68 says not to run the linker. Then the output consists of object files
69 output by the assembler.
71 Other options are passed on to one stage of processing. Some options
72 control the preprocessor and others the compiler itself. Yet other
73 options control the assembler and linker; most of these are not
74 documented here, since you rarely need to use any of them.
76 @cindex C compilation options
77 Most of the command line options that you can use with GCC are useful
78 for C programs; when an option is only useful with another language
79 (usually C++), the explanation says so explicitly. If the description
80 for a particular option does not mention a source language, you can use
81 that option with all supported languages.
83 @cindex C++ compilation options
84 @xref{Invoking G++,,Compiling C++ Programs}, for a summary of special
85 options for compiling C++ programs.
87 @cindex grouping options
88 @cindex options, grouping
89 The @command{gcc} program accepts options and file names as operands. Many
90 options have multi-letter names; therefore multiple single-letter options
91 may @emph{not} be grouped: @option{-dr} is very different from @w{@samp{-d
94 @cindex order of options
95 @cindex options, order
96 You can mix options and other arguments. For the most part, the order
97 you use doesn't matter. Order does matter when you use several options
98 of the same kind; for example, if you specify @option{-L} more than once,
99 the directories are searched in the order specified.
101 Many options have long names starting with @samp{-f} or with
102 @samp{-W}---for example, @option{-fforce-mem},
103 @option{-fstrength-reduce}, @option{-Wformat} and so on. Most of
104 these have both positive and negative forms; the negative form of
105 @option{-ffoo} would be @option{-fno-foo}. This manual documents
106 only one of these two forms, whichever one is not the default.
110 @xref{Option Index}, for an index to GCC's options.
113 * Option Summary:: Brief list of all options, without explanations.
114 * Overall Options:: Controlling the kind of output:
115 an executable, object files, assembler files,
116 or preprocessed source.
117 * Invoking G++:: Compiling C++ programs.
118 * C Dialect Options:: Controlling the variant of C language compiled.
119 * C++ Dialect Options:: Variations on C++.
120 * Objective-C Dialect Options:: Variations on Objective-C.
121 * Language Independent Options:: Controlling how diagnostics should be
123 * Warning Options:: How picky should the compiler be?
124 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
125 * Optimize Options:: How much optimization?
126 * Preprocessor Options:: Controlling header files and macro definitions.
127 Also, getting dependency information for Make.
128 * Assembler Options:: Passing options to the assembler.
129 * Link Options:: Specifying libraries and so on.
130 * Directory Options:: Where to find header files and libraries.
131 Where to find the compiler executable files.
132 * Spec Files:: How to pass switches to sub-processes.
133 * Target Options:: Running a cross-compiler, or an old version of GCC.
134 * Submodel Options:: Specifying minor hardware or convention variations,
135 such as 68010 vs 68020.
136 * Code Gen Options:: Specifying conventions for function calls, data layout
138 * Environment Variables:: Env vars that affect GCC.
139 * Running Protoize:: Automatically adding or removing function prototypes.
145 @section Option Summary
147 Here is a summary of all the options, grouped by type. Explanations are
148 in the following sections.
151 @item Overall Options
152 @xref{Overall Options,,Options Controlling the Kind of Output}.
154 -c -S -E -o @var{file} -pipe -pass-exit-codes -x @var{language} @gol
155 -v --target-help --help}
157 @item C Language Options
158 @xref{C Dialect Options,,Options Controlling C Dialect}.
160 -ansi -std=@var{standard} -aux-info @var{filename} @gol
161 -fno-asm -fno-builtin @gol
162 -fhosted -ffreestanding @gol
163 -trigraphs -traditional -traditional-cpp @gol
164 -fallow-single-precision -fcond-mismatch @gol
165 -fsigned-bitfields -fsigned-char @gol
166 -funsigned-bitfields -funsigned-char @gol
167 -fwritable-strings -fshort-wchar}
169 @item C++ Language Options
170 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
172 -fno-access-control -fcheck-new -fconserve-space @gol
173 -fno-const-strings -fdollars-in-identifiers @gol
174 -fno-elide-constructors @gol
175 -fno-enforce-eh-specs -fexternal-templates @gol
176 -falt-external-templates @gol
177 -ffor-scope -fno-for-scope -fno-gnu-keywords @gol
178 -fno-implicit-templates @gol
179 -fno-implicit-inline-templates @gol
180 -fno-implement-inlines -fms-extensions @gol
181 -fno-nonansi-builtins -fno-operator-names @gol
182 -fno-optional-diags -fpermissive @gol
183 -frepo -fno-rtti -fstats -ftemplate-depth-@var{n} @gol
184 -fuse-cxa-atexit -fvtable-gc -fno-weak -nostdinc++ @gol
185 -fno-default-inline -Wctor-dtor-privacy @gol
186 -Wnon-virtual-dtor -Wreorder @gol
187 -Weffc++ -Wno-deprecated @gol
188 -Wno-non-template-friend -Wold-style-cast @gol
189 -Woverloaded-virtual -Wno-pmf-conversions @gol
190 -Wsign-promo -Wsynth}
192 @item Objective-C Language Options
193 @xref{Objective-C Dialect Options,,Options Controlling Objective-C Dialect}.
195 -fconstant-string-class=@var{class-name} @gol
196 -fgnu-runtime -fnext-runtime -gen-decls @gol
197 -Wno-protocol -Wselector}
199 @item Language Independent Options
200 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
202 -fmessage-length=@var{n} @gol
203 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]}}
205 @item Warning Options
206 @xref{Warning Options,,Options to Request or Suppress Warnings}.
208 -fsyntax-only -pedantic -pedantic-errors @gol
209 -w -W -Wall -Waggregate-return @gol
210 -Wcast-align -Wcast-qual -Wchar-subscripts -Wcomment @gol
211 -Wconversion -Wdisabled-optimization -Werror @gol
212 -Wfloat-equal -Wformat -Wformat=2 @gol
213 -Wformat-nonliteral -Wformat-security @gol
214 -Wimplicit -Wimplicit-int @gol
215 -Wimplicit-function-declaration @gol
216 -Werror-implicit-function-declaration @gol
217 -Wimport -Winline @gol
218 -Wlarger-than-@var{len} -Wlong-long @gol
219 -Wmain -Wmissing-braces -Wmissing-declarations @gol
220 -Wmissing-format-attribute -Wmissing-noreturn @gol
221 -Wmultichar -Wno-format-extra-args -Wno-format-y2k @gol
222 -Wno-import -Wpacked -Wpadded @gol
223 -Wparentheses -Wpointer-arith -Wredundant-decls @gol
224 -Wreturn-type -Wsequence-point -Wshadow @gol
225 -Wsign-compare -Wswitch -Wsystem-headers @gol
226 -Wtrigraphs -Wundef -Wuninitialized @gol
227 -Wunknown-pragmas -Wunreachable-code @gol
228 -Wunused -Wunused-function -Wunused-label -Wunused-parameter @gol
229 -Wunused-value -Wunused-variable -Wwrite-strings}
231 @item C-only Warning Options
233 -Wbad-function-cast -Wmissing-prototypes -Wnested-externs @gol
234 -Wstrict-prototypes -Wtraditional}
236 @item Debugging Options
237 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
239 -a -ax -d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
240 -fdump-unnumbered -fdump-translation-unit@r{[}-@var{n}@r{]} @gol
241 -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
242 -fdump-tree-original@r{[}-@var{n}@r{]} -fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
243 -fdump-tree-inlined@r{[}-@var{n}@r{]} @gol
244 -fmem-report -fpretend-float @gol
245 -fprofile-arcs -ftest-coverage -ftime-report @gol
246 -g -g@var{level} -gcoff -gdwarf -gdwarf-1 -gdwarf-1+ -gdwarf-2 @gol
247 -ggdb -gstabs -gstabs+ -gxcoff -gxcoff+ @gol
248 -p -pg -print-file-name=@var{library} -print-libgcc-file-name @gol
249 -print-multi-directory -print-multi-lib @gol
250 -print-prog-name=@var{program} -print-search-dirs -Q @gol
253 @item Optimization Options
254 @xref{Optimize Options,,Options that Control Optimization}.
256 -falign-functions=@var{n} -falign-jumps=@var{n} @gol
257 -falign-labels=@var{n} -falign-loops=@var{n} @gol
258 -fbranch-probabilities -fcaller-saves @gol
259 -fcse-follow-jumps -fcse-skip-blocks -fdata-sections @gol
260 -fdelayed-branch -fdelete-null-pointer-checks @gol
261 -fexpensive-optimizations -ffast-math -ffloat-store @gol
262 -fforce-addr -fforce-mem -ffunction-sections @gol
263 -fgcse -fgcse-lm -fgcse-sm @gol
264 -finline-functions -finline-limit=@var{n} -fkeep-inline-functions @gol
265 -fkeep-static-consts -fmerge-constants -fmerge-all-constants @gol
266 -fmove-all-movables -fno-default-inline -fno-defer-pop @gol
267 -fno-function-cse -fno-guess-branch-probability @gol
268 -fno-inline -fno-math-errno -fno-peephole -fno-peephole2 @gol
269 -funsafe-math-optimizations -fno-trapping-math @gol
270 -fomit-frame-pointer -foptimize-register-move @gol
271 -foptimize-sibling-calls -freduce-all-givs @gol
272 -fregmove -frename-registers @gol
273 -frerun-cse-after-loop -frerun-loop-opt @gol
274 -fschedule-insns -fschedule-insns2 @gol
275 -fsingle-precision-constant -fssa -fssa-ccp -fssa-dce @gol
276 -fstrength-reduce -fstrict-aliasing -fthread-jumps -ftrapv @gol
277 -funroll-all-loops -funroll-loops -fmessy-debugging @gol
278 --param @var{name}=@var{value}
279 -O -O0 -O1 -O2 -O3 -Os}
281 @item Preprocessor Options
282 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
284 -$ -A@var{question}=@var{answer} -A-@var{question}@r{[}=@var{answer}@r{]} @gol
285 -C -dD -dI -dM -dN @gol
286 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
287 -idirafter @var{dir} @gol
288 -include @var{file} -imacros @var{file} @gol
289 -iprefix @var{file} -iwithprefix @var{dir} @gol
290 -iwithprefixbefore @var{dir} -isystem @var{dir} @gol
291 -M -MM -MF -MG -MP -MQ -MT -nostdinc -P -remap @gol
292 -trigraphs -undef -U@var{macro} -Wp,@var{option}}
294 @item Assembler Option
295 @xref{Assembler Options,,Passing Options to the Assembler}.
300 @xref{Link Options,,Options for Linking}.
302 @var{object-file-name} -l@var{library} @gol
303 -nostartfiles -nodefaultlibs -nostdlib @gol
304 -s -static -static-libgcc -shared -shared-libgcc -symbolic @gol
305 -Wl,@var{option} -Xlinker @var{option} @gol
308 @item Directory Options
309 @xref{Directory Options,,Options for Directory Search}.
311 -B@var{prefix} -I@var{dir} -I- -L@var{dir} -specs=@var{file}}
314 @c I wrote this xref this way to avoid overfull hbox. -- rms
315 @xref{Target Options}.
317 -b @var{machine} -V @var{version}}
319 @item Machine Dependent Options
320 @xref{Submodel Options,,Hardware Models and Configurations}.
321 @emph{M680x0 Options}
323 -m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
324 -m68060 -mcpu32 -m5200 -m68881 -mbitfield -mc68000 -mc68020 @gol
325 -mfpa -mnobitfield -mrtd -mshort -msoft-float -mpcrel @gol
326 -malign-int -mstrict-align}
328 @emph{M68hc1x Options}
330 -m6811 -m6812 -m68hc11 -m68hc12 @gol
331 -mauto-incdec -mshort -msoft-reg-count=@var{count}}
339 -mcpu=@var{cpu-type} @gol
340 -mtune=@var{cpu-type} @gol
341 -mcmodel=@var{code-model} @gol
343 -mapp-regs -mbroken-saverestore -mcypress @gol
344 -mepilogue -mfaster-structs -mflat @gol
345 -mfpu -mhard-float -mhard-quad-float @gol
346 -mimpure-text -mlive-g0 -mno-app-regs @gol
347 -mno-epilogue -mno-faster-structs -mno-flat -mno-fpu @gol
348 -mno-impure-text -mno-stack-bias -mno-unaligned-doubles @gol
349 -msoft-float -msoft-quad-float -msparclite -mstack-bias @gol
350 -msupersparc -munaligned-doubles -mv8}
352 @emph{Convex Options}
354 -mc1 -mc2 -mc32 -mc34 -mc38 @gol
355 -margcount -mnoargcount @gol
356 -mlong32 -mlong64 @gol
357 -mvolatile-cache -mvolatile-nocache}
359 @emph{AMD29K Options}
361 -m29000 -m29050 -mbw -mnbw -mdw -mndw @gol
362 -mlarge -mnormal -msmall @gol
363 -mkernel-registers -mno-reuse-arg-regs @gol
364 -mno-stack-check -mno-storem-bug @gol
365 -mreuse-arg-regs -msoft-float -mstack-check @gol
366 -mstorem-bug -muser-registers}
370 -mapcs-frame -mno-apcs-frame @gol
371 -mapcs-26 -mapcs-32 @gol
372 -mapcs-stack-check -mno-apcs-stack-check @gol
373 -mapcs-float -mno-apcs-float @gol
374 -mapcs-reentrant -mno-apcs-reentrant @gol
375 -msched-prolog -mno-sched-prolog @gol
376 -mlittle-endian -mbig-endian -mwords-little-endian @gol
377 -malignment-traps -mno-alignment-traps @gol
378 -msoft-float -mhard-float -mfpe @gol
379 -mthumb-interwork -mno-thumb-interwork @gol
380 -mcpu=@var{name} -march=@var{name} -mfpe=@var{name} @gol
381 -mstructure-size-boundary=@var{n} @gol
382 -mbsd -mxopen -mno-symrename @gol
383 -mabort-on-noreturn @gol
384 -mlong-calls -mno-long-calls @gol
385 -msingle-pic-base -mno-single-pic-base @gol
386 -mpic-register=@var{reg} @gol
387 -mnop-fun-dllimport @gol
388 -mpoke-function-name @gol
390 -mtpcs-frame -mtpcs-leaf-frame @gol
391 -mcaller-super-interworking -mcallee-super-interworking }
393 @emph{MN10200 Options}
397 @emph{MN10300 Options}
399 -mmult-bug -mno-mult-bug @gol
400 -mam33 -mno-am33 @gol
403 @emph{M32R/D Options}
405 -mcode-model=@var{model-type} -msdata=@var{sdata-type} @gol
410 -m88000 -m88100 -m88110 -mbig-pic @gol
411 -mcheck-zero-division -mhandle-large-shift @gol
412 -midentify-revision -mno-check-zero-division @gol
413 -mno-ocs-debug-info -mno-ocs-frame-position @gol
414 -mno-optimize-arg-area -mno-serialize-volatile @gol
415 -mno-underscores -mocs-debug-info @gol
416 -mocs-frame-position -moptimize-arg-area @gol
417 -mserialize-volatile -mshort-data-@var{num} -msvr3 @gol
418 -msvr4 -mtrap-large-shift -muse-div-instruction @gol
419 -mversion-03.00 -mwarn-passed-structs}
421 @emph{RS/6000 and PowerPC Options}
423 -mcpu=@var{cpu-type} @gol
424 -mtune=@var{cpu-type} @gol
425 -mpower -mno-power -mpower2 -mno-power2 @gol
426 -mpowerpc -mpowerpc64 -mno-powerpc @gol
427 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
428 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
429 -mnew-mnemonics -mold-mnemonics @gol
430 -mfull-toc -mminimal-toc -mno-fop-in-toc -mno-sum-in-toc @gol
431 -m64 -m32 -mxl-call -mno-xl-call -mthreads -mpe @gol
432 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
433 -mstring -mno-string -mupdate -mno-update @gol
434 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
435 -mstrict-align -mno-strict-align -mrelocatable @gol
436 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
437 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
438 -mcall-aix -mcall-sysv -mcall-netbsd @gol
439 -mprototype -mno-prototype @gol
440 -msim -mmvme -mads -myellowknife -memb -msdata @gol
441 -msdata=@var{opt} -mvxworks -G @var{num}}
445 -mcall-lib-mul -mfp-arg-in-fpregs -mfp-arg-in-gregs @gol
446 -mfull-fp-blocks -mhc-struct-return -min-line-mul @gol
447 -mminimum-fp-blocks -mnohc-struct-return}
451 -mabicalls -march=@var{cpu-type} -mtune=@var{cpu=type} @gol
452 -mcpu=@var{cpu-type} -membedded-data -muninit-const-in-rodata @gol
453 -membedded-pic -mfp32 -mfp64 -mfused-madd -mno-fused-madd @gol
454 -mgas -mgp32 -mgp64 @gol
455 -mgpopt -mhalf-pic -mhard-float -mint64 -mips1 @gol
456 -mips2 -mips3 -mips4 -mlong64 -mlong32 -mlong-calls -mmemcpy @gol
457 -mmips-as -mmips-tfile -mno-abicalls @gol
458 -mno-embedded-data -mno-uninit-const-in-rodata @gol
459 -mno-embedded-pic -mno-gpopt -mno-long-calls @gol
460 -mno-memcpy -mno-mips-tfile -mno-rnames -mno-stats @gol
461 -mrnames -msoft-float @gol
462 -m4650 -msingle-float -mmad @gol
463 -mstats -EL -EB -G @var{num} -nocpp @gol
464 -mabi=32 -mabi=n32 -mabi=64 -mabi=eabi @gol
469 -mcpu=@var{cpu-type} -march=@var{cpu-type} @gol
470 -mintel-syntax -mieee-fp -mno-fancy-math-387 @gol
471 -mno-fp-ret-in-387 -msoft-float -msvr3-shlib @gol
472 -mno-wide-multiply -mrtd -malign-double @gol
473 -mpreferred-stack-boundary=@var{num} @gol
474 -mthreads -mno-align-stringops -minline-all-stringops @gol
475 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
476 -m96bit-long-double -mregparm=@var{num} -momit-leaf-frame-pointer}
480 -march=@var{architecture-type} @gol
481 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
482 -mfast-indirect-calls -mgas -mjump-in-delay @gol
483 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
484 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
485 -mno-jump-in-delay -mno-long-load-store @gol
486 -mno-portable-runtime -mno-soft-float @gol
487 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
488 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
489 -mschedule=@var{cpu-type} -mspace-regs}
491 @emph{Intel 960 Options}
493 -m@var{cpu-type} -masm-compat -mclean-linkage @gol
494 -mcode-align -mcomplex-addr -mleaf-procedures @gol
495 -mic-compat -mic2.0-compat -mic3.0-compat @gol
496 -mintel-asm -mno-clean-linkage -mno-code-align @gol
497 -mno-complex-addr -mno-leaf-procedures @gol
498 -mno-old-align -mno-strict-align -mno-tail-call @gol
499 -mnumerics -mold-align -msoft-float -mstrict-align @gol
502 @emph{DEC Alpha Options}
504 -mfp-regs -mno-fp-regs -mno-soft-float -msoft-float @gol
505 -malpha-as -mgas @gol
506 -mieee -mieee-with-inexact -mieee-conformant @gol
507 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
508 -mtrap-precision=@var{mode} -mbuild-constants @gol
509 -mcpu=@var{cpu-type} @gol
510 -mbwx -mno-bwx -mcix -mno-cix -mmax -mno-max @gol
511 -mmemory-latency=@var{time}}
513 @emph{Clipper Options}
517 @emph{H8/300 Options}
519 -mrelax -mh -ms -mint32 -malign-300}
523 -m1 -m2 -m3 -m3e @gol
524 -m4-nofpu -m4-single-only -m4-single -m4 @gol
525 -mb -ml -mdalign -mrelax @gol
526 -mbigtable -mfmovd -mhitachi -mnomacsave @gol
527 -mieee -misize -mpadstruct -mspace @gol
528 -mprefergot -musermode}
530 @emph{System V Options}
532 -Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
537 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
538 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
540 @emph{TMS320C3x/C4x Options}
542 -mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm @gol
543 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload @gol
544 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned @gol
545 -mparallel-insns -mparallel-mpy -mpreserve-float}
549 -mlong-calls -mno-long-calls -mep -mno-ep @gol
550 -mprolog-function -mno-prolog-function -mspace @gol
551 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
556 -m32032 -m32332 -m32532 -m32081 -m32381 @gol
557 -mmult-add -mnomult-add -msoft-float -mrtd -mnortd @gol
558 -mregparam -mnoregparam -msb -mnosb @gol
559 -mbitfield -mnobitfield -mhimem -mnohimem}
563 -mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts @gol
564 -mcall-prologues -mno-tablejump -mtiny-stack}
568 -mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
569 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
570 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
571 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
572 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
576 -mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
577 -mvolatile-asm-stop -mb-step -mregister-names -mno-sdata @gol
578 -mconstant-gp -mauto-pic -minline-divide-min-latency @gol
579 -minline-divide-max-throughput -mno-dwarf2-asm @gol
580 -mfixed-range=@var{register-range}}
582 @emph{S/390 and zSeries Options}
584 -mhard-float -msoft-float -mbackchain -mno-backchain @gol
585 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
586 -m64 -m31 -mdebug -mno-debug}
588 @item Code Generation Options
589 @xref{Code Gen Options,,Options for Code Generation Conventions}.
591 -fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
592 -ffixed-@var{reg} -fexceptions @gol
593 -fnon-call-exceptions -funwind-tables @gol
594 -finhibit-size-directive -finstrument-functions @gol
595 -fcheck-memory-usage -fprefix-function-name @gol
596 -fno-common -fno-ident -fno-gnu-linker @gol
597 -fpcc-struct-return -fpic -fPIC @gol
598 -freg-struct-return -fshared-data -fshort-enums @gol
599 -fshort-double -fvolatile @gol
600 -fvolatile-global -fvolatile-static @gol
601 -fverbose-asm -fpack-struct -fstack-check @gol
602 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
603 -fargument-alias -fargument-noalias @gol
604 -fargument-noalias-global -fleading-underscore}
608 * Overall Options:: Controlling the kind of output:
609 an executable, object files, assembler files,
610 or preprocessed source.
611 * C Dialect Options:: Controlling the variant of C language compiled.
612 * C++ Dialect Options:: Variations on C++.
613 * Objective-C Dialect Options:: Variations on Objective-C.
614 * Language Independent Options:: Controlling how diagnostics should be
616 * Warning Options:: How picky should the compiler be?
617 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
618 * Optimize Options:: How much optimization?
619 * Preprocessor Options:: Controlling header files and macro definitions.
620 Also, getting dependency information for Make.
621 * Assembler Options:: Passing options to the assembler.
622 * Link Options:: Specifying libraries and so on.
623 * Directory Options:: Where to find header files and libraries.
624 Where to find the compiler executable files.
625 * Spec Files:: How to pass switches to sub-processes.
626 * Target Options:: Running a cross-compiler, or an old version of GCC.
629 @node Overall Options
630 @section Options Controlling the Kind of Output
632 Compilation can involve up to four stages: preprocessing, compilation
633 proper, assembly and linking, always in that order. The first three
634 stages apply to an individual source file, and end by producing an
635 object file; linking combines all the object files (those newly
636 compiled, and those specified as input) into an executable file.
638 @cindex file name suffix
639 For any given input file, the file name suffix determines what kind of
644 C source code which must be preprocessed.
647 C source code which should not be preprocessed.
650 C++ source code which should not be preprocessed.
653 Objective-C source code. Note that you must link with the library
654 @file{libobjc.a} to make an Objective-C program work.
657 Objective-C source code which should not be preprocessed.
660 C header file (not to be compiled or linked).
664 @itemx @var{file}.cxx
665 @itemx @var{file}.cpp
666 @itemx @var{file}.c++
668 C++ source code which must be preprocessed. Note that in @samp{.cxx},
669 the last two letters must both be literally @samp{x}. Likewise,
670 @samp{.C} refers to a literal capital C@.
673 @itemx @var{file}.for
674 @itemx @var{file}.FOR
675 Fortran source code which should not be preprocessed.
678 @itemx @var{file}.fpp
679 @itemx @var{file}.FPP
680 Fortran source code which must be preprocessed (with the traditional
684 Fortran source code which must be preprocessed with a RATFOR
685 preprocessor (not included with GCC)@.
687 @xref{Overall Options,,Options Controlling the Kind of Output, g77,
688 Using and Porting GNU Fortran}, for more details of the handling of
691 @c FIXME: Descriptions of Java file types.
697 @c GCC also knows about some suffixes for languages not yet included:
707 @itemx @var{file}.chi
708 CHILL source code (preprocessed with the traditional preprocessor).
714 Assembler code which must be preprocessed.
717 An object file to be fed straight into linking.
718 Any file name with no recognized suffix is treated this way.
722 You can specify the input language explicitly with the @option{-x} option:
725 @item -x @var{language}
726 Specify explicitly the @var{language} for the following input files
727 (rather than letting the compiler choose a default based on the file
728 name suffix). This option applies to all following input files until
729 the next @option{-x} option. Possible values for @var{language} are:
731 c c-header cpp-output
733 objective-c objc-cpp-output
734 assembler assembler-with-cpp
735 f77 f77-cpp-input ratfor
738 @c Also f77-version, for internal use only.
741 Turn off any specification of a language, so that subsequent files are
742 handled according to their file name suffixes (as they are if @option{-x}
743 has not been used at all).
745 @item -pass-exit-codes
746 @opindex pass-exit-codes
747 Normally the @command{gcc} program will exit with the code of 1 if any
748 phase of the compiler returns a non-success return code. If you specify
749 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
750 numerically highest error produced by any phase that returned an error
754 If you only want some of the stages of compilation, you can use
755 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
756 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
757 @command{gcc} is to stop. Note that some combinations (for example,
758 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
763 Compile or assemble the source files, but do not link. The linking
764 stage simply is not done. The ultimate output is in the form of an
765 object file for each source file.
767 By default, the object file name for a source file is made by replacing
768 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
770 Unrecognized input files, not requiring compilation or assembly, are
775 Stop after the stage of compilation proper; do not assemble. The output
776 is in the form of an assembler code file for each non-assembler input
779 By default, the assembler file name for a source file is made by
780 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
782 Input files that don't require compilation are ignored.
786 Stop after the preprocessing stage; do not run the compiler proper. The
787 output is in the form of preprocessed source code, which is sent to the
790 Input files which don't require preprocessing are ignored.
792 @cindex output file option
795 Place output in file @var{file}. This applies regardless to whatever
796 sort of output is being produced, whether it be an executable file,
797 an object file, an assembler file or preprocessed C code.
799 Since only one output file can be specified, it does not make sense to
800 use @option{-o} when compiling more than one input file, unless you are
801 producing an executable file as output.
803 If @option{-o} is not specified, the default is to put an executable file
804 in @file{a.out}, the object file for @file{@var{source}.@var{suffix}} in
805 @file{@var{source}.o}, its assembler file in @file{@var{source}.s}, and
806 all preprocessed C source on standard output.
810 Print (on standard error output) the commands executed to run the stages
811 of compilation. Also print the version number of the compiler driver
812 program and of the preprocessor and the compiler proper.
816 Use pipes rather than temporary files for communication between the
817 various stages of compilation. This fails to work on some systems where
818 the assembler is unable to read from a pipe; but the GNU assembler has
823 Print (on the standard output) a description of the command line options
824 understood by @command{gcc}. If the @option{-v} option is also specified
825 then @option{--help} will also be passed on to the various processes
826 invoked by @command{gcc}, so that they can display the command line options
827 they accept. If the @option{-W} option is also specified then command
828 line options which have no documentation associated with them will also
833 Print (on the standard output) a description of target specific command
834 line options for each tool.
838 @section Compiling C++ Programs
840 @cindex suffixes for C++ source
841 @cindex C++ source file suffixes
842 C++ source files conventionally use one of the suffixes @samp{.C},
843 @samp{.cc}, @samp{.cpp}, @samp{.c++}, @samp{.cp}, or @samp{.cxx};
844 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
845 files with these names and compiles them as C++ programs even if you
846 call the compiler the same way as for compiling C programs (usually with
847 the name @command{gcc}).
851 However, C++ programs often require class libraries as well as a
852 compiler that understands the C++ language---and under some
853 circumstances, you might want to compile programs from standard input,
854 or otherwise without a suffix that flags them as C++ programs.
855 @command{g++} is a program that calls GCC with the default language
856 set to C++, and automatically specifies linking against the C++
857 library. On many systems, @command{g++} is also
858 installed with the name @command{c++}.
860 @cindex invoking @command{g++}
861 When you compile C++ programs, you may specify many of the same
862 command-line options that you use for compiling programs in any
863 language; or command-line options meaningful for C and related
864 languages; or options that are meaningful only for C++ programs.
865 @xref{C Dialect Options,,Options Controlling C Dialect}, for
866 explanations of options for languages related to C@.
867 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
868 explanations of options that are meaningful only for C++ programs.
870 @node C Dialect Options
871 @section Options Controlling C Dialect
872 @cindex dialect options
873 @cindex language dialect options
874 @cindex options, dialect
876 The following options control the dialect of C (or languages derived
877 from C, such as C++ and Objective-C) that the compiler accepts:
884 In C mode, support all ISO C89 programs. In C++ mode,
885 remove GNU extensions that conflict with ISO C++.
887 This turns off certain features of GCC that are incompatible with ISO
888 C89 (when compiling C code), or of standard C++ (when compiling C++ code),
889 such as the @code{asm} and @code{typeof} keywords, and
890 predefined macros such as @code{unix} and @code{vax} that identify the
891 type of system you are using. It also enables the undesirable and
892 rarely used ISO trigraph feature. For the C compiler,
893 it disables recognition of C++ style @samp{//} comments as well as
894 the @code{inline} keyword.
896 The alternate keywords @code{__asm__}, @code{__extension__},
897 @code{__inline__} and @code{__typeof__} continue to work despite
898 @option{-ansi}. You would not want to use them in an ISO C program, of
899 course, but it is useful to put them in header files that might be included
900 in compilations done with @option{-ansi}. Alternate predefined macros
901 such as @code{__unix__} and @code{__vax__} are also available, with or
902 without @option{-ansi}.
904 The @option{-ansi} option does not cause non-ISO programs to be
905 rejected gratuitously. For that, @option{-pedantic} is required in
906 addition to @option{-ansi}. @xref{Warning Options}.
908 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
909 option is used. Some header files may notice this macro and refrain
910 from declaring certain functions or defining certain macros that the
911 ISO standard doesn't call for; this is to avoid interfering with any
912 programs that might use these names for other things.
914 Functions which would normally be built in but do not have semantics
915 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
916 functions with @option{-ansi} is used. @xref{Other Builtins,,Other
917 built-in functions provided by GCC}, for details of the functions
922 Determine the language standard. This option is currently only
923 supported when compiling C@. A value for this option must be provided;
929 ISO C89 (same as @option{-ansi}).
932 ISO C89 as modified in amendment 1.
938 ISO C99. Note that this standard is not yet fully supported; see
939 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
940 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
943 Default, ISO C89 plus GNU extensions (including some C99 features).
947 ISO C99 plus GNU extensions. When ISO C99 is fully implemented in GCC,
948 this will become the default. The name @samp{gnu9x} is deprecated.
952 Even when this option is not specified, you can still use some of the
953 features of newer standards in so far as they do not conflict with
954 previous C standards. For example, you may use @code{__restrict__} even
955 when @option{-std=c99} is not specified.
957 The @option{-std} options specifying some version of ISO C have the same
958 effects as @option{-ansi}, except that features that were not in ISO C89
959 but are in the specified version (for example, @samp{//} comments and
960 the @code{inline} keyword in ISO C99) are not disabled.
962 @xref{Standards,,Language Standards Supported by GCC}, for details of
963 these standard versions.
965 @item -aux-info @var{filename}
967 Output to the given filename prototyped declarations for all functions
968 declared and/or defined in a translation unit, including those in header
969 files. This option is silently ignored in any language other than C@.
971 Besides declarations, the file indicates, in comments, the origin of
972 each declaration (source file and line), whether the declaration was
973 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
974 @samp{O} for old, respectively, in the first character after the line
975 number and the colon), and whether it came from a declaration or a
976 definition (@samp{C} or @samp{F}, respectively, in the following
977 character). In the case of function definitions, a K&R-style list of
978 arguments followed by their declarations is also provided, inside
979 comments, after the declaration.
983 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
984 keyword, so that code can use these words as identifiers. You can use
985 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
986 instead. @option{-ansi} implies @option{-fno-asm}.
988 In C++, this switch only affects the @code{typeof} keyword, since
989 @code{asm} and @code{inline} are standard keywords. You may want to
990 use the @option{-fno-gnu-keywords} flag instead, which has the same
991 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
992 switch only affects the @code{asm} and @code{typeof} keywords, since
993 @code{inline} is a standard keyword in ISO C99.
997 @cindex built-in functions
998 Don't recognize built-in functions that do not begin with
999 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1000 functions provided by GCC}, for details of the functions affected,
1001 including those which are not built-in functions when @option{-ansi} or
1002 @option{-std} options for strict ISO C conformance are used because they
1003 do not have an ISO standard meaning.
1005 GCC normally generates special code to handle certain built-in functions
1006 more efficiently; for instance, calls to @code{alloca} may become single
1007 instructions that adjust the stack directly, and calls to @code{memcpy}
1008 may become inline copy loops. The resulting code is often both smaller
1009 and faster, but since the function calls no longer appear as such, you
1010 cannot set a breakpoint on those calls, nor can you change the behavior
1011 of the functions by linking with a different library.
1013 In C++, @option{-fno-builtin} is always in effect. The @option{-fbuiltin}
1014 option has no effect. Therefore, in C++, the only way to get the
1015 optimization benefits of built-in functions is to call the function
1016 using the @samp{__builtin_} prefix. The GNU C++ Standard Library uses
1017 built-in functions to implement many functions (like
1018 @code{std::strchr}), so that you automatically get efficient code.
1022 @cindex hosted environment
1024 Assert that compilation takes place in a hosted environment. This implies
1025 @option{-fbuiltin}. A hosted environment is one in which the
1026 entire standard library is available, and in which @code{main} has a return
1027 type of @code{int}. Examples are nearly everything except a kernel.
1028 This is equivalent to @option{-fno-freestanding}.
1030 @item -ffreestanding
1031 @opindex ffreestanding
1032 @cindex hosted environment
1034 Assert that compilation takes place in a freestanding environment. This
1035 implies @option{-fno-builtin}. A freestanding environment
1036 is one in which the standard library may not exist, and program startup may
1037 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1038 This is equivalent to @option{-fno-hosted}.
1040 @xref{Standards,,Language Standards Supported by GCC}, for details of
1041 freestanding and hosted environments.
1045 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1046 options for strict ISO C conformance) implies @option{-trigraphs}.
1048 @cindex traditional C language
1049 @cindex C language, traditional
1051 @opindex traditional
1052 Attempt to support some aspects of traditional C compilers.
1057 All @code{extern} declarations take effect globally even if they
1058 are written inside of a function definition. This includes implicit
1059 declarations of functions.
1062 The newer keywords @code{typeof}, @code{inline}, @code{signed}, @code{const}
1063 and @code{volatile} are not recognized. (You can still use the
1064 alternative keywords such as @code{__typeof__}, @code{__inline__}, and
1068 Comparisons between pointers and integers are always allowed.
1071 Integer types @code{unsigned short} and @code{unsigned char} promote
1072 to @code{unsigned int}.
1075 Out-of-range floating point literals are not an error.
1078 Certain constructs which ISO regards as a single invalid preprocessing
1079 number, such as @samp{0xe-0xd}, are treated as expressions instead.
1082 String ``constants'' are not necessarily constant; they are stored in
1083 writable space, and identical looking constants are allocated
1084 separately. (This is the same as the effect of
1085 @option{-fwritable-strings}.)
1087 @cindex @code{longjmp} and automatic variables
1089 All automatic variables not declared @code{register} are preserved by
1090 @code{longjmp}. Ordinarily, GNU C follows ISO C: automatic variables
1091 not declared @code{volatile} may be clobbered.
1096 @cindex escape sequences, traditional
1097 The character escape sequences @samp{\x} and @samp{\a} evaluate as the
1098 literal characters @samp{x} and @samp{a} respectively. Without
1099 @w{@option{-traditional}}, @samp{\x} is a prefix for the hexadecimal
1100 representation of a character, and @samp{\a} produces a bell.
1103 You may wish to use @option{-fno-builtin} as well as @option{-traditional}
1104 if your program uses names that are normally GNU C built-in functions for
1105 other purposes of its own.
1107 You cannot use @option{-traditional} if you include any header files that
1108 rely on ISO C features. Some vendors are starting to ship systems with
1109 ISO C header files and you cannot use @option{-traditional} on such
1110 systems to compile files that include any system headers.
1112 The @option{-traditional} option also enables @option{-traditional-cpp}.
1114 @item -traditional-cpp
1115 @opindex traditional-cpp
1116 Attempt to support some aspects of traditional C preprocessors.
1117 See the GNU CPP manual for details.
1119 @item -fcond-mismatch
1120 @opindex fcond-mismatch
1121 Allow conditional expressions with mismatched types in the second and
1122 third arguments. The value of such an expression is void. This option
1123 is not supported for C++.
1125 @item -funsigned-char
1126 @opindex funsigned-char
1127 Let the type @code{char} be unsigned, like @code{unsigned char}.
1129 Each kind of machine has a default for what @code{char} should
1130 be. It is either like @code{unsigned char} by default or like
1131 @code{signed char} by default.
1133 Ideally, a portable program should always use @code{signed char} or
1134 @code{unsigned char} when it depends on the signedness of an object.
1135 But many programs have been written to use plain @code{char} and
1136 expect it to be signed, or expect it to be unsigned, depending on the
1137 machines they were written for. This option, and its inverse, let you
1138 make such a program work with the opposite default.
1140 The type @code{char} is always a distinct type from each of
1141 @code{signed char} or @code{unsigned char}, even though its behavior
1142 is always just like one of those two.
1145 @opindex fsigned-char
1146 Let the type @code{char} be signed, like @code{signed char}.
1148 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1149 the negative form of @option{-funsigned-char}. Likewise, the option
1150 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1152 @item -fsigned-bitfields
1153 @itemx -funsigned-bitfields
1154 @itemx -fno-signed-bitfields
1155 @itemx -fno-unsigned-bitfields
1156 @opindex fsigned-bitfields
1157 @opindex funsigned-bitfields
1158 @opindex fno-signed-bitfields
1159 @opindex fno-unsigned-bitfields
1160 These options control whether a bit-field is signed or unsigned, when the
1161 declaration does not use either @code{signed} or @code{unsigned}. By
1162 default, such a bit-field is signed, because this is consistent: the
1163 basic integer types such as @code{int} are signed types.
1165 However, when @option{-traditional} is used, bit-fields are all unsigned
1168 @item -fwritable-strings
1169 @opindex fwritable-strings
1170 Store string constants in the writable data segment and don't uniquize
1171 them. This is for compatibility with old programs which assume they can
1172 write into string constants. The option @option{-traditional} also has
1175 Writing into string constants is a very bad idea; ``constants'' should
1178 @item -fallow-single-precision
1179 @opindex fallow-single-precision
1180 Do not promote single precision math operations to double precision,
1181 even when compiling with @option{-traditional}.
1183 Traditional K&R C promotes all floating point operations to double
1184 precision, regardless of the sizes of the operands. On the
1185 architecture for which you are compiling, single precision may be faster
1186 than double precision. If you must use @option{-traditional}, but want
1187 to use single precision operations when the operands are single
1188 precision, use this option. This option has no effect when compiling
1189 with ISO or GNU C conventions (the default).
1192 @opindex fshort-wchar
1193 Override the underlying type for @samp{wchar_t} to be @samp{short
1194 unsigned int} instead of the default for the target. This option is
1195 useful for building programs to run under WINE@.
1198 @node C++ Dialect Options
1199 @section Options Controlling C++ Dialect
1201 @cindex compiler options, C++
1202 @cindex C++ options, command line
1203 @cindex options, C++
1204 This section describes the command-line options that are only meaningful
1205 for C++ programs; but you can also use most of the GNU compiler options
1206 regardless of what language your program is in. For example, you
1207 might compile a file @code{firstClass.C} like this:
1210 g++ -g -frepo -O -c firstClass.C
1214 In this example, only @option{-frepo} is an option meant
1215 only for C++ programs; you can use the other options with any
1216 language supported by GCC@.
1218 Here is a list of options that are @emph{only} for compiling C++ programs:
1221 @item -fno-access-control
1222 @opindex fno-access-control
1223 Turn off all access checking. This switch is mainly useful for working
1224 around bugs in the access control code.
1228 Check that the pointer returned by @code{operator new} is non-null
1229 before attempting to modify the storage allocated. The current Working
1230 Paper requires that @code{operator new} never return a null pointer, so
1231 this check is normally unnecessary.
1233 An alternative to using this option is to specify that your
1234 @code{operator new} does not throw any exceptions; if you declare it
1235 @samp{throw()}, G++ will check the return value. See also @samp{new
1238 @item -fconserve-space
1239 @opindex fconserve-space
1240 Put uninitialized or runtime-initialized global variables into the
1241 common segment, as C does. This saves space in the executable at the
1242 cost of not diagnosing duplicate definitions. If you compile with this
1243 flag and your program mysteriously crashes after @code{main()} has
1244 completed, you may have an object that is being destroyed twice because
1245 two definitions were merged.
1247 This option is no longer useful on most targets, now that support has
1248 been added for putting variables into BSS without making them common.
1250 @item -fno-const-strings
1251 @opindex fno-const-strings
1252 Give string constants type @code{char *} instead of type @code{const
1253 char *}. By default, G++ uses type @code{const char *} as required by
1254 the standard. Even if you use @option{-fno-const-strings}, you cannot
1255 actually modify the value of a string constant, unless you also use
1256 @option{-fwritable-strings}.
1258 This option might be removed in a future release of G++. For maximum
1259 portability, you should structure your code so that it works with
1260 string constants that have type @code{const char *}.
1262 @item -fdollars-in-identifiers
1263 @opindex fdollars-in-identifiers
1264 Accept @samp{$} in identifiers. You can also explicitly prohibit use of
1265 @samp{$} with the option @option{-fno-dollars-in-identifiers}. (GNU C allows
1266 @samp{$} by default on most target systems, but there are a few exceptions.)
1267 Traditional C allowed the character @samp{$} to form part of
1268 identifiers. However, ISO C and C++ forbid @samp{$} in identifiers.
1270 @item -fno-elide-constructors
1271 @opindex fno-elide-constructors
1272 The C++ standard allows an implementation to omit creating a temporary
1273 which is only used to initialize another object of the same type.
1274 Specifying this option disables that optimization, and forces G++ to
1275 call the copy constructor in all cases.
1277 @item -fno-enforce-eh-specs
1278 @opindex fno-enforce-eh-specs
1279 Don't check for violation of exception specifications at runtime. This
1280 option violates the C++ standard, but may be useful for reducing code
1281 size in production builds, much like defining @samp{NDEBUG}. The compiler
1282 will still optimize based on the exception specifications.
1284 @item -fexternal-templates
1285 @opindex fexternal-templates
1287 Cause @samp{#pragma interface} and @samp{implementation} to apply to
1288 template instantiation; template instances are emitted or not according
1289 to the location of the template definition. @xref{Template
1290 Instantiation}, for more information.
1292 This option is deprecated.
1294 @item -falt-external-templates
1295 @opindex falt-external-templates
1296 Similar to @option{-fexternal-templates}, but template instances are
1297 emitted or not according to the place where they are first instantiated.
1298 @xref{Template Instantiation}, for more information.
1300 This option is deprecated.
1303 @itemx -fno-for-scope
1305 @opindex fno-for-scope
1306 If @option{-ffor-scope} is specified, the scope of variables declared in
1307 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1308 as specified by the C++ standard.
1309 If @option{-fno-for-scope} is specified, the scope of variables declared in
1310 a @i{for-init-statement} extends to the end of the enclosing scope,
1311 as was the case in old versions of G++, and other (traditional)
1312 implementations of C++.
1314 The default if neither flag is given to follow the standard,
1315 but to allow and give a warning for old-style code that would
1316 otherwise be invalid, or have different behavior.
1318 @item -fno-gnu-keywords
1319 @opindex fno-gnu-keywords
1320 Do not recognize @code{typeof} as a keyword, so that code can use this
1321 word as an identifier. You can use the keyword @code{__typeof__} instead.
1322 @option{-ansi} implies @option{-fno-gnu-keywords}.
1324 @item -fno-implicit-templates
1325 @opindex fno-implicit-templates
1326 Never emit code for non-inline templates which are instantiated
1327 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1328 @xref{Template Instantiation}, for more information.
1330 @item -fno-implicit-inline-templates
1331 @opindex fno-implicit-inline-templates
1332 Don't emit code for implicit instantiations of inline templates, either.
1333 The default is to handle inlines differently so that compiles with and
1334 without optimization will need the same set of explicit instantiations.
1336 @item -fno-implement-inlines
1337 @opindex fno-implement-inlines
1338 To save space, do not emit out-of-line copies of inline functions
1339 controlled by @samp{#pragma implementation}. This will cause linker
1340 errors if these functions are not inlined everywhere they are called.
1342 @item -fms-extensions
1343 @opindex fms-extensions
1344 Disable pedantic warnings about constructs used in MFC, such as implicit
1345 int and getting a pointer to member function via non-standard syntax.
1347 @item -fno-nonansi-builtins
1348 @opindex fno-nonansi-builtins
1349 Disable built-in declarations of functions that are not mandated by
1350 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1351 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1353 @item -fno-operator-names
1354 @opindex fno-operator-names
1355 Do not treat the operator name keywords @code{and}, @code{bitand},
1356 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1357 synonyms as keywords.
1359 @item -fno-optional-diags
1360 @opindex fno-optional-diags
1361 Disable diagnostics that the standard says a compiler does not need to
1362 issue. Currently, the only such diagnostic issued by G++ is the one for
1363 a name having multiple meanings within a class.
1366 @opindex fpermissive
1367 Downgrade messages about nonconformant code from errors to warnings. By
1368 default, G++ effectively sets @option{-pedantic-errors} without
1369 @option{-pedantic}; this option reverses that. This behavior and this
1370 option are superseded by @option{-pedantic}, which works as it does for GNU C@.
1374 Enable automatic template instantiation at link time. This option also
1375 implies @option{-fno-implicit-templates}. @xref{Template
1376 Instantiation}, for more information.
1380 Disable generation of information about every class with virtual
1381 functions for use by the C++ runtime type identification features
1382 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1383 of the language, you can save some space by using this flag. Note that
1384 exception handling uses the same information, but it will generate it as
1389 Emit statistics about front-end processing at the end of the compilation.
1390 This information is generally only useful to the G++ development team.
1392 @item -ftemplate-depth-@var{n}
1393 @opindex ftemplate-depth
1394 Set the maximum instantiation depth for template classes to @var{n}.
1395 A limit on the template instantiation depth is needed to detect
1396 endless recursions during template class instantiation. ANSI/ISO C++
1397 conforming programs must not rely on a maximum depth greater than 17.
1399 @item -fuse-cxa-atexit
1400 @opindex fuse-cxa-atexit
1401 Register destructors for objects with static storage duration with the
1402 @code{__cxa_atexit} function rather than the @code{atexit} function.
1403 This option is required for fully standards-compliant handling of static
1404 destructors, but will only work if your C library supports
1405 @code{__cxa_atexit}.
1409 Emit special relocations for vtables and virtual function references
1410 so that the linker can identify unused virtual functions and zero out
1411 vtable slots that refer to them. This is most useful with
1412 @option{-ffunction-sections} and @option{-Wl,--gc-sections}, in order to
1413 also discard the functions themselves.
1415 This optimization requires GNU as and GNU ld. Not all systems support
1416 this option. @option{-Wl,--gc-sections} is ignored without @option{-static}.
1420 Do not use weak symbol support, even if it is provided by the linker.
1421 By default, G++ will use weak symbols if they are available. This
1422 option exists only for testing, and should not be used by end-users;
1423 it will result in inferior code and has no benefits. This option may
1424 be removed in a future release of G++.
1428 Do not search for header files in the standard directories specific to
1429 C++, but do still search the other standard directories. (This option
1430 is used when building the C++ library.)
1433 In addition, these optimization, warning, and code generation options
1434 have meanings only for C++ programs:
1437 @item -fno-default-inline
1438 @opindex fno-default-inline
1439 Do not assume @samp{inline} for functions defined inside a class scope.
1440 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1441 functions will have linkage like inline functions; they just won't be
1444 @item -Wctor-dtor-privacy @r{(C++ only)}
1445 @opindex Wctor-dtor-privacy
1446 Warn when a class seems unusable, because all the constructors or
1447 destructors in a class are private and the class has no friends or
1448 public static member functions.
1450 @item -Wnon-virtual-dtor @r{(C++ only)}
1451 @opindex Wnon-virtual-dtor
1452 Warn when a class declares a non-virtual destructor that should probably
1453 be virtual, because it looks like the class will be used polymorphically.
1455 @item -Wreorder @r{(C++ only)}
1457 @cindex reordering, warning
1458 @cindex warning for reordering of member initializers
1459 Warn when the order of member initializers given in the code does not
1460 match the order in which they must be executed. For instance:
1466 A(): j (0), i (1) @{ @}
1470 Here the compiler will warn that the member initializers for @samp{i}
1471 and @samp{j} will be rearranged to match the declaration order of the
1475 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
1478 @item -Weffc++ @r{(C++ only)}
1480 Warn about violations of various style guidelines from Scott Meyers'
1481 @cite{Effective C++} books. If you use this option, you should be aware
1482 that the standard library headers do not obey all of these guidelines;
1483 you can use @samp{grep -v} to filter out those warnings.
1485 @item -Wno-deprecated @r{(C++ only)}
1486 @opindex Wno-deprecated
1487 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1489 @item -Wno-non-template-friend @r{(C++ only)}
1490 @opindex Wno-non-template-friend
1491 Disable warnings when non-templatized friend functions are declared
1492 within a template. With the advent of explicit template specification
1493 support in G++, if the name of the friend is an unqualified-id (i.e.,
1494 @samp{friend foo(int)}), the C++ language specification demands that the
1495 friend declare or define an ordinary, nontemplate function. (Section
1496 14.5.3). Before G++ implemented explicit specification, unqualified-ids
1497 could be interpreted as a particular specialization of a templatized
1498 function. Because this non-conforming behavior is no longer the default
1499 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
1500 check existing code for potential trouble spots, and is on by default.
1501 This new compiler behavior can be turned off with
1502 @option{-Wno-non-template-friend} which keeps the conformant compiler code
1503 but disables the helpful warning.
1505 @item -Wold-style-cast @r{(C++ only)}
1506 @opindex Wold-style-cast
1507 Warn if an old-style (C-style) cast is used within a C++ program. The
1508 new-style casts (@samp{static_cast}, @samp{reinterpret_cast}, and
1509 @samp{const_cast}) are less vulnerable to unintended effects, and much
1512 @item -Woverloaded-virtual @r{(C++ only)}
1513 @opindex Woverloaded-virtual
1514 @cindex overloaded virtual fn, warning
1515 @cindex warning for overloaded virtual fn
1516 Warn when a function declaration hides virtual functions from a
1517 base class. For example, in:
1524 struct B: public A @{
1529 the @code{A} class version of @code{f} is hidden in @code{B}, and code
1537 will fail to compile.
1539 @item -Wno-pmf-conversions @r{(C++ only)}
1540 @opindex Wno-pmf-conversions
1541 Disable the diagnostic for converting a bound pointer to member function
1544 @item -Wsign-promo @r{(C++ only)}
1545 @opindex Wsign-promo
1546 Warn when overload resolution chooses a promotion from unsigned or
1547 enumeral type to a signed type over a conversion to an unsigned type of
1548 the same size. Previous versions of G++ would try to preserve
1549 unsignedness, but the standard mandates the current behavior.
1551 @item -Wsynth @r{(C++ only)}
1553 @cindex warning for synthesized methods
1554 @cindex synthesized methods, warning
1555 Warn when G++'s synthesis behavior does not match that of cfront. For
1561 A& operator = (int);
1571 In this example, G++ will synthesize a default @samp{A& operator =
1572 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1575 @node Objective-C Dialect Options
1576 @section Options Controlling Objective-C Dialect
1578 @cindex compiler options, Objective-C
1579 @cindex Objective-C options, command line
1580 @cindex options, Objective-C
1581 This section describes the command-line options that are only meaningful
1582 for Objective-C programs; but you can also use most of the GNU compiler
1583 options regardless of what language your program is in. For example,
1584 you might compile a file @code{some_class.m} like this:
1587 gcc -g -fgnu-runtime -O -c some_class.m
1591 In this example, only @option{-fgnu-runtime} is an option meant only for
1592 Objective-C programs; you can use the other options with any language
1595 Here is a list of options that are @emph{only} for compiling Objective-C
1599 @item -fconstant-string-class=@var{class-name}
1600 @opindex fconstant-string-class
1601 Use @var{class-name} as the name of the class to instantiate for each
1602 literal string specified with the syntax @code{@@"@dots{}"}. The default
1603 class name is @code{NXConstantString}.
1606 @opindex fgnu-runtime
1607 Generate object code compatible with the standard GNU Objective-C
1608 runtime. This is the default for most types of systems.
1610 @item -fnext-runtime
1611 @opindex fnext-runtime
1612 Generate output compatible with the NeXT runtime. This is the default
1613 for NeXT-based systems, including Darwin and Mac OS X@.
1617 Dump interface declarations for all classes seen in the source file to a
1618 file named @file{@var{sourcename}.decl}.
1621 @opindex Wno-protocol
1622 Do not warn if methods required by a protocol are not implemented
1623 in the class adopting it.
1627 Warn if a selector has multiple methods of different types defined.
1629 @c not documented because only avail via -Wp
1630 @c @item -print-objc-runtime-info
1634 @node Language Independent Options
1635 @section Options to Control Diagnostic Messages Formatting
1636 @cindex options to control diagnostics formatting
1637 @cindex diagnostic messages
1638 @cindex message formatting
1640 Traditionally, diagnostic messages have been formatted irrespective of
1641 the output device's aspect (e.g.@: its width, @dots{}). The options described
1642 below can be used to control the diagnostic messages formatting
1643 algorithm, e.g.@: how many characters per line, how often source location
1644 information should be reported. Right now, only the C++ front end can
1645 honor these options. However it is expected, in the near future, that
1646 the remaining front ends would be able to digest them correctly.
1649 @item -fmessage-length=@var{n}
1650 @opindex fmessage-length
1651 Try to format error messages so that they fit on lines of about @var{n}
1652 characters. The default is 72 characters for @command{g++} and 0 for the rest of
1653 the front ends supported by GCC@. If @var{n} is zero, then no
1654 line-wrapping will be done; each error message will appear on a single
1657 @opindex fdiagnostics-show-location
1658 @item -fdiagnostics-show-location=once
1659 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
1660 reporter to emit @emph{once} source location information; that is, in
1661 case the message is too long to fit on a single physical line and has to
1662 be wrapped, the source location won't be emitted (as prefix) again,
1663 over and over, in subsequent continuation lines. This is the default
1666 @item -fdiagnostics-show-location=every-line
1667 Only meaningful in line-wrapping mode. Instructs the diagnostic
1668 messages reporter to emit the same source location information (as
1669 prefix) for physical lines that result from the process of breaking a
1670 a message which is too long to fit on a single line.
1674 @node Warning Options
1675 @section Options to Request or Suppress Warnings
1676 @cindex options to control warnings
1677 @cindex warning messages
1678 @cindex messages, warning
1679 @cindex suppressing warnings
1681 Warnings are diagnostic messages that report constructions which
1682 are not inherently erroneous but which are risky or suggest there
1683 may have been an error.
1685 You can request many specific warnings with options beginning @samp{-W},
1686 for example @option{-Wimplicit} to request warnings on implicit
1687 declarations. Each of these specific warning options also has a
1688 negative form beginning @samp{-Wno-} to turn off warnings;
1689 for example, @option{-Wno-implicit}. This manual lists only one of the
1690 two forms, whichever is not the default.
1692 These options control the amount and kinds of warnings produced by GCC:
1695 @cindex syntax checking
1697 @opindex fsyntax-only
1698 Check the code for syntax errors, but don't do anything beyond that.
1702 Issue all the warnings demanded by strict ISO C and ISO C++;
1703 reject all programs that use forbidden extensions, and some other
1704 programs that do not follow ISO C and ISO C++. For ISO C, follows the
1705 version of the ISO C standard specified by any @option{-std} option used.
1707 Valid ISO C and ISO C++ programs should compile properly with or without
1708 this option (though a rare few will require @option{-ansi} or a
1709 @option{-std} option specifying the required version of ISO C)@. However,
1710 without this option, certain GNU extensions and traditional C and C++
1711 features are supported as well. With this option, they are rejected.
1713 @option{-pedantic} does not cause warning messages for use of the
1714 alternate keywords whose names begin and end with @samp{__}. Pedantic
1715 warnings are also disabled in the expression that follows
1716 @code{__extension__}. However, only system header files should use
1717 these escape routes; application programs should avoid them.
1718 @xref{Alternate Keywords}.
1720 Some users try to use @option{-pedantic} to check programs for strict ISO
1721 C conformance. They soon find that it does not do quite what they want:
1722 it finds some non-ISO practices, but not all---only those for which
1723 ISO C @emph{requires} a diagnostic, and some others for which
1724 diagnostics have been added.
1726 A feature to report any failure to conform to ISO C might be useful in
1727 some instances, but would require considerable additional work and would
1728 be quite different from @option{-pedantic}. We don't have plans to
1729 support such a feature in the near future.
1731 Where the standard specified with @option{-std} represents a GNU
1732 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
1733 corresponding @dfn{base standard}, the version of ISO C on which the GNU
1734 extended dialect is based. Warnings from @option{-pedantic} are given
1735 where they are required by the base standard. (It would not make sense
1736 for such warnings to be given only for features not in the specified GNU
1737 C dialect, since by definition the GNU dialects of C include all
1738 features the compiler supports with the given option, and there would be
1739 nothing to warn about.)
1741 @item -pedantic-errors
1742 @opindex pedantic-errors
1743 Like @option{-pedantic}, except that errors are produced rather than
1748 Inhibit all warning messages.
1752 Inhibit warning messages about the use of @samp{#import}.
1754 @item -Wchar-subscripts
1755 @opindex Wchar-subscripts
1756 Warn if an array subscript has type @code{char}. This is a common cause
1757 of error, as programmers often forget that this type is signed on some
1762 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
1763 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
1767 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
1768 the arguments supplied have types appropriate to the format string
1769 specified, and that the conversions specified in the format string make
1770 sense. This includes standard functions, and others specified by format
1771 attributes (@pxref{Function Attributes}), in the @code{printf},
1772 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
1773 not in the C standard) families.
1775 The formats are checked against the format features supported by GNU
1776 libc version 2.2. These include all ISO C89 and C99 features, as well
1777 as features from the Single Unix Specification and some BSD and GNU
1778 extensions. Other library implementations may not support all these
1779 features; GCC does not support warning about features that go beyond a
1780 particular library's limitations. However, if @option{-pedantic} is used
1781 with @option{-Wformat}, warnings will be given about format features not
1782 in the selected standard version (but not for @code{strfmon} formats,
1783 since those are not in any version of the C standard). @xref{C Dialect
1784 Options,,Options Controlling C Dialect}.
1786 @option{-Wformat} is included in @option{-Wall}. For more control over some
1787 aspects of format checking, the options @option{-Wno-format-y2k},
1788 @option{-Wno-format-extra-args}, @option{-Wformat-nonliteral},
1789 @option{-Wformat-security} and @option{-Wformat=2} are available, but are
1790 not included in @option{-Wall}.
1792 @item -Wno-format-y2k
1793 @opindex Wno-format-y2k
1794 If @option{-Wformat} is specified, do not warn about @code{strftime}
1795 formats which may yield only a two-digit year.
1797 @item -Wno-format-extra-args
1798 @opindex Wno-format-extra-args
1799 If @option{-Wformat} is specified, do not warn about excess arguments to a
1800 @code{printf} or @code{scanf} format function. The C standard specifies
1801 that such arguments are ignored.
1803 @item -Wformat-nonliteral
1804 @opindex Wformat-nonliteral
1805 If @option{-Wformat} is specified, also warn if the format string is not a
1806 string literal and so cannot be checked, unless the format function
1807 takes its format arguments as a @code{va_list}.
1809 @item -Wformat-security
1810 @opindex Wformat-security
1811 If @option{-Wformat} is specified, also warn about uses of format
1812 functions that represent possible security problems. At present, this
1813 warns about calls to @code{printf} and @code{scanf} functions where the
1814 format string is not a string literal and there are no format arguments,
1815 as in @code{printf (foo);}. This may be a security hole if the format
1816 string came from untrusted input and contains @samp{%n}. (This is
1817 currently a subset of what @option{-Wformat-nonliteral} warns about, but
1818 in future warnings may be added to @option{-Wformat-security} that are not
1819 included in @option{-Wformat-nonliteral}.)
1823 Enable @option{-Wformat} plus format checks not included in
1824 @option{-Wformat}. Currently equivalent to @samp{-Wformat
1825 -Wformat-nonliteral -Wformat-security}.
1827 @item -Wimplicit-int
1828 @opindex Wimplicit-int
1829 Warn when a declaration does not specify a type.
1831 @item -Wimplicit-function-declaration
1832 @itemx -Werror-implicit-function-declaration
1833 @opindex Wimplicit-function-declaration
1834 @opindex Werror-implicit-function-declaration
1835 Give a warning (or error) whenever a function is used before being
1840 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
1844 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
1845 function with external linkage, returning int, taking either zero
1846 arguments, two, or three arguments of appropriate types.
1848 @item -Wmissing-braces
1849 @opindex Wmissing-braces
1850 Warn if an aggregate or union initializer is not fully bracketed. In
1851 the following example, the initializer for @samp{a} is not fully
1852 bracketed, but that for @samp{b} is fully bracketed.
1855 int a[2][2] = @{ 0, 1, 2, 3 @};
1856 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
1861 Warn if a multicharacter constant (@samp{'FOOF'}) is used. Usually they
1862 indicate a typo in the user's code, as they have implementation-defined
1863 values, and should not be used in portable code.
1866 @opindex Wparentheses
1867 Warn if parentheses are omitted in certain contexts, such
1868 as when there is an assignment in a context where a truth value
1869 is expected, or when operators are nested whose precedence people
1870 often get confused about.
1872 Also warn about constructions where there may be confusion to which
1873 @code{if} statement an @code{else} branch belongs. Here is an example of
1888 In C, every @code{else} branch belongs to the innermost possible @code{if}
1889 statement, which in this example is @code{if (b)}. This is often not
1890 what the programmer expected, as illustrated in the above example by
1891 indentation the programmer chose. When there is the potential for this
1892 confusion, GCC will issue a warning when this flag is specified.
1893 To eliminate the warning, add explicit braces around the innermost
1894 @code{if} statement so there is no way the @code{else} could belong to
1895 the enclosing @code{if}. The resulting code would look like this:
1911 @item -Wsequence-point
1912 @opindex Wsequence-point
1913 Warn about code that may have undefined semantics because of violations
1914 of sequence point rules in the C standard.
1916 The C standard defines the order in which expressions in a C program are
1917 evaluated in terms of @dfn{sequence points}, which represent a partial
1918 ordering between the execution of parts of the program: those executed
1919 before the sequence point, and those executed after it. These occur
1920 after the evaluation of a full expression (one which is not part of a
1921 larger expression), after the evaluation of the first operand of a
1922 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
1923 function is called (but after the evaluation of its arguments and the
1924 expression denoting the called function), and in certain other places.
1925 Other than as expressed by the sequence point rules, the order of
1926 evaluation of subexpressions of an expression is not specified. All
1927 these rules describe only a partial order rather than a total order,
1928 since, for example, if two functions are called within one expression
1929 with no sequence point between them, the order in which the functions
1930 are called is not specified. However, the standards committee have
1931 ruled that function calls do not overlap.
1933 It is not specified when between sequence points modifications to the
1934 values of objects take effect. Programs whose behavior depends on this
1935 have undefined behavior; the C standard specifies that ``Between the
1936 previous and next sequence point an object shall have its stored value
1937 modified at most once by the evaluation of an expression. Furthermore,
1938 the prior value shall be read only to determine the value to be
1939 stored.''. If a program breaks these rules, the results on any
1940 particular implementation are entirely unpredictable.
1942 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
1943 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
1944 diagnosed by this option, and it may give an occasional false positive
1945 result, but in general it has been found fairly effective at detecting
1946 this sort of problem in programs.
1948 The present implementation of this option only works for C programs. A
1949 future implementation may also work for C++ programs.
1951 The C standard is worded confusingly, therefore there is some debate
1952 over the precise meaning of the sequence point rules in subtle cases.
1953 Links to discussions of the problem, including proposed formal
1954 definitions, may be found on our readings page, at
1955 @w{@uref{http://gcc.gnu.org/readings.html}}.
1958 @opindex Wreturn-type
1959 Warn whenever a function is defined with a return-type that defaults to
1960 @code{int}. Also warn about any @code{return} statement with no
1961 return-value in a function whose return-type is not @code{void}.
1963 For C++, a function without return type always produces a diagnostic
1964 message, even when @option{-Wno-return-type} is specified. The only
1965 exceptions are @samp{main} and functions defined in system headers.
1969 Warn whenever a @code{switch} statement has an index of enumeral type
1970 and lacks a @code{case} for one or more of the named codes of that
1971 enumeration. (The presence of a @code{default} label prevents this
1972 warning.) @code{case} labels outside the enumeration range also
1973 provoke warnings when this option is used.
1977 Warn if any trigraphs are encountered that might change the meaning of
1978 the program (trigraphs within comments are not warned about).
1980 @item -Wunused-function
1981 @opindex Wunused-function
1982 Warn whenever a static function is declared but not defined or a
1983 non\-inline static function is unused.
1985 @item -Wunused-label
1986 @opindex Wunused-label
1987 Warn whenever a label is declared but not used.
1989 To suppress this warning use the @samp{unused} attribute
1990 (@pxref{Variable Attributes}).
1992 @item -Wunused-parameter
1993 @opindex Wunused-parameter
1994 Warn whenever a function parameter is unused aside from its declaration.
1996 To suppress this warning use the @samp{unused} attribute
1997 (@pxref{Variable Attributes}).
1999 @item -Wunused-variable
2000 @opindex Wunused-variable
2001 Warn whenever a local variable or non-constant static variable is unused
2002 aside from its declaration
2004 To suppress this warning use the @samp{unused} attribute
2005 (@pxref{Variable Attributes}).
2007 @item -Wunused-value
2008 @opindex Wunused-value
2009 Warn whenever a statement computes a result that is explicitly not used.
2011 To suppress this warning cast the expression to @samp{void}.
2015 All all the above @option{-Wunused} options combined.
2017 In order to get a warning about an unused function parameter, you must
2018 either specify @samp{-W -Wunused} or separately specify
2019 @option{-Wunused-parameter}.
2021 @item -Wuninitialized
2022 @opindex Wuninitialized
2023 Warn if an automatic variable is used without first being initialized or
2024 if a variable may be clobbered by a @code{setjmp} call.
2026 These warnings are possible only in optimizing compilation,
2027 because they require data flow information that is computed only
2028 when optimizing. If you don't specify @option{-O}, you simply won't
2031 These warnings occur only for variables that are candidates for
2032 register allocation. Therefore, they do not occur for a variable that
2033 is declared @code{volatile}, or whose address is taken, or whose size
2034 is other than 1, 2, 4 or 8 bytes. Also, they do not occur for
2035 structures, unions or arrays, even when they are in registers.
2037 Note that there may be no warning about a variable that is used only
2038 to compute a value that itself is never used, because such
2039 computations may be deleted by data flow analysis before the warnings
2042 These warnings are made optional because GCC is not smart
2043 enough to see all the reasons why the code might be correct
2044 despite appearing to have an error. Here is one example of how
2065 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
2066 always initialized, but GCC doesn't know this. Here is
2067 another common case:
2072 if (change_y) save_y = y, y = new_y;
2074 if (change_y) y = save_y;
2079 This has no bug because @code{save_y} is used only if it is set.
2081 @cindex @code{longjmp} warnings
2082 This option also warns when a non-volatile automatic variable might be
2083 changed by a call to @code{longjmp}. These warnings as well are possible
2084 only in optimizing compilation.
2086 The compiler sees only the calls to @code{setjmp}. It cannot know
2087 where @code{longjmp} will be called; in fact, a signal handler could
2088 call it at any point in the code. As a result, you may get a warning
2089 even when there is in fact no problem because @code{longjmp} cannot
2090 in fact be called at the place which would cause a problem.
2092 Some spurious warnings can be avoided if you declare all the functions
2093 you use that never return as @code{noreturn}. @xref{Function
2096 @item -Wreorder @r{(C++ only)}
2098 @cindex reordering, warning
2099 @cindex warning for reordering of member initializers
2100 Warn when the order of member initializers given in the code does not
2101 match the order in which they must be executed. For instance:
2103 @item -Wunknown-pragmas
2104 @opindex Wunknown-pragmas
2105 @cindex warning for unknown pragmas
2106 @cindex unknown pragmas, warning
2107 @cindex pragmas, warning of unknown
2108 Warn when a #pragma directive is encountered which is not understood by
2109 GCC@. If this command line option is used, warnings will even be issued
2110 for unknown pragmas in system header files. This is not the case if
2111 the warnings were only enabled by the @option{-Wall} command line option.
2115 All of the above @samp{-W} options combined. This enables all the
2116 warnings about constructions that some users consider questionable, and
2117 that are easy to avoid (or modify to prevent the warning), even in
2118 conjunction with macros.
2120 @item -Wsystem-headers
2121 @opindex Wsystem-headers
2122 @cindex warnings from system headers
2123 @cindex system headers, warnings from
2124 Print warning messages for constructs found in system header files.
2125 Warnings from system headers are normally suppressed, on the assumption
2126 that they usually do not indicate real problems and would only make the
2127 compiler output harder to read. Using this command line option tells
2128 GCC to emit warnings from system headers as if they occurred in user
2129 code. However, note that using @option{-Wall} in conjunction with this
2130 option will @emph{not} warn about unknown pragmas in system
2131 headers---for that, @option{-Wunknown-pragmas} must also be used.
2134 The following @option{-W@dots{}} options are not implied by @option{-Wall}.
2135 Some of them warn about constructions that users generally do not
2136 consider questionable, but which occasionally you might wish to check
2137 for; others warn about constructions that are necessary or hard to avoid
2138 in some cases, and there is no simple way to modify the code to suppress
2144 Print extra warning messages for these events:
2148 A function can return either with or without a value. (Falling
2149 off the end of the function body is considered returning without
2150 a value.) For example, this function would evoke such a
2164 An expression-statement or the left-hand side of a comma expression
2165 contains no side effects.
2166 To suppress the warning, cast the unused expression to void.
2167 For example, an expression such as @samp{x[i,j]} will cause a warning,
2168 but @samp{x[(void)i,j]} will not.
2171 An unsigned value is compared against zero with @samp{<} or @samp{<=}.
2174 A comparison like @samp{x<=y<=z} appears; this is equivalent to
2175 @samp{(x<=y ? 1 : 0) <= z}, which is a different interpretation from
2176 that of ordinary mathematical notation.
2179 Storage-class specifiers like @code{static} are not the first things in
2180 a declaration. According to the C Standard, this usage is obsolescent.
2183 The return type of a function has a type qualifier such as @code{const}.
2184 Such a type qualifier has no effect, since the value returned by a
2185 function is not an lvalue. (But don't warn about the GNU extension of
2186 @code{volatile void} return types. That extension will be warned about
2187 if @option{-pedantic} is specified.)
2190 If @option{-Wall} or @option{-Wunused} is also specified, warn about unused
2194 A comparison between signed and unsigned values could produce an
2195 incorrect result when the signed value is converted to unsigned.
2196 (But don't warn if @option{-Wno-sign-compare} is also specified.)
2199 An aggregate has a partly bracketed initializer.
2200 For example, the following code would evoke such a warning,
2201 because braces are missing around the initializer for @code{x.h}:
2204 struct s @{ int f, g; @};
2205 struct t @{ struct s h; int i; @};
2206 struct t x = @{ 1, 2, 3 @};
2210 An aggregate has an initializer which does not initialize all members.
2211 For example, the following code would cause such a warning, because
2212 @code{x.h} would be implicitly initialized to zero:
2215 struct s @{ int f, g, h; @};
2216 struct s x = @{ 3, 4 @};
2221 @opindex Wfloat-equal
2222 Warn if floating point values are used in equality comparisons.
2224 The idea behind this is that sometimes it is convenient (for the
2225 programmer) to consider floating-point values as approximations to
2226 infinitely precise real numbers. If you are doing this, then you need
2227 to compute (by analysing the code, or in some other way) the maximum or
2228 likely maximum error that the computation introduces, and allow for it
2229 when performing comparisons (and when producing output, but that's a
2230 different problem). In particular, instead of testing for equality, you
2231 would check to see whether the two values have ranges that overlap; and
2232 this is done with the relational operators, so equality comparisons are
2235 @item -Wtraditional @r{(C only)}
2236 @opindex Wtraditional
2237 Warn about certain constructs that behave differently in traditional and
2238 ISO C@. Also warn about ISO C constructs that have no traditional C
2239 equivalent, and/or problematic constructs which should be avoided.
2243 Macro parameters that appear within string literals in the macro body.
2244 In traditional C macro replacement takes place within string literals,
2245 but does not in ISO C@.
2248 In traditional C, some preprocessor directives did not exist.
2249 Traditional preprocessors would only consider a line to be a directive
2250 if the @samp{#} appeared in column 1 on the line. Therefore
2251 @option{-Wtraditional} warns about directives that traditional C
2252 understands but would ignore because the @samp{#} does not appear as the
2253 first character on the line. It also suggests you hide directives like
2254 @samp{#pragma} not understood by traditional C by indenting them. Some
2255 traditional implementations would not recognise @samp{#elif}, so it
2256 suggests avoiding it altogether.
2259 A function-like macro that appears without arguments.
2262 The unary plus operator.
2265 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
2266 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
2267 constants.) Note, these suffixes appear in macros defined in the system
2268 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
2269 Use of these macros in user code might normally lead to spurious
2270 warnings, however gcc's integrated preprocessor has enough context to
2271 avoid warning in these cases.
2274 A function declared external in one block and then used after the end of
2278 A @code{switch} statement has an operand of type @code{long}.
2281 A non-@code{static} function declaration follows a @code{static} one.
2282 This construct is not accepted by some traditional C compilers.
2285 The ISO type of an integer constant has a different width or
2286 signedness from its traditional type. This warning is only issued if
2287 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
2288 typically represent bit patterns, are not warned about.
2291 Usage of ISO string concatenation is detected.
2294 Initialization of automatic aggregates.
2297 Identifier conflicts with labels. Traditional C lacks a separate
2298 namespace for labels.
2301 Initialization of unions. If the initializer is zero, the warning is
2302 omitted. This is done under the assumption that the zero initializer in
2303 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
2304 initializer warnings and relies on default initialization to zero in the
2308 Conversions by prototypes between fixed/floating point values and vice
2309 versa. The absence of these prototypes when compiling with traditional
2310 C would cause serious problems. This is a subset of the possible
2311 conversion warnings, for the full set use @option{-Wconversion}.
2316 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
2320 Warn whenever a local variable shadows another local variable, parameter or
2321 global variable or whenever a built-in function is shadowed.
2323 @item -Wlarger-than-@var{len}
2324 @opindex Wlarger-than
2325 Warn whenever an object of larger than @var{len} bytes is defined.
2327 @item -Wpointer-arith
2328 @opindex Wpointer-arith
2329 Warn about anything that depends on the ``size of'' a function type or
2330 of @code{void}. GNU C assigns these types a size of 1, for
2331 convenience in calculations with @code{void *} pointers and pointers
2334 @item -Wbad-function-cast @r{(C only)}
2335 @opindex Wbad-function-cast
2336 Warn whenever a function call is cast to a non-matching type.
2337 For example, warn if @code{int malloc()} is cast to @code{anything *}.
2341 Warn whenever a pointer is cast so as to remove a type qualifier from
2342 the target type. For example, warn if a @code{const char *} is cast
2343 to an ordinary @code{char *}.
2346 @opindex Wcast-align
2347 Warn whenever a pointer is cast such that the required alignment of the
2348 target is increased. For example, warn if a @code{char *} is cast to
2349 an @code{int *} on machines where integers can only be accessed at
2350 two- or four-byte boundaries.
2352 @item -Wwrite-strings
2353 @opindex Wwrite-strings
2354 When compiling C, give string constants the type @code{const
2355 char[@var{length}]} so that
2356 copying the address of one into a non-@code{const} @code{char *}
2357 pointer will get a warning; when compiling C++, warn about the
2358 deprecated conversion from string constants to @code{char *}.
2359 These warnings will help you find at
2360 compile time code that can try to write into a string constant, but
2361 only if you have been very careful about using @code{const} in
2362 declarations and prototypes. Otherwise, it will just be a nuisance;
2363 this is why we did not make @option{-Wall} request these warnings.
2366 @opindex Wconversion
2367 Warn if a prototype causes a type conversion that is different from what
2368 would happen to the same argument in the absence of a prototype. This
2369 includes conversions of fixed point to floating and vice versa, and
2370 conversions changing the width or signedness of a fixed point argument
2371 except when the same as the default promotion.
2373 Also, warn if a negative integer constant expression is implicitly
2374 converted to an unsigned type. For example, warn about the assignment
2375 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
2376 casts like @code{(unsigned) -1}.
2378 @item -Wsign-compare
2379 @opindex Wsign-compare
2380 @cindex warning for comparison of signed and unsigned values
2381 @cindex comparison of signed and unsigned values, warning
2382 @cindex signed and unsigned values, comparison warning
2383 Warn when a comparison between signed and unsigned values could produce
2384 an incorrect result when the signed value is converted to unsigned.
2385 This warning is also enabled by @option{-W}; to get the other warnings
2386 of @option{-W} without this warning, use @samp{-W -Wno-sign-compare}.
2388 @item -Waggregate-return
2389 @opindex Waggregate-return
2390 Warn if any functions that return structures or unions are defined or
2391 called. (In languages where you can return an array, this also elicits
2394 @item -Wstrict-prototypes @r{(C only)}
2395 @opindex Wstrict-prototypes
2396 Warn if a function is declared or defined without specifying the
2397 argument types. (An old-style function definition is permitted without
2398 a warning if preceded by a declaration which specifies the argument
2401 @item -Wmissing-prototypes @r{(C only)}
2402 @opindex Wmissing-prototypes
2403 Warn if a global function is defined without a previous prototype
2404 declaration. This warning is issued even if the definition itself
2405 provides a prototype. The aim is to detect global functions that fail
2406 to be declared in header files.
2408 @item -Wmissing-declarations
2409 @opindex Wmissing-declarations
2410 Warn if a global function is defined without a previous declaration.
2411 Do so even if the definition itself provides a prototype.
2412 Use this option to detect global functions that are not declared in
2415 @item -Wmissing-noreturn
2416 @opindex Wmissing-noreturn
2417 Warn about functions which might be candidates for attribute @code{noreturn}.
2418 Note these are only possible candidates, not absolute ones. Care should
2419 be taken to manually verify functions actually do not ever return before
2420 adding the @code{noreturn} attribute, otherwise subtle code generation
2421 bugs could be introduced. You will not get a warning for @code{main} in
2422 hosted C environments.
2424 @item -Wmissing-format-attribute
2425 @opindex Wmissing-format-attribute
2427 If @option{-Wformat} is enabled, also warn about functions which might be
2428 candidates for @code{format} attributes. Note these are only possible
2429 candidates, not absolute ones. GCC will guess that @code{format}
2430 attributes might be appropriate for any function that calls a function
2431 like @code{vprintf} or @code{vscanf}, but this might not always be the
2432 case, and some functions for which @code{format} attributes are
2433 appropriate may not be detected. This option has no effect unless
2434 @option{-Wformat} is enabled (possibly by @option{-Wall}).
2438 Warn if a structure is given the packed attribute, but the packed
2439 attribute has no effect on the layout or size of the structure.
2440 Such structures may be mis-aligned for little benefit. For
2441 instance, in this code, the variable @code{f.x} in @code{struct bar}
2442 will be misaligned even though @code{struct bar} does not itself
2443 have the packed attribute:
2450 @} __attribute__((packed));
2460 Warn if padding is included in a structure, either to align an element
2461 of the structure or to align the whole structure. Sometimes when this
2462 happens it is possible to rearrange the fields of the structure to
2463 reduce the padding and so make the structure smaller.
2465 @item -Wredundant-decls
2466 @opindex Wredundant-decls
2467 Warn if anything is declared more than once in the same scope, even in
2468 cases where multiple declaration is valid and changes nothing.
2470 @item -Wnested-externs @r{(C only)}
2471 @opindex Wnested-externs
2472 Warn if an @code{extern} declaration is encountered within a function.
2474 @item -Wunreachable-code
2475 @opindex Wunreachable-code
2476 Warn if the compiler detects that code will never be executed.
2478 This option is intended to warn when the compiler detects that at
2479 least a whole line of source code will never be executed, because
2480 some condition is never satisfied or because it is after a
2481 procedure that never returns.
2483 It is possible for this option to produce a warning even though there
2484 are circumstances under which part of the affected line can be executed,
2485 so care should be taken when removing apparently-unreachable code.
2487 For instance, when a function is inlined, a warning may mean that the
2488 line is unreachable in only one inlined copy of the function.
2490 This option is not made part of @option{-Wall} because in a debugging
2491 version of a program there is often substantial code which checks
2492 correct functioning of the program and is, hopefully, unreachable
2493 because the program does work. Another common use of unreachable
2494 code is to provide behaviour which is selectable at compile-time.
2498 Warn if a function can not be inlined and it was declared as inline.
2502 @opindex Wno-long-long
2503 Warn if @samp{long long} type is used. This is default. To inhibit
2504 the warning messages, use @option{-Wno-long-long}. Flags
2505 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
2506 only when @option{-pedantic} flag is used.
2508 @item -Wdisabled-optimization
2509 @opindex Wdisabled-optimization
2510 Warn if a requested optimization pass is disabled. This warning does
2511 not generally indicate that there is anything wrong with your code; it
2512 merely indicates that GCC's optimizers were unable to handle the code
2513 effectively. Often, the problem is that your code is too big or too
2514 complex; GCC will refuse to optimize programs when the optimization
2515 itself is likely to take inordinate amounts of time.
2519 Make all warnings into errors.
2522 @node Debugging Options
2523 @section Options for Debugging Your Program or GCC
2524 @cindex options, debugging
2525 @cindex debugging information options
2527 GCC has various special options that are used for debugging
2528 either your program or GCC:
2533 Produce debugging information in the operating system's native format
2534 (stabs, COFF, XCOFF, or DWARF)@. GDB can work with this debugging
2537 On most systems that use stabs format, @option{-g} enables use of extra
2538 debugging information that only GDB can use; this extra information
2539 makes debugging work better in GDB but will probably make other debuggers
2541 refuse to read the program. If you want to control for certain whether
2542 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
2543 @option{-gxcoff+}, @option{-gxcoff}, @option{-gdwarf-1+}, or @option{-gdwarf-1}
2546 Unlike most other C compilers, GCC allows you to use @option{-g} with
2547 @option{-O}. The shortcuts taken by optimized code may occasionally
2548 produce surprising results: some variables you declared may not exist
2549 at all; flow of control may briefly move where you did not expect it;
2550 some statements may not be executed because they compute constant
2551 results or their values were already at hand; some statements may
2552 execute in different places because they were moved out of loops.
2554 Nevertheless it proves possible to debug optimized output. This makes
2555 it reasonable to use the optimizer for programs that might have bugs.
2557 The following options are useful when GCC is generated with the
2558 capability for more than one debugging format.
2562 Produce debugging information for use by GDB@. This means to use the
2563 most expressive format available (DWARF 2, stabs, or the native format
2564 if neither of those are supported), including GDB extensions if at all
2569 Produce debugging information in stabs format (if that is supported),
2570 without GDB extensions. This is the format used by DBX on most BSD
2571 systems. On MIPS, Alpha and System V Release 4 systems this option
2572 produces stabs debugging output which is not understood by DBX or SDB@.
2573 On System V Release 4 systems this option requires the GNU assembler.
2577 Produce debugging information in stabs format (if that is supported),
2578 using GNU extensions understood only by the GNU debugger (GDB)@. The
2579 use of these extensions is likely to make other debuggers crash or
2580 refuse to read the program.
2584 Produce debugging information in COFF format (if that is supported).
2585 This is the format used by SDB on most System V systems prior to
2590 Produce debugging information in XCOFF format (if that is supported).
2591 This is the format used by the DBX debugger on IBM RS/6000 systems.
2595 Produce debugging information in XCOFF format (if that is supported),
2596 using GNU extensions understood only by the GNU debugger (GDB)@. The
2597 use of these extensions is likely to make other debuggers crash or
2598 refuse to read the program, and may cause assemblers other than the GNU
2599 assembler (GAS) to fail with an error.
2603 Produce debugging information in DWARF version 1 format (if that is
2604 supported). This is the format used by SDB on most System V Release 4
2609 Produce debugging information in DWARF version 1 format (if that is
2610 supported), using GNU extensions understood only by the GNU debugger
2611 (GDB)@. The use of these extensions is likely to make other debuggers
2612 crash or refuse to read the program.
2616 Produce debugging information in DWARF version 2 format (if that is
2617 supported). This is the format used by DBX on IRIX 6.
2620 @itemx -ggdb@var{level}
2621 @itemx -gstabs@var{level}
2622 @itemx -gcoff@var{level}
2623 @itemx -gxcoff@var{level}
2624 @itemx -gdwarf@var{level}
2625 @itemx -gdwarf-2@var{level}
2626 Request debugging information and also use @var{level} to specify how
2627 much information. The default level is 2.
2629 Level 1 produces minimal information, enough for making backtraces in
2630 parts of the program that you don't plan to debug. This includes
2631 descriptions of functions and external variables, but no information
2632 about local variables and no line numbers.
2634 Level 3 includes extra information, such as all the macro definitions
2635 present in the program. Some debuggers support macro expansion when
2636 you use @option{-g3}.
2641 Generate extra code to write profile information suitable for the
2642 analysis program @code{prof}. You must use this option when compiling
2643 the source files you want data about, and you must also use it when
2646 @cindex @code{gprof}
2649 Generate extra code to write profile information suitable for the
2650 analysis program @code{gprof}. You must use this option when compiling
2651 the source files you want data about, and you must also use it when
2657 Generate extra code to write profile information for basic blocks, which will
2658 record the number of times each basic block is executed, the basic block start
2659 address, and the function name containing the basic block. If @option{-g} is
2660 used, the line number and filename of the start of the basic block will also be
2661 recorded. If not overridden by the machine description, the default action is
2662 to append to the text file @file{bb.out}.
2664 This data could be analyzed by a program like @code{tcov}. Note,
2665 however, that the format of the data is not what @code{tcov} expects.
2666 Eventually GNU @code{gprof} should be extended to process this data.
2670 Makes the compiler print out each function name as it is compiled, and
2671 print some statistics about each pass when it finishes.
2674 @opindex ftime-report
2675 Makes the compiler print some statistics about the time consumed by each
2676 pass when it finishes.
2679 @opindex fmem-report
2680 Makes the compiler print some statistics about permanent memory
2681 allocation when it finishes.
2685 Generate extra code to profile basic blocks. Your executable will
2686 produce output that is a superset of that produced when @option{-a} is
2687 used. Additional output is the source and target address of the basic
2688 blocks where a jump takes place, the number of times a jump is executed,
2689 and (optionally) the complete sequence of basic blocks being executed.
2690 The output is appended to file @file{bb.out}.
2692 You can examine different profiling aspects without recompilation. Your
2693 executable will read a list of function names from file @file{bb.in}.
2694 Profiling starts when a function on the list is entered and stops when
2695 that invocation is exited. To exclude a function from profiling, prefix
2696 its name with @samp{-}. If a function name is not unique, you can
2697 disambiguate it by writing it in the form
2698 @samp{/path/filename.d:functionname}. Your executable will write the
2699 available paths and filenames in file @file{bb.out}.
2701 Several function names have a special meaning:
2704 Write source, target and frequency of jumps to file @file{bb.out}.
2705 @item __bb_hidecall__
2706 Exclude function calls from frequency count.
2707 @item __bb_showret__
2708 Include function returns in frequency count.
2710 Write the sequence of basic blocks executed to file @file{bbtrace.gz}.
2711 The file will be compressed using the program @samp{gzip}, which must
2712 exist in your @env{PATH}. On systems without the @samp{popen}
2713 function, the file will be named @file{bbtrace} and will not be
2714 compressed. @strong{Profiling for even a few seconds on these systems
2715 will produce a very large file.} Note: @code{__bb_hidecall__} and
2716 @code{__bb_showret__} will not affect the sequence written to
2720 Here's a short example using different profiling parameters
2721 in file @file{bb.in}. Assume function @code{foo} consists of basic blocks
2722 1 and 2 and is called twice from block 3 of function @code{main}. After
2723 the calls, block 3 transfers control to block 4 of @code{main}.
2725 With @code{__bb_trace__} and @code{main} contained in file @file{bb.in},
2726 the following sequence of blocks is written to file @file{bbtrace.gz}:
2727 0 3 1 2 1 2 4. The return from block 2 to block 3 is not shown, because
2728 the return is to a point inside the block and not to the top. The
2729 block address 0 always indicates, that control is transferred
2730 to the trace from somewhere outside the observed functions. With
2731 @samp{-foo} added to @file{bb.in}, the blocks of function
2732 @code{foo} are removed from the trace, so only 0 3 4 remains.
2734 With @code{__bb_jumps__} and @code{main} contained in file @file{bb.in},
2735 jump frequencies will be written to file @file{bb.out}. The
2736 frequencies are obtained by constructing a trace of blocks
2737 and incrementing a counter for every neighbouring pair of blocks
2738 in the trace. The trace 0 3 1 2 1 2 4 displays the following
2742 Jump from block 0x0 to block 0x3 executed 1 time(s)
2743 Jump from block 0x3 to block 0x1 executed 1 time(s)
2744 Jump from block 0x1 to block 0x2 executed 2 time(s)
2745 Jump from block 0x2 to block 0x1 executed 1 time(s)
2746 Jump from block 0x2 to block 0x4 executed 1 time(s)
2749 With @code{__bb_hidecall__}, control transfer due to call instructions
2750 is removed from the trace, that is the trace is cut into three parts: 0
2751 3 4, 0 1 2 and 0 1 2. With @code{__bb_showret__}, control transfer due
2752 to return instructions is added to the trace. The trace becomes: 0 3 1
2753 2 3 1 2 3 4. Note, that this trace is not the same, as the sequence
2754 written to @file{bbtrace.gz}. It is solely used for counting jump
2757 @item -fprofile-arcs
2758 @opindex fprofile-arcs
2759 Instrument @dfn{arcs} during compilation to generate coverage data
2760 or for profile-directed block ordering. During execution the program
2761 records how many times each branch is executed and how many times it is
2762 taken. When the compiled program exits it saves this data to a file
2763 called @file{@var{sourcename}.da} for each source file.
2765 For profile-directed block ordering, compile the program with
2766 @option{-fprofile-arcs} plus optimization and code generation options,
2767 generate the arc profile information by running the program on a
2768 selected workload, and then compile the program again with the same
2769 optimization and code generation options plus
2770 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
2771 Control Optimization}).
2773 The other use of @option{-fprofile-arcs} is for use with @code{gcov},
2774 when it is used with the @option{-ftest-coverage} option. GCC
2775 supports two methods of determining code coverage: the options that
2776 support @code{gcov}, and options @option{-a} and @option{-ax}, which
2777 write information to text files. The options that support @code{gcov}
2778 do not need to instrument every arc in the program, so a program compiled
2779 with them runs faster than a program compiled with @option{-a}, which
2780 adds instrumentation code to every basic block in the program. The
2781 tradeoff: since @code{gcov} does not have execution counts for all
2782 branches, it must start with the execution counts for the instrumented
2783 branches, and then iterate over the program flow graph until the entire
2784 graph has been solved. Hence, @code{gcov} runs a little more slowly than
2785 a program which uses information from @option{-a} and @option{-ax}.
2787 With @option{-fprofile-arcs}, for each function of your program GCC
2788 creates a program flow graph, then finds a spanning tree for the graph.
2789 Only arcs that are not on the spanning tree have to be instrumented: the
2790 compiler adds code to count the number of times that these arcs are
2791 executed. When an arc is the only exit or only entrance to a block, the
2792 instrumentation code can be added to the block; otherwise, a new basic
2793 block must be created to hold the instrumentation code.
2795 This option makes it possible to estimate branch probabilities and to
2796 calculate basic block execution counts. In general, basic block
2797 execution counts as provided by @option{-a} do not give enough
2798 information to estimate all branch probabilities.
2801 @item -ftest-coverage
2802 @opindex ftest-coverage
2803 Create data files for the @code{gcov} code-coverage utility
2804 (@pxref{Gcov,, @code{gcov}: a GCC Test Coverage Program}).
2805 The data file names begin with the name of your source file:
2808 @item @var{sourcename}.bb
2809 A mapping from basic blocks to line numbers, which @code{gcov} uses to
2810 associate basic block execution counts with line numbers.
2812 @item @var{sourcename}.bbg
2813 A list of all arcs in the program flow graph. This allows @code{gcov}
2814 to reconstruct the program flow graph, so that it can compute all basic
2815 block and arc execution counts from the information in the
2816 @code{@var{sourcename}.da} file.
2819 Use @option{-ftest-coverage} with @option{-fprofile-arcs}; the latter
2820 option adds instrumentation to the program, which then writes
2821 execution counts to another data file:
2824 @item @var{sourcename}.da
2825 Runtime arc execution counts, used in conjunction with the arc
2826 information in the file @code{@var{sourcename}.bbg}.
2829 Coverage data will map better to the source files if
2830 @option{-ftest-coverage} is used without optimization.
2832 @item -d@var{letters}
2834 Says to make debugging dumps during compilation at times specified by
2835 @var{letters}. This is used for debugging the compiler. The file names
2836 for most of the dumps are made by appending a pass number and a word to
2837 the source file name (e.g. @file{foo.c.00.rtl} or @file{foo.c.01.sibling}).
2838 Here are the possible letters for use in @var{letters}, and their meanings:
2843 Annotate the assembler output with miscellaneous debugging information.
2846 Dump after computing branch probabilities, to @file{@var{file}.14.bp}.
2849 Dump after block reordering, to @file{@var{file}.28.bbro}.
2852 Dump after instruction combination, to the file @file{@var{file}.16.combine}.
2855 Dump after the first if conversion, to the file @file{@var{file}.17.ce}.
2858 Dump after delayed branch scheduling, to @file{@var{file}.31.dbr}.
2861 Dump all macro definitions, at the end of preprocessing, in addition to
2865 Dump after SSA optimizations, to @file{@var{file}.04.ssa} and
2866 @file{@var{file}.07.ussa}.
2869 Dump after the second if conversion, to @file{@var{file}.26.ce2}.
2872 Dump after life analysis, to @file{@var{file}.15.life}.
2875 Dump after purging @code{ADDRESSOF} codes, to @file{@var{file}.09.addressof}.
2878 Dump after global register allocation, to @file{@var{file}.21.greg}.
2881 Dump after finalization of EH handling code, to @file{@var{file}.02.eh}.
2884 Dump after post-reload optimizations, to @file{@var{file}.22.postreload}.
2887 Dump after GCSE, to @file{@var{file}.10.gcse}.
2890 Dump after sibling call optimizations, to @file{@var{file}.01.sibling}.
2893 Dump after the first jump optimization, to @file{@var{file}.03.jump}.
2896 Dump after conversion from registers to stack, to @file{@var{file}.32.stack}.
2899 Dump after local register allocation, to @file{@var{file}.20.lreg}.
2902 Dump after loop optimization, to @file{@var{file}.11.loop}.
2905 Dump after performing the machine dependent reorganisation pass, to
2906 @file{@var{file}.30.mach}.
2909 Dump after register renumbering, to @file{@var{file}.25.rnreg}.
2912 Dump after the register move pass, to @file{@var{file}.18.regmove}.
2915 Dump after RTL generation, to @file{@var{file}.00.rtl}.
2918 Dump after the second scheduling pass, to @file{@var{file}.27.sched2}.
2921 Dump after CSE (including the jump optimization that sometimes follows
2922 CSE), to @file{@var{file}.08.cse}.
2925 Dump after the first scheduling pass, to @file{@var{file}.19.sched}.
2928 Dump after the second CSE pass (including the jump optimization that
2929 sometimes follows CSE), to @file{@var{file}.12.cse2}.
2932 Dump after the second flow pass, to @file{@var{file}.23.flow2}.
2935 Dump after SSA dead code elimination, to @file{@var{file}.06.ssadce}.
2938 Dump after the peephole pass, to @file{@var{file}.24.peephole2}.
2941 Produce all the dumps listed above.
2944 Print statistics on memory usage, at the end of the run, to
2948 Annotate the assembler output with a comment indicating which
2949 pattern and alternative was used. The length of each instruction is
2953 Dump the RTL in the assembler output as a comment before each instruction.
2954 Also turns on @option{-dp} annotation.
2957 For each of the other indicated dump files (except for
2958 @file{@var{file}.00.rtl}), dump a representation of the control flow graph
2959 suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
2962 Just generate RTL for a function instead of compiling it. Usually used
2966 Dump debugging information during parsing, to standard error.
2969 @item -fdump-unnumbered
2970 @opindex fdump-unnumbered
2971 When doing debugging dumps (see @option{-d} option above), suppress instruction
2972 numbers and line number note output. This makes it more feasible to
2973 use diff on debugging dumps for compiler invocations with different
2974 options, in particular with and without @option{-g}.
2976 @item -fdump-translation-unit @r{(C and C++ only)}
2977 @itemx -fdump-translation-unit-@var{options} @r{(C and C++ only)}
2978 @opindex fdump-translation-unit
2979 Dump a representation of the tree structure for the entire translation
2980 unit to a file. The file name is made by appending @file{.tu} to the
2981 source file name. If the @samp{-@var{options}} form is used, @var{options}
2982 controls the details of the dump as described for the
2983 @option{-fdump-tree} options.
2985 @item -fdump-class-hierarchy @r{(C++ only)}
2986 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
2987 @opindex fdump-class-hierarchy
2988 Dump a representation of each class's hierarchy and virtual function
2989 table layout to a file. The file name is made by appending @file{.class}
2990 to the source file name. If the @samp{-@var{options}} form is used,
2991 @var{options} controls the details of the dump as described for the
2992 @option{-fdump-tree} options.
2994 @item -fdump-tree-@var{switch} @r{(C++ only)}
2995 @itemx -fdump-tree-@var{switch}-@var{options} @r{(C++ only)}
2997 Control the dumping at various stages of processing the intermediate
2998 language tree to a file. The file name is generated by appending a switch
2999 specific suffix to the source file name. If the @samp{-@var{options}}
3000 form is used, @var{options} is a list of @samp{-} separated options that
3001 control the details of the dump. Not all options are applicable to all
3002 dumps, those which are not meaningful will be ignored. The following
3003 options are available
3007 Print the address of each node. Usually this is not meaningful as it
3008 changes according to the environment and source file. Its primary use
3009 is for tying up a dump file with a debug environment.
3011 Inhibit dumping of members of a scope or body of a function merely
3012 because that scope has been reached. Only dump such items when they
3013 are directly reachable by some other path.
3015 Turn on all options.
3018 The following tree dumps are possible:
3021 Dump before any tree based optimization, to @file{@var{file}.original}.
3023 Dump after all tree based optimization, to @file{@var{file}.optimized}.
3025 Dump after function inlining, to @file{@var{file}.inlined}.
3028 @item -fpretend-float
3029 @opindex fpretend-float
3030 When running a cross-compiler, pretend that the target machine uses the
3031 same floating point format as the host machine. This causes incorrect
3032 output of the actual floating constants, but the actual instruction
3033 sequence will probably be the same as GCC would make when running on
3038 Store the usual ``temporary'' intermediate files permanently; place them
3039 in the current directory and name them based on the source file. Thus,
3040 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
3041 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
3042 preprocessed @file{foo.i} output file even though the compiler now
3043 normally uses an integrated preprocessor.
3047 Report the CPU time taken by each subprocess in the compilation
3048 sequence. For C source files, this is the compiler proper and assembler
3049 (plus the linker if linking is done). The output looks like this:
3056 The first number on each line is the ``user time,'' that is time spent
3057 executing the program itself. The second number is ``system time,''
3058 time spent executing operating system routines on behalf of the program.
3059 Both numbers are in seconds.
3061 @item -print-file-name=@var{library}
3062 @opindex print-file-name
3063 Print the full absolute name of the library file @var{library} that
3064 would be used when linking---and don't do anything else. With this
3065 option, GCC does not compile or link anything; it just prints the
3068 @item -print-multi-directory
3069 @opindex print-multi-directory
3070 Print the directory name corresponding to the multilib selected by any
3071 other switches present in the command line. This directory is supposed
3072 to exist in @env{GCC_EXEC_PREFIX}.
3074 @item -print-multi-lib
3075 @opindex print-multi-lib
3076 Print the mapping from multilib directory names to compiler switches
3077 that enable them. The directory name is separated from the switches by
3078 @samp{;}, and each switch starts with an @samp{@@} instead of the
3079 @samp{-}, without spaces between multiple switches. This is supposed to
3080 ease shell-processing.
3082 @item -print-prog-name=@var{program}
3083 @opindex print-prog-name
3084 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
3086 @item -print-libgcc-file-name
3087 @opindex print-libgcc-file-name
3088 Same as @option{-print-file-name=libgcc.a}.
3090 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
3091 but you do want to link with @file{libgcc.a}. You can do
3094 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
3097 @item -print-search-dirs
3098 @opindex print-search-dirs
3099 Print the name of the configured installation directory and a list of
3100 program and library directories gcc will search---and don't do anything else.
3102 This is useful when gcc prints the error message
3103 @samp{installation problem, cannot exec cpp0: No such file or directory}.
3104 To resolve this you either need to put @file{cpp0} and the other compiler
3105 components where gcc expects to find them, or you can set the environment
3106 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
3107 Don't forget the trailing '/'.
3108 @xref{Environment Variables}.
3111 @opindex dumpmachine
3112 Print the compiler's target machine (for example,
3113 @samp{i686-pc-linux-gnu})---and don't do anything else.
3116 @opindex dumpversion
3117 Print the compiler version (for example, @samp{3.0})---and don't do
3122 Print the compiler's built-in specs---and don't do anything else. (This
3123 is used when GCC itself is being built.) @xref{Spec Files}.
3126 @node Optimize Options
3127 @section Options That Control Optimization
3128 @cindex optimize options
3129 @cindex options, optimization
3131 These options control various sorts of optimizations:
3138 Optimize. Optimizing compilation takes somewhat more time, and a lot
3139 more memory for a large function.
3141 Without @option{-O}, the compiler's goal is to reduce the cost of
3142 compilation and to make debugging produce the expected results.
3143 Statements are independent: if you stop the program with a breakpoint
3144 between statements, you can then assign a new value to any variable or
3145 change the program counter to any other statement in the function and
3146 get exactly the results you would expect from the source code.
3148 With @option{-O}, the compiler tries to reduce code size and execution
3149 time, without performing any optimizations that take a great deal of
3154 Optimize even more. GCC performs nearly all supported optimizations
3155 that do not involve a space-speed tradeoff. The compiler does not
3156 perform loop unrolling or function inlining when you specify @option{-O2}.
3157 As compared to @option{-O}, this option increases both compilation time
3158 and the performance of the generated code.
3160 @option{-O2} turns on all optional optimizations except for loop unrolling,
3161 function inlining, and register renaming. It also turns on the
3162 @option{-fforce-mem} option on all machines and frame pointer elimination
3163 on machines where doing so does not interfere with debugging.
3165 Please note the warning under @option{-fgcse} about
3166 invoking @option{-O2} on programs that use computed gotos.
3170 Optimize yet more. @option{-O3} turns on all optimizations specified by
3171 @option{-O2} and also turns on the @option{-finline-functions} and
3172 @option{-frename-registers} options.
3180 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
3181 do not typically increase code size. It also performs further
3182 optimizations designed to reduce code size.
3184 If you use multiple @option{-O} options, with or without level numbers,
3185 the last such option is the one that is effective.
3188 Options of the form @option{-f@var{flag}} specify machine-independent
3189 flags. Most flags have both positive and negative forms; the negative
3190 form of @option{-ffoo} would be @option{-fno-foo}. In the table below,
3191 only one of the forms is listed---the one which is not the default.
3192 You can figure out the other form by either removing @samp{no-} or
3197 @opindex ffloat-store
3198 Do not store floating point variables in registers, and inhibit other
3199 options that might change whether a floating point value is taken from a
3202 @cindex floating point precision
3203 This option prevents undesirable excess precision on machines such as
3204 the 68000 where the floating registers (of the 68881) keep more
3205 precision than a @code{double} is supposed to have. Similarly for the
3206 x86 architecture. For most programs, the excess precision does only
3207 good, but a few programs rely on the precise definition of IEEE floating
3208 point. Use @option{-ffloat-store} for such programs, after modifying
3209 them to store all pertinent intermediate computations into variables.
3211 @item -fno-default-inline
3212 @opindex fno-default-inline
3213 Do not make member functions inline by default merely because they are
3214 defined inside the class scope (C++ only). Otherwise, when you specify
3215 @w{@option{-O}}, member functions defined inside class scope are compiled
3216 inline by default; i.e., you don't need to add @samp{inline} in front of
3217 the member function name.
3219 @item -fno-defer-pop
3220 @opindex fno-defer-pop
3221 Always pop the arguments to each function call as soon as that function
3222 returns. For machines which must pop arguments after a function call,
3223 the compiler normally lets arguments accumulate on the stack for several
3224 function calls and pops them all at once.
3228 Force memory operands to be copied into registers before doing
3229 arithmetic on them. This produces better code by making all memory
3230 references potential common subexpressions. When they are not common
3231 subexpressions, instruction combination should eliminate the separate
3232 register-load. The @option{-O2} option turns on this option.
3235 @opindex fforce-addr
3236 Force memory address constants to be copied into registers before
3237 doing arithmetic on them. This may produce better code just as
3238 @option{-fforce-mem} may.
3240 @item -fomit-frame-pointer
3241 @opindex fomit-frame-pointer
3242 Don't keep the frame pointer in a register for functions that
3243 don't need one. This avoids the instructions to save, set up and
3244 restore frame pointers; it also makes an extra register available
3245 in many functions. @strong{It also makes debugging impossible on
3249 On some machines, such as the VAX, this flag has no effect, because
3250 the standard calling sequence automatically handles the frame pointer
3251 and nothing is saved by pretending it doesn't exist. The
3252 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
3253 whether a target machine supports this flag. @xref{Registers}.
3256 On some machines, such as the VAX, this flag has no effect, because
3257 the standard calling sequence automatically handles the frame pointer
3258 and nothing is saved by pretending it doesn't exist. The
3259 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
3260 whether a target machine supports this flag. @xref{Registers,,Register
3261 Usage, gcc.info, Using and Porting GCC}.
3264 @item -foptimize-sibling-calls
3265 @opindex foptimize-sibling-calls
3266 Optimize sibling and tail recursive calls.
3270 This option generates traps for signed overflow on addition, subtraction,
3271 multiplication operations.
3275 Don't pay attention to the @code{inline} keyword. Normally this option
3276 is used to keep the compiler from expanding any functions inline.
3277 Note that if you are not optimizing, no functions can be expanded inline.
3279 @item -finline-functions
3280 @opindex finline-functions
3281 Integrate all simple functions into their callers. The compiler
3282 heuristically decides which functions are simple enough to be worth
3283 integrating in this way.
3285 If all calls to a given function are integrated, and the function is
3286 declared @code{static}, then the function is normally not output as
3287 assembler code in its own right.
3289 @item -finline-limit=@var{n}
3290 @opindex finline-limit
3291 By default, gcc limits the size of functions that can be inlined. This flag
3292 allows the control of this limit for functions that are explicitly marked as
3293 inline (ie marked with the inline keyword or defined within the class
3294 definition in c++). @var{n} is the size of functions that can be inlined in
3295 number of pseudo instructions (not counting parameter handling). The default
3296 value of @var{n} is 600.
3297 Increasing this value can result in more inlined code at
3298 the cost of compilation time and memory consumption. Decreasing usually makes
3299 the compilation faster and less code will be inlined (which presumably
3300 means slower programs). This option is particularly useful for programs that
3301 use inlining heavily such as those based on recursive templates with C++.
3303 @emph{Note:} pseudo instruction represents, in this particular context, an
3304 abstract measurement of function's size. In no way, it represents a count
3305 of assembly instructions and as such its exact meaning might change from one
3306 release to an another.
3308 @item -fkeep-inline-functions
3309 @opindex fkeep-inline-functions
3310 Even if all calls to a given function are integrated, and the function
3311 is declared @code{static}, nevertheless output a separate run-time
3312 callable version of the function. This switch does not affect
3313 @code{extern inline} functions.
3315 @item -fkeep-static-consts
3316 @opindex fkeep-static-consts
3317 Emit variables declared @code{static const} when optimization isn't turned
3318 on, even if the variables aren't referenced.
3320 GCC enables this option by default. If you want to force the compiler to
3321 check if the variable was referenced, regardless of whether or not
3322 optimization is turned on, use the @option{-fno-keep-static-consts} option.
3324 @item -fmerge-constants
3325 Attempt to merge identical constants (string constants and floating point
3326 constants) accross compilation units.
3328 This option is default for optimized compilation if assembler and linker
3329 support it. Use @option{-fno-merge-constants} to inhibit this behaviour.
3331 @item -fmerge-all-constants
3332 Attempt to merge identical constants and identical variables.
3334 This option implies @option{-fmerge-constants}. In addition to
3335 @option{-fmerge-constants} this considers e.g. even constant initialized
3336 arrays or initialized constant variables with integral or floating point
3337 types. Languages like C or C++ require each non-automatic variable to
3338 have distinct location, so using this option will result in non-conforming
3341 @item -fno-function-cse
3342 @opindex fno-function-cse
3343 Do not put function addresses in registers; make each instruction that
3344 calls a constant function contain the function's address explicitly.
3346 This option results in less efficient code, but some strange hacks
3347 that alter the assembler output may be confused by the optimizations
3348 performed when this option is not used.
3352 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations}, and @*
3353 @option{-fno-trapping-math}.
3355 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
3357 This option should never be turned on by any @option{-O} option since
3358 it can result in incorrect output for programs which depend on
3359 an exact implementation of IEEE or ISO rules/specifications for
3362 @item -fno-math-errno
3363 @opindex fno-math-errno
3364 Do not set ERRNO after calling math functions that are executed
3365 with a single instruction, e.g., sqrt. A program that relies on
3366 IEEE exceptions for math error handling may want to use this flag
3367 for speed while maintaining IEEE arithmetic compatibility.
3369 This option should never be turned on by any @option{-O} option since
3370 it can result in incorrect output for programs which depend on
3371 an exact implementation of IEEE or ISO rules/specifications for
3374 The default is @option{-fmath-errno}.
3376 @item -funsafe-math-optimizations
3377 @opindex funsafe-math-optimizations
3378 Allow optimizations for floating-point arithmetic that (a) assume
3379 that arguments and results are valid and (b) may violate IEEE or
3380 ANSI standards. When used at link-time, it may include libraries
3381 or startup files that change the default FPU control word or other
3382 similar optimizations.
3384 This option should never be turned on by any @option{-O} option since
3385 it can result in incorrect output for programs which depend on
3386 an exact implementation of IEEE or ISO rules/specifications for
3389 The default is @option{-fno-unsafe-math-optimizations}.
3391 @item -fno-trapping-math
3392 @opindex fno-trapping-math
3393 Compile code assuming that floating-point operations cannot generate
3394 user-visible traps. Setting this option may allow faster code
3395 if one relies on ``non-stop'' IEEE arithmetic, for example.
3397 This option should never be turned on by any @option{-O} option since
3398 it can result in incorrect output for programs which depend on
3399 an exact implementation of IEEE or ISO rules/specifications for
3402 The default is @option{-ftrapping-math}.
3405 The following options control specific optimizations. The @option{-O2}
3406 option turns on all of these optimizations except @option{-funroll-loops}
3407 and @option{-funroll-all-loops}. On most machines, the @option{-O} option
3408 turns on the @option{-fthread-jumps} and @option{-fdelayed-branch} options,
3409 but specific machines may handle it differently.
3411 You can use the following flags in the rare cases when ``fine-tuning''
3412 of optimizations to be performed is desired.
3414 Not all of the optimizations performed by GCC have @option{-f} options
3418 @item -fmessy-debugging
3419 @opindex fmessy-debugging
3420 Some optimizations can be stronger if compiler give up ability to generate
3421 at least approximately usefull debug information for optimized programs.
3422 These transformation are not done when @option{-g} switch is not specified,
3423 as politics is to get the code with @option{-g} and without equivalent,
3424 except for debugging section.
3426 This optimization enables such transformations. It is still compatible with
3427 @option{-g}, but debugging resulting programs will be even higher challenge.
3429 @item -fstrength-reduce
3430 @opindex fstrength-reduce
3431 Perform the optimizations of loop strength reduction and
3432 elimination of iteration variables.
3434 @item -fthread-jumps
3435 @opindex fthread-jumps
3436 Perform optimizations where we check to see if a jump branches to a
3437 location where another comparison subsumed by the first is found. If
3438 so, the first branch is redirected to either the destination of the
3439 second branch or a point immediately following it, depending on whether
3440 the condition is known to be true or false.
3442 @item -fcse-follow-jumps
3443 @opindex fcse-follow-jumps
3444 In common subexpression elimination, scan through jump instructions
3445 when the target of the jump is not reached by any other path. For
3446 example, when CSE encounters an @code{if} statement with an
3447 @code{else} clause, CSE will follow the jump when the condition
3450 @item -fcse-skip-blocks
3451 @opindex fcse-skip-blocks
3452 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
3453 follow jumps which conditionally skip over blocks. When CSE
3454 encounters a simple @code{if} statement with no else clause,
3455 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
3456 body of the @code{if}.
3458 @item -frerun-cse-after-loop
3459 @opindex frerun-cse-after-loop
3460 Re-run common subexpression elimination after loop optimizations has been
3463 @item -frerun-loop-opt
3464 @opindex frerun-loop-opt
3465 Run the loop optimizer twice.
3469 Perform a global common subexpression elimination pass.
3470 This pass also performs global constant and copy propagation.
3472 @emph{Note:} When compiling a program using computed gotos, a GCC
3473 extension, you may get better runtime performance if you disable
3474 the global common subexpression elmination pass by adding
3475 @option{-fno-gcse} to the command line.
3479 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
3480 attempt to move loads which are only killed by stores into themselves. This
3481 allows a loop containing a load/store sequence to be changed to a load outside
3482 the loop, and a copy/store within the loop.
3486 When @option{-fgcse-sm} is enabled, A store motion pass is run after global common
3487 subexpression elimination. This pass will attempt to move stores out of loops.
3488 When used in conjunction with @option{-fgcse-lm}, loops containing a load/store sequence
3489 can be changed to a load before the loop and a store after the loop.
3491 @item -fdelete-null-pointer-checks
3492 @opindex fdelete-null-pointer-checks
3493 Use global dataflow analysis to identify and eliminate useless checks
3494 for null pointers. The compiler assumes that dereferencing a null
3495 pointer would have halted the program. If a pointer is checked after
3496 it has already been dereferenced, it cannot be null.
3498 In some environments, this assumption is not true, and programs can
3499 safely dereference null pointers. Use
3500 @option{-fno-delete-null-pointer-checks} to disable this optimization
3501 for programs which depend on that behavior.
3503 @item -fexpensive-optimizations
3504 @opindex fexpensive-optimizations
3505 Perform a number of minor optimizations that are relatively expensive.
3507 @item -foptimize-register-move
3509 @opindex foptimize-register-move
3511 Attempt to reassign register numbers in move instructions and as
3512 operands of other simple instructions in order to maximize the amount of
3513 register tying. This is especially helpful on machines with two-operand
3514 instructions. GCC enables this optimization by default with @option{-O2}
3517 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
3520 @item -fdelayed-branch
3521 @opindex fdelayed-branch
3522 If supported for the target machine, attempt to reorder instructions
3523 to exploit instruction slots available after delayed branch
3526 @item -fschedule-insns
3527 @opindex fschedule-insns
3528 If supported for the target machine, attempt to reorder instructions to
3529 eliminate execution stalls due to required data being unavailable. This
3530 helps machines that have slow floating point or memory load instructions
3531 by allowing other instructions to be issued until the result of the load
3532 or floating point instruction is required.
3534 @item -fschedule-insns2
3535 @opindex fschedule-insns2
3536 Similar to @option{-fschedule-insns}, but requests an additional pass of
3537 instruction scheduling after register allocation has been done. This is
3538 especially useful on machines with a relatively small number of
3539 registers and where memory load instructions take more than one cycle.
3541 @item -ffunction-sections
3542 @itemx -fdata-sections
3543 @opindex ffunction-sections
3544 @opindex fdata-sections
3545 Place each function or data item into its own section in the output
3546 file if the target supports arbitrary sections. The name of the
3547 function or the name of the data item determines the section's name
3550 Use these options on systems where the linker can perform optimizations
3551 to improve locality of reference in the instruction space. HPPA
3552 processors running HP-UX and Sparc processors running Solaris 2 have
3553 linkers with such optimizations. Other systems using the ELF object format
3554 as well as AIX may have these optimizations in the future.
3556 Only use these options when there are significant benefits from doing
3557 so. When you specify these options, the assembler and linker will
3558 create larger object and executable files and will also be slower.
3559 You will not be able to use @code{gprof} on all systems if you
3560 specify this option and you may have problems with debugging if
3561 you specify both this option and @option{-g}.
3563 @item -fcaller-saves
3564 @opindex fcaller-saves
3565 Enable values to be allocated in registers that will be clobbered by
3566 function calls, by emitting extra instructions to save and restore the
3567 registers around such calls. Such allocation is done only when it
3568 seems to result in better code than would otherwise be produced.
3570 This option is always enabled by default on certain machines, usually
3571 those which have no call-preserved registers to use instead.
3573 For all machines, optimization level 2 and higher enables this flag by
3576 @item -funroll-loops
3577 @opindex funroll-loops
3578 Unroll loops whose number of iterations can be determined at compile
3579 time or upon entry to the loop. @option{-funroll-loops} implies both
3580 @option{-fstrength-reduce} and @option{-frerun-cse-after-loop}. This
3581 option makes code larger, and may or may not make it run faster.
3583 @item -funroll-all-loops
3584 @opindex funroll-all-loops
3585 Unroll all loops, even if their number of iterations is uncertain when
3586 the loop is entered. This usually makes programs run more slowly.
3587 @option{-funroll-all-loops} implies the same options as
3588 @option{-funroll-loops},
3591 @item -fmove-all-movables
3592 @opindex fmove-all-movables
3593 Forces all invariant computations in loops to be moved
3596 @item -freduce-all-givs
3597 @opindex freduce-all-givs
3598 Forces all general-induction variables in loops to be
3601 @emph{Note:} When compiling programs written in Fortran,
3602 @option{-fmove-all-movables} and @option{-freduce-all-givs} are enabled
3603 by default when you use the optimizer.
3605 These options may generate better or worse code; results are highly
3606 dependent on the structure of loops within the source code.
3608 These two options are intended to be removed someday, once
3609 they have helped determine the efficacy of various
3610 approaches to improving loop optimizations.
3612 Please let us (@w{@email{gcc@@gcc.gnu.org}} and @w{@email{fortran@@gnu.org}})
3613 know how use of these options affects
3614 the performance of your production code.
3615 We're very interested in code that runs @emph{slower}
3616 when these options are @emph{enabled}.
3619 @itemx -fno-peephole2
3620 @opindex fno-peephole
3621 @opindex fno-peephole2
3622 Disable any machine-specific peephole optimizations. The difference
3623 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
3624 are implemented in the compiler; some targets use one, some use the
3625 other, a few use both.
3627 @item -fbranch-probabilities
3628 @opindex fbranch-probabilities
3629 After running a program compiled with @option{-fprofile-arcs}
3630 (@pxref{Debugging Options,, Options for Debugging Your Program or
3631 @command{gcc}}), you can compile it a second time using
3632 @option{-fbranch-probabilities}, to improve optimizations based on
3633 the number of times each branch was taken. When the program
3634 compiled with @option{-fprofile-arcs} exits it saves arc execution
3635 counts to a file called @file{@var{sourcename}.da} for each source
3636 file The information in this data file is very dependent on the
3637 structure of the generated code, so you must use the same source code
3638 and the same optimization options for both compilations.
3641 With @option{-fbranch-probabilities}, GCC puts a @samp{REG_EXEC_COUNT}
3642 note on the first instruction of each basic block, and a
3643 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
3644 These can be used to improve optimization. Currently, they are only
3645 used in one place: in @file{reorg.c}, instead of guessing which path a
3646 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
3647 exactly determine which path is taken more often.
3650 @item -fno-guess-branch-probability
3651 @opindex fno-guess-branch-probability
3652 Do not guess branch probabilities using a randomized model.
3654 Sometimes gcc will opt to use a randomized model to guess branch
3655 probabilities, when none are available from either profiling feedback
3656 (@option{-fprofile-arcs}) or @samp{__builtin_expect}. This means that
3657 different runs of the compiler on the same program may produce different
3660 In a hard real-time system, people don't want different runs of the
3661 compiler to produce code that has different behavior; minimizing
3662 non-determinism is of paramount import. This switch allows users to
3663 reduce non-determinism, possibly at the expense of inferior
3666 @item -fstrict-aliasing
3667 @opindex fstrict-aliasing
3668 Allows the compiler to assume the strictest aliasing rules applicable to
3669 the language being compiled. For C (and C++), this activates
3670 optimizations based on the type of expressions. In particular, an
3671 object of one type is assumed never to reside at the same address as an
3672 object of a different type, unless the types are almost the same. For
3673 example, an @code{unsigned int} can alias an @code{int}, but not a
3674 @code{void*} or a @code{double}. A character type may alias any other
3677 Pay special attention to code like this:
3690 The practice of reading from a different union member than the one most
3691 recently written to (called ``type-punning'') is common. Even with
3692 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
3693 is accessed through the union type. So, the code above will work as
3694 expected. However, this code might not:
3706 Every language that wishes to perform language-specific alias analysis
3707 should define a function that computes, given an @code{tree}
3708 node, an alias set for the node. Nodes in different alias sets are not
3709 allowed to alias. For an example, see the C front-end function
3710 @code{c_get_alias_set}.
3713 @item -falign-functions
3714 @itemx -falign-functions=@var{n}
3715 @opindex falign-functions
3716 Align the start of functions to the next power-of-two greater than
3717 @var{n}, skipping up to @var{n} bytes. For instance,
3718 @option{-falign-functions=32} aligns functions to the next 32-byte
3719 boundary, but @option{-falign-functions=24} would align to the next
3720 32-byte boundary only if this can be done by skipping 23 bytes or less.
3722 @option{-fno-align-functions} and @option{-falign-functions=1} are
3723 equivalent and mean that functions will not be aligned.
3725 Some assemblers only support this flag when @var{n} is a power of two;
3726 in that case, it is rounded up.
3728 If @var{n} is not specified, use a machine-dependent default.
3730 @item -falign-labels
3731 @itemx -falign-labels=@var{n}
3732 @opindex falign-labels
3733 Align all branch targets to a power-of-two boundary, skipping up to
3734 @var{n} bytes like @option{-falign-functions}. This option can easily
3735 make code slower, because it must insert dummy operations for when the
3736 branch target is reached in the usual flow of the code.
3738 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
3739 are greater than this value, then their values are used instead.
3741 If @var{n} is not specified, use a machine-dependent default which is
3742 very likely to be @samp{1}, meaning no alignment.
3745 @itemx -falign-loops=@var{n}
3746 @opindex falign-loops
3747 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
3748 like @option{-falign-functions}. The hope is that the loop will be
3749 executed many times, which will make up for any execution of the dummy
3752 If @var{n} is not specified, use a machine-dependent default.
3755 @itemx -falign-jumps=@var{n}
3756 @opindex falign-jumps
3757 Align branch targets to a power-of-two boundary, for branch targets
3758 where the targets can only be reached by jumping, skipping up to @var{n}
3759 bytes like @option{-falign-functions}. In this case, no dummy operations
3762 If @var{n} is not specified, use a machine-dependent default.
3766 Perform optimizations in static single assignment form. Each function's
3767 flow graph is translated into SSA form, optimizations are performed, and
3768 the flow graph is translated back from SSA form. Users should not
3769 specify this option, since it is not yet ready for production use.
3773 Perform Sparse Conditional Constant Propagation in SSA form. Requires
3774 @option{-fssa}. Like @option{-fssa}, this is an experimental feature.
3778 Perform aggressive dead-code elimination in SSA form. Requires @option{-fssa}.
3779 Like @option{-fssa}, this is an experimental feature.
3781 @item -fsingle-precision-constant
3782 @opindex fsingle-precision-constant
3783 Treat floating point constant as single precision constant instead of
3784 implicitly converting it to double precision constant.
3786 @item -frename-registers
3787 @opindex frename-registers
3788 Attempt to avoid false dependencies in scheduled code by making use
3789 of registers left over after register allocation. This optimization
3790 will most benefit processors with lots of registers. It can, however,
3791 make debugging impossible, since variables will no longer stay in
3792 a ``home register''.
3794 @item --param @var{name}=@var{value}
3796 In some places, GCC uses various constants to control the amount of
3797 optimization that is done. For example, GCC will not inline functions
3798 that contain more that a certain number of instructions. You can
3799 control some of these constants on the command-line using the
3800 @option{--param} option.
3802 In each case, the @var{value} is a integer. The allowable choices for
3803 @var{name} are given in the following table:
3806 @item max-delay-slot-insn-search
3807 The maximum number of instructions to consider when looking for an
3808 instruction to fill a delay slot. If more than this arbitrary number of
3809 instructions is searched, the time savings from filling the delay slot
3810 will be minimal so stop searching. Increasing values mean more
3811 aggressive optimization, making the compile time increase with probably
3812 small improvement in executable run time.
3814 @item max-delay-slot-live-search
3815 When trying to fill delay slots, the maximum number of instructions to
3816 consider when searching for a block with valid live register
3817 information. Increasing this arbitrarily chosen value means more
3818 aggressive optimization, increasing the compile time. This parameter
3819 should be removed when the delay slot code is rewritten to maintain the
3822 @item max-gcse-memory
3823 The approximate maximum amount of memory that will be allocated in
3824 order to perform the global common subexpression elimination
3825 optimization. If more memory than specified is required, the
3826 optimization will not be done.
3828 @item max-gcse-passes
3829 The maximum number of passes of GCSE to run.
3831 @item max-pending-list-length
3832 The maximum number of pending dependancies scheduling will allow
3833 before flushing the current state and starting over. Large functions
3834 with few branches or calls can create excessively large lists which
3835 needlessly consume memory and resources.
3837 @item max-inline-insns
3838 If an function contains more than this many instructions, it
3839 will not be inlined. This option is precisely equivalent to
3840 @option{-finline-limit}.
3845 @node Preprocessor Options
3846 @section Options Controlling the Preprocessor
3847 @cindex preprocessor options
3848 @cindex options, preprocessor
3850 These options control the C preprocessor, which is run on each C source
3851 file before actual compilation.
3853 If you use the @option{-E} option, nothing is done except preprocessing.
3854 Some of these options make sense only together with @option{-E} because
3855 they cause the preprocessor output to be unsuitable for actual
3859 @item -include @var{file}
3861 Process @var{file} as input before processing the regular input file.
3862 In effect, the contents of @var{file} are compiled first. Any @option{-D}
3863 and @option{-U} options on the command line are always processed before
3864 @option{-include @var{file}}, regardless of the order in which they are
3865 written. All the @option{-include} and @option{-imacros} options are
3866 processed in the order in which they are written.
3868 @item -imacros @var{file}
3870 Process @var{file} as input, discarding the resulting output, before
3871 processing the regular input file. Because the output generated from
3872 @var{file} is discarded, the only effect of @option{-imacros @var{file}}
3873 is to make the macros defined in @var{file} available for use in the
3874 main input. All the @option{-include} and @option{-imacros} options are
3875 processed in the order in which they are written.
3877 @item -idirafter @var{dir}
3879 @cindex second include path
3880 Add the directory @var{dir} to the second include path. The directories
3881 on the second include path are searched when a header file is not found
3882 in any of the directories in the main include path (the one that
3883 @option{-I} adds to).
3885 @item -iprefix @var{prefix}
3887 Specify @var{prefix} as the prefix for subsequent @option{-iwithprefix}
3890 @item -iwithprefix @var{dir}
3891 @opindex iwithprefix
3892 Add a directory to the second include path. The directory's name is
3893 made by concatenating @var{prefix} and @var{dir}, where @var{prefix} was
3894 specified previously with @option{-iprefix}. If you have not specified a
3895 prefix yet, the directory containing the installed passes of the
3896 compiler is used as the default.
3898 @item -iwithprefixbefore @var{dir}
3899 @opindex iwithprefixbefore
3900 Add a directory to the main include path. The directory's name is made
3901 by concatenating @var{prefix} and @var{dir}, as in the case of
3902 @option{-iwithprefix}.
3904 @item -isystem @var{dir}
3906 Add a directory to the beginning of the second include path, marking it
3907 as a system directory, so that it gets the same special treatment as
3908 is applied to the standard system directories.
3912 Do not search the standard system directories for header files. Only
3913 the directories you have specified with @option{-I} options (and the
3914 current directory, if appropriate) are searched. @xref{Directory
3915 Options}, for information on @option{-I}.
3917 By using both @option{-nostdinc} and @option{-I-}, you can limit the include-file
3918 search path to only those directories you specify explicitly.
3922 When searching for a header file in a directory, remap file names if a
3923 file named @file{header.gcc} exists in that directory. This can be used
3924 to work around limitations of file systems with file name restrictions.
3925 The @file{header.gcc} file should contain a series of lines with two
3926 tokens on each line: the first token is the name to map, and the second
3927 token is the actual name to use.
3931 Do not predefine any nonstandard macros. (Including architecture flags).
3935 Run only the C preprocessor. Preprocess all the C source files
3936 specified and output the results to standard output or to the
3937 specified output file.
3941 Tell the preprocessor not to discard comments. Used with the
3946 Tell the preprocessor not to generate @samp{#line} directives.
3947 Used with the @option{-E} option.
3950 @cindex dependencies, make
3953 Instead of outputting the result of preprocessing, output a rule
3954 suitable for @code{make} describing the dependencies of the main source
3955 file. The preprocessor outputs one @code{make} rule containing the
3956 object file name for that source file, a colon, and the names of all the
3957 included files. Unless overridden explicitly, the object file name
3958 consists of the basename of the source file with any suffix replaced with
3959 object file suffix. If there are many included files then the
3960 rule is split into several lines using @samp{\}-newline.
3962 @option{-M} implies @option{-E}.
3966 Like @option{-M}, but mention only the files included with @samp{#include
3967 "@var{file}"}. System header files included with @samp{#include
3968 <@var{file}>} are omitted.
3972 Like @option{-M} but the dependency information is written to a file
3973 rather than stdout. @code{gcc} will use the same file name and
3974 directory as the object file, but with the suffix @file{.d} instead.
3976 This is in addition to compiling the main file as specified---@option{-MD}
3977 does not inhibit ordinary compilation the way @option{-M} does,
3978 unless you also specify @option{-MG}.
3980 With Mach, you can use the utility @code{md} to merge multiple
3981 dependency files into a single dependency file suitable for using with
3982 the @samp{make} command.
3986 Like @option{-MD} except mention only user header files, not system
3989 @item -MF @var{file}
3991 When used with @option{-M} or @option{-MM}, specifies a file to write the
3992 dependencies to. This allows the preprocessor to write the preprocessed
3993 file to stdout normally. If no @option{-MF} switch is given, CPP sends
3994 the rules to stdout and suppresses normal preprocessed output.
3996 Another way to specify output of a @code{make} rule is by setting
3997 the environment variable @env{DEPENDENCIES_OUTPUT} (@pxref{Environment
4002 When used with @option{-M} or @option{-MM}, @option{-MG} says to treat missing
4003 header files as generated files and assume they live in the same
4004 directory as the source file. It suppresses preprocessed output, as a
4005 missing header file is ordinarily an error.
4007 This feature is used in automatic updating of makefiles.
4011 This option instructs CPP to add a phony target for each dependency
4012 other than the main file, causing each to depend on nothing. These
4013 dummy rules work around errors @code{make} gives if you remove header
4014 files without updating the @code{Makefile} to match.
4016 This is typical output:-
4019 /tmp/test.o: /tmp/test.c /tmp/test.h
4024 @item -MQ @var{target}
4025 @item -MT @var{target}
4028 By default CPP uses the main file name, including any path, and appends
4029 the object suffix, normally ``.o'', to it to obtain the name of the
4030 target for dependency generation. With @option{-MT} you can specify a
4031 target yourself, overriding the default one.
4033 If you want multiple targets, you can specify them as a single argument
4034 to @option{-MT}, or use multiple @option{-MT} options.
4036 The targets you specify are output in the order they appear on the
4037 command line. @option{-MQ} is identical to @option{-MT}, except that the
4038 target name is quoted for Make, but with @option{-MT} it isn't. For
4039 example, @option{-MT '$(objpfx)foo.o'} gives
4042 $(objpfx)foo.o: /tmp/foo.c
4045 but @option{-MQ '$(objpfx)foo.o'} gives
4048 $$(objpfx)foo.o: /tmp/foo.c
4051 The default target is automatically quoted, as if it were given with
4056 Print the name of each header file used, in addition to other normal
4059 @item -A@var{question}(@var{answer})
4061 Assert the answer @var{answer} for @var{question}, in case it is tested
4062 with a preprocessing conditional such as @samp{#if
4063 #@var{question}(@var{answer})}. @option{-A-} disables the standard
4064 assertions that normally describe the target machine.
4068 Define macro @var{macro} with the string @samp{1} as its definition.
4070 @item -D@var{macro}=@var{defn}
4071 Define macro @var{macro} as @var{defn}. All instances of @option{-D} on
4072 the command line are processed before any @option{-U} options.
4074 Any @option{-D} and @option{-U} options on the command line are processed in
4075 order, and always before @option{-imacros @var{file}}, regardless of the
4076 order in which they are written.
4080 Undefine macro @var{macro}. @option{-U} options are evaluated after all
4081 @option{-D} options, but before any @option{-include} and @option{-imacros}
4084 Any @option{-D} and @option{-U} options on the command line are processed in
4085 order, and always before @option{-imacros @var{file}}, regardless of the
4086 order in which they are written.
4090 Tell the preprocessor to output only a list of the macro definitions
4091 that are in effect at the end of preprocessing. Used with the @option{-E}
4096 Tell the preprocessing to pass all macro definitions into the output, in
4097 their proper sequence in the rest of the output.
4101 Like @option{-dD} except that the macro arguments and contents are omitted.
4102 Only @samp{#define @var{name}} is included in the output.
4106 Output @samp{#include} directives in addition to the result of
4109 @item -fpreprocessed
4110 @opindex fpreprocessed
4111 Indicate to the preprocessor that the input file has already been
4112 preprocessed. This suppresses things like macro expansion, trigraph
4113 conversion, escaped newline splicing, and processing of most directives.
4114 The preprocessor still recognizes and removes comments, so that you can
4115 pass a file preprocessed with @option{-C} to the compiler without
4116 problems. In this mode the integrated preprocessor is little more than
4117 a tokenizer for the front ends.
4119 @option{-fpreprocessed} is implicit if the input file has one of the
4120 extensions @samp{i}, @samp{ii} or @samp{mi}. These are the extensions
4121 that GCC uses for preprocessed files created by @option{-save-temps}.
4125 Process ISO standard trigraph sequences. These are three-character
4126 sequences, all starting with @samp{??}, that are defined by ISO C to
4127 stand for single characters. For example, @samp{??/} stands for
4128 @samp{\}, so @samp{'??/n'} is a character constant for a newline. By
4129 default, GCC ignores trigraphs, but in standard-conforming modes it
4130 converts them. See the @option{-std} and @option{-ansi} options.
4132 The nine trigraph sequences are
4135 @expansion{} @samp{[}
4138 @expansion{} @samp{]}
4141 @expansion{} @samp{@{}
4144 @expansion{} @samp{@}}
4147 @expansion{} @samp{#}
4150 @expansion{} @samp{\}
4153 @expansion{} @samp{^}
4156 @expansion{} @samp{|}
4159 @expansion{} @samp{~}
4163 Trigraph support is not popular, so many compilers do not implement it
4164 properly. Portable code should not rely on trigraphs being either
4165 converted or ignored.
4167 @item -Wp,@var{option}
4169 Pass @var{option} as an option to the preprocessor. If @var{option}
4170 contains commas, it is split into multiple options at the commas.
4173 @node Assembler Options
4174 @section Passing Options to the Assembler
4176 @c prevent bad page break with this line
4177 You can pass options to the assembler.
4180 @item -Wa,@var{option}
4182 Pass @var{option} as an option to the assembler. If @var{option}
4183 contains commas, it is split into multiple options at the commas.
4187 @section Options for Linking
4188 @cindex link options
4189 @cindex options, linking
4191 These options come into play when the compiler links object files into
4192 an executable output file. They are meaningless if the compiler is
4193 not doing a link step.
4197 @item @var{object-file-name}
4198 A file name that does not end in a special recognized suffix is
4199 considered to name an object file or library. (Object files are
4200 distinguished from libraries by the linker according to the file
4201 contents.) If linking is done, these object files are used as input
4210 If any of these options is used, then the linker is not run, and
4211 object file names should not be used as arguments. @xref{Overall
4215 @item -l@var{library}
4216 @itemx -l @var{library}
4218 Search the library named @var{library} when linking. (The second
4219 alternative with the library as a separate argument is only for
4220 POSIX compliance and is not recommended.)
4222 It makes a difference where in the command you write this option; the
4223 linker searches and processes libraries and object files in the order they
4224 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
4225 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
4226 to functions in @samp{z}, those functions may not be loaded.
4228 The linker searches a standard list of directories for the library,
4229 which is actually a file named @file{lib@var{library}.a}. The linker
4230 then uses this file as if it had been specified precisely by name.
4232 The directories searched include several standard system directories
4233 plus any that you specify with @option{-L}.
4235 Normally the files found this way are library files---archive files
4236 whose members are object files. The linker handles an archive file by
4237 scanning through it for members which define symbols that have so far
4238 been referenced but not defined. But if the file that is found is an
4239 ordinary object file, it is linked in the usual fashion. The only
4240 difference between using an @option{-l} option and specifying a file name
4241 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
4242 and searches several directories.
4246 You need this special case of the @option{-l} option in order to
4247 link an Objective-C program.
4250 @opindex nostartfiles
4251 Do not use the standard system startup files when linking.
4252 The standard system libraries are used normally, unless @option{-nostdlib}
4253 or @option{-nodefaultlibs} is used.
4255 @item -nodefaultlibs
4256 @opindex nodefaultlibs
4257 Do not use the standard system libraries when linking.
4258 Only the libraries you specify will be passed to the linker.
4259 The standard startup files are used normally, unless @option{-nostartfiles}
4260 is used. The compiler may generate calls to memcmp, memset, and memcpy
4261 for System V (and ISO C) environments or to bcopy and bzero for
4262 BSD environments. These entries are usually resolved by entries in
4263 libc. These entry points should be supplied through some other
4264 mechanism when this option is specified.
4268 Do not use the standard system startup files or libraries when linking.
4269 No startup files and only the libraries you specify will be passed to
4270 the linker. The compiler may generate calls to memcmp, memset, and memcpy
4271 for System V (and ISO C) environments or to bcopy and bzero for
4272 BSD environments. These entries are usually resolved by entries in
4273 libc. These entry points should be supplied through some other
4274 mechanism when this option is specified.
4276 @cindex @option{-lgcc}, use with @option{-nostdlib}
4277 @cindex @option{-nostdlib} and unresolved references
4278 @cindex unresolved references and @option{-nostdlib}
4279 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
4280 @cindex @option{-nodefaultlibs} and unresolved references
4281 @cindex unresolved references and @option{-nodefaultlibs}
4282 One of the standard libraries bypassed by @option{-nostdlib} and
4283 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
4284 that GCC uses to overcome shortcomings of particular machines, or special
4285 needs for some languages.
4287 (@xref{Interface,,Interfacing to GCC Output}, for more discussion of
4291 (@xref{Interface,,Interfacing to GCC Output,gcc.info,Porting GCC},
4292 for more discussion of @file{libgcc.a}.)
4294 In most cases, you need @file{libgcc.a} even when you want to avoid
4295 other standard libraries. In other words, when you specify @option{-nostdlib}
4296 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
4297 This ensures that you have no unresolved references to internal GCC
4298 library subroutines. (For example, @samp{__main}, used to ensure C++
4299 constructors will be called; @pxref{Collect2,,@command{collect2}}.)
4303 Remove all symbol table and relocation information from the executable.
4307 On systems that support dynamic linking, this prevents linking with the shared
4308 libraries. On other systems, this option has no effect.
4312 Produce a shared object which can then be linked with other objects to
4313 form an executable. Not all systems support this option. For predictable
4314 results, you must also specify the same set of options that were used to
4315 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
4316 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
4317 needs to build supplementary stub code for constructors to work. On
4318 multi-libbed systems, @samp{gcc -shared} must select the correct support
4319 libraries to link against. Failing to supply the correct flags may lead
4320 to subtle defects. Supplying them in cases where they are not necessary
4323 @item -shared-libgcc
4324 @itemx -static-libgcc
4325 @opindex shared-libgcc
4326 @opindex static-libgcc
4327 On systems that provide @file{libgcc} as a shared library, these options
4328 force the use of either the shared or static version respectively.
4329 If no shared version of @file{libgcc} was built when the compiler was
4330 configured, these options have no effect.
4332 There are several situations in which an application should use the
4333 shared @file{libgcc} instead of the static version. The most common
4334 of these is when the application wishes to throw and catch exceptions
4335 across different shared libraries. In that case, each of the libraries
4336 as well as the application itself should use the shared @file{libgcc}.
4338 Therefore, whenever you specify the @option{-shared} option, the GCC
4339 driver automatically adds @option{-shared-libgcc}, unless you explicitly
4340 specify @option{-static-libgcc}. The G++ driver automatically adds
4341 @option{-shared-libgcc} when you build a main executable as well because
4342 for C++ programs that is typically the right thing to do.
4343 (Exception-handling will not work reliably otherwise.)
4345 However, when linking a main executable written in C, you must
4346 explicitly say @option{-shared-libgcc} if you want to use the shared
4351 Bind references to global symbols when building a shared object. Warn
4352 about any unresolved references (unless overridden by the link editor
4353 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
4356 @item -Xlinker @var{option}
4358 Pass @var{option} as an option to the linker. You can use this to
4359 supply system-specific linker options which GCC does not know how to
4362 If you want to pass an option that takes an argument, you must use
4363 @option{-Xlinker} twice, once for the option and once for the argument.
4364 For example, to pass @option{-assert definitions}, you must write
4365 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
4366 @option{-Xlinker "-assert definitions"}, because this passes the entire
4367 string as a single argument, which is not what the linker expects.
4369 @item -Wl,@var{option}
4371 Pass @var{option} as an option to the linker. If @var{option} contains
4372 commas, it is split into multiple options at the commas.
4374 @item -u @var{symbol}
4376 Pretend the symbol @var{symbol} is undefined, to force linking of
4377 library modules to define it. You can use @option{-u} multiple times with
4378 different symbols to force loading of additional library modules.
4381 @node Directory Options
4382 @section Options for Directory Search
4383 @cindex directory options
4384 @cindex options, directory search
4387 These options specify directories to search for header files, for
4388 libraries and for parts of the compiler:
4393 Add the directory @var{dir} to the head of the list of directories to be
4394 searched for header files. This can be used to override a system header
4395 file, substituting your own version, since these directories are
4396 searched before the system header file directories. However, you should
4397 not use this option to add directories that contain vendor-supplied
4398 system header files (use @option{-isystem} for that). If you use more than
4399 one @option{-I} option, the directories are scanned in left-to-right
4400 order; the standard system directories come after.
4402 If a standard system include directory, or a directory specified with
4403 @option{-isystem}, is also specified with @option{-I}, it will be
4404 searched only in the position requested by @option{-I}. Also, it will
4405 not be considered a system include directory. If that directory really
4406 does contain system headers, there is a good chance that they will
4407 break. For instance, if GCC's installation procedure edited the headers
4408 in @file{/usr/include} to fix bugs, @samp{-I/usr/include} will cause the
4409 original, buggy headers to be found instead of the corrected ones. GCC
4410 will issue a warning when a system include directory is hidden in this
4415 Any directories you specify with @option{-I} options before the @option{-I-}
4416 option are searched only for the case of @samp{#include "@var{file}"};
4417 they are not searched for @samp{#include <@var{file}>}.
4419 If additional directories are specified with @option{-I} options after
4420 the @option{-I-}, these directories are searched for all @samp{#include}
4421 directives. (Ordinarily @emph{all} @option{-I} directories are used
4424 In addition, the @option{-I-} option inhibits the use of the current
4425 directory (where the current input file came from) as the first search
4426 directory for @samp{#include "@var{file}"}. There is no way to
4427 override this effect of @option{-I-}. With @option{-I.} you can specify
4428 searching the directory which was current when the compiler was
4429 invoked. That is not exactly the same as what the preprocessor does
4430 by default, but it is often satisfactory.
4432 @option{-I-} does not inhibit the use of the standard system directories
4433 for header files. Thus, @option{-I-} and @option{-nostdinc} are
4438 Add directory @var{dir} to the list of directories to be searched
4441 @item -B@var{prefix}
4443 This option specifies where to find the executables, libraries,
4444 include files, and data files of the compiler itself.
4446 The compiler driver program runs one or more of the subprograms
4447 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
4448 @var{prefix} as a prefix for each program it tries to run, both with and
4449 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
4451 For each subprogram to be run, the compiler driver first tries the
4452 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
4453 was not specified, the driver tries two standard prefixes, which are
4454 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc-lib/}. If neither of
4455 those results in a file name that is found, the unmodified program
4456 name is searched for using the directories specified in your
4457 @env{PATH} environment variable.
4459 The compiler will check to see if the path provided by the @option{-B}
4460 refers to a directory, and if necessary it will add a directory
4461 separator character at the end of the path.
4463 @option{-B} prefixes that effectively specify directory names also apply
4464 to libraries in the linker, because the compiler translates these
4465 options into @option{-L} options for the linker. They also apply to
4466 includes files in the preprocessor, because the compiler translates these
4467 options into @option{-isystem} options for the preprocessor. In this case,
4468 the compiler appends @samp{include} to the prefix.
4470 The run-time support file @file{libgcc.a} can also be searched for using
4471 the @option{-B} prefix, if needed. If it is not found there, the two
4472 standard prefixes above are tried, and that is all. The file is left
4473 out of the link if it is not found by those means.
4475 Another way to specify a prefix much like the @option{-B} prefix is to use
4476 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
4479 As a special kludge, if the path provided by @option{-B} is
4480 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
4481 9, then it will be replaced by @file{[dir/]include}. This is to help
4482 with boot-strapping the compiler.
4484 @item -specs=@var{file}
4486 Process @var{file} after the compiler reads in the standard @file{specs}
4487 file, in order to override the defaults that the @file{gcc} driver
4488 program uses when determining what switches to pass to @file{cc1},
4489 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
4490 @option{-specs=@var{file}} can be specified on the command line, and they
4491 are processed in order, from left to right.
4497 @section Specifying subprocesses and the switches to pass to them
4499 @command{gcc} is a driver program. It performs its job by invoking a
4500 sequence of other programs to do the work of compiling, assembling and
4501 linking. GCC interprets its command-line parameters and uses these to
4502 deduce which programs it should invoke, and which command-line options
4503 it ought to place on their command lines. This behaviour is controlled
4504 by @dfn{spec strings}. In most cases there is one spec string for each
4505 program that GCC can invoke, but a few programs have multiple spec
4506 strings to control their behaviour. The spec strings built into GCC can
4507 be overridden by using the @option{-specs=} command-line switch to specify
4510 @dfn{Spec files} are plaintext files that are used to construct spec
4511 strings. They consist of a sequence of directives separated by blank
4512 lines. The type of directive is determined by the first non-whitespace
4513 character on the line and it can be one of the following:
4516 @item %@var{command}
4517 Issues a @var{command} to the spec file processor. The commands that can
4521 @item %include <@var{file}>
4523 Search for @var{file} and insert its text at the current point in the
4526 @item %include_noerr <@var{file}>
4527 @cindex %include_noerr
4528 Just like @samp{%include}, but do not generate an error message if the include
4529 file cannot be found.
4531 @item %rename @var{old_name} @var{new_name}
4533 Rename the spec string @var{old_name} to @var{new_name}.
4537 @item *[@var{spec_name}]:
4538 This tells the compiler to create, override or delete the named spec
4539 string. All lines after this directive up to the next directive or
4540 blank line are considered to be the text for the spec string. If this
4541 results in an empty string then the spec will be deleted. (Or, if the
4542 spec did not exist, then nothing will happened.) Otherwise, if the spec
4543 does not currently exist a new spec will be created. If the spec does
4544 exist then its contents will be overridden by the text of this
4545 directive, unless the first character of that text is the @samp{+}
4546 character, in which case the text will be appended to the spec.
4548 @item [@var{suffix}]:
4549 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
4550 and up to the next directive or blank line are considered to make up the
4551 spec string for the indicated suffix. When the compiler encounters an
4552 input file with the named suffix, it will processes the spec string in
4553 order to work out how to compile that file. For example:
4560 This says that any input file whose name ends in @samp{.ZZ} should be
4561 passed to the program @samp{z-compile}, which should be invoked with the
4562 command-line switch @option{-input} and with the result of performing the
4563 @samp{%i} substitution. (See below.)
4565 As an alternative to providing a spec string, the text that follows a
4566 suffix directive can be one of the following:
4569 @item @@@var{language}
4570 This says that the suffix is an alias for a known @var{language}. This is
4571 similar to using the @option{-x} command-line switch to GCC to specify a
4572 language explicitly. For example:
4579 Says that .ZZ files are, in fact, C++ source files.
4582 This causes an error messages saying:
4585 @var{name} compiler not installed on this system.
4589 GCC already has an extensive list of suffixes built into it.
4590 This directive will add an entry to the end of the list of suffixes, but
4591 since the list is searched from the end backwards, it is effectively
4592 possible to override earlier entries using this technique.
4596 GCC has the following spec strings built into it. Spec files can
4597 override these strings or create their own. Note that individual
4598 targets can also add their own spec strings to this list.
4601 asm Options to pass to the assembler
4602 asm_final Options to pass to the assembler post-processor
4603 cpp Options to pass to the C preprocessor
4604 cc1 Options to pass to the C compiler
4605 cc1plus Options to pass to the C++ compiler
4606 endfile Object files to include at the end of the link
4607 link Options to pass to the linker
4608 lib Libraries to include on the command line to the linker
4609 libgcc Decides which GCC support library to pass to the linker
4610 linker Sets the name of the linker
4611 predefines Defines to be passed to the C preprocessor
4612 signed_char Defines to pass to CPP to say whether @code{char} is signed
4614 startfile Object files to include at the start of the link
4617 Here is a small example of a spec file:
4623 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
4626 This example renames the spec called @samp{lib} to @samp{old_lib} and
4627 then overrides the previous definition of @samp{lib} with a new one.
4628 The new definition adds in some extra command-line options before
4629 including the text of the old definition.
4631 @dfn{Spec strings} are a list of command-line options to be passed to their
4632 corresponding program. In addition, the spec strings can contain
4633 @samp{%}-prefixed sequences to substitute variable text or to
4634 conditionally insert text into the command line. Using these constructs
4635 it is possible to generate quite complex command lines.
4637 Here is a table of all defined @samp{%}-sequences for spec
4638 strings. Note that spaces are not generated automatically around the
4639 results of expanding these sequences. Therefore you can concatenate them
4640 together or combine them with constant text in a single argument.
4644 Substitute one @samp{%} into the program name or argument.
4647 Substitute the name of the input file being processed.
4650 Substitute the basename of the input file being processed.
4651 This is the substring up to (and not including) the last period
4652 and not including the directory.
4655 This is the same as @samp{%b}, but include the file suffix (text after
4659 Marks the argument containing or following the @samp{%d} as a
4660 temporary file name, so that that file will be deleted if GCC exits
4661 successfully. Unlike @samp{%g}, this contributes no text to the
4664 @item %g@var{suffix}
4665 Substitute a file name that has suffix @var{suffix} and is chosen
4666 once per compilation, and mark the argument in the same way as
4667 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
4668 name is now chosen in a way that is hard to predict even when previously
4669 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
4670 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
4671 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
4672 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
4673 was simply substituted with a file name chosen once per compilation,
4674 without regard to any appended suffix (which was therefore treated
4675 just like ordinary text), making such attacks more likely to succeed.
4677 @item %u@var{suffix}
4678 Like @samp{%g}, but generates a new temporary file name even if
4679 @samp{%u@var{suffix}} was already seen.
4681 @item %U@var{suffix}
4682 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
4683 new one if there is no such last file name. In the absence of any
4684 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
4685 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
4686 would involve the generation of two distinct file names, one
4687 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
4688 simply substituted with a file name chosen for the previous @samp{%u},
4689 without regard to any appended suffix.
4691 @item %j@var{SUFFIX}
4692 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
4693 writable, and if save-temps is off; otherwise, substitute the name
4694 of a temporary file, just like @samp{%u}. This temporary file is not
4695 meant for communication between processes, but rather as a junk
4698 @item %.@var{SUFFIX}
4699 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
4700 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
4701 terminated by the next space or %.
4704 Marks the argument containing or following the @samp{%w} as the
4705 designated output file of this compilation. This puts the argument
4706 into the sequence of arguments that @samp{%o} will substitute later.
4709 Substitutes the names of all the output files, with spaces
4710 automatically placed around them. You should write spaces
4711 around the @samp{%o} as well or the results are undefined.
4712 @samp{%o} is for use in the specs for running the linker.
4713 Input files whose names have no recognized suffix are not compiled
4714 at all, but they are included among the output files, so they will
4718 Substitutes the suffix for object files. Note that this is
4719 handled specially when it immediately follows @samp{%g, %u, or %U},
4720 because of the need for those to form complete file names. The
4721 handling is such that @samp{%O} is treated exactly as if it had already
4722 been substituted, except that @samp{%g, %u, and %U} do not currently
4723 support additional @var{suffix} characters following @samp{%O} as they would
4724 following, for example, @samp{.o}.
4727 Substitutes the standard macro predefinitions for the
4728 current target machine. Use this when running @code{cpp}.
4731 Like @samp{%p}, but puts @samp{__} before and after the name of each
4732 predefined macro, except for macros that start with @samp{__} or with
4733 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
4737 Substitute a @option{-iprefix} option made from @env{GCC_EXEC_PREFIX}.
4740 Current argument is the name of a library or startup file of some sort.
4741 Search for that file in a standard list of directories and substitute
4742 the full name found.
4745 Print @var{str} as an error message. @var{str} is terminated by a newline.
4746 Use this when inconsistent options are detected.
4749 Output @samp{-} if the input for the current command is coming from a pipe.
4752 Substitute the contents of spec string @var{name} at this point.
4755 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
4757 @item %x@{@var{option}@}
4758 Accumulate an option for @samp{%X}.
4761 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
4765 Output the accumulated assembler options specified by @option{-Wa}.
4768 Output the accumulated preprocessor options specified by @option{-Wp}.
4771 Substitute the major version number of GCC@.
4772 (For version 2.9.5, this is 2.)
4775 Substitute the minor version number of GCC@.
4776 (For version 2.9.5, this is 9.)
4779 Substitute the patch level number of GCC@.
4780 (For version 2.9.5, this is 5.)
4783 Process the @code{asm} spec. This is used to compute the
4784 switches to be passed to the assembler.
4787 Process the @code{asm_final} spec. This is a spec string for
4788 passing switches to an assembler post-processor, if such a program is
4792 Process the @code{link} spec. This is the spec for computing the
4793 command line passed to the linker. Typically it will make use of the
4794 @samp{%L %G %S %D and %E} sequences.
4797 Dump out a @option{-L} option for each directory that GCC believes might
4798 contain startup files. If the target supports multilibs then the
4799 current multilib directory will be prepended to each of these paths.
4802 Output the multilib directory with directory separators replaced with
4803 @samp{_}. If multilib directories are not set, or the multilib directory is
4804 @file{.} then this option emits nothing.
4807 Process the @code{lib} spec. This is a spec string for deciding which
4808 libraries should be included on the command line to the linker.
4811 Process the @code{libgcc} spec. This is a spec string for deciding
4812 which GCC support library should be included on the command line to the linker.
4815 Process the @code{startfile} spec. This is a spec for deciding which
4816 object files should be the first ones passed to the linker. Typically
4817 this might be a file named @file{crt0.o}.
4820 Process the @code{endfile} spec. This is a spec string that specifies
4821 the last object files that will be passed to the linker.
4824 Process the @code{cpp} spec. This is used to construct the arguments
4825 to be passed to the C preprocessor.
4828 Process the @code{signed_char} spec. This is intended to be used
4829 to tell cpp whether a char is signed. It typically has the definition:
4831 %@{funsigned-char:-D__CHAR_UNSIGNED__@}
4835 Process the @code{cc1} spec. This is used to construct the options to be
4836 passed to the actual C compiler (@samp{cc1}).
4839 Process the @code{cc1plus} spec. This is used to construct the options to be
4840 passed to the actual C++ compiler (@samp{cc1plus}).
4843 Substitute the variable part of a matched option. See below.
4844 Note that each comma in the substituted string is replaced by
4848 Substitutes the @code{-S} switch, if that switch was given to GCC@.
4849 If that switch was not specified, this substitutes nothing. Note that
4850 the leading dash is omitted when specifying this option, and it is
4851 automatically inserted if the substitution is performed. Thus the spec
4852 string @samp{%@{foo@}} would match the command-line option @option{-foo}
4853 and would output the command line option @option{-foo}.
4855 @item %W@{@code{S}@}
4856 Like %@{@code{S}@} but mark last argument supplied within as a file to be
4859 @item %@{@code{S}*@}
4860 Substitutes all the switches specified to GCC whose names start
4861 with @code{-S}, but which also take an argument. This is used for
4862 switches like @option{-o}, @option{-D}, @option{-I}, etc.
4863 GCC considers @option{-o foo} as being
4864 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
4865 text, including the space. Thus two arguments would be generated.
4867 @item %@{^@code{S}*@}
4868 Like %@{@code{S}*@}, but don't put a blank between a switch and its
4869 argument. Thus %@{^o*@} would only generate one argument, not two.
4871 @item %@{@code{S}*&@code{T}*@}
4872 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
4873 (the order of @code{S} and @code{T} in the spec is not significant).
4874 There can be any number of ampersand-separated variables; for each the
4875 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
4877 @item %@{<@code{S}@}
4878 Remove all occurrences of @code{-S} from the command line. Note---this
4879 command is position dependent. @samp{%} commands in the spec string
4880 before this option will see @code{-S}, @samp{%} commands in the spec
4881 string after this option will not.
4883 @item %@{@code{S}*:@code{X}@}
4884 Substitutes @code{X} if one or more switches whose names start with
4885 @code{-S} are specified to GCC@. Note that the tail part of the
4886 @code{-S} option (i.e.@: the part matched by the @samp{*}) will be substituted
4887 for each occurrence of @samp{%*} within @code{X}.
4889 @item %@{@code{S}:@code{X}@}
4890 Substitutes @code{X}, but only if the @samp{-S} switch was given to GCC@.
4892 @item %@{!@code{S}:@code{X}@}
4893 Substitutes @code{X}, but only if the @samp{-S} switch was @emph{not} given to GCC@.
4895 @item %@{|@code{S}:@code{X}@}
4896 Like %@{@code{S}:@code{X}@}, but if no @code{S} switch, substitute @samp{-}.
4898 @item %@{|!@code{S}:@code{X}@}
4899 Like %@{!@code{S}:@code{X}@}, but if there is an @code{S} switch, substitute @samp{-}.
4901 @item %@{.@code{S}:@code{X}@}
4902 Substitutes @code{X}, but only if processing a file with suffix @code{S}.
4904 @item %@{!.@code{S}:@code{X}@}
4905 Substitutes @code{X}, but only if @emph{not} processing a file with suffix @code{S}.
4907 @item %@{@code{S}|@code{P}:@code{X}@}
4908 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC@. This may be
4909 combined with @samp{!} and @samp{.} sequences as well, although they
4910 have a stronger binding than the @samp{|}. For example a spec string
4914 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
4917 will output the following command-line options from the following input
4918 command-line options:
4923 -d fred.c -foo -baz -boggle
4924 -d jim.d -bar -baz -boggle
4929 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or
4930 %@{!@code{S}:@code{X}@} construct may contain other nested @samp{%} constructs
4931 or spaces, or even newlines. They are processed as usual, as described
4934 The @option{-O}, @option{-f}, @option{-m}, and @option{-W}
4935 switches are handled specifically in these
4936 constructs. If another value of @option{-O} or the negated form of a @option{-f}, @option{-m}, or
4937 @option{-W} switch is found later in the command line, the earlier switch
4938 value is ignored, except with @{@code{S}*@} where @code{S} is just one
4939 letter, which passes all matching options.
4941 The character @samp{|} at the beginning of the predicate text is used to indicate
4942 that a command should be piped to the following command, but only if @option{-pipe}
4945 It is built into GCC which switches take arguments and which do not.
4946 (You might think it would be useful to generalize this to allow each
4947 compiler's spec to say which switches take arguments. But this cannot
4948 be done in a consistent fashion. GCC cannot even decide which input
4949 files have been specified without knowing which switches take arguments,
4950 and it must know which input files to compile in order to tell which
4953 GCC also knows implicitly that arguments starting in @option{-l} are to be
4954 treated as compiler output files, and passed to the linker in their
4955 proper position among the other output files.
4957 @c man begin OPTIONS
4959 @node Target Options
4960 @section Specifying Target Machine and Compiler Version
4961 @cindex target options
4962 @cindex cross compiling
4963 @cindex specifying machine version
4964 @cindex specifying compiler version and target machine
4965 @cindex compiler version, specifying
4966 @cindex target machine, specifying
4968 By default, GCC compiles code for the same type of machine that you
4969 are using. However, it can also be installed as a cross-compiler, to
4970 compile for some other type of machine. In fact, several different
4971 configurations of GCC, for different target machines, can be
4972 installed side by side. Then you specify which one to use with the
4975 In addition, older and newer versions of GCC can be installed side
4976 by side. One of them (probably the newest) will be the default, but
4977 you may sometimes wish to use another.
4980 @item -b @var{machine}
4982 The argument @var{machine} specifies the target machine for compilation.
4983 This is useful when you have installed GCC as a cross-compiler.
4985 The value to use for @var{machine} is the same as was specified as the
4986 machine type when configuring GCC as a cross-compiler. For
4987 example, if a cross-compiler was configured with @samp{configure
4988 i386v}, meaning to compile for an 80386 running System V, then you
4989 would specify @option{-b i386v} to run that cross compiler.
4991 When you do not specify @option{-b}, it normally means to compile for
4992 the same type of machine that you are using.
4994 @item -V @var{version}
4996 The argument @var{version} specifies which version of GCC to run.
4997 This is useful when multiple versions are installed. For example,
4998 @var{version} might be @samp{2.0}, meaning to run GCC version 2.0.
5000 The default version, when you do not specify @option{-V}, is the last
5001 version of GCC that you installed.
5004 The @option{-b} and @option{-V} options actually work by controlling part of
5005 the file name used for the executable files and libraries used for
5006 compilation. A given version of GCC, for a given target machine, is
5007 normally kept in the directory @file{/usr/local/lib/gcc-lib/@var{machine}/@var{version}}.
5009 Thus, sites can customize the effect of @option{-b} or @option{-V} either by
5010 changing the names of these directories or adding alternate names (or
5011 symbolic links). If in directory @file{/usr/local/lib/gcc-lib/} the
5012 file @file{80386} is a link to the file @file{i386v}, then @option{-b
5013 80386} becomes an alias for @option{-b i386v}.
5015 In one respect, the @option{-b} or @option{-V} do not completely change
5016 to a different compiler: the top-level driver program @command{gcc}
5017 that you originally invoked continues to run and invoke the other
5018 executables (preprocessor, compiler per se, assembler and linker)
5019 that do the real work. However, since no real work is done in the
5020 driver program, it usually does not matter that the driver program
5021 in use is not the one for the specified target. It is common for the
5022 interface to the other executables to change incompatibly between
5023 compiler versions, so unless the version specified is very close to that
5024 of the driver (for example, @option{-V 3.0} with a driver program from GCC
5025 version 3.0.1), use of @option{-V} may not work; for example, using
5026 @option{-V 2.95.2} will not work with a driver program from GCC 3.0.
5028 The only way that the driver program depends on the target machine is
5029 in the parsing and handling of special machine-specific options.
5030 However, this is controlled by a file which is found, along with the
5031 other executables, in the directory for the specified version and
5032 target machine. As a result, a single installed driver program adapts
5033 to any specified target machine, and sufficiently similar compiler
5036 The driver program executable does control one significant thing,
5037 however: the default version and target machine. Therefore, you can
5038 install different instances of the driver program, compiled for
5039 different targets or versions, under different names.
5041 For example, if the driver for version 2.0 is installed as @command{ogcc}
5042 and that for version 2.1 is installed as @command{gcc}, then the command
5043 @command{gcc} will use version 2.1 by default, while @command{ogcc} will use
5044 2.0 by default. However, you can choose either version with either
5045 command with the @option{-V} option.
5047 @node Submodel Options
5048 @section Hardware Models and Configurations
5049 @cindex submodel options
5050 @cindex specifying hardware config
5051 @cindex hardware models and configurations, specifying
5052 @cindex machine dependent options
5054 Earlier we discussed the standard option @option{-b} which chooses among
5055 different installed compilers for completely different target
5056 machines, such as VAX vs.@: 68000 vs.@: 80386.
5058 In addition, each of these target machine types can have its own
5059 special options, starting with @samp{-m}, to choose among various
5060 hardware models or configurations---for example, 68010 vs 68020,
5061 floating coprocessor or none. A single installed version of the
5062 compiler can compile for any model or configuration, according to the
5065 Some configurations of the compiler also support additional special
5066 options, usually for compatibility with other compilers on the same
5070 These options are defined by the macro @code{TARGET_SWITCHES} in the
5071 machine description. The default for the options is also defined by
5072 that macro, which enables you to change the defaults.
5087 * RS/6000 and PowerPC Options::
5092 * Intel 960 Options::
5093 * DEC Alpha Options::
5097 * System V Options::
5098 * TMS320C3x/C4x Options::
5106 * S/390 and zSeries Options::
5109 @node M680x0 Options
5110 @subsection M680x0 Options
5111 @cindex M680x0 options
5113 These are the @samp{-m} options defined for the 68000 series. The default
5114 values for these options depends on which style of 68000 was selected when
5115 the compiler was configured; the defaults for the most common choices are
5123 Generate output for a 68000. This is the default
5124 when the compiler is configured for 68000-based systems.
5126 Use this option for microcontrollers with a 68000 or EC000 core,
5127 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
5133 Generate output for a 68020. This is the default
5134 when the compiler is configured for 68020-based systems.
5138 Generate output containing 68881 instructions for floating point.
5139 This is the default for most 68020 systems unless @option{--nfp} was
5140 specified when the compiler was configured.
5144 Generate output for a 68030. This is the default when the compiler is
5145 configured for 68030-based systems.
5149 Generate output for a 68040. This is the default when the compiler is
5150 configured for 68040-based systems.
5152 This option inhibits the use of 68881/68882 instructions that have to be
5153 emulated by software on the 68040. Use this option if your 68040 does not
5154 have code to emulate those instructions.
5158 Generate output for a 68060. This is the default when the compiler is
5159 configured for 68060-based systems.
5161 This option inhibits the use of 68020 and 68881/68882 instructions that
5162 have to be emulated by software on the 68060. Use this option if your 68060
5163 does not have code to emulate those instructions.
5167 Generate output for a CPU32. This is the default
5168 when the compiler is configured for CPU32-based systems.
5170 Use this option for microcontrollers with a
5171 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
5172 68336, 68340, 68341, 68349 and 68360.
5176 Generate output for a 520X ``coldfire'' family cpu. This is the default
5177 when the compiler is configured for 520X-based systems.
5179 Use this option for microcontroller with a 5200 core, including
5180 the MCF5202, MCF5203, MCF5204 and MCF5202.
5185 Generate output for a 68040, without using any of the new instructions.
5186 This results in code which can run relatively efficiently on either a
5187 68020/68881 or a 68030 or a 68040. The generated code does use the
5188 68881 instructions that are emulated on the 68040.
5192 Generate output for a 68060, without using any of the new instructions.
5193 This results in code which can run relatively efficiently on either a
5194 68020/68881 or a 68030 or a 68040. The generated code does use the
5195 68881 instructions that are emulated on the 68060.
5199 Generate output containing Sun FPA instructions for floating point.
5202 @opindex msoft-float
5203 Generate output containing library calls for floating point.
5204 @strong{Warning:} the requisite libraries are not available for all m68k
5205 targets. Normally the facilities of the machine's usual C compiler are
5206 used, but this can't be done directly in cross-compilation. You must
5207 make your own arrangements to provide suitable library functions for
5208 cross-compilation. The embedded targets @samp{m68k-*-aout} and
5209 @samp{m68k-*-coff} do provide software floating point support.
5213 Consider type @code{int} to be 16 bits wide, like @code{short int}.
5216 @opindex mnobitfield
5217 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
5218 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
5222 Do use the bit-field instructions. The @option{-m68020} option implies
5223 @option{-mbitfield}. This is the default if you use a configuration
5224 designed for a 68020.
5228 Use a different function-calling convention, in which functions
5229 that take a fixed number of arguments return with the @code{rtd}
5230 instruction, which pops their arguments while returning. This
5231 saves one instruction in the caller since there is no need to pop
5232 the arguments there.
5234 This calling convention is incompatible with the one normally
5235 used on Unix, so you cannot use it if you need to call libraries
5236 compiled with the Unix compiler.
5238 Also, you must provide function prototypes for all functions that
5239 take variable numbers of arguments (including @code{printf});
5240 otherwise incorrect code will be generated for calls to those
5243 In addition, seriously incorrect code will result if you call a
5244 function with too many arguments. (Normally, extra arguments are
5245 harmlessly ignored.)
5247 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
5248 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
5251 @itemx -mno-align-int
5253 @opindex mno-align-int
5254 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
5255 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
5256 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
5257 Aligning variables on 32-bit boundaries produces code that runs somewhat
5258 faster on processors with 32-bit busses at the expense of more memory.
5260 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
5261 align structures containing the above types differently than
5262 most published application binary interface specifications for the m68k.
5266 Use the pc-relative addressing mode of the 68000 directly, instead of
5267 using a global offset table. At present, this option implies @option{-fpic},
5268 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
5269 not presently supported with @option{-mpcrel}, though this could be supported for
5270 68020 and higher processors.
5272 @item -mno-strict-align
5273 @itemx -mstrict-align
5274 @opindex mno-strict-align
5275 @opindex mstrict-align
5276 Do not (do) assume that unaligned memory references will be handled by
5281 @node M68hc1x Options
5282 @subsection M68hc1x Options
5283 @cindex M68hc1x options
5285 These are the @samp{-m} options defined for the 68hc11 and 68hc12
5286 microcontrollers. The default values for these options depends on
5287 which style of microcontroller was selected when the compiler was configured;
5288 the defaults for the most common choices are given below.
5295 Generate output for a 68HC11. This is the default
5296 when the compiler is configured for 68HC11-based systems.
5302 Generate output for a 68HC12. This is the default
5303 when the compiler is configured for 68HC12-based systems.
5306 @opindex mauto-incdec
5307 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
5312 Consider type @code{int} to be 16 bits wide, like @code{short int}.
5314 @item -msoft-reg-count=@var{count}
5315 @opindex msoft-reg-count
5316 Specify the number of pseudo-soft registers which are used for the
5317 code generation. The maximum number is 32. Using more pseudo-soft
5318 register may or may not result in better code depending on the program.
5319 The default is 4 for 68HC11 and 2 for 68HC12.
5324 @subsection VAX Options
5327 These @samp{-m} options are defined for the VAX:
5332 Do not output certain jump instructions (@code{aobleq} and so on)
5333 that the Unix assembler for the VAX cannot handle across long
5338 Do output those jump instructions, on the assumption that you
5339 will assemble with the GNU assembler.
5343 Output code for g-format floating point numbers instead of d-format.
5347 @subsection SPARC Options
5348 @cindex SPARC options
5350 These @samp{-m} switches are supported on the SPARC:
5355 @opindex mno-app-regs
5357 Specify @option{-mapp-regs} to generate output using the global registers
5358 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
5361 To be fully SVR4 ABI compliant at the cost of some performance loss,
5362 specify @option{-mno-app-regs}. You should compile libraries and system
5363 software with this option.
5368 @opindex mhard-float
5369 Generate output containing floating point instructions. This is the
5375 @opindex msoft-float
5376 Generate output containing library calls for floating point.
5377 @strong{Warning:} the requisite libraries are not available for all SPARC
5378 targets. Normally the facilities of the machine's usual C compiler are
5379 used, but this cannot be done directly in cross-compilation. You must make
5380 your own arrangements to provide suitable library functions for
5381 cross-compilation. The embedded targets @samp{sparc-*-aout} and
5382 @samp{sparclite-*-*} do provide software floating point support.
5384 @option{-msoft-float} changes the calling convention in the output file;
5385 therefore, it is only useful if you compile @emph{all} of a program with
5386 this option. In particular, you need to compile @file{libgcc.a}, the
5387 library that comes with GCC, with @option{-msoft-float} in order for
5390 @item -mhard-quad-float
5391 @opindex mhard-quad-float
5392 Generate output containing quad-word (long double) floating point
5395 @item -msoft-quad-float
5396 @opindex msoft-quad-float
5397 Generate output containing library calls for quad-word (long double)
5398 floating point instructions. The functions called are those specified
5399 in the SPARC ABI@. This is the default.
5401 As of this writing, there are no sparc implementations that have hardware
5402 support for the quad-word floating point instructions. They all invoke
5403 a trap handler for one of these instructions, and then the trap handler
5404 emulates the effect of the instruction. Because of the trap handler overhead,
5405 this is much slower than calling the ABI library routines. Thus the
5406 @option{-msoft-quad-float} option is the default.
5410 @opindex mno-epilogue
5412 With @option{-mepilogue} (the default), the compiler always emits code for
5413 function exit at the end of each function. Any function exit in
5414 the middle of the function (such as a return statement in C) will
5415 generate a jump to the exit code at the end of the function.
5417 With @option{-mno-epilogue}, the compiler tries to emit exit code inline
5418 at every function exit.
5424 With @option{-mflat}, the compiler does not generate save/restore instructions
5425 and will use a ``flat'' or single register window calling convention.
5426 This model uses %i7 as the frame pointer and is compatible with the normal
5427 register window model. Code from either may be intermixed.
5428 The local registers and the input registers (0--5) are still treated as
5429 ``call saved'' registers and will be saved on the stack as necessary.
5431 With @option{-mno-flat} (the default), the compiler emits save/restore
5432 instructions (except for leaf functions) and is the normal mode of operation.
5434 @item -mno-unaligned-doubles
5435 @itemx -munaligned-doubles
5436 @opindex mno-unaligned-doubles
5437 @opindex munaligned-doubles
5438 Assume that doubles have 8 byte alignment. This is the default.
5440 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
5441 alignment only if they are contained in another type, or if they have an
5442 absolute address. Otherwise, it assumes they have 4 byte alignment.
5443 Specifying this option avoids some rare compatibility problems with code
5444 generated by other compilers. It is not the default because it results
5445 in a performance loss, especially for floating point code.
5447 @item -mno-faster-structs
5448 @itemx -mfaster-structs
5449 @opindex mno-faster-structs
5450 @opindex mfaster-structs
5451 With @option{-mfaster-structs}, the compiler assumes that structures
5452 should have 8 byte alignment. This enables the use of pairs of
5453 @code{ldd} and @code{std} instructions for copies in structure
5454 assignment, in place of twice as many @code{ld} and @code{st} pairs.
5455 However, the use of this changed alignment directly violates the Sparc
5456 ABI@. Thus, it's intended only for use on targets where the developer
5457 acknowledges that their resulting code will not be directly in line with
5458 the rules of the ABI@.
5464 These two options select variations on the SPARC architecture.
5466 By default (unless specifically configured for the Fujitsu SPARClite),
5467 GCC generates code for the v7 variant of the SPARC architecture.
5469 @option{-mv8} will give you SPARC v8 code. The only difference from v7
5470 code is that the compiler emits the integer multiply and integer
5471 divide instructions which exist in SPARC v8 but not in SPARC v7.
5473 @option{-msparclite} will give you SPARClite code. This adds the integer
5474 multiply, integer divide step and scan (@code{ffs}) instructions which
5475 exist in SPARClite but not in SPARC v7.
5477 These options are deprecated and will be deleted in a future GCC release.
5478 They have been replaced with @option{-mcpu=xxx}.
5483 @opindex msupersparc
5484 These two options select the processor for which the code is optimised.
5486 With @option{-mcypress} (the default), the compiler optimizes code for the
5487 Cypress CY7C602 chip, as used in the SparcStation/SparcServer 3xx series.
5488 This is also appropriate for the older SparcStation 1, 2, IPX etc.
5490 With @option{-msupersparc} the compiler optimizes code for the SuperSparc cpu, as
5491 used in the SparcStation 10, 1000 and 2000 series. This flag also enables use
5492 of the full SPARC v8 instruction set.
5494 These options are deprecated and will be deleted in a future GCC release.
5495 They have been replaced with @option{-mcpu=xxx}.
5497 @item -mcpu=@var{cpu_type}
5499 Set the instruction set, register set, and instruction scheduling parameters
5500 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
5501 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
5502 @samp{hypersparc}, @samp{sparclite86x}, @samp{f930}, @samp{f934},
5503 @samp{sparclet}, @samp{tsc701}, @samp{v9}, and @samp{ultrasparc}.
5505 Default instruction scheduling parameters are used for values that select
5506 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
5507 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
5509 Here is a list of each supported architecture and their supported
5514 v8: supersparc, hypersparc
5515 sparclite: f930, f934, sparclite86x
5520 @item -mtune=@var{cpu_type}
5522 Set the instruction scheduling parameters for machine type
5523 @var{cpu_type}, but do not set the instruction set or register set that the
5524 option @option{-mcpu=@var{cpu_type}} would.
5526 The same values for @option{-mcpu=@var{cpu_type}} can be used for
5527 @option{-mtune=@var{cpu_type}}, but the only useful values are those
5528 that select a particular cpu implementation. Those are @samp{cypress},
5529 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
5530 @samp{sparclite86x}, @samp{tsc701}, and @samp{ultrasparc}.
5534 These @samp{-m} switches are supported in addition to the above
5535 on the SPARCLET processor.
5538 @item -mlittle-endian
5539 @opindex mlittle-endian
5540 Generate code for a processor running in little-endian mode.
5544 Treat register @code{%g0} as a normal register.
5545 GCC will continue to clobber it as necessary but will not assume
5546 it always reads as 0.
5548 @item -mbroken-saverestore
5549 @opindex mbroken-saverestore
5550 Generate code that does not use non-trivial forms of the @code{save} and
5551 @code{restore} instructions. Early versions of the SPARCLET processor do
5552 not correctly handle @code{save} and @code{restore} instructions used with
5553 arguments. They correctly handle them used without arguments. A @code{save}
5554 instruction used without arguments increments the current window pointer
5555 but does not allocate a new stack frame. It is assumed that the window
5556 overflow trap handler will properly handle this case as will interrupt
5560 These @samp{-m} switches are supported in addition to the above
5561 on SPARC V9 processors in 64-bit environments.
5564 @item -mlittle-endian
5565 @opindex mlittle-endian
5566 Generate code for a processor running in little-endian mode.
5572 Generate code for a 32-bit or 64-bit environment.
5573 The 32-bit environment sets int, long and pointer to 32 bits.
5574 The 64-bit environment sets int to 32 bits and long and pointer
5577 @item -mcmodel=medlow
5578 @opindex mcmodel=medlow
5579 Generate code for the Medium/Low code model: the program must be linked
5580 in the low 32 bits of the address space. Pointers are 64 bits.
5581 Programs can be statically or dynamically linked.
5583 @item -mcmodel=medmid
5584 @opindex mcmodel=medmid
5585 Generate code for the Medium/Middle code model: the program must be linked
5586 in the low 44 bits of the address space, the text segment must be less than
5587 2G bytes, and data segment must be within 2G of the text segment.
5588 Pointers are 64 bits.
5590 @item -mcmodel=medany
5591 @opindex mcmodel=medany
5592 Generate code for the Medium/Anywhere code model: the program may be linked
5593 anywhere in the address space, the text segment must be less than
5594 2G bytes, and data segment must be within 2G of the text segment.
5595 Pointers are 64 bits.
5597 @item -mcmodel=embmedany
5598 @opindex mcmodel=embmedany
5599 Generate code for the Medium/Anywhere code model for embedded systems:
5600 assume a 32-bit text and a 32-bit data segment, both starting anywhere
5601 (determined at link time). Register %g4 points to the base of the
5602 data segment. Pointers are still 64 bits.
5603 Programs are statically linked, PIC is not supported.
5606 @itemx -mno-stack-bias
5607 @opindex mstack-bias
5608 @opindex mno-stack-bias
5609 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
5610 frame pointer if present, are offset by @minus{}2047 which must be added back
5611 when making stack frame references.
5612 Otherwise, assume no such offset is present.
5615 @node Convex Options
5616 @subsection Convex Options
5617 @cindex Convex options
5619 These @samp{-m} options are defined for Convex:
5624 Generate output for C1. The code will run on any Convex machine.
5625 The preprocessor symbol @code{__convex__c1__} is defined.
5629 Generate output for C2. Uses instructions not available on C1.
5630 Scheduling and other optimizations are chosen for max performance on C2.
5631 The preprocessor symbol @code{__convex_c2__} is defined.
5635 Generate output for C32xx. Uses instructions not available on C1.
5636 Scheduling and other optimizations are chosen for max performance on C32.
5637 The preprocessor symbol @code{__convex_c32__} is defined.
5641 Generate output for C34xx. Uses instructions not available on C1.
5642 Scheduling and other optimizations are chosen for max performance on C34.
5643 The preprocessor symbol @code{__convex_c34__} is defined.
5647 Generate output for C38xx. Uses instructions not available on C1.
5648 Scheduling and other optimizations are chosen for max performance on C38.
5649 The preprocessor symbol @code{__convex_c38__} is defined.
5653 Generate code which puts an argument count in the word preceding each
5654 argument list. This is compatible with regular CC, and a few programs
5655 may need the argument count word. GDB and other source-level debuggers
5656 do not need it; this info is in the symbol table.
5659 @opindex mnoargcount
5660 Omit the argument count word. This is the default.
5662 @item -mvolatile-cache
5663 @opindex mvolatile-cache
5664 Allow volatile references to be cached. This is the default.
5666 @item -mvolatile-nocache
5667 @opindex mvolatile-nocache
5668 Volatile references bypass the data cache, going all the way to memory.
5669 This is only needed for multi-processor code that does not use standard
5670 synchronization instructions. Making non-volatile references to volatile
5671 locations will not necessarily work.
5675 Type long is 32 bits, the same as type int. This is the default.
5679 Type long is 64 bits, the same as type long long. This option is useless,
5680 because no library support exists for it.
5683 @node AMD29K Options
5684 @subsection AMD29K Options
5685 @cindex AMD29K options
5687 These @samp{-m} options are defined for the AMD Am29000:
5692 @cindex DW bit (29k)
5693 Generate code that assumes the @code{DW} bit is set, i.e., that byte and
5694 halfword operations are directly supported by the hardware. This is the
5699 Generate code that assumes the @code{DW} bit is not set.
5703 @cindex byte writes (29k)
5704 Generate code that assumes the system supports byte and halfword write
5705 operations. This is the default.
5709 Generate code that assumes the systems does not support byte and
5710 halfword write operations. @option{-mnbw} implies @option{-mndw}.
5714 @cindex memory model (29k)
5715 Use a small memory model that assumes that all function addresses are
5716 either within a single 256 KB segment or at an absolute address of less
5717 than 256k. This allows the @code{call} instruction to be used instead
5718 of a @code{const}, @code{consth}, @code{calli} sequence.
5722 Use the normal memory model: Generate @code{call} instructions only when
5723 calling functions in the same file and @code{calli} instructions
5724 otherwise. This works if each file occupies less than 256 KB but allows
5725 the entire executable to be larger than 256 KB@. This is the default.
5729 Always use @code{calli} instructions. Specify this option if you expect
5730 a single file to compile into more than 256 KB of code.
5734 @cindex processor selection (29k)
5735 Generate code for the Am29050.
5739 Generate code for the Am29000. This is the default.
5741 @item -mkernel-registers
5742 @opindex mkernel-registers
5743 @cindex kernel and user registers (29k)
5744 Generate references to registers @code{gr64-gr95} instead of to
5745 registers @code{gr96-gr127}. This option can be used when compiling
5746 kernel code that wants a set of global registers disjoint from that used
5749 Note that when this option is used, register names in @samp{-f} flags
5750 must use the normal, user-mode, names.
5752 @item -muser-registers
5753 @opindex muser-registers
5754 Use the normal set of global registers, @code{gr96-gr127}. This is the
5758 @itemx -mno-stack-check
5759 @opindex mstack-check
5760 @opindex mno-stack-check
5761 @cindex stack checks (29k)
5762 Insert (or do not insert) a call to @code{__msp_check} after each stack
5763 adjustment. This is often used for kernel code.
5766 @itemx -mno-storem-bug
5767 @opindex mstorem-bug
5768 @opindex mno-storem-bug
5769 @cindex storem bug (29k)
5770 @option{-mstorem-bug} handles 29k processors which cannot handle the
5771 separation of a mtsrim insn and a storem instruction (most 29000 chips
5772 to date, but not the 29050).
5774 @item -mno-reuse-arg-regs
5775 @itemx -mreuse-arg-regs
5776 @opindex mno-reuse-arg-regs
5777 @opindex mreuse-arg-regs
5778 @option{-mno-reuse-arg-regs} tells the compiler to only use incoming argument
5779 registers for copying out arguments. This helps detect calling a function
5780 with fewer arguments than it was declared with.
5782 @item -mno-impure-text
5783 @itemx -mimpure-text
5784 @opindex mno-impure-text
5785 @opindex mimpure-text
5786 @option{-mimpure-text}, used in addition to @option{-shared}, tells the compiler to
5787 not pass @option{-assert pure-text} to the linker when linking a shared object.
5790 @opindex msoft-float
5791 Generate output containing library calls for floating point.
5792 @strong{Warning:} the requisite libraries are not part of GCC@.
5793 Normally the facilities of the machine's usual C compiler are used, but
5794 this can't be done directly in cross-compilation. You must make your
5795 own arrangements to provide suitable library functions for
5800 Do not generate multm or multmu instructions. This is useful for some embedded
5801 systems which do not have trap handlers for these instructions.
5805 @subsection ARM Options
5808 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
5813 @opindex mapcs-frame
5814 Generate a stack frame that is compliant with the ARM Procedure Call
5815 Standard for all functions, even if this is not strictly necessary for
5816 correct execution of the code. Specifying @option{-fomit-frame-pointer}
5817 with this option will cause the stack frames not to be generated for
5818 leaf functions. The default is @option{-mno-apcs-frame}.
5822 This is a synonym for @option{-mapcs-frame}.
5826 Generate code for a processor running with a 26-bit program counter,
5827 and conforming to the function calling standards for the APCS 26-bit
5828 option. This option replaces the @option{-m2} and @option{-m3} options
5829 of previous releases of the compiler.
5833 Generate code for a processor running with a 32-bit program counter,
5834 and conforming to the function calling standards for the APCS 32-bit
5835 option. This option replaces the @option{-m6} option of previous releases
5839 @c not currently implemented
5840 @item -mapcs-stack-check
5841 @opindex mapcs-stack-check
5842 Generate code to check the amount of stack space available upon entry to
5843 every function (that actually uses some stack space). If there is
5844 insufficient space available then either the function
5845 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
5846 called, depending upon the amount of stack space required. The run time
5847 system is required to provide these functions. The default is
5848 @option{-mno-apcs-stack-check}, since this produces smaller code.
5850 @c not currently implemented
5852 @opindex mapcs-float
5853 Pass floating point arguments using the float point registers. This is
5854 one of the variants of the APCS@. This option is recommended if the
5855 target hardware has a floating point unit or if a lot of floating point
5856 arithmetic is going to be performed by the code. The default is
5857 @option{-mno-apcs-float}, since integer only code is slightly increased in
5858 size if @option{-mapcs-float} is used.
5860 @c not currently implemented
5861 @item -mapcs-reentrant
5862 @opindex mapcs-reentrant
5863 Generate reentrant, position independent code. The default is
5864 @option{-mno-apcs-reentrant}.
5867 @item -mthumb-interwork
5868 @opindex mthumb-interwork
5869 Generate code which supports calling between the ARM and Thumb
5870 instruction sets. Without this option the two instruction sets cannot
5871 be reliably used inside one program. The default is
5872 @option{-mno-thumb-interwork}, since slightly larger code is generated
5873 when @option{-mthumb-interwork} is specified.
5875 @item -mno-sched-prolog
5876 @opindex mno-sched-prolog
5877 Prevent the reordering of instructions in the function prolog, or the
5878 merging of those instruction with the instructions in the function's
5879 body. This means that all functions will start with a recognizable set
5880 of instructions (or in fact one of a choice from a small set of
5881 different function prologues), and this information can be used to
5882 locate the start if functions inside an executable piece of code. The
5883 default is @option{-msched-prolog}.
5886 @opindex mhard-float
5887 Generate output containing floating point instructions. This is the
5891 @opindex msoft-float
5892 Generate output containing library calls for floating point.
5893 @strong{Warning:} the requisite libraries are not available for all ARM
5894 targets. Normally the facilities of the machine's usual C compiler are
5895 used, but this cannot be done directly in cross-compilation. You must make
5896 your own arrangements to provide suitable library functions for
5899 @option{-msoft-float} changes the calling convention in the output file;
5900 therefore, it is only useful if you compile @emph{all} of a program with
5901 this option. In particular, you need to compile @file{libgcc.a}, the
5902 library that comes with GCC, with @option{-msoft-float} in order for
5905 @item -mlittle-endian
5906 @opindex mlittle-endian
5907 Generate code for a processor running in little-endian mode. This is
5908 the default for all standard configurations.
5911 @opindex mbig-endian
5912 Generate code for a processor running in big-endian mode; the default is
5913 to compile code for a little-endian processor.
5915 @item -mwords-little-endian
5916 @opindex mwords-little-endian
5917 This option only applies when generating code for big-endian processors.
5918 Generate code for a little-endian word order but a big-endian byte
5919 order. That is, a byte order of the form @samp{32107654}. Note: this
5920 option should only be used if you require compatibility with code for
5921 big-endian ARM processors generated by versions of the compiler prior to
5924 @item -malignment-traps
5925 @opindex malignment-traps
5926 Generate code that will not trap if the MMU has alignment traps enabled.
5927 On ARM architectures prior to ARMv4, there were no instructions to
5928 access half-word objects stored in memory. However, when reading from
5929 memory a feature of the ARM architecture allows a word load to be used,
5930 even if the address is unaligned, and the processor core will rotate the
5931 data as it is being loaded. This option tells the compiler that such
5932 misaligned accesses will cause a MMU trap and that it should instead
5933 synthesise the access as a series of byte accesses. The compiler can
5934 still use word accesses to load half-word data if it knows that the
5935 address is aligned to a word boundary.
5937 This option is ignored when compiling for ARM architecture 4 or later,
5938 since these processors have instructions to directly access half-word
5941 @item -mno-alignment-traps
5942 @opindex mno-alignment-traps
5943 Generate code that assumes that the MMU will not trap unaligned
5944 accesses. This produces better code when the target instruction set
5945 does not have half-word memory operations (i.e.@: implementations prior to
5948 Note that you cannot use this option to access unaligned word objects,
5949 since the processor will only fetch one 32-bit aligned object from
5952 The default setting for most targets is @option{-mno-alignment-traps}, since
5953 this produces better code when there are no half-word memory
5954 instructions available.
5956 @item -mshort-load-bytes
5957 @itemx -mno-short-load-words
5958 @opindex mshort-load-bytes
5959 @opindex mno-short-load-words
5960 These are deprecated aliases for @option{-malignment-traps}.
5962 @item -mno-short-load-bytes
5963 @itemx -mshort-load-words
5964 @opindex mno-short-load-bytes
5965 @opindex mshort-load-words
5966 This are deprecated aliases for @option{-mno-alignment-traps}.
5970 This option only applies to RISC iX@. Emulate the native BSD-mode
5971 compiler. This is the default if @option{-ansi} is not specified.
5975 This option only applies to RISC iX@. Emulate the native X/Open-mode
5978 @item -mno-symrename
5979 @opindex mno-symrename
5980 This option only applies to RISC iX@. Do not run the assembler
5981 post-processor, @samp{symrename}, after code has been assembled.
5982 Normally it is necessary to modify some of the standard symbols in
5983 preparation for linking with the RISC iX C library; this option
5984 suppresses this pass. The post-processor is never run when the
5985 compiler is built for cross-compilation.
5987 @item -mcpu=@var{name}
5989 This specifies the name of the target ARM processor. GCC uses this name
5990 to determine what kind of instructions it can emit when generating
5991 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
5992 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
5993 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
5994 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
5995 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
5996 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm8},
5997 @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
5998 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
5999 @samp{arm920t}, @samp{arm940t}, @samp{arm9tdmi}, @samp{arm10tdmi},
6000 @samp{arm1020t}, @samp{xscale}.
6002 @itemx -mtune=@var{name}
6004 This option is very similar to the @option{-mcpu=} option, except that
6005 instead of specifying the actual target processor type, and hence
6006 restricting which instructions can be used, it specifies that GCC should
6007 tune the performance of the code as if the target were of the type
6008 specified in this option, but still choosing the instructions that it
6009 will generate based on the cpu specified by a @option{-mcpu=} option.
6010 For some ARM implementations better performance can be obtained by using
6013 @item -march=@var{name}
6015 This specifies the name of the target ARM architecture. GCC uses this
6016 name to determine what kind of instructions it can emit when generating
6017 assembly code. This option can be used in conjunction with or instead
6018 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
6019 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
6020 @samp{armv5}, @samp{armv5t}, @samp{armv5te}.
6022 @item -mfpe=@var{number}
6023 @itemx -mfp=@var{number}
6026 This specifies the version of the floating point emulation available on
6027 the target. Permissible values are 2 and 3. @option{-mfp=} is a synonym
6028 for @option{-mfpe=}, for compatibility with older versions of GCC@.
6030 @item -mstructure-size-boundary=@var{n}
6031 @opindex mstructure-size-boundary
6032 The size of all structures and unions will be rounded up to a multiple
6033 of the number of bits set by this option. Permissible values are 8 and
6034 32. The default value varies for different toolchains. For the COFF
6035 targeted toolchain the default value is 8. Specifying the larger number
6036 can produce faster, more efficient code, but can also increase the size
6037 of the program. The two values are potentially incompatible. Code
6038 compiled with one value cannot necessarily expect to work with code or
6039 libraries compiled with the other value, if they exchange information
6040 using structures or unions.
6042 @item -mabort-on-noreturn
6043 @opindex mabort-on-noreturn
6044 Generate a call to the function @code{abort} at the end of a
6045 @code{noreturn} function. It will be executed if the function tries to
6049 @itemx -mno-long-calls
6050 @opindex mlong-calls
6051 @opindex mno-long-calls
6052 Tells the compiler to perform function calls by first loading the
6053 address of the function into a register and then performing a subroutine
6054 call on this register. This switch is needed if the target function
6055 will lie outside of the 64 megabyte addressing range of the offset based
6056 version of subroutine call instruction.
6058 Even if this switch is enabled, not all function calls will be turned
6059 into long calls. The heuristic is that static functions, functions
6060 which have the @samp{short-call} attribute, functions that are inside
6061 the scope of a @samp{#pragma no_long_calls} directive and functions whose
6062 definitions have already been compiled within the current compilation
6063 unit, will not be turned into long calls. The exception to this rule is
6064 that weak function definitions, functions with the @samp{long-call}
6065 attribute or the @samp{section} attribute, and functions that are within
6066 the scope of a @samp{#pragma long_calls} directive, will always be
6067 turned into long calls.
6069 This feature is not enabled by default. Specifying
6070 @option{-mno-long-calls} will restore the default behaviour, as will
6071 placing the function calls within the scope of a @samp{#pragma
6072 long_calls_off} directive. Note these switches have no effect on how
6073 the compiler generates code to handle function calls via function
6076 @item -mnop-fun-dllimport
6077 @opindex mnop-fun-dllimport
6078 Disable support for the @code{dllimport} attribute.
6080 @item -msingle-pic-base
6081 @opindex msingle-pic-base
6082 Treat the register used for PIC addressing as read-only, rather than
6083 loading it in the prologue for each function. The run-time system is
6084 responsible for initialising this register with an appropriate value
6085 before execution begins.
6087 @item -mpic-register=@var{reg}
6088 @opindex mpic-register
6089 Specify the register to be used for PIC addressing. The default is R10
6090 unless stack-checking is enabled, when R9 is used.
6092 @item -mpoke-function-name
6093 @opindex mpoke-function-name
6094 Write the name of each function into the text section, directly
6095 preceding the function prologue. The generated code is similar to this:
6099 .ascii "arm_poke_function_name", 0
6102 .word 0xff000000 + (t1 - t0)
6103 arm_poke_function_name
6105 stmfd sp!, @{fp, ip, lr, pc@}
6109 When performing a stack backtrace, code can inspect the value of
6110 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
6111 location @code{pc - 12} and the top 8 bits are set, then we know that
6112 there is a function name embedded immediately preceding this location
6113 and has length @code{((pc[-3]) & 0xff000000)}.
6117 Generate code for the 16-bit Thumb instruction set. The default is to
6118 use the 32-bit ARM instruction set.
6121 @opindex mtpcs-frame
6122 Generate a stack frame that is compliant with the Thumb Procedure Call
6123 Standard for all non-leaf functions. (A leaf function is one that does
6124 not call any other functions.) The default is @option{-mno-tpcs-frame}.
6126 @item -mtpcs-leaf-frame
6127 @opindex mtpcs-leaf-frame
6128 Generate a stack frame that is compliant with the Thumb Procedure Call
6129 Standard for all leaf functions. (A leaf function is one that does
6130 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
6132 @item -mcallee-super-interworking
6133 @opindex mcallee-super-interworking
6134 Gives all externally visible functions in the file being compiled an ARM
6135 instruction set header which switches to Thumb mode before executing the
6136 rest of the function. This allows these functions to be called from
6137 non-interworking code.
6139 @item -mcaller-super-interworking
6140 @opindex mcaller-super-interworking
6141 Allows calls via function pointers (including virtual functions) to
6142 execute correctly regardless of whether the target code has been
6143 compiled for interworking or not. There is a small overhead in the cost
6144 of executing a function pointer if this option is enabled.
6148 @node MN10200 Options
6149 @subsection MN10200 Options
6150 @cindex MN10200 options
6151 These @option{-m} options are defined for Matsushita MN10200 architectures:
6156 Indicate to the linker that it should perform a relaxation optimization pass
6157 to shorten branches, calls and absolute memory addresses. This option only
6158 has an effect when used on the command line for the final link step.
6160 This option makes symbolic debugging impossible.
6163 @node MN10300 Options
6164 @subsection MN10300 Options
6165 @cindex MN10300 options
6166 These @option{-m} options are defined for Matsushita MN10300 architectures:
6171 Generate code to avoid bugs in the multiply instructions for the MN10300
6172 processors. This is the default.
6175 @opindex mno-mult-bug
6176 Do not generate code to avoid bugs in the multiply instructions for the
6181 Generate code which uses features specific to the AM33 processor.
6185 Do not generate code which uses features specific to the AM33 processor. This
6190 Do not link in the C run-time initialization object file.
6194 Indicate to the linker that it should perform a relaxation optimization pass
6195 to shorten branches, calls and absolute memory addresses. This option only
6196 has an effect when used on the command line for the final link step.
6198 This option makes symbolic debugging impossible.
6202 @node M32R/D Options
6203 @subsection M32R/D Options
6204 @cindex M32R/D options
6206 These @option{-m} options are defined for Mitsubishi M32R/D architectures:
6209 @item -mcode-model=small
6210 @opindex mcode-model=small
6211 Assume all objects live in the lower 16MB of memory (so that their addresses
6212 can be loaded with the @code{ld24} instruction), and assume all subroutines
6213 are reachable with the @code{bl} instruction.
6214 This is the default.
6216 The addressability of a particular object can be set with the
6217 @code{model} attribute.
6219 @item -mcode-model=medium
6220 @opindex mcode-model=medium
6221 Assume objects may be anywhere in the 32-bit address space (the compiler
6222 will generate @code{seth/add3} instructions to load their addresses), and
6223 assume all subroutines are reachable with the @code{bl} instruction.
6225 @item -mcode-model=large
6226 @opindex mcode-model=large
6227 Assume objects may be anywhere in the 32-bit address space (the compiler
6228 will generate @code{seth/add3} instructions to load their addresses), and
6229 assume subroutines may not be reachable with the @code{bl} instruction
6230 (the compiler will generate the much slower @code{seth/add3/jl}
6231 instruction sequence).
6234 @opindex msdata=none
6235 Disable use of the small data area. Variables will be put into
6236 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
6237 @code{section} attribute has been specified).
6238 This is the default.
6240 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
6241 Objects may be explicitly put in the small data area with the
6242 @code{section} attribute using one of these sections.
6245 @opindex msdata=sdata
6246 Put small global and static data in the small data area, but do not
6247 generate special code to reference them.
6251 Put small global and static data in the small data area, and generate
6252 special instructions to reference them.
6256 @cindex smaller data references
6257 Put global and static objects less than or equal to @var{num} bytes
6258 into the small data or bss sections instead of the normal data or bss
6259 sections. The default value of @var{num} is 8.
6260 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
6261 for this option to have any effect.
6263 All modules should be compiled with the same @option{-G @var{num}} value.
6264 Compiling with different values of @var{num} may or may not work; if it
6265 doesn't the linker will give an error message---incorrect code will not be
6271 @subsection M88K Options
6272 @cindex M88k options
6274 These @samp{-m} options are defined for Motorola 88k architectures:
6279 Generate code that works well on both the m88100 and the
6284 Generate code that works best for the m88100, but that also
6289 Generate code that works best for the m88110, and may not run
6294 Obsolete option to be removed from the next revision.
6297 @item -midentify-revision
6298 @opindex midentify-revision
6299 @cindex identifying source, compiler (88k)
6300 Include an @code{ident} directive in the assembler output recording the
6301 source file name, compiler name and version, timestamp, and compilation
6304 @item -mno-underscores
6305 @opindex mno-underscores
6306 @cindex underscores, avoiding (88k)
6307 In assembler output, emit symbol names without adding an underscore
6308 character at the beginning of each name. The default is to use an
6309 underscore as prefix on each name.
6311 @item -mocs-debug-info
6312 @itemx -mno-ocs-debug-info
6313 @opindex mocs-debug-info
6314 @opindex mno-ocs-debug-info
6316 @cindex debugging, 88k OCS
6317 Include (or omit) additional debugging information (about registers used
6318 in each stack frame) as specified in the 88open Object Compatibility
6319 Standard, ``OCS''@. This extra information allows debugging of code that
6320 has had the frame pointer eliminated. The default for DG/UX, SVr4, and
6321 Delta 88 SVr3.2 is to include this information; other 88k configurations
6322 omit this information by default.
6324 @item -mocs-frame-position
6325 @opindex mocs-frame-position
6326 @cindex register positions in frame (88k)
6327 When emitting COFF debugging information for automatic variables and
6328 parameters stored on the stack, use the offset from the canonical frame
6329 address, which is the stack pointer (register 31) on entry to the
6330 function. The DG/UX, SVr4, Delta88 SVr3.2, and BCS configurations use
6331 @option{-mocs-frame-position}; other 88k configurations have the default
6332 @option{-mno-ocs-frame-position}.
6334 @item -mno-ocs-frame-position
6335 @opindex mno-ocs-frame-position
6336 @cindex register positions in frame (88k)
6337 When emitting COFF debugging information for automatic variables and
6338 parameters stored on the stack, use the offset from the frame pointer
6339 register (register 30). When this option is in effect, the frame
6340 pointer is not eliminated when debugging information is selected by the
6343 @item -moptimize-arg-area
6344 @opindex moptimize-arg-area
6345 @cindex arguments in frame (88k)
6346 Save space by reorganizing the stack frame. This option generates code
6347 that does not agree with the 88open specifications, but uses less
6350 @itemx -mno-optimize-arg-area
6351 @opindex mno-optimize-arg-area
6352 Do not reorganize the stack frame to save space. This is the default.
6353 The generated conforms to the specification, but uses more memory.
6355 @item -mshort-data-@var{num}
6356 @opindex mshort-data
6357 @cindex smaller data references (88k)
6358 @cindex r0-relative references (88k)
6359 Generate smaller data references by making them relative to @code{r0},
6360 which allows loading a value using a single instruction (rather than the
6361 usual two). You control which data references are affected by
6362 specifying @var{num} with this option. For example, if you specify
6363 @option{-mshort-data-512}, then the data references affected are those
6364 involving displacements of less than 512 bytes.
6365 @option{-mshort-data-@var{num}} is not effective for @var{num} greater
6368 @item -mserialize-volatile
6369 @opindex mserialize-volatile
6370 @itemx -mno-serialize-volatile
6371 @opindex mno-serialize-volatile
6372 @cindex sequential consistency on 88k
6373 Do, or don't, generate code to guarantee sequential consistency
6374 of volatile memory references. By default, consistency is
6377 The order of memory references made by the MC88110 processor does
6378 not always match the order of the instructions requesting those
6379 references. In particular, a load instruction may execute before
6380 a preceding store instruction. Such reordering violates
6381 sequential consistency of volatile memory references, when there
6382 are multiple processors. When consistency must be guaranteed,
6383 GCC generates special instructions, as needed, to force
6384 execution in the proper order.
6386 The MC88100 processor does not reorder memory references and so
6387 always provides sequential consistency. However, by default, GCC
6388 generates the special instructions to guarantee consistency
6389 even when you use @option{-m88100}, so that the code may be run on an
6390 MC88110 processor. If you intend to run your code only on the
6391 MC88100 processor, you may use @option{-mno-serialize-volatile}.
6393 The extra code generated to guarantee consistency may affect the
6394 performance of your application. If you know that you can safely
6395 forgo this guarantee, you may use @option{-mno-serialize-volatile}.
6401 @cindex assembler syntax, 88k
6403 Turn on (@option{-msvr4}) or off (@option{-msvr3}) compiler extensions
6404 related to System V release 4 (SVr4). This controls the following:
6408 Which variant of the assembler syntax to emit.
6410 @option{-msvr4} makes the C preprocessor recognize @samp{#pragma weak}
6411 that is used on System V release 4.
6413 @option{-msvr4} makes GCC issue additional declaration directives used in
6417 @option{-msvr4} is the default for the m88k-motorola-sysv4 and
6418 m88k-dg-dgux m88k configurations. @option{-msvr3} is the default for all
6419 other m88k configurations.
6421 @item -mversion-03.00
6422 @opindex mversion-03.00
6423 This option is obsolete, and is ignored.
6424 @c ??? which asm syntax better for GAS? option there too?
6426 @item -mno-check-zero-division
6427 @itemx -mcheck-zero-division
6428 @opindex mno-check-zero-division
6429 @opindex mcheck-zero-division
6430 @cindex zero division on 88k
6431 Do, or don't, generate code to guarantee that integer division by
6432 zero will be detected. By default, detection is guaranteed.
6434 Some models of the MC88100 processor fail to trap upon integer
6435 division by zero under certain conditions. By default, when
6436 compiling code that might be run on such a processor, GCC
6437 generates code that explicitly checks for zero-valued divisors
6438 and traps with exception number 503 when one is detected. Use of
6439 @option{-mno-check-zero-division} suppresses such checking for code
6440 generated to run on an MC88100 processor.
6442 GCC assumes that the MC88110 processor correctly detects all instances
6443 of integer division by zero. When @option{-m88110} is specified, no
6444 explicit checks for zero-valued divisors are generated, and both
6445 @option{-mcheck-zero-division} and @option{-mno-check-zero-division} are
6448 @item -muse-div-instruction
6449 @opindex muse-div-instruction
6450 @cindex divide instruction, 88k
6451 Use the div instruction for signed integer division on the
6452 MC88100 processor. By default, the div instruction is not used.
6454 On the MC88100 processor the signed integer division instruction
6455 div) traps to the operating system on a negative operand. The
6456 operating system transparently completes the operation, but at a
6457 large cost in execution time. By default, when compiling code
6458 that might be run on an MC88100 processor, GCC emulates signed
6459 integer division using the unsigned integer division instruction
6460 divu), thereby avoiding the large penalty of a trap to the
6461 operating system. Such emulation has its own, smaller, execution
6462 cost in both time and space. To the extent that your code's
6463 important signed integer division operations are performed on two
6464 nonnegative operands, it may be desirable to use the div
6465 instruction directly.
6467 On the MC88110 processor the div instruction (also known as the
6468 divs instruction) processes negative operands without trapping to
6469 the operating system. When @option{-m88110} is specified,
6470 @option{-muse-div-instruction} is ignored, and the div instruction is used
6471 for signed integer division.
6473 Note that the result of dividing @code{INT_MIN} by @minus{}1 is undefined. In
6474 particular, the behavior of such a division with and without
6475 @option{-muse-div-instruction} may differ.
6477 @item -mtrap-large-shift
6478 @itemx -mhandle-large-shift
6479 @opindex mtrap-large-shift
6480 @opindex mhandle-large-shift
6481 @cindex bit shift overflow (88k)
6482 @cindex large bit shifts (88k)
6483 Include code to detect bit-shifts of more than 31 bits; respectively,
6484 trap such shifts or emit code to handle them properly. By default GCC
6485 makes no special provision for large bit shifts.
6487 @item -mwarn-passed-structs
6488 @opindex mwarn-passed-structs
6489 @cindex structure passing (88k)
6490 Warn when a function passes a struct as an argument or result.
6491 Structure-passing conventions have changed during the evolution of the C
6492 language, and are often the source of portability problems. By default,
6493 GCC issues no such warning.
6496 @c break page here to avoid unsightly interparagraph stretch.
6500 @node RS/6000 and PowerPC Options
6501 @subsection IBM RS/6000 and PowerPC Options
6502 @cindex RS/6000 and PowerPC Options
6503 @cindex IBM RS/6000 and PowerPC Options
6505 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
6513 @itemx -mpowerpc-gpopt
6514 @itemx -mno-powerpc-gpopt
6515 @itemx -mpowerpc-gfxopt
6516 @itemx -mno-powerpc-gfxopt
6518 @itemx -mno-powerpc64
6524 @opindex mno-powerpc
6525 @opindex mpowerpc-gpopt
6526 @opindex mno-powerpc-gpopt
6527 @opindex mpowerpc-gfxopt
6528 @opindex mno-powerpc-gfxopt
6530 @opindex mno-powerpc64
6531 GCC supports two related instruction set architectures for the
6532 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
6533 instructions supported by the @samp{rios} chip set used in the original
6534 RS/6000 systems and the @dfn{PowerPC} instruction set is the
6535 architecture of the Motorola MPC5xx, MPC6xx, MPC8xx microprocessors, and
6536 the IBM 4xx microprocessors.
6538 Neither architecture is a subset of the other. However there is a
6539 large common subset of instructions supported by both. An MQ
6540 register is included in processors supporting the POWER architecture.
6542 You use these options to specify which instructions are available on the
6543 processor you are using. The default value of these options is
6544 determined when configuring GCC@. Specifying the
6545 @option{-mcpu=@var{cpu_type}} overrides the specification of these
6546 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
6547 rather than the options listed above.
6549 The @option{-mpower} option allows GCC to generate instructions that
6550 are found only in the POWER architecture and to use the MQ register.
6551 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
6552 to generate instructions that are present in the POWER2 architecture but
6553 not the original POWER architecture.
6555 The @option{-mpowerpc} option allows GCC to generate instructions that
6556 are found only in the 32-bit subset of the PowerPC architecture.
6557 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
6558 GCC to use the optional PowerPC architecture instructions in the
6559 General Purpose group, including floating-point square root. Specifying
6560 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
6561 use the optional PowerPC architecture instructions in the Graphics
6562 group, including floating-point select.
6564 The @option{-mpowerpc64} option allows GCC to generate the additional
6565 64-bit instructions that are found in the full PowerPC64 architecture
6566 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
6567 @option{-mno-powerpc64}.
6569 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
6570 will use only the instructions in the common subset of both
6571 architectures plus some special AIX common-mode calls, and will not use
6572 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
6573 permits GCC to use any instruction from either architecture and to
6574 allow use of the MQ register; specify this for the Motorola MPC601.
6576 @item -mnew-mnemonics
6577 @itemx -mold-mnemonics
6578 @opindex mnew-mnemonics
6579 @opindex mold-mnemonics
6580 Select which mnemonics to use in the generated assembler code. With
6581 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
6582 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
6583 assembler mnemonics defined for the POWER architecture. Instructions
6584 defined in only one architecture have only one mnemonic; GCC uses that
6585 mnemonic irrespective of which of these options is specified.
6587 GCC defaults to the mnemonics appropriate for the architecture in
6588 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
6589 value of these option. Unless you are building a cross-compiler, you
6590 should normally not specify either @option{-mnew-mnemonics} or
6591 @option{-mold-mnemonics}, but should instead accept the default.
6593 @item -mcpu=@var{cpu_type}
6595 Set architecture type, register usage, choice of mnemonics, and
6596 instruction scheduling parameters for machine type @var{cpu_type}.
6597 Supported values for @var{cpu_type} are @samp{rios}, @samp{rios1},
6598 @samp{rsc}, @samp{rios2}, @samp{rs64a}, @samp{601}, @samp{602},
6599 @samp{603}, @samp{603e}, @samp{604}, @samp{604e}, @samp{620},
6600 @samp{630}, @samp{740}, @samp{750}, @samp{power}, @samp{power2},
6601 @samp{powerpc}, @samp{403}, @samp{505}, @samp{801}, @samp{821},
6602 @samp{823}, and @samp{860} and @samp{common}.
6604 @option{-mcpu=common} selects a completely generic processor. Code
6605 generated under this option will run on any POWER or PowerPC processor.
6606 GCC will use only the instructions in the common subset of both
6607 architectures, and will not use the MQ register. GCC assumes a generic
6608 processor model for scheduling purposes.
6610 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
6611 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
6612 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
6613 types, with an appropriate, generic processor model assumed for
6614 scheduling purposes.
6616 The other options specify a specific processor. Code generated under
6617 those options will run best on that processor, and may not run at all on
6620 The @option{-mcpu} options automatically enable or disable other
6621 @option{-m} options as follows:
6625 @option{-mno-power}, @option{-mno-powerc}
6632 @option{-mpower}, @option{-mno-powerpc}, @option{-mno-new-mnemonics}
6645 @option{-mno-power}, @option{-mpowerpc}, @option{-mnew-mnemonics}
6648 @option{-mpower}, @option{-mpowerpc}, @option{-mnew-mnemonics}
6653 @option{-mno-power}, @option{-mpowerpc}, @option{-mnew-mnemonics}, @option{-msoft-float}
6656 @item -mtune=@var{cpu_type}
6658 Set the instruction scheduling parameters for machine type
6659 @var{cpu_type}, but do not set the architecture type, register usage, or
6660 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
6661 values for @var{cpu_type} are used for @option{-mtune} as for
6662 @option{-mcpu}. If both are specified, the code generated will use the
6663 architecture, registers, and mnemonics set by @option{-mcpu}, but the
6664 scheduling parameters set by @option{-mtune}.
6667 @itemx -mno-fp-in-toc
6668 @itemx -mno-sum-in-toc
6669 @itemx -mminimal-toc
6671 @opindex mno-fp-in-toc
6672 @opindex mno-sum-in-toc
6673 @opindex mminimal-toc
6674 Modify generation of the TOC (Table Of Contents), which is created for
6675 every executable file. The @option{-mfull-toc} option is selected by
6676 default. In that case, GCC will allocate at least one TOC entry for
6677 each unique non-automatic variable reference in your program. GCC
6678 will also place floating-point constants in the TOC@. However, only
6679 16,384 entries are available in the TOC@.
6681 If you receive a linker error message that saying you have overflowed
6682 the available TOC space, you can reduce the amount of TOC space used
6683 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
6684 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
6685 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
6686 generate code to calculate the sum of an address and a constant at
6687 run-time instead of putting that sum into the TOC@. You may specify one
6688 or both of these options. Each causes GCC to produce very slightly
6689 slower and larger code at the expense of conserving TOC space.
6691 If you still run out of space in the TOC even when you specify both of
6692 these options, specify @option{-mminimal-toc} instead. This option causes
6693 GCC to make only one TOC entry for every file. When you specify this
6694 option, GCC will produce code that is slower and larger but which
6695 uses extremely little TOC space. You may wish to use this option
6696 only on files that contain less frequently executed code.
6702 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
6703 @code{long} type, and the infrastructure needed to support them.
6704 Specifying @option{-maix64} implies @option{-mpowerpc64} and
6705 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
6706 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
6711 @opindex mno-xl-call
6712 On AIX, pass floating-point arguments to prototyped functions beyond the
6713 register save area (RSA) on the stack in addition to argument FPRs. The
6714 AIX calling convention was extended but not initially documented to
6715 handle an obscure K&R C case of calling a function that takes the
6716 address of its arguments with fewer arguments than declared. AIX XL
6717 compilers access floating point arguments which do not fit in the
6718 RSA from the stack when a subroutine is compiled without
6719 optimization. Because always storing floating-point arguments on the
6720 stack is inefficient and rarely needed, this option is not enabled by
6721 default and only is necessary when calling subroutines compiled by AIX
6722 XL compilers without optimization.
6726 Support @dfn{AIX Threads}. Link an application written to use
6727 @dfn{pthreads} with special libraries and startup code to enable the
6732 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
6733 application written to use message passing with special startup code to
6734 enable the application to run. The system must have PE installed in the
6735 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
6736 must be overridden with the @option{-specs=} option to specify the
6737 appropriate directory location. The Parallel Environment does not
6738 support threads, so the @option{-mpe} option and the @option{-mthreads}
6739 option are incompatible.
6743 @opindex msoft-float
6744 @opindex mhard-float
6745 Generate code that does not use (uses) the floating-point register set.
6746 Software floating point emulation is provided if you use the
6747 @option{-msoft-float} option, and pass the option to GCC when linking.
6750 @itemx -mno-multiple
6752 @opindex mno-multiple
6753 Generate code that uses (does not use) the load multiple word
6754 instructions and the store multiple word instructions. These
6755 instructions are generated by default on POWER systems, and not
6756 generated on PowerPC systems. Do not use @option{-mmultiple} on little
6757 endian PowerPC systems, since those instructions do not work when the
6758 processor is in little endian mode. The exceptions are PPC740 and
6759 PPC750 which permit the instructions usage in little endian mode.
6765 Generate code that uses (does not use) the load string instructions
6766 and the store string word instructions to save multiple registers and
6767 do small block moves. These instructions are generated by default on
6768 POWER systems, and not generated on PowerPC systems. Do not use
6769 @option{-mstring} on little endian PowerPC systems, since those
6770 instructions do not work when the processor is in little endian mode.
6771 The exceptions are PPC740 and PPC750 which permit the instructions
6772 usage in little endian mode.
6778 Generate code that uses (does not use) the load or store instructions
6779 that update the base register to the address of the calculated memory
6780 location. These instructions are generated by default. If you use
6781 @option{-mno-update}, there is a small window between the time that the
6782 stack pointer is updated and the address of the previous frame is
6783 stored, which means code that walks the stack frame across interrupts or
6784 signals may get corrupted data.
6787 @itemx -mno-fused-madd
6788 @opindex mfused-madd
6789 @opindex mno-fused-madd
6790 Generate code that uses (does not use) the floating point multiply and
6791 accumulate instructions. These instructions are generated by default if
6792 hardware floating is used.
6794 @item -mno-bit-align
6796 @opindex mno-bit-align
6798 On System V.4 and embedded PowerPC systems do not (do) force structures
6799 and unions that contain bit-fields to be aligned to the base type of the
6802 For example, by default a structure containing nothing but 8
6803 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
6804 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
6805 the structure would be aligned to a 1 byte boundary and be one byte in
6808 @item -mno-strict-align
6809 @itemx -mstrict-align
6810 @opindex mno-strict-align
6811 @opindex mstrict-align
6812 On System V.4 and embedded PowerPC systems do not (do) assume that
6813 unaligned memory references will be handled by the system.
6816 @itemx -mno-relocatable
6817 @opindex mrelocatable
6818 @opindex mno-relocatable
6819 On embedded PowerPC systems generate code that allows (does not allow)
6820 the program to be relocated to a different address at runtime. If you
6821 use @option{-mrelocatable} on any module, all objects linked together must
6822 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
6824 @item -mrelocatable-lib
6825 @itemx -mno-relocatable-lib
6826 @opindex mrelocatable-lib
6827 @opindex mno-relocatable-lib
6828 On embedded PowerPC systems generate code that allows (does not allow)
6829 the program to be relocated to a different address at runtime. Modules
6830 compiled with @option{-mrelocatable-lib} can be linked with either modules
6831 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
6832 with modules compiled with the @option{-mrelocatable} options.
6838 On System V.4 and embedded PowerPC systems do not (do) assume that
6839 register 2 contains a pointer to a global area pointing to the addresses
6840 used in the program.
6843 @itemx -mlittle-endian
6845 @opindex mlittle-endian
6846 On System V.4 and embedded PowerPC systems compile code for the
6847 processor in little endian mode. The @option{-mlittle-endian} option is
6848 the same as @option{-mlittle}.
6853 @opindex mbig-endian
6854 On System V.4 and embedded PowerPC systems compile code for the
6855 processor in big endian mode. The @option{-mbig-endian} option is
6856 the same as @option{-mbig}.
6860 On System V.4 and embedded PowerPC systems compile code using calling
6861 conventions that adheres to the March 1995 draft of the System V
6862 Application Binary Interface, PowerPC processor supplement. This is the
6863 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
6865 @item -mcall-sysv-eabi
6866 @opindex mcall-sysv-eabi
6867 Specify both @option{-mcall-sysv} and @option{-meabi} options.
6869 @item -mcall-sysv-noeabi
6870 @opindex mcall-sysv-noeabi
6871 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
6875 On System V.4 and embedded PowerPC systems compile code using calling
6876 conventions that are similar to those used on AIX@. This is the
6877 default if you configured GCC using @samp{powerpc-*-eabiaix}.
6879 @item -mcall-solaris
6880 @opindex mcall-solaris
6881 On System V.4 and embedded PowerPC systems compile code for the Solaris
6885 @opindex mcall-linux
6886 On System V.4 and embedded PowerPC systems compile code for the
6887 Linux-based GNU system.
6890 @opindex mcall-netbsd
6891 On System V.4 and embedded PowerPC systems compile code for the
6892 NetBSD operating system.
6895 @itemx -mno-prototype
6897 @opindex mno-prototype
6898 On System V.4 and embedded PowerPC systems assume that all calls to
6899 variable argument functions are properly prototyped. Otherwise, the
6900 compiler must insert an instruction before every non prototyped call to
6901 set or clear bit 6 of the condition code register (@var{CR}) to
6902 indicate whether floating point values were passed in the floating point
6903 registers in case the function takes a variable arguments. With
6904 @option{-mprototype}, only calls to prototyped variable argument functions
6905 will set or clear the bit.
6909 On embedded PowerPC systems, assume that the startup module is called
6910 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
6911 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
6916 On embedded PowerPC systems, assume that the startup module is called
6917 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
6922 On embedded PowerPC systems, assume that the startup module is called
6923 @file{crt0.o} and the standard C libraries are @file{libads.a} and
6927 @opindex myellowknife
6928 On embedded PowerPC systems, assume that the startup module is called
6929 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
6934 On System V.4 and embedded PowerPC systems, specify that you are
6935 compiling for a VxWorks system.
6939 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
6940 header to indicate that @samp{eabi} extended relocations are used.
6946 On System V.4 and embedded PowerPC systems do (do not) adhere to the
6947 Embedded Applications Binary Interface (eabi) which is a set of
6948 modifications to the System V.4 specifications. Selecting @option{-meabi}
6949 means that the stack is aligned to an 8 byte boundary, a function
6950 @code{__eabi} is called to from @code{main} to set up the eabi
6951 environment, and the @option{-msdata} option can use both @code{r2} and
6952 @code{r13} to point to two separate small data areas. Selecting
6953 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
6954 do not call an initialization function from @code{main}, and the
6955 @option{-msdata} option will only use @code{r13} to point to a single
6956 small data area. The @option{-meabi} option is on by default if you
6957 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
6960 @opindex msdata=eabi
6961 On System V.4 and embedded PowerPC systems, put small initialized
6962 @code{const} global and static data in the @samp{.sdata2} section, which
6963 is pointed to by register @code{r2}. Put small initialized
6964 non-@code{const} global and static data in the @samp{.sdata} section,
6965 which is pointed to by register @code{r13}. Put small uninitialized
6966 global and static data in the @samp{.sbss} section, which is adjacent to
6967 the @samp{.sdata} section. The @option{-msdata=eabi} option is
6968 incompatible with the @option{-mrelocatable} option. The
6969 @option{-msdata=eabi} option also sets the @option{-memb} option.
6972 @opindex msdata=sysv
6973 On System V.4 and embedded PowerPC systems, put small global and static
6974 data in the @samp{.sdata} section, which is pointed to by register
6975 @code{r13}. Put small uninitialized global and static data in the
6976 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
6977 The @option{-msdata=sysv} option is incompatible with the
6978 @option{-mrelocatable} option.
6980 @item -msdata=default
6982 @opindex msdata=default
6984 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
6985 compile code the same as @option{-msdata=eabi}, otherwise compile code the
6986 same as @option{-msdata=sysv}.
6989 @opindex msdata-data
6990 On System V.4 and embedded PowerPC systems, put small global and static
6991 data in the @samp{.sdata} section. Put small uninitialized global and
6992 static data in the @samp{.sbss} section. Do not use register @code{r13}
6993 to address small data however. This is the default behavior unless
6994 other @option{-msdata} options are used.
6998 @opindex msdata=none
7000 On embedded PowerPC systems, put all initialized global and static data
7001 in the @samp{.data} section, and all uninitialized data in the
7002 @samp{.bss} section.
7006 @cindex smaller data references (PowerPC)
7007 @cindex .sdata/.sdata2 references (PowerPC)
7008 On embedded PowerPC systems, put global and static items less than or
7009 equal to @var{num} bytes into the small data or bss sections instead of
7010 the normal data or bss section. By default, @var{num} is 8. The
7011 @option{-G @var{num}} switch is also passed to the linker.
7012 All modules should be compiled with the same @option{-G @var{num}} value.
7015 @itemx -mno-regnames
7017 @opindex mno-regnames
7018 On System V.4 and embedded PowerPC systems do (do not) emit register
7019 names in the assembly language output using symbolic forms.
7024 @subsection IBM RT Options
7026 @cindex IBM RT options
7028 These @samp{-m} options are defined for the IBM RT PC:
7032 @opindex min-line-mul
7033 Use an in-line code sequence for integer multiplies. This is the
7036 @item -mcall-lib-mul
7037 @opindex mcall-lib-mul
7038 Call @code{lmul$$} for integer multiples.
7040 @item -mfull-fp-blocks
7041 @opindex mfull-fp-blocks
7042 Generate full-size floating point data blocks, including the minimum
7043 amount of scratch space recommended by IBM@. This is the default.
7045 @item -mminimum-fp-blocks
7046 @opindex mminimum-fp-blocks
7047 Do not include extra scratch space in floating point data blocks. This
7048 results in smaller code, but slower execution, since scratch space must
7049 be allocated dynamically.
7051 @cindex @file{varargs.h} and RT PC
7052 @cindex @file{stdarg.h} and RT PC
7053 @item -mfp-arg-in-fpregs
7054 @opindex mfp-arg-in-fpregs
7055 Use a calling sequence incompatible with the IBM calling convention in
7056 which floating point arguments are passed in floating point registers.
7057 Note that @code{varargs.h} and @code{stdarg.h} will not work with
7058 floating point operands if this option is specified.
7060 @item -mfp-arg-in-gregs
7061 @opindex mfp-arg-in-gregs
7062 Use the normal calling convention for floating point arguments. This is
7065 @item -mhc-struct-return
7066 @opindex mhc-struct-return
7067 Return structures of more than one word in memory, rather than in a
7068 register. This provides compatibility with the MetaWare HighC (hc)
7069 compiler. Use the option @option{-fpcc-struct-return} for compatibility
7070 with the Portable C Compiler (pcc).
7072 @item -mnohc-struct-return
7073 @opindex mnohc-struct-return
7074 Return some structures of more than one word in registers, when
7075 convenient. This is the default. For compatibility with the
7076 IBM-supplied compilers, use the option @option{-fpcc-struct-return} or the
7077 option @option{-mhc-struct-return}.
7081 @subsection MIPS Options
7082 @cindex MIPS options
7084 These @samp{-m} options are defined for the MIPS family of computers:
7088 @item -march=@var{cpu-type}
7090 Assume the defaults for the machine type @var{cpu-type} when generating
7091 instructions. The choices for @var{cpu-type} are @samp{r2000}, @samp{r3000},
7092 @samp{r3900}, @samp{r4000}, @samp{r4100}, @samp{r4300}, @samp{r4400},
7093 @samp{r4600}, @samp{r4650}, @samp{r5000}, @samp{r6000}, @samp{r8000},
7094 and @samp{orion}. Additionally, the @samp{r2000}, @samp{r3000},
7095 @samp{r4000}, @samp{r5000}, and @samp{r6000} can be abbreviated as
7096 @samp{r2k} (or @samp{r2K}), @samp{r3k}, etc.
7098 @item -mtune=@var{cpu-type}
7100 Assume the defaults for the machine type @var{cpu-type} when scheduling
7101 instructions. The choices for @var{cpu-type} are @samp{r2000}, @samp{r3000},
7102 @samp{r3900}, @samp{r4000}, @samp{r4100}, @samp{r4300}, @samp{r4400},
7103 @samp{r4600}, @samp{r4650}, @samp{r5000}, @samp{r6000}, @samp{r8000},
7104 and @samp{orion}. Additionally, the @samp{r2000}, @samp{r3000},
7105 @samp{r4000}, @samp{r5000}, and @samp{r6000} can be abbreviated as
7106 @samp{r2k} (or @samp{r2K}), @samp{r3k}, etc. While picking a specific
7107 @var{cpu-type} will schedule things appropriately for that particular
7108 chip, the compiler will not generate any code that does not meet level 1
7109 of the MIPS ISA (instruction set architecture) without a @option{-mipsX}
7110 or @option{-mabi} switch being used.
7112 @item -mcpu=@var{cpu-type}
7114 This is identical to specifying both @option{-march} and @option{-mtune}.
7118 Issue instructions from level 1 of the MIPS ISA@. This is the default.
7119 @samp{r3000} is the default @var{cpu-type} at this ISA level.
7123 Issue instructions from level 2 of the MIPS ISA (branch likely, square
7124 root instructions). @samp{r6000} is the default @var{cpu-type} at this
7129 Issue instructions from level 3 of the MIPS ISA (64-bit instructions).
7130 @samp{r4000} is the default @var{cpu-type} at this ISA level.
7134 Issue instructions from level 4 of the MIPS ISA (conditional move,
7135 prefetch, enhanced FPU instructions). @samp{r8000} is the default
7136 @var{cpu-type} at this ISA level.
7140 Assume that 32 32-bit floating point registers are available. This is
7145 Assume that 32 64-bit floating point registers are available. This is
7146 the default when the @option{-mips3} option is used.
7149 @itemx -mno-fused-madd
7150 @opindex mfused-madd
7151 @opindex mno-fused-madd
7152 Generate code that uses (does not use) the floating point multiply and
7153 accumulate instructions, when they are available. These instructions
7154 are generated by default if they are available, but this may be
7155 undesirable if the extra precision causes problems or on certain chips
7156 in the mode where denormals are rounded to zero where denormals
7157 generated by multiply and accumulate instructions cause exceptions
7162 Assume that 32 32-bit general purpose registers are available. This is
7167 Assume that 32 64-bit general purpose registers are available. This is
7168 the default when the @option{-mips3} option is used.
7172 Force int and long types to be 64 bits wide. See @option{-mlong32} for an
7173 explanation of the default, and the width of pointers.
7177 Force long types to be 64 bits wide. See @option{-mlong32} for an
7178 explanation of the default, and the width of pointers.
7182 Force long, int, and pointer types to be 32 bits wide.
7184 If none of @option{-mlong32}, @option{-mlong64}, or @option{-mint64} are set,
7185 the size of ints, longs, and pointers depends on the ABI and ISA chosen.
7186 For @option{-mabi=32}, and @option{-mabi=n32}, ints and longs are 32 bits
7187 wide. For @option{-mabi=64}, ints are 32 bits, and longs are 64 bits wide.
7188 For @option{-mabi=eabi} and either @option{-mips1} or @option{-mips2}, ints
7189 and longs are 32 bits wide. For @option{-mabi=eabi} and higher ISAs, ints
7190 are 32 bits, and longs are 64 bits wide. The width of pointer types is
7191 the smaller of the width of longs or the width of general purpose
7192 registers (which in turn depends on the ISA)@.
7204 Generate code for the indicated ABI@. The default instruction level is
7205 @option{-mips1} for @samp{32}, @option{-mips3} for @samp{n32}, and
7206 @option{-mips4} otherwise. Conversely, with @option{-mips1} or
7207 @option{-mips2}, the default ABI is @samp{32}; otherwise, the default ABI
7212 Generate code for the MIPS assembler, and invoke @file{mips-tfile} to
7213 add normal debug information. This is the default for all
7214 platforms except for the OSF/1 reference platform, using the OSF/rose
7215 object format. If the either of the @option{-gstabs} or @option{-gstabs+}
7216 switches are used, the @file{mips-tfile} program will encapsulate the
7217 stabs within MIPS ECOFF@.
7221 Generate code for the GNU assembler. This is the default on the OSF/1
7222 reference platform, using the OSF/rose object format. Also, this is
7223 the default if the configure option @option{--with-gnu-as} is used.
7225 @item -msplit-addresses
7226 @itemx -mno-split-addresses
7227 @opindex msplit-addresses
7228 @opindex mno-split-addresses
7229 Generate code to load the high and low parts of address constants separately.
7230 This allows GCC to optimize away redundant loads of the high order
7231 bits of addresses. This optimization requires GNU as and GNU ld.
7232 This optimization is enabled by default for some embedded targets where
7233 GNU as and GNU ld are standard.
7239 The @option{-mrnames} switch says to output code using the MIPS software
7240 names for the registers, instead of the hardware names (ie, @var{a0}
7241 instead of @var{$4}). The only known assembler that supports this option
7242 is the Algorithmics assembler.
7248 The @option{-mgpopt} switch says to write all of the data declarations
7249 before the instructions in the text section, this allows the MIPS
7250 assembler to generate one word memory references instead of using two
7251 words for short global or static data items. This is on by default if
7252 optimization is selected.
7258 For each non-inline function processed, the @option{-mstats} switch
7259 causes the compiler to emit one line to the standard error file to
7260 print statistics about the program (number of registers saved, stack
7267 The @option{-mmemcpy} switch makes all block moves call the appropriate
7268 string function (@samp{memcpy} or @samp{bcopy}) instead of possibly
7269 generating inline code.
7272 @itemx -mno-mips-tfile
7273 @opindex mmips-tfile
7274 @opindex mno-mips-tfile
7275 The @option{-mno-mips-tfile} switch causes the compiler not
7276 postprocess the object file with the @file{mips-tfile} program,
7277 after the MIPS assembler has generated it to add debug support. If
7278 @file{mips-tfile} is not run, then no local variables will be
7279 available to the debugger. In addition, @file{stage2} and
7280 @file{stage3} objects will have the temporary file names passed to the
7281 assembler embedded in the object file, which means the objects will
7282 not compare the same. The @option{-mno-mips-tfile} switch should only
7283 be used when there are bugs in the @file{mips-tfile} program that
7284 prevents compilation.
7287 @opindex msoft-float
7288 Generate output containing library calls for floating point.
7289 @strong{Warning:} the requisite libraries are not part of GCC@.
7290 Normally the facilities of the machine's usual C compiler are used, but
7291 this can't be done directly in cross-compilation. You must make your
7292 own arrangements to provide suitable library functions for
7296 @opindex mhard-float
7297 Generate output containing floating point instructions. This is the
7298 default if you use the unmodified sources.
7301 @itemx -mno-abicalls
7303 @opindex mno-abicalls
7304 Emit (or do not emit) the pseudo operations @samp{.abicalls},
7305 @samp{.cpload}, and @samp{.cprestore} that some System V.4 ports use for
7306 position independent code.
7309 @itemx -mno-long-calls
7310 @opindex mlong-calls
7311 @opindex mno-long-calls
7312 Do all calls with the @samp{JALR} instruction, which requires
7313 loading up a function's address into a register before the call.
7314 You need to use this switch, if you call outside of the current
7315 512 megabyte segment to functions that are not through pointers.
7318 @itemx -mno-half-pic
7320 @opindex mno-half-pic
7321 Put pointers to extern references into the data section and load them
7322 up, rather than put the references in the text section.
7324 @item -membedded-pic
7325 @itemx -mno-embedded-pic
7326 @opindex membedded-pic
7327 @opindex mno-embedded-pic
7328 Generate PIC code suitable for some embedded systems. All calls are
7329 made using PC relative address, and all data is addressed using the $gp
7330 register. No more than 65536 bytes of global data may be used. This
7331 requires GNU as and GNU ld which do most of the work. This currently
7332 only works on targets which use ECOFF; it does not work with ELF@.
7334 @item -membedded-data
7335 @itemx -mno-embedded-data
7336 @opindex membedded-data
7337 @opindex mno-embedded-data
7338 Allocate variables to the read-only data section first if possible, then
7339 next in the small data section if possible, otherwise in data. This gives
7340 slightly slower code than the default, but reduces the amount of RAM required
7341 when executing, and thus may be preferred for some embedded systems.
7343 @item -muninit-const-in-rodata
7344 @itemx -mno-uninit-const-in-rodata
7345 @opindex muninit-const-in-rodata
7346 @opindex mno-uninit-const-in-rodata
7347 When used together with @option{-membedded-data}, it will always store uninitialized
7348 const variables in the read-only data section.
7350 @item -msingle-float
7351 @itemx -mdouble-float
7352 @opindex msingle-float
7353 @opindex mdouble-float
7354 The @option{-msingle-float} switch tells gcc to assume that the floating
7355 point coprocessor only supports single precision operations, as on the
7356 @samp{r4650} chip. The @option{-mdouble-float} switch permits gcc to use
7357 double precision operations. This is the default.
7363 Permit use of the @samp{mad}, @samp{madu} and @samp{mul} instructions,
7364 as on the @samp{r4650} chip.
7368 Turns on @option{-msingle-float}, @option{-mmad}, and, at least for now,
7369 @option{-mcpu=r4650}.
7375 Enable 16-bit instructions.
7379 Use the entry and exit pseudo ops. This option can only be used with
7384 Compile code for the processor in little endian mode.
7385 The requisite libraries are assumed to exist.
7389 Compile code for the processor in big endian mode.
7390 The requisite libraries are assumed to exist.
7394 @cindex smaller data references (MIPS)
7395 @cindex gp-relative references (MIPS)
7396 Put global and static items less than or equal to @var{num} bytes into
7397 the small data or bss sections instead of the normal data or bss
7398 section. This allows the assembler to emit one word memory reference
7399 instructions based on the global pointer (@var{gp} or @var{$28}),
7400 instead of the normal two words used. By default, @var{num} is 8 when
7401 the MIPS assembler is used, and 0 when the GNU assembler is used. The
7402 @option{-G @var{num}} switch is also passed to the assembler and linker.
7403 All modules should be compiled with the same @option{-G @var{num}}
7408 Tell the MIPS assembler to not run its preprocessor over user
7409 assembler files (with a @samp{.s} suffix) when assembling them.
7413 Pass an option to gas which will cause nops to be inserted if
7414 the read of the destination register of an mfhi or mflo instruction
7415 occurs in the following two instructions.
7419 Do not include the default crt0.
7423 These options are defined by the macro
7424 @code{TARGET_SWITCHES} in the machine description. The default for the
7425 options is also defined by that macro, which enables you to change the
7430 @subsection Intel 386 Options
7431 @cindex i386 Options
7432 @cindex Intel 386 Options
7434 These @samp{-m} options are defined for the i386 family of computers:
7437 @item -mcpu=@var{cpu-type}
7439 Assume the defaults for the machine type @var{cpu-type} when scheduling
7440 instructions. The choices for @var{cpu-type} are @samp{i386},
7441 @samp{i486}, @samp{i586}, @samp{i686}, @samp{pentium},
7442 @samp{pentiumpro}, @samp{pentium4}, @samp{k6}, and @samp{athlon}
7444 While picking a specific @var{cpu-type} will schedule things appropriately
7445 for that particular chip, the compiler will not generate any code that
7446 does not run on the i386 without the @option{-march=@var{cpu-type}} option
7447 being used. @samp{i586} is equivalent to @samp{pentium} and @samp{i686}
7448 is equivalent to @samp{pentiumpro}. @samp{k6} and @samp{athlon} are the
7449 AMD chips as opposed to the Intel ones.
7451 @item -march=@var{cpu-type}
7453 Generate instructions for the machine type @var{cpu-type}. The choices
7454 for @var{cpu-type} are the same as for @option{-mcpu}. Moreover,
7455 specifying @option{-march=@var{cpu-type}} implies @option{-mcpu=@var{cpu-type}}.
7464 @opindex mpentiumpro
7465 These options are synonyms for @option{-mcpu=i386}, @option{-mcpu=i486},
7466 @option{-mcpu=pentium}, and @option{-mcpu=pentiumpro} respectively.
7467 These synonyms are deprecated.
7469 @item -mintel-syntax
7470 @opindex mintel-syntax
7471 Emit assembly using Intel syntax opcodes instead of AT&T syntax.
7476 @opindex mno-ieee-fp
7477 Control whether or not the compiler uses IEEE floating point
7478 comparisons. These handle correctly the case where the result of a
7479 comparison is unordered.
7482 @opindex msoft-float
7483 Generate output containing library calls for floating point.
7484 @strong{Warning:} the requisite libraries are not part of GCC@.
7485 Normally the facilities of the machine's usual C compiler are used, but
7486 this can't be done directly in cross-compilation. You must make your
7487 own arrangements to provide suitable library functions for
7490 On machines where a function returns floating point results in the 80387
7491 register stack, some floating point opcodes may be emitted even if
7492 @option{-msoft-float} is used.
7494 @item -mno-fp-ret-in-387
7495 @opindex mno-fp-ret-in-387
7496 Do not use the FPU registers for return values of functions.
7498 The usual calling convention has functions return values of types
7499 @code{float} and @code{double} in an FPU register, even if there
7500 is no FPU@. The idea is that the operating system should emulate
7503 The option @option{-mno-fp-ret-in-387} causes such values to be returned
7504 in ordinary CPU registers instead.
7506 @item -mno-fancy-math-387
7507 @opindex mno-fancy-math-387
7508 Some 387 emulators do not support the @code{sin}, @code{cos} and
7509 @code{sqrt} instructions for the 387. Specify this option to avoid
7510 generating those instructions. This option is the default on FreeBSD@.
7511 As of revision 2.6.1, these instructions are not generated unless you
7512 also use the @option{-funsafe-math-optimizations} switch.
7514 @item -malign-double
7515 @itemx -mno-align-double
7516 @opindex malign-double
7517 @opindex mno-align-double
7518 Control whether GCC aligns @code{double}, @code{long double}, and
7519 @code{long long} variables on a two word boundary or a one word
7520 boundary. Aligning @code{double} variables on a two word boundary will
7521 produce code that runs somewhat faster on a @samp{Pentium} at the
7522 expense of more memory.
7524 @item -m128bit-long-double
7525 @opindex m128bit-long-double
7526 Control the size of @code{long double} type. i386 application binary interface
7527 specify the size to be 12 bytes, while modern architectures (Pentium and newer)
7528 prefer @code{long double} aligned to 8 or 16 byte boundary. This is
7529 impossible to reach with 12 byte long doubles in the array accesses.
7531 @strong{Warning:} if you use the @option{-m128bit-long-double} switch, the
7532 structures and arrays containing @code{long double} will change their size as
7533 well as function calling convention for function taking @code{long double}
7536 @item -m96bit-long-double
7537 @opindex m96bit-long-double
7538 Set the size of @code{long double} to 96 bits as required by the i386
7539 application binary interface. This is the default.
7542 @itemx -mno-svr3-shlib
7543 @opindex msvr3-shlib
7544 @opindex mno-svr3-shlib
7545 Control whether GCC places uninitialized local variables into the
7546 @code{bss} or @code{data} segments. @option{-msvr3-shlib} places them
7547 into @code{bss}. These options are meaningful only on System V Release 3.
7551 Use a different function-calling convention, in which functions that
7552 take a fixed number of arguments return with the @code{ret} @var{num}
7553 instruction, which pops their arguments while returning. This saves one
7554 instruction in the caller since there is no need to pop the arguments
7557 You can specify that an individual function is called with this calling
7558 sequence with the function attribute @samp{stdcall}. You can also
7559 override the @option{-mrtd} option by using the function attribute
7560 @samp{cdecl}. @xref{Function Attributes}.
7562 @strong{Warning:} this calling convention is incompatible with the one
7563 normally used on Unix, so you cannot use it if you need to call
7564 libraries compiled with the Unix compiler.
7566 Also, you must provide function prototypes for all functions that
7567 take variable numbers of arguments (including @code{printf});
7568 otherwise incorrect code will be generated for calls to those
7571 In addition, seriously incorrect code will result if you call a
7572 function with too many arguments. (Normally, extra arguments are
7573 harmlessly ignored.)
7575 @item -mregparm=@var{num}
7577 Control how many registers are used to pass integer arguments. By
7578 default, no registers are used to pass arguments, and at most 3
7579 registers can be used. You can control this behavior for a specific
7580 function by using the function attribute @samp{regparm}.
7581 @xref{Function Attributes}.
7583 @strong{Warning:} if you use this switch, and
7584 @var{num} is nonzero, then you must build all modules with the same
7585 value, including any libraries. This includes the system libraries and
7588 @item -mpreferred-stack-boundary=@var{num}
7589 @opindex mpreferred-stack-boundary
7590 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
7591 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
7592 the default is 4 (16 bytes or 128 bits).
7594 The stack is required to be aligned on a 4 byte boundary. On Pentium
7595 and PentiumPro, @code{double} and @code{long double} values should be
7596 aligned to an 8 byte boundary (see @option{-malign-double}) or suffer
7597 significant run time performance penalties. On Pentium III, the
7598 Streaming SIMD Extension (SSE) data type @code{__m128} suffers similar
7599 penalties if it is not 16 byte aligned.
7601 To ensure proper alignment of this values on the stack, the stack boundary
7602 must be as aligned as that required by any value stored on the stack.
7603 Further, every function must be generated such that it keeps the stack
7604 aligned. Thus calling a function compiled with a higher preferred
7605 stack boundary from a function compiled with a lower preferred stack
7606 boundary will most likely misalign the stack. It is recommended that
7607 libraries that use callbacks always use the default setting.
7609 This extra alignment does consume extra stack space. Code that is sensitive
7610 to stack space usage, such as embedded systems and operating system kernels,
7611 may want to reduce the preferred alignment to
7612 @option{-mpreferred-stack-boundary=2}.
7615 @itemx -mno-push-args
7617 @opindex mno-push-args
7618 Use PUSH operations to store outgoing parameters. This method is shorter
7619 and usually equally fast as method using SUB/MOV operations and is enabled
7620 by default. In some cases disabling it may improve performance because of
7621 improved scheduling and reduced dependencies.
7623 @item -maccumulate-outgoing-args
7624 @opindex maccumulate-outgoing-args
7625 If enabled, the maximum amount of space required for outgoing arguments will be
7626 computed in the function prologue. This is faster on most modern CPUs
7627 because of reduced dependencies, improved scheduling and reduced stack usage
7628 when preferred stack boundary is not equal to 2. The drawback is a notable
7629 increase in code size. This switch implies @option{-mno-push-args}.
7633 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
7634 on thread-safe exception handling must compile and link all code with the
7635 @option{-mthreads} option. When compiling, @option{-mthreads} defines
7636 @option{-D_MT}; when linking, it links in a special thread helper library
7637 @option{-lmingwthrd} which cleans up per thread exception handling data.
7639 @item -mno-align-stringops
7640 @opindex mno-align-stringops
7641 Do not align destination of inlined string operations. This switch reduces
7642 code size and improves performance in case the destination is already aligned,
7643 but gcc don't know about it.
7645 @item -minline-all-stringops
7646 @opindex minline-all-stringops
7647 By default GCC inlines string operations only when destination is known to be
7648 aligned at least to 4 byte boundary. This enables more inlining, increase code
7649 size, but may improve performance of code that depends on fast memcpy, strlen
7650 and memset for short lengths.
7652 @item -momit-leaf-frame-pointer
7653 @opindex momit-leaf-frame-pointer
7654 Don't keep the frame pointer in a register for leaf functions. This
7655 avoids the instructions to save, set up and restore frame pointers and
7656 makes an extra register available in leaf functions. The option
7657 @option{-fomit-frame-pointer} removes the frame pointer for all functions
7658 which might make debugging harder.
7662 @subsection HPPA Options
7663 @cindex HPPA Options
7665 These @samp{-m} options are defined for the HPPA family of computers:
7668 @item -march=@var{architecture-type}
7670 Generate code for the specified architecture. The choices for
7671 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
7672 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
7673 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
7674 architecture option for your machine. Code compiled for lower numbered
7675 architectures will run on higher numbered architectures, but not the
7678 PA 2.0 support currently requires gas snapshot 19990413 or later. The
7679 next release of binutils (current is 2.9.1) will probably contain PA 2.0
7683 @itemx -mpa-risc-1-1
7684 @itemx -mpa-risc-2-0
7685 @opindex mpa-risc-1-0
7686 @opindex mpa-risc-1-1
7687 @opindex mpa-risc-2-0
7688 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
7691 @opindex mbig-switch
7692 Generate code suitable for big switch tables. Use this option only if
7693 the assembler/linker complain about out of range branches within a switch
7696 @item -mjump-in-delay
7697 @opindex mjump-in-delay
7698 Fill delay slots of function calls with unconditional jump instructions
7699 by modifying the return pointer for the function call to be the target
7700 of the conditional jump.
7702 @item -mdisable-fpregs
7703 @opindex mdisable-fpregs
7704 Prevent floating point registers from being used in any manner. This is
7705 necessary for compiling kernels which perform lazy context switching of
7706 floating point registers. If you use this option and attempt to perform
7707 floating point operations, the compiler will abort.
7709 @item -mdisable-indexing
7710 @opindex mdisable-indexing
7711 Prevent the compiler from using indexing address modes. This avoids some
7712 rather obscure problems when compiling MIG generated code under MACH@.
7714 @item -mno-space-regs
7715 @opindex mno-space-regs
7716 Generate code that assumes the target has no space registers. This allows
7717 GCC to generate faster indirect calls and use unscaled index address modes.
7719 Such code is suitable for level 0 PA systems and kernels.
7721 @item -mfast-indirect-calls
7722 @opindex mfast-indirect-calls
7723 Generate code that assumes calls never cross space boundaries. This
7724 allows GCC to emit code which performs faster indirect calls.
7726 This option will not work in the presence of shared libraries or nested
7729 @item -mlong-load-store
7730 @opindex mlong-load-store
7731 Generate 3-instruction load and store sequences as sometimes required by
7732 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
7735 @item -mportable-runtime
7736 @opindex mportable-runtime
7737 Use the portable calling conventions proposed by HP for ELF systems.
7741 Enable the use of assembler directives only GAS understands.
7743 @item -mschedule=@var{cpu-type}
7745 Schedule code according to the constraints for the machine type
7746 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
7747 @samp{7100}, @samp{7100LC}, @samp{7200}, and @samp{8000}. Refer to
7748 @file{/usr/lib/sched.models} on an HP-UX system to determine the
7749 proper scheduling option for your machine.
7752 @opindex mlinker-opt
7753 Enable the optimization pass in the HPUX linker. Note this makes symbolic
7754 debugging impossible. It also triggers a bug in the HPUX 8 and HPUX 9 linkers
7755 in which they give bogus error messages when linking some programs.
7758 @opindex msoft-float
7759 Generate output containing library calls for floating point.
7760 @strong{Warning:} the requisite libraries are not available for all HPPA
7761 targets. Normally the facilities of the machine's usual C compiler are
7762 used, but this cannot be done directly in cross-compilation. You must make
7763 your own arrangements to provide suitable library functions for
7764 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
7765 does provide software floating point support.
7767 @option{-msoft-float} changes the calling convention in the output file;
7768 therefore, it is only useful if you compile @emph{all} of a program with
7769 this option. In particular, you need to compile @file{libgcc.a}, the
7770 library that comes with GCC, with @option{-msoft-float} in order for
7774 @node Intel 960 Options
7775 @subsection Intel 960 Options
7777 These @samp{-m} options are defined for the Intel 960 implementations:
7780 @item -m@var{cpu-type}
7788 Assume the defaults for the machine type @var{cpu-type} for some of
7789 the other options, including instruction scheduling, floating point
7790 support, and addressing modes. The choices for @var{cpu-type} are
7791 @samp{ka}, @samp{kb}, @samp{mc}, @samp{ca}, @samp{cf},
7792 @samp{sa}, and @samp{sb}.
7799 @opindex msoft-float
7800 The @option{-mnumerics} option indicates that the processor does support
7801 floating-point instructions. The @option{-msoft-float} option indicates
7802 that floating-point support should not be assumed.
7804 @item -mleaf-procedures
7805 @itemx -mno-leaf-procedures
7806 @opindex mleaf-procedures
7807 @opindex mno-leaf-procedures
7808 Do (or do not) attempt to alter leaf procedures to be callable with the
7809 @code{bal} instruction as well as @code{call}. This will result in more
7810 efficient code for explicit calls when the @code{bal} instruction can be
7811 substituted by the assembler or linker, but less efficient code in other
7812 cases, such as calls via function pointers, or using a linker that doesn't
7813 support this optimization.
7816 @itemx -mno-tail-call
7818 @opindex mno-tail-call
7819 Do (or do not) make additional attempts (beyond those of the
7820 machine-independent portions of the compiler) to optimize tail-recursive
7821 calls into branches. You may not want to do this because the detection of
7822 cases where this is not valid is not totally complete. The default is
7823 @option{-mno-tail-call}.
7825 @item -mcomplex-addr
7826 @itemx -mno-complex-addr
7827 @opindex mcomplex-addr
7828 @opindex mno-complex-addr
7829 Assume (or do not assume) that the use of a complex addressing mode is a
7830 win on this implementation of the i960. Complex addressing modes may not
7831 be worthwhile on the K-series, but they definitely are on the C-series.
7832 The default is currently @option{-mcomplex-addr} for all processors except
7836 @itemx -mno-code-align
7837 @opindex mcode-align
7838 @opindex mno-code-align
7839 Align code to 8-byte boundaries for faster fetching (or don't bother).
7840 Currently turned on by default for C-series implementations only.
7843 @item -mclean-linkage
7844 @itemx -mno-clean-linkage
7845 @opindex mclean-linkage
7846 @opindex mno-clean-linkage
7847 These options are not fully implemented.
7851 @itemx -mic2.0-compat
7852 @itemx -mic3.0-compat
7854 @opindex mic2.0-compat
7855 @opindex mic3.0-compat
7856 Enable compatibility with iC960 v2.0 or v3.0.
7860 @opindex masm-compat
7862 Enable compatibility with the iC960 assembler.
7864 @item -mstrict-align
7865 @itemx -mno-strict-align
7866 @opindex mstrict-align
7867 @opindex mno-strict-align
7868 Do not permit (do permit) unaligned accesses.
7872 Enable structure-alignment compatibility with Intel's gcc release version
7873 1.3 (based on gcc 1.37). This option implies @option{-mstrict-align}.
7875 @item -mlong-double-64
7876 @opindex mlong-double-64
7877 Implement type @samp{long double} as 64-bit floating point numbers.
7878 Without the option @samp{long double} is implemented by 80-bit
7879 floating point numbers. The only reason we have it because there is
7880 no 128-bit @samp{long double} support in @samp{fp-bit.c} yet. So it
7881 is only useful for people using soft-float targets. Otherwise, we
7882 should recommend against use of it.
7886 @node DEC Alpha Options
7887 @subsection DEC Alpha Options
7889 These @samp{-m} options are defined for the DEC Alpha implementations:
7892 @item -mno-soft-float
7894 @opindex mno-soft-float
7895 @opindex msoft-float
7896 Use (do not use) the hardware floating-point instructions for
7897 floating-point operations. When @option{-msoft-float} is specified,
7898 functions in @file{libgcc.a} will be used to perform floating-point
7899 operations. Unless they are replaced by routines that emulate the
7900 floating-point operations, or compiled in such a way as to call such
7901 emulations routines, these routines will issue floating-point
7902 operations. If you are compiling for an Alpha without floating-point
7903 operations, you must ensure that the library is built so as not to call
7906 Note that Alpha implementations without floating-point operations are
7907 required to have floating-point registers.
7912 @opindex mno-fp-regs
7913 Generate code that uses (does not use) the floating-point register set.
7914 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
7915 register set is not used, floating point operands are passed in integer
7916 registers as if they were integers and floating-point results are passed
7917 in $0 instead of $f0. This is a non-standard calling sequence, so any
7918 function with a floating-point argument or return value called by code
7919 compiled with @option{-mno-fp-regs} must also be compiled with that
7922 A typical use of this option is building a kernel that does not use,
7923 and hence need not save and restore, any floating-point registers.
7927 The Alpha architecture implements floating-point hardware optimized for
7928 maximum performance. It is mostly compliant with the IEEE floating
7929 point standard. However, for full compliance, software assistance is
7930 required. This option generates code fully IEEE compliant code
7931 @emph{except} that the @var{inexact-flag} is not maintained (see below).
7932 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
7933 defined during compilation. The resulting code is less efficient but is
7934 able to correctly support denormalized numbers and exceptional IEEE
7935 values such as not-a-number and plus/minus infinity. Other Alpha
7936 compilers call this option @option{-ieee_with_no_inexact}.
7938 @item -mieee-with-inexact
7939 @opindex mieee-with-inexact
7940 This is like @option{-mieee} except the generated code also maintains
7941 the IEEE @var{inexact-flag}. Turning on this option causes the
7942 generated code to implement fully-compliant IEEE math. In addition to
7943 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
7944 macro. On some Alpha implementations the resulting code may execute
7945 significantly slower than the code generated by default. Since there is
7946 very little code that depends on the @var{inexact-flag}, you should
7947 normally not specify this option. Other Alpha compilers call this
7948 option @option{-ieee_with_inexact}.
7950 @item -mfp-trap-mode=@var{trap-mode}
7951 @opindex mfp-trap-mode
7952 This option controls what floating-point related traps are enabled.
7953 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
7954 The trap mode can be set to one of four values:
7958 This is the default (normal) setting. The only traps that are enabled
7959 are the ones that cannot be disabled in software (e.g., division by zero
7963 In addition to the traps enabled by @samp{n}, underflow traps are enabled
7967 Like @samp{su}, but the instructions are marked to be safe for software
7968 completion (see Alpha architecture manual for details).
7971 Like @samp{su}, but inexact traps are enabled as well.
7974 @item -mfp-rounding-mode=@var{rounding-mode}
7975 @opindex mfp-rounding-mode
7976 Selects the IEEE rounding mode. Other Alpha compilers call this option
7977 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
7982 Normal IEEE rounding mode. Floating point numbers are rounded towards
7983 the nearest machine number or towards the even machine number in case
7987 Round towards minus infinity.
7990 Chopped rounding mode. Floating point numbers are rounded towards zero.
7993 Dynamic rounding mode. A field in the floating point control register
7994 (@var{fpcr}, see Alpha architecture reference manual) controls the
7995 rounding mode in effect. The C library initializes this register for
7996 rounding towards plus infinity. Thus, unless your program modifies the
7997 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
8000 @item -mtrap-precision=@var{trap-precision}
8001 @opindex mtrap-precision
8002 In the Alpha architecture, floating point traps are imprecise. This
8003 means without software assistance it is impossible to recover from a
8004 floating trap and program execution normally needs to be terminated.
8005 GCC can generate code that can assist operating system trap handlers
8006 in determining the exact location that caused a floating point trap.
8007 Depending on the requirements of an application, different levels of
8008 precisions can be selected:
8012 Program precision. This option is the default and means a trap handler
8013 can only identify which program caused a floating point exception.
8016 Function precision. The trap handler can determine the function that
8017 caused a floating point exception.
8020 Instruction precision. The trap handler can determine the exact
8021 instruction that caused a floating point exception.
8024 Other Alpha compilers provide the equivalent options called
8025 @option{-scope_safe} and @option{-resumption_safe}.
8027 @item -mieee-conformant
8028 @opindex mieee-conformant
8029 This option marks the generated code as IEEE conformant. You must not
8030 use this option unless you also specify @option{-mtrap-precision=i} and either
8031 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
8032 is to emit the line @samp{.eflag 48} in the function prologue of the
8033 generated assembly file. Under DEC Unix, this has the effect that
8034 IEEE-conformant math library routines will be linked in.
8036 @item -mbuild-constants
8037 @opindex mbuild-constants
8038 Normally GCC examines a 32- or 64-bit integer constant to
8039 see if it can construct it from smaller constants in two or three
8040 instructions. If it cannot, it will output the constant as a literal and
8041 generate code to load it from the data segment at runtime.
8043 Use this option to require GCC to construct @emph{all} integer constants
8044 using code, even if it takes more instructions (the maximum is six).
8046 You would typically use this option to build a shared library dynamic
8047 loader. Itself a shared library, it must relocate itself in memory
8048 before it can find the variables and constants in its own data segment.
8054 Select whether to generate code to be assembled by the vendor-supplied
8055 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
8069 Indicate whether GCC should generate code to use the optional BWX,
8070 CIX, and MAX instruction sets. The default is to use the instruction sets
8071 supported by the CPU type specified via @option{-mcpu=} option or that
8072 of the CPU on which GCC was built if none was specified.
8074 @item -mcpu=@var{cpu_type}
8076 Set the instruction set, register set, and instruction scheduling
8077 parameters for machine type @var{cpu_type}. You can specify either the
8078 @samp{EV} style name or the corresponding chip number. GCC
8079 supports scheduling parameters for the EV4 and EV5 family of processors
8080 and will choose the default values for the instruction set from
8081 the processor you specify. If you do not specify a processor type,
8082 GCC will default to the processor on which the compiler was built.
8084 Supported values for @var{cpu_type} are
8089 Schedules as an EV4 and has no instruction set extensions.
8093 Schedules as an EV5 and has no instruction set extensions.
8097 Schedules as an EV5 and supports the BWX extension.
8102 Schedules as an EV5 and supports the BWX and MAX extensions.
8106 Schedules as an EV5 (until Digital releases the scheduling parameters
8107 for the EV6) and supports the BWX, CIX, and MAX extensions.
8110 @item -mmemory-latency=@var{time}
8111 @opindex mmemory-latency
8112 Sets the latency the scheduler should assume for typical memory
8113 references as seen by the application. This number is highly
8114 dependent on the memory access patterns used by the application
8115 and the size of the external cache on the machine.
8117 Valid options for @var{time} are
8121 A decimal number representing clock cycles.
8127 The compiler contains estimates of the number of clock cycles for
8128 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
8129 (also called Dcache, Scache, and Bcache), as well as to main memory.
8130 Note that L3 is only valid for EV5.
8135 @node Clipper Options
8136 @subsection Clipper Options
8138 These @samp{-m} options are defined for the Clipper implementations:
8143 Produce code for a C300 Clipper processor. This is the default.
8147 Produce code for a C400 Clipper processor, i.e.@: use floating point
8151 @node H8/300 Options
8152 @subsection H8/300 Options
8154 These @samp{-m} options are defined for the H8/300 implementations:
8159 Shorten some address references at link time, when possible; uses the
8160 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
8161 ld.info, Using ld}, for a fuller description.
8165 Generate code for the H8/300H@.
8169 Generate code for the H8/S@.
8173 Generate code for the H8/S2600. This switch must be used with @option{-ms}.
8177 Make @code{int} data 32 bits by default.
8181 On the H8/300H and H8/S, use the same alignment rules as for the H8/300.
8182 The default for the H8/300H and H8/S is to align longs and floats on 4
8184 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
8185 This option has no effect on the H8/300.
8189 @subsection SH Options
8191 These @samp{-m} options are defined for the SH implementations:
8196 Generate code for the SH1.
8200 Generate code for the SH2.
8204 Generate code for the SH3.
8208 Generate code for the SH3e.
8212 Generate code for the SH4 without a floating-point unit.
8214 @item -m4-single-only
8215 @opindex m4-single-only
8216 Generate code for the SH4 with a floating-point unit that only
8217 supports single-precision arithmetic.
8221 Generate code for the SH4 assuming the floating-point unit is in
8222 single-precision mode by default.
8226 Generate code for the SH4.
8230 Compile code for the processor in big endian mode.
8234 Compile code for the processor in little endian mode.
8238 Align doubles at 64-bit boundaries. Note that this changes the calling
8239 conventions, and thus some functions from the standard C library will
8240 not work unless you recompile it first with @option{-mdalign}.
8244 Shorten some address references at link time, when possible; uses the
8245 linker option @option{-relax}.
8249 Use 32-bit offsets in @code{switch} tables. The default is to use
8254 Enable the use of the instruction @code{fmovd}.
8258 Comply with the calling conventions defined by Hitachi.
8262 Mark the @code{MAC} register as call-clobbered, even if
8263 @option{-mhitachi} is given.
8267 Increase IEEE-compliance of floating-point code.
8271 Dump instruction size and location in the assembly code.
8275 This option is deprecated. It pads structures to multiple of 4 bytes,
8276 which is incompatible with the SH ABI@.
8280 Optimize for space instead of speed. Implied by @option{-Os}.
8284 When generating position-independent code, emit function calls using
8285 the Global Offset Table instead of the Procedure Linkage Table.
8289 Generate a library function call to invalidate instruction cache
8290 entries, after fixing up a trampoline. This library function call
8291 doesn't assume it can write to the whole memory address space. This
8292 is the default when the target is @code{sh-*-linux*}.
8295 @node System V Options
8296 @subsection Options for System V
8298 These additional options are available on System V Release 4 for
8299 compatibility with other compilers on those systems:
8304 Create a shared object.
8305 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
8309 Identify the versions of each tool used by the compiler, in a
8310 @code{.ident} assembler directive in the output.
8314 Refrain from adding @code{.ident} directives to the output file (this is
8317 @item -YP,@var{dirs}
8319 Search the directories @var{dirs}, and no others, for libraries
8320 specified with @option{-l}.
8324 Look in the directory @var{dir} to find the M4 preprocessor.
8325 The assembler uses this option.
8326 @c This is supposed to go with a -Yd for predefined M4 macro files, but
8327 @c the generic assembler that comes with Solaris takes just -Ym.
8330 @node TMS320C3x/C4x Options
8331 @subsection TMS320C3x/C4x Options
8332 @cindex TMS320C3x/C4x Options
8334 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
8338 @item -mcpu=@var{cpu_type}
8340 Set the instruction set, register set, and instruction scheduling
8341 parameters for machine type @var{cpu_type}. Supported values for
8342 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
8343 @samp{c44}. The default is @samp{c40} to generate code for the
8348 @itemx -msmall-memory
8350 @opindex mbig-memory
8352 @opindex msmall-memory
8354 Generates code for the big or small memory model. The small memory
8355 model assumed that all data fits into one 64K word page. At run-time
8356 the data page (DP) register must be set to point to the 64K page
8357 containing the .bss and .data program sections. The big memory model is
8358 the default and requires reloading of the DP register for every direct
8365 Allow (disallow) allocation of general integer operands into the block
8372 Enable (disable) generation of code using decrement and branch,
8373 DBcond(D), instructions. This is enabled by default for the C4x. To be
8374 on the safe side, this is disabled for the C3x, since the maximum
8375 iteration count on the C3x is @math{2^23 + 1} (but who iterates loops more than
8376 @math{2^23} times on the C3x?). Note that GCC will try to reverse a loop so
8377 that it can utilise the decrement and branch instruction, but will give
8378 up if there is more than one memory reference in the loop. Thus a loop
8379 where the loop counter is decremented can generate slightly more
8380 efficient code, in cases where the RPTB instruction cannot be utilised.
8382 @item -mdp-isr-reload
8384 @opindex mdp-isr-reload
8386 Force the DP register to be saved on entry to an interrupt service
8387 routine (ISR), reloaded to point to the data section, and restored on
8388 exit from the ISR@. This should not be required unless someone has
8389 violated the small memory model by modifying the DP register, say within
8396 For the C3x use the 24-bit MPYI instruction for integer multiplies
8397 instead of a library call to guarantee 32-bit results. Note that if one
8398 of the operands is a constant, then the multiplication will be performed
8399 using shifts and adds. If the @option{-mmpyi} option is not specified for the C3x,
8400 then squaring operations are performed inline instead of a library call.
8403 @itemx -mno-fast-fix
8405 @opindex mno-fast-fix
8406 The C3x/C4x FIX instruction to convert a floating point value to an
8407 integer value chooses the nearest integer less than or equal to the
8408 floating point value rather than to the nearest integer. Thus if the
8409 floating point number is negative, the result will be incorrectly
8410 truncated an additional code is necessary to detect and correct this
8411 case. This option can be used to disable generation of the additional
8412 code required to correct the result.
8418 Enable (disable) generation of repeat block sequences using the RPTB
8419 instruction for zero overhead looping. The RPTB construct is only used
8420 for innermost loops that do not call functions or jump across the loop
8421 boundaries. There is no advantage having nested RPTB loops due to the
8422 overhead required to save and restore the RC, RS, and RE registers.
8423 This is enabled by default with @option{-O2}.
8425 @item -mrpts=@var{count}
8429 Enable (disable) the use of the single instruction repeat instruction
8430 RPTS@. If a repeat block contains a single instruction, and the loop
8431 count can be guaranteed to be less than the value @var{count}, GCC will
8432 emit a RPTS instruction instead of a RPTB@. If no value is specified,
8433 then a RPTS will be emitted even if the loop count cannot be determined
8434 at compile time. Note that the repeated instruction following RPTS does
8435 not have to be reloaded from memory each iteration, thus freeing up the
8436 CPU buses for operands. However, since interrupts are blocked by this
8437 instruction, it is disabled by default.
8439 @item -mloop-unsigned
8440 @itemx -mno-loop-unsigned
8441 @opindex mloop-unsigned
8442 @opindex mno-loop-unsigned
8443 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
8444 is @math{2^31 + 1} since these instructions test if the iteration count is
8445 negative to terminate the loop. If the iteration count is unsigned
8446 there is a possibility than the @math{2^31 + 1} maximum iteration count may be
8447 exceeded. This switch allows an unsigned iteration count.
8451 Try to emit an assembler syntax that the TI assembler (asm30) is happy
8452 with. This also enforces compatibility with the API employed by the TI
8453 C3x C compiler. For example, long doubles are passed as structures
8454 rather than in floating point registers.
8460 Generate code that uses registers (stack) for passing arguments to functions.
8461 By default, arguments are passed in registers where possible rather
8462 than by pushing arguments on to the stack.
8464 @item -mparallel-insns
8465 @itemx -mno-parallel-insns
8466 @opindex mparallel-insns
8467 @opindex mno-parallel-insns
8468 Allow the generation of parallel instructions. This is enabled by
8469 default with @option{-O2}.
8471 @item -mparallel-mpy
8472 @itemx -mno-parallel-mpy
8473 @opindex mparallel-mpy
8474 @opindex mno-parallel-mpy
8475 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
8476 provided @option{-mparallel-insns} is also specified. These instructions have
8477 tight register constraints which can pessimize the code generation
8483 @subsection V850 Options
8484 @cindex V850 Options
8486 These @samp{-m} options are defined for V850 implementations:
8490 @itemx -mno-long-calls
8491 @opindex mlong-calls
8492 @opindex mno-long-calls
8493 Treat all calls as being far away (near). If calls are assumed to be
8494 far away, the compiler will always load the functions address up into a
8495 register, and call indirect through the pointer.
8501 Do not optimize (do optimize) basic blocks that use the same index
8502 pointer 4 or more times to copy pointer into the @code{ep} register, and
8503 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
8504 option is on by default if you optimize.
8506 @item -mno-prolog-function
8507 @itemx -mprolog-function
8508 @opindex mno-prolog-function
8509 @opindex mprolog-function
8510 Do not use (do use) external functions to save and restore registers at
8511 the prolog and epilog of a function. The external functions are slower,
8512 but use less code space if more than one function saves the same number
8513 of registers. The @option{-mprolog-function} option is on by default if
8518 Try to make the code as small as possible. At present, this just turns
8519 on the @option{-mep} and @option{-mprolog-function} options.
8523 Put static or global variables whose size is @var{n} bytes or less into
8524 the tiny data area that register @code{ep} points to. The tiny data
8525 area can hold up to 256 bytes in total (128 bytes for byte references).
8529 Put static or global variables whose size is @var{n} bytes or less into
8530 the small data area that register @code{gp} points to. The small data
8531 area can hold up to 64 kilobytes.
8535 Put static or global variables whose size is @var{n} bytes or less into
8536 the first 32 kilobytes of memory.
8540 Specify that the target processor is the V850.
8543 @opindex mbig-switch
8544 Generate code suitable for big switch tables. Use this option only if
8545 the assembler/linker complain about out of range branches within a switch
8550 @subsection ARC Options
8553 These options are defined for ARC implementations:
8558 Compile code for little endian mode. This is the default.
8562 Compile code for big endian mode.
8565 @opindex mmangle-cpu
8566 Prepend the name of the cpu to all public symbol names.
8567 In multiple-processor systems, there are many ARC variants with different
8568 instruction and register set characteristics. This flag prevents code
8569 compiled for one cpu to be linked with code compiled for another.
8570 No facility exists for handling variants that are ``almost identical''.
8571 This is an all or nothing option.
8573 @item -mcpu=@var{cpu}
8575 Compile code for ARC variant @var{cpu}.
8576 Which variants are supported depend on the configuration.
8577 All variants support @option{-mcpu=base}, this is the default.
8579 @item -mtext=@var{text-section}
8580 @itemx -mdata=@var{data-section}
8581 @itemx -mrodata=@var{readonly-data-section}
8585 Put functions, data, and readonly data in @var{text-section},
8586 @var{data-section}, and @var{readonly-data-section} respectively
8587 by default. This can be overridden with the @code{section} attribute.
8588 @xref{Variable Attributes}.
8593 @subsection NS32K Options
8594 @cindex NS32K options
8596 These are the @samp{-m} options defined for the 32000 series. The default
8597 values for these options depends on which style of 32000 was selected when
8598 the compiler was configured; the defaults for the most common choices are
8606 Generate output for a 32032. This is the default
8607 when the compiler is configured for 32032 and 32016 based systems.
8613 Generate output for a 32332. This is the default
8614 when the compiler is configured for 32332-based systems.
8620 Generate output for a 32532. This is the default
8621 when the compiler is configured for 32532-based systems.
8625 Generate output containing 32081 instructions for floating point.
8626 This is the default for all systems.
8630 Generate output containing 32381 instructions for floating point. This
8631 also implies @option{-m32081}. The 32381 is only compatible with the 32332
8632 and 32532 cpus. This is the default for the pc532-netbsd configuration.
8636 Try and generate multiply-add floating point instructions @code{polyF}
8637 and @code{dotF}. This option is only available if the @option{-m32381}
8638 option is in effect. Using these instructions requires changes to to
8639 register allocation which generally has a negative impact on
8640 performance. This option should only be enabled when compiling code
8641 particularly likely to make heavy use of multiply-add instructions.
8644 @opindex mnomulti-add
8645 Do not try and generate multiply-add floating point instructions
8646 @code{polyF} and @code{dotF}. This is the default on all platforms.
8649 @opindex msoft-float
8650 Generate output containing library calls for floating point.
8651 @strong{Warning:} the requisite libraries may not be available.
8654 @opindex mnobitfield
8655 Do not use the bit-field instructions. On some machines it is faster to
8656 use shifting and masking operations. This is the default for the pc532.
8660 Do use the bit-field instructions. This is the default for all platforms
8665 Use a different function-calling convention, in which functions
8666 that take a fixed number of arguments return pop their
8667 arguments on return with the @code{ret} instruction.
8669 This calling convention is incompatible with the one normally
8670 used on Unix, so you cannot use it if you need to call libraries
8671 compiled with the Unix compiler.
8673 Also, you must provide function prototypes for all functions that
8674 take variable numbers of arguments (including @code{printf});
8675 otherwise incorrect code will be generated for calls to those
8678 In addition, seriously incorrect code will result if you call a
8679 function with too many arguments. (Normally, extra arguments are
8680 harmlessly ignored.)
8682 This option takes its name from the 680x0 @code{rtd} instruction.
8687 Use a different function-calling convention where the first two arguments
8688 are passed in registers.
8690 This calling convention is incompatible with the one normally
8691 used on Unix, so you cannot use it if you need to call libraries
8692 compiled with the Unix compiler.
8695 @opindex mnoregparam
8696 Do not pass any arguments in registers. This is the default for all
8701 It is OK to use the sb as an index register which is always loaded with
8702 zero. This is the default for the pc532-netbsd target.
8706 The sb register is not available for use or has not been initialized to
8707 zero by the run time system. This is the default for all targets except
8708 the pc532-netbsd. It is also implied whenever @option{-mhimem} or
8709 @option{-fpic} is set.
8713 Many ns32000 series addressing modes use displacements of up to 512MB@.
8714 If an address is above 512MB then displacements from zero can not be used.
8715 This option causes code to be generated which can be loaded above 512MB@.
8716 This may be useful for operating systems or ROM code.
8720 Assume code will be loaded in the first 512MB of virtual address space.
8721 This is the default for all platforms.
8727 @subsection AVR Options
8730 These options are defined for AVR implementations:
8733 @item -mmcu=@var{mcu}
8735 Specify ATMEL AVR instruction set or MCU type.
8737 Instruction set avr1 is for the minimal AVR core, not supported by the C
8738 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
8739 attiny11, attiny12, attiny15, attiny28).
8741 Instruction set avr2 (default) is for the classic AVR core with up to
8742 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
8743 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
8744 at90c8534, at90s8535).
8746 Instruction set avr3 is for the classic AVR core with up to 128K program
8747 memory space (MCU types: atmega103, atmega603).
8749 Instruction set avr4 is for the enhanced AVR core with up to 8K program
8750 memory space (MCU types: atmega83, atmega85).
8752 Instruction set avr5 is for the enhanced AVR core with up to 128K program
8753 memory space (MCU types: atmega161, atmega163, atmega32, at94k).
8757 Output instruction sizes to the asm file.
8759 @item -minit-stack=@var{N}
8760 @opindex minit-stack
8761 Specify the initial stack address, which may be a symbol or numeric value,
8762 @samp{__stack} is the default.
8764 @item -mno-interrupts
8765 @opindex mno-interrupts
8766 Generated code is not compatible with hardware interrupts.
8767 Code size will be smaller.
8769 @item -mcall-prologues
8770 @opindex mcall-prologues
8771 Functions prologues/epilogues expanded as call to appropriate
8772 subroutines. Code size will be smaller.
8774 @item -mno-tablejump
8775 @opindex mno-tablejump
8776 Do not generate tablejump insns which sometimes increase code size.
8779 @opindex mtiny-stack
8780 Change only the low 8 bits of the stack pointer.
8784 @subsection MCore Options
8785 @cindex MCore options
8787 These are the @samp{-m} options defined for the Motorola M*Core
8797 @opindex mno-hardlit
8798 Inline constants into the code stream if it can be done in two
8799 instructions or less.
8807 Use the divide instruction. (Enabled by default).
8809 @item -mrelax-immediate
8810 @itemx -mrelax-immediate
8811 @itemx -mno-relax-immediate
8812 @opindex mrelax-immediate
8813 @opindex mrelax-immediate
8814 @opindex mno-relax-immediate
8815 Allow arbitrary sized immediates in bit operations.
8817 @item -mwide-bitfields
8818 @itemx -mwide-bitfields
8819 @itemx -mno-wide-bitfields
8820 @opindex mwide-bitfields
8821 @opindex mwide-bitfields
8822 @opindex mno-wide-bitfields
8823 Always treat bit-fields as int-sized.
8825 @item -m4byte-functions
8826 @itemx -m4byte-functions
8827 @itemx -mno-4byte-functions
8828 @opindex m4byte-functions
8829 @opindex m4byte-functions
8830 @opindex mno-4byte-functions
8831 Force all functions to be aligned to a four byte boundary.
8833 @item -mcallgraph-data
8834 @itemx -mcallgraph-data
8835 @itemx -mno-callgraph-data
8836 @opindex mcallgraph-data
8837 @opindex mcallgraph-data
8838 @opindex mno-callgraph-data
8839 Emit callgraph information.
8843 @itemx -mno-slow-bytes
8844 @opindex mslow-bytes
8845 @opindex mslow-bytes
8846 @opindex mno-slow-bytes
8847 Prefer word access when reading byte quantities.
8849 @item -mlittle-endian
8850 @itemx -mlittle-endian
8852 @opindex mlittle-endian
8853 @opindex mlittle-endian
8854 @opindex mbig-endian
8855 Generate code for a little endian target.
8863 Generate code for the 210 processor.
8867 @subsection IA-64 Options
8868 @cindex IA-64 Options
8870 These are the @samp{-m} options defined for the Intel IA-64 architecture.
8874 @opindex mbig-endian
8875 Generate code for a big endian target. This is the default for HPUX@.
8877 @item -mlittle-endian
8878 @opindex mlittle-endian
8879 Generate code for a little endian target. This is the default for AIX5
8886 Generate (or don't) code for the GNU assembler. This is the default.
8887 @c Also, this is the default if the configure option @option{--with-gnu-as}
8894 Generate (or don't) code for the GNU linker. This is the default.
8895 @c Also, this is the default if the configure option @option{--with-gnu-ld}
8900 Generate code that does not use a global pointer register. The result
8901 is not position independent code, and violates the IA-64 ABI@.
8903 @item -mvolatile-asm-stop
8904 @itemx -mno-volatile-asm-stop
8905 @opindex mvolatile-asm-stop
8906 @opindex mno-volatile-asm-stop
8907 Generate (or don't) a stop bit immediately before and after volatile asm
8912 Generate code that works around Itanium B step errata.
8914 @item -mregister-names
8915 @itemx -mno-register-names
8916 @opindex mregister-names
8917 @opindex mno-register-names
8918 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
8919 the stacked registers. This may make assembler output more readable.
8925 Disable (or enable) optimizations that use the small data section. This may
8926 be useful for working around optimizer bugs.
8929 @opindex mconstant-gp
8930 Generate code that uses a single constant global pointer value. This is
8931 useful when compiling kernel code.
8935 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
8936 This is useful when compiling firmware code.
8938 @item -minline-divide-min-latency
8939 @opindex minline-divide-min-latency
8940 Generate code for inline divides using the minimum latency algorithm.
8942 @item -minline-divide-max-throughput
8943 @opindex minline-divide-max-throughput
8944 Generate code for inline divides using the maximum throughput algorithm.
8946 @item -mno-dwarf2-asm
8948 @opindex mno-dwarf2-asm
8949 @opindex mdwarf2-asm
8950 Don't (or do) generate assembler code for the DWARF2 line number debugging
8951 info. This may be useful when not using the GNU assembler.
8953 @item -mfixed-range=@var{register-range}
8954 @opindex mfixed-range
8955 Generate code treating the given register range as fixed registers.
8956 A fixed register is one that the register allocator can not use. This is
8957 useful when compiling kernel code. A register range is specified as
8958 two registers separated by a dash. Multiple register ranges can be
8959 specified separated by a comma.
8963 @subsection D30V Options
8964 @cindex D30V Options
8966 These @samp{-m} options are defined for D30V implementations:
8971 Link the @samp{.text}, @samp{.data}, @samp{.bss}, @samp{.strings},
8972 @samp{.rodata}, @samp{.rodata1}, @samp{.data1} sections into external
8973 memory, which starts at location @code{0x80000000}.
8977 Same as the @option{-mextmem} switch.
8981 Link the @samp{.text} section into onchip text memory, which starts at
8982 location @code{0x0}. Also link @samp{.data}, @samp{.bss},
8983 @samp{.strings}, @samp{.rodata}, @samp{.rodata1}, @samp{.data1} sections
8984 into onchip data memory, which starts at location @code{0x20000000}.
8986 @item -mno-asm-optimize
8987 @itemx -masm-optimize
8988 @opindex mno-asm-optimize
8989 @opindex masm-optimize
8990 Disable (enable) passing @option{-O} to the assembler when optimizing.
8991 The assembler uses the @option{-O} option to automatically parallelize
8992 adjacent short instructions where possible.
8994 @item -mbranch-cost=@var{n}
8995 @opindex mbranch-cost
8996 Increase the internal costs of branches to @var{n}. Higher costs means
8997 that the compiler will issue more instructions to avoid doing a branch.
9000 @item -mcond-exec=@var{n}
9002 Specify the maximum number of conditionally executed instructions that
9003 replace a branch. The default is 4.
9006 @node S/390 and zSeries Options
9007 @subsection S/390 and zSeries Options
9008 @cindex S/390 and zSeries Options
9010 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
9015 @opindex mhard-float
9016 @opindex msoft-float
9017 Use (do not use) the hardware floating-point instructions and registers
9018 for floating-point operations. When @option{-msoft-float} is specified,
9019 functions in @file{libgcc.a} will be used to perform floating-point
9020 operations. When @option{-mhard-float} is specified, the compiler
9021 generates IEEE floating-point instructions. This is the default.
9024 @itemx -mno-backchain
9026 @opindex mno-backchain
9027 Generate (or do not generate) code which maintains an explicit
9028 backchain within the stack frame that points to the caller's frame.
9029 This is currently needed to allow debugging. The default is to
9030 generate the backchain.
9033 @itemx -mno-small-exec
9034 @opindex msmall-exec
9035 @opindex mno-small-exec
9036 Generate (or do not generate) code using the @code{bras} instruction
9037 to do subroutine calls.
9038 This only works reliably if the total executable size does not
9039 exceed 64k. The default is to use the @code{basr} instruction instead,
9040 which does not have this limitation.
9046 When @option{-m31} is specified, generate code compliant to the
9047 Linux for S/390 ABI@. When @option{-m64} is specified, generate
9048 code compliant to the Linux for zSeries ABI@. This allows GCC in
9049 particular to generate 64-bit instructions. For the @samp{s390}
9050 targets, the default is @option{-m31}, while the @samp{s390x}
9051 targets default to @option{-m64}.
9057 Generate (or do not generate) code using the @code{mvcle} instruction
9058 to perform block moves. When @option{-mno-mvcle} is specifed,
9059 use a @code{mvc} loop instead. This is the default.
9065 Print (or do not print) additional debug information when compiling.
9066 The default is to not print debug information.
9071 @node Code Gen Options
9072 @section Options for Code Generation Conventions
9073 @cindex code generation conventions
9074 @cindex options, code generation
9075 @cindex run-time options
9077 These machine-independent options control the interface conventions
9078 used in code generation.
9080 Most of them have both positive and negative forms; the negative form
9081 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
9082 one of the forms is listed---the one which is not the default. You
9083 can figure out the other form by either removing @samp{no-} or adding
9088 @opindex fexceptions
9089 Enable exception handling. Generates extra code needed to propagate
9090 exceptions. For some targets, this implies GCC will generate frame
9091 unwind information for all functions, which can produce significant data
9092 size overhead, although it does not affect execution. If you do not
9093 specify this option, GCC will enable it by default for languages like
9094 C++ which normally require exception handling, and disable it for
9095 languages like C that do not normally require it. However, you may need
9096 to enable this option when compiling C code that needs to interoperate
9097 properly with exception handlers written in C++. You may also wish to
9098 disable this option if you are compiling older C++ programs that don't
9099 use exception handling.
9101 @item -fnon-call-exceptions
9102 @opindex fnon-call-exceptions
9103 Generate code that allows trapping instructions to throw exceptions.
9104 Note that this requires platform-specific runtime support that does
9105 not exist everywhere. Moreover, it only allows @emph{trapping}
9106 instructions to throw exceptions, i.e.@: memory references or floating
9107 point instructions. It does not allow exceptions to be thrown from
9108 arbitrary signal handlers such as @code{SIGALRM}.
9110 @item -funwind-tables
9111 @opindex funwind-tables
9112 Similar to @option{-fexceptions}, except that it will just generate any needed
9113 static data, but will not affect the generated code in any other way.
9114 You will normally not enable this option; instead, a language processor
9115 that needs this handling would enable it on your behalf.
9117 @item -fpcc-struct-return
9118 @opindex fpcc-struct-return
9119 Return ``short'' @code{struct} and @code{union} values in memory like
9120 longer ones, rather than in registers. This convention is less
9121 efficient, but it has the advantage of allowing intercallability between
9122 GCC-compiled files and files compiled with other compilers.
9124 The precise convention for returning structures in memory depends
9125 on the target configuration macros.
9127 Short structures and unions are those whose size and alignment match
9128 that of some integer type.
9130 @item -freg-struct-return
9131 @opindex freg-struct-return
9132 Return @code{struct} and @code{union} values in registers when possible.
9133 This is more efficient for small structures than
9134 @option{-fpcc-struct-return}.
9136 If you specify neither @option{-fpcc-struct-return} nor
9137 @option{-freg-struct-return}, GCC defaults to whichever convention is
9138 standard for the target. If there is no standard convention, GCC
9139 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
9140 the principal compiler. In those cases, we can choose the standard, and
9141 we chose the more efficient register return alternative.
9144 @opindex fshort-enums
9145 Allocate to an @code{enum} type only as many bytes as it needs for the
9146 declared range of possible values. Specifically, the @code{enum} type
9147 will be equivalent to the smallest integer type which has enough room.
9149 @item -fshort-double
9150 @opindex fshort-double
9151 Use the same size for @code{double} as for @code{float}.
9154 @opindex fshared-data
9155 Requests that the data and non-@code{const} variables of this
9156 compilation be shared data rather than private data. The distinction
9157 makes sense only on certain operating systems, where shared data is
9158 shared between processes running the same program, while private data
9159 exists in one copy per process.
9163 In C, allocate even uninitialized global variables in the data section of the
9164 object file, rather than generating them as common blocks. This has the
9165 effect that if the same variable is declared (without @code{extern}) in
9166 two different compilations, you will get an error when you link them.
9167 The only reason this might be useful is if you wish to verify that the
9168 program will work on other systems which always work this way.
9172 Ignore the @samp{#ident} directive.
9174 @item -fno-gnu-linker
9175 @opindex fno-gnu-linker
9176 Do not output global initializations (such as C++ constructors and
9177 destructors) in the form used by the GNU linker (on systems where the GNU
9178 linker is the standard method of handling them). Use this option when
9179 you want to use a non-GNU linker, which also requires using the
9180 @command{collect2} program to make sure the system linker includes
9181 constructors and destructors. (@command{collect2} is included in the GCC
9182 distribution.) For systems which @emph{must} use @command{collect2}, the
9183 compiler driver @command{gcc} is configured to do this automatically.
9185 @item -finhibit-size-directive
9186 @opindex finhibit-size-directive
9187 Don't output a @code{.size} assembler directive, or anything else that
9188 would cause trouble if the function is split in the middle, and the
9189 two halves are placed at locations far apart in memory. This option is
9190 used when compiling @file{crtstuff.c}; you should not need to use it
9194 @opindex fverbose-asm
9195 Put extra commentary information in the generated assembly code to
9196 make it more readable. This option is generally only of use to those
9197 who actually need to read the generated assembly code (perhaps while
9198 debugging the compiler itself).
9200 @option{-fno-verbose-asm}, the default, causes the
9201 extra information to be omitted and is useful when comparing two assembler
9206 Consider all memory references through pointers to be volatile.
9208 @item -fvolatile-global
9209 @opindex fvolatile-global
9210 Consider all memory references to extern and global data items to
9211 be volatile. GCC does not consider static data items to be volatile
9212 because of this switch.
9214 @item -fvolatile-static
9215 @opindex fvolatile-static
9216 Consider all memory references to static data to be volatile.
9220 @cindex global offset table
9222 Generate position-independent code (PIC) suitable for use in a shared
9223 library, if supported for the target machine. Such code accesses all
9224 constant addresses through a global offset table (GOT)@. The dynamic
9225 loader resolves the GOT entries when the program starts (the dynamic
9226 loader is not part of GCC; it is part of the operating system). If
9227 the GOT size for the linked executable exceeds a machine-specific
9228 maximum size, you get an error message from the linker indicating that
9229 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
9230 instead. (These maximums are 16k on the m88k, 8k on the Sparc, and 32k
9231 on the m68k and RS/6000. The 386 has no such limit.)
9233 Position-independent code requires special support, and therefore works
9234 only on certain machines. For the 386, GCC supports PIC for System V
9235 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
9236 position-independent.
9240 If supported for the target machine, emit position-independent code,
9241 suitable for dynamic linking and avoiding any limit on the size of the
9242 global offset table. This option makes a difference on the m68k, m88k,
9245 Position-independent code requires special support, and therefore works
9246 only on certain machines.
9248 @item -ffixed-@var{reg}
9250 Treat the register named @var{reg} as a fixed register; generated code
9251 should never refer to it (except perhaps as a stack pointer, frame
9252 pointer or in some other fixed role).
9254 @var{reg} must be the name of a register. The register names accepted
9255 are machine-specific and are defined in the @code{REGISTER_NAMES}
9256 macro in the machine description macro file.
9258 This flag does not have a negative form, because it specifies a
9261 @item -fcall-used-@var{reg}
9263 Treat the register named @var{reg} as an allocable register that is
9264 clobbered by function calls. It may be allocated for temporaries or
9265 variables that do not live across a call. Functions compiled this way
9266 will not save and restore the register @var{reg}.
9268 It is an error to used this flag with the frame pointer or stack pointer.
9269 Use of this flag for other registers that have fixed pervasive roles in
9270 the machine's execution model will produce disastrous results.
9272 This flag does not have a negative form, because it specifies a
9275 @item -fcall-saved-@var{reg}
9276 @opindex fcall-saved
9277 Treat the register named @var{reg} as an allocable register saved by
9278 functions. It may be allocated even for temporaries or variables that
9279 live across a call. Functions compiled this way will save and restore
9280 the register @var{reg} if they use it.
9282 It is an error to used this flag with the frame pointer or stack pointer.
9283 Use of this flag for other registers that have fixed pervasive roles in
9284 the machine's execution model will produce disastrous results.
9286 A different sort of disaster will result from the use of this flag for
9287 a register in which function values may be returned.
9289 This flag does not have a negative form, because it specifies a
9293 @opindex fpack-struct
9294 Pack all structure members together without holes. Usually you would
9295 not want to use this option, since it makes the code suboptimal, and
9296 the offsets of structure members won't agree with system libraries.
9298 @item -fcheck-memory-usage
9299 @opindex fcheck-memory-usage
9300 Generate extra code to check each memory access. GCC will generate
9301 code that is suitable for a detector of bad memory accesses such as
9304 Normally, you should compile all, or none, of your code with this option.
9306 If you do mix code compiled with and without this option,
9307 you must ensure that all code that has side effects
9308 and that is called by code compiled with this option
9309 is, itself, compiled with this option.
9310 If you do not, you might get erroneous messages from the detector.
9312 If you use functions from a library that have side-effects (such as
9313 @code{read}), you might not be able to recompile the library and
9314 specify this option. In that case, you can enable the
9315 @option{-fprefix-function-name} option, which requests GCC to encapsulate
9316 your code and make other functions look as if they were compiled with
9317 @option{-fcheck-memory-usage}. This is done by calling ``stubs'',
9318 which are provided by the detector. If you cannot find or build
9319 stubs for every function you call, you might have to specify
9320 @option{-fcheck-memory-usage} without @option{-fprefix-function-name}.
9322 If you specify this option, you can not use the @code{asm} or
9323 @code{__asm__} keywords in functions with memory checking enabled. GCC
9324 cannot understand what the @code{asm} statement may do, and therefore
9325 cannot generate the appropriate code, so it will reject it. However, if
9326 you specify the function attribute @code{no_check_memory_usage}
9327 (@pxref{Function Attributes}), GCC will disable memory checking within a
9328 function; you may use @code{asm} statements inside such functions. You
9329 may have an inline expansion of a non-checked function within a checked
9330 function; in that case GCC will not generate checks for the inlined
9331 function's memory accesses.
9333 If you move your @code{asm} statements to non-checked inline functions
9334 and they do access memory, you can add calls to the support code in your
9335 inline function, to indicate any reads, writes, or copies being done.
9336 These calls would be similar to those done in the stubs described above.
9338 @item -fprefix-function-name
9339 @opindex fprefix-function-name
9340 Request GCC to add a prefix to the symbols generated for function names.
9341 GCC adds a prefix to the names of functions defined as well as
9342 functions called. Code compiled with this option and code compiled
9343 without the option can't be linked together, unless stubs are used.
9345 If you compile the following code with @option{-fprefix-function-name}
9347 extern void bar (int);
9356 GCC will compile the code as if it was written:
9358 extern void prefix_bar (int);
9362 return prefix_bar (a + 5);
9365 This option is designed to be used with @option{-fcheck-memory-usage}.
9367 @item -finstrument-functions
9368 @opindex finstrument-functions
9369 Generate instrumentation calls for entry and exit to functions. Just
9370 after function entry and just before function exit, the following
9371 profiling functions will be called with the address of the current
9372 function and its call site. (On some platforms,
9373 @code{__builtin_return_address} does not work beyond the current
9374 function, so the call site information may not be available to the
9375 profiling functions otherwise.)
9378 void __cyg_profile_func_enter (void *this_fn,
9380 void __cyg_profile_func_exit (void *this_fn,
9384 The first argument is the address of the start of the current function,
9385 which may be looked up exactly in the symbol table.
9387 This instrumentation is also done for functions expanded inline in other
9388 functions. The profiling calls will indicate where, conceptually, the
9389 inline function is entered and exited. This means that addressable
9390 versions of such functions must be available. If all your uses of a
9391 function are expanded inline, this may mean an additional expansion of
9392 code size. If you use @samp{extern inline} in your C code, an
9393 addressable version of such functions must be provided. (This is
9394 normally the case anyways, but if you get lucky and the optimizer always
9395 expands the functions inline, you might have gotten away without
9396 providing static copies.)
9398 A function may be given the attribute @code{no_instrument_function}, in
9399 which case this instrumentation will not be done. This can be used, for
9400 example, for the profiling functions listed above, high-priority
9401 interrupt routines, and any functions from which the profiling functions
9402 cannot safely be called (perhaps signal handlers, if the profiling
9403 routines generate output or allocate memory).
9406 @opindex fstack-check
9407 Generate code to verify that you do not go beyond the boundary of the
9408 stack. You should specify this flag if you are running in an
9409 environment with multiple threads, but only rarely need to specify it in
9410 a single-threaded environment since stack overflow is automatically
9411 detected on nearly all systems if there is only one stack.
9413 Note that this switch does not actually cause checking to be done; the
9414 operating system must do that. The switch causes generation of code
9415 to ensure that the operating system sees the stack being extended.
9417 @item -fstack-limit-register=@var{reg}
9418 @itemx -fstack-limit-symbol=@var{sym}
9419 @itemx -fno-stack-limit
9420 @opindex fstack-limit-register
9421 @opindex fstack-limit-symbol
9422 @opindex fno-stack-limit
9423 Generate code to ensure that the stack does not grow beyond a certain value,
9424 either the value of a register or the address of a symbol. If the stack
9425 would grow beyond the value, a signal is raised. For most targets,
9426 the signal is raised before the stack overruns the boundary, so
9427 it is possible to catch the signal without taking special precautions.
9429 For instance, if the stack starts at absolute address @samp{0x80000000}
9430 and grows downwards, you can use the flags
9431 @option{-fstack-limit-symbol=__stack_limit} and
9432 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
9433 of 128KB@. Note that this may only work with the GNU linker.
9435 @cindex aliasing of parameters
9436 @cindex parameters, aliased
9437 @item -fargument-alias
9438 @itemx -fargument-noalias
9439 @itemx -fargument-noalias-global
9440 @opindex fargument-alias
9441 @opindex fargument-noalias
9442 @opindex fargument-noalias-global
9443 Specify the possible relationships among parameters and between
9444 parameters and global data.
9446 @option{-fargument-alias} specifies that arguments (parameters) may
9447 alias each other and may alias global storage.@*
9448 @option{-fargument-noalias} specifies that arguments do not alias
9449 each other, but may alias global storage.@*
9450 @option{-fargument-noalias-global} specifies that arguments do not
9451 alias each other and do not alias global storage.
9453 Each language will automatically use whatever option is required by
9454 the language standard. You should not need to use these options yourself.
9456 @item -fleading-underscore
9457 @opindex fleading-underscore
9458 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
9459 change the way C symbols are represented in the object file. One use
9460 is to help link with legacy assembly code.
9462 Be warned that you should know what you are doing when invoking this
9463 option, and that not all targets provide complete support for it.
9468 @node Environment Variables
9469 @section Environment Variables Affecting GCC
9470 @cindex environment variables
9472 @c man begin ENVIRONMENT
9474 This section describes several environment variables that affect how GCC
9475 operates. Some of them work by specifying directories or prefixes to use
9476 when searching for various kinds of files. Some are used to specify other
9477 aspects of the compilation environment.
9480 Note that you can also specify places to search using options such as
9481 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
9482 take precedence over places specified using environment variables, which
9483 in turn take precedence over those specified by the configuration of GCC@.
9487 Note that you can also specify places to search using options such as
9488 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
9489 take precedence over places specified using environment variables, which
9490 in turn take precedence over those specified by the configuration of GCC@.
9497 @c @itemx LC_COLLATE
9499 @c @itemx LC_MONETARY
9500 @c @itemx LC_NUMERIC
9505 @c @findex LC_COLLATE
9507 @c @findex LC_MONETARY
9508 @c @findex LC_NUMERIC
9512 These environment variables control the way that GCC uses
9513 localization information that allow GCC to work with different
9514 national conventions. GCC inspects the locale categories
9515 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
9516 so. These locale categories can be set to any value supported by your
9517 installation. A typical value is @samp{en_UK} for English in the United
9520 The @env{LC_CTYPE} environment variable specifies character
9521 classification. GCC uses it to determine the character boundaries in
9522 a string; this is needed for some multibyte encodings that contain quote
9523 and escape characters that would otherwise be interpreted as a string
9526 The @env{LC_MESSAGES} environment variable specifies the language to
9527 use in diagnostic messages.
9529 If the @env{LC_ALL} environment variable is set, it overrides the value
9530 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
9531 and @env{LC_MESSAGES} default to the value of the @env{LANG}
9532 environment variable. If none of these variables are set, GCC
9533 defaults to traditional C English behavior.
9537 If @env{TMPDIR} is set, it specifies the directory to use for temporary
9538 files. GCC uses temporary files to hold the output of one stage of
9539 compilation which is to be used as input to the next stage: for example,
9540 the output of the preprocessor, which is the input to the compiler
9543 @item GCC_EXEC_PREFIX
9544 @findex GCC_EXEC_PREFIX
9545 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
9546 names of the subprograms executed by the compiler. No slash is added
9547 when this prefix is combined with the name of a subprogram, but you can
9548 specify a prefix that ends with a slash if you wish.
9550 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
9551 an appropriate prefix to use based on the pathname it was invoked with.
9553 If GCC cannot find the subprogram using the specified prefix, it
9554 tries looking in the usual places for the subprogram.
9556 The default value of @env{GCC_EXEC_PREFIX} is
9557 @file{@var{prefix}/lib/gcc-lib/} where @var{prefix} is the value
9558 of @code{prefix} when you ran the @file{configure} script.
9560 Other prefixes specified with @option{-B} take precedence over this prefix.
9562 This prefix is also used for finding files such as @file{crt0.o} that are
9565 In addition, the prefix is used in an unusual way in finding the
9566 directories to search for header files. For each of the standard
9567 directories whose name normally begins with @samp{/usr/local/lib/gcc-lib}
9568 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
9569 replacing that beginning with the specified prefix to produce an
9570 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
9571 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
9572 These alternate directories are searched first; the standard directories
9576 @findex COMPILER_PATH
9577 The value of @env{COMPILER_PATH} is a colon-separated list of
9578 directories, much like @env{PATH}. GCC tries the directories thus
9579 specified when searching for subprograms, if it can't find the
9580 subprograms using @env{GCC_EXEC_PREFIX}.
9583 @findex LIBRARY_PATH
9584 The value of @env{LIBRARY_PATH} is a colon-separated list of
9585 directories, much like @env{PATH}. When configured as a native compiler,
9586 GCC tries the directories thus specified when searching for special
9587 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
9588 using GCC also uses these directories when searching for ordinary
9589 libraries for the @option{-l} option (but directories specified with
9590 @option{-L} come first).
9592 @item C_INCLUDE_PATH
9593 @itemx CPLUS_INCLUDE_PATH
9594 @itemx OBJC_INCLUDE_PATH
9595 @findex C_INCLUDE_PATH
9596 @findex CPLUS_INCLUDE_PATH
9597 @findex OBJC_INCLUDE_PATH
9598 @c @itemx OBJCPLUS_INCLUDE_PATH
9599 These environment variables pertain to particular languages. Each
9600 variable's value is a colon-separated list of directories, much like
9601 @env{PATH}. When GCC searches for header files, it tries the
9602 directories listed in the variable for the language you are using, after
9603 the directories specified with @option{-I} but before the standard header
9606 @item DEPENDENCIES_OUTPUT
9607 @findex DEPENDENCIES_OUTPUT
9608 @cindex dependencies for make as output
9609 If this variable is set, its value specifies how to output dependencies
9610 for Make based on the header files processed by the compiler. This
9611 output looks much like the output from the @option{-M} option
9612 (@pxref{Preprocessor Options}), but it goes to a separate file, and is
9613 in addition to the usual results of compilation.
9615 The value of @env{DEPENDENCIES_OUTPUT} can be just a file name, in
9616 which case the Make rules are written to that file, guessing the target
9617 name from the source file name. Or the value can have the form
9618 @samp{@var{file} @var{target}}, in which case the rules are written to
9619 file @var{file} using @var{target} as the target name.
9623 @cindex locale definition
9624 This variable is used to pass locale information to the compiler. One way in
9625 which this information is used is to determine the character set to be used
9626 when character literals, string literals and comments are parsed in C and C++.
9627 When the compiler is configured to allow multibyte characters,
9628 the following values for @env{LANG} are recognized:
9632 Recognize JIS characters.
9634 Recognize SJIS characters.
9636 Recognize EUCJP characters.
9639 If @env{LANG} is not defined, or if it has some other value, then the
9640 compiler will use mblen and mbtowc as defined by the default locale to
9641 recognize and translate multibyte characters.
9646 @node Running Protoize
9647 @section Running Protoize
9649 The program @code{protoize} is an optional part of GCC@. You can use
9650 it to add prototypes to a program, thus converting the program to ISO
9651 C in one respect. The companion program @code{unprotoize} does the
9652 reverse: it removes argument types from any prototypes that are found.
9654 When you run these programs, you must specify a set of source files as
9655 command line arguments. The conversion programs start out by compiling
9656 these files to see what functions they define. The information gathered
9657 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
9659 After scanning comes actual conversion. The specified files are all
9660 eligible to be converted; any files they include (whether sources or
9661 just headers) are eligible as well.
9663 But not all the eligible files are converted. By default,
9664 @code{protoize} and @code{unprotoize} convert only source and header
9665 files in the current directory. You can specify additional directories
9666 whose files should be converted with the @option{-d @var{directory}}
9667 option. You can also specify particular files to exclude with the
9668 @option{-x @var{file}} option. A file is converted if it is eligible, its
9669 directory name matches one of the specified directory names, and its
9670 name within the directory has not been excluded.
9672 Basic conversion with @code{protoize} consists of rewriting most
9673 function definitions and function declarations to specify the types of
9674 the arguments. The only ones not rewritten are those for varargs
9677 @code{protoize} optionally inserts prototype declarations at the
9678 beginning of the source file, to make them available for any calls that
9679 precede the function's definition. Or it can insert prototype
9680 declarations with block scope in the blocks where undeclared functions
9683 Basic conversion with @code{unprotoize} consists of rewriting most
9684 function declarations to remove any argument types, and rewriting
9685 function definitions to the old-style pre-ISO form.
9687 Both conversion programs print a warning for any function declaration or
9688 definition that they can't convert. You can suppress these warnings
9691 The output from @code{protoize} or @code{unprotoize} replaces the
9692 original source file. The original file is renamed to a name ending
9693 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
9694 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
9695 for DOS) file already exists, then the source file is simply discarded.
9697 @code{protoize} and @code{unprotoize} both depend on GCC itself to
9698 scan the program and collect information about the functions it uses.
9699 So neither of these programs will work until GCC is installed.
9701 Here is a table of the options you can use with @code{protoize} and
9702 @code{unprotoize}. Each option works with both programs unless
9706 @item -B @var{directory}
9707 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
9708 usual directory (normally @file{/usr/local/lib}). This file contains
9709 prototype information about standard system functions. This option
9710 applies only to @code{protoize}.
9712 @item -c @var{compilation-options}
9713 Use @var{compilation-options} as the options when running @code{gcc} to
9714 produce the @samp{.X} files. The special option @option{-aux-info} is
9715 always passed in addition, to tell @code{gcc} to write a @samp{.X} file.
9717 Note that the compilation options must be given as a single argument to
9718 @code{protoize} or @code{unprotoize}. If you want to specify several
9719 @code{gcc} options, you must quote the entire set of compilation options
9720 to make them a single word in the shell.
9722 There are certain @code{gcc} arguments that you cannot use, because they
9723 would produce the wrong kind of output. These include @option{-g},
9724 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
9725 the @var{compilation-options}, they are ignored.
9728 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
9729 systems) instead of @samp{.c}. This is convenient if you are converting
9730 a C program to C++. This option applies only to @code{protoize}.
9733 Add explicit global declarations. This means inserting explicit
9734 declarations at the beginning of each source file for each function
9735 that is called in the file and was not declared. These declarations
9736 precede the first function definition that contains a call to an
9737 undeclared function. This option applies only to @code{protoize}.
9739 @item -i @var{string}
9740 Indent old-style parameter declarations with the string @var{string}.
9741 This option applies only to @code{protoize}.
9743 @code{unprotoize} converts prototyped function definitions to old-style
9744 function definitions, where the arguments are declared between the
9745 argument list and the initial @samp{@{}. By default, @code{unprotoize}
9746 uses five spaces as the indentation. If you want to indent with just
9747 one space instead, use @option{-i " "}.
9750 Keep the @samp{.X} files. Normally, they are deleted after conversion
9754 Add explicit local declarations. @code{protoize} with @option{-l} inserts
9755 a prototype declaration for each function in each block which calls the
9756 function without any declaration. This option applies only to
9760 Make no real changes. This mode just prints information about the conversions
9761 that would have been done without @option{-n}.
9764 Make no @samp{.save} files. The original files are simply deleted.
9765 Use this option with caution.
9767 @item -p @var{program}
9768 Use the program @var{program} as the compiler. Normally, the name
9772 Work quietly. Most warnings are suppressed.
9775 Print the version number, just like @option{-v} for @code{gcc}.
9778 If you need special compiler options to compile one of your program's
9779 source files, then you should generate that file's @samp{.X} file
9780 specially, by running @code{gcc} on that source file with the
9781 appropriate options and the option @option{-aux-info}. Then run
9782 @code{protoize} on the entire set of files. @code{protoize} will use
9783 the existing @samp{.X} file because it is newer than the source file.
9787 gcc -Dfoo=bar file1.c -aux-info file1.X
9792 You need to include the special files along with the rest in the
9793 @code{protoize} command, even though their @samp{.X} files already
9794 exist, because otherwise they won't get converted.
9796 @xref{Protoize Caveats}, for more information on how to use
9797 @code{protoize} successfully.