1 @c Copyright (C) 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
2 @c 2000, 2001, 2002, 2003, 2004 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, 2002, 2003, 2004 Free Software Foundation, Inc.
11 Permission is granted to copy, distribute and/or modify this document
12 under the terms of the GNU Free Documentation License, Version 1.2 or
13 any later version published by the Free Software Foundation; with the
14 Invariant Sections being ``GNU General Public License'' and ``Funding
15 Free Software'', the Front-Cover texts being (a) (see below), and with
16 the Back-Cover Texts being (b) (see below). A copy of the license is
17 included in the gfdl(7) man page.
19 (a) The FSF's Front-Cover Text is:
23 (b) The FSF's Back-Cover Text is:
25 You have freedom to copy and modify this GNU Manual, like GNU
26 software. Copies published by the Free Software Foundation raise
27 funds for GNU development.
29 @c Set file name and title for the man page.
31 @settitle GNU project C and C++ compiler
33 gcc [@option{-c}|@option{-S}|@option{-E}] [@option{-std=}@var{standard}]
34 [@option{-g}] [@option{-pg}] [@option{-O}@var{level}]
35 [@option{-W}@var{warn}@dots{}] [@option{-pedantic}]
36 [@option{-I}@var{dir}@dots{}] [@option{-L}@var{dir}@dots{}]
37 [@option{-D}@var{macro}[=@var{defn}]@dots{}] [@option{-U}@var{macro}]
38 [@option{-f}@var{option}@dots{}] [@option{-m}@var{machine-option}@dots{}]
39 [@option{-o} @var{outfile}] @var{infile}@dots{}
41 Only the most useful options are listed here; see below for the
42 remainder. @samp{g++} accepts mostly the same options as @samp{gcc}.
45 gpl(7), gfdl(7), fsf-funding(7),
46 cpp(1), gcov(1), g77(1), as(1), ld(1), gdb(1), adb(1), dbx(1), sdb(1)
47 and the Info entries for @file{gcc}, @file{cpp}, @file{g77}, @file{as},
48 @file{ld}, @file{binutils} and @file{gdb}.
51 For instructions on reporting bugs, see
52 @w{@uref{http://gcc.gnu.org/bugs.html}}. Use of the @command{gccbug}
53 script to report bugs is recommended.
56 See the Info entry for @command{gcc}, or
57 @w{@uref{http://gcc.gnu.org/onlinedocs/gcc/Contributors.html}},
58 for contributors to GCC@.
63 @chapter GCC Command Options
64 @cindex GCC command options
65 @cindex command options
66 @cindex options, GCC command
68 @c man begin DESCRIPTION
69 When you invoke GCC, it normally does preprocessing, compilation,
70 assembly and linking. The ``overall options'' allow you to stop this
71 process at an intermediate stage. For example, the @option{-c} option
72 says not to run the linker. Then the output consists of object files
73 output by the assembler.
75 Other options are passed on to one stage of processing. Some options
76 control the preprocessor and others the compiler itself. Yet other
77 options control the assembler and linker; most of these are not
78 documented here, since you rarely need to use any of them.
80 @cindex C compilation options
81 Most of the command line options that you can use with GCC are useful
82 for C programs; when an option is only useful with another language
83 (usually C++), the explanation says so explicitly. If the description
84 for a particular option does not mention a source language, you can use
85 that option with all supported languages.
87 @cindex C++ compilation options
88 @xref{Invoking G++,,Compiling C++ Programs}, for a summary of special
89 options for compiling C++ programs.
91 @cindex grouping options
92 @cindex options, grouping
93 The @command{gcc} program accepts options and file names as operands. Many
94 options have multi-letter names; therefore multiple single-letter options
95 may @emph{not} be grouped: @option{-dr} is very different from @w{@samp{-d
98 @cindex order of options
99 @cindex options, order
100 You can mix options and other arguments. For the most part, the order
101 you use doesn't matter. Order does matter when you use several options
102 of the same kind; for example, if you specify @option{-L} more than once,
103 the directories are searched in the order specified.
105 Many options have long names starting with @samp{-f} or with
106 @samp{-W}---for example, @option{-fforce-mem},
107 @option{-fstrength-reduce}, @option{-Wformat} and so on. Most of
108 these have both positive and negative forms; the negative form of
109 @option{-ffoo} would be @option{-fno-foo}. This manual documents
110 only one of these two forms, whichever one is not the default.
114 @xref{Option Index}, for an index to GCC's options.
117 * Option Summary:: Brief list of all options, without explanations.
118 * Overall Options:: Controlling the kind of output:
119 an executable, object files, assembler files,
120 or preprocessed source.
121 * Invoking G++:: Compiling C++ programs.
122 * C Dialect Options:: Controlling the variant of C language compiled.
123 * C++ Dialect Options:: Variations on C++.
124 * Objective-C Dialect Options:: Variations on Objective-C.
125 * Language Independent Options:: Controlling how diagnostics should be
127 * Warning Options:: How picky should the compiler be?
128 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
129 * Optimize Options:: How much optimization?
130 * Preprocessor Options:: Controlling header files and macro definitions.
131 Also, getting dependency information for Make.
132 * Assembler Options:: Passing options to the assembler.
133 * Link Options:: Specifying libraries and so on.
134 * Directory Options:: Where to find header files and libraries.
135 Where to find the compiler executable files.
136 * Spec Files:: How to pass switches to sub-processes.
137 * Target Options:: Running a cross-compiler, or an old version of GCC.
138 * Submodel Options:: Specifying minor hardware or convention variations,
139 such as 68010 vs 68020.
140 * Code Gen Options:: Specifying conventions for function calls, data layout
142 * Environment Variables:: Env vars that affect GCC.
143 * Precompiled Headers:: Compiling a header once, and using it many times.
144 * Running Protoize:: Automatically adding or removing function prototypes.
150 @section Option Summary
152 Here is a summary of all the options, grouped by type. Explanations are
153 in the following sections.
156 @item Overall Options
157 @xref{Overall Options,,Options Controlling the Kind of Output}.
158 @gccoptlist{-c -S -E -o @var{file} -combine -pipe -pass-exit-codes @gol
159 -x @var{language} -v -### --help --target-help --version}
161 @item C Language Options
162 @xref{C Dialect Options,,Options Controlling C Dialect}.
163 @gccoptlist{-ansi -std=@var{standard} -aux-info @var{filename} @gol
164 -fno-asm -fno-builtin -fno-builtin-@var{function} @gol
165 -fhosted -ffreestanding -fms-extensions @gol
166 -trigraphs -no-integrated-cpp -traditional -traditional-cpp @gol
167 -fallow-single-precision -fcond-mismatch @gol
168 -fsigned-bitfields -fsigned-char @gol
169 -funsigned-bitfields -funsigned-char}
171 @item C++ Language Options
172 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
173 @gccoptlist{-fabi-version=@var{n} -fno-access-control -fcheck-new @gol
174 -fconserve-space -fno-const-strings @gol
175 -fno-elide-constructors @gol
176 -fno-enforce-eh-specs @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 -fno-weak -nostdinc++ @gol
185 -fno-default-inline -Wabi -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 @gol
197 -fno-nil-receivers @gol
198 -fobjc-exceptions @gol
199 -freplace-objc-classes @gol
202 -Wno-protocol -Wselector -Wundeclared-selector}
204 @item Language Independent Options
205 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
206 @gccoptlist{-fmessage-length=@var{n} @gol
207 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]}}
209 @item Warning Options
210 @xref{Warning Options,,Options to Request or Suppress Warnings}.
211 @gccoptlist{-fsyntax-only -pedantic -pedantic-errors @gol
212 -w -Wextra -Wall -Waggregate-return @gol
213 -Wcast-align -Wcast-qual -Wchar-subscripts -Wcomment @gol
214 -Wconversion -Wno-deprecated-declarations @gol
215 -Wdisabled-optimization -Wno-div-by-zero -Wendif-labels @gol
216 -Werror -Werror-implicit-function-declaration @gol
217 -Wfatal-errors -Wfloat-equal -Wformat -Wformat=2 @gol
218 -Wno-format-extra-args -Wformat-nonliteral @gol
219 -Wformat-security -Wformat-y2k @gol
220 -Wimplicit -Wimplicit-function-declaration -Wimplicit-int @gol
221 -Wimport -Wno-import -Winit-self -Winline @gol
222 -Wno-invalid-offsetof -Winvalid-pch @gol
223 -Wlarger-than-@var{len} -Wlong-long @gol
224 -Wmain -Wmissing-braces @gol
225 -Wmissing-format-attribute -Wmissing-noreturn @gol
226 -Wno-multichar -Wnonnull -Wpacked -Wpadded @gol
227 -Wparentheses -Wpointer-arith -Wredundant-decls @gol
228 -Wreturn-type -Wsequence-point -Wshadow @gol
229 -Wsign-compare -Wstrict-aliasing -Wstrict-aliasing=2 @gol
230 -Wswitch -Wswitch-default -Wswitch-enum @gol
231 -Wsystem-headers -Wtrigraphs -Wundef -Wuninitialized @gol
232 -Wunknown-pragmas -Wunreachable-code @gol
233 -Wunused -Wunused-function -Wunused-label -Wunused-parameter @gol
234 -Wunused-value -Wunused-variable -Wwrite-strings @gol
237 @item C-only Warning Options
238 @gccoptlist{-Wbad-function-cast -Wmissing-declarations @gol
239 -Wmissing-prototypes -Wnested-externs -Wold-style-definition @gol
240 -Wstrict-prototypes -Wtraditional @gol
241 -Wdeclaration-after-statement}
243 @item Debugging Options
244 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
245 @gccoptlist{-d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
246 -fdump-unnumbered -fdump-translation-unit@r{[}-@var{n}@r{]} @gol
247 -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
249 -fdump-tree-original@r{[}-@var{n}@r{]} @gol
250 -fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
251 -fdump-tree-inlined@r{[}-@var{n}@r{]} @gol
252 -fdump-tree-cfg -fdump-tree-vcg -fdump-tree-alias @gol
254 -fdump-tree-ssa@r{[}-@var{n}@r{]} -fdump-tree-pre@r{[}-@var{n}@r{]} @gol
255 -fdump-tree-ccp@r{[}-@var{n}@r{]} -fdump-tree-dce@r{[}-@var{n}@r{]} @gol
256 -fdump-tree-gimple@r{[}-raw@r{]} -fdump-tree-mudflap@r{[}-@var{n}@r{]} @gol
257 -fdump-tree-dom@r{[}-@var{n}@r{]} @gol
258 -fdump-tree-dse@r{[}-@var{n}@r{]} @gol
259 -fdump-tree-phiopt@r{[}-@var{n}@r{]} @gol
260 -fdump-tree-forwprop@r{[}-@var{n}@r{]} @gol
261 -fdump-tree-copyrename@r{[}-@var{n}@r{]} @gol
263 -fdump-tree-sra@r{[}-@var{n}@r{]} @gol
264 -feliminate-dwarf2-dups -feliminate-unused-debug-types @gol
265 -feliminate-unused-debug-symbols -fmem-report -fprofile-arcs -ftree-based-profiling @gol
266 -frandom-seed=@var{string} -fsched-verbose=@var{n} @gol
267 -ftest-coverage -ftime-report -fvar-tracking @gol
268 -g -g@var{level} -gcoff -gdwarf-2 @gol
269 -ggdb -gstabs -gstabs+ -gvms -gxcoff -gxcoff+ @gol
270 -p -pg -print-file-name=@var{library} -print-libgcc-file-name @gol
271 -print-multi-directory -print-multi-lib @gol
272 -print-prog-name=@var{program} -print-search-dirs -Q @gol
275 @item Optimization Options
276 @xref{Optimize Options,,Options that Control Optimization}.
277 @gccoptlist{-falign-functions=@var{n} -falign-jumps=@var{n} @gol
278 -falign-labels=@var{n} -falign-loops=@var{n} @gol
279 -fbounds-check -fmudflap -fmudflapth -fmudflapir @gol
280 -fbranch-probabilities -fprofile-values -fvpt -fbranch-target-load-optimize @gol
281 -fbranch-target-load-optimize2 -fbtr-bb-exclusive @gol
282 -fcaller-saves -fcprop-registers @gol
283 -fcse-follow-jumps -fcse-skip-blocks -fdata-sections @gol
284 -fdelayed-branch -fdelete-null-pointer-checks @gol
285 -fexpensive-optimizations -ffast-math -ffloat-store @gol
286 -fforce-addr -fforce-mem -ffunction-sections @gol
287 -fgcse -fgcse-lm -fgcse-sm -fgcse-las -fgcse-after-reload @gol
288 -floop-optimize -fcrossjumping -fif-conversion -fif-conversion2 @gol
289 -finline-functions -finline-limit=@var{n} -fkeep-inline-functions @gol
290 -fkeep-static-consts -fmerge-constants -fmerge-all-constants @gol
291 -fmove-all-movables -fnew-ra -fno-branch-count-reg @gol
292 -fno-default-inline -fno-defer-pop @gol
293 -fno-function-cse -fno-guess-branch-probability @gol
294 -fno-inline -fno-math-errno -fno-peephole -fno-peephole2 @gol
295 -funsafe-math-optimizations -ffinite-math-only @gol
296 -fno-trapping-math -fno-zero-initialized-in-bss @gol
297 -fomit-frame-pointer -foptimize-register-move @gol
298 -foptimize-sibling-calls -fprefetch-loop-arrays @gol
299 -fprofile-generate -fprofile-use @gol
300 -freduce-all-givs -fregmove -frename-registers @gol
301 -freorder-blocks -freorder-blocks-and-partition -freorder-functions @gol
302 -frerun-cse-after-loop -frerun-loop-opt @gol
303 -frounding-math -fschedule-insns -fschedule-insns2 @gol
304 -fno-sched-interblock -fno-sched-spec -fsched-spec-load @gol
305 -fsched-spec-load-dangerous @gol
306 -fsched-stalled-insns=@var{n} -sched-stalled-insns-dep=@var{n} @gol
307 -fsched2-use-superblocks @gol
308 -fsched2-use-traces -fsignaling-nans @gol
309 -fsingle-precision-constant @gol
310 -fstrength-reduce -fstrict-aliasing -ftracer -fthread-jumps @gol
311 -funroll-all-loops -funroll-loops -fpeel-loops @gol
312 -funswitch-loops -fold-unroll-loops -fold-unroll-all-loops @gol
313 -ftree-pre -ftree-ccp -ftree-dce @gol
314 -ftree-dominator-opts -ftree-dse -ftree-copyrename @gol
315 -ftree-ch -ftree-sra -ftree-ter -ftree-lrs @gol
316 --param @var{name}=@var{value}
317 -O -O0 -O1 -O2 -O3 -Os}
319 @item Preprocessor Options
320 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
321 @gccoptlist{-A@var{question}=@var{answer} @gol
322 -A-@var{question}@r{[}=@var{answer}@r{]} @gol
323 -C -dD -dI -dM -dN @gol
324 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
325 -idirafter @var{dir} @gol
326 -include @var{file} -imacros @var{file} @gol
327 -iprefix @var{file} -iwithprefix @var{dir} @gol
328 -iwithprefixbefore @var{dir} -isystem @var{dir} @gol
329 -M -MM -MF -MG -MP -MQ -MT -nostdinc @gol
330 -P -fworking-directory -remap @gol
331 -trigraphs -undef -U@var{macro} -Wp,@var{option} @gol
332 -Xpreprocessor @var{option}}
334 @item Assembler Option
335 @xref{Assembler Options,,Passing Options to the Assembler}.
336 @gccoptlist{-Wa,@var{option} -Xassembler @var{option}}
339 @xref{Link Options,,Options for Linking}.
340 @gccoptlist{@var{object-file-name} -l@var{library} @gol
341 -nostartfiles -nodefaultlibs -nostdlib -pie @gol
342 -s -static -static-libgcc -shared -shared-libgcc -symbolic @gol
343 -Wl,@var{option} -Xlinker @var{option} @gol
346 @item Directory Options
347 @xref{Directory Options,,Options for Directory Search}.
348 @gccoptlist{-B@var{prefix} -I@var{dir} -iquote@var{dir} -L@var{dir} -specs=@var{file} -I-}
351 @c I wrote this xref this way to avoid overfull hbox. -- rms
352 @xref{Target Options}.
353 @gccoptlist{-V @var{version} -b @var{machine}}
355 @item Machine Dependent Options
356 @xref{Submodel Options,,Hardware Models and Configurations}.
358 @emph{M680x0 Options}
359 @gccoptlist{-m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
360 -m68060 -mcpu32 -m5200 -m68881 -mbitfield -mc68000 -mc68020 @gol
361 -mnobitfield -mrtd -mshort -msoft-float -mpcrel @gol
362 -malign-int -mstrict-align -msep-data -mno-sep-data @gol
363 -mshared-library-id=n -mid-shared-library -mno-id-shared-library}
365 @emph{M68hc1x Options}
366 @gccoptlist{-m6811 -m6812 -m68hc11 -m68hc12 -m68hcs12 @gol
367 -mauto-incdec -minmax -mlong-calls -mshort @gol
368 -msoft-reg-count=@var{count}}
371 @gccoptlist{-mg -mgnu -munix}
374 @gccoptlist{-mcpu=@var{cpu-type} @gol
375 -mtune=@var{cpu-type} @gol
376 -mcmodel=@var{code-model} @gol
377 -m32 -m64 -mapp-regs -mno-app-regs @gol
378 -mfaster-structs -mno-faster-structs @gol
379 -mfpu -mno-fpu -mhard-float -msoft-float @gol
380 -mhard-quad-float -msoft-quad-float @gol
381 -mimpure-text -mno-impure-text -mlittle-endian @gol
382 -mstack-bias -mno-stack-bias @gol
383 -munaligned-doubles -mno-unaligned-doubles @gol
384 -mv8plus -mno-v8plus -mvis -mno-vis}
387 @gccoptlist{-mapcs-frame -mno-apcs-frame @gol
388 -mabi=@var{name} @gol
389 -mapcs-26 -mapcs-32 @gol
390 -mapcs-stack-check -mno-apcs-stack-check @gol
391 -mapcs-float -mno-apcs-float @gol
392 -mapcs-reentrant -mno-apcs-reentrant @gol
393 -msched-prolog -mno-sched-prolog @gol
394 -mlittle-endian -mbig-endian -mwords-little-endian @gol
395 -malignment-traps -mno-alignment-traps @gol
396 -mfloat-abi=@var{name} soft-float -mhard-float -mfpe @gol
397 -mthumb-interwork -mno-thumb-interwork @gol
398 -mcpu=@var{name} -march=@var{name} -mfpu=@var{name} @gol
399 -mstructure-size-boundary=@var{n} @gol
400 -mabort-on-noreturn @gol
401 -mlong-calls -mno-long-calls @gol
402 -msingle-pic-base -mno-single-pic-base @gol
403 -mpic-register=@var{reg} @gol
404 -mnop-fun-dllimport @gol
405 -mcirrus-fix-invalid-insns -mno-cirrus-fix-invalid-insns @gol
406 -mpoke-function-name @gol
408 -mtpcs-frame -mtpcs-leaf-frame @gol
409 -mcaller-super-interworking -mcallee-super-interworking}
411 @emph{MN10300 Options}
412 @gccoptlist{-mmult-bug -mno-mult-bug @gol
413 -mam33 -mno-am33 @gol
414 -mam33-2 -mno-am33-2 @gol
417 @emph{M32R/D Options}
418 @gccoptlist{-m32r2 -m32rx -m32r @gol
420 -malign-loops -mno-align-loops @gol
421 -missue-rate=@var{number} @gol
422 -mbranch-cost=@var{number} @gol
423 -mmodel=@var{code-size-model-type} @gol
424 -msdata=@var{sdata-type} @gol
425 -mno-flush-func -mflush-func=@var{name} @gol
426 -mno-flush-trap -mflush-trap=@var{number} @gol
429 @emph{RS/6000 and PowerPC Options}
430 @gccoptlist{-mcpu=@var{cpu-type} @gol
431 -mtune=@var{cpu-type} @gol
432 -mpower -mno-power -mpower2 -mno-power2 @gol
433 -mpowerpc -mpowerpc64 -mno-powerpc @gol
434 -maltivec -mno-altivec @gol
435 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
436 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
437 -mnew-mnemonics -mold-mnemonics @gol
438 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
439 -m64 -m32 -mxl-call -mno-xl-call -mpe @gol
440 -malign-power -malign-natural @gol
441 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
442 -mstring -mno-string -mupdate -mno-update @gol
443 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
444 -mstrict-align -mno-strict-align -mrelocatable @gol
445 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
446 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
447 -mdynamic-no-pic @gol
448 -mprioritize-restricted-insns=@var{priority} @gol
449 -msched-costly-dep=@var{dependence_type} @gol
450 -minsert-sched-nops=@var{scheme} @gol
451 -mcall-sysv -mcall-netbsd @gol
452 -maix-struct-return -msvr4-struct-return @gol
453 -mabi=altivec -mabi=no-altivec @gol
454 -mabi=spe -mabi=no-spe @gol
455 -misel=yes -misel=no @gol
456 -mspe=yes -mspe=no @gol
457 -mfloat-gprs=yes -mfloat-gprs=no @gol
458 -mprototype -mno-prototype @gol
459 -msim -mmvme -mads -myellowknife -memb -msdata @gol
460 -msdata=@var{opt} -mvxworks -mwindiss -G @var{num} -pthread}
462 @emph{Darwin Options}
463 @gccoptlist{-all_load -allowable_client -arch -arch_errors_fatal @gol
464 -arch_only -bind_at_load -bundle -bundle_loader @gol
465 -client_name -compatibility_version -current_version @gol
466 -dependency-file -dylib_file -dylinker_install_name @gol
467 -dynamic -dynamiclib -exported_symbols_list @gol
468 -filelist -flat_namespace -force_cpusubtype_ALL @gol
469 -force_flat_namespace -headerpad_max_install_names @gol
470 -image_base -init -install_name -keep_private_externs @gol
471 -multi_module -multiply_defined -multiply_defined_unused @gol
472 -noall_load -nofixprebinding -nomultidefs -noprebind -noseglinkedit @gol
473 -pagezero_size -prebind -prebind_all_twolevel_modules @gol
474 -private_bundle -read_only_relocs -sectalign @gol
475 -sectobjectsymbols -whyload -seg1addr @gol
476 -sectcreate -sectobjectsymbols -sectorder @gol
477 -seg_addr_table -seg_addr_table_filename -seglinkedit @gol
478 -segprot -segs_read_only_addr -segs_read_write_addr @gol
479 -single_module -static -sub_library -sub_umbrella @gol
480 -twolevel_namespace -umbrella -undefined @gol
481 -unexported_symbols_list -weak_reference_mismatches @gol
485 @gccoptlist{-EL -EB -march=@var{arch} -mtune=@var{arch} @gol
486 -mips1 -mips2 -mips3 -mips4 -mips32 -mips32r2 -mips64 @gol
487 -mips16 -mno-mips16 -mabi=@var{abi} -mabicalls -mno-abicalls @gol
488 -mxgot -mno-xgot -mgp32 -mgp64 -mfp32 -mfp64 @gol
489 -mhard-float -msoft-float -msingle-float -mdouble-float @gol
490 -mint64 -mlong64 -mlong32 @gol
491 -G@var{num} -membedded-data -mno-embedded-data @gol
492 -muninit-const-in-rodata -mno-uninit-const-in-rodata @gol
493 -msplit-addresses -mno-split-addresses @gol
494 -mexplicit-relocs -mno-explicit-relocs @gol
495 -mrnames -mno-rnames @gol
496 -mcheck-zero-division -mno-check-zero-division @gol
497 -mmemcpy -mno-memcpy -mlong-calls -mno-long-calls @gol
498 -mmad -mno-mad -mfused-madd -mno-fused-madd -nocpp @gol
499 -mfix-r4000 -mno-fix-r4000 -mfix-r4400 -mno-fix-r4400 @gol
500 -mfix-vr4120 -mno-fix-vr4120 -mfix-sb1 -mno-fix-sb1 @gol
501 -mflush-func=@var{func} -mno-flush-func @gol
502 -mbranch-likely -mno-branch-likely @gol
503 -mfp-exceptions -mno-fp-exceptions @gol
504 -mvr4130-align -mno-vr4130-align}
506 @emph{i386 and x86-64 Options}
507 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
508 -mfpmath=@var{unit} @gol
509 -masm=@var{dialect} -mno-fancy-math-387 @gol
510 -mno-fp-ret-in-387 -msoft-float -msvr3-shlib @gol
511 -mno-wide-multiply -mrtd -malign-double @gol
512 -mpreferred-stack-boundary=@var{num} @gol
513 -mmmx -msse -msse2 -msse3 -m3dnow @gol
514 -mthreads -mno-align-stringops -minline-all-stringops @gol
515 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
516 -m96bit-long-double -mregparm=@var{num} -momit-leaf-frame-pointer @gol
517 -mno-red-zone -mno-tls-direct-seg-refs @gol
518 -mcmodel=@var{code-model} @gol
522 @gccoptlist{-march=@var{architecture-type} @gol
523 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
524 -mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
525 -mjump-in-delay -mlinker-opt -mlong-calls @gol
526 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
527 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
528 -mno-jump-in-delay -mno-long-load-store @gol
529 -mno-portable-runtime -mno-soft-float @gol
530 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
531 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
532 -mschedule=@var{cpu-type} -mspace-regs -msio -mwsio @gol
533 -nolibdld -static -threads}
535 @emph{DEC Alpha Options}
536 @gccoptlist{-mno-fp-regs -msoft-float -malpha-as -mgas @gol
537 -mieee -mieee-with-inexact -mieee-conformant @gol
538 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
539 -mtrap-precision=@var{mode} -mbuild-constants @gol
540 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} @gol
541 -mbwx -mmax -mfix -mcix @gol
542 -mfloat-vax -mfloat-ieee @gol
543 -mexplicit-relocs -msmall-data -mlarge-data @gol
544 -msmall-text -mlarge-text @gol
545 -mmemory-latency=@var{time}}
547 @emph{DEC Alpha/VMS Options}
548 @gccoptlist{-mvms-return-codes}
550 @emph{H8/300 Options}
551 @gccoptlist{-mrelax -mh -ms -mn -mint32 -malign-300}
554 @gccoptlist{-m1 -m2 -m2e -m3 -m3e @gol
555 -m4-nofpu -m4-single-only -m4-single -m4 @gol
556 -m5-64media -m5-64media-nofpu @gol
557 -m5-32media -m5-32media-nofpu @gol
558 -m5-compact -m5-compact-nofpu @gol
559 -mb -ml -mdalign -mrelax @gol
560 -mbigtable -mfmovd -mhitachi -mnomacsave @gol
561 -mieee -misize -mpadstruct -mspace @gol
562 -mprefergot -musermode}
564 @emph{System V Options}
565 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
568 @gccoptlist{-EB -EL @gol
569 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
570 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
572 @emph{TMS320C3x/C4x Options}
573 @gccoptlist{-mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm @gol
574 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload @gol
575 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned @gol
576 -mparallel-insns -mparallel-mpy -mpreserve-float}
579 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
580 -mprolog-function -mno-prolog-function -mspace @gol
581 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
582 -mapp-regs -mno-app-regs @gol
583 -mdisable-callt -mno-disable-callt @gol
589 @gccoptlist{-m32032 -m32332 -m32532 -m32081 -m32381 @gol
590 -mmult-add -mnomult-add -msoft-float -mrtd -mnortd @gol
591 -mregparam -mnoregparam -msb -mnosb @gol
592 -mbitfield -mnobitfield -mhimem -mnohimem}
595 @gccoptlist{-mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts @gol
596 -mcall-prologues -mno-tablejump -mtiny-stack}
599 @gccoptlist{-mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
600 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
601 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
602 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
603 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
606 @gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
607 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
608 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
609 -mno-base-addresses -msingle-exit -mno-single-exit}
612 @gccoptlist{-mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
613 -mvolatile-asm-stop -mb-step -mregister-names -mno-sdata @gol
614 -mconstant-gp -mauto-pic -minline-float-divide-min-latency @gol
615 -minline-float-divide-max-throughput @gol
616 -minline-int-divide-min-latency @gol
617 -minline-int-divide-max-throughput -mno-dwarf2-asm @gol
618 -mfixed-range=@var{register-range}}
620 @emph{S/390 and zSeries Options}
621 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
622 -mhard-float -msoft-float -mbackchain -mno-backchain @gol
623 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
624 -m64 -m31 -mdebug -mno-debug -mesa -mzarch -mfused-madd -mno-fused-madd}
627 @gccoptlist{-mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
628 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
629 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
630 -mstack-align -mdata-align -mconst-align @gol
631 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
632 -melf -maout -melinux -mlinux -sim -sim2 @gol
633 -mmul-bug-workaround -mno-mul-bug-workaround}
635 @emph{PDP-11 Options}
636 @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
637 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
638 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
639 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
640 -mbranch-expensive -mbranch-cheap @gol
641 -msplit -mno-split -munix-asm -mdec-asm}
643 @emph{Xstormy16 Options}
646 @emph{Xtensa Options}
647 @gccoptlist{-mconst16 -mno-const16 @gol
648 -mfused-madd -mno-fused-madd @gol
649 -mtext-section-literals -mno-text-section-literals @gol
650 -mtarget-align -mno-target-align @gol
651 -mlongcalls -mno-longcalls}
654 @gccoptlist{-mgpr-32 -mgpr-64 -mfpr-32 -mfpr-64 @gol
655 -mhard-float -msoft-float @gol
656 -malloc-cc -mfixed-cc -mdword -mno-dword @gol
657 -mdouble -mno-double @gol
658 -mmedia -mno-media -mmuladd -mno-muladd @gol
659 -mfdpic -minline-plt -mgprel-ro -multilib-library-pic -mlinked-fp @gol
660 -mlibrary-pic -macc-4 -macc-8 @gol
661 -mpack -mno-pack -mno-eflags -mcond-move -mno-cond-move @gol
662 -mscc -mno-scc -mcond-exec -mno-cond-exec @gol
663 -mvliw-branch -mno-vliw-branch @gol
664 -mmulti-cond-exec -mno-multi-cond-exec -mnested-cond-exec @gol
665 -mno-nested-cond-exec -mtomcat-stats @gol
668 @item Code Generation Options
669 @xref{Code Gen Options,,Options for Code Generation Conventions}.
670 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
671 -ffixed-@var{reg} -fexceptions @gol
672 -fnon-call-exceptions -funwind-tables @gol
673 -fasynchronous-unwind-tables @gol
674 -finhibit-size-directive -finstrument-functions @gol
675 -fno-common -fno-ident @gol
676 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
677 -freg-struct-return -fshared-data -fshort-enums @gol
678 -fshort-double -fshort-wchar @gol
679 -fverbose-asm -fpack-struct -fstack-check @gol
680 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
681 -fargument-alias -fargument-noalias @gol
682 -fargument-noalias-global -fleading-underscore @gol
683 -ftls-model=@var{model} @gol
684 -ftrapv -fwrapv -fbounds-check}
688 * Overall Options:: Controlling the kind of output:
689 an executable, object files, assembler files,
690 or preprocessed source.
691 * C Dialect Options:: Controlling the variant of C language compiled.
692 * C++ Dialect Options:: Variations on C++.
693 * Objective-C Dialect Options:: Variations on Objective-C.
694 * Language Independent Options:: Controlling how diagnostics should be
696 * Warning Options:: How picky should the compiler be?
697 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
698 * Optimize Options:: How much optimization?
699 * Preprocessor Options:: Controlling header files and macro definitions.
700 Also, getting dependency information for Make.
701 * Assembler Options:: Passing options to the assembler.
702 * Link Options:: Specifying libraries and so on.
703 * Directory Options:: Where to find header files and libraries.
704 Where to find the compiler executable files.
705 * Spec Files:: How to pass switches to sub-processes.
706 * Target Options:: Running a cross-compiler, or an old version of GCC.
709 @node Overall Options
710 @section Options Controlling the Kind of Output
712 Compilation can involve up to four stages: preprocessing, compilation
713 proper, assembly and linking, always in that order. GCC is capable of
714 preprocessing and compiling several files either into several
715 assembler input files, or into one assembler input file; then each
716 assembler input file produces an object file, and linking combines all
717 the object files (those newly compiled, and those specified as input)
718 into an executable file.
720 @cindex file name suffix
721 For any given input file, the file name suffix determines what kind of
726 C source code which must be preprocessed.
729 C source code which should not be preprocessed.
732 C++ source code which should not be preprocessed.
735 Objective-C source code. Note that you must link with the library
736 @file{libobjc.a} to make an Objective-C program work.
739 Objective-C source code which should not be preprocessed.
742 C or C++ header file to be turned into a precompiled header.
746 @itemx @var{file}.cxx
747 @itemx @var{file}.cpp
748 @itemx @var{file}.CPP
749 @itemx @var{file}.c++
751 C++ source code which must be preprocessed. Note that in @samp{.cxx},
752 the last two letters must both be literally @samp{x}. Likewise,
753 @samp{.C} refers to a literal capital C@.
757 C++ header file to be turned into a precompiled header.
760 @itemx @var{file}.for
761 @itemx @var{file}.FOR
762 Fortran source code which should not be preprocessed.
765 @itemx @var{file}.fpp
766 @itemx @var{file}.FPP
767 Fortran source code which must be preprocessed (with the traditional
771 Fortran source code which must be preprocessed with a RATFOR
772 preprocessor (not included with GCC)@.
775 @itemx @var{file}.f95
776 Fortran 90/95 source code which should not be preprocessed.
778 @xref{Overall Options,,Options Controlling the Kind of Output, g77,
779 Using and Porting GNU Fortran}, for more details of the handling of
782 @c FIXME: Descriptions of Java file types.
789 Ada source code file which contains a library unit declaration (a
790 declaration of a package, subprogram, or generic, or a generic
791 instantiation), or a library unit renaming declaration (a package,
792 generic, or subprogram renaming declaration). Such files are also
795 @itemx @var{file}.adb
796 Ada source code file containing a library unit body (a subprogram or
797 package body). Such files are also called @dfn{bodies}.
799 @c GCC also knows about some suffixes for languages not yet included:
808 Assembler code which must be preprocessed.
811 An object file to be fed straight into linking.
812 Any file name with no recognized suffix is treated this way.
816 You can specify the input language explicitly with the @option{-x} option:
819 @item -x @var{language}
820 Specify explicitly the @var{language} for the following input files
821 (rather than letting the compiler choose a default based on the file
822 name suffix). This option applies to all following input files until
823 the next @option{-x} option. Possible values for @var{language} are:
825 c c-header cpp-output
826 c++ c++-header c++-cpp-output
827 objective-c objective-c-header objc-cpp-output
828 assembler assembler-with-cpp
830 f77 f77-cpp-input ratfor
837 Turn off any specification of a language, so that subsequent files are
838 handled according to their file name suffixes (as they are if @option{-x}
839 has not been used at all).
841 @item -pass-exit-codes
842 @opindex pass-exit-codes
843 Normally the @command{gcc} program will exit with the code of 1 if any
844 phase of the compiler returns a non-success return code. If you specify
845 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
846 numerically highest error produced by any phase that returned an error
850 If you only want some of the stages of compilation, you can use
851 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
852 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
853 @command{gcc} is to stop. Note that some combinations (for example,
854 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
859 Compile or assemble the source files, but do not link. The linking
860 stage simply is not done. The ultimate output is in the form of an
861 object file for each source file.
863 By default, the object file name for a source file is made by replacing
864 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
866 Unrecognized input files, not requiring compilation or assembly, are
871 Stop after the stage of compilation proper; do not assemble. The output
872 is in the form of an assembler code file for each non-assembler input
875 By default, the assembler file name for a source file is made by
876 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
878 Input files that don't require compilation are ignored.
882 Stop after the preprocessing stage; do not run the compiler proper. The
883 output is in the form of preprocessed source code, which is sent to the
886 Input files which don't require preprocessing are ignored.
888 @cindex output file option
891 Place output in file @var{file}. This applies regardless to whatever
892 sort of output is being produced, whether it be an executable file,
893 an object file, an assembler file or preprocessed C code.
895 If @option{-o} is not specified, the default is to put an executable
896 file in @file{a.out}, the object file for
897 @file{@var{source}.@var{suffix}} in @file{@var{source}.o}, its
898 assembler file in @file{@var{source}.s}, a precompiled header file in
899 @file{@var{source}.@var{suffix}.gch}, and all preprocessed C source on
904 Print (on standard error output) the commands executed to run the stages
905 of compilation. Also print the version number of the compiler driver
906 program and of the preprocessor and the compiler proper.
910 Like @option{-v} except the commands are not executed and all command
911 arguments are quoted. This is useful for shell scripts to capture the
912 driver-generated command lines.
916 Use pipes rather than temporary files for communication between the
917 various stages of compilation. This fails to work on some systems where
918 the assembler is unable to read from a pipe; but the GNU assembler has
923 If you are compiling multiple source files, this option tells the driver
924 to pass all the source files to the compiler at once (for those
925 languages for which the compiler can handle this). This will allow
926 intermodule analysis (IMA) to be performed by the compiler. Currently the only
927 language for which this is supported is C. If you pass source files for
928 multiple languages to the driver, using this option, the driver will invoke
929 the compiler(s) that support IMA once each, passing each compiler all the
930 source files appropriate for it. For those languages that do not support
931 IMA this option will be ignored, and the compiler will be invoked once for
932 each source file in that language. If you use this option in conjunction
933 with -save-temps, the compiler will generate multiple pre-processed files
934 (one for each source file), but only one (combined) .o or .s file.
938 Print (on the standard output) a description of the command line options
939 understood by @command{gcc}. If the @option{-v} option is also specified
940 then @option{--help} will also be passed on to the various processes
941 invoked by @command{gcc}, so that they can display the command line options
942 they accept. If the @option{-Wextra} option is also specified then command
943 line options which have no documentation associated with them will also
948 Print (on the standard output) a description of target specific command
949 line options for each tool.
953 Display the version number and copyrights of the invoked GCC.
957 @section Compiling C++ Programs
959 @cindex suffixes for C++ source
960 @cindex C++ source file suffixes
961 C++ source files conventionally use one of the suffixes @samp{.C},
962 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
963 @samp{.cxx}; C++ header files often use @samp{.hh} or @samp{.H}; and
964 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
965 files with these names and compiles them as C++ programs even if you
966 call the compiler the same way as for compiling C programs (usually
967 with the name @command{gcc}).
971 However, C++ programs often require class libraries as well as a
972 compiler that understands the C++ language---and under some
973 circumstances, you might want to compile programs or header files from
974 standard input, or otherwise without a suffix that flags them as C++
975 programs. You might also like to precompile a C header file with a
976 @samp{.h} extension to be used in C++ compilations. @command{g++} is a
977 program that calls GCC with the default language set to C++, and
978 automatically specifies linking against the C++ library. On many
979 systems, @command{g++} is also installed with the name @command{c++}.
981 @cindex invoking @command{g++}
982 When you compile C++ programs, you may specify many of the same
983 command-line options that you use for compiling programs in any
984 language; or command-line options meaningful for C and related
985 languages; or options that are meaningful only for C++ programs.
986 @xref{C Dialect Options,,Options Controlling C Dialect}, for
987 explanations of options for languages related to C@.
988 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
989 explanations of options that are meaningful only for C++ programs.
991 @node C Dialect Options
992 @section Options Controlling C Dialect
993 @cindex dialect options
994 @cindex language dialect options
995 @cindex options, dialect
997 The following options control the dialect of C (or languages derived
998 from C, such as C++ and Objective-C) that the compiler accepts:
1001 @cindex ANSI support
1005 In C mode, support all ISO C90 programs. In C++ mode,
1006 remove GNU extensions that conflict with ISO C++.
1008 This turns off certain features of GCC that are incompatible with ISO
1009 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
1010 such as the @code{asm} and @code{typeof} keywords, and
1011 predefined macros such as @code{unix} and @code{vax} that identify the
1012 type of system you are using. It also enables the undesirable and
1013 rarely used ISO trigraph feature. For the C compiler,
1014 it disables recognition of C++ style @samp{//} comments as well as
1015 the @code{inline} keyword.
1017 The alternate keywords @code{__asm__}, @code{__extension__},
1018 @code{__inline__} and @code{__typeof__} continue to work despite
1019 @option{-ansi}. You would not want to use them in an ISO C program, of
1020 course, but it is useful to put them in header files that might be included
1021 in compilations done with @option{-ansi}. Alternate predefined macros
1022 such as @code{__unix__} and @code{__vax__} are also available, with or
1023 without @option{-ansi}.
1025 The @option{-ansi} option does not cause non-ISO programs to be
1026 rejected gratuitously. For that, @option{-pedantic} is required in
1027 addition to @option{-ansi}. @xref{Warning Options}.
1029 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1030 option is used. Some header files may notice this macro and refrain
1031 from declaring certain functions or defining certain macros that the
1032 ISO standard doesn't call for; this is to avoid interfering with any
1033 programs that might use these names for other things.
1035 Functions which would normally be built in but do not have semantics
1036 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1037 functions with @option{-ansi} is used. @xref{Other Builtins,,Other
1038 built-in functions provided by GCC}, for details of the functions
1043 Determine the language standard. This option is currently only
1044 supported when compiling C or C++. A value for this option must be
1045 provided; possible values are
1050 ISO C90 (same as @option{-ansi}).
1052 @item iso9899:199409
1053 ISO C90 as modified in amendment 1.
1059 ISO C99. Note that this standard is not yet fully supported; see
1060 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1061 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1064 Default, ISO C90 plus GNU extensions (including some C99 features).
1068 ISO C99 plus GNU extensions. When ISO C99 is fully implemented in GCC,
1069 this will become the default. The name @samp{gnu9x} is deprecated.
1072 The 1998 ISO C++ standard plus amendments.
1075 The same as @option{-std=c++98} plus GNU extensions. This is the
1076 default for C++ code.
1079 Even when this option is not specified, you can still use some of the
1080 features of newer standards in so far as they do not conflict with
1081 previous C standards. For example, you may use @code{__restrict__} even
1082 when @option{-std=c99} is not specified.
1084 The @option{-std} options specifying some version of ISO C have the same
1085 effects as @option{-ansi}, except that features that were not in ISO C90
1086 but are in the specified version (for example, @samp{//} comments and
1087 the @code{inline} keyword in ISO C99) are not disabled.
1089 @xref{Standards,,Language Standards Supported by GCC}, for details of
1090 these standard versions.
1092 @item -aux-info @var{filename}
1094 Output to the given filename prototyped declarations for all functions
1095 declared and/or defined in a translation unit, including those in header
1096 files. This option is silently ignored in any language other than C@.
1098 Besides declarations, the file indicates, in comments, the origin of
1099 each declaration (source file and line), whether the declaration was
1100 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1101 @samp{O} for old, respectively, in the first character after the line
1102 number and the colon), and whether it came from a declaration or a
1103 definition (@samp{C} or @samp{F}, respectively, in the following
1104 character). In the case of function definitions, a K&R-style list of
1105 arguments followed by their declarations is also provided, inside
1106 comments, after the declaration.
1110 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1111 keyword, so that code can use these words as identifiers. You can use
1112 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1113 instead. @option{-ansi} implies @option{-fno-asm}.
1115 In C++, this switch only affects the @code{typeof} keyword, since
1116 @code{asm} and @code{inline} are standard keywords. You may want to
1117 use the @option{-fno-gnu-keywords} flag instead, which has the same
1118 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1119 switch only affects the @code{asm} and @code{typeof} keywords, since
1120 @code{inline} is a standard keyword in ISO C99.
1123 @itemx -fno-builtin-@var{function}
1124 @opindex fno-builtin
1125 @cindex built-in functions
1126 Don't recognize built-in functions that do not begin with
1127 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1128 functions provided by GCC}, for details of the functions affected,
1129 including those which are not built-in functions when @option{-ansi} or
1130 @option{-std} options for strict ISO C conformance are used because they
1131 do not have an ISO standard meaning.
1133 GCC normally generates special code to handle certain built-in functions
1134 more efficiently; for instance, calls to @code{alloca} may become single
1135 instructions that adjust the stack directly, and calls to @code{memcpy}
1136 may become inline copy loops. The resulting code is often both smaller
1137 and faster, but since the function calls no longer appear as such, you
1138 cannot set a breakpoint on those calls, nor can you change the behavior
1139 of the functions by linking with a different library.
1141 With the @option{-fno-builtin-@var{function}} option
1142 only the built-in function @var{function} is
1143 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1144 function is named this is not built-in in this version of GCC, this
1145 option is ignored. There is no corresponding
1146 @option{-fbuiltin-@var{function}} option; if you wish to enable
1147 built-in functions selectively when using @option{-fno-builtin} or
1148 @option{-ffreestanding}, you may define macros such as:
1151 #define abs(n) __builtin_abs ((n))
1152 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1157 @cindex hosted environment
1159 Assert that compilation takes place in a hosted environment. This implies
1160 @option{-fbuiltin}. A hosted environment is one in which the
1161 entire standard library is available, and in which @code{main} has a return
1162 type of @code{int}. Examples are nearly everything except a kernel.
1163 This is equivalent to @option{-fno-freestanding}.
1165 @item -ffreestanding
1166 @opindex ffreestanding
1167 @cindex hosted environment
1169 Assert that compilation takes place in a freestanding environment. This
1170 implies @option{-fno-builtin}. A freestanding environment
1171 is one in which the standard library may not exist, and program startup may
1172 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1173 This is equivalent to @option{-fno-hosted}.
1175 @xref{Standards,,Language Standards Supported by GCC}, for details of
1176 freestanding and hosted environments.
1178 @item -fms-extensions
1179 @opindex fms-extensions
1180 Accept some non-standard constructs used in Microsoft header files.
1184 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1185 options for strict ISO C conformance) implies @option{-trigraphs}.
1187 @item -no-integrated-cpp
1188 @opindex no-integrated-cpp
1189 Performs a compilation in two passes: preprocessing and compiling. This
1190 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1191 @option{-B} option. The user supplied compilation step can then add in
1192 an additional preprocessing step after normal preprocessing but before
1193 compiling. The default is to use the integrated cpp (internal cpp)
1195 The semantics of this option will change if "cc1", "cc1plus", and
1196 "cc1obj" are merged.
1198 @cindex traditional C language
1199 @cindex C language, traditional
1201 @itemx -traditional-cpp
1202 @opindex traditional-cpp
1203 @opindex traditional
1204 Formerly, these options caused GCC to attempt to emulate a pre-standard
1205 C compiler. They are now only supported with the @option{-E} switch.
1206 The preprocessor continues to support a pre-standard mode. See the GNU
1207 CPP manual for details.
1209 @item -fcond-mismatch
1210 @opindex fcond-mismatch
1211 Allow conditional expressions with mismatched types in the second and
1212 third arguments. The value of such an expression is void. This option
1213 is not supported for C++.
1215 @item -funsigned-char
1216 @opindex funsigned-char
1217 Let the type @code{char} be unsigned, like @code{unsigned char}.
1219 Each kind of machine has a default for what @code{char} should
1220 be. It is either like @code{unsigned char} by default or like
1221 @code{signed char} by default.
1223 Ideally, a portable program should always use @code{signed char} or
1224 @code{unsigned char} when it depends on the signedness of an object.
1225 But many programs have been written to use plain @code{char} and
1226 expect it to be signed, or expect it to be unsigned, depending on the
1227 machines they were written for. This option, and its inverse, let you
1228 make such a program work with the opposite default.
1230 The type @code{char} is always a distinct type from each of
1231 @code{signed char} or @code{unsigned char}, even though its behavior
1232 is always just like one of those two.
1235 @opindex fsigned-char
1236 Let the type @code{char} be signed, like @code{signed char}.
1238 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1239 the negative form of @option{-funsigned-char}. Likewise, the option
1240 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1242 @item -fsigned-bitfields
1243 @itemx -funsigned-bitfields
1244 @itemx -fno-signed-bitfields
1245 @itemx -fno-unsigned-bitfields
1246 @opindex fsigned-bitfields
1247 @opindex funsigned-bitfields
1248 @opindex fno-signed-bitfields
1249 @opindex fno-unsigned-bitfields
1250 These options control whether a bit-field is signed or unsigned, when the
1251 declaration does not use either @code{signed} or @code{unsigned}. By
1252 default, such a bit-field is signed, because this is consistent: the
1253 basic integer types such as @code{int} are signed types.
1256 @node C++ Dialect Options
1257 @section Options Controlling C++ Dialect
1259 @cindex compiler options, C++
1260 @cindex C++ options, command line
1261 @cindex options, C++
1262 This section describes the command-line options that are only meaningful
1263 for C++ programs; but you can also use most of the GNU compiler options
1264 regardless of what language your program is in. For example, you
1265 might compile a file @code{firstClass.C} like this:
1268 g++ -g -frepo -O -c firstClass.C
1272 In this example, only @option{-frepo} is an option meant
1273 only for C++ programs; you can use the other options with any
1274 language supported by GCC@.
1276 Here is a list of options that are @emph{only} for compiling C++ programs:
1280 @item -fabi-version=@var{n}
1281 @opindex fabi-version
1282 Use version @var{n} of the C++ ABI. Version 2 is the version of the
1283 C++ ABI that first appeared in G++ 3.4. Version 1 is the version of
1284 the C++ ABI that first appeared in G++ 3.2. Version 0 will always be
1285 the version that conforms most closely to the C++ ABI specification.
1286 Therefore, the ABI obtained using version 0 will change as ABI bugs
1289 The default is version 2.
1291 @item -fno-access-control
1292 @opindex fno-access-control
1293 Turn off all access checking. This switch is mainly useful for working
1294 around bugs in the access control code.
1298 Check that the pointer returned by @code{operator new} is non-null
1299 before attempting to modify the storage allocated. This check is
1300 normally unnecessary because the C++ standard specifies that
1301 @code{operator new} will only return @code{0} if it is declared
1302 @samp{throw()}, in which case the compiler will always check the
1303 return value even without this option. In all other cases, when
1304 @code{operator new} has a non-empty exception specification, memory
1305 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1306 @samp{new (nothrow)}.
1308 @item -fconserve-space
1309 @opindex fconserve-space
1310 Put uninitialized or runtime-initialized global variables into the
1311 common segment, as C does. This saves space in the executable at the
1312 cost of not diagnosing duplicate definitions. If you compile with this
1313 flag and your program mysteriously crashes after @code{main()} has
1314 completed, you may have an object that is being destroyed twice because
1315 two definitions were merged.
1317 This option is no longer useful on most targets, now that support has
1318 been added for putting variables into BSS without making them common.
1320 @item -fno-const-strings
1321 @opindex fno-const-strings
1322 Give string constants type @code{char *} instead of type @code{const
1323 char *}. By default, G++ uses type @code{const char *} as required by
1324 the standard. Even if you use @option{-fno-const-strings}, you cannot
1325 actually modify the value of a string constant.
1327 This option might be removed in a future release of G++. For maximum
1328 portability, you should structure your code so that it works with
1329 string constants that have type @code{const char *}.
1331 @item -fno-elide-constructors
1332 @opindex fno-elide-constructors
1333 The C++ standard allows an implementation to omit creating a temporary
1334 which is only used to initialize another object of the same type.
1335 Specifying this option disables that optimization, and forces G++ to
1336 call the copy constructor in all cases.
1338 @item -fno-enforce-eh-specs
1339 @opindex fno-enforce-eh-specs
1340 Don't check for violation of exception specifications at runtime. This
1341 option violates the C++ standard, but may be useful for reducing code
1342 size in production builds, much like defining @samp{NDEBUG}. The compiler
1343 will still optimize based on the exception specifications.
1346 @itemx -fno-for-scope
1348 @opindex fno-for-scope
1349 If @option{-ffor-scope} is specified, the scope of variables declared in
1350 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1351 as specified by the C++ standard.
1352 If @option{-fno-for-scope} is specified, the scope of variables declared in
1353 a @i{for-init-statement} extends to the end of the enclosing scope,
1354 as was the case in old versions of G++, and other (traditional)
1355 implementations of C++.
1357 The default if neither flag is given to follow the standard,
1358 but to allow and give a warning for old-style code that would
1359 otherwise be invalid, or have different behavior.
1361 @item -fno-gnu-keywords
1362 @opindex fno-gnu-keywords
1363 Do not recognize @code{typeof} as a keyword, so that code can use this
1364 word as an identifier. You can use the keyword @code{__typeof__} instead.
1365 @option{-ansi} implies @option{-fno-gnu-keywords}.
1367 @item -fno-implicit-templates
1368 @opindex fno-implicit-templates
1369 Never emit code for non-inline templates which are instantiated
1370 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1371 @xref{Template Instantiation}, for more information.
1373 @item -fno-implicit-inline-templates
1374 @opindex fno-implicit-inline-templates
1375 Don't emit code for implicit instantiations of inline templates, either.
1376 The default is to handle inlines differently so that compiles with and
1377 without optimization will need the same set of explicit instantiations.
1379 @item -fno-implement-inlines
1380 @opindex fno-implement-inlines
1381 To save space, do not emit out-of-line copies of inline functions
1382 controlled by @samp{#pragma implementation}. This will cause linker
1383 errors if these functions are not inlined everywhere they are called.
1385 @item -fms-extensions
1386 @opindex fms-extensions
1387 Disable pedantic warnings about constructs used in MFC, such as implicit
1388 int and getting a pointer to member function via non-standard syntax.
1390 @item -fno-nonansi-builtins
1391 @opindex fno-nonansi-builtins
1392 Disable built-in declarations of functions that are not mandated by
1393 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1394 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1396 @item -fno-operator-names
1397 @opindex fno-operator-names
1398 Do not treat the operator name keywords @code{and}, @code{bitand},
1399 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1400 synonyms as keywords.
1402 @item -fno-optional-diags
1403 @opindex fno-optional-diags
1404 Disable diagnostics that the standard says a compiler does not need to
1405 issue. Currently, the only such diagnostic issued by G++ is the one for
1406 a name having multiple meanings within a class.
1409 @opindex fpermissive
1410 Downgrade some diagnostics about nonconformant code from errors to
1411 warnings. Thus, using @option{-fpermissive} will allow some
1412 nonconforming code to compile.
1416 Enable automatic template instantiation at link time. This option also
1417 implies @option{-fno-implicit-templates}. @xref{Template
1418 Instantiation}, for more information.
1422 Disable generation of information about every class with virtual
1423 functions for use by the C++ runtime type identification features
1424 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1425 of the language, you can save some space by using this flag. Note that
1426 exception handling uses the same information, but it will generate it as
1431 Emit statistics about front-end processing at the end of the compilation.
1432 This information is generally only useful to the G++ development team.
1434 @item -ftemplate-depth-@var{n}
1435 @opindex ftemplate-depth
1436 Set the maximum instantiation depth for template classes to @var{n}.
1437 A limit on the template instantiation depth is needed to detect
1438 endless recursions during template class instantiation. ANSI/ISO C++
1439 conforming programs must not rely on a maximum depth greater than 17.
1441 @item -fuse-cxa-atexit
1442 @opindex fuse-cxa-atexit
1443 Register destructors for objects with static storage duration with the
1444 @code{__cxa_atexit} function rather than the @code{atexit} function.
1445 This option is required for fully standards-compliant handling of static
1446 destructors, but will only work if your C library supports
1447 @code{__cxa_atexit}.
1451 Do not use weak symbol support, even if it is provided by the linker.
1452 By default, G++ will use weak symbols if they are available. This
1453 option exists only for testing, and should not be used by end-users;
1454 it will result in inferior code and has no benefits. This option may
1455 be removed in a future release of G++.
1459 Do not search for header files in the standard directories specific to
1460 C++, but do still search the other standard directories. (This option
1461 is used when building the C++ library.)
1464 In addition, these optimization, warning, and code generation options
1465 have meanings only for C++ programs:
1468 @item -fno-default-inline
1469 @opindex fno-default-inline
1470 Do not assume @samp{inline} for functions defined inside a class scope.
1471 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1472 functions will have linkage like inline functions; they just won't be
1475 @item -Wabi @r{(C++ only)}
1477 Warn when G++ generates code that is probably not compatible with the
1478 vendor-neutral C++ ABI. Although an effort has been made to warn about
1479 all such cases, there are probably some cases that are not warned about,
1480 even though G++ is generating incompatible code. There may also be
1481 cases where warnings are emitted even though the code that is generated
1484 You should rewrite your code to avoid these warnings if you are
1485 concerned about the fact that code generated by G++ may not be binary
1486 compatible with code generated by other compilers.
1488 The known incompatibilities at this point include:
1493 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
1494 pack data into the same byte as a base class. For example:
1497 struct A @{ virtual void f(); int f1 : 1; @};
1498 struct B : public A @{ int f2 : 1; @};
1502 In this case, G++ will place @code{B::f2} into the same byte
1503 as@code{A::f1}; other compilers will not. You can avoid this problem
1504 by explicitly padding @code{A} so that its size is a multiple of the
1505 byte size on your platform; that will cause G++ and other compilers to
1506 layout @code{B} identically.
1509 Incorrect handling of tail-padding for virtual bases. G++ does not use
1510 tail padding when laying out virtual bases. For example:
1513 struct A @{ virtual void f(); char c1; @};
1514 struct B @{ B(); char c2; @};
1515 struct C : public A, public virtual B @{@};
1519 In this case, G++ will not place @code{B} into the tail-padding for
1520 @code{A}; other compilers will. You can avoid this problem by
1521 explicitly padding @code{A} so that its size is a multiple of its
1522 alignment (ignoring virtual base classes); that will cause G++ and other
1523 compilers to layout @code{C} identically.
1526 Incorrect handling of bit-fields with declared widths greater than that
1527 of their underlying types, when the bit-fields appear in a union. For
1531 union U @{ int i : 4096; @};
1535 Assuming that an @code{int} does not have 4096 bits, G++ will make the
1536 union too small by the number of bits in an @code{int}.
1539 Empty classes can be placed at incorrect offsets. For example:
1549 struct C : public B, public A @{@};
1553 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
1554 it should be placed at offset zero. G++ mistakenly believes that the
1555 @code{A} data member of @code{B} is already at offset zero.
1558 Names of template functions whose types involve @code{typename} or
1559 template template parameters can be mangled incorrectly.
1562 template <typename Q>
1563 void f(typename Q::X) @{@}
1565 template <template <typename> class Q>
1566 void f(typename Q<int>::X) @{@}
1570 Instantiations of these templates may be mangled incorrectly.
1574 @item -Wctor-dtor-privacy @r{(C++ only)}
1575 @opindex Wctor-dtor-privacy
1576 Warn when a class seems unusable because all the constructors or
1577 destructors in that class are private, and it has neither friends nor
1578 public static member functions.
1580 @item -Wnon-virtual-dtor @r{(C++ only)}
1581 @opindex Wnon-virtual-dtor
1582 Warn when a class appears to be polymorphic, thereby requiring a virtual
1583 destructor, yet it declares a non-virtual one.
1584 This warning is enabled by @option{-Wall}.
1586 @item -Wreorder @r{(C++ only)}
1588 @cindex reordering, warning
1589 @cindex warning for reordering of member initializers
1590 Warn when the order of member initializers given in the code does not
1591 match the order in which they must be executed. For instance:
1597 A(): j (0), i (1) @{ @}
1601 The compiler will rearrange the member initializers for @samp{i}
1602 and @samp{j} to match the declaration order of the members, emitting
1603 a warning to that effect. This warning is enabled by @option{-Wall}.
1606 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
1609 @item -Weffc++ @r{(C++ only)}
1611 Warn about violations of the following style guidelines from Scott Meyers'
1612 @cite{Effective C++} book:
1616 Item 11: Define a copy constructor and an assignment operator for classes
1617 with dynamically allocated memory.
1620 Item 12: Prefer initialization to assignment in constructors.
1623 Item 14: Make destructors virtual in base classes.
1626 Item 15: Have @code{operator=} return a reference to @code{*this}.
1629 Item 23: Don't try to return a reference when you must return an object.
1633 Also warn about violations of the following style guidelines from
1634 Scott Meyers' @cite{More Effective C++} book:
1638 Item 6: Distinguish between prefix and postfix forms of increment and
1639 decrement operators.
1642 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
1646 When selecting this option, be aware that the standard library
1647 headers do not obey all of these guidelines; use @samp{grep -v}
1648 to filter out those warnings.
1650 @item -Wno-deprecated @r{(C++ only)}
1651 @opindex Wno-deprecated
1652 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1654 @item -Wno-non-template-friend @r{(C++ only)}
1655 @opindex Wno-non-template-friend
1656 Disable warnings when non-templatized friend functions are declared
1657 within a template. Since the advent of explicit template specification
1658 support in G++, if the name of the friend is an unqualified-id (i.e.,
1659 @samp{friend foo(int)}), the C++ language specification demands that the
1660 friend declare or define an ordinary, nontemplate function. (Section
1661 14.5.3). Before G++ implemented explicit specification, unqualified-ids
1662 could be interpreted as a particular specialization of a templatized
1663 function. Because this non-conforming behavior is no longer the default
1664 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
1665 check existing code for potential trouble spots and is on by default.
1666 This new compiler behavior can be turned off with
1667 @option{-Wno-non-template-friend} which keeps the conformant compiler code
1668 but disables the helpful warning.
1670 @item -Wold-style-cast @r{(C++ only)}
1671 @opindex Wold-style-cast
1672 Warn if an old-style (C-style) cast to a non-void type is used within
1673 a C++ program. The new-style casts (@samp{static_cast},
1674 @samp{reinterpret_cast}, and @samp{const_cast}) are less vulnerable to
1675 unintended effects and much easier to search for.
1677 @item -Woverloaded-virtual @r{(C++ only)}
1678 @opindex Woverloaded-virtual
1679 @cindex overloaded virtual fn, warning
1680 @cindex warning for overloaded virtual fn
1681 Warn when a function declaration hides virtual functions from a
1682 base class. For example, in:
1689 struct B: public A @{
1694 the @code{A} class version of @code{f} is hidden in @code{B}, and code
1702 will fail to compile.
1704 @item -Wno-pmf-conversions @r{(C++ only)}
1705 @opindex Wno-pmf-conversions
1706 Disable the diagnostic for converting a bound pointer to member function
1709 @item -Wsign-promo @r{(C++ only)}
1710 @opindex Wsign-promo
1711 Warn when overload resolution chooses a promotion from unsigned or
1712 enumeral type to a signed type, over a conversion to an unsigned type of
1713 the same size. Previous versions of G++ would try to preserve
1714 unsignedness, but the standard mandates the current behavior.
1716 @item -Wsynth @r{(C++ only)}
1718 @cindex warning for synthesized methods
1719 @cindex synthesized methods, warning
1720 Warn when G++'s synthesis behavior does not match that of cfront. For
1726 A& operator = (int);
1736 In this example, G++ will synthesize a default @samp{A& operator =
1737 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1740 @node Objective-C Dialect Options
1741 @section Options Controlling Objective-C Dialect
1743 @cindex compiler options, Objective-C
1744 @cindex Objective-C options, command line
1745 @cindex options, Objective-C
1746 (NOTE: This manual does not describe the Objective-C language itself. See
1747 @w{@uref{http://gcc.gnu.org/readings.html}} for references.)
1749 This section describes the command-line options that are only meaningful
1750 for Objective-C programs, but you can also use most of the GNU compiler
1751 options regardless of what language your program is in. For example,
1752 you might compile a file @code{some_class.m} like this:
1755 gcc -g -fgnu-runtime -O -c some_class.m
1759 In this example, @option{-fgnu-runtime} is an option meant only for
1760 Objective-C programs; you can use the other options with any language
1763 Here is a list of options that are @emph{only} for compiling Objective-C
1767 @item -fconstant-string-class=@var{class-name}
1768 @opindex fconstant-string-class
1769 Use @var{class-name} as the name of the class to instantiate for each
1770 literal string specified with the syntax @code{@@"@dots{}"}. The default
1771 class name is @code{NXConstantString} if the GNU runtime is being used, and
1772 @code{NSConstantString} if the NeXT runtime is being used (see below). The
1773 @option{-fconstant-cfstrings} option, if also present, will override the
1774 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
1775 to be laid out as constant CoreFoundation strings.
1778 @opindex fgnu-runtime
1779 Generate object code compatible with the standard GNU Objective-C
1780 runtime. This is the default for most types of systems.
1782 @item -fnext-runtime
1783 @opindex fnext-runtime
1784 Generate output compatible with the NeXT runtime. This is the default
1785 for NeXT-based systems, including Darwin and Mac OS X@. The macro
1786 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
1789 @item -fno-nil-receivers
1790 @opindex fno-nil-receivers
1791 Assume that all Objective-C message dispatches (e.g.,
1792 @code{[receiver message:arg]}) in this translation unit ensure that the receiver
1793 is not @code{nil}. This allows for more efficient entry points in the runtime to be
1794 used. Currently, this option is only available in conjunction with
1795 the NeXT runtime on Mac OS X 10.3 and later.
1797 @item -fobjc-exceptions
1798 @opindex fobjc-exceptions
1799 Enable syntactic support for structured exception handling in Objective-C,
1800 similar to what is offered by C++ and Java. Currently, this option is only
1801 available in conjunction with the NeXT runtime on Mac OS X 10.3 and later.
1809 @@catch (AnObjCClass *exc) @{
1816 @@catch (AnotherClass *exc) @{
1819 @@catch (id allOthers) @{
1829 The @code{@@throw} statement may appear anywhere in an Objective-C or
1830 Objective-C++ program; when used inside of a @code{@@catch} block, the
1831 @code{@@throw} may appear without an argument (as shown above), in which case
1832 the object caught by the @code{@@catch} will be rethrown.
1834 Note that only (pointers to) Objective-C objects may be thrown and
1835 caught using this scheme. When an object is thrown, it will be caught
1836 by the nearest @code{@@catch} clause capable of handling objects of that type,
1837 analogously to how @code{catch} blocks work in C++ and Java. A
1838 @code{@@catch(id @dots{})} clause (as shown above) may also be provided to catch
1839 any and all Objective-C exceptions not caught by previous @code{@@catch}
1842 The @code{@@finally} clause, if present, will be executed upon exit from the
1843 immediately preceding @code{@@try @dots{} @@catch} section. This will happen
1844 regardless of whether any exceptions are thrown, caught or rethrown
1845 inside the @code{@@try @dots{} @@catch} section, analogously to the behavior
1846 of the @code{finally} clause in Java.
1848 There are several caveats to using the new exception mechanism:
1852 Although currently designed to be binary compatible with @code{NS_HANDLER}-style
1853 idioms provided by the @code{NSException} class, the new
1854 exceptions can only be used on Mac OS X 10.3 (Panther) and later
1855 systems, due to additional functionality needed in the (NeXT) Objective-C
1859 As mentioned above, the new exceptions do not support handling
1860 types other than Objective-C objects. Furthermore, when used from
1861 Objective-C++, the Objective-C exception model does not interoperate with C++
1862 exceptions at this time. This means you cannot @code{@@throw} an exception
1863 from Objective-C and @code{catch} it in C++, or vice versa
1864 (i.e., @code{throw @dots{} @@catch}).
1867 The @option{-fobjc-exceptions} switch also enables the use of synchronization
1868 blocks for thread-safe execution:
1871 @@synchronized (ObjCClass *guard) @{
1876 Upon entering the @code{@@synchronized} block, a thread of execution shall
1877 first check whether a lock has been placed on the corresponding @code{guard}
1878 object by another thread. If it has, the current thread shall wait until
1879 the other thread relinquishes its lock. Once @code{guard} becomes available,
1880 the current thread will place its own lock on it, execute the code contained in
1881 the @code{@@synchronized} block, and finally relinquish the lock (thereby
1882 making @code{guard} available to other threads).
1884 Unlike Java, Objective-C does not allow for entire methods to be marked
1885 @code{@@synchronized}. Note that throwing exceptions out of
1886 @code{@@synchronized} blocks is allowed, and will cause the guarding object
1887 to be unlocked properly.
1889 @item -freplace-objc-classes
1890 @opindex freplace-objc-classes
1891 Emit a special marker instructing @command{ld(1)} not to statically link in
1892 the resulting object file, and allow @command{dyld(1)} to load it in at
1893 run time instead. This is used in conjunction with the Fix-and-Continue
1894 debugging mode, where the object file in question may be recompiled and
1895 dynamically reloaded in the course of program execution, without the need
1896 to restart the program itself. Currently, Fix-and-Continue functionality
1897 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
1902 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
1903 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
1904 compile time) with static class references that get initialized at load time,
1905 which improves run-time performance. Specifying the @option{-fzero-link} flag
1906 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
1907 to be retained. This is useful in Zero-Link debugging mode, since it allows
1908 for individual class implementations to be modified during program execution.
1912 Dump interface declarations for all classes seen in the source file to a
1913 file named @file{@var{sourcename}.decl}.
1916 @opindex Wno-protocol
1917 If a class is declared to implement a protocol, a warning is issued for
1918 every method in the protocol that is not implemented by the class. The
1919 default behavior is to issue a warning for every method not explicitly
1920 implemented in the class, even if a method implementation is inherited
1921 from the superclass. If you use the @code{-Wno-protocol} option, then
1922 methods inherited from the superclass are considered to be implemented,
1923 and no warning is issued for them.
1927 Warn if multiple methods of different types for the same selector are
1928 found during compilation. The check is performed on the list of methods
1929 in the final stage of compilation. Additionally, a check is performed
1930 for each selector appearing in a @code{@@selector(@dots{})}
1931 expression, and a corresponding method for that selector has been found
1932 during compilation. Because these checks scan the method table only at
1933 the end of compilation, these warnings are not produced if the final
1934 stage of compilation is not reached, for example because an error is
1935 found during compilation, or because the @code{-fsyntax-only} option is
1938 @item -Wundeclared-selector
1939 @opindex Wundeclared-selector
1940 Warn if a @code{@@selector(@dots{})} expression referring to an
1941 undeclared selector is found. A selector is considered undeclared if no
1942 method with that name has been declared before the
1943 @code{@@selector(@dots{})} expression, either explicitly in an
1944 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
1945 an @code{@@implementation} section. This option always performs its
1946 checks as soon as a @code{@@selector(@dots{})} expression is found,
1947 while @code{-Wselector} only performs its checks in the final stage of
1948 compilation. This also enforces the coding style convention
1949 that methods and selectors must be declared before being used.
1951 @item -print-objc-runtime-info
1952 @opindex print-objc-runtime-info
1953 Generate C header describing the largest structure that is passed by
1958 @node Language Independent Options
1959 @section Options to Control Diagnostic Messages Formatting
1960 @cindex options to control diagnostics formatting
1961 @cindex diagnostic messages
1962 @cindex message formatting
1964 Traditionally, diagnostic messages have been formatted irrespective of
1965 the output device's aspect (e.g.@: its width, @dots{}). The options described
1966 below can be used to control the diagnostic messages formatting
1967 algorithm, e.g.@: how many characters per line, how often source location
1968 information should be reported. Right now, only the C++ front end can
1969 honor these options. However it is expected, in the near future, that
1970 the remaining front ends would be able to digest them correctly.
1973 @item -fmessage-length=@var{n}
1974 @opindex fmessage-length
1975 Try to format error messages so that they fit on lines of about @var{n}
1976 characters. The default is 72 characters for @command{g++} and 0 for the rest of
1977 the front ends supported by GCC@. If @var{n} is zero, then no
1978 line-wrapping will be done; each error message will appear on a single
1981 @opindex fdiagnostics-show-location
1982 @item -fdiagnostics-show-location=once
1983 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
1984 reporter to emit @emph{once} source location information; that is, in
1985 case the message is too long to fit on a single physical line and has to
1986 be wrapped, the source location won't be emitted (as prefix) again,
1987 over and over, in subsequent continuation lines. This is the default
1990 @item -fdiagnostics-show-location=every-line
1991 Only meaningful in line-wrapping mode. Instructs the diagnostic
1992 messages reporter to emit the same source location information (as
1993 prefix) for physical lines that result from the process of breaking
1994 a message which is too long to fit on a single line.
1998 @node Warning Options
1999 @section Options to Request or Suppress Warnings
2000 @cindex options to control warnings
2001 @cindex warning messages
2002 @cindex messages, warning
2003 @cindex suppressing warnings
2005 Warnings are diagnostic messages that report constructions which
2006 are not inherently erroneous but which are risky or suggest there
2007 may have been an error.
2009 You can request many specific warnings with options beginning @samp{-W},
2010 for example @option{-Wimplicit} to request warnings on implicit
2011 declarations. Each of these specific warning options also has a
2012 negative form beginning @samp{-Wno-} to turn off warnings;
2013 for example, @option{-Wno-implicit}. This manual lists only one of the
2014 two forms, whichever is not the default.
2016 The following options control the amount and kinds of warnings produced
2017 by GCC; for further, language-specific options also refer to
2018 @ref{C++ Dialect Options} and @ref{Objective-C Dialect Options}.
2021 @cindex syntax checking
2023 @opindex fsyntax-only
2024 Check the code for syntax errors, but don't do anything beyond that.
2028 Issue all the warnings demanded by strict ISO C and ISO C++;
2029 reject all programs that use forbidden extensions, and some other
2030 programs that do not follow ISO C and ISO C++. For ISO C, follows the
2031 version of the ISO C standard specified by any @option{-std} option used.
2033 Valid ISO C and ISO C++ programs should compile properly with or without
2034 this option (though a rare few will require @option{-ansi} or a
2035 @option{-std} option specifying the required version of ISO C)@. However,
2036 without this option, certain GNU extensions and traditional C and C++
2037 features are supported as well. With this option, they are rejected.
2039 @option{-pedantic} does not cause warning messages for use of the
2040 alternate keywords whose names begin and end with @samp{__}. Pedantic
2041 warnings are also disabled in the expression that follows
2042 @code{__extension__}. However, only system header files should use
2043 these escape routes; application programs should avoid them.
2044 @xref{Alternate Keywords}.
2046 Some users try to use @option{-pedantic} to check programs for strict ISO
2047 C conformance. They soon find that it does not do quite what they want:
2048 it finds some non-ISO practices, but not all---only those for which
2049 ISO C @emph{requires} a diagnostic, and some others for which
2050 diagnostics have been added.
2052 A feature to report any failure to conform to ISO C might be useful in
2053 some instances, but would require considerable additional work and would
2054 be quite different from @option{-pedantic}. We don't have plans to
2055 support such a feature in the near future.
2057 Where the standard specified with @option{-std} represents a GNU
2058 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
2059 corresponding @dfn{base standard}, the version of ISO C on which the GNU
2060 extended dialect is based. Warnings from @option{-pedantic} are given
2061 where they are required by the base standard. (It would not make sense
2062 for such warnings to be given only for features not in the specified GNU
2063 C dialect, since by definition the GNU dialects of C include all
2064 features the compiler supports with the given option, and there would be
2065 nothing to warn about.)
2067 @item -pedantic-errors
2068 @opindex pedantic-errors
2069 Like @option{-pedantic}, except that errors are produced rather than
2074 Inhibit all warning messages.
2078 Inhibit warning messages about the use of @samp{#import}.
2080 @item -Wchar-subscripts
2081 @opindex Wchar-subscripts
2082 Warn if an array subscript has type @code{char}. This is a common cause
2083 of error, as programmers often forget that this type is signed on some
2088 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
2089 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
2091 @item -Wfatal-errors
2092 @opindex Wfatal-errors
2093 This option causes the compiler to abort compilation on the first error
2094 occurred rather than trying to keep going and printing further error
2099 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
2100 the arguments supplied have types appropriate to the format string
2101 specified, and that the conversions specified in the format string make
2102 sense. This includes standard functions, and others specified by format
2103 attributes (@pxref{Function Attributes}), in the @code{printf},
2104 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
2105 not in the C standard) families.
2107 The formats are checked against the format features supported by GNU
2108 libc version 2.2. These include all ISO C90 and C99 features, as well
2109 as features from the Single Unix Specification and some BSD and GNU
2110 extensions. Other library implementations may not support all these
2111 features; GCC does not support warning about features that go beyond a
2112 particular library's limitations. However, if @option{-pedantic} is used
2113 with @option{-Wformat}, warnings will be given about format features not
2114 in the selected standard version (but not for @code{strfmon} formats,
2115 since those are not in any version of the C standard). @xref{C Dialect
2116 Options,,Options Controlling C Dialect}.
2118 Since @option{-Wformat} also checks for null format arguments for
2119 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
2121 @option{-Wformat} is included in @option{-Wall}. For more control over some
2122 aspects of format checking, the options @option{-Wformat-y2k},
2123 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
2124 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
2125 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
2128 @opindex Wformat-y2k
2129 If @option{-Wformat} is specified, also warn about @code{strftime}
2130 formats which may yield only a two-digit year.
2132 @item -Wno-format-extra-args
2133 @opindex Wno-format-extra-args
2134 If @option{-Wformat} is specified, do not warn about excess arguments to a
2135 @code{printf} or @code{scanf} format function. The C standard specifies
2136 that such arguments are ignored.
2138 Where the unused arguments lie between used arguments that are
2139 specified with @samp{$} operand number specifications, normally
2140 warnings are still given, since the implementation could not know what
2141 type to pass to @code{va_arg} to skip the unused arguments. However,
2142 in the case of @code{scanf} formats, this option will suppress the
2143 warning if the unused arguments are all pointers, since the Single
2144 Unix Specification says that such unused arguments are allowed.
2146 @item -Wno-format-zero-length
2147 @opindex Wno-format-zero-length
2148 If @option{-Wformat} is specified, do not warn about zero-length formats.
2149 The C standard specifies that zero-length formats are allowed.
2151 @item -Wformat-nonliteral
2152 @opindex Wformat-nonliteral
2153 If @option{-Wformat} is specified, also warn if the format string is not a
2154 string literal and so cannot be checked, unless the format function
2155 takes its format arguments as a @code{va_list}.
2157 @item -Wformat-security
2158 @opindex Wformat-security
2159 If @option{-Wformat} is specified, also warn about uses of format
2160 functions that represent possible security problems. At present, this
2161 warns about calls to @code{printf} and @code{scanf} functions where the
2162 format string is not a string literal and there are no format arguments,
2163 as in @code{printf (foo);}. This may be a security hole if the format
2164 string came from untrusted input and contains @samp{%n}. (This is
2165 currently a subset of what @option{-Wformat-nonliteral} warns about, but
2166 in future warnings may be added to @option{-Wformat-security} that are not
2167 included in @option{-Wformat-nonliteral}.)
2171 Enable @option{-Wformat} plus format checks not included in
2172 @option{-Wformat}. Currently equivalent to @samp{-Wformat
2173 -Wformat-nonliteral -Wformat-security -Wformat-y2k}.
2177 Warn about passing a null pointer for arguments marked as
2178 requiring a non-null value by the @code{nonnull} function attribute.
2180 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
2181 can be disabled with the @option{-Wno-nonnull} option.
2183 @item -Winit-self @r{(C, C++, and Objective-C only)}
2185 Warn about uninitialized variables which are initialized with themselves.
2186 Note this option can only be used with the @option{-Wuninitialized} option,
2187 which in turn only works with @option{-O1} and above.
2189 For example, GCC will warn about @code{i} being uninitialized in the
2190 following snippet only when @option{-Winit-self} has been specified:
2201 @item -Wimplicit-int
2202 @opindex Wimplicit-int
2203 Warn when a declaration does not specify a type.
2205 @item -Wimplicit-function-declaration
2206 @itemx -Werror-implicit-function-declaration
2207 @opindex Wimplicit-function-declaration
2208 @opindex Werror-implicit-function-declaration
2209 Give a warning (or error) whenever a function is used before being
2214 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
2218 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
2219 function with external linkage, returning int, taking either zero
2220 arguments, two, or three arguments of appropriate types.
2222 @item -Wmissing-braces
2223 @opindex Wmissing-braces
2224 Warn if an aggregate or union initializer is not fully bracketed. In
2225 the following example, the initializer for @samp{a} is not fully
2226 bracketed, but that for @samp{b} is fully bracketed.
2229 int a[2][2] = @{ 0, 1, 2, 3 @};
2230 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
2234 @opindex Wparentheses
2235 Warn if parentheses are omitted in certain contexts, such
2236 as when there is an assignment in a context where a truth value
2237 is expected, or when operators are nested whose precedence people
2238 often get confused about.
2240 Also warn about constructions where there may be confusion to which
2241 @code{if} statement an @code{else} branch belongs. Here is an example of
2256 In C, every @code{else} branch belongs to the innermost possible @code{if}
2257 statement, which in this example is @code{if (b)}. This is often not
2258 what the programmer expected, as illustrated in the above example by
2259 indentation the programmer chose. When there is the potential for this
2260 confusion, GCC will issue a warning when this flag is specified.
2261 To eliminate the warning, add explicit braces around the innermost
2262 @code{if} statement so there is no way the @code{else} could belong to
2263 the enclosing @code{if}. The resulting code would look like this:
2279 @item -Wsequence-point
2280 @opindex Wsequence-point
2281 Warn about code that may have undefined semantics because of violations
2282 of sequence point rules in the C standard.
2284 The C standard defines the order in which expressions in a C program are
2285 evaluated in terms of @dfn{sequence points}, which represent a partial
2286 ordering between the execution of parts of the program: those executed
2287 before the sequence point, and those executed after it. These occur
2288 after the evaluation of a full expression (one which is not part of a
2289 larger expression), after the evaluation of the first operand of a
2290 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
2291 function is called (but after the evaluation of its arguments and the
2292 expression denoting the called function), and in certain other places.
2293 Other than as expressed by the sequence point rules, the order of
2294 evaluation of subexpressions of an expression is not specified. All
2295 these rules describe only a partial order rather than a total order,
2296 since, for example, if two functions are called within one expression
2297 with no sequence point between them, the order in which the functions
2298 are called is not specified. However, the standards committee have
2299 ruled that function calls do not overlap.
2301 It is not specified when between sequence points modifications to the
2302 values of objects take effect. Programs whose behavior depends on this
2303 have undefined behavior; the C standard specifies that ``Between the
2304 previous and next sequence point an object shall have its stored value
2305 modified at most once by the evaluation of an expression. Furthermore,
2306 the prior value shall be read only to determine the value to be
2307 stored.''. If a program breaks these rules, the results on any
2308 particular implementation are entirely unpredictable.
2310 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
2311 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
2312 diagnosed by this option, and it may give an occasional false positive
2313 result, but in general it has been found fairly effective at detecting
2314 this sort of problem in programs.
2316 The present implementation of this option only works for C programs. A
2317 future implementation may also work for C++ programs.
2319 The C standard is worded confusingly, therefore there is some debate
2320 over the precise meaning of the sequence point rules in subtle cases.
2321 Links to discussions of the problem, including proposed formal
2322 definitions, may be found on our readings page, at
2323 @w{@uref{http://gcc.gnu.org/readings.html}}.
2326 @opindex Wreturn-type
2327 Warn whenever a function is defined with a return-type that defaults to
2328 @code{int}. Also warn about any @code{return} statement with no
2329 return-value in a function whose return-type is not @code{void}.
2331 For C++, a function without return type always produces a diagnostic
2332 message, even when @option{-Wno-return-type} is specified. The only
2333 exceptions are @samp{main} and functions defined in system headers.
2337 Warn whenever a @code{switch} statement has an index of enumeral type
2338 and lacks a @code{case} for one or more of the named codes of that
2339 enumeration. (The presence of a @code{default} label prevents this
2340 warning.) @code{case} labels outside the enumeration range also
2341 provoke warnings when this option is used.
2343 @item -Wswitch-default
2344 @opindex Wswitch-switch
2345 Warn whenever a @code{switch} statement does not have a @code{default}
2349 @opindex Wswitch-enum
2350 Warn whenever a @code{switch} statement has an index of enumeral type
2351 and lacks a @code{case} for one or more of the named codes of that
2352 enumeration. @code{case} labels outside the enumeration range also
2353 provoke warnings when this option is used.
2357 Warn if any trigraphs are encountered that might change the meaning of
2358 the program (trigraphs within comments are not warned about).
2360 @item -Wunused-function
2361 @opindex Wunused-function
2362 Warn whenever a static function is declared but not defined or a
2363 non\-inline static function is unused.
2365 @item -Wunused-label
2366 @opindex Wunused-label
2367 Warn whenever a label is declared but not used.
2369 To suppress this warning use the @samp{unused} attribute
2370 (@pxref{Variable Attributes}).
2372 @item -Wunused-parameter
2373 @opindex Wunused-parameter
2374 Warn whenever a function parameter is unused aside from its declaration.
2376 To suppress this warning use the @samp{unused} attribute
2377 (@pxref{Variable Attributes}).
2379 @item -Wunused-variable
2380 @opindex Wunused-variable
2381 Warn whenever a local variable or non-constant static variable is unused
2382 aside from its declaration
2384 To suppress this warning use the @samp{unused} attribute
2385 (@pxref{Variable Attributes}).
2387 @item -Wunused-value
2388 @opindex Wunused-value
2389 Warn whenever a statement computes a result that is explicitly not used.
2391 To suppress this warning cast the expression to @samp{void}.
2395 All the above @option{-Wunused} options combined.
2397 In order to get a warning about an unused function parameter, you must
2398 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
2399 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
2401 @item -Wuninitialized
2402 @opindex Wuninitialized
2403 Warn if an automatic variable is used without first being initialized or
2404 if a variable may be clobbered by a @code{setjmp} call.
2406 These warnings are possible only in optimizing compilation,
2407 because they require data flow information that is computed only
2408 when optimizing. If you don't specify @option{-O}, you simply won't
2411 If you want to warn about code which uses the uninitialized value of the
2412 variable in its own initializer, use the @option{-Winit-self} option.
2414 These warnings occur only for variables that are candidates for
2415 register allocation. Therefore, they do not occur for a variable that
2416 is declared @code{volatile}, or whose address is taken, or whose size
2417 is other than 1, 2, 4 or 8 bytes. Also, they do not occur for
2418 structures, unions or arrays, even when they are in registers.
2420 Note that there may be no warning about a variable that is used only
2421 to compute a value that itself is never used, because such
2422 computations may be deleted by data flow analysis before the warnings
2425 These warnings are made optional because GCC is not smart
2426 enough to see all the reasons why the code might be correct
2427 despite appearing to have an error. Here is one example of how
2448 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
2449 always initialized, but GCC doesn't know this. Here is
2450 another common case:
2455 if (change_y) save_y = y, y = new_y;
2457 if (change_y) y = save_y;
2462 This has no bug because @code{save_y} is used only if it is set.
2464 @cindex @code{longjmp} warnings
2465 This option also warns when a non-volatile automatic variable might be
2466 changed by a call to @code{longjmp}. These warnings as well are possible
2467 only in optimizing compilation.
2469 The compiler sees only the calls to @code{setjmp}. It cannot know
2470 where @code{longjmp} will be called; in fact, a signal handler could
2471 call it at any point in the code. As a result, you may get a warning
2472 even when there is in fact no problem because @code{longjmp} cannot
2473 in fact be called at the place which would cause a problem.
2475 Some spurious warnings can be avoided if you declare all the functions
2476 you use that never return as @code{noreturn}. @xref{Function
2479 @item -Wunknown-pragmas
2480 @opindex Wunknown-pragmas
2481 @cindex warning for unknown pragmas
2482 @cindex unknown pragmas, warning
2483 @cindex pragmas, warning of unknown
2484 Warn when a #pragma directive is encountered which is not understood by
2485 GCC@. If this command line option is used, warnings will even be issued
2486 for unknown pragmas in system header files. This is not the case if
2487 the warnings were only enabled by the @option{-Wall} command line option.
2489 @item -Wstrict-aliasing
2490 @opindex Wstrict-aliasing
2491 This option is only active when @option{-fstrict-aliasing} is active.
2492 It warns about code which might break the strict aliasing rules that the
2493 compiler is using for optimization. The warning does not catch all
2494 cases, but does attempt to catch the more common pitfalls. It is
2495 included in @option{-Wall}.
2497 @item -Wstrict-aliasing=2
2498 @opindex Wstrict-aliasing=2
2499 This option is only active when @option{-fstrict-aliasing} is active.
2500 It warns about all code which might break the strict aliasing rules that the
2501 compiler is using for optimization. This warning catches all cases, but
2502 it will also give a warning for some ambiguous cases that are safe.
2506 All of the above @samp{-W} options combined. This enables all the
2507 warnings about constructions that some users consider questionable, and
2508 that are easy to avoid (or modify to prevent the warning), even in
2509 conjunction with macros. This also enables some language-specific
2510 warnings described in @ref{C++ Dialect Options} and
2511 @ref{Objective-C Dialect Options}.
2514 The following @option{-W@dots{}} options are not implied by @option{-Wall}.
2515 Some of them warn about constructions that users generally do not
2516 consider questionable, but which occasionally you might wish to check
2517 for; others warn about constructions that are necessary or hard to avoid
2518 in some cases, and there is no simple way to modify the code to suppress
2525 (This option used to be called @option{-W}. The older name is still
2526 supported, but the newer name is more descriptive.) Print extra warning
2527 messages for these events:
2531 A function can return either with or without a value. (Falling
2532 off the end of the function body is considered returning without
2533 a value.) For example, this function would evoke such a
2547 An expression-statement or the left-hand side of a comma expression
2548 contains no side effects.
2549 To suppress the warning, cast the unused expression to void.
2550 For example, an expression such as @samp{x[i,j]} will cause a warning,
2551 but @samp{x[(void)i,j]} will not.
2554 An unsigned value is compared against zero with @samp{<} or @samp{>=}.
2557 A comparison like @samp{x<=y<=z} appears; this is equivalent to
2558 @samp{(x<=y ? 1 : 0) <= z}, which is a different interpretation from
2559 that of ordinary mathematical notation.
2562 Storage-class specifiers like @code{static} are not the first things in
2563 a declaration. According to the C Standard, this usage is obsolescent.
2566 The return type of a function has a type qualifier such as @code{const}.
2567 Such a type qualifier has no effect, since the value returned by a
2568 function is not an lvalue. (But don't warn about the GNU extension of
2569 @code{volatile void} return types. That extension will be warned about
2570 if @option{-pedantic} is specified.)
2573 If @option{-Wall} or @option{-Wunused} is also specified, warn about unused
2577 A comparison between signed and unsigned values could produce an
2578 incorrect result when the signed value is converted to unsigned.
2579 (But don't warn if @option{-Wno-sign-compare} is also specified.)
2582 An aggregate has an initializer which does not initialize all members.
2583 For example, the following code would cause such a warning, because
2584 @code{x.h} would be implicitly initialized to zero:
2587 struct s @{ int f, g, h; @};
2588 struct s x = @{ 3, 4 @};
2592 A function parameter is declared without a type specifier in K&R-style
2600 An empty body occurs in an @samp{if} or @samp{else} statement.
2603 A pointer is compared against integer zero with @samp{<}, @samp{<=},
2604 @samp{>}, or @samp{>=}.
2607 A variable might be changed by @samp{longjmp} or @samp{vfork}.
2610 Any of several floating-point events that often indicate errors, such as
2611 overflow, underflow, loss of precision, etc.
2613 @item @r{(C++ only)}
2614 An enumerator and a non-enumerator both appear in a conditional expression.
2616 @item @r{(C++ only)}
2617 A non-static reference or non-static @samp{const} member appears in a
2618 class without constructors.
2620 @item @r{(C++ only)}
2621 Ambiguous virtual bases.
2623 @item @r{(C++ only)}
2624 Subscripting an array which has been declared @samp{register}.
2626 @item @r{(C++ only)}
2627 Taking the address of a variable which has been declared @samp{register}.
2629 @item @r{(C++ only)}
2630 A base class is not initialized in a derived class' copy constructor.
2633 @item -Wno-div-by-zero
2634 @opindex Wno-div-by-zero
2635 @opindex Wdiv-by-zero
2636 Do not warn about compile-time integer division by zero. Floating point
2637 division by zero is not warned about, as it can be a legitimate way of
2638 obtaining infinities and NaNs.
2640 @item -Wsystem-headers
2641 @opindex Wsystem-headers
2642 @cindex warnings from system headers
2643 @cindex system headers, warnings from
2644 Print warning messages for constructs found in system header files.
2645 Warnings from system headers are normally suppressed, on the assumption
2646 that they usually do not indicate real problems and would only make the
2647 compiler output harder to read. Using this command line option tells
2648 GCC to emit warnings from system headers as if they occurred in user
2649 code. However, note that using @option{-Wall} in conjunction with this
2650 option will @emph{not} warn about unknown pragmas in system
2651 headers---for that, @option{-Wunknown-pragmas} must also be used.
2654 @opindex Wfloat-equal
2655 Warn if floating point values are used in equality comparisons.
2657 The idea behind this is that sometimes it is convenient (for the
2658 programmer) to consider floating-point values as approximations to
2659 infinitely precise real numbers. If you are doing this, then you need
2660 to compute (by analyzing the code, or in some other way) the maximum or
2661 likely maximum error that the computation introduces, and allow for it
2662 when performing comparisons (and when producing output, but that's a
2663 different problem). In particular, instead of testing for equality, you
2664 would check to see whether the two values have ranges that overlap; and
2665 this is done with the relational operators, so equality comparisons are
2668 @item -Wtraditional @r{(C only)}
2669 @opindex Wtraditional
2670 Warn about certain constructs that behave differently in traditional and
2671 ISO C@. Also warn about ISO C constructs that have no traditional C
2672 equivalent, and/or problematic constructs which should be avoided.
2676 Macro parameters that appear within string literals in the macro body.
2677 In traditional C macro replacement takes place within string literals,
2678 but does not in ISO C@.
2681 In traditional C, some preprocessor directives did not exist.
2682 Traditional preprocessors would only consider a line to be a directive
2683 if the @samp{#} appeared in column 1 on the line. Therefore
2684 @option{-Wtraditional} warns about directives that traditional C
2685 understands but would ignore because the @samp{#} does not appear as the
2686 first character on the line. It also suggests you hide directives like
2687 @samp{#pragma} not understood by traditional C by indenting them. Some
2688 traditional implementations would not recognize @samp{#elif}, so it
2689 suggests avoiding it altogether.
2692 A function-like macro that appears without arguments.
2695 The unary plus operator.
2698 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
2699 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
2700 constants.) Note, these suffixes appear in macros defined in the system
2701 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
2702 Use of these macros in user code might normally lead to spurious
2703 warnings, however GCC's integrated preprocessor has enough context to
2704 avoid warning in these cases.
2707 A function declared external in one block and then used after the end of
2711 A @code{switch} statement has an operand of type @code{long}.
2714 A non-@code{static} function declaration follows a @code{static} one.
2715 This construct is not accepted by some traditional C compilers.
2718 The ISO type of an integer constant has a different width or
2719 signedness from its traditional type. This warning is only issued if
2720 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
2721 typically represent bit patterns, are not warned about.
2724 Usage of ISO string concatenation is detected.
2727 Initialization of automatic aggregates.
2730 Identifier conflicts with labels. Traditional C lacks a separate
2731 namespace for labels.
2734 Initialization of unions. If the initializer is zero, the warning is
2735 omitted. This is done under the assumption that the zero initializer in
2736 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
2737 initializer warnings and relies on default initialization to zero in the
2741 Conversions by prototypes between fixed/floating point values and vice
2742 versa. The absence of these prototypes when compiling with traditional
2743 C would cause serious problems. This is a subset of the possible
2744 conversion warnings, for the full set use @option{-Wconversion}.
2747 Use of ISO C style function definitions. This warning intentionally is
2748 @emph{not} issued for prototype declarations or variadic functions
2749 because these ISO C features will appear in your code when using
2750 libiberty's traditional C compatibility macros, @code{PARAMS} and
2751 @code{VPARAMS}. This warning is also bypassed for nested functions
2752 because that feature is already a GCC extension and thus not relevant to
2753 traditional C compatibility.
2756 @item -Wdeclaration-after-statement @r{(C only)}
2757 @opindex Wdeclaration-after-statement
2758 Warn when a declaration is found after a statement in a block. This
2759 construct, known from C++, was introduced with ISO C99 and is by default
2760 allowed in GCC@. It is not supported by ISO C90 and was not supported by
2761 GCC versions before GCC 3.0. @xref{Mixed Declarations}.
2765 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
2767 @item -Wendif-labels
2768 @opindex Wendif-labels
2769 Warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
2773 Warn whenever a local variable shadows another local variable, parameter or
2774 global variable or whenever a built-in function is shadowed.
2776 @item -Wlarger-than-@var{len}
2777 @opindex Wlarger-than
2778 Warn whenever an object of larger than @var{len} bytes is defined.
2780 @item -Wpointer-arith
2781 @opindex Wpointer-arith
2782 Warn about anything that depends on the ``size of'' a function type or
2783 of @code{void}. GNU C assigns these types a size of 1, for
2784 convenience in calculations with @code{void *} pointers and pointers
2787 @item -Wbad-function-cast @r{(C only)}
2788 @opindex Wbad-function-cast
2789 Warn whenever a function call is cast to a non-matching type.
2790 For example, warn if @code{int malloc()} is cast to @code{anything *}.
2794 Warn whenever a pointer is cast so as to remove a type qualifier from
2795 the target type. For example, warn if a @code{const char *} is cast
2796 to an ordinary @code{char *}.
2799 @opindex Wcast-align
2800 Warn whenever a pointer is cast such that the required alignment of the
2801 target is increased. For example, warn if a @code{char *} is cast to
2802 an @code{int *} on machines where integers can only be accessed at
2803 two- or four-byte boundaries.
2805 @item -Wwrite-strings
2806 @opindex Wwrite-strings
2807 When compiling C, give string constants the type @code{const
2808 char[@var{length}]} so that
2809 copying the address of one into a non-@code{const} @code{char *}
2810 pointer will get a warning; when compiling C++, warn about the
2811 deprecated conversion from string constants to @code{char *}.
2812 These warnings will help you find at
2813 compile time code that can try to write into a string constant, but
2814 only if you have been very careful about using @code{const} in
2815 declarations and prototypes. Otherwise, it will just be a nuisance;
2816 this is why we did not make @option{-Wall} request these warnings.
2819 @opindex Wconversion
2820 Warn if a prototype causes a type conversion that is different from what
2821 would happen to the same argument in the absence of a prototype. This
2822 includes conversions of fixed point to floating and vice versa, and
2823 conversions changing the width or signedness of a fixed point argument
2824 except when the same as the default promotion.
2826 Also, warn if a negative integer constant expression is implicitly
2827 converted to an unsigned type. For example, warn about the assignment
2828 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
2829 casts like @code{(unsigned) -1}.
2831 @item -Wsign-compare
2832 @opindex Wsign-compare
2833 @cindex warning for comparison of signed and unsigned values
2834 @cindex comparison of signed and unsigned values, warning
2835 @cindex signed and unsigned values, comparison warning
2836 Warn when a comparison between signed and unsigned values could produce
2837 an incorrect result when the signed value is converted to unsigned.
2838 This warning is also enabled by @option{-Wextra}; to get the other warnings
2839 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
2841 @item -Waggregate-return
2842 @opindex Waggregate-return
2843 Warn if any functions that return structures or unions are defined or
2844 called. (In languages where you can return an array, this also elicits
2847 @item -Wstrict-prototypes @r{(C only)}
2848 @opindex Wstrict-prototypes
2849 Warn if a function is declared or defined without specifying the
2850 argument types. (An old-style function definition is permitted without
2851 a warning if preceded by a declaration which specifies the argument
2854 @item -Wold-style-definition @r{(C only)}
2855 @opindex Wold-style-definition
2856 Warn if an old-style function definition is used. A warning is given
2857 even if there is a previous prototype.
2859 @item -Wmissing-prototypes @r{(C only)}
2860 @opindex Wmissing-prototypes
2861 Warn if a global function is defined without a previous prototype
2862 declaration. This warning is issued even if the definition itself
2863 provides a prototype. The aim is to detect global functions that fail
2864 to be declared in header files.
2866 @item -Wmissing-declarations @r{(C only)}
2867 @opindex Wmissing-declarations
2868 Warn if a global function is defined without a previous declaration.
2869 Do so even if the definition itself provides a prototype.
2870 Use this option to detect global functions that are not declared in
2873 @item -Wmissing-noreturn
2874 @opindex Wmissing-noreturn
2875 Warn about functions which might be candidates for attribute @code{noreturn}.
2876 Note these are only possible candidates, not absolute ones. Care should
2877 be taken to manually verify functions actually do not ever return before
2878 adding the @code{noreturn} attribute, otherwise subtle code generation
2879 bugs could be introduced. You will not get a warning for @code{main} in
2880 hosted C environments.
2882 @item -Wmissing-format-attribute
2883 @opindex Wmissing-format-attribute
2885 If @option{-Wformat} is enabled, also warn about functions which might be
2886 candidates for @code{format} attributes. Note these are only possible
2887 candidates, not absolute ones. GCC will guess that @code{format}
2888 attributes might be appropriate for any function that calls a function
2889 like @code{vprintf} or @code{vscanf}, but this might not always be the
2890 case, and some functions for which @code{format} attributes are
2891 appropriate may not be detected. This option has no effect unless
2892 @option{-Wformat} is enabled (possibly by @option{-Wall}).
2894 @item -Wno-multichar
2895 @opindex Wno-multichar
2897 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
2898 Usually they indicate a typo in the user's code, as they have
2899 implementation-defined values, and should not be used in portable code.
2901 @item -Wno-deprecated-declarations
2902 @opindex Wno-deprecated-declarations
2903 Do not warn about uses of functions, variables, and types marked as
2904 deprecated by using the @code{deprecated} attribute.
2905 (@pxref{Function Attributes}, @pxref{Variable Attributes},
2906 @pxref{Type Attributes}.)
2910 Warn if a structure is given the packed attribute, but the packed
2911 attribute has no effect on the layout or size of the structure.
2912 Such structures may be mis-aligned for little benefit. For
2913 instance, in this code, the variable @code{f.x} in @code{struct bar}
2914 will be misaligned even though @code{struct bar} does not itself
2915 have the packed attribute:
2922 @} __attribute__((packed));
2932 Warn if padding is included in a structure, either to align an element
2933 of the structure or to align the whole structure. Sometimes when this
2934 happens it is possible to rearrange the fields of the structure to
2935 reduce the padding and so make the structure smaller.
2937 @item -Wredundant-decls
2938 @opindex Wredundant-decls
2939 Warn if anything is declared more than once in the same scope, even in
2940 cases where multiple declaration is valid and changes nothing.
2942 @item -Wnested-externs @r{(C only)}
2943 @opindex Wnested-externs
2944 Warn if an @code{extern} declaration is encountered within a function.
2946 @item -Wunreachable-code
2947 @opindex Wunreachable-code
2948 Warn if the compiler detects that code will never be executed.
2950 This option is intended to warn when the compiler detects that at
2951 least a whole line of source code will never be executed, because
2952 some condition is never satisfied or because it is after a
2953 procedure that never returns.
2955 It is possible for this option to produce a warning even though there
2956 are circumstances under which part of the affected line can be executed,
2957 so care should be taken when removing apparently-unreachable code.
2959 For instance, when a function is inlined, a warning may mean that the
2960 line is unreachable in only one inlined copy of the function.
2962 This option is not made part of @option{-Wall} because in a debugging
2963 version of a program there is often substantial code which checks
2964 correct functioning of the program and is, hopefully, unreachable
2965 because the program does work. Another common use of unreachable
2966 code is to provide behavior which is selectable at compile-time.
2970 Warn if a function can not be inlined and it was declared as inline.
2971 Even with this option, the compiler will not warn about failures to
2972 inline functions declared in system headers.
2974 The compiler uses a variety of heuristics to determine whether or not
2975 to inline a function. For example, the compiler takes into account
2976 the size of the function being inlined and the the amount of inlining
2977 that has already been done in the current function. Therefore,
2978 seemingly insignificant changes in the source program can cause the
2979 warnings produced by @option{-Winline} to appear or disappear.
2981 @item -Wno-invalid-offsetof @r{(C++ only)}
2982 @opindex Wno-invalid-offsetof
2983 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
2984 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
2985 to a non-POD type is undefined. In existing C++ implementations,
2986 however, @samp{offsetof} typically gives meaningful results even when
2987 applied to certain kinds of non-POD types. (Such as a simple
2988 @samp{struct} that fails to be a POD type only by virtue of having a
2989 constructor.) This flag is for users who are aware that they are
2990 writing nonportable code and who have deliberately chosen to ignore the
2993 The restrictions on @samp{offsetof} may be relaxed in a future version
2994 of the C++ standard.
2997 @opindex Winvalid-pch
2998 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
2999 the search path but can't be used.
3003 @opindex Wno-long-long
3004 Warn if @samp{long long} type is used. This is default. To inhibit
3005 the warning messages, use @option{-Wno-long-long}. Flags
3006 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
3007 only when @option{-pedantic} flag is used.
3009 @item -Wvariadic-macros
3010 @opindex Wvariadic-macros
3011 @opindex Wno-variadic-macros
3012 Warn if variadic macros are used in pedantic ISO C90 mode, or the GNU
3013 alternate syntax when in pedantic ISO C99 mode. This is default.
3014 To inhibit the warning messages, use @option{-Wno-variadic-macros}.
3016 @item -Wdisabled-optimization
3017 @opindex Wdisabled-optimization
3018 Warn if a requested optimization pass is disabled. This warning does
3019 not generally indicate that there is anything wrong with your code; it
3020 merely indicates that GCC's optimizers were unable to handle the code
3021 effectively. Often, the problem is that your code is too big or too
3022 complex; GCC will refuse to optimize programs when the optimization
3023 itself is likely to take inordinate amounts of time.
3027 Make all warnings into errors.
3030 @node Debugging Options
3031 @section Options for Debugging Your Program or GCC
3032 @cindex options, debugging
3033 @cindex debugging information options
3035 GCC has various special options that are used for debugging
3036 either your program or GCC:
3041 Produce debugging information in the operating system's native format
3042 (stabs, COFF, XCOFF, or DWARF)@. GDB can work with this debugging
3045 On most systems that use stabs format, @option{-g} enables use of extra
3046 debugging information that only GDB can use; this extra information
3047 makes debugging work better in GDB but will probably make other debuggers
3049 refuse to read the program. If you want to control for certain whether
3050 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
3051 @option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
3053 Unlike most other C compilers, GCC allows you to use @option{-g} with
3054 @option{-O}. The shortcuts taken by optimized code may occasionally
3055 produce surprising results: some variables you declared may not exist
3056 at all; flow of control may briefly move where you did not expect it;
3057 some statements may not be executed because they compute constant
3058 results or their values were already at hand; some statements may
3059 execute in different places because they were moved out of loops.
3061 Nevertheless it proves possible to debug optimized output. This makes
3062 it reasonable to use the optimizer for programs that might have bugs.
3064 The following options are useful when GCC is generated with the
3065 capability for more than one debugging format.
3069 Produce debugging information for use by GDB@. This means to use the
3070 most expressive format available (DWARF 2, stabs, or the native format
3071 if neither of those are supported), including GDB extensions if at all
3076 Produce debugging information in stabs format (if that is supported),
3077 without GDB extensions. This is the format used by DBX on most BSD
3078 systems. On MIPS, Alpha and System V Release 4 systems this option
3079 produces stabs debugging output which is not understood by DBX or SDB@.
3080 On System V Release 4 systems this option requires the GNU assembler.
3082 @item -feliminate-unused-debug-symbols
3083 @opindex feliminate-unused-debug-symbols
3084 Produce debugging information in stabs format (if that is supported),
3085 for only symbols that are actually used.
3089 Produce debugging information in stabs format (if that is supported),
3090 using GNU extensions understood only by the GNU debugger (GDB)@. The
3091 use of these extensions is likely to make other debuggers crash or
3092 refuse to read the program.
3096 Produce debugging information in COFF format (if that is supported).
3097 This is the format used by SDB on most System V systems prior to
3102 Produce debugging information in XCOFF format (if that is supported).
3103 This is the format used by the DBX debugger on IBM RS/6000 systems.
3107 Produce debugging information in XCOFF format (if that is supported),
3108 using GNU extensions understood only by the GNU debugger (GDB)@. The
3109 use of these extensions is likely to make other debuggers crash or
3110 refuse to read the program, and may cause assemblers other than the GNU
3111 assembler (GAS) to fail with an error.
3115 Produce debugging information in DWARF version 2 format (if that is
3116 supported). This is the format used by DBX on IRIX 6.
3120 Produce debugging information in VMS debug format (if that is
3121 supported). This is the format used by DEBUG on VMS systems.
3124 @itemx -ggdb@var{level}
3125 @itemx -gstabs@var{level}
3126 @itemx -gcoff@var{level}
3127 @itemx -gxcoff@var{level}
3128 @itemx -gvms@var{level}
3129 Request debugging information and also use @var{level} to specify how
3130 much information. The default level is 2.
3132 Level 1 produces minimal information, enough for making backtraces in
3133 parts of the program that you don't plan to debug. This includes
3134 descriptions of functions and external variables, but no information
3135 about local variables and no line numbers.
3137 Level 3 includes extra information, such as all the macro definitions
3138 present in the program. Some debuggers support macro expansion when
3139 you use @option{-g3}.
3141 Note that in order to avoid confusion between DWARF1 debug level 2,
3142 and DWARF2 @option{-gdwarf-2} does not accept a concatenated debug
3143 level. Instead use an additional @option{-g@var{level}} option to
3144 change the debug level for DWARF2.
3146 @item -feliminate-dwarf2-dups
3147 @opindex feliminate-dwarf2-dups
3148 Compress DWARF2 debugging information by eliminating duplicated
3149 information about each symbol. This option only makes sense when
3150 generating DWARF2 debugging information with @option{-gdwarf-2}.
3152 @cindex @command{prof}
3155 Generate extra code to write profile information suitable for the
3156 analysis program @command{prof}. You must use this option when compiling
3157 the source files you want data about, and you must also use it when
3160 @cindex @command{gprof}
3163 Generate extra code to write profile information suitable for the
3164 analysis program @command{gprof}. You must use this option when compiling
3165 the source files you want data about, and you must also use it when
3170 Makes the compiler print out each function name as it is compiled, and
3171 print some statistics about each pass when it finishes.
3174 @opindex ftime-report
3175 Makes the compiler print some statistics about the time consumed by each
3176 pass when it finishes.
3179 @opindex fmem-report
3180 Makes the compiler print some statistics about permanent memory
3181 allocation when it finishes.
3183 @item -fprofile-arcs
3184 @opindex fprofile-arcs
3185 Add code so that program flow @dfn{arcs} are instrumented. During
3186 execution the program records how many times each branch and call is
3187 executed and how many times it is taken or returns. When the compiled
3188 program exits it saves this data to a file called
3189 @file{@var{auxname}.gcda} for each source file. The data may be used for
3190 profile-directed optimizations (@option{-fbranch-probabilities}), or for
3191 test coverage analysis (@option{-ftest-coverage}). Each object file's
3192 @var{auxname} is generated from the name of the output file, if
3193 explicitly specified and it is not the final executable, otherwise it is
3194 the basename of the source file. In both cases any suffix is removed
3195 (e.g. @file{foo.gcda} for input file @file{dir/foo.c}, or
3196 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
3201 Compile the source files with @option{-fprofile-arcs} plus optimization
3202 and code generation options. For test coverage analysis, use the
3203 additional @option{-ftest-coverage} option. You do not need to profile
3204 every source file in a program.
3207 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
3208 (the latter implies the former).
3211 Run the program on a representative workload to generate the arc profile
3212 information. This may be repeated any number of times. You can run
3213 concurrent instances of your program, and provided that the file system
3214 supports locking, the data files will be correctly updated. Also
3215 @code{fork} calls are detected and correctly handled (double counting
3219 For profile-directed optimizations, compile the source files again with
3220 the same optimization and code generation options plus
3221 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
3222 Control Optimization}).
3225 For test coverage analysis, use @command{gcov} to produce human readable
3226 information from the @file{.gcno} and @file{.gcda} files. Refer to the
3227 @command{gcov} documentation for further information.
3231 With @option{-fprofile-arcs}, for each function of your program GCC
3232 creates a program flow graph, then finds a spanning tree for the graph.
3233 Only arcs that are not on the spanning tree have to be instrumented: the
3234 compiler adds code to count the number of times that these arcs are
3235 executed. When an arc is the only exit or only entrance to a block, the
3236 instrumentation code can be added to the block; otherwise, a new basic
3237 block must be created to hold the instrumentation code.
3239 @item -ftree-based-profiling
3240 @opindex ftree-based-profiling
3241 This option is used in addition to @option{-fprofile-arcs} or
3242 @option{-fbranch-probabilities} to control whether those optimizations
3243 are performed on a tree-based or rtl-based internal representation.
3244 If you use this option when compiling with @option{-fprofile-arcs},
3245 you must also use it when compiling later with @option{-fbranch-probabilities}.
3246 Currently the tree-based optimization is in an early stage of
3247 development, and this option is recommended only for those people
3248 working on improving it.
3251 @item -ftest-coverage
3252 @opindex ftest-coverage
3253 Produce a notes file that the @command{gcov} code-coverage utility
3254 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
3255 show program coverage. Each source file's note file is called
3256 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
3257 above for a description of @var{auxname} and instructions on how to
3258 generate test coverage data. Coverage data will match the source files
3259 more closely, if you do not optimize.
3261 @item -d@var{letters}
3263 Says to make debugging dumps during compilation at times specified by
3264 @var{letters}. This is used for debugging the compiler. The file names
3265 for most of the dumps are made by appending a pass number and a word to
3266 the @var{dumpname}. @var{dumpname} is generated from the name of the
3267 output file, if explicitly specified and it is not an executable,
3268 otherwise it is the basename of the source file. In both cases any
3269 suffix is removed (e.g. @file{foo.01.rtl} or @file{foo.02.sibling}).
3270 Here are the possible letters for use in @var{letters}, and their
3276 Annotate the assembler output with miscellaneous debugging information.
3279 Dump after computing branch probabilities, to @file{@var{file}.12.bp}.
3282 Dump after block reordering, to @file{@var{file}.31.bbro}.
3285 Dump after instruction combination, to the file @file{@var{file}.20.combine}.
3288 Dump after the first if conversion, to the file @file{@var{file}.14.ce1}.
3289 Also dump after the second if conversion, to the file @file{@var{file}.21.ce2}.
3292 Dump after branch target load optimization, to to @file{@var{file}.32.btl}.
3293 Also dump after delayed branch scheduling, to @file{@var{file}.36.dbr}.
3296 Dump all macro definitions, at the end of preprocessing, in addition to
3300 Dump after the third if conversion, to @file{@var{file}.30.ce3}.
3303 Dump after control and data flow analysis, to @file{@var{file}.11.cfg}.
3304 Also dump after life analysis, to @file{@var{file}.19.life}.
3307 Dump after purging @code{ADDRESSOF} codes, to @file{@var{file}.07.addressof}.
3310 Dump after global register allocation, to @file{@var{file}.25.greg}.
3313 Dump after GCSE, to @file{@var{file}.08.gcse}.
3314 Also dump after jump bypassing and control flow optimizations, to
3315 @file{@var{file}.10.bypass}.
3318 Dump after finalization of EH handling code, to @file{@var{file}.03.eh}.
3321 Dump after sibling call optimizations, to @file{@var{file}.02.sibling}.
3324 Dump after the first jump optimization, to @file{@var{file}.04.jump}.
3327 Dump after conversion from registers to stack, to @file{@var{file}.34.stack}.
3330 Dump after local register allocation, to @file{@var{file}.24.lreg}.
3333 Dump after loop optimization passes, to @file{@var{file}.09.loop} and
3334 @file{@var{file}.16.loop2}.
3337 Dump after performing the machine dependent reorganization pass, to
3338 @file{@var{file}.35.mach}.
3341 Dump after register renumbering, to @file{@var{file}.29.rnreg}.
3344 Dump after the register move pass, to @file{@var{file}.22.regmove}.
3347 Dump after post-reload optimizations, to @file{@var{file}.26.postreload}.
3350 Dump after RTL generation, to @file{@var{file}.01.rtl}.
3353 Dump after the second scheduling pass, to @file{@var{file}.33.sched2}.
3356 Dump after CSE (including the jump optimization that sometimes follows
3357 CSE), to @file{@var{file}.06.cse}.
3360 Dump after the first scheduling pass, to @file{@var{file}.23.sched}.
3363 Dump after the second CSE pass (including the jump optimization that
3364 sometimes follows CSE), to @file{@var{file}.18.cse2}.
3367 Dump after running tracer, to @file{@var{file}.15.tracer}.
3370 Dump after null pointer elimination pass to @file{@var{file}.05.null}.
3373 Dump callgraph and unit-at-a-time optimization @file{@var{file}.00.unit}.
3376 Dump after the value profile transformations, to @file{@var{file}.13.vpt}.
3377 Also dump after variable tracking, to @file{@var{file}.35.vartrack}.
3380 Dump after the second flow pass, to @file{@var{file}.27.flow2}.
3383 Dump after the peephole pass, to @file{@var{file}.28.peephole2}.
3386 Dump after constructing the web, to @file{@var{file}.17.web}.
3389 Produce all the dumps listed above.
3392 Produce a core dump whenever an error occurs.
3395 Print statistics on memory usage, at the end of the run, to
3399 Annotate the assembler output with a comment indicating which
3400 pattern and alternative was used. The length of each instruction is
3404 Dump the RTL in the assembler output as a comment before each instruction.
3405 Also turns on @option{-dp} annotation.
3408 For each of the other indicated dump files (except for
3409 @file{@var{file}.01.rtl}), dump a representation of the control flow graph
3410 suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
3413 Just generate RTL for a function instead of compiling it. Usually used
3417 Dump debugging information during parsing, to standard error.
3420 @item -fdump-unnumbered
3421 @opindex fdump-unnumbered
3422 When doing debugging dumps (see @option{-d} option above), suppress instruction
3423 numbers and line number note output. This makes it more feasible to
3424 use diff on debugging dumps for compiler invocations with different
3425 options, in particular with and without @option{-g}.
3427 @item -fdump-translation-unit @r{(C and C++ only)}
3428 @itemx -fdump-translation-unit-@var{options} @r{(C and C++ only)}
3429 @opindex fdump-translation-unit
3430 Dump a representation of the tree structure for the entire translation
3431 unit to a file. The file name is made by appending @file{.tu} to the
3432 source file name. If the @samp{-@var{options}} form is used, @var{options}
3433 controls the details of the dump as described for the
3434 @option{-fdump-tree} options.
3436 @item -fdump-class-hierarchy @r{(C++ only)}
3437 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
3438 @opindex fdump-class-hierarchy
3439 Dump a representation of each class's hierarchy and virtual function
3440 table layout to a file. The file name is made by appending @file{.class}
3441 to the source file name. If the @samp{-@var{options}} form is used,
3442 @var{options} controls the details of the dump as described for the
3443 @option{-fdump-tree} options.
3445 @item -fdump-tree-@var{switch} @r{(C and C++ only)}
3446 @itemx -fdump-tree-@var{switch}-@var{options} @r{(C and C++ only)}
3448 Control the dumping at various stages of processing the intermediate
3449 language tree to a file. The file name is generated by appending a switch
3450 specific suffix to the source file name. If the @samp{-@var{options}}
3451 form is used, @var{options} is a list of @samp{-} separated options that
3452 control the details of the dump. Not all options are applicable to all
3453 dumps, those which are not meaningful will be ignored. The following
3454 options are available
3458 Print the address of each node. Usually this is not meaningful as it
3459 changes according to the environment and source file. Its primary use
3460 is for tying up a dump file with a debug environment.
3462 Inhibit dumping of members of a scope or body of a function merely
3463 because that scope has been reached. Only dump such items when they
3464 are directly reachable by some other path. When dumping pretty-printed
3465 trees, this option inhibits dumping the bodies of control structures.
3467 Print a raw representation of the tree. By default, trees are
3468 pretty-printed into a C-like representation.
3470 Enable more detailed dumps (not honored by every dump option).
3472 Enable dumping various statistics about the pass (not honored by every dump
3475 Enable showing basic block boundaries (disabled in raw dumps).
3477 Enable showing virtual operands for every statement.
3479 Enable showing line numbers for statements.
3481 Enable showing the unique ID (@code{DECL_UID}) for each variable.
3483 Turn on all options, except @option{raw}, @option{slim} and @option{lineno}.
3486 The following tree dumps are possible:
3490 Dump before any tree based optimization, to @file{@var{file}.original}.
3493 Dump after all tree based optimization, to @file{@var{file}.optimized}.
3496 Dump after function inlining, to @file{@var{file}.inlined}.
3499 @opindex fdump-tree-gimple
3500 Dump each function before and after the gimplification pass to a file. The
3501 file name is made by appending @file{.gimple} to the source file name.
3504 @opindex fdump-tree-cfg
3505 Dump the control flow graph of each function to a file. The file name is
3506 made by appending @file{.cfg} to the source file name.
3509 @opindex fdump-tree-vcg
3510 Dump the control flow graph of each function to a file in VCG format. The
3511 file name is made by appending @file{.vcg} to the source file name. Note
3512 that if the file contains more than one function, the generated file cannot
3513 be used directly by VCG. You will need to cut and paste each function's
3514 graph into its own separate file first.
3517 @opindex fdump-tree-ch
3518 Dump each function after copying loop headers. The file name is made by
3519 appending @file{.ch} to the source file name.
3522 @opindex fdump-tree-ssa
3523 Dump SSA related information to a file. The file name is made by appending
3524 @file{.ssa} to the source file name.
3527 @opindex fdump-tree-alias
3528 Dump aliasing information for each function. The file name is made by
3529 appending @file{.alias} to the source file name.
3532 @opindex fdump-tree-ccp
3533 Dump each function after CCP. The file name is made by appending
3534 @file{.ccp} to the source file name.
3537 @opindex fdump-tree-pre
3538 Dump trees after partial redundancy elimination. The file name is made
3539 by appending @file{.pre} to the source file name.
3542 @opindex fdump-tree-dce
3543 Dump each function after dead code elimination. The file name is made by
3544 appending @file{.dce} to the source file name.
3547 @opindex fdump-tree-mudflap
3548 Dump each function after adding mudflap instrumentation. The file name is
3549 made by appending @file{.mudflap} to the source file name.
3552 @opindex fdump-tree-sra
3553 Dump each function after performing scalar replacement of aggregates. The
3554 file name is made by appending @file{.sra} to the source file name.
3557 @opindex fdump-tree-dom
3558 Dump each function after applying dominator tree optimizations. The file
3559 name is made by appending @file{.dom} to the source file name.
3562 @opindex fdump-tree-dse
3563 Dump each function after applying dead store elimination. The file
3564 name is made by appending @file{.dse} to the source file name.
3567 @opindex fdump-tree-phiopt
3568 Dump each function after optimizing PHI nodes into straightline code. The file
3569 name is made by appending @file{.phiopt} to the source file name.
3572 @opindex fdump-tree-forwprop
3573 Dump each function after forward propagating single use variables. The file
3574 name is made by appending @file{.forwprop} to the source file name.
3577 @opindex fdump-tree-copyrename
3578 Dump each function after applying the copy rename optimization. The file
3579 name is made by appending @file{.copyrename} to the source file name.
3582 @opindex fdump-tree-nrv
3583 Dump each function after applying the named return value optimization on
3584 generic trees. The file name is made by appending @file{.nrv} to the source
3588 @opindex fdump-tree-all
3589 Enable all the available tree dumps with the flags provided in this option.
3592 @item -frandom-seed=@var{string}
3593 @opindex frandom-string
3594 This option provides a seed that GCC uses when it would otherwise use
3595 random numbers. It is used to generate certain symbol names
3596 that have to be different in every compiled file. It is also used to
3597 place unique stamps in coverage data files and the object files that
3598 produce them. You can use the @option{-frandom-seed} option to produce
3599 reproducibly identical object files.
3601 The @var{string} should be different for every file you compile.
3603 @item -fsched-verbose=@var{n}
3604 @opindex fsched-verbose
3605 On targets that use instruction scheduling, this option controls the
3606 amount of debugging output the scheduler prints. This information is
3607 written to standard error, unless @option{-dS} or @option{-dR} is
3608 specified, in which case it is output to the usual dump
3609 listing file, @file{.sched} or @file{.sched2} respectively. However
3610 for @var{n} greater than nine, the output is always printed to standard
3613 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
3614 same information as @option{-dRS}. For @var{n} greater than one, it
3615 also output basic block probabilities, detailed ready list information
3616 and unit/insn info. For @var{n} greater than two, it includes RTL
3617 at abort point, control-flow and regions info. And for @var{n} over
3618 four, @option{-fsched-verbose} also includes dependence info.
3622 Store the usual ``temporary'' intermediate files permanently; place them
3623 in the current directory and name them based on the source file. Thus,
3624 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
3625 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
3626 preprocessed @file{foo.i} output file even though the compiler now
3627 normally uses an integrated preprocessor.
3631 Report the CPU time taken by each subprocess in the compilation
3632 sequence. For C source files, this is the compiler proper and assembler
3633 (plus the linker if linking is done). The output looks like this:
3640 The first number on each line is the ``user time,'' that is time spent
3641 executing the program itself. The second number is ``system time,''
3642 time spent executing operating system routines on behalf of the program.
3643 Both numbers are in seconds.
3645 @item -fvar-tracking
3646 @opindex fvar-tracking
3647 Run variable tracking pass. It computes where variables are stored at each
3648 position in code. Better debugging information is then generated
3649 (if the debugging information format supports this information).
3651 It is enabled by default when compiling with optimization (@option{-Os},
3652 @option{-O}, @option{-O2}, ...), debugging information (@option{-g}) and
3653 the debug info format supports it.
3655 @item -print-file-name=@var{library}
3656 @opindex print-file-name
3657 Print the full absolute name of the library file @var{library} that
3658 would be used when linking---and don't do anything else. With this
3659 option, GCC does not compile or link anything; it just prints the
3662 @item -print-multi-directory
3663 @opindex print-multi-directory
3664 Print the directory name corresponding to the multilib selected by any
3665 other switches present in the command line. This directory is supposed
3666 to exist in @env{GCC_EXEC_PREFIX}.
3668 @item -print-multi-lib
3669 @opindex print-multi-lib
3670 Print the mapping from multilib directory names to compiler switches
3671 that enable them. The directory name is separated from the switches by
3672 @samp{;}, and each switch starts with an @samp{@@} instead of the
3673 @samp{-}, without spaces between multiple switches. This is supposed to
3674 ease shell-processing.
3676 @item -print-prog-name=@var{program}
3677 @opindex print-prog-name
3678 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
3680 @item -print-libgcc-file-name
3681 @opindex print-libgcc-file-name
3682 Same as @option{-print-file-name=libgcc.a}.
3684 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
3685 but you do want to link with @file{libgcc.a}. You can do
3688 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
3691 @item -print-search-dirs
3692 @opindex print-search-dirs
3693 Print the name of the configured installation directory and a list of
3694 program and library directories @command{gcc} will search---and don't do anything else.
3696 This is useful when @command{gcc} prints the error message
3697 @samp{installation problem, cannot exec cpp0: No such file or directory}.
3698 To resolve this you either need to put @file{cpp0} and the other compiler
3699 components where @command{gcc} expects to find them, or you can set the environment
3700 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
3701 Don't forget the trailing '/'.
3702 @xref{Environment Variables}.
3705 @opindex dumpmachine
3706 Print the compiler's target machine (for example,
3707 @samp{i686-pc-linux-gnu})---and don't do anything else.
3710 @opindex dumpversion
3711 Print the compiler version (for example, @samp{3.0})---and don't do
3716 Print the compiler's built-in specs---and don't do anything else. (This
3717 is used when GCC itself is being built.) @xref{Spec Files}.
3719 @item -feliminate-unused-debug-types
3720 @opindex feliminate-unused-debug-types
3721 Normally, when producing DWARF2 output, GCC will emit debugging
3722 information for all types declared in a compilation
3723 unit, regardless of whether or not they are actually used
3724 in that compilation unit. Sometimes this is useful, such as
3725 if, in the debugger, you want to cast a value to a type that is
3726 not actually used in your program (but is declared). More often,
3727 however, this results in a significant amount of wasted space.
3728 With this option, GCC will avoid producing debug symbol output
3729 for types that are nowhere used in the source file being compiled.
3732 @node Optimize Options
3733 @section Options That Control Optimization
3734 @cindex optimize options
3735 @cindex options, optimization
3737 These options control various sorts of optimizations.
3739 Without any optimization option, the compiler's goal is to reduce the
3740 cost of compilation and to make debugging produce the expected
3741 results. Statements are independent: if you stop the program with a
3742 breakpoint between statements, you can then assign a new value to any
3743 variable or change the program counter to any other statement in the
3744 function and get exactly the results you would expect from the source
3747 Turning on optimization flags makes the compiler attempt to improve
3748 the performance and/or code size at the expense of compilation time
3749 and possibly the ability to debug the program.
3751 The compiler performs optimization based on the knowledge it has of
3752 the program. Optimization levels @option{-O2} and above, in
3753 particular, enable @emph{unit-at-a-time} mode, which allows the
3754 compiler to consider information gained from later functions in
3755 the file when compiling a function. Compiling multiple files at
3756 once to a single output file in @emph{unit-at-a-time} mode allows
3757 the compiler to use information gained from all of the files when
3758 compiling each of them.
3760 Not all optimizations are controlled directly by a flag. Only
3761 optimizations that have a flag are listed.
3768 Optimize. Optimizing compilation takes somewhat more time, and a lot
3769 more memory for a large function.
3771 With @option{-O}, the compiler tries to reduce code size and execution
3772 time, without performing any optimizations that take a great deal of
3775 @option{-O} turns on the following optimization flags:
3776 @gccoptlist{-fdefer-pop @gol
3777 -fmerge-constants @gol
3779 -floop-optimize @gol
3780 -fif-conversion @gol
3781 -fif-conversion2 @gol
3782 -fdelayed-branch @gol
3783 -fguess-branch-probability @gol
3786 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
3787 where doing so does not interfere with debugging.
3791 Optimize even more. GCC performs nearly all supported optimizations
3792 that do not involve a space-speed tradeoff. The compiler does not
3793 perform loop unrolling or function inlining when you specify @option{-O2}.
3794 As compared to @option{-O}, this option increases both compilation time
3795 and the performance of the generated code.
3797 @option{-O2} turns on all optimization flags specified by @option{-O}. It
3798 also turns on the following optimization flags:
3799 @gccoptlist{-fforce-mem @gol
3800 -foptimize-sibling-calls @gol
3801 -fstrength-reduce @gol
3802 -fcse-follow-jumps -fcse-skip-blocks @gol
3803 -frerun-cse-after-loop -frerun-loop-opt @gol
3804 -fgcse -fgcse-lm -fgcse-sm -fgcse-las @gol
3805 -fdelete-null-pointer-checks @gol
3806 -fexpensive-optimizations @gol
3808 -fschedule-insns -fschedule-insns2 @gol
3809 -fsched-interblock -fsched-spec @gol
3812 -freorder-blocks -freorder-functions @gol
3813 -fstrict-aliasing @gol
3814 -funit-at-a-time @gol
3815 -falign-functions -falign-jumps @gol
3816 -falign-loops -falign-labels @gol
3819 Please note the warning under @option{-fgcse} about
3820 invoking @option{-O2} on programs that use computed gotos.
3824 Optimize yet more. @option{-O3} turns on all optimizations specified by
3825 @option{-O2} and also turns on the @option{-finline-functions},
3826 @option{-fweb}, @option{-frename-registers}
3827 and @option{-fgcse-after-reload} options.
3831 Do not optimize. This is the default.
3835 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
3836 do not typically increase code size. It also performs further
3837 optimizations designed to reduce code size.
3839 @option{-Os} disables the following optimization flags:
3840 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
3841 -falign-labels -freorder-blocks -freorder-blocks-and-partition -fprefetch-loop-arrays}
3843 If you use multiple @option{-O} options, with or without level numbers,
3844 the last such option is the one that is effective.
3847 Options of the form @option{-f@var{flag}} specify machine-independent
3848 flags. Most flags have both positive and negative forms; the negative
3849 form of @option{-ffoo} would be @option{-fno-foo}. In the table
3850 below, only one of the forms is listed---the one you typically will
3851 use. You can figure out the other form by either removing @samp{no-}
3854 The following options control specific optimizations. They are either
3855 activated by @option{-O} options or are related to ones that are. You
3856 can use the following flags in the rare cases when ``fine-tuning'' of
3857 optimizations to be performed is desired.
3860 @item -fno-default-inline
3861 @opindex fno-default-inline
3862 Do not make member functions inline by default merely because they are
3863 defined inside the class scope (C++ only). Otherwise, when you specify
3864 @w{@option{-O}}, member functions defined inside class scope are compiled
3865 inline by default; i.e., you don't need to add @samp{inline} in front of
3866 the member function name.
3868 @item -fno-defer-pop
3869 @opindex fno-defer-pop
3870 Always pop the arguments to each function call as soon as that function
3871 returns. For machines which must pop arguments after a function call,
3872 the compiler normally lets arguments accumulate on the stack for several
3873 function calls and pops them all at once.
3875 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3879 Force memory operands to be copied into registers before doing
3880 arithmetic on them. This produces better code by making all memory
3881 references potential common subexpressions. When they are not common
3882 subexpressions, instruction combination should eliminate the separate
3885 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3888 @opindex fforce-addr
3889 Force memory address constants to be copied into registers before
3890 doing arithmetic on them. This may produce better code just as
3891 @option{-fforce-mem} may.
3893 @item -fomit-frame-pointer
3894 @opindex fomit-frame-pointer
3895 Don't keep the frame pointer in a register for functions that
3896 don't need one. This avoids the instructions to save, set up and
3897 restore frame pointers; it also makes an extra register available
3898 in many functions. @strong{It also makes debugging impossible on
3901 On some machines, such as the VAX, this flag has no effect, because
3902 the standard calling sequence automatically handles the frame pointer
3903 and nothing is saved by pretending it doesn't exist. The
3904 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
3905 whether a target machine supports this flag. @xref{Registers,,Register
3906 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
3908 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3910 @item -foptimize-sibling-calls
3911 @opindex foptimize-sibling-calls
3912 Optimize sibling and tail recursive calls.
3914 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3918 Don't pay attention to the @code{inline} keyword. Normally this option
3919 is used to keep the compiler from expanding any functions inline.
3920 Note that if you are not optimizing, no functions can be expanded inline.
3922 @item -finline-functions
3923 @opindex finline-functions
3924 Integrate all simple functions into their callers. The compiler
3925 heuristically decides which functions are simple enough to be worth
3926 integrating in this way.
3928 If all calls to a given function are integrated, and the function is
3929 declared @code{static}, then the function is normally not output as
3930 assembler code in its own right.
3932 Enabled at level @option{-O3}.
3934 @item -finline-limit=@var{n}
3935 @opindex finline-limit
3936 By default, GCC limits the size of functions that can be inlined. This flag
3937 allows the control of this limit for functions that are explicitly marked as
3938 inline (i.e., marked with the inline keyword or defined within the class
3939 definition in c++). @var{n} is the size of functions that can be inlined in
3940 number of pseudo instructions (not counting parameter handling). The default
3941 value of @var{n} is 600.
3942 Increasing this value can result in more inlined code at
3943 the cost of compilation time and memory consumption. Decreasing usually makes
3944 the compilation faster and less code will be inlined (which presumably
3945 means slower programs). This option is particularly useful for programs that
3946 use inlining heavily such as those based on recursive templates with C++.
3948 Inlining is actually controlled by a number of parameters, which may be
3949 specified individually by using @option{--param @var{name}=@var{value}}.
3950 The @option{-finline-limit=@var{n}} option sets some of these parameters
3954 @item max-inline-insns-single
3955 is set to @var{n}/2.
3956 @item max-inline-insns-auto
3957 is set to @var{n}/2.
3958 @item min-inline-insns
3959 is set to 130 or @var{n}/4, whichever is smaller.
3960 @item max-inline-insns-rtl
3964 See below for a documentation of the individual
3965 parameters controlling inlining.
3967 @emph{Note:} pseudo instruction represents, in this particular context, an
3968 abstract measurement of function's size. In no way, it represents a count
3969 of assembly instructions and as such its exact meaning might change from one
3970 release to an another.
3972 @item -fkeep-inline-functions
3973 @opindex fkeep-inline-functions
3974 Even if all calls to a given function are integrated, and the function
3975 is declared @code{static}, nevertheless output a separate run-time
3976 callable version of the function. This switch does not affect
3977 @code{extern inline} functions.
3979 @item -fkeep-static-consts
3980 @opindex fkeep-static-consts
3981 Emit variables declared @code{static const} when optimization isn't turned
3982 on, even if the variables aren't referenced.
3984 GCC enables this option by default. If you want to force the compiler to
3985 check if the variable was referenced, regardless of whether or not
3986 optimization is turned on, use the @option{-fno-keep-static-consts} option.
3988 @item -fmerge-constants
3989 Attempt to merge identical constants (string constants and floating point
3990 constants) across compilation units.
3992 This option is the default for optimized compilation if the assembler and
3993 linker support it. Use @option{-fno-merge-constants} to inhibit this
3996 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3998 @item -fmerge-all-constants
3999 Attempt to merge identical constants and identical variables.
4001 This option implies @option{-fmerge-constants}. In addition to
4002 @option{-fmerge-constants} this considers e.g. even constant initialized
4003 arrays or initialized constant variables with integral or floating point
4004 types. Languages like C or C++ require each non-automatic variable to
4005 have distinct location, so using this option will result in non-conforming
4010 Use a graph coloring register allocator. Currently this option is meant
4011 only for testing. Users should not specify this option, since it is not
4012 yet ready for production use.
4014 @item -fno-branch-count-reg
4015 @opindex fno-branch-count-reg
4016 Do not use ``decrement and branch'' instructions on a count register,
4017 but instead generate a sequence of instructions that decrement a
4018 register, compare it against zero, then branch based upon the result.
4019 This option is only meaningful on architectures that support such
4020 instructions, which include x86, PowerPC, IA-64 and S/390.
4022 The default is @option{-fbranch-count-reg}, enabled when
4023 @option{-fstrength-reduce} is enabled.
4025 @item -fno-function-cse
4026 @opindex fno-function-cse
4027 Do not put function addresses in registers; make each instruction that
4028 calls a constant function contain the function's address explicitly.
4030 This option results in less efficient code, but some strange hacks
4031 that alter the assembler output may be confused by the optimizations
4032 performed when this option is not used.
4034 The default is @option{-ffunction-cse}
4036 @item -fno-zero-initialized-in-bss
4037 @opindex fno-zero-initialized-in-bss
4038 If the target supports a BSS section, GCC by default puts variables that
4039 are initialized to zero into BSS@. This can save space in the resulting
4042 This option turns off this behavior because some programs explicitly
4043 rely on variables going to the data section. E.g., so that the
4044 resulting executable can find the beginning of that section and/or make
4045 assumptions based on that.
4047 The default is @option{-fzero-initialized-in-bss}.
4049 @item -fbounds-check
4050 @opindex fbounds-check
4051 For front-ends that support it, generate additional code to check that
4052 indices used to access arrays are within the declared range. This is
4053 currently only supported by the Java and Fortran front-ends, where
4054 this option defaults to true and false respectively.
4056 @item -fmudflap -fmudflapth -fmudflapir
4060 @cindex bounds checking
4062 For front-ends that support it (C and C++), instrument all risky
4063 pointer/array dereferencing operations, some standard library
4064 string/heap functions, and some other associated constructs with
4065 range/validity tests. Modules so instrumented should be immune to
4066 buffer overflows, invalid heap use, and some other classes of C/C++
4067 programming errors. The instrumentation relies on a separate runtime
4068 library (@file{libmudflap}), which will be linked into a program if
4069 @option{-fmudflap} is given at link time. Run-time behavior of the
4070 instrumented program is controlled by the @env{MUDFLAP_OPTIONS}
4071 environment variable. See @code{env MUDFLAP_OPTIONS=-help a.out}
4074 Use @option{-fmudflapth} instead of @option{-fmudflap} to compile and to
4075 link if your program is multi-threaded. Use @option{-fmudflapir}, in
4076 addition to @option{-fmudflap} or @option{-fmudflapth}, if
4077 instrumentation should ignore pointer reads. This produces less
4078 instrumentation (and therefore faster execution) and still provides
4079 some protection against outright memory corrupting writes, but allows
4080 erroneously read data to propagate within a program.
4082 @item -fstrength-reduce
4083 @opindex fstrength-reduce
4084 Perform the optimizations of loop strength reduction and
4085 elimination of iteration variables.
4087 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4089 @item -fthread-jumps
4090 @opindex fthread-jumps
4091 Perform optimizations where we check to see if a jump branches to a
4092 location where another comparison subsumed by the first is found. If
4093 so, the first branch is redirected to either the destination of the
4094 second branch or a point immediately following it, depending on whether
4095 the condition is known to be true or false.
4097 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4099 @item -fcse-follow-jumps
4100 @opindex fcse-follow-jumps
4101 In common subexpression elimination, scan through jump instructions
4102 when the target of the jump is not reached by any other path. For
4103 example, when CSE encounters an @code{if} statement with an
4104 @code{else} clause, CSE will follow the jump when the condition
4107 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4109 @item -fcse-skip-blocks
4110 @opindex fcse-skip-blocks
4111 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
4112 follow jumps which conditionally skip over blocks. When CSE
4113 encounters a simple @code{if} statement with no else clause,
4114 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
4115 body of the @code{if}.
4117 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4119 @item -frerun-cse-after-loop
4120 @opindex frerun-cse-after-loop
4121 Re-run common subexpression elimination after loop optimizations has been
4124 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4126 @item -frerun-loop-opt
4127 @opindex frerun-loop-opt
4128 Run the loop optimizer twice.
4130 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4134 Perform a global common subexpression elimination pass.
4135 This pass also performs global constant and copy propagation.
4137 @emph{Note:} When compiling a program using computed gotos, a GCC
4138 extension, you may get better runtime performance if you disable
4139 the global common subexpression elimination pass by adding
4140 @option{-fno-gcse} to the command line.
4142 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4146 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
4147 attempt to move loads which are only killed by stores into themselves. This
4148 allows a loop containing a load/store sequence to be changed to a load outside
4149 the loop, and a copy/store within the loop.
4151 Enabled by default when gcse is enabled.
4155 When @option{-fgcse-sm} is enabled, a store motion pass is run after
4156 global common subexpression elimination. This pass will attempt to move
4157 stores out of loops. When used in conjunction with @option{-fgcse-lm},
4158 loops containing a load/store sequence can be changed to a load before
4159 the loop and a store after the loop.
4161 Enabled by default when gcse is enabled.
4165 When @option{-fgcse-las} is enabled, the global common subexpression
4166 elimination pass eliminates redundant loads that come after stores to the
4167 same memory location (both partial and full redundancies).
4169 Enabled by default when gcse is enabled.
4171 @item -fgcse-after-reload
4172 @opindex fgcse-after-reload
4173 When @option{-fgcse-after-reload} is enabled, a redundant load elimination
4174 pass is performed after reload. The purpose of this pass is to cleanup
4177 @item -floop-optimize
4178 @opindex floop-optimize
4179 Perform loop optimizations: move constant expressions out of loops, simplify
4180 exit test conditions and optionally do strength-reduction and loop unrolling as
4183 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4185 @item -fcrossjumping
4186 @opindex crossjumping
4187 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
4188 resulting code may or may not perform better than without cross-jumping.
4190 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4192 @item -fif-conversion
4193 @opindex if-conversion
4194 Attempt to transform conditional jumps into branch-less equivalents. This
4195 include use of conditional moves, min, max, set flags and abs instructions, and
4196 some tricks doable by standard arithmetics. The use of conditional execution
4197 on chips where it is available is controlled by @code{if-conversion2}.
4199 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4201 @item -fif-conversion2
4202 @opindex if-conversion2
4203 Use conditional execution (where available) to transform conditional jumps into
4204 branch-less equivalents.
4206 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4208 @item -fdelete-null-pointer-checks
4209 @opindex fdelete-null-pointer-checks
4210 Use global dataflow analysis to identify and eliminate useless checks
4211 for null pointers. The compiler assumes that dereferencing a null
4212 pointer would have halted the program. If a pointer is checked after
4213 it has already been dereferenced, it cannot be null.
4215 In some environments, this assumption is not true, and programs can
4216 safely dereference null pointers. Use
4217 @option{-fno-delete-null-pointer-checks} to disable this optimization
4218 for programs which depend on that behavior.
4220 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4222 @item -fexpensive-optimizations
4223 @opindex fexpensive-optimizations
4224 Perform a number of minor optimizations that are relatively expensive.
4226 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4228 @item -foptimize-register-move
4230 @opindex foptimize-register-move
4232 Attempt to reassign register numbers in move instructions and as
4233 operands of other simple instructions in order to maximize the amount of
4234 register tying. This is especially helpful on machines with two-operand
4237 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
4240 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4242 @item -fdelayed-branch
4243 @opindex fdelayed-branch
4244 If supported for the target machine, attempt to reorder instructions
4245 to exploit instruction slots available after delayed branch
4248 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4250 @item -fschedule-insns
4251 @opindex fschedule-insns
4252 If supported for the target machine, attempt to reorder instructions to
4253 eliminate execution stalls due to required data being unavailable. This
4254 helps machines that have slow floating point or memory load instructions
4255 by allowing other instructions to be issued until the result of the load
4256 or floating point instruction is required.
4258 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4260 @item -fschedule-insns2
4261 @opindex fschedule-insns2
4262 Similar to @option{-fschedule-insns}, but requests an additional pass of
4263 instruction scheduling after register allocation has been done. This is
4264 especially useful on machines with a relatively small number of
4265 registers and where memory load instructions take more than one cycle.
4267 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4269 @item -fno-sched-interblock
4270 @opindex fno-sched-interblock
4271 Don't schedule instructions across basic blocks. This is normally
4272 enabled by default when scheduling before register allocation, i.e.@:
4273 with @option{-fschedule-insns} or at @option{-O2} or higher.
4275 @item -fno-sched-spec
4276 @opindex fno-sched-spec
4277 Don't allow speculative motion of non-load instructions. This is normally
4278 enabled by default when scheduling before register allocation, i.e.@:
4279 with @option{-fschedule-insns} or at @option{-O2} or higher.
4281 @item -fsched-spec-load
4282 @opindex fsched-spec-load
4283 Allow speculative motion of some load instructions. This only makes
4284 sense when scheduling before register allocation, i.e.@: with
4285 @option{-fschedule-insns} or at @option{-O2} or higher.
4287 @item -fsched-spec-load-dangerous
4288 @opindex fsched-spec-load-dangerous
4289 Allow speculative motion of more load instructions. This only makes
4290 sense when scheduling before register allocation, i.e.@: with
4291 @option{-fschedule-insns} or at @option{-O2} or higher.
4293 @item -fsched-stalled-insns=@var{n}
4294 @opindex fsched-stalled-insns
4295 Define how many insns (if any) can be moved prematurely from the queue
4296 of stalled insns into the ready list, during the second scheduling pass.
4298 @item -fsched-stalled-insns-dep=@var{n}
4299 @opindex fsched-stalled-insns-dep
4300 Define how many insn groups (cycles) will be examined for a dependency
4301 on a stalled insn that is candidate for premature removal from the queue
4302 of stalled insns. Has an effect only during the second scheduling pass,
4303 and only if @option{-fsched-stalled-insns} is used and its value is not zero.
4305 @item -fsched2-use-superblocks
4306 @opindex fsched2-use-superblocks
4307 When scheduling after register allocation, do use superblock scheduling
4308 algorithm. Superblock scheduling allows motion across basic block boundaries
4309 resulting on faster schedules. This option is experimental, as not all machine
4310 descriptions used by GCC model the CPU closely enough to avoid unreliable
4311 results from the algorithm.
4313 This only makes sense when scheduling after register allocation, i.e.@: with
4314 @option{-fschedule-insns2} or at @option{-O2} or higher.
4316 @item -fsched2-use-traces
4317 @opindex fsched2-use-traces
4318 Use @option{-fsched2-use-superblocks} algorithm when scheduling after register
4319 allocation and additionally perform code duplication in order to increase the
4320 size of superblocks using tracer pass. See @option{-ftracer} for details on
4323 This mode should produce faster but significantly longer programs. Also
4324 without @code{-fbranch-probabilities} the traces constructed may not match the
4325 reality and hurt the performance. This only makes
4326 sense when scheduling after register allocation, i.e.@: with
4327 @option{-fschedule-insns2} or at @option{-O2} or higher.
4329 @item -fcaller-saves
4330 @opindex fcaller-saves
4331 Enable values to be allocated in registers that will be clobbered by
4332 function calls, by emitting extra instructions to save and restore the
4333 registers around such calls. Such allocation is done only when it
4334 seems to result in better code than would otherwise be produced.
4336 This option is always enabled by default on certain machines, usually
4337 those which have no call-preserved registers to use instead.
4339 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4342 Perform Partial Redundancy Elimination (PRE) on trees. This flag is
4343 enabled by default at -O and higher.
4346 Perform sparse conditional constant propagation (CCP) on trees. This flag
4347 is enabled by default at -O and higher.
4350 Perform dead code elimination (DCE) on trees. This flag is enabled by
4351 default at -O and higher.
4353 @item -ftree-dominator-opts
4354 Perform dead code elimination (DCE) on trees. This flag is enabled by
4355 default at -O and higher.
4358 Perform loop header copying on trees. This is beneficial since it increases
4359 effectivity of code motion optimizations. It also saves one jump. This flag
4360 is enabled by default at -O and higher. It is not enabled for -Os, since it
4361 usually increases code size.
4364 Perform scalar replacement of aggregates. This pass replaces structure
4365 references with scalars to prevent committing structures to memory too
4366 early. This flag is enabled by default at -O and higher.
4368 @item -ftree-copyrename
4369 Perform copy renaming on trees. This pass attempts to rename compiler
4370 temporaries to other variables at copy locations, usually resulting in
4371 variable names which more closely resemble the original variables. This flag
4372 is enabled by default at -O and higher.
4375 Perform temporary expression replacement during the SSA->normal phase. Single
4376 use/single def temporaries are replaced at their use location with their
4377 defining expression. This results in non-GIMPLE code, but gives the expanders
4378 much more complex trees to work on resulting in better RTL generation. This is
4379 enabled by default at -O and higher.
4382 Perform live range splitting during the SSA->normal phase. Distinct live
4383 ranges of a variable are split into unique variables, allowing for better
4384 optimization later. This is enabled by default at -O and higher.
4388 Perform tail duplication to enlarge superblock size. This transformation
4389 simplifies the control flow of the function allowing other optimizations to do
4392 @item -funroll-loops
4393 @opindex funroll-loops
4394 Unroll loops whose number of iterations can be determined at compile
4395 time or upon entry to the loop. @option{-funroll-loops} implies both
4396 @option{-fstrength-reduce} and @option{-frerun-cse-after-loop}. This
4397 option makes code larger, and may or may not make it run faster.
4399 @item -funroll-all-loops
4400 @opindex funroll-all-loops
4401 Unroll all loops, even if their number of iterations is uncertain when
4402 the loop is entered. This usually makes programs run more slowly.
4403 @option{-funroll-all-loops} implies the same options as
4404 @option{-funroll-loops},
4406 @item -fprefetch-loop-arrays
4407 @opindex fprefetch-loop-arrays
4408 If supported by the target machine, generate instructions to prefetch
4409 memory to improve the performance of loops that access large arrays.
4411 @item -fmove-all-movables
4412 @opindex fmove-all-movables
4413 Forces all invariant computations in loops to be moved
4416 @item -freduce-all-givs
4417 @opindex freduce-all-givs
4418 Forces all general-induction variables in loops to be
4421 @emph{Note:} When compiling programs written in Fortran,
4422 @option{-fmove-all-movables} and @option{-freduce-all-givs} are enabled
4423 by default when you use the optimizer.
4425 These options may generate better or worse code; results are highly
4426 dependent on the structure of loops within the source code.
4428 These two options are intended to be removed someday, once
4429 they have helped determine the efficacy of various
4430 approaches to improving loop optimizations.
4432 Please let us (@w{@email{gcc@@gcc.gnu.org}} and @w{@email{fortran@@gnu.org}})
4433 know how use of these options affects
4434 the performance of your production code.
4435 We're very interested in code that runs @emph{slower}
4436 when these options are @emph{enabled}.
4439 @itemx -fno-peephole2
4440 @opindex fno-peephole
4441 @opindex fno-peephole2
4442 Disable any machine-specific peephole optimizations. The difference
4443 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
4444 are implemented in the compiler; some targets use one, some use the
4445 other, a few use both.
4447 @option{-fpeephole} is enabled by default.
4448 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4450 @item -fno-guess-branch-probability
4451 @opindex fno-guess-branch-probability
4452 Do not guess branch probabilities using a randomized model.
4454 Sometimes GCC will opt to use a randomized model to guess branch
4455 probabilities, when none are available from either profiling feedback
4456 (@option{-fprofile-arcs}) or @samp{__builtin_expect}. This means that
4457 different runs of the compiler on the same program may produce different
4460 In a hard real-time system, people don't want different runs of the
4461 compiler to produce code that has different behavior; minimizing
4462 non-determinism is of paramount import. This switch allows users to
4463 reduce non-determinism, possibly at the expense of inferior
4466 The default is @option{-fguess-branch-probability} at levels
4467 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4469 @item -freorder-blocks
4470 @opindex freorder-blocks
4471 Reorder basic blocks in the compiled function in order to reduce number of
4472 taken branches and improve code locality.
4474 Enabled at levels @option{-O2}, @option{-O3}.
4476 @item -freorder-blocks-and-partition
4477 @opindex freorder-blocks-and-partition
4478 In addition to reordering basic blocks in the compiled function, in order
4479 to reduce number of taken branches, partitions hot and cold basic blocks
4480 into separate sections of the assembly and .o files, to improve
4481 paging and cache locality performance.
4483 @item -freorder-functions
4484 @opindex freorder-functions
4485 Reorder basic blocks in the compiled function in order to reduce number of
4486 taken branches and improve code locality. This is implemented by using special
4487 subsections @code{.text.hot} for most frequently executed functions and
4488 @code{.text.unlikely} for unlikely executed functions. Reordering is done by
4489 the linker so object file format must support named sections and linker must
4490 place them in a reasonable way.
4492 Also profile feedback must be available in to make this option effective. See
4493 @option{-fprofile-arcs} for details.
4495 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4497 @item -fstrict-aliasing
4498 @opindex fstrict-aliasing
4499 Allows the compiler to assume the strictest aliasing rules applicable to
4500 the language being compiled. For C (and C++), this activates
4501 optimizations based on the type of expressions. In particular, an
4502 object of one type is assumed never to reside at the same address as an
4503 object of a different type, unless the types are almost the same. For
4504 example, an @code{unsigned int} can alias an @code{int}, but not a
4505 @code{void*} or a @code{double}. A character type may alias any other
4508 Pay special attention to code like this:
4521 The practice of reading from a different union member than the one most
4522 recently written to (called ``type-punning'') is common. Even with
4523 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
4524 is accessed through the union type. So, the code above will work as
4525 expected. However, this code might not:
4536 Every language that wishes to perform language-specific alias analysis
4537 should define a function that computes, given an @code{tree}
4538 node, an alias set for the node. Nodes in different alias sets are not
4539 allowed to alias. For an example, see the C front-end function
4540 @code{c_get_alias_set}.
4542 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4544 @item -falign-functions
4545 @itemx -falign-functions=@var{n}
4546 @opindex falign-functions
4547 Align the start of functions to the next power-of-two greater than
4548 @var{n}, skipping up to @var{n} bytes. For instance,
4549 @option{-falign-functions=32} aligns functions to the next 32-byte
4550 boundary, but @option{-falign-functions=24} would align to the next
4551 32-byte boundary only if this can be done by skipping 23 bytes or less.
4553 @option{-fno-align-functions} and @option{-falign-functions=1} are
4554 equivalent and mean that functions will not be aligned.
4556 Some assemblers only support this flag when @var{n} is a power of two;
4557 in that case, it is rounded up.
4559 If @var{n} is not specified or is zero, use a machine-dependent default.
4561 Enabled at levels @option{-O2}, @option{-O3}.
4563 @item -falign-labels
4564 @itemx -falign-labels=@var{n}
4565 @opindex falign-labels
4566 Align all branch targets to a power-of-two boundary, skipping up to
4567 @var{n} bytes like @option{-falign-functions}. This option can easily
4568 make code slower, because it must insert dummy operations for when the
4569 branch target is reached in the usual flow of the code.
4571 @option{-fno-align-labels} and @option{-falign-labels=1} are
4572 equivalent and mean that labels will not be aligned.
4574 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
4575 are greater than this value, then their values are used instead.
4577 If @var{n} is not specified or is zero, use a machine-dependent default
4578 which is very likely to be @samp{1}, meaning no alignment.
4580 Enabled at levels @option{-O2}, @option{-O3}.
4583 @itemx -falign-loops=@var{n}
4584 @opindex falign-loops
4585 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
4586 like @option{-falign-functions}. The hope is that the loop will be
4587 executed many times, which will make up for any execution of the dummy
4590 @option{-fno-align-loops} and @option{-falign-loops=1} are
4591 equivalent and mean that loops will not be aligned.
4593 If @var{n} is not specified or is zero, use a machine-dependent default.
4595 Enabled at levels @option{-O2}, @option{-O3}.
4598 @itemx -falign-jumps=@var{n}
4599 @opindex falign-jumps
4600 Align branch targets to a power-of-two boundary, for branch targets
4601 where the targets can only be reached by jumping, skipping up to @var{n}
4602 bytes like @option{-falign-functions}. In this case, no dummy operations
4605 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
4606 equivalent and mean that loops will not be aligned.
4608 If @var{n} is not specified or is zero, use a machine-dependent default.
4610 Enabled at levels @option{-O2}, @option{-O3}.
4612 @item -funit-at-a-time
4613 @opindex funit-at-a-time
4614 Parse the whole compilation unit before starting to produce code.
4615 This allows some extra optimizations to take place but consumes
4616 more memory (in general). There are some compatibility issues
4617 with @emph{unit-at-at-time} mode:
4620 enabling @emph{unit-at-a-time} mode may change the order
4621 in which functions, variables, and top-level @code{asm} statements
4622 are emitted, and will likely break code relying on some particular
4623 ordering. The majority of such top-level @code{asm} statements,
4624 though, can be replaced by @code{section} attributes.
4627 @emph{unit-at-a-time} mode removes unreferenced static variables
4628 and functions are removed. This may result in undefined references
4629 when an @code{asm} statement refers directly to variables or functions
4630 that are otherwise unused. In that case either the variable/function
4631 shall be listed as an operand of the @code{asm} statement operand or,
4632 in the case of top-level @code{asm} statements the attribute @code{used}
4633 shall be used on the declaration.
4636 Static functions now can use non-standard passing conventions that
4637 may break @code{asm} statements calling functions directly. Again,
4638 attribute @code{used} will prevent this behavior.
4641 As a temporary workaround, @option{-fno-unit-at-a-time} can be used,
4642 but this scheme may not be supported by future releases of GCC.
4644 Enabled at levels @option{-O2}, @option{-O3}.
4648 Constructs webs as commonly used for register allocation purposes and assign
4649 each web individual pseudo register. This allows our register allocation pass
4650 to operate on pseudos directly, but also strengthens several other optimization
4651 passes, such as CSE, loop optimizer and trivial dead code remover. It can,
4652 however, make debugging impossible, since variables will no longer stay in a
4655 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os},
4656 on targets where the default format for debugging information supports
4659 @item -fno-cprop-registers
4660 @opindex fno-cprop-registers
4661 After register allocation and post-register allocation instruction splitting,
4662 we perform a copy-propagation pass to try to reduce scheduling dependencies
4663 and occasionally eliminate the copy.
4665 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4667 @item -fprofile-generate
4668 @opindex fprofile-generate
4670 Enable options usually used for instrumenting application to produce
4671 profile useful for later recompilation with profile feedback based
4672 optimization. You must use @code{-fprofile-generate} both when
4673 compiling and when linking your program.
4675 The following options are enabled: @code{-fprofile-arcs}, @code{-fprofile-values}, @code{-fvpt}.
4678 @opindex fprofile-use
4679 Enable profile feedback directed optimizations, and optimizations
4680 generally profitable only with profile feedback available.
4682 The following options are enabled: @code{-fbranch-probabilities},
4683 @code{-fvpt}, @code{-funroll-loops}, @code{-fpeel-loops}, @code{-ftracer}.
4687 The following options control compiler behavior regarding floating
4688 point arithmetic. These options trade off between speed and
4689 correctness. All must be specifically enabled.
4693 @opindex ffloat-store
4694 Do not store floating point variables in registers, and inhibit other
4695 options that might change whether a floating point value is taken from a
4698 @cindex floating point precision
4699 This option prevents undesirable excess precision on machines such as
4700 the 68000 where the floating registers (of the 68881) keep more
4701 precision than a @code{double} is supposed to have. Similarly for the
4702 x86 architecture. For most programs, the excess precision does only
4703 good, but a few programs rely on the precise definition of IEEE floating
4704 point. Use @option{-ffloat-store} for such programs, after modifying
4705 them to store all pertinent intermediate computations into variables.
4709 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations}, @*
4710 @option{-fno-trapping-math}, @option{-ffinite-math-only},
4711 @option{-fno-rounding-math} and @option{-fno-signaling-nans}.
4713 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
4715 This option should never be turned on by any @option{-O} option since
4716 it can result in incorrect output for programs which depend on
4717 an exact implementation of IEEE or ISO rules/specifications for
4720 @item -fno-math-errno
4721 @opindex fno-math-errno
4722 Do not set ERRNO after calling math functions that are executed
4723 with a single instruction, e.g., sqrt. A program that relies on
4724 IEEE exceptions for math error handling may want to use this flag
4725 for speed while maintaining IEEE arithmetic compatibility.
4727 This option should never be turned on by any @option{-O} option since
4728 it can result in incorrect output for programs which depend on
4729 an exact implementation of IEEE or ISO rules/specifications for
4732 The default is @option{-fmath-errno}.
4734 @item -funsafe-math-optimizations
4735 @opindex funsafe-math-optimizations
4736 Allow optimizations for floating-point arithmetic that (a) assume
4737 that arguments and results are valid and (b) may violate IEEE or
4738 ANSI standards. When used at link-time, it may include libraries
4739 or startup files that change the default FPU control word or other
4740 similar optimizations.
4742 This option should never be turned on by any @option{-O} option since
4743 it can result in incorrect output for programs which depend on
4744 an exact implementation of IEEE or ISO rules/specifications for
4747 The default is @option{-fno-unsafe-math-optimizations}.
4749 @item -ffinite-math-only
4750 @opindex ffinite-math-only
4751 Allow optimizations for floating-point arithmetic that assume
4752 that arguments and results are not NaNs or +-Infs.
4754 This option should never be turned on by any @option{-O} option since
4755 it can result in incorrect output for programs which depend on
4756 an exact implementation of IEEE or ISO rules/specifications.
4758 The default is @option{-fno-finite-math-only}.
4760 @item -fno-trapping-math
4761 @opindex fno-trapping-math
4762 Compile code assuming that floating-point operations cannot generate
4763 user-visible traps. These traps include division by zero, overflow,
4764 underflow, inexact result and invalid operation. This option implies
4765 @option{-fno-signaling-nans}. Setting this option may allow faster
4766 code if one relies on ``non-stop'' IEEE arithmetic, for example.
4768 This option should never be turned on by any @option{-O} option since
4769 it can result in incorrect output for programs which depend on
4770 an exact implementation of IEEE or ISO rules/specifications for
4773 The default is @option{-ftrapping-math}.
4775 @item -frounding-math
4776 @opindex frounding-math
4777 Disable transformations and optimizations that assume default floating
4778 point rounding behavior. This is round-to-zero for all floating point
4779 to integer conversions, and round-to-nearest for all other arithmetic
4780 truncations. This option should be specified for programs that change
4781 the FP rounding mode dynamically, or that may be executed with a
4782 non-default rounding mode. This option disables constant folding of
4783 floating point expressions at compile-time (which may be affected by
4784 rounding mode) and arithmetic transformations that are unsafe in the
4785 presence of sign-dependent rounding modes.
4787 The default is @option{-fno-rounding-math}.
4789 This option is experimental and does not currently guarantee to
4790 disable all GCC optimizations that are affected by rounding mode.
4791 Future versions of GCC may provide finer control of this setting
4792 using C99's @code{FENV_ACCESS} pragma. This command line option
4793 will be used to specify the default state for @code{FENV_ACCESS}.
4795 @item -fsignaling-nans
4796 @opindex fsignaling-nans
4797 Compile code assuming that IEEE signaling NaNs may generate user-visible
4798 traps during floating-point operations. Setting this option disables
4799 optimizations that may change the number of exceptions visible with
4800 signaling NaNs. This option implies @option{-ftrapping-math}.
4802 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
4805 The default is @option{-fno-signaling-nans}.
4807 This option is experimental and does not currently guarantee to
4808 disable all GCC optimizations that affect signaling NaN behavior.
4810 @item -fsingle-precision-constant
4811 @opindex fsingle-precision-constant
4812 Treat floating point constant as single precision constant instead of
4813 implicitly converting it to double precision constant.
4818 The following options control optimizations that may improve
4819 performance, but are not enabled by any @option{-O} options. This
4820 section includes experimental options that may produce broken code.
4823 @item -fbranch-probabilities
4824 @opindex fbranch-probabilities
4825 After running a program compiled with @option{-fprofile-arcs}
4826 (@pxref{Debugging Options,, Options for Debugging Your Program or
4827 @command{gcc}}), you can compile it a second time using
4828 @option{-fbranch-probabilities}, to improve optimizations based on
4829 the number of times each branch was taken. When the program
4830 compiled with @option{-fprofile-arcs} exits it saves arc execution
4831 counts to a file called @file{@var{sourcename}.gcda} for each source
4832 file The information in this data file is very dependent on the
4833 structure of the generated code, so you must use the same source code
4834 and the same optimization options for both compilations.
4836 With @option{-fbranch-probabilities}, GCC puts a
4837 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
4838 These can be used to improve optimization. Currently, they are only
4839 used in one place: in @file{reorg.c}, instead of guessing which path a
4840 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
4841 exactly determine which path is taken more often.
4843 @item -fprofile-values
4844 @opindex fprofile-values
4845 If combined with @option{-fprofile-arcs}, it adds code so that some
4846 data about values of expressions in the program is gathered.
4848 With @option{-fbranch-probabilities}, it reads back the data gathered
4849 from profiling values of expressions and adds @samp{REG_VALUE_PROFILE}
4850 notes to instructions for their later usage in optimizations.
4852 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
4856 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
4857 a code to gather information about values of expressions.
4859 With @option{-fbranch-probabilities}, it reads back the data gathered
4860 and actually performs the optimizations based on them.
4861 Currently the optimizations include specialization of division operation
4862 using the knowledge about the value of the denominator.
4864 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
4866 @item -frename-registers
4867 @opindex frename-registers
4868 Attempt to avoid false dependencies in scheduled code by making use
4869 of registers left over after register allocation. This optimization
4870 will most benefit processors with lots of registers. Depending on the
4871 debug information format adopted by the target, however, it can
4872 make debugging impossible, since variables will no longer stay in
4873 a ``home register''.
4875 Not enabled by default at any level because it has known bugs.
4879 Use a graph coloring register allocator. Currently this option is meant
4880 for testing, so we are interested to hear about miscompilations with
4885 Perform tail duplication to enlarge superblock size. This transformation
4886 simplifies the control flow of the function allowing other optimizations to do
4889 Enabled with @option{-fprofile-use}.
4891 @item -funroll-loops
4892 @opindex funroll-loops
4893 Unroll loops whose number of iterations can be determined at compile time or
4894 upon entry to the loop. @option{-funroll-loops} implies
4895 @option{-frerun-cse-after-loop}. It also turns on complete loop peeling
4896 (i.e. complete removal of loops with small constant number of iterations).
4897 This option makes code larger, and may or may not make it run faster.
4899 Enabled with @option{-fprofile-use}.
4901 @item -funroll-all-loops
4902 @opindex funroll-all-loops
4903 Unroll all loops, even if their number of iterations is uncertain when
4904 the loop is entered. This usually makes programs run more slowly.
4905 @option{-funroll-all-loops} implies the same options as
4906 @option{-funroll-loops}.
4909 @opindex fpeel-loops
4910 Peels the loops for that there is enough information that they do not
4911 roll much (from profile feedback). It also turns on complete loop peeling
4912 (i.e. complete removal of loops with small constant number of iterations).
4914 Enabled with @option{-fprofile-use}.
4916 @item -funswitch-loops
4917 @opindex funswitch-loops
4918 Move branches with loop invariant conditions out of the loop, with duplicates
4919 of the loop on both branches (modified according to result of the condition).
4921 @item -fold-unroll-loops
4922 @opindex fold-unroll-loops
4923 Unroll loops whose number of iterations can be determined at compile
4924 time or upon entry to the loop, using the old loop unroller whose loop
4925 recognition is based on notes from frontend. @option{-fold-unroll-loops} implies
4926 both @option{-fstrength-reduce} and @option{-frerun-cse-after-loop}. This
4927 option makes code larger, and may or may not make it run faster.
4929 @item -fold-unroll-all-loops
4930 @opindex fold-unroll-all-loops
4931 Unroll all loops, even if their number of iterations is uncertain when
4932 the loop is entered. This is done using the old loop unroller whose loop
4933 recognition is based on notes from frontend. This usually makes programs run more slowly.
4934 @option{-fold-unroll-all-loops} implies the same options as
4935 @option{-fold-unroll-loops}.
4937 @item -fprefetch-loop-arrays
4938 @opindex fprefetch-loop-arrays
4939 If supported by the target machine, generate instructions to prefetch
4940 memory to improve the performance of loops that access large arrays.
4942 Disabled at level @option{-Os}.
4944 @item -ffunction-sections
4945 @itemx -fdata-sections
4946 @opindex ffunction-sections
4947 @opindex fdata-sections
4948 Place each function or data item into its own section in the output
4949 file if the target supports arbitrary sections. The name of the
4950 function or the name of the data item determines the section's name
4953 Use these options on systems where the linker can perform optimizations
4954 to improve locality of reference in the instruction space. Most systems
4955 using the ELF object format and SPARC processors running Solaris 2 have
4956 linkers with such optimizations. AIX may have these optimizations in
4959 Only use these options when there are significant benefits from doing
4960 so. When you specify these options, the assembler and linker will
4961 create larger object and executable files and will also be slower.
4962 You will not be able to use @code{gprof} on all systems if you
4963 specify this option and you may have problems with debugging if
4964 you specify both this option and @option{-g}.
4966 @item -fbranch-target-load-optimize
4967 @opindex fbranch-target-load-optimize
4968 Perform branch target register load optimization before prologue / epilogue
4970 The use of target registers can typically be exposed only during reload,
4971 thus hoisting loads out of loops and doing inter-block scheduling needs
4972 a separate optimization pass.
4974 @item -fbranch-target-load-optimize2
4975 @opindex fbranch-target-load-optimize2
4976 Perform branch target register load optimization after prologue / epilogue
4979 @item -fbtr-bb-exclusive
4980 @opindex fbtr-bb-exclusive
4981 When performing branch target register load optimization, don't reuse
4982 branch target registers in within any basic block.
4984 @item --param @var{name}=@var{value}
4986 In some places, GCC uses various constants to control the amount of
4987 optimization that is done. For example, GCC will not inline functions
4988 that contain more that a certain number of instructions. You can
4989 control some of these constants on the command-line using the
4990 @option{--param} option.
4992 The names of specific parameters, and the meaning of the values, are
4993 tied to the internals of the compiler, and are subject to change
4994 without notice in future releases.
4996 In each case, the @var{value} is an integer. The allowable choices for
4997 @var{name} are given in the following table:
5000 @item max-crossjump-edges
5001 The maximum number of incoming edges to consider for crossjumping.
5002 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
5003 the number of edges incoming to each block. Increasing values mean
5004 more aggressive optimization, making the compile time increase with
5005 probably small improvement in executable size.
5007 @item max-delay-slot-insn-search
5008 The maximum number of instructions to consider when looking for an
5009 instruction to fill a delay slot. If more than this arbitrary number of
5010 instructions is searched, the time savings from filling the delay slot
5011 will be minimal so stop searching. Increasing values mean more
5012 aggressive optimization, making the compile time increase with probably
5013 small improvement in executable run time.
5015 @item max-delay-slot-live-search
5016 When trying to fill delay slots, the maximum number of instructions to
5017 consider when searching for a block with valid live register
5018 information. Increasing this arbitrarily chosen value means more
5019 aggressive optimization, increasing the compile time. This parameter
5020 should be removed when the delay slot code is rewritten to maintain the
5023 @item max-gcse-memory
5024 The approximate maximum amount of memory that will be allocated in
5025 order to perform the global common subexpression elimination
5026 optimization. If more memory than specified is required, the
5027 optimization will not be done.
5029 @item max-gcse-passes
5030 The maximum number of passes of GCSE to run. The default is 1.
5032 @item max-pending-list-length
5033 The maximum number of pending dependencies scheduling will allow
5034 before flushing the current state and starting over. Large functions
5035 with few branches or calls can create excessively large lists which
5036 needlessly consume memory and resources.
5038 @item max-inline-insns-single
5039 Several parameters control the tree inliner used in gcc.
5040 This number sets the maximum number of instructions (counted in GCC's
5041 internal representation) in a single function that the tree inliner
5042 will consider for inlining. This only affects functions declared
5043 inline and methods implemented in a class declaration (C++).
5044 The default value is 500.
5046 @item max-inline-insns-auto
5047 When you use @option{-finline-functions} (included in @option{-O3}),
5048 a lot of functions that would otherwise not be considered for inlining
5049 by the compiler will be investigated. To those functions, a different
5050 (more restrictive) limit compared to functions declared inline can
5052 The default value is 120.
5054 @item large-function-insns
5055 The limit specifying really large functions. For functions greater than this
5056 limit inlining is constrained by @option{--param large-function-growth}.
5057 This parameter is useful primarily to avoid extreme compilation time caused by non-linear
5058 algorithms used by the backend.
5059 This parameter is ignored when @option{-funit-at-a-time} is not used.
5060 The default value is 3000.
5062 @item large-function-growth
5063 Specifies maximal growth of large function caused by inlining in percents.
5064 This parameter is ignored when @option{-funit-at-a-time} is not used.
5065 The default value is 200.
5067 @item inline-unit-growth
5068 Specifies maximal overall growth of the compilation unit caused by inlining.
5069 This parameter is ignored when @option{-funit-at-a-time} is not used.
5070 The default value is 150.
5072 @item max-inline-insns-recursive
5073 @itemx max-inline-insns-recursive-auto
5074 Specifies maximum number of instructions out-of-line copy of self recursive inline
5075 function can grow into by performing recursive inlining.
5077 For functions declared inline @option{--param max-inline-insns-recursive} is
5078 taken into acount. For function not declared inline, recursive inlining
5079 happens only when @option{-finline-functions} (included in @option{-O3}) is
5080 enabled and @option{--param max-inline-insns-recursive-auto} is used. The
5081 default value is 500.
5083 @item max-inline-recursive-depth
5084 @itemx max-inline-recursive-depth-auto
5085 Specifies maximum recursion depth used by the recursive inlining.
5087 For functions declared inline @option{--param max-inline-recursive-depth} is
5088 taken into acount. For function not declared inline, recursive inlining
5089 happens only when @option{-finline-functions} (included in @option{-O3}) is
5090 enabled and @option{--param max-inline-recursive-depth-auto} is used. The
5091 default value is 500.
5093 @item max-inline-insns-rtl
5094 For languages that use the RTL inliner (this happens at a later stage
5095 than tree inlining), you can set the maximum allowable size (counted
5096 in RTL instructions) for the RTL inliner with this parameter.
5097 The default value is 600.
5099 @item max-unrolled-insns
5100 The maximum number of instructions that a loop should have if that loop
5101 is unrolled, and if the loop is unrolled, it determines how many times
5102 the loop code is unrolled.
5104 @item max-average-unrolled-insns
5105 The maximum number of instructions biased by probabilities of their execution
5106 that a loop should have if that loop is unrolled, and if the loop is unrolled,
5107 it determines how many times the loop code is unrolled.
5109 @item max-unroll-times
5110 The maximum number of unrollings of a single loop.
5112 @item max-peeled-insns
5113 The maximum number of instructions that a loop should have if that loop
5114 is peeled, and if the loop is peeled, it determines how many times
5115 the loop code is peeled.
5117 @item max-peel-times
5118 The maximum number of peelings of a single loop.
5120 @item max-completely-peeled-insns
5121 The maximum number of insns of a completely peeled loop.
5123 @item max-completely-peel-times
5124 The maximum number of iterations of a loop to be suitable for complete peeling.
5126 @item max-unswitch-insns
5127 The maximum number of insns of an unswitched loop.
5129 @item max-unswitch-level
5130 The maximum number of branches unswitched in a single loop.
5132 @item hot-bb-count-fraction
5133 Select fraction of the maximal count of repetitions of basic block in program
5134 given basic block needs to have to be considered hot.
5136 @item hot-bb-frequency-fraction
5137 Select fraction of the maximal frequency of executions of basic block in
5138 function given basic block needs to have to be considered hot
5140 @item tracer-dynamic-coverage
5141 @itemx tracer-dynamic-coverage-feedback
5143 This value is used to limit superblock formation once the given percentage of
5144 executed instructions is covered. This limits unnecessary code size
5147 The @option{tracer-dynamic-coverage-feedback} is used only when profile
5148 feedback is available. The real profiles (as opposed to statically estimated
5149 ones) are much less balanced allowing the threshold to be larger value.
5151 @item tracer-max-code-growth
5152 Stop tail duplication once code growth has reached given percentage. This is
5153 rather hokey argument, as most of the duplicates will be eliminated later in
5154 cross jumping, so it may be set to much higher values than is the desired code
5157 @item tracer-min-branch-ratio
5159 Stop reverse growth when the reverse probability of best edge is less than this
5160 threshold (in percent).
5162 @item tracer-min-branch-ratio
5163 @itemx tracer-min-branch-ratio-feedback
5165 Stop forward growth if the best edge do have probability lower than this
5168 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
5169 compilation for profile feedback and one for compilation without. The value
5170 for compilation with profile feedback needs to be more conservative (higher) in
5171 order to make tracer effective.
5173 @item max-cse-path-length
5175 Maximum number of basic blocks on path that cse considers. The default is 10.
5177 @item global-var-threshold
5179 Counts the number of function calls (N) and the number of
5180 call-clobbered variables (V). If NxV is larger than this limit, a
5181 single artificial variable will be created to represent all the
5182 call-clobbered variables at function call sites. This artificial
5183 variable will then be made to alias every call-clobbered variable.
5184 (done as int * size_t on the host machine; beware overflow).
5186 @item max-aliased-vops
5188 Maxiumum number of virtual operands allowed to represent aliases
5189 before triggering the alias grouping heuristic. Alias grouping
5190 reduces compile times and memory consumption needed for aliasing at
5191 the expense of precision loss in alias information.
5193 @item ggc-min-expand
5195 GCC uses a garbage collector to manage its own memory allocation. This
5196 parameter specifies the minimum percentage by which the garbage
5197 collector's heap should be allowed to expand between collections.
5198 Tuning this may improve compilation speed; it has no effect on code
5201 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
5202 RAM >= 1GB. If @code{getrlimit} is available, the notion of "RAM" is
5203 the smallest of actual RAM, RLIMIT_RSS, RLIMIT_DATA and RLIMIT_AS. If
5204 GCC is not able to calculate RAM on a particular platform, the lower
5205 bound of 30% is used. Setting this parameter and
5206 @option{ggc-min-heapsize} to zero causes a full collection to occur at
5207 every opportunity. This is extremely slow, but can be useful for
5210 @item ggc-min-heapsize
5212 Minimum size of the garbage collector's heap before it begins bothering
5213 to collect garbage. The first collection occurs after the heap expands
5214 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
5215 tuning this may improve compilation speed, and has no effect on code
5218 The default is RAM/8, with a lower bound of 4096 (four megabytes) and an
5219 upper bound of 131072 (128 megabytes). If @code{getrlimit} is
5220 available, the notion of "RAM" is the smallest of actual RAM,
5221 RLIMIT_RSS, RLIMIT_DATA and RLIMIT_AS. If GCC is not able to calculate
5222 RAM on a particular platform, the lower bound is used. Setting this
5223 parameter very large effectively disables garbage collection. Setting
5224 this parameter and @option{ggc-min-expand} to zero causes a full
5225 collection to occur at every opportunity.
5227 @item max-reload-search-insns
5228 The maximum number of instruction reload should look backward for equivalent
5229 register. Increasing values mean more aggressive optimization, making the
5230 compile time increase with probably slightly better performance. The default
5233 @item max-cselib-memory-location
5234 The maximum number of memory locations cselib should take into acount.
5235 Increasing values mean more aggressive optimization, making the compile time
5236 increase with probably slightly better performance. The default value is 500.
5238 @item reorder-blocks-duplicate
5239 @itemx reorder-blocks-duplicate-feedback
5241 Used by basic block reordering pass to decide whether to use unconditional
5242 branch or duplicate the code on its destination. Code is duplicated when its
5243 estimated size is smaller than this value multiplied by the estimated size of
5244 unconditional jump in the hot spots of the program.
5246 The @option{reorder-block-duplicate-feedback} is used only when profile
5247 feedback is available and may be set to higher values than
5248 @option{reorder-block-duplicate} since information about the hot spots is more
5251 @item max-sched-region-blocks
5252 The maximum number of blocks in a region to be considered for
5253 interblock scheduling. The default value is 10.
5255 @item max-sched-region-insns
5256 The maximum number of insns in a region to be considered for
5257 interblock scheduling. The default value is 100.
5261 @node Preprocessor Options
5262 @section Options Controlling the Preprocessor
5263 @cindex preprocessor options
5264 @cindex options, preprocessor
5266 These options control the C preprocessor, which is run on each C source
5267 file before actual compilation.
5269 If you use the @option{-E} option, nothing is done except preprocessing.
5270 Some of these options make sense only together with @option{-E} because
5271 they cause the preprocessor output to be unsuitable for actual
5276 You can use @option{-Wp,@var{option}} to bypass the compiler driver
5277 and pass @var{option} directly through to the preprocessor. If
5278 @var{option} contains commas, it is split into multiple options at the
5279 commas. However, many options are modified, translated or interpreted
5280 by the compiler driver before being passed to the preprocessor, and
5281 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
5282 interface is undocumented and subject to change, so whenever possible
5283 you should avoid using @option{-Wp} and let the driver handle the
5286 @item -Xpreprocessor @var{option}
5287 @opindex preprocessor
5288 Pass @var{option} as an option to the preprocessor. You can use this to
5289 supply system-specific preprocessor options which GCC does not know how to
5292 If you want to pass an option that takes an argument, you must use
5293 @option{-Xpreprocessor} twice, once for the option and once for the argument.
5296 @include cppopts.texi
5298 @node Assembler Options
5299 @section Passing Options to the Assembler
5301 @c prevent bad page break with this line
5302 You can pass options to the assembler.
5305 @item -Wa,@var{option}
5307 Pass @var{option} as an option to the assembler. If @var{option}
5308 contains commas, it is split into multiple options at the commas.
5310 @item -Xassembler @var{option}
5312 Pass @var{option} as an option to the assembler. You can use this to
5313 supply system-specific assembler options which GCC does not know how to
5316 If you want to pass an option that takes an argument, you must use
5317 @option{-Xassembler} twice, once for the option and once for the argument.
5322 @section Options for Linking
5323 @cindex link options
5324 @cindex options, linking
5326 These options come into play when the compiler links object files into
5327 an executable output file. They are meaningless if the compiler is
5328 not doing a link step.
5332 @item @var{object-file-name}
5333 A file name that does not end in a special recognized suffix is
5334 considered to name an object file or library. (Object files are
5335 distinguished from libraries by the linker according to the file
5336 contents.) If linking is done, these object files are used as input
5345 If any of these options is used, then the linker is not run, and
5346 object file names should not be used as arguments. @xref{Overall
5350 @item -l@var{library}
5351 @itemx -l @var{library}
5353 Search the library named @var{library} when linking. (The second
5354 alternative with the library as a separate argument is only for
5355 POSIX compliance and is not recommended.)
5357 It makes a difference where in the command you write this option; the
5358 linker searches and processes libraries and object files in the order they
5359 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
5360 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
5361 to functions in @samp{z}, those functions may not be loaded.
5363 The linker searches a standard list of directories for the library,
5364 which is actually a file named @file{lib@var{library}.a}. The linker
5365 then uses this file as if it had been specified precisely by name.
5367 The directories searched include several standard system directories
5368 plus any that you specify with @option{-L}.
5370 Normally the files found this way are library files---archive files
5371 whose members are object files. The linker handles an archive file by
5372 scanning through it for members which define symbols that have so far
5373 been referenced but not defined. But if the file that is found is an
5374 ordinary object file, it is linked in the usual fashion. The only
5375 difference between using an @option{-l} option and specifying a file name
5376 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
5377 and searches several directories.
5381 You need this special case of the @option{-l} option in order to
5382 link an Objective-C program.
5385 @opindex nostartfiles
5386 Do not use the standard system startup files when linking.
5387 The standard system libraries are used normally, unless @option{-nostdlib}
5388 or @option{-nodefaultlibs} is used.
5390 @item -nodefaultlibs
5391 @opindex nodefaultlibs
5392 Do not use the standard system libraries when linking.
5393 Only the libraries you specify will be passed to the linker.
5394 The standard startup files are used normally, unless @option{-nostartfiles}
5395 is used. The compiler may generate calls to memcmp, memset, and memcpy
5396 for System V (and ISO C) environments or to bcopy and bzero for
5397 BSD environments. These entries are usually resolved by entries in
5398 libc. These entry points should be supplied through some other
5399 mechanism when this option is specified.
5403 Do not use the standard system startup files or libraries when linking.
5404 No startup files and only the libraries you specify will be passed to
5405 the linker. The compiler may generate calls to memcmp, memset, and memcpy
5406 for System V (and ISO C) environments or to bcopy and bzero for
5407 BSD environments. These entries are usually resolved by entries in
5408 libc. These entry points should be supplied through some other
5409 mechanism when this option is specified.
5411 @cindex @option{-lgcc}, use with @option{-nostdlib}
5412 @cindex @option{-nostdlib} and unresolved references
5413 @cindex unresolved references and @option{-nostdlib}
5414 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
5415 @cindex @option{-nodefaultlibs} and unresolved references
5416 @cindex unresolved references and @option{-nodefaultlibs}
5417 One of the standard libraries bypassed by @option{-nostdlib} and
5418 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
5419 that GCC uses to overcome shortcomings of particular machines, or special
5420 needs for some languages.
5421 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
5422 Collection (GCC) Internals},
5423 for more discussion of @file{libgcc.a}.)
5424 In most cases, you need @file{libgcc.a} even when you want to avoid
5425 other standard libraries. In other words, when you specify @option{-nostdlib}
5426 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
5427 This ensures that you have no unresolved references to internal GCC
5428 library subroutines. (For example, @samp{__main}, used to ensure C++
5429 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
5430 GNU Compiler Collection (GCC) Internals}.)
5434 Produce a position independent executable on targets which support it.
5435 For predictable results, you must also specify the same set of options
5436 that were used to generate code (@option{-fpie}, @option{-fPIE},
5437 or model suboptions) when you specify this option.
5441 Remove all symbol table and relocation information from the executable.
5445 On systems that support dynamic linking, this prevents linking with the shared
5446 libraries. On other systems, this option has no effect.
5450 Produce a shared object which can then be linked with other objects to
5451 form an executable. Not all systems support this option. For predictable
5452 results, you must also specify the same set of options that were used to
5453 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
5454 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
5455 needs to build supplementary stub code for constructors to work. On
5456 multi-libbed systems, @samp{gcc -shared} must select the correct support
5457 libraries to link against. Failing to supply the correct flags may lead
5458 to subtle defects. Supplying them in cases where they are not necessary
5461 @item -shared-libgcc
5462 @itemx -static-libgcc
5463 @opindex shared-libgcc
5464 @opindex static-libgcc
5465 On systems that provide @file{libgcc} as a shared library, these options
5466 force the use of either the shared or static version respectively.
5467 If no shared version of @file{libgcc} was built when the compiler was
5468 configured, these options have no effect.
5470 There are several situations in which an application should use the
5471 shared @file{libgcc} instead of the static version. The most common
5472 of these is when the application wishes to throw and catch exceptions
5473 across different shared libraries. In that case, each of the libraries
5474 as well as the application itself should use the shared @file{libgcc}.
5476 Therefore, the G++ and GCJ drivers automatically add
5477 @option{-shared-libgcc} whenever you build a shared library or a main
5478 executable, because C++ and Java programs typically use exceptions, so
5479 this is the right thing to do.
5481 If, instead, you use the GCC driver to create shared libraries, you may
5482 find that they will not always be linked with the shared @file{libgcc}.
5483 If GCC finds, at its configuration time, that you have a GNU linker that
5484 does not support option @option{--eh-frame-hdr}, it will link the shared
5485 version of @file{libgcc} into shared libraries by default. Otherwise,
5486 it will take advantage of the linker and optimize away the linking with
5487 the shared version of @file{libgcc}, linking with the static version of
5488 libgcc by default. This allows exceptions to propagate through such
5489 shared libraries, without incurring relocation costs at library load
5492 However, if a library or main executable is supposed to throw or catch
5493 exceptions, you must link it using the G++ or GCJ driver, as appropriate
5494 for the languages used in the program, or using the option
5495 @option{-shared-libgcc}, such that it is linked with the shared
5500 Bind references to global symbols when building a shared object. Warn
5501 about any unresolved references (unless overridden by the link editor
5502 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
5505 @item -Xlinker @var{option}
5507 Pass @var{option} as an option to the linker. You can use this to
5508 supply system-specific linker options which GCC does not know how to
5511 If you want to pass an option that takes an argument, you must use
5512 @option{-Xlinker} twice, once for the option and once for the argument.
5513 For example, to pass @option{-assert definitions}, you must write
5514 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
5515 @option{-Xlinker "-assert definitions"}, because this passes the entire
5516 string as a single argument, which is not what the linker expects.
5518 @item -Wl,@var{option}
5520 Pass @var{option} as an option to the linker. If @var{option} contains
5521 commas, it is split into multiple options at the commas.
5523 @item -u @var{symbol}
5525 Pretend the symbol @var{symbol} is undefined, to force linking of
5526 library modules to define it. You can use @option{-u} multiple times with
5527 different symbols to force loading of additional library modules.
5530 @node Directory Options
5531 @section Options for Directory Search
5532 @cindex directory options
5533 @cindex options, directory search
5536 These options specify directories to search for header files, for
5537 libraries and for parts of the compiler:
5542 Add the directory @var{dir} to the head of the list of directories to be
5543 searched for header files. This can be used to override a system header
5544 file, substituting your own version, since these directories are
5545 searched before the system header file directories. However, you should
5546 not use this option to add directories that contain vendor-supplied
5547 system header files (use @option{-isystem} for that). If you use more than
5548 one @option{-I} option, the directories are scanned in left-to-right
5549 order; the standard system directories come after.
5551 If a standard system include directory, or a directory specified with
5552 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
5553 option will be ignored. The directory will still be searched but as a
5554 system directory at its normal position in the system include chain.
5555 This is to ensure that GCC's procedure to fix buggy system headers and
5556 the ordering for the include_next directive are not inadvertently changed.
5557 If you really need to change the search order for system directories,
5558 use the @option{-nostdinc} and/or @option{-isystem} options.
5560 @item -iquote@var{dir}
5562 Add the directory @var{dir} to the head of the list of directories to
5563 be searched for header files only for the case of @samp{#include
5564 "@var{file}"}; they are not searched for @samp{#include <@var{file}>},
5565 otherwise just like @option{-I}.
5569 Add directory @var{dir} to the list of directories to be searched
5572 @item -B@var{prefix}
5574 This option specifies where to find the executables, libraries,
5575 include files, and data files of the compiler itself.
5577 The compiler driver program runs one or more of the subprograms
5578 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
5579 @var{prefix} as a prefix for each program it tries to run, both with and
5580 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
5582 For each subprogram to be run, the compiler driver first tries the
5583 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
5584 was not specified, the driver tries two standard prefixes, which are
5585 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
5586 those results in a file name that is found, the unmodified program
5587 name is searched for using the directories specified in your
5588 @env{PATH} environment variable.
5590 The compiler will check to see if the path provided by the @option{-B}
5591 refers to a directory, and if necessary it will add a directory
5592 separator character at the end of the path.
5594 @option{-B} prefixes that effectively specify directory names also apply
5595 to libraries in the linker, because the compiler translates these
5596 options into @option{-L} options for the linker. They also apply to
5597 includes files in the preprocessor, because the compiler translates these
5598 options into @option{-isystem} options for the preprocessor. In this case,
5599 the compiler appends @samp{include} to the prefix.
5601 The run-time support file @file{libgcc.a} can also be searched for using
5602 the @option{-B} prefix, if needed. If it is not found there, the two
5603 standard prefixes above are tried, and that is all. The file is left
5604 out of the link if it is not found by those means.
5606 Another way to specify a prefix much like the @option{-B} prefix is to use
5607 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
5610 As a special kludge, if the path provided by @option{-B} is
5611 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
5612 9, then it will be replaced by @file{[dir/]include}. This is to help
5613 with boot-strapping the compiler.
5615 @item -specs=@var{file}
5617 Process @var{file} after the compiler reads in the standard @file{specs}
5618 file, in order to override the defaults that the @file{gcc} driver
5619 program uses when determining what switches to pass to @file{cc1},
5620 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
5621 @option{-specs=@var{file}} can be specified on the command line, and they
5622 are processed in order, from left to right.
5626 This option has been deprecated. Please use @option{-iquote} instead for
5627 @option{-I} directories before the @option{-I-} and remove the @option{-I-}.
5628 Any directories you specify with @option{-I} options before the @option{-I-}
5629 option are searched only for the case of @samp{#include "@var{file}"};
5630 they are not searched for @samp{#include <@var{file}>}.
5632 If additional directories are specified with @option{-I} options after
5633 the @option{-I-}, these directories are searched for all @samp{#include}
5634 directives. (Ordinarily @emph{all} @option{-I} directories are used
5637 In addition, the @option{-I-} option inhibits the use of the current
5638 directory (where the current input file came from) as the first search
5639 directory for @samp{#include "@var{file}"}. There is no way to
5640 override this effect of @option{-I-}. With @option{-I.} you can specify
5641 searching the directory which was current when the compiler was
5642 invoked. That is not exactly the same as what the preprocessor does
5643 by default, but it is often satisfactory.
5645 @option{-I-} does not inhibit the use of the standard system directories
5646 for header files. Thus, @option{-I-} and @option{-nostdinc} are
5653 @section Specifying subprocesses and the switches to pass to them
5656 @command{gcc} is a driver program. It performs its job by invoking a
5657 sequence of other programs to do the work of compiling, assembling and
5658 linking. GCC interprets its command-line parameters and uses these to
5659 deduce which programs it should invoke, and which command-line options
5660 it ought to place on their command lines. This behavior is controlled
5661 by @dfn{spec strings}. In most cases there is one spec string for each
5662 program that GCC can invoke, but a few programs have multiple spec
5663 strings to control their behavior. The spec strings built into GCC can
5664 be overridden by using the @option{-specs=} command-line switch to specify
5667 @dfn{Spec files} are plaintext files that are used to construct spec
5668 strings. They consist of a sequence of directives separated by blank
5669 lines. The type of directive is determined by the first non-whitespace
5670 character on the line and it can be one of the following:
5673 @item %@var{command}
5674 Issues a @var{command} to the spec file processor. The commands that can
5678 @item %include <@var{file}>
5680 Search for @var{file} and insert its text at the current point in the
5683 @item %include_noerr <@var{file}>
5684 @cindex %include_noerr
5685 Just like @samp{%include}, but do not generate an error message if the include
5686 file cannot be found.
5688 @item %rename @var{old_name} @var{new_name}
5690 Rename the spec string @var{old_name} to @var{new_name}.
5694 @item *[@var{spec_name}]:
5695 This tells the compiler to create, override or delete the named spec
5696 string. All lines after this directive up to the next directive or
5697 blank line are considered to be the text for the spec string. If this
5698 results in an empty string then the spec will be deleted. (Or, if the
5699 spec did not exist, then nothing will happened.) Otherwise, if the spec
5700 does not currently exist a new spec will be created. If the spec does
5701 exist then its contents will be overridden by the text of this
5702 directive, unless the first character of that text is the @samp{+}
5703 character, in which case the text will be appended to the spec.
5705 @item [@var{suffix}]:
5706 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
5707 and up to the next directive or blank line are considered to make up the
5708 spec string for the indicated suffix. When the compiler encounters an
5709 input file with the named suffix, it will processes the spec string in
5710 order to work out how to compile that file. For example:
5717 This says that any input file whose name ends in @samp{.ZZ} should be
5718 passed to the program @samp{z-compile}, which should be invoked with the
5719 command-line switch @option{-input} and with the result of performing the
5720 @samp{%i} substitution. (See below.)
5722 As an alternative to providing a spec string, the text that follows a
5723 suffix directive can be one of the following:
5726 @item @@@var{language}
5727 This says that the suffix is an alias for a known @var{language}. This is
5728 similar to using the @option{-x} command-line switch to GCC to specify a
5729 language explicitly. For example:
5736 Says that .ZZ files are, in fact, C++ source files.
5739 This causes an error messages saying:
5742 @var{name} compiler not installed on this system.
5746 GCC already has an extensive list of suffixes built into it.
5747 This directive will add an entry to the end of the list of suffixes, but
5748 since the list is searched from the end backwards, it is effectively
5749 possible to override earlier entries using this technique.
5753 GCC has the following spec strings built into it. Spec files can
5754 override these strings or create their own. Note that individual
5755 targets can also add their own spec strings to this list.
5758 asm Options to pass to the assembler
5759 asm_final Options to pass to the assembler post-processor
5760 cpp Options to pass to the C preprocessor
5761 cc1 Options to pass to the C compiler
5762 cc1plus Options to pass to the C++ compiler
5763 endfile Object files to include at the end of the link
5764 link Options to pass to the linker
5765 lib Libraries to include on the command line to the linker
5766 libgcc Decides which GCC support library to pass to the linker
5767 linker Sets the name of the linker
5768 predefines Defines to be passed to the C preprocessor
5769 signed_char Defines to pass to CPP to say whether @code{char} is signed
5771 startfile Object files to include at the start of the link
5774 Here is a small example of a spec file:
5780 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
5783 This example renames the spec called @samp{lib} to @samp{old_lib} and
5784 then overrides the previous definition of @samp{lib} with a new one.
5785 The new definition adds in some extra command-line options before
5786 including the text of the old definition.
5788 @dfn{Spec strings} are a list of command-line options to be passed to their
5789 corresponding program. In addition, the spec strings can contain
5790 @samp{%}-prefixed sequences to substitute variable text or to
5791 conditionally insert text into the command line. Using these constructs
5792 it is possible to generate quite complex command lines.
5794 Here is a table of all defined @samp{%}-sequences for spec
5795 strings. Note that spaces are not generated automatically around the
5796 results of expanding these sequences. Therefore you can concatenate them
5797 together or combine them with constant text in a single argument.
5801 Substitute one @samp{%} into the program name or argument.
5804 Substitute the name of the input file being processed.
5807 Substitute the basename of the input file being processed.
5808 This is the substring up to (and not including) the last period
5809 and not including the directory.
5812 This is the same as @samp{%b}, but include the file suffix (text after
5816 Marks the argument containing or following the @samp{%d} as a
5817 temporary file name, so that that file will be deleted if GCC exits
5818 successfully. Unlike @samp{%g}, this contributes no text to the
5821 @item %g@var{suffix}
5822 Substitute a file name that has suffix @var{suffix} and is chosen
5823 once per compilation, and mark the argument in the same way as
5824 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
5825 name is now chosen in a way that is hard to predict even when previously
5826 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
5827 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
5828 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
5829 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
5830 was simply substituted with a file name chosen once per compilation,
5831 without regard to any appended suffix (which was therefore treated
5832 just like ordinary text), making such attacks more likely to succeed.
5834 @item %u@var{suffix}
5835 Like @samp{%g}, but generates a new temporary file name even if
5836 @samp{%u@var{suffix}} was already seen.
5838 @item %U@var{suffix}
5839 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
5840 new one if there is no such last file name. In the absence of any
5841 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
5842 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
5843 would involve the generation of two distinct file names, one
5844 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
5845 simply substituted with a file name chosen for the previous @samp{%u},
5846 without regard to any appended suffix.
5848 @item %j@var{suffix}
5849 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
5850 writable, and if save-temps is off; otherwise, substitute the name
5851 of a temporary file, just like @samp{%u}. This temporary file is not
5852 meant for communication between processes, but rather as a junk
5855 @item %|@var{suffix}
5856 @itemx %m@var{suffix}
5857 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
5858 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
5859 all. These are the two most common ways to instruct a program that it
5860 should read from standard input or write to standard output. If you
5861 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
5862 construct: see for example @file{f/lang-specs.h}.
5864 @item %.@var{SUFFIX}
5865 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
5866 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
5867 terminated by the next space or %.
5870 Marks the argument containing or following the @samp{%w} as the
5871 designated output file of this compilation. This puts the argument
5872 into the sequence of arguments that @samp{%o} will substitute later.
5875 Substitutes the names of all the output files, with spaces
5876 automatically placed around them. You should write spaces
5877 around the @samp{%o} as well or the results are undefined.
5878 @samp{%o} is for use in the specs for running the linker.
5879 Input files whose names have no recognized suffix are not compiled
5880 at all, but they are included among the output files, so they will
5884 Substitutes the suffix for object files. Note that this is
5885 handled specially when it immediately follows @samp{%g, %u, or %U},
5886 because of the need for those to form complete file names. The
5887 handling is such that @samp{%O} is treated exactly as if it had already
5888 been substituted, except that @samp{%g, %u, and %U} do not currently
5889 support additional @var{suffix} characters following @samp{%O} as they would
5890 following, for example, @samp{.o}.
5893 Substitutes the standard macro predefinitions for the
5894 current target machine. Use this when running @code{cpp}.
5897 Like @samp{%p}, but puts @samp{__} before and after the name of each
5898 predefined macro, except for macros that start with @samp{__} or with
5899 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
5903 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
5904 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}), and
5905 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
5909 Current argument is the name of a library or startup file of some sort.
5910 Search for that file in a standard list of directories and substitute
5911 the full name found.
5914 Print @var{str} as an error message. @var{str} is terminated by a newline.
5915 Use this when inconsistent options are detected.
5918 Substitute the contents of spec string @var{name} at this point.
5921 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
5923 @item %x@{@var{option}@}
5924 Accumulate an option for @samp{%X}.
5927 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
5931 Output the accumulated assembler options specified by @option{-Wa}.
5934 Output the accumulated preprocessor options specified by @option{-Wp}.
5937 Process the @code{asm} spec. This is used to compute the
5938 switches to be passed to the assembler.
5941 Process the @code{asm_final} spec. This is a spec string for
5942 passing switches to an assembler post-processor, if such a program is
5946 Process the @code{link} spec. This is the spec for computing the
5947 command line passed to the linker. Typically it will make use of the
5948 @samp{%L %G %S %D and %E} sequences.
5951 Dump out a @option{-L} option for each directory that GCC believes might
5952 contain startup files. If the target supports multilibs then the
5953 current multilib directory will be prepended to each of these paths.
5956 Output the multilib directory with directory separators replaced with
5957 @samp{_}. If multilib directories are not set, or the multilib directory is
5958 @file{.} then this option emits nothing.
5961 Process the @code{lib} spec. This is a spec string for deciding which
5962 libraries should be included on the command line to the linker.
5965 Process the @code{libgcc} spec. This is a spec string for deciding
5966 which GCC support library should be included on the command line to the linker.
5969 Process the @code{startfile} spec. This is a spec for deciding which
5970 object files should be the first ones passed to the linker. Typically
5971 this might be a file named @file{crt0.o}.
5974 Process the @code{endfile} spec. This is a spec string that specifies
5975 the last object files that will be passed to the linker.
5978 Process the @code{cpp} spec. This is used to construct the arguments
5979 to be passed to the C preprocessor.
5982 Process the @code{cc1} spec. This is used to construct the options to be
5983 passed to the actual C compiler (@samp{cc1}).
5986 Process the @code{cc1plus} spec. This is used to construct the options to be
5987 passed to the actual C++ compiler (@samp{cc1plus}).
5990 Substitute the variable part of a matched option. See below.
5991 Note that each comma in the substituted string is replaced by
5995 Remove all occurrences of @code{-S} from the command line. Note---this
5996 command is position dependent. @samp{%} commands in the spec string
5997 before this one will see @code{-S}, @samp{%} commands in the spec string
5998 after this one will not.
6000 @item %:@var{function}(@var{args})
6001 Call the named function @var{function}, passing it @var{args}.
6002 @var{args} is first processed as a nested spec string, then split
6003 into an argument vector in the usual fashion. The function returns
6004 a string which is processed as if it had appeared literally as part
6005 of the current spec.
6007 The following built-in spec functions are provided:
6010 @item @code{if-exists}
6011 The @code{if-exists} spec function takes one argument, an absolute
6012 pathname to a file. If the file exists, @code{if-exists} returns the
6013 pathname. Here is a small example of its usage:
6017 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
6020 @item @code{if-exists-else}
6021 The @code{if-exists-else} spec function is similar to the @code{if-exists}
6022 spec function, except that it takes two arguments. The first argument is
6023 an absolute pathname to a file. If the file exists, @code{if-exists-else}
6024 returns the pathname. If it does not exist, it returns the second argument.
6025 This way, @code{if-exists-else} can be used to select one file or another,
6026 based on the existence of the first. Here is a small example of its usage:
6030 crt0%O%s %:if-exists(crti%O%s) \
6031 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
6036 Substitutes the @code{-S} switch, if that switch was given to GCC@.
6037 If that switch was not specified, this substitutes nothing. Note that
6038 the leading dash is omitted when specifying this option, and it is
6039 automatically inserted if the substitution is performed. Thus the spec
6040 string @samp{%@{foo@}} would match the command-line option @option{-foo}
6041 and would output the command line option @option{-foo}.
6043 @item %W@{@code{S}@}
6044 Like %@{@code{S}@} but mark last argument supplied within as a file to be
6047 @item %@{@code{S}*@}
6048 Substitutes all the switches specified to GCC whose names start
6049 with @code{-S}, but which also take an argument. This is used for
6050 switches like @option{-o}, @option{-D}, @option{-I}, etc.
6051 GCC considers @option{-o foo} as being
6052 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
6053 text, including the space. Thus two arguments would be generated.
6055 @item %@{@code{S}*&@code{T}*@}
6056 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
6057 (the order of @code{S} and @code{T} in the spec is not significant).
6058 There can be any number of ampersand-separated variables; for each the
6059 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
6061 @item %@{@code{S}:@code{X}@}
6062 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
6064 @item %@{!@code{S}:@code{X}@}
6065 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
6067 @item %@{@code{S}*:@code{X}@}
6068 Substitutes @code{X} if one or more switches whose names start with
6069 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
6070 once, no matter how many such switches appeared. However, if @code{%*}
6071 appears somewhere in @code{X}, then @code{X} will be substituted once
6072 for each matching switch, with the @code{%*} replaced by the part of
6073 that switch that matched the @code{*}.
6075 @item %@{.@code{S}:@code{X}@}
6076 Substitutes @code{X}, if processing a file with suffix @code{S}.
6078 @item %@{!.@code{S}:@code{X}@}
6079 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
6081 @item %@{@code{S}|@code{P}:@code{X}@}
6082 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC@.
6083 This may be combined with @samp{!}, @samp{.}, and @code{*} sequences as well,
6084 although they have a stronger binding than the @samp{|}. If @code{%*}
6085 appears in @code{X}, all of the alternatives must be starred, and only
6086 the first matching alternative is substituted.
6088 For example, a spec string like this:
6091 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
6094 will output the following command-line options from the following input
6095 command-line options:
6100 -d fred.c -foo -baz -boggle
6101 -d jim.d -bar -baz -boggle
6104 @item %@{S:X; T:Y; :D@}
6106 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
6107 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
6108 be as many clauses as you need. This may be combined with @code{.},
6109 @code{!}, @code{|}, and @code{*} as needed.
6114 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
6115 construct may contain other nested @samp{%} constructs or spaces, or
6116 even newlines. They are processed as usual, as described above.
6117 Trailing white space in @code{X} is ignored. White space may also
6118 appear anywhere on the left side of the colon in these constructs,
6119 except between @code{.} or @code{*} and the corresponding word.
6121 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
6122 handled specifically in these constructs. If another value of
6123 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
6124 @option{-W} switch is found later in the command line, the earlier
6125 switch value is ignored, except with @{@code{S}*@} where @code{S} is
6126 just one letter, which passes all matching options.
6128 The character @samp{|} at the beginning of the predicate text is used to
6129 indicate that a command should be piped to the following command, but
6130 only if @option{-pipe} is specified.
6132 It is built into GCC which switches take arguments and which do not.
6133 (You might think it would be useful to generalize this to allow each
6134 compiler's spec to say which switches take arguments. But this cannot
6135 be done in a consistent fashion. GCC cannot even decide which input
6136 files have been specified without knowing which switches take arguments,
6137 and it must know which input files to compile in order to tell which
6140 GCC also knows implicitly that arguments starting in @option{-l} are to be
6141 treated as compiler output files, and passed to the linker in their
6142 proper position among the other output files.
6144 @c man begin OPTIONS
6146 @node Target Options
6147 @section Specifying Target Machine and Compiler Version
6148 @cindex target options
6149 @cindex cross compiling
6150 @cindex specifying machine version
6151 @cindex specifying compiler version and target machine
6152 @cindex compiler version, specifying
6153 @cindex target machine, specifying
6155 The usual way to run GCC is to run the executable called @file{gcc}, or
6156 @file{<machine>-gcc} when cross-compiling, or
6157 @file{<machine>-gcc-<version>} to run a version other than the one that
6158 was installed last. Sometimes this is inconvenient, so GCC provides
6159 options that will switch to another cross-compiler or version.
6162 @item -b @var{machine}
6164 The argument @var{machine} specifies the target machine for compilation.
6166 The value to use for @var{machine} is the same as was specified as the
6167 machine type when configuring GCC as a cross-compiler. For
6168 example, if a cross-compiler was configured with @samp{configure
6169 i386v}, meaning to compile for an 80386 running System V, then you
6170 would specify @option{-b i386v} to run that cross compiler.
6172 @item -V @var{version}
6174 The argument @var{version} specifies which version of GCC to run.
6175 This is useful when multiple versions are installed. For example,
6176 @var{version} might be @samp{2.0}, meaning to run GCC version 2.0.
6179 The @option{-V} and @option{-b} options work by running the
6180 @file{<machine>-gcc-<version>} executable, so there's no real reason to
6181 use them if you can just run that directly.
6183 @node Submodel Options
6184 @section Hardware Models and Configurations
6185 @cindex submodel options
6186 @cindex specifying hardware config
6187 @cindex hardware models and configurations, specifying
6188 @cindex machine dependent options
6190 Earlier we discussed the standard option @option{-b} which chooses among
6191 different installed compilers for completely different target
6192 machines, such as VAX vs.@: 68000 vs.@: 80386.
6194 In addition, each of these target machine types can have its own
6195 special options, starting with @samp{-m}, to choose among various
6196 hardware models or configurations---for example, 68010 vs 68020,
6197 floating coprocessor or none. A single installed version of the
6198 compiler can compile for any model or configuration, according to the
6201 Some configurations of the compiler also support additional special
6202 options, usually for compatibility with other compilers on the same
6205 These options are defined by the macro @code{TARGET_SWITCHES} in the
6206 machine description. The default for the options is also defined by
6207 that macro, which enables you to change the defaults.
6217 * RS/6000 and PowerPC Options::
6220 * i386 and x86-64 Options::
6222 * DEC Alpha Options::
6223 * DEC Alpha/VMS Options::
6226 * System V Options::
6227 * TMS320C3x/C4x Options::
6234 * S/390 and zSeries Options::
6238 * Xstormy16 Options::
6243 @node M680x0 Options
6244 @subsection M680x0 Options
6245 @cindex M680x0 options
6247 These are the @samp{-m} options defined for the 68000 series. The default
6248 values for these options depends on which style of 68000 was selected when
6249 the compiler was configured; the defaults for the most common choices are
6257 Generate output for a 68000. This is the default
6258 when the compiler is configured for 68000-based systems.
6260 Use this option for microcontrollers with a 68000 or EC000 core,
6261 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
6267 Generate output for a 68020. This is the default
6268 when the compiler is configured for 68020-based systems.
6272 Generate output containing 68881 instructions for floating point.
6273 This is the default for most 68020 systems unless @option{--nfp} was
6274 specified when the compiler was configured.
6278 Generate output for a 68030. This is the default when the compiler is
6279 configured for 68030-based systems.
6283 Generate output for a 68040. This is the default when the compiler is
6284 configured for 68040-based systems.
6286 This option inhibits the use of 68881/68882 instructions that have to be
6287 emulated by software on the 68040. Use this option if your 68040 does not
6288 have code to emulate those instructions.
6292 Generate output for a 68060. This is the default when the compiler is
6293 configured for 68060-based systems.
6295 This option inhibits the use of 68020 and 68881/68882 instructions that
6296 have to be emulated by software on the 68060. Use this option if your 68060
6297 does not have code to emulate those instructions.
6301 Generate output for a CPU32. This is the default
6302 when the compiler is configured for CPU32-based systems.
6304 Use this option for microcontrollers with a
6305 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
6306 68336, 68340, 68341, 68349 and 68360.
6310 Generate output for a 520X ``coldfire'' family cpu. This is the default
6311 when the compiler is configured for 520X-based systems.
6313 Use this option for microcontroller with a 5200 core, including
6314 the MCF5202, MCF5203, MCF5204 and MCF5202.
6319 Generate output for a 68040, without using any of the new instructions.
6320 This results in code which can run relatively efficiently on either a
6321 68020/68881 or a 68030 or a 68040. The generated code does use the
6322 68881 instructions that are emulated on the 68040.
6326 Generate output for a 68060, without using any of the new instructions.
6327 This results in code which can run relatively efficiently on either a
6328 68020/68881 or a 68030 or a 68040. The generated code does use the
6329 68881 instructions that are emulated on the 68060.
6332 @opindex msoft-float
6333 Generate output containing library calls for floating point.
6334 @strong{Warning:} the requisite libraries are not available for all m68k
6335 targets. Normally the facilities of the machine's usual C compiler are
6336 used, but this can't be done directly in cross-compilation. You must
6337 make your own arrangements to provide suitable library functions for
6338 cross-compilation. The embedded targets @samp{m68k-*-aout} and
6339 @samp{m68k-*-coff} do provide software floating point support.
6343 Consider type @code{int} to be 16 bits wide, like @code{short int}.
6346 @opindex mnobitfield
6347 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
6348 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
6352 Do use the bit-field instructions. The @option{-m68020} option implies
6353 @option{-mbitfield}. This is the default if you use a configuration
6354 designed for a 68020.
6358 Use a different function-calling convention, in which functions
6359 that take a fixed number of arguments return with the @code{rtd}
6360 instruction, which pops their arguments while returning. This
6361 saves one instruction in the caller since there is no need to pop
6362 the arguments there.
6364 This calling convention is incompatible with the one normally
6365 used on Unix, so you cannot use it if you need to call libraries
6366 compiled with the Unix compiler.
6368 Also, you must provide function prototypes for all functions that
6369 take variable numbers of arguments (including @code{printf});
6370 otherwise incorrect code will be generated for calls to those
6373 In addition, seriously incorrect code will result if you call a
6374 function with too many arguments. (Normally, extra arguments are
6375 harmlessly ignored.)
6377 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
6378 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
6381 @itemx -mno-align-int
6383 @opindex mno-align-int
6384 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
6385 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
6386 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
6387 Aligning variables on 32-bit boundaries produces code that runs somewhat
6388 faster on processors with 32-bit busses at the expense of more memory.
6390 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
6391 align structures containing the above types differently than
6392 most published application binary interface specifications for the m68k.
6396 Use the pc-relative addressing mode of the 68000 directly, instead of
6397 using a global offset table. At present, this option implies @option{-fpic},
6398 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
6399 not presently supported with @option{-mpcrel}, though this could be supported for
6400 68020 and higher processors.
6402 @item -mno-strict-align
6403 @itemx -mstrict-align
6404 @opindex mno-strict-align
6405 @opindex mstrict-align
6406 Do not (do) assume that unaligned memory references will be handled by
6410 Generate code that allows the data segment to be located in a different
6411 area of memory from the text segment. This allows for execute in place in
6412 an environment without virtual memory management. This option implies -fPIC.
6415 Generate code that assumes that the data segment follows the text segment.
6416 This is the default.
6418 @item -mid-shared-library
6419 Generate code that supports shared libraries via the library ID method.
6420 This allows for execute in place and shared libraries in an environment
6421 without virtual memory management. This option implies -fPIC.
6423 @item -mno-id-shared-library
6424 Generate code that doesn't assume ID based shared libraries are being used.
6425 This is the default.
6427 @item -mshared-library-id=n
6428 Specified the identification number of the ID based shared library being
6429 compiled. Specifying a value of 0 will generate more compact code, specifying
6430 other values will force the allocation of that number to the current
6431 library but is no more space or time efficient than omitting this option.
6435 @node M68hc1x Options
6436 @subsection M68hc1x Options
6437 @cindex M68hc1x options
6439 These are the @samp{-m} options defined for the 68hc11 and 68hc12
6440 microcontrollers. The default values for these options depends on
6441 which style of microcontroller was selected when the compiler was configured;
6442 the defaults for the most common choices are given below.
6449 Generate output for a 68HC11. This is the default
6450 when the compiler is configured for 68HC11-based systems.
6456 Generate output for a 68HC12. This is the default
6457 when the compiler is configured for 68HC12-based systems.
6463 Generate output for a 68HCS12.
6466 @opindex mauto-incdec
6467 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
6474 Enable the use of 68HC12 min and max instructions.
6477 @itemx -mno-long-calls
6478 @opindex mlong-calls
6479 @opindex mno-long-calls
6480 Treat all calls as being far away (near). If calls are assumed to be
6481 far away, the compiler will use the @code{call} instruction to
6482 call a function and the @code{rtc} instruction for returning.
6486 Consider type @code{int} to be 16 bits wide, like @code{short int}.
6488 @item -msoft-reg-count=@var{count}
6489 @opindex msoft-reg-count
6490 Specify the number of pseudo-soft registers which are used for the
6491 code generation. The maximum number is 32. Using more pseudo-soft
6492 register may or may not result in better code depending on the program.
6493 The default is 4 for 68HC11 and 2 for 68HC12.
6498 @subsection VAX Options
6501 These @samp{-m} options are defined for the VAX:
6506 Do not output certain jump instructions (@code{aobleq} and so on)
6507 that the Unix assembler for the VAX cannot handle across long
6512 Do output those jump instructions, on the assumption that you
6513 will assemble with the GNU assembler.
6517 Output code for g-format floating point numbers instead of d-format.
6521 @subsection SPARC Options
6522 @cindex SPARC options
6524 These @samp{-m} options are supported on the SPARC:
6529 @opindex mno-app-regs
6531 Specify @option{-mapp-regs} to generate output using the global registers
6532 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
6535 To be fully SVR4 ABI compliant at the cost of some performance loss,
6536 specify @option{-mno-app-regs}. You should compile libraries and system
6537 software with this option.
6542 @opindex mhard-float
6543 Generate output containing floating point instructions. This is the
6549 @opindex msoft-float
6550 Generate output containing library calls for floating point.
6551 @strong{Warning:} the requisite libraries are not available for all SPARC
6552 targets. Normally the facilities of the machine's usual C compiler are
6553 used, but this cannot be done directly in cross-compilation. You must make
6554 your own arrangements to provide suitable library functions for
6555 cross-compilation. The embedded targets @samp{sparc-*-aout} and
6556 @samp{sparclite-*-*} do provide software floating point support.
6558 @option{-msoft-float} changes the calling convention in the output file;
6559 therefore, it is only useful if you compile @emph{all} of a program with
6560 this option. In particular, you need to compile @file{libgcc.a}, the
6561 library that comes with GCC, with @option{-msoft-float} in order for
6564 @item -mhard-quad-float
6565 @opindex mhard-quad-float
6566 Generate output containing quad-word (long double) floating point
6569 @item -msoft-quad-float
6570 @opindex msoft-quad-float
6571 Generate output containing library calls for quad-word (long double)
6572 floating point instructions. The functions called are those specified
6573 in the SPARC ABI@. This is the default.
6575 As of this writing, there are no SPARC implementations that have hardware
6576 support for the quad-word floating point instructions. They all invoke
6577 a trap handler for one of these instructions, and then the trap handler
6578 emulates the effect of the instruction. Because of the trap handler overhead,
6579 this is much slower than calling the ABI library routines. Thus the
6580 @option{-msoft-quad-float} option is the default.
6582 @item -mno-unaligned-doubles
6583 @itemx -munaligned-doubles
6584 @opindex mno-unaligned-doubles
6585 @opindex munaligned-doubles
6586 Assume that doubles have 8 byte alignment. This is the default.
6588 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
6589 alignment only if they are contained in another type, or if they have an
6590 absolute address. Otherwise, it assumes they have 4 byte alignment.
6591 Specifying this option avoids some rare compatibility problems with code
6592 generated by other compilers. It is not the default because it results
6593 in a performance loss, especially for floating point code.
6595 @item -mno-faster-structs
6596 @itemx -mfaster-structs
6597 @opindex mno-faster-structs
6598 @opindex mfaster-structs
6599 With @option{-mfaster-structs}, the compiler assumes that structures
6600 should have 8 byte alignment. This enables the use of pairs of
6601 @code{ldd} and @code{std} instructions for copies in structure
6602 assignment, in place of twice as many @code{ld} and @code{st} pairs.
6603 However, the use of this changed alignment directly violates the SPARC
6604 ABI@. Thus, it's intended only for use on targets where the developer
6605 acknowledges that their resulting code will not be directly in line with
6606 the rules of the ABI@.
6609 @opindex mimpure-text
6610 @option{-mimpure-text}, used in addition to @option{-shared}, tells
6611 the compiler to not pass @option{-z text} to the linker when linking a
6612 shared object. Using this option, you can link position-dependent
6613 code into a shared object.
6615 @option{-mimpure-text} suppresses the ``relocations remain against
6616 allocatable but non-writable sections'' linker error message.
6617 However, the necessary relocations will trigger copy-on-write, and the
6618 shared object is not actually shared across processes. Instead of
6619 using @option{-mimpure-text}, you should compile all source code with
6620 @option{-fpic} or @option{-fPIC}.
6622 This option is only available on SunOS and Solaris.
6624 @item -mcpu=@var{cpu_type}
6626 Set the instruction set, register set, and instruction scheduling parameters
6627 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
6628 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
6629 @samp{f930}, @samp{f934}, @samp{hypersparc}, @samp{sparclite86x},
6630 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc}, and
6633 Default instruction scheduling parameters are used for values that select
6634 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
6635 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
6637 Here is a list of each supported architecture and their supported
6642 v8: supersparc, hypersparc
6643 sparclite: f930, f934, sparclite86x
6645 v9: ultrasparc, ultrasparc3
6648 By default (unless configured otherwise), GCC generates code for the V7
6649 variant of the SPARC architecture. With @option{-mcpu=cypress}, the compiler
6650 additionally optimizes it for the Cypress CY7C602 chip, as used in the
6651 SPARCStation/SPARCServer 3xx series. This is also appropriate for the older
6652 SPARCStation 1, 2, IPX etc.
6654 With @option{-mcpu=v8}, GCC generates code for the V8 variant of the SPARC
6655 architecture. The only difference from V7 code is that the compiler emits
6656 the integer multiply and integer divide instructions which exist in SPARC-V8
6657 but not in SPARC-V7. With @option{-mcpu=supersparc}, the compiler additionally
6658 optimizes it for the SuperSPARC chip, as used in the SPARCStation 10, 1000 and
6661 With @option{-mcpu=sparclite}, GCC generates code for the SPARClite variant of
6662 the SPARC architecture. This adds the integer multiply, integer divide step
6663 and scan (@code{ffs}) instructions which exist in SPARClite but not in SPARC-V7.
6664 With @option{-mcpu=f930}, the compiler additionally optimizes it for the
6665 Fujitsu MB86930 chip, which is the original SPARClite, with no FPU. With
6666 @option{-mcpu=f934}, the compiler additionally optimizes it for the Fujitsu
6667 MB86934 chip, which is the more recent SPARClite with FPU.
6669 With @option{-mcpu=sparclet}, GCC generates code for the SPARClet variant of
6670 the SPARC architecture. This adds the integer multiply, multiply/accumulate,
6671 integer divide step and scan (@code{ffs}) instructions which exist in SPARClet
6672 but not in SPARC-V7. With @option{-mcpu=tsc701}, the compiler additionally
6673 optimizes it for the TEMIC SPARClet chip.
6675 With @option{-mcpu=v9}, GCC generates code for the V9 variant of the SPARC
6676 architecture. This adds 64-bit integer and floating-point move instructions,
6677 3 additional floating-point condition code registers and conditional move
6678 instructions. With @option{-mcpu=ultrasparc}, the compiler additionally
6679 optimizes it for the Sun UltraSPARC I/II chips. With
6680 @option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
6681 Sun UltraSPARC III chip.
6683 @item -mtune=@var{cpu_type}
6685 Set the instruction scheduling parameters for machine type
6686 @var{cpu_type}, but do not set the instruction set or register set that the
6687 option @option{-mcpu=@var{cpu_type}} would.
6689 The same values for @option{-mcpu=@var{cpu_type}} can be used for
6690 @option{-mtune=@var{cpu_type}}, but the only useful values are those
6691 that select a particular cpu implementation. Those are @samp{cypress},
6692 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
6693 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc}, and
6700 With @option{-mv8plus}, GCC generates code for the SPARC-V8+ ABI. The
6701 difference from the V8 ABI is that the global and out registers are
6702 considered 64-bit wide. This is enabled by default on Solaris in 32-bit
6703 mode for all SPARC-V9 processors.
6709 With @option{-mvis}, GCC generates code that takes advantage of the UltraSPARC
6710 Visual Instruction Set extensions. The default is @option{-mno-vis}.
6713 These @samp{-m} options are supported in addition to the above
6714 on SPARC-V9 processors in 64-bit environments:
6717 @item -mlittle-endian
6718 @opindex mlittle-endian
6719 Generate code for a processor running in little-endian mode. It is only
6720 available for a few configurations and most notably not on Solaris.
6726 Generate code for a 32-bit or 64-bit environment.
6727 The 32-bit environment sets int, long and pointer to 32 bits.
6728 The 64-bit environment sets int to 32 bits and long and pointer
6731 @item -mcmodel=medlow
6732 @opindex mcmodel=medlow
6733 Generate code for the Medium/Low code model: 64-bit addresses, programs
6734 must be linked in the low 32 bits of memory. Programs can be statically
6735 or dynamically linked.
6737 @item -mcmodel=medmid
6738 @opindex mcmodel=medmid
6739 Generate code for the Medium/Middle code model: 64-bit addresses, programs
6740 must be linked in the low 44 bits of memory, the text and data segments must
6741 be less than 2GB in size and the data segment must be located within 2GB of
6744 @item -mcmodel=medany
6745 @opindex mcmodel=medany
6746 Generate code for the Medium/Anywhere code model: 64-bit addresses, programs
6747 may be linked anywhere in memory, the text and data segments must be less
6748 than 2GB in size and the data segment must be located within 2GB of the
6751 @item -mcmodel=embmedany
6752 @opindex mcmodel=embmedany
6753 Generate code for the Medium/Anywhere code model for embedded systems:
6754 64-bit addresses, the text and data segments must be less than 2GB in
6755 size, both starting anywhere in memory (determined at link time). The
6756 global register %g4 points to the base of the data segment. Programs
6757 are statically linked and PIC is not supported.
6760 @itemx -mno-stack-bias
6761 @opindex mstack-bias
6762 @opindex mno-stack-bias
6763 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
6764 frame pointer if present, are offset by @minus{}2047 which must be added back
6765 when making stack frame references. This is the default in 64-bit mode.
6766 Otherwise, assume no such offset is present.
6770 @subsection ARM Options
6773 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
6777 @item -mabi=@var{name}
6779 Generate code for the specified ABI. Permissible values are: @samp{apcs-gnu},
6780 @samp{atpcs}, @samp{aapcs} and @samp{iwmmxt}.
6783 @opindex mapcs-frame
6784 Generate a stack frame that is compliant with the ARM Procedure Call
6785 Standard for all functions, even if this is not strictly necessary for
6786 correct execution of the code. Specifying @option{-fomit-frame-pointer}
6787 with this option will cause the stack frames not to be generated for
6788 leaf functions. The default is @option{-mno-apcs-frame}.
6792 This is a synonym for @option{-mapcs-frame}.
6796 Generate code for a processor running with a 26-bit program counter,
6797 and conforming to the function calling standards for the APCS 26-bit
6798 option. This option replaces the @option{-m2} and @option{-m3} options
6799 of previous releases of the compiler.
6803 Generate code for a processor running with a 32-bit program counter,
6804 and conforming to the function calling standards for the APCS 32-bit
6805 option. This option replaces the @option{-m6} option of previous releases
6809 @c not currently implemented
6810 @item -mapcs-stack-check
6811 @opindex mapcs-stack-check
6812 Generate code to check the amount of stack space available upon entry to
6813 every function (that actually uses some stack space). If there is
6814 insufficient space available then either the function
6815 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
6816 called, depending upon the amount of stack space required. The run time
6817 system is required to provide these functions. The default is
6818 @option{-mno-apcs-stack-check}, since this produces smaller code.
6820 @c not currently implemented
6822 @opindex mapcs-float
6823 Pass floating point arguments using the float point registers. This is
6824 one of the variants of the APCS@. This option is recommended if the
6825 target hardware has a floating point unit or if a lot of floating point
6826 arithmetic is going to be performed by the code. The default is
6827 @option{-mno-apcs-float}, since integer only code is slightly increased in
6828 size if @option{-mapcs-float} is used.
6830 @c not currently implemented
6831 @item -mapcs-reentrant
6832 @opindex mapcs-reentrant
6833 Generate reentrant, position independent code. The default is
6834 @option{-mno-apcs-reentrant}.
6837 @item -mthumb-interwork
6838 @opindex mthumb-interwork
6839 Generate code which supports calling between the ARM and Thumb
6840 instruction sets. Without this option the two instruction sets cannot
6841 be reliably used inside one program. The default is
6842 @option{-mno-thumb-interwork}, since slightly larger code is generated
6843 when @option{-mthumb-interwork} is specified.
6845 @item -mno-sched-prolog
6846 @opindex mno-sched-prolog
6847 Prevent the reordering of instructions in the function prolog, or the
6848 merging of those instruction with the instructions in the function's
6849 body. This means that all functions will start with a recognizable set
6850 of instructions (or in fact one of a choice from a small set of
6851 different function prologues), and this information can be used to
6852 locate the start if functions inside an executable piece of code. The
6853 default is @option{-msched-prolog}.
6856 @opindex mhard-float
6857 Generate output containing floating point instructions. This is the
6861 @opindex msoft-float
6862 Generate output containing library calls for floating point.
6863 @strong{Warning:} the requisite libraries are not available for all ARM
6864 targets. Normally the facilities of the machine's usual C compiler are
6865 used, but this cannot be done directly in cross-compilation. You must make
6866 your own arrangements to provide suitable library functions for
6869 @option{-msoft-float} changes the calling convention in the output file;
6870 therefore, it is only useful if you compile @emph{all} of a program with
6871 this option. In particular, you need to compile @file{libgcc.a}, the
6872 library that comes with GCC, with @option{-msoft-float} in order for
6875 @item -mfloat-abi=@var{name}
6877 Specifies which ABI to use for floating point values. Permissible values
6878 are: @samp{soft}, @samp{softfp} and @samp{hard}.
6880 @samp{soft} and @samp{hard} are equivalent to @option{-msoft-float}
6881 and @option{-mhard-float} respectively. @samp{softfp} allows the generation
6882 of floating point instructions, but still uses the soft-float calling
6885 @item -mlittle-endian
6886 @opindex mlittle-endian
6887 Generate code for a processor running in little-endian mode. This is
6888 the default for all standard configurations.
6891 @opindex mbig-endian
6892 Generate code for a processor running in big-endian mode; the default is
6893 to compile code for a little-endian processor.
6895 @item -mwords-little-endian
6896 @opindex mwords-little-endian
6897 This option only applies when generating code for big-endian processors.
6898 Generate code for a little-endian word order but a big-endian byte
6899 order. That is, a byte order of the form @samp{32107654}. Note: this
6900 option should only be used if you require compatibility with code for
6901 big-endian ARM processors generated by versions of the compiler prior to
6904 @item -malignment-traps
6905 @opindex malignment-traps
6906 Generate code that will not trap if the MMU has alignment traps enabled.
6907 On ARM architectures prior to ARMv4, there were no instructions to
6908 access half-word objects stored in memory. However, when reading from
6909 memory a feature of the ARM architecture allows a word load to be used,
6910 even if the address is unaligned, and the processor core will rotate the
6911 data as it is being loaded. This option tells the compiler that such
6912 misaligned accesses will cause a MMU trap and that it should instead
6913 synthesize the access as a series of byte accesses. The compiler can
6914 still use word accesses to load half-word data if it knows that the
6915 address is aligned to a word boundary.
6917 This option is ignored when compiling for ARM architecture 4 or later,
6918 since these processors have instructions to directly access half-word
6921 @item -mno-alignment-traps
6922 @opindex mno-alignment-traps
6923 Generate code that assumes that the MMU will not trap unaligned
6924 accesses. This produces better code when the target instruction set
6925 does not have half-word memory operations (i.e.@: implementations prior to
6928 Note that you cannot use this option to access unaligned word objects,
6929 since the processor will only fetch one 32-bit aligned object from
6932 The default setting for most targets is @option{-mno-alignment-traps}, since
6933 this produces better code when there are no half-word memory
6934 instructions available.
6936 @item -mshort-load-bytes
6937 @itemx -mno-short-load-words
6938 @opindex mshort-load-bytes
6939 @opindex mno-short-load-words
6940 These are deprecated aliases for @option{-malignment-traps}.
6942 @item -mno-short-load-bytes
6943 @itemx -mshort-load-words
6944 @opindex mno-short-load-bytes
6945 @opindex mshort-load-words
6946 This are deprecated aliases for @option{-mno-alignment-traps}.
6948 @item -mcpu=@var{name}
6950 This specifies the name of the target ARM processor. GCC uses this name
6951 to determine what kind of instructions it can emit when generating
6952 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
6953 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
6954 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
6955 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
6956 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
6957 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm8},
6958 @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
6959 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
6960 @samp{arm920t}, @samp{arm926ejs}, @samp{arm940t}, @samp{arm9tdmi},
6961 @samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ejs},
6962 @samp{arm1136js}, @samp{arm1136jfs} ,@samp{xscale}, @samp{iwmmxt},
6965 @itemx -mtune=@var{name}
6967 This option is very similar to the @option{-mcpu=} option, except that
6968 instead of specifying the actual target processor type, and hence
6969 restricting which instructions can be used, it specifies that GCC should
6970 tune the performance of the code as if the target were of the type
6971 specified in this option, but still choosing the instructions that it
6972 will generate based on the cpu specified by a @option{-mcpu=} option.
6973 For some ARM implementations better performance can be obtained by using
6976 @item -march=@var{name}
6978 This specifies the name of the target ARM architecture. GCC uses this
6979 name to determine what kind of instructions it can emit when generating
6980 assembly code. This option can be used in conjunction with or instead
6981 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
6982 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
6983 @samp{armv5}, @samp{armv5t}, @samp{armv5te}, @samp{armv6}, @samp{armv6j},
6984 @samp{iwmmxt}, @samp{ep9312}.
6986 @item -mfpu=@var{name}
6987 @itemx -mfpe=@var{number}
6988 @itemx -mfp=@var{number}
6992 This specifies what floating point hardware (or hardware emulation) is
6993 available on the target. Permissible names are: @samp{fpa}, @samp{fpe2},
6994 @samp{fpe3}, @samp{maverick}, @samp{vfp}. @option{-mfp} and @option{-mfpe}
6995 are synonyms for @option{-mfpu}=@samp{fpe}@var{number}, for compatibility
6996 with older versions of GCC@.
6998 If @option{-msoft-float} is specified this specifies the format of
6999 floating point values.
7001 @item -mstructure-size-boundary=@var{n}
7002 @opindex mstructure-size-boundary
7003 The size of all structures and unions will be rounded up to a multiple
7004 of the number of bits set by this option. Permissible values are 8, 32
7005 and 64. The default value varies for different toolchains. For the COFF
7006 targeted toolchain the default value is 8. A value of 64 is only allowed
7007 if the underlying ABI supports it.
7009 Specifying the larger number can produce faster, more efficient code, but
7010 can also increase the size of the program. Different values are potentially
7011 incompatible. Code compiled with one value cannot necessarily expect to
7012 work with code or libraries compiled with another value, if they exchange
7013 information using structures or unions.
7015 @item -mabort-on-noreturn
7016 @opindex mabort-on-noreturn
7017 Generate a call to the function @code{abort} at the end of a
7018 @code{noreturn} function. It will be executed if the function tries to
7022 @itemx -mno-long-calls
7023 @opindex mlong-calls
7024 @opindex mno-long-calls
7025 Tells the compiler to perform function calls by first loading the
7026 address of the function into a register and then performing a subroutine
7027 call on this register. This switch is needed if the target function
7028 will lie outside of the 64 megabyte addressing range of the offset based
7029 version of subroutine call instruction.
7031 Even if this switch is enabled, not all function calls will be turned
7032 into long calls. The heuristic is that static functions, functions
7033 which have the @samp{short-call} attribute, functions that are inside
7034 the scope of a @samp{#pragma no_long_calls} directive and functions whose
7035 definitions have already been compiled within the current compilation
7036 unit, will not be turned into long calls. The exception to this rule is
7037 that weak function definitions, functions with the @samp{long-call}
7038 attribute or the @samp{section} attribute, and functions that are within
7039 the scope of a @samp{#pragma long_calls} directive, will always be
7040 turned into long calls.
7042 This feature is not enabled by default. Specifying
7043 @option{-mno-long-calls} will restore the default behavior, as will
7044 placing the function calls within the scope of a @samp{#pragma
7045 long_calls_off} directive. Note these switches have no effect on how
7046 the compiler generates code to handle function calls via function
7049 @item -mnop-fun-dllimport
7050 @opindex mnop-fun-dllimport
7051 Disable support for the @code{dllimport} attribute.
7053 @item -msingle-pic-base
7054 @opindex msingle-pic-base
7055 Treat the register used for PIC addressing as read-only, rather than
7056 loading it in the prologue for each function. The run-time system is
7057 responsible for initializing this register with an appropriate value
7058 before execution begins.
7060 @item -mpic-register=@var{reg}
7061 @opindex mpic-register
7062 Specify the register to be used for PIC addressing. The default is R10
7063 unless stack-checking is enabled, when R9 is used.
7065 @item -mcirrus-fix-invalid-insns
7066 @opindex mcirrus-fix-invalid-insns
7067 @opindex mno-cirrus-fix-invalid-insns
7068 Insert NOPs into the instruction stream to in order to work around
7069 problems with invalid Maverick instruction combinations. This option
7070 is only valid if the @option{-mcpu=ep9312} option has been used to
7071 enable generation of instructions for the Cirrus Maverick floating
7072 point co-processor. This option is not enabled by default, since the
7073 problem is only present in older Maverick implementations. The default
7074 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
7077 @item -mpoke-function-name
7078 @opindex mpoke-function-name
7079 Write the name of each function into the text section, directly
7080 preceding the function prologue. The generated code is similar to this:
7084 .ascii "arm_poke_function_name", 0
7087 .word 0xff000000 + (t1 - t0)
7088 arm_poke_function_name
7090 stmfd sp!, @{fp, ip, lr, pc@}
7094 When performing a stack backtrace, code can inspect the value of
7095 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
7096 location @code{pc - 12} and the top 8 bits are set, then we know that
7097 there is a function name embedded immediately preceding this location
7098 and has length @code{((pc[-3]) & 0xff000000)}.
7102 Generate code for the 16-bit Thumb instruction set. The default is to
7103 use the 32-bit ARM instruction set.
7106 @opindex mtpcs-frame
7107 Generate a stack frame that is compliant with the Thumb Procedure Call
7108 Standard for all non-leaf functions. (A leaf function is one that does
7109 not call any other functions.) The default is @option{-mno-tpcs-frame}.
7111 @item -mtpcs-leaf-frame
7112 @opindex mtpcs-leaf-frame
7113 Generate a stack frame that is compliant with the Thumb Procedure Call
7114 Standard for all leaf functions. (A leaf function is one that does
7115 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
7117 @item -mcallee-super-interworking
7118 @opindex mcallee-super-interworking
7119 Gives all externally visible functions in the file being compiled an ARM
7120 instruction set header which switches to Thumb mode before executing the
7121 rest of the function. This allows these functions to be called from
7122 non-interworking code.
7124 @item -mcaller-super-interworking
7125 @opindex mcaller-super-interworking
7126 Allows calls via function pointers (including virtual functions) to
7127 execute correctly regardless of whether the target code has been
7128 compiled for interworking or not. There is a small overhead in the cost
7129 of executing a function pointer if this option is enabled.
7133 @node MN10300 Options
7134 @subsection MN10300 Options
7135 @cindex MN10300 options
7137 These @option{-m} options are defined for Matsushita MN10300 architectures:
7142 Generate code to avoid bugs in the multiply instructions for the MN10300
7143 processors. This is the default.
7146 @opindex mno-mult-bug
7147 Do not generate code to avoid bugs in the multiply instructions for the
7152 Generate code which uses features specific to the AM33 processor.
7156 Do not generate code which uses features specific to the AM33 processor. This
7161 Do not link in the C run-time initialization object file.
7165 Indicate to the linker that it should perform a relaxation optimization pass
7166 to shorten branches, calls and absolute memory addresses. This option only
7167 has an effect when used on the command line for the final link step.
7169 This option makes symbolic debugging impossible.
7173 @node M32R/D Options
7174 @subsection M32R/D Options
7175 @cindex M32R/D options
7177 These @option{-m} options are defined for Renesas M32R/D architectures:
7182 Generate code for the M32R/2@.
7186 Generate code for the M32R/X@.
7190 Generate code for the M32R@. This is the default.
7193 @opindex mmodel=small
7194 Assume all objects live in the lower 16MB of memory (so that their addresses
7195 can be loaded with the @code{ld24} instruction), and assume all subroutines
7196 are reachable with the @code{bl} instruction.
7197 This is the default.
7199 The addressability of a particular object can be set with the
7200 @code{model} attribute.
7202 @item -mmodel=medium
7203 @opindex mmodel=medium
7204 Assume objects may be anywhere in the 32-bit address space (the compiler
7205 will generate @code{seth/add3} instructions to load their addresses), and
7206 assume all subroutines are reachable with the @code{bl} instruction.
7209 @opindex mmodel=large
7210 Assume objects may be anywhere in the 32-bit address space (the compiler
7211 will generate @code{seth/add3} instructions to load their addresses), and
7212 assume subroutines may not be reachable with the @code{bl} instruction
7213 (the compiler will generate the much slower @code{seth/add3/jl}
7214 instruction sequence).
7217 @opindex msdata=none
7218 Disable use of the small data area. Variables will be put into
7219 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
7220 @code{section} attribute has been specified).
7221 This is the default.
7223 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
7224 Objects may be explicitly put in the small data area with the
7225 @code{section} attribute using one of these sections.
7228 @opindex msdata=sdata
7229 Put small global and static data in the small data area, but do not
7230 generate special code to reference them.
7234 Put small global and static data in the small data area, and generate
7235 special instructions to reference them.
7239 @cindex smaller data references
7240 Put global and static objects less than or equal to @var{num} bytes
7241 into the small data or bss sections instead of the normal data or bss
7242 sections. The default value of @var{num} is 8.
7243 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
7244 for this option to have any effect.
7246 All modules should be compiled with the same @option{-G @var{num}} value.
7247 Compiling with different values of @var{num} may or may not work; if it
7248 doesn't the linker will give an error message---incorrect code will not be
7253 Makes the M32R specific code in the compiler display some statistics
7254 that might help in debugging programs.
7257 @opindex malign-loops
7258 Align all loops to a 32-byte boundary.
7260 @item -mno-align-loops
7261 @opindex mno-align-loops
7262 Do not enforce a 32-byte alignment for loops. This is the default.
7264 @item -missue-rate=@var{number}
7265 @opindex missue-rate=@var{number}
7266 Issue @var{number} instructions per cycle. @var{number} can only be 1
7269 @item -mbranch-cost=@var{number}
7270 @opindex mbranch-cost=@var{number}
7271 @var{number} can only be 1 or 2. If it is 1 then branches will be
7272 preferred over conditional code, if it is 2, then the opposite will
7275 @item -mflush-trap=@var{number}
7276 @opindex mflush-trap=@var{number}
7277 Specifies the trap number to use to flush the cache. The default is
7278 12. Valid numbers are between 0 and 15 inclusive.
7280 @item -mno-flush-trap
7281 @opindex mno-flush-trap
7282 Specifies that the cache cannot be flushed by using a trap.
7284 @item -mflush-func=@var{name}
7285 @opindex mflush-func=@var{name}
7286 Specifies the name of the operating system function to call to flush
7287 the cache. The default is @emph{_flush_cache}, but a function call
7288 will only be used if a trap is not available.
7290 @item -mno-flush-func
7291 @opindex mno-flush-func
7292 Indicates that there is no OS function for flushing the cache.
7296 @node RS/6000 and PowerPC Options
7297 @subsection IBM RS/6000 and PowerPC Options
7298 @cindex RS/6000 and PowerPC Options
7299 @cindex IBM RS/6000 and PowerPC Options
7301 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
7309 @itemx -mpowerpc-gpopt
7310 @itemx -mno-powerpc-gpopt
7311 @itemx -mpowerpc-gfxopt
7312 @itemx -mno-powerpc-gfxopt
7314 @itemx -mno-powerpc64
7320 @opindex mno-powerpc
7321 @opindex mpowerpc-gpopt
7322 @opindex mno-powerpc-gpopt
7323 @opindex mpowerpc-gfxopt
7324 @opindex mno-powerpc-gfxopt
7326 @opindex mno-powerpc64
7327 GCC supports two related instruction set architectures for the
7328 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
7329 instructions supported by the @samp{rios} chip set used in the original
7330 RS/6000 systems and the @dfn{PowerPC} instruction set is the
7331 architecture of the Motorola MPC5xx, MPC6xx, MPC8xx microprocessors, and
7332 the IBM 4xx microprocessors.
7334 Neither architecture is a subset of the other. However there is a
7335 large common subset of instructions supported by both. An MQ
7336 register is included in processors supporting the POWER architecture.
7338 You use these options to specify which instructions are available on the
7339 processor you are using. The default value of these options is
7340 determined when configuring GCC@. Specifying the
7341 @option{-mcpu=@var{cpu_type}} overrides the specification of these
7342 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
7343 rather than the options listed above.
7345 The @option{-mpower} option allows GCC to generate instructions that
7346 are found only in the POWER architecture and to use the MQ register.
7347 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
7348 to generate instructions that are present in the POWER2 architecture but
7349 not the original POWER architecture.
7351 The @option{-mpowerpc} option allows GCC to generate instructions that
7352 are found only in the 32-bit subset of the PowerPC architecture.
7353 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
7354 GCC to use the optional PowerPC architecture instructions in the
7355 General Purpose group, including floating-point square root. Specifying
7356 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
7357 use the optional PowerPC architecture instructions in the Graphics
7358 group, including floating-point select.
7360 The @option{-mpowerpc64} option allows GCC to generate the additional
7361 64-bit instructions that are found in the full PowerPC64 architecture
7362 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
7363 @option{-mno-powerpc64}.
7365 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
7366 will use only the instructions in the common subset of both
7367 architectures plus some special AIX common-mode calls, and will not use
7368 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
7369 permits GCC to use any instruction from either architecture and to
7370 allow use of the MQ register; specify this for the Motorola MPC601.
7372 @item -mnew-mnemonics
7373 @itemx -mold-mnemonics
7374 @opindex mnew-mnemonics
7375 @opindex mold-mnemonics
7376 Select which mnemonics to use in the generated assembler code. With
7377 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
7378 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
7379 assembler mnemonics defined for the POWER architecture. Instructions
7380 defined in only one architecture have only one mnemonic; GCC uses that
7381 mnemonic irrespective of which of these options is specified.
7383 GCC defaults to the mnemonics appropriate for the architecture in
7384 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
7385 value of these option. Unless you are building a cross-compiler, you
7386 should normally not specify either @option{-mnew-mnemonics} or
7387 @option{-mold-mnemonics}, but should instead accept the default.
7389 @item -mcpu=@var{cpu_type}
7391 Set architecture type, register usage, choice of mnemonics, and
7392 instruction scheduling parameters for machine type @var{cpu_type}.
7393 Supported values for @var{cpu_type} are @samp{401}, @samp{403},
7394 @samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{505},
7395 @samp{601}, @samp{602}, @samp{603}, @samp{603e}, @samp{604},
7396 @samp{604e}, @samp{620}, @samp{630}, @samp{740}, @samp{7400},
7397 @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
7398 @samp{860}, @samp{970}, @samp{common}, @samp{ec603e}, @samp{G3},
7399 @samp{G4}, @samp{G5}, @samp{power}, @samp{power2}, @samp{power3},
7400 @samp{power4}, @samp{power5}, @samp{powerpc}, @samp{powerpc64},
7401 @samp{rios}, @samp{rios1}, @samp{rios2}, @samp{rsc}, and @samp{rs64a}.
7403 @option{-mcpu=common} selects a completely generic processor. Code
7404 generated under this option will run on any POWER or PowerPC processor.
7405 GCC will use only the instructions in the common subset of both
7406 architectures, and will not use the MQ register. GCC assumes a generic
7407 processor model for scheduling purposes.
7409 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
7410 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
7411 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
7412 types, with an appropriate, generic processor model assumed for
7413 scheduling purposes.
7415 The other options specify a specific processor. Code generated under
7416 those options will run best on that processor, and may not run at all on
7419 The @option{-mcpu} options automatically enable or disable the
7420 following options: @option{-maltivec}, @option{-mhard-float},
7421 @option{-mmfcrf}, @option{-mmultiple}, @option{-mnew-mnemonics},
7422 @option{-mpower}, @option{-mpower2}, @option{-mpowerpc64},
7423 @option{-mpowerpc-gpopt}, @option{-mpowerpc-gfxopt},
7424 @option{-mstring}. The particular options set for any particular CPU
7425 will vary between compiler versions, depending on what setting seems
7426 to produce optimal code for that CPU; it doesn't necessarily reflect
7427 the actual hardware's capabilities. If you wish to set an individual
7428 option to a particular value, you may specify it after the
7429 @option{-mcpu} option, like @samp{-mcpu=970 -mno-altivec}.
7431 On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
7432 not enabled or disabled by the @option{-mcpu} option at present, since
7433 AIX does not have full support for these options. You may still
7434 enable or disable them individually if you're sure it'll work in your
7437 @item -mtune=@var{cpu_type}
7439 Set the instruction scheduling parameters for machine type
7440 @var{cpu_type}, but do not set the architecture type, register usage, or
7441 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
7442 values for @var{cpu_type} are used for @option{-mtune} as for
7443 @option{-mcpu}. If both are specified, the code generated will use the
7444 architecture, registers, and mnemonics set by @option{-mcpu}, but the
7445 scheduling parameters set by @option{-mtune}.
7450 @opindex mno-altivec
7451 These switches enable or disable the use of built-in functions that
7452 allow access to the AltiVec instruction set. You may also need to set
7453 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
7458 Extend the current ABI with SPE ABI extensions. This does not change
7459 the default ABI, instead it adds the SPE ABI extensions to the current
7463 @opindex mabi=no-spe
7464 Disable Booke SPE ABI extensions for the current ABI.
7466 @item -misel=@var{yes/no}
7469 This switch enables or disables the generation of ISEL instructions.
7471 @item -mspe=@var{yes/no}
7474 This switch enables or disables the generation of SPE simd
7477 @item -mfloat-gprs=@var{yes/no}
7479 @opindex mfloat-gprs
7480 This switch enables or disables the generation of floating point
7481 operations on the general purpose registers for architectures that
7482 support it. This option is currently only available on the MPC8540.
7485 @itemx -mno-fp-in-toc
7486 @itemx -mno-sum-in-toc
7487 @itemx -mminimal-toc
7489 @opindex mno-fp-in-toc
7490 @opindex mno-sum-in-toc
7491 @opindex mminimal-toc
7492 Modify generation of the TOC (Table Of Contents), which is created for
7493 every executable file. The @option{-mfull-toc} option is selected by
7494 default. In that case, GCC will allocate at least one TOC entry for
7495 each unique non-automatic variable reference in your program. GCC
7496 will also place floating-point constants in the TOC@. However, only
7497 16,384 entries are available in the TOC@.
7499 If you receive a linker error message that saying you have overflowed
7500 the available TOC space, you can reduce the amount of TOC space used
7501 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
7502 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
7503 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
7504 generate code to calculate the sum of an address and a constant at
7505 run-time instead of putting that sum into the TOC@. You may specify one
7506 or both of these options. Each causes GCC to produce very slightly
7507 slower and larger code at the expense of conserving TOC space.
7509 If you still run out of space in the TOC even when you specify both of
7510 these options, specify @option{-mminimal-toc} instead. This option causes
7511 GCC to make only one TOC entry for every file. When you specify this
7512 option, GCC will produce code that is slower and larger but which
7513 uses extremely little TOC space. You may wish to use this option
7514 only on files that contain less frequently executed code.
7520 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
7521 @code{long} type, and the infrastructure needed to support them.
7522 Specifying @option{-maix64} implies @option{-mpowerpc64} and
7523 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
7524 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
7529 @opindex mno-xl-call
7530 On AIX, pass floating-point arguments to prototyped functions beyond the
7531 register save area (RSA) on the stack in addition to argument FPRs. The
7532 AIX calling convention was extended but not initially documented to
7533 handle an obscure K&R C case of calling a function that takes the
7534 address of its arguments with fewer arguments than declared. AIX XL
7535 compilers access floating point arguments which do not fit in the
7536 RSA from the stack when a subroutine is compiled without
7537 optimization. Because always storing floating-point arguments on the
7538 stack is inefficient and rarely needed, this option is not enabled by
7539 default and only is necessary when calling subroutines compiled by AIX
7540 XL compilers without optimization.
7544 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
7545 application written to use message passing with special startup code to
7546 enable the application to run. The system must have PE installed in the
7547 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
7548 must be overridden with the @option{-specs=} option to specify the
7549 appropriate directory location. The Parallel Environment does not
7550 support threads, so the @option{-mpe} option and the @option{-pthread}
7551 option are incompatible.
7553 @item -malign-natural
7554 @itemx -malign-power
7555 @opindex malign-natural
7556 @opindex malign-power
7557 On AIX, Darwin, and 64-bit PowerPC GNU/Linux, the option
7558 @option{-malign-natural} overrides the ABI-defined alignment of larger
7559 types, such as floating-point doubles, on their natural size-based boundary.
7560 The option @option{-malign-power} instructs GCC to follow the ABI-specified
7561 alignment rules. GCC defaults to the standard alignment defined in the ABI.
7565 @opindex msoft-float
7566 @opindex mhard-float
7567 Generate code that does not use (uses) the floating-point register set.
7568 Software floating point emulation is provided if you use the
7569 @option{-msoft-float} option, and pass the option to GCC when linking.
7572 @itemx -mno-multiple
7574 @opindex mno-multiple
7575 Generate code that uses (does not use) the load multiple word
7576 instructions and the store multiple word instructions. These
7577 instructions are generated by default on POWER systems, and not
7578 generated on PowerPC systems. Do not use @option{-mmultiple} on little
7579 endian PowerPC systems, since those instructions do not work when the
7580 processor is in little endian mode. The exceptions are PPC740 and
7581 PPC750 which permit the instructions usage in little endian mode.
7587 Generate code that uses (does not use) the load string instructions
7588 and the store string word instructions to save multiple registers and
7589 do small block moves. These instructions are generated by default on
7590 POWER systems, and not generated on PowerPC systems. Do not use
7591 @option{-mstring} on little endian PowerPC systems, since those
7592 instructions do not work when the processor is in little endian mode.
7593 The exceptions are PPC740 and PPC750 which permit the instructions
7594 usage in little endian mode.
7600 Generate code that uses (does not use) the load or store instructions
7601 that update the base register to the address of the calculated memory
7602 location. These instructions are generated by default. If you use
7603 @option{-mno-update}, there is a small window between the time that the
7604 stack pointer is updated and the address of the previous frame is
7605 stored, which means code that walks the stack frame across interrupts or
7606 signals may get corrupted data.
7609 @itemx -mno-fused-madd
7610 @opindex mfused-madd
7611 @opindex mno-fused-madd
7612 Generate code that uses (does not use) the floating point multiply and
7613 accumulate instructions. These instructions are generated by default if
7614 hardware floating is used.
7616 @item -mno-bit-align
7618 @opindex mno-bit-align
7620 On System V.4 and embedded PowerPC systems do not (do) force structures
7621 and unions that contain bit-fields to be aligned to the base type of the
7624 For example, by default a structure containing nothing but 8
7625 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
7626 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
7627 the structure would be aligned to a 1 byte boundary and be one byte in
7630 @item -mno-strict-align
7631 @itemx -mstrict-align
7632 @opindex mno-strict-align
7633 @opindex mstrict-align
7634 On System V.4 and embedded PowerPC systems do not (do) assume that
7635 unaligned memory references will be handled by the system.
7638 @itemx -mno-relocatable
7639 @opindex mrelocatable
7640 @opindex mno-relocatable
7641 On embedded PowerPC systems generate code that allows (does not allow)
7642 the program to be relocated to a different address at runtime. If you
7643 use @option{-mrelocatable} on any module, all objects linked together must
7644 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
7646 @item -mrelocatable-lib
7647 @itemx -mno-relocatable-lib
7648 @opindex mrelocatable-lib
7649 @opindex mno-relocatable-lib
7650 On embedded PowerPC systems generate code that allows (does not allow)
7651 the program to be relocated to a different address at runtime. Modules
7652 compiled with @option{-mrelocatable-lib} can be linked with either modules
7653 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
7654 with modules compiled with the @option{-mrelocatable} options.
7660 On System V.4 and embedded PowerPC systems do not (do) assume that
7661 register 2 contains a pointer to a global area pointing to the addresses
7662 used in the program.
7665 @itemx -mlittle-endian
7667 @opindex mlittle-endian
7668 On System V.4 and embedded PowerPC systems compile code for the
7669 processor in little endian mode. The @option{-mlittle-endian} option is
7670 the same as @option{-mlittle}.
7675 @opindex mbig-endian
7676 On System V.4 and embedded PowerPC systems compile code for the
7677 processor in big endian mode. The @option{-mbig-endian} option is
7678 the same as @option{-mbig}.
7680 @item -mdynamic-no-pic
7681 @opindex mdynamic-no-pic
7682 On Darwin and Mac OS X systems, compile code so that it is not
7683 relocatable, but that its external references are relocatable. The
7684 resulting code is suitable for applications, but not shared
7687 @item -mprioritize-restricted-insns=@var{priority}
7688 @opindex mprioritize-restricted-insns
7689 This option controls the priority that is assigned to
7690 dispatch-slot restricted instructions during the second scheduling
7691 pass. The argument @var{priority} takes the value @var{0/1/2} to assign
7692 @var{no/highest/second-highest} priority to dispatch slot restricted
7695 @item -msched-costly-dep=@var{dependence_type}
7696 @opindex msched-costly-dep
7697 This option controls which dependences are considered costly
7698 by the target during instruction scheduling. The argument
7699 @var{dependence_type} takes one of the following values:
7700 @var{no}: no dependence is costly,
7701 @var{all}: all dependences are costly,
7702 @var{true_store_to_load}: a true dependence from store to load is costly,
7703 @var{store_to_load}: any dependence from store to load is costly,
7704 @var{number}: any dependence which latency >= @var{number} is costly.
7706 @item -minsert-sched-nops=@var{scheme}
7707 @opindex minsert-sched-nops
7708 This option controls which nop insertion scheme will be used during
7709 the second scheduling pass. The argument @var{scheme} takes one of the
7711 @var{no}: Don't insert nops.
7712 @var{pad}: Pad with nops any dispatch group which has vacant issue slots,
7713 according to the scheduler's grouping.
7714 @var{regroup_exact}: Insert nops to force costly dependent insns into
7715 separate groups. Insert exactly as many nops as needed to force an insn
7716 to a new group, according to the estimated processor grouping.
7717 @var{number}: Insert nops to force costly dependent insns into
7718 separate groups. Insert @var{number} nops to force an insn to a new group.
7722 On System V.4 and embedded PowerPC systems compile code using calling
7723 conventions that adheres to the March 1995 draft of the System V
7724 Application Binary Interface, PowerPC processor supplement. This is the
7725 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
7727 @item -mcall-sysv-eabi
7728 @opindex mcall-sysv-eabi
7729 Specify both @option{-mcall-sysv} and @option{-meabi} options.
7731 @item -mcall-sysv-noeabi
7732 @opindex mcall-sysv-noeabi
7733 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
7735 @item -mcall-solaris
7736 @opindex mcall-solaris
7737 On System V.4 and embedded PowerPC systems compile code for the Solaris
7741 @opindex mcall-linux
7742 On System V.4 and embedded PowerPC systems compile code for the
7743 Linux-based GNU system.
7747 On System V.4 and embedded PowerPC systems compile code for the
7748 Hurd-based GNU system.
7751 @opindex mcall-netbsd
7752 On System V.4 and embedded PowerPC systems compile code for the
7753 NetBSD operating system.
7755 @item -maix-struct-return
7756 @opindex maix-struct-return
7757 Return all structures in memory (as specified by the AIX ABI)@.
7759 @item -msvr4-struct-return
7760 @opindex msvr4-struct-return
7761 Return structures smaller than 8 bytes in registers (as specified by the
7765 @opindex mabi=altivec
7766 Extend the current ABI with AltiVec ABI extensions. This does not
7767 change the default ABI, instead it adds the AltiVec ABI extensions to
7770 @item -mabi=no-altivec
7771 @opindex mabi=no-altivec
7772 Disable AltiVec ABI extensions for the current ABI.
7775 @itemx -mno-prototype
7777 @opindex mno-prototype
7778 On System V.4 and embedded PowerPC systems assume that all calls to
7779 variable argument functions are properly prototyped. Otherwise, the
7780 compiler must insert an instruction before every non prototyped call to
7781 set or clear bit 6 of the condition code register (@var{CR}) to
7782 indicate whether floating point values were passed in the floating point
7783 registers in case the function takes a variable arguments. With
7784 @option{-mprototype}, only calls to prototyped variable argument functions
7785 will set or clear the bit.
7789 On embedded PowerPC systems, assume that the startup module is called
7790 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
7791 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
7796 On embedded PowerPC systems, assume that the startup module is called
7797 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
7802 On embedded PowerPC systems, assume that the startup module is called
7803 @file{crt0.o} and the standard C libraries are @file{libads.a} and
7807 @opindex myellowknife
7808 On embedded PowerPC systems, assume that the startup module is called
7809 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
7814 On System V.4 and embedded PowerPC systems, specify that you are
7815 compiling for a VxWorks system.
7819 Specify that you are compiling for the WindISS simulation environment.
7823 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
7824 header to indicate that @samp{eabi} extended relocations are used.
7830 On System V.4 and embedded PowerPC systems do (do not) adhere to the
7831 Embedded Applications Binary Interface (eabi) which is a set of
7832 modifications to the System V.4 specifications. Selecting @option{-meabi}
7833 means that the stack is aligned to an 8 byte boundary, a function
7834 @code{__eabi} is called to from @code{main} to set up the eabi
7835 environment, and the @option{-msdata} option can use both @code{r2} and
7836 @code{r13} to point to two separate small data areas. Selecting
7837 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
7838 do not call an initialization function from @code{main}, and the
7839 @option{-msdata} option will only use @code{r13} to point to a single
7840 small data area. The @option{-meabi} option is on by default if you
7841 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
7844 @opindex msdata=eabi
7845 On System V.4 and embedded PowerPC systems, put small initialized
7846 @code{const} global and static data in the @samp{.sdata2} section, which
7847 is pointed to by register @code{r2}. Put small initialized
7848 non-@code{const} global and static data in the @samp{.sdata} section,
7849 which is pointed to by register @code{r13}. Put small uninitialized
7850 global and static data in the @samp{.sbss} section, which is adjacent to
7851 the @samp{.sdata} section. The @option{-msdata=eabi} option is
7852 incompatible with the @option{-mrelocatable} option. The
7853 @option{-msdata=eabi} option also sets the @option{-memb} option.
7856 @opindex msdata=sysv
7857 On System V.4 and embedded PowerPC systems, put small global and static
7858 data in the @samp{.sdata} section, which is pointed to by register
7859 @code{r13}. Put small uninitialized global and static data in the
7860 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
7861 The @option{-msdata=sysv} option is incompatible with the
7862 @option{-mrelocatable} option.
7864 @item -msdata=default
7866 @opindex msdata=default
7868 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
7869 compile code the same as @option{-msdata=eabi}, otherwise compile code the
7870 same as @option{-msdata=sysv}.
7873 @opindex msdata-data
7874 On System V.4 and embedded PowerPC systems, put small global and static
7875 data in the @samp{.sdata} section. Put small uninitialized global and
7876 static data in the @samp{.sbss} section. Do not use register @code{r13}
7877 to address small data however. This is the default behavior unless
7878 other @option{-msdata} options are used.
7882 @opindex msdata=none
7884 On embedded PowerPC systems, put all initialized global and static data
7885 in the @samp{.data} section, and all uninitialized data in the
7886 @samp{.bss} section.
7890 @cindex smaller data references (PowerPC)
7891 @cindex .sdata/.sdata2 references (PowerPC)
7892 On embedded PowerPC systems, put global and static items less than or
7893 equal to @var{num} bytes into the small data or bss sections instead of
7894 the normal data or bss section. By default, @var{num} is 8. The
7895 @option{-G @var{num}} switch is also passed to the linker.
7896 All modules should be compiled with the same @option{-G @var{num}} value.
7899 @itemx -mno-regnames
7901 @opindex mno-regnames
7902 On System V.4 and embedded PowerPC systems do (do not) emit register
7903 names in the assembly language output using symbolic forms.
7906 @itemx -mno-longcall
7908 @opindex mno-longcall
7909 Default to making all function calls indirectly, using a register, so
7910 that functions which reside further than 32 megabytes (33,554,432
7911 bytes) from the current location can be called. This setting can be
7912 overridden by the @code{shortcall} function attribute, or by
7913 @code{#pragma longcall(0)}.
7915 Some linkers are capable of detecting out-of-range calls and generating
7916 glue code on the fly. On these systems, long calls are unnecessary and
7917 generate slower code. As of this writing, the AIX linker can do this,
7918 as can the GNU linker for PowerPC/64. It is planned to add this feature
7919 to the GNU linker for 32-bit PowerPC systems as well.
7921 On Darwin/PPC systems, @code{#pragma longcall} will generate ``jbsr
7922 callee, L42'', plus a ``branch island'' (glue code). The two target
7923 addresses represent the callee and the ``branch island.'' The
7924 Darwin/PPC linker will prefer the first address and generate a ``bl
7925 callee'' if the PPC ``bl'' instruction will reach the callee directly;
7926 otherwise, the linker will generate ``bl L42'' to call the ``branch
7927 island.'' The ``branch island'' is appended to the body of the
7928 calling function; it computes the full 32-bit address of the callee
7931 On Mach-O (Darwin) systems, this option directs the compiler emit to
7932 the glue for every direct call, and the Darwin linker decides whether
7933 to use or discard it.
7935 In the future, we may cause GCC to ignore all longcall specifications
7936 when the linker is known to generate glue.
7940 Adds support for multithreading with the @dfn{pthreads} library.
7941 This option sets flags for both the preprocessor and linker.
7945 @node Darwin Options
7946 @subsection Darwin Options
7947 @cindex Darwin options
7949 These options are defined for all architectures running the Darwin operating
7950 system. They are useful for compatibility with other Mac OS compilers.
7955 Add the framework directory @var{dir} to the head of the list of
7956 directories to be searched for header files. These directories are
7957 interleaved with those specified by @option{-I} options and are
7958 scanned in a left-to-right order.
7960 A framework directory is a directory with frameworks in it. A
7961 framework is a directory with a @samp{"Headers"} and/or
7962 @samp{"PrivateHeaders"} directory contained directly in it that ends
7963 in @samp{".framework"}. The name of a framework is the name of this
7964 directory excluding the @samp{".framework"}. Headers associated with
7965 the framework are found in one of those two directories, with
7966 @samp{"Headers"} being searched first. A subframework is a framework
7967 directory that is in a framework's @samp{"Frameworks"} directory.
7968 Includes of subframework headers can only appear in a header of a
7969 framework that contains the subframework, or in a sibling subframework
7970 header. Two subframeworks are siblings if they occur in the same
7971 framework. A subframework should not have the same name as a
7972 framework, a warning will be issued if this is violated. Currently a
7973 subframework cannot have subframeworks, in the future, the mechanism
7974 may be extended to support this. The standard frameworks can be found
7975 in @samp{"/System/Library/Frameworks"}, @samp{"/Library/Frameworks"}
7976 and @samp{"/Local/Library/Frameworks"}. An example include looks like
7977 @code{#include <Framework/header.h>}, where @samp{Framework} denotes
7978 the name of the framework and header.h is found in the
7979 @samp{"PrivateHeaders"} or @samp{"Headers"} directory.
7983 Loads all members of static archive libraries.
7984 See man ld(1) for more information.
7986 @item -arch_errors_fatal
7987 @opindex arch_errors_fatal
7988 Cause the errors having to do with files that have the wrong architecture
7992 @opindex bind_at_load
7993 Causes the output file to be marked such that the dynamic linker will
7994 bind all undefined references when the file is loaded or launched.
7998 Produce a Mach-o bundle format file.
7999 See man ld(1) for more information.
8001 @item -bundle_loader @var{executable}
8002 @opindex bundle_loader
8003 This specifies the @var{executable} that will be loading the build
8004 output file being linked. See man ld(1) for more information.
8006 @item -allowable_client @var{client_name}
8010 @itemx -compatibility_version
8011 @itemx -current_version
8012 @itemx -dependency-file
8014 @itemx -dylinker_install_name
8017 @itemx -exported_symbols_list
8019 @itemx -flat_namespace
8020 @itemx -force_cpusubtype_ALL
8021 @itemx -force_flat_namespace
8022 @itemx -headerpad_max_install_names
8025 @itemx -install_name
8026 @itemx -keep_private_externs
8027 @itemx -multi_module
8028 @itemx -multiply_defined
8029 @itemx -multiply_defined_unused
8031 @itemx -nofixprebinding
8034 @itemx -noseglinkedit
8035 @itemx -pagezero_size
8037 @itemx -prebind_all_twolevel_modules
8038 @itemx -private_bundle
8039 @itemx -read_only_relocs
8041 @itemx -sectobjectsymbols
8045 @itemx -sectobjectsymbols
8047 @itemx -seg_addr_table
8048 @itemx -seg_addr_table_filename
8051 @itemx -segs_read_only_addr
8052 @itemx -segs_read_write_addr
8053 @itemx -single_module
8056 @itemx -sub_umbrella
8057 @itemx -twolevel_namespace
8060 @itemx -unexported_symbols_list
8061 @itemx -weak_reference_mismatches
8064 @opindex allowable_client
8066 @opindex client_name
8067 @opindex compatibility_version
8068 @opindex current_version
8069 @opindex dependency-file
8071 @opindex dylinker_install_name
8074 @opindex exported_symbols_list
8076 @opindex flat_namespace
8077 @opindex force_cpusubtype_ALL
8078 @opindex force_flat_namespace
8079 @opindex headerpad_max_install_names
8082 @opindex install_name
8083 @opindex keep_private_externs
8084 @opindex multi_module
8085 @opindex multiply_defined
8086 @opindex multiply_defined_unused
8088 @opindex nofixprebinding
8089 @opindex nomultidefs
8091 @opindex noseglinkedit
8092 @opindex pagezero_size
8094 @opindex prebind_all_twolevel_modules
8095 @opindex private_bundle
8096 @opindex read_only_relocs
8098 @opindex sectobjectsymbols
8102 @opindex sectobjectsymbols
8104 @opindex seg_addr_table
8105 @opindex seg_addr_table_filename
8106 @opindex seglinkedit
8108 @opindex segs_read_only_addr
8109 @opindex segs_read_write_addr
8110 @opindex single_module
8112 @opindex sub_library
8113 @opindex sub_umbrella
8114 @opindex twolevel_namespace
8117 @opindex unexported_symbols_list
8118 @opindex weak_reference_mismatches
8119 @opindex whatsloaded
8121 These options are available for Darwin linker. Darwin linker man page
8122 describes them in detail.
8127 @subsection MIPS Options
8128 @cindex MIPS options
8134 Generate big-endian code.
8138 Generate little-endian code. This is the default for @samp{mips*el-*-*}
8141 @item -march=@var{arch}
8143 Generate code that will run on @var{arch}, which can be the name of a
8144 generic MIPS ISA, or the name of a particular processor.
8146 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
8147 @samp{mips32}, @samp{mips32r2}, and @samp{mips64}.
8148 The processor names are:
8149 @samp{4kc}, @samp{4kp}, @samp{5kc}, @samp{20kc},
8151 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
8152 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000}, @samp{rm7000},
8156 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4130}, @samp{vr4300},
8157 @samp{vr5000}, @samp{vr5400} and @samp{vr5500}.
8158 The special value @samp{from-abi} selects the
8159 most compatible architecture for the selected ABI (that is,
8160 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
8162 In processor names, a final @samp{000} can be abbreviated as @samp{k}
8163 (for example, @samp{-march=r2k}). Prefixes are optional, and
8164 @samp{vr} may be written @samp{r}.
8166 GCC defines two macros based on the value of this option. The first
8167 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
8168 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
8169 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
8170 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
8171 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
8173 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
8174 above. In other words, it will have the full prefix and will not
8175 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
8176 the macro names the resolved architecture (either @samp{"mips1"} or
8177 @samp{"mips3"}). It names the default architecture when no
8178 @option{-march} option is given.
8180 @item -mtune=@var{arch}
8182 Optimize for @var{arch}. Among other things, this option controls
8183 the way instructions are scheduled, and the perceived cost of arithmetic
8184 operations. The list of @var{arch} values is the same as for
8187 When this option is not used, GCC will optimize for the processor
8188 specified by @option{-march}. By using @option{-march} and
8189 @option{-mtune} together, it is possible to generate code that will
8190 run on a family of processors, but optimize the code for one
8191 particular member of that family.
8193 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
8194 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
8195 @samp{-march} ones described above.
8199 Equivalent to @samp{-march=mips1}.
8203 Equivalent to @samp{-march=mips2}.
8207 Equivalent to @samp{-march=mips3}.
8211 Equivalent to @samp{-march=mips4}.
8215 Equivalent to @samp{-march=mips32}.
8219 Equivalent to @samp{-march=mips32r2}.
8223 Equivalent to @samp{-march=mips64}.
8229 Use (do not use) the MIPS16 ISA.
8241 Generate code for the given ABI@.
8243 Note that the EABI has a 32-bit and a 64-bit variant. GCC normally
8244 generates 64-bit code when you select a 64-bit architecture, but you
8245 can use @option{-mgp32} to get 32-bit code instead.
8247 For information about the O64 ABI, see
8248 @w{@uref{http://gcc.gnu.org/projects/mipso64-abi.html}}.
8251 @itemx -mno-abicalls
8253 @opindex mno-abicalls
8254 Generate (do not generate) SVR4-style position-independent code.
8255 @option{-mabicalls} is the default for SVR4-based systems.
8261 Lift (do not lift) the usual restrictions on the size of the global
8264 GCC normally uses a single instruction to load values from the GOT.
8265 While this is relatively efficient, it will only work if the GOT
8266 is smaller than about 64k. Anything larger will cause the linker
8267 to report an error such as:
8269 @cindex relocation truncated to fit (MIPS)
8271 relocation truncated to fit: R_MIPS_GOT16 foobar
8274 If this happens, you should recompile your code with @option{-mxgot}.
8275 It should then work with very large GOTs, although it will also be
8276 less efficient, since it will take three instructions to fetch the
8277 value of a global symbol.
8279 Note that some linkers can create multiple GOTs. If you have such a
8280 linker, you should only need to use @option{-mxgot} when a single object
8281 file accesses more than 64k's worth of GOT entries. Very few do.
8283 These options have no effect unless GCC is generating position
8288 Assume that general-purpose registers are 32 bits wide.
8292 Assume that general-purpose registers are 64 bits wide.
8296 Assume that floating-point registers are 32 bits wide.
8300 Assume that floating-point registers are 64 bits wide.
8303 @opindex mhard-float
8304 Use floating-point coprocessor instructions.
8307 @opindex msoft-float
8308 Do not use floating-point coprocessor instructions. Implement
8309 floating-point calculations using library calls instead.
8311 @item -msingle-float
8312 @opindex msingle-float
8313 Assume that the floating-point coprocessor only supports single-precision
8316 @itemx -mdouble-float
8317 @opindex mdouble-float
8318 Assume that the floating-point coprocessor supports double-precision
8319 operations. This is the default.
8323 Force @code{int} and @code{long} types to be 64 bits wide. See
8324 @option{-mlong32} for an explanation of the default and the way
8325 that the pointer size is determined.
8329 Force @code{long} types to be 64 bits wide. See @option{-mlong32} for
8330 an explanation of the default and the way that the pointer size is
8335 Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
8337 The default size of @code{int}s, @code{long}s and pointers depends on
8338 the ABI@. All the supported ABIs use 32-bit @code{int}s. The n64 ABI
8339 uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
8340 32-bit @code{long}s. Pointers are the same size as @code{long}s,
8341 or the same size as integer registers, whichever is smaller.
8345 @cindex smaller data references (MIPS)
8346 @cindex gp-relative references (MIPS)
8347 Put global and static items less than or equal to @var{num} bytes into
8348 the small data or bss section instead of the normal data or bss section.
8349 This allows the data to be accessed using a single instruction.
8351 All modules should be compiled with the same @option{-G @var{num}}
8354 @item -membedded-data
8355 @itemx -mno-embedded-data
8356 @opindex membedded-data
8357 @opindex mno-embedded-data
8358 Allocate variables to the read-only data section first if possible, then
8359 next in the small data section if possible, otherwise in data. This gives
8360 slightly slower code than the default, but reduces the amount of RAM required
8361 when executing, and thus may be preferred for some embedded systems.
8363 @item -muninit-const-in-rodata
8364 @itemx -mno-uninit-const-in-rodata
8365 @opindex muninit-const-in-rodata
8366 @opindex mno-uninit-const-in-rodata
8367 Put uninitialized @code{const} variables in the read-only data section.
8368 This option is only meaningful in conjunction with @option{-membedded-data}.
8370 @item -msplit-addresses
8371 @itemx -mno-split-addresses
8372 @opindex msplit-addresses
8373 @opindex mno-split-addresses
8374 Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
8375 relocation operators. This option has been superceded by
8376 @option{-mexplicit-relocs} but is retained for backwards compatibility.
8378 @item -mexplicit-relocs
8379 @itemx -mno-explicit-relocs
8380 @opindex mexplicit-relocs
8381 @opindex mno-explicit-relocs
8382 Use (do not use) assembler relocation operators when dealing with symbolic
8383 addresses. The alternative, selected by @option{-mno-explicit-relocs},
8384 is to use assembler macros instead.
8386 @option{-mexplicit-relocs} is usually the default if GCC was configured
8387 to use an assembler that supports relocation operators. However, the
8388 combination of @option{-mabicalls} and @option{-fno-unit-at-a-time}
8389 implies @option{-mno-explicit-relocs} unless explicitly overridden.
8390 This is because, when generating abicalls, the choice of relocation
8391 depends on whether a symbol is local or global. In some rare cases,
8392 GCC will not be able to decide this until the whole compilation unit
8399 Generate (do not generate) code that refers to registers using their
8400 software names. The default is @option{-mno-rnames}, which tells GCC
8401 to use hardware names like @samp{$4} instead of software names like
8402 @samp{a0}. The only assembler known to support @option{-rnames} is
8403 the Algorithmics assembler.
8405 @item -mcheck-zero-division
8406 @itemx -mno-check-zero-division
8407 @opindex mcheck-zero-division
8408 @opindex mno-check-zero-division
8409 Trap (do not trap) on integer division by zero. The default is
8410 @option{-mcheck-zero-division}.
8416 Force (do not force) the use of @code{memcpy()} for non-trivial block
8417 moves. The default is @option{-mno-memcpy}, which allows GCC to inline
8418 most constant-sized copies.
8421 @itemx -mno-long-calls
8422 @opindex mlong-calls
8423 @opindex mno-long-calls
8424 Disable (do not disable) use of the @code{jal} instruction. Calling
8425 functions using @code{jal} is more efficient but requires the caller
8426 and callee to be in the same 256 megabyte segment.
8428 This option has no effect on abicalls code. The default is
8429 @option{-mno-long-calls}.
8435 Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
8436 instructions, as provided by the R4650 ISA.
8439 @itemx -mno-fused-madd
8440 @opindex mfused-madd
8441 @opindex mno-fused-madd
8442 Enable (disable) use of the floating point multiply-accumulate
8443 instructions, when they are available. The default is
8444 @option{-mfused-madd}.
8446 When multiply-accumulate instructions are used, the intermediate
8447 product is calculated to infinite precision and is not subject to
8448 the FCSR Flush to Zero bit. This may be undesirable in some
8453 Tell the MIPS assembler to not run its preprocessor over user
8454 assembler files (with a @samp{.s} suffix) when assembling them.
8457 @itemx -mno-fix-r4000
8459 @opindex mno-fix-r4000
8460 Work around certain R4000 CPU errata:
8463 A double-word or a variable shift may give an incorrect result if executed
8464 immediately after starting an integer division.
8466 A double-word or a variable shift may give an incorrect result if executed
8467 while an integer multiplication is in progress.
8469 An integer division may give an incorrect result if started in a delay slot
8470 of a taken branch or a jump.
8474 @itemx -mno-fix-r4400
8476 @opindex mno-fix-r4400
8477 Work around certain R4400 CPU errata:
8480 A double-word or a variable shift may give an incorrect result if executed
8481 immediately after starting an integer division.
8485 @itemx -mno-fix-vr4120
8486 @opindex mfix-vr4120
8487 Work around certain VR4120 errata:
8490 @code{dmultu} does not always produce the correct result.
8492 @code{div} and @code{ddiv} do not always produce the correct result if one
8493 of the operands is negative.
8495 The workarounds for the division errata rely on special functions in
8496 @file{libgcc.a}. At present, these functions are only provided by
8497 the @code{mips64vr*-elf} configurations.
8499 Other VR4120 errata require a nop to be inserted between certain pairs of
8500 instructions. These errata are handled by the assembler, not by GCC itself.
8505 Work around certain SB-1 CPU core errata.
8506 (This flag currently works around the SB-1 revision 2
8507 ``F1'' and ``F2'' floating point errata.)
8509 @item -mflush-func=@var{func}
8510 @itemx -mno-flush-func
8511 @opindex mflush-func
8512 Specifies the function to call to flush the I and D caches, or to not
8513 call any such function. If called, the function must take the same
8514 arguments as the common @code{_flush_func()}, that is, the address of the
8515 memory range for which the cache is being flushed, the size of the
8516 memory range, and the number 3 (to flush both caches). The default
8517 depends on the target GCC was configured for, but commonly is either
8518 @samp{_flush_func} or @samp{__cpu_flush}.
8520 @item -mbranch-likely
8521 @itemx -mno-branch-likely
8522 @opindex mbranch-likely
8523 @opindex mno-branch-likely
8524 Enable or disable use of Branch Likely instructions, regardless of the
8525 default for the selected architecture. By default, Branch Likely
8526 instructions may be generated if they are supported by the selected
8527 architecture. An exception is for the MIPS32 and MIPS64 architectures
8528 and processors which implement those architectures; for those, Branch
8529 Likely instructions will not be generated by default because the MIPS32
8530 and MIPS64 architectures specifically deprecate their use.
8532 @item -mfp-exceptions
8533 @itemx -mno-fp-exceptions
8534 @opindex mfp-exceptions
8535 Specifies whether FP exceptions are enabled. This affects how we schedule
8536 FP instructions for some processors. The default is that FP exceptions are
8539 For instance, on the SB-1, if FP exceptions are disabled, and we are emitting
8540 64-bit code, then we can use both FP pipes. Otherwise, we can only use one
8543 @item -mvr4130-align
8544 @itemx -mno-vr4130-align
8545 @opindex mvr4130-align
8546 The VR4130 pipeline is two-way superscalar, but can only issue two
8547 instructions together if the first one is 8-byte aligned. When this
8548 option is enabled, GCC will align pairs of instructions that it
8549 thinks should execute in parallel.
8551 This option only has an effect when optimizing for the VR4130.
8552 It normally makes code faster, but at the expense of making it bigger.
8553 It is enabled by default at optimization level @option{-O3}.
8556 @node i386 and x86-64 Options
8557 @subsection Intel 386 and AMD x86-64 Options
8558 @cindex i386 Options
8559 @cindex x86-64 Options
8560 @cindex Intel 386 Options
8561 @cindex AMD x86-64 Options
8563 These @samp{-m} options are defined for the i386 and x86-64 family of
8567 @item -mtune=@var{cpu-type}
8569 Tune to @var{cpu-type} everything applicable about the generated code, except
8570 for the ABI and the set of available instructions. The choices for
8574 Original Intel's i386 CPU.
8576 Intel's i486 CPU. (No scheduling is implemented for this chip.)
8578 Intel Pentium CPU with no MMX support.
8580 Intel PentiumMMX CPU based on Pentium core with MMX instruction set support.
8581 @item i686, pentiumpro
8582 Intel PentiumPro CPU.
8584 Intel Pentium2 CPU based on PentiumPro core with MMX instruction set support.
8585 @item pentium3, pentium3m
8586 Intel Pentium3 CPU based on PentiumPro core with MMX and SSE instruction set
8589 Low power version of Intel Pentium3 CPU with MMX, SSE and SSE2 instruction set
8590 support. Used by Centrino notebooks.
8591 @item pentium4, pentium4m
8592 Intel Pentium4 CPU with MMX, SSE and SSE2 instruction set support.
8594 Improved version of Intel Pentium4 CPU with MMX, SSE, SSE2 and SSE3 instruction
8597 Improved version of Intel Pentium4 CPU with 64-bit extensions, MMX, SSE,
8598 SSE2 and SSE3 instruction set support.
8600 AMD K6 CPU with MMX instruction set support.
8602 Improved versions of AMD K6 CPU with MMX and 3dNOW! instruction set support.
8603 @item athlon, athlon-tbird
8604 AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and SSE prefetch instructions
8606 @item athlon-4, athlon-xp, athlon-mp
8607 Improved AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and full SSE
8608 instruction set support.
8609 @item k8, opteron, athlon64, athlon-fx
8610 AMD K8 core based CPUs with x86-64 instruction set support. (This supersets
8611 MMX, SSE, SSE2, 3dNOW!, enhanced 3dNOW! and 64-bit instruction set extensions.)
8613 IDT Winchip C6 CPU, dealt in same way as i486 with additional MMX instruction
8616 IDT Winchip2 CPU, dealt in same way as i486 with additional MMX and 3dNOW!
8617 instruction set support.
8619 Via C3 CPU with MMX and 3dNOW! instruction set support. (No scheduling is
8620 implemented for this chip.)
8622 Via C3-2 CPU with MMX and SSE instruction set support. (No scheduling is
8623 implemented for this chip.)
8626 While picking a specific @var{cpu-type} will schedule things appropriately
8627 for that particular chip, the compiler will not generate any code that
8628 does not run on the i386 without the @option{-march=@var{cpu-type}} option
8631 @item -march=@var{cpu-type}
8633 Generate instructions for the machine type @var{cpu-type}. The choices
8634 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
8635 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
8637 @item -mcpu=@var{cpu-type}
8639 A deprecated synonym for @option{-mtune}.
8648 @opindex mpentiumpro
8649 These options are synonyms for @option{-mtune=i386}, @option{-mtune=i486},
8650 @option{-mtune=pentium}, and @option{-mtune=pentiumpro} respectively.
8651 These synonyms are deprecated.
8653 @item -mfpmath=@var{unit}
8655 Generate floating point arithmetics for selected unit @var{unit}. The choices
8660 Use the standard 387 floating point coprocessor present majority of chips and
8661 emulated otherwise. Code compiled with this option will run almost everywhere.
8662 The temporary results are computed in 80bit precision instead of precision
8663 specified by the type resulting in slightly different results compared to most
8664 of other chips. See @option{-ffloat-store} for more detailed description.
8666 This is the default choice for i386 compiler.
8669 Use scalar floating point instructions present in the SSE instruction set.
8670 This instruction set is supported by Pentium3 and newer chips, in the AMD line
8671 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
8672 instruction set supports only single precision arithmetics, thus the double and
8673 extended precision arithmetics is still done using 387. Later version, present
8674 only in Pentium4 and the future AMD x86-64 chips supports double precision
8677 For i387 you need to use @option{-march=@var{cpu-type}}, @option{-msse} or
8678 @option{-msse2} switches to enable SSE extensions and make this option
8679 effective. For x86-64 compiler, these extensions are enabled by default.
8681 The resulting code should be considerably faster in the majority of cases and avoid
8682 the numerical instability problems of 387 code, but may break some existing
8683 code that expects temporaries to be 80bit.
8685 This is the default choice for the x86-64 compiler.
8688 Attempt to utilize both instruction sets at once. This effectively double the
8689 amount of available registers and on chips with separate execution units for
8690 387 and SSE the execution resources too. Use this option with care, as it is
8691 still experimental, because the GCC register allocator does not model separate
8692 functional units well resulting in instable performance.
8695 @item -masm=@var{dialect}
8696 @opindex masm=@var{dialect}
8697 Output asm instructions using selected @var{dialect}. Supported choices are
8698 @samp{intel} or @samp{att} (the default one).
8703 @opindex mno-ieee-fp
8704 Control whether or not the compiler uses IEEE floating point
8705 comparisons. These handle correctly the case where the result of a
8706 comparison is unordered.
8709 @opindex msoft-float
8710 Generate output containing library calls for floating point.
8711 @strong{Warning:} the requisite libraries are not part of GCC@.
8712 Normally the facilities of the machine's usual C compiler are used, but
8713 this can't be done directly in cross-compilation. You must make your
8714 own arrangements to provide suitable library functions for
8717 On machines where a function returns floating point results in the 80387
8718 register stack, some floating point opcodes may be emitted even if
8719 @option{-msoft-float} is used.
8721 @item -mno-fp-ret-in-387
8722 @opindex mno-fp-ret-in-387
8723 Do not use the FPU registers for return values of functions.
8725 The usual calling convention has functions return values of types
8726 @code{float} and @code{double} in an FPU register, even if there
8727 is no FPU@. The idea is that the operating system should emulate
8730 The option @option{-mno-fp-ret-in-387} causes such values to be returned
8731 in ordinary CPU registers instead.
8733 @item -mno-fancy-math-387
8734 @opindex mno-fancy-math-387
8735 Some 387 emulators do not support the @code{sin}, @code{cos} and
8736 @code{sqrt} instructions for the 387. Specify this option to avoid
8737 generating those instructions. This option is the default on FreeBSD,
8738 OpenBSD and NetBSD@. This option is overridden when @option{-march}
8739 indicates that the target cpu will always have an FPU and so the
8740 instruction will not need emulation. As of revision 2.6.1, these
8741 instructions are not generated unless you also use the
8742 @option{-funsafe-math-optimizations} switch.
8744 @item -malign-double
8745 @itemx -mno-align-double
8746 @opindex malign-double
8747 @opindex mno-align-double
8748 Control whether GCC aligns @code{double}, @code{long double}, and
8749 @code{long long} variables on a two word boundary or a one word
8750 boundary. Aligning @code{double} variables on a two word boundary will
8751 produce code that runs somewhat faster on a @samp{Pentium} at the
8752 expense of more memory.
8754 @strong{Warning:} if you use the @option{-malign-double} switch,
8755 structures containing the above types will be aligned differently than
8756 the published application binary interface specifications for the 386
8757 and will not be binary compatible with structures in code compiled
8758 without that switch.
8760 @item -m96bit-long-double
8761 @itemx -m128bit-long-double
8762 @opindex m96bit-long-double
8763 @opindex m128bit-long-double
8764 These switches control the size of @code{long double} type. The i386
8765 application binary interface specifies the size to be 96 bits,
8766 so @option{-m96bit-long-double} is the default in 32 bit mode.
8768 Modern architectures (Pentium and newer) would prefer @code{long double}
8769 to be aligned to an 8 or 16 byte boundary. In arrays or structures
8770 conforming to the ABI, this would not be possible. So specifying a
8771 @option{-m128bit-long-double} will align @code{long double}
8772 to a 16 byte boundary by padding the @code{long double} with an additional
8775 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
8776 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
8778 Notice that neither of these options enable any extra precision over the x87
8779 standard of 80 bits for a @code{long double}.
8781 @strong{Warning:} if you override the default value for your target ABI, the
8782 structures and arrays containing @code{long double} variables will change
8783 their size as well as function calling convention for function taking
8784 @code{long double} will be modified. Hence they will not be binary
8785 compatible with arrays or structures in code compiled without that switch.
8789 @itemx -mno-svr3-shlib
8790 @opindex msvr3-shlib
8791 @opindex mno-svr3-shlib
8792 Control whether GCC places uninitialized local variables into the
8793 @code{bss} or @code{data} segments. @option{-msvr3-shlib} places them
8794 into @code{bss}. These options are meaningful only on System V Release 3.
8798 Use a different function-calling convention, in which functions that
8799 take a fixed number of arguments return with the @code{ret} @var{num}
8800 instruction, which pops their arguments while returning. This saves one
8801 instruction in the caller since there is no need to pop the arguments
8804 You can specify that an individual function is called with this calling
8805 sequence with the function attribute @samp{stdcall}. You can also
8806 override the @option{-mrtd} option by using the function attribute
8807 @samp{cdecl}. @xref{Function Attributes}.
8809 @strong{Warning:} this calling convention is incompatible with the one
8810 normally used on Unix, so you cannot use it if you need to call
8811 libraries compiled with the Unix compiler.
8813 Also, you must provide function prototypes for all functions that
8814 take variable numbers of arguments (including @code{printf});
8815 otherwise incorrect code will be generated for calls to those
8818 In addition, seriously incorrect code will result if you call a
8819 function with too many arguments. (Normally, extra arguments are
8820 harmlessly ignored.)
8822 @item -mregparm=@var{num}
8824 Control how many registers are used to pass integer arguments. By
8825 default, no registers are used to pass arguments, and at most 3
8826 registers can be used. You can control this behavior for a specific
8827 function by using the function attribute @samp{regparm}.
8828 @xref{Function Attributes}.
8830 @strong{Warning:} if you use this switch, and
8831 @var{num} is nonzero, then you must build all modules with the same
8832 value, including any libraries. This includes the system libraries and
8835 @item -mpreferred-stack-boundary=@var{num}
8836 @opindex mpreferred-stack-boundary
8837 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
8838 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
8839 the default is 4 (16 bytes or 128 bits), except when optimizing for code
8840 size (@option{-Os}), in which case the default is the minimum correct
8841 alignment (4 bytes for x86, and 8 bytes for x86-64).
8843 On Pentium and PentiumPro, @code{double} and @code{long double} values
8844 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
8845 suffer significant run time performance penalties. On Pentium III, the
8846 Streaming SIMD Extension (SSE) data type @code{__m128} suffers similar
8847 penalties if it is not 16 byte aligned.
8849 To ensure proper alignment of this values on the stack, the stack boundary
8850 must be as aligned as that required by any value stored on the stack.
8851 Further, every function must be generated such that it keeps the stack
8852 aligned. Thus calling a function compiled with a higher preferred
8853 stack boundary from a function compiled with a lower preferred stack
8854 boundary will most likely misalign the stack. It is recommended that
8855 libraries that use callbacks always use the default setting.
8857 This extra alignment does consume extra stack space, and generally
8858 increases code size. Code that is sensitive to stack space usage, such
8859 as embedded systems and operating system kernels, may want to reduce the
8860 preferred alignment to @option{-mpreferred-stack-boundary=2}.
8878 These switches enable or disable the use of built-in functions that allow
8879 direct access to the MMX, SSE, SSE2, SSE3 and 3Dnow extensions of the
8882 @xref{X86 Built-in Functions}, for details of the functions enabled
8883 and disabled by these switches.
8885 To have SSE/SSE2 instructions generated automatically from floating-point
8886 code, see @option{-mfpmath=sse}.
8889 @itemx -mno-push-args
8891 @opindex mno-push-args
8892 Use PUSH operations to store outgoing parameters. This method is shorter
8893 and usually equally fast as method using SUB/MOV operations and is enabled
8894 by default. In some cases disabling it may improve performance because of
8895 improved scheduling and reduced dependencies.
8897 @item -maccumulate-outgoing-args
8898 @opindex maccumulate-outgoing-args
8899 If enabled, the maximum amount of space required for outgoing arguments will be
8900 computed in the function prologue. This is faster on most modern CPUs
8901 because of reduced dependencies, improved scheduling and reduced stack usage
8902 when preferred stack boundary is not equal to 2. The drawback is a notable
8903 increase in code size. This switch implies @option{-mno-push-args}.
8907 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
8908 on thread-safe exception handling must compile and link all code with the
8909 @option{-mthreads} option. When compiling, @option{-mthreads} defines
8910 @option{-D_MT}; when linking, it links in a special thread helper library
8911 @option{-lmingwthrd} which cleans up per thread exception handling data.
8913 @item -mno-align-stringops
8914 @opindex mno-align-stringops
8915 Do not align destination of inlined string operations. This switch reduces
8916 code size and improves performance in case the destination is already aligned,
8917 but GCC doesn't know about it.
8919 @item -minline-all-stringops
8920 @opindex minline-all-stringops
8921 By default GCC inlines string operations only when destination is known to be
8922 aligned at least to 4 byte boundary. This enables more inlining, increase code
8923 size, but may improve performance of code that depends on fast memcpy, strlen
8924 and memset for short lengths.
8926 @item -momit-leaf-frame-pointer
8927 @opindex momit-leaf-frame-pointer
8928 Don't keep the frame pointer in a register for leaf functions. This
8929 avoids the instructions to save, set up and restore frame pointers and
8930 makes an extra register available in leaf functions. The option
8931 @option{-fomit-frame-pointer} removes the frame pointer for all functions
8932 which might make debugging harder.
8934 @item -mtls-direct-seg-refs
8935 @itemx -mno-tls-direct-seg-refs
8936 @opindex mtls-direct-seg-refs
8937 Controls whether TLS variables may be accessed with offsets from the
8938 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
8939 or whether the thread base pointer must be added. Whether or not this
8940 is legal depends on the operating system, and whether it maps the
8941 segment to cover the entire TLS area.
8943 For systems that use GNU libc, the default is on.
8946 These @samp{-m} switches are supported in addition to the above
8947 on AMD x86-64 processors in 64-bit environments.
8954 Generate code for a 32-bit or 64-bit environment.
8955 The 32-bit environment sets int, long and pointer to 32 bits and
8956 generates code that runs on any i386 system.
8957 The 64-bit environment sets int to 32 bits and long and pointer
8958 to 64 bits and generates code for AMD's x86-64 architecture.
8961 @opindex no-red-zone
8962 Do not use a so called red zone for x86-64 code. The red zone is mandated
8963 by the x86-64 ABI, it is a 128-byte area beyond the location of the
8964 stack pointer that will not be modified by signal or interrupt handlers
8965 and therefore can be used for temporary data without adjusting the stack
8966 pointer. The flag @option{-mno-red-zone} disables this red zone.
8968 @item -mcmodel=small
8969 @opindex mcmodel=small
8970 Generate code for the small code model: the program and its symbols must
8971 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
8972 Programs can be statically or dynamically linked. This is the default
8975 @item -mcmodel=kernel
8976 @opindex mcmodel=kernel
8977 Generate code for the kernel code model. The kernel runs in the
8978 negative 2 GB of the address space.
8979 This model has to be used for Linux kernel code.
8981 @item -mcmodel=medium
8982 @opindex mcmodel=medium
8983 Generate code for the medium model: The program is linked in the lower 2
8984 GB of the address space but symbols can be located anywhere in the
8985 address space. Programs can be statically or dynamically linked, but
8986 building of shared libraries are not supported with the medium model.
8988 @item -mcmodel=large
8989 @opindex mcmodel=large
8990 Generate code for the large model: This model makes no assumptions
8991 about addresses and sizes of sections. Currently GCC does not implement
8996 @subsection HPPA Options
8997 @cindex HPPA Options
8999 These @samp{-m} options are defined for the HPPA family of computers:
9002 @item -march=@var{architecture-type}
9004 Generate code for the specified architecture. The choices for
9005 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
9006 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
9007 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
9008 architecture option for your machine. Code compiled for lower numbered
9009 architectures will run on higher numbered architectures, but not the
9012 PA 2.0 support currently requires gas snapshot 19990413 or later. The
9013 next release of binutils (current is 2.9.1) will probably contain PA 2.0
9017 @itemx -mpa-risc-1-1
9018 @itemx -mpa-risc-2-0
9019 @opindex mpa-risc-1-0
9020 @opindex mpa-risc-1-1
9021 @opindex mpa-risc-2-0
9022 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
9025 @opindex mbig-switch
9026 Generate code suitable for big switch tables. Use this option only if
9027 the assembler/linker complain about out of range branches within a switch
9030 @item -mjump-in-delay
9031 @opindex mjump-in-delay
9032 Fill delay slots of function calls with unconditional jump instructions
9033 by modifying the return pointer for the function call to be the target
9034 of the conditional jump.
9036 @item -mdisable-fpregs
9037 @opindex mdisable-fpregs
9038 Prevent floating point registers from being used in any manner. This is
9039 necessary for compiling kernels which perform lazy context switching of
9040 floating point registers. If you use this option and attempt to perform
9041 floating point operations, the compiler will abort.
9043 @item -mdisable-indexing
9044 @opindex mdisable-indexing
9045 Prevent the compiler from using indexing address modes. This avoids some
9046 rather obscure problems when compiling MIG generated code under MACH@.
9048 @item -mno-space-regs
9049 @opindex mno-space-regs
9050 Generate code that assumes the target has no space registers. This allows
9051 GCC to generate faster indirect calls and use unscaled index address modes.
9053 Such code is suitable for level 0 PA systems and kernels.
9055 @item -mfast-indirect-calls
9056 @opindex mfast-indirect-calls
9057 Generate code that assumes calls never cross space boundaries. This
9058 allows GCC to emit code which performs faster indirect calls.
9060 This option will not work in the presence of shared libraries or nested
9063 @item -mlong-load-store
9064 @opindex mlong-load-store
9065 Generate 3-instruction load and store sequences as sometimes required by
9066 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
9069 @item -mportable-runtime
9070 @opindex mportable-runtime
9071 Use the portable calling conventions proposed by HP for ELF systems.
9075 Enable the use of assembler directives only GAS understands.
9077 @item -mschedule=@var{cpu-type}
9079 Schedule code according to the constraints for the machine type
9080 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
9081 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
9082 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
9083 proper scheduling option for your machine. The default scheduling is
9087 @opindex mlinker-opt
9088 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
9089 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
9090 linkers in which they give bogus error messages when linking some programs.
9093 @opindex msoft-float
9094 Generate output containing library calls for floating point.
9095 @strong{Warning:} the requisite libraries are not available for all HPPA
9096 targets. Normally the facilities of the machine's usual C compiler are
9097 used, but this cannot be done directly in cross-compilation. You must make
9098 your own arrangements to provide suitable library functions for
9099 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
9100 does provide software floating point support.
9102 @option{-msoft-float} changes the calling convention in the output file;
9103 therefore, it is only useful if you compile @emph{all} of a program with
9104 this option. In particular, you need to compile @file{libgcc.a}, the
9105 library that comes with GCC, with @option{-msoft-float} in order for
9110 Generate the predefine, @code{_SIO}, for server IO. The default is
9111 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
9112 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO. These
9113 options are available under HP-UX and HI-UX.
9117 Use GNU ld specific options. This passes @option{-shared} to ld when
9118 building a shared library. It is the default when GCC is configured,
9119 explicitly or implicitly, with the GNU linker. This option does not
9120 have any affect on which ld is called, it only changes what parameters
9121 are passed to that ld. The ld that is called is determined by the
9122 @option{--with-ld} configure option, GCC's program search path, and
9123 finally by the user's @env{PATH}. The linker used by GCC can be printed
9124 using @samp{which `gcc -print-prog-name=ld`}.
9128 Use HP ld specific options. This passes @option{-b} to ld when building
9129 a shared library and passes @option{+Accept TypeMismatch} to ld on all
9130 links. It is the default when GCC is configured, explicitly or
9131 implicitly, with the HP linker. This option does not have any affect on
9132 which ld is called, it only changes what parameters are passed to that
9133 ld. The ld that is called is determined by the @option{--with-ld}
9134 configure option, GCC's program search path, and finally by the user's
9135 @env{PATH}. The linker used by GCC can be printed using @samp{which
9136 `gcc -print-prog-name=ld`}.
9141 Select the FDPIC ABI, that uses function descriptors to represent
9142 pointers to functions. Without any PIC/PIE-related options, it
9143 implies @option{-fPIE}. With @option{-fpic} or @option{-fpie}, it
9144 assumes GOT entries and small data are within a 12-bit range from the
9145 GOT base address; with @option{-fPIC} or @option{-fPIE}, GOT offsets
9146 are computed with 32 bits.
9149 @opindex minline-plt
9151 Enable inlining of PLT entries in function calls to functions that are
9152 not known to bind locally. It has no effect without @option{-mfdpic}.
9153 It's enabled by default if optimizing for speed and compiling for
9154 shared libraries (i.e., @option{-fPIC} or @option{-fpic}), or when an
9155 optimization option such as @option{-O3} or above is present in the
9161 Enable the use of @code{GPREL} relocations in the FDPIC ABI for data
9162 that is known to be in read-only sections. It's enabled by default,
9163 except for @option{-fpic} or @option{-fpie}: even though it may help
9164 make the global offset table smaller, it trades 1 instruction for 4.
9165 With @option{-fPIC} or @option{-fPIE}, it trades 3 instructions for 4,
9166 one of which may be shared by multiple symbols, and it avoids the need
9167 for a GOT entry for the referenced symbol, so it's more likely to be a
9168 win. If it is not, @option{-mno-gprel-ro} can be used to disable it.
9170 @item -multilib-library-pic
9171 @opindex multilib-library-pic
9173 Link with the (library, not FD) pic libraries. It's implied by
9174 @option{-mlibrary-pic}, as well as by @option{-fPIC} and
9175 @option{-fpic} without @option{-mfdpic}. You should never have to use
9181 Follow the EABI requirement of always creating a frame pointer whenever
9182 a stack frame is allocated. This option is enabled by default and can
9183 be disabled with @option{-mno-linked-fp}.
9186 @opindex mno-long-calls
9187 Generate code that uses long call sequences. This ensures that a call
9188 is always able to reach linker generated stubs. The default is to generate
9189 long calls only when the distance from the call site to the beginning
9190 of the function or translation unit, as the case may be, exceeds a
9191 predefined limit set by the branch type being used. The limits for
9192 normal calls are 7,600,000 and 240,000 bytes, respectively for the
9193 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
9196 Distances are measured from the beginning of functions when using the
9197 @option{-ffunction-sections} option, or when using the @option{-mgas}
9198 and @option{-mno-portable-runtime} options together under HP-UX with
9201 It is normally not desirable to use this option as it will degrade
9202 performance. However, it may be useful in large applications,
9203 particularly when partial linking is used to build the application.
9205 The types of long calls used depends on the capabilities of the
9206 assembler and linker, and the type of code being generated. The
9207 impact on systems that support long absolute calls, and long pic
9208 symbol-difference or pc-relative calls should be relatively small.
9209 However, an indirect call is used on 32-bit ELF systems in pic code
9210 and it is quite long.
9214 Suppress the generation of link options to search libdld.sl when the
9215 @option{-static} option is specified on HP-UX 10 and later.
9219 The HP-UX implementation of setlocale in libc has a dependency on
9220 libdld.sl. There isn't an archive version of libdld.sl. Thus,
9221 when the @option{-static} option is specified, special link options
9222 are needed to resolve this dependency.
9224 On HP-UX 10 and later, the GCC driver adds the necessary options to
9225 link with libdld.sl when the @option{-static} option is specified.
9226 This causes the resulting binary to be dynamic. On the 64-bit port,
9227 the linkers generate dynamic binaries by default in any case. The
9228 @option{-nolibdld} option can be used to prevent the GCC driver from
9229 adding these link options.
9233 Add support for multithreading with the @dfn{dce thread} library
9234 under HP-UX. This option sets flags for both the preprocessor and
9238 @node DEC Alpha Options
9239 @subsection DEC Alpha Options
9241 These @samp{-m} options are defined for the DEC Alpha implementations:
9244 @item -mno-soft-float
9246 @opindex mno-soft-float
9247 @opindex msoft-float
9248 Use (do not use) the hardware floating-point instructions for
9249 floating-point operations. When @option{-msoft-float} is specified,
9250 functions in @file{libgcc.a} will be used to perform floating-point
9251 operations. Unless they are replaced by routines that emulate the
9252 floating-point operations, or compiled in such a way as to call such
9253 emulations routines, these routines will issue floating-point
9254 operations. If you are compiling for an Alpha without floating-point
9255 operations, you must ensure that the library is built so as not to call
9258 Note that Alpha implementations without floating-point operations are
9259 required to have floating-point registers.
9264 @opindex mno-fp-regs
9265 Generate code that uses (does not use) the floating-point register set.
9266 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
9267 register set is not used, floating point operands are passed in integer
9268 registers as if they were integers and floating-point results are passed
9269 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
9270 so any function with a floating-point argument or return value called by code
9271 compiled with @option{-mno-fp-regs} must also be compiled with that
9274 A typical use of this option is building a kernel that does not use,
9275 and hence need not save and restore, any floating-point registers.
9279 The Alpha architecture implements floating-point hardware optimized for
9280 maximum performance. It is mostly compliant with the IEEE floating
9281 point standard. However, for full compliance, software assistance is
9282 required. This option generates code fully IEEE compliant code
9283 @emph{except} that the @var{inexact-flag} is not maintained (see below).
9284 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
9285 defined during compilation. The resulting code is less efficient but is
9286 able to correctly support denormalized numbers and exceptional IEEE
9287 values such as not-a-number and plus/minus infinity. Other Alpha
9288 compilers call this option @option{-ieee_with_no_inexact}.
9290 @item -mieee-with-inexact
9291 @opindex mieee-with-inexact
9292 This is like @option{-mieee} except the generated code also maintains
9293 the IEEE @var{inexact-flag}. Turning on this option causes the
9294 generated code to implement fully-compliant IEEE math. In addition to
9295 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
9296 macro. On some Alpha implementations the resulting code may execute
9297 significantly slower than the code generated by default. Since there is
9298 very little code that depends on the @var{inexact-flag}, you should
9299 normally not specify this option. Other Alpha compilers call this
9300 option @option{-ieee_with_inexact}.
9302 @item -mfp-trap-mode=@var{trap-mode}
9303 @opindex mfp-trap-mode
9304 This option controls what floating-point related traps are enabled.
9305 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
9306 The trap mode can be set to one of four values:
9310 This is the default (normal) setting. The only traps that are enabled
9311 are the ones that cannot be disabled in software (e.g., division by zero
9315 In addition to the traps enabled by @samp{n}, underflow traps are enabled
9319 Like @samp{su}, but the instructions are marked to be safe for software
9320 completion (see Alpha architecture manual for details).
9323 Like @samp{su}, but inexact traps are enabled as well.
9326 @item -mfp-rounding-mode=@var{rounding-mode}
9327 @opindex mfp-rounding-mode
9328 Selects the IEEE rounding mode. Other Alpha compilers call this option
9329 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
9334 Normal IEEE rounding mode. Floating point numbers are rounded towards
9335 the nearest machine number or towards the even machine number in case
9339 Round towards minus infinity.
9342 Chopped rounding mode. Floating point numbers are rounded towards zero.
9345 Dynamic rounding mode. A field in the floating point control register
9346 (@var{fpcr}, see Alpha architecture reference manual) controls the
9347 rounding mode in effect. The C library initializes this register for
9348 rounding towards plus infinity. Thus, unless your program modifies the
9349 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
9352 @item -mtrap-precision=@var{trap-precision}
9353 @opindex mtrap-precision
9354 In the Alpha architecture, floating point traps are imprecise. This
9355 means without software assistance it is impossible to recover from a
9356 floating trap and program execution normally needs to be terminated.
9357 GCC can generate code that can assist operating system trap handlers
9358 in determining the exact location that caused a floating point trap.
9359 Depending on the requirements of an application, different levels of
9360 precisions can be selected:
9364 Program precision. This option is the default and means a trap handler
9365 can only identify which program caused a floating point exception.
9368 Function precision. The trap handler can determine the function that
9369 caused a floating point exception.
9372 Instruction precision. The trap handler can determine the exact
9373 instruction that caused a floating point exception.
9376 Other Alpha compilers provide the equivalent options called
9377 @option{-scope_safe} and @option{-resumption_safe}.
9379 @item -mieee-conformant
9380 @opindex mieee-conformant
9381 This option marks the generated code as IEEE conformant. You must not
9382 use this option unless you also specify @option{-mtrap-precision=i} and either
9383 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
9384 is to emit the line @samp{.eflag 48} in the function prologue of the
9385 generated assembly file. Under DEC Unix, this has the effect that
9386 IEEE-conformant math library routines will be linked in.
9388 @item -mbuild-constants
9389 @opindex mbuild-constants
9390 Normally GCC examines a 32- or 64-bit integer constant to
9391 see if it can construct it from smaller constants in two or three
9392 instructions. If it cannot, it will output the constant as a literal and
9393 generate code to load it from the data segment at runtime.
9395 Use this option to require GCC to construct @emph{all} integer constants
9396 using code, even if it takes more instructions (the maximum is six).
9398 You would typically use this option to build a shared library dynamic
9399 loader. Itself a shared library, it must relocate itself in memory
9400 before it can find the variables and constants in its own data segment.
9406 Select whether to generate code to be assembled by the vendor-supplied
9407 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
9425 Indicate whether GCC should generate code to use the optional BWX,
9426 CIX, FIX and MAX instruction sets. The default is to use the instruction
9427 sets supported by the CPU type specified via @option{-mcpu=} option or that
9428 of the CPU on which GCC was built if none was specified.
9433 @opindex mfloat-ieee
9434 Generate code that uses (does not use) VAX F and G floating point
9435 arithmetic instead of IEEE single and double precision.
9437 @item -mexplicit-relocs
9438 @itemx -mno-explicit-relocs
9439 @opindex mexplicit-relocs
9440 @opindex mno-explicit-relocs
9441 Older Alpha assemblers provided no way to generate symbol relocations
9442 except via assembler macros. Use of these macros does not allow
9443 optimal instruction scheduling. GNU binutils as of version 2.12
9444 supports a new syntax that allows the compiler to explicitly mark
9445 which relocations should apply to which instructions. This option
9446 is mostly useful for debugging, as GCC detects the capabilities of
9447 the assembler when it is built and sets the default accordingly.
9451 @opindex msmall-data
9452 @opindex mlarge-data
9453 When @option{-mexplicit-relocs} is in effect, static data is
9454 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
9455 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
9456 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
9457 16-bit relocations off of the @code{$gp} register. This limits the
9458 size of the small data area to 64KB, but allows the variables to be
9459 directly accessed via a single instruction.
9461 The default is @option{-mlarge-data}. With this option the data area
9462 is limited to just below 2GB. Programs that require more than 2GB of
9463 data must use @code{malloc} or @code{mmap} to allocate the data in the
9464 heap instead of in the program's data segment.
9466 When generating code for shared libraries, @option{-fpic} implies
9467 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
9471 @opindex msmall-text
9472 @opindex mlarge-text
9473 When @option{-msmall-text} is used, the compiler assumes that the
9474 code of the entire program (or shared library) fits in 4MB, and is
9475 thus reachable with a branch instruction. When @option{-msmall-data}
9476 is used, the compiler can assume that all local symbols share the
9477 same @code{$gp} value, and thus reduce the number of instructions
9478 required for a function call from 4 to 1.
9480 The default is @option{-mlarge-text}.
9482 @item -mcpu=@var{cpu_type}
9484 Set the instruction set and instruction scheduling parameters for
9485 machine type @var{cpu_type}. You can specify either the @samp{EV}
9486 style name or the corresponding chip number. GCC supports scheduling
9487 parameters for the EV4, EV5 and EV6 family of processors and will
9488 choose the default values for the instruction set from the processor
9489 you specify. If you do not specify a processor type, GCC will default
9490 to the processor on which the compiler was built.
9492 Supported values for @var{cpu_type} are
9498 Schedules as an EV4 and has no instruction set extensions.
9502 Schedules as an EV5 and has no instruction set extensions.
9506 Schedules as an EV5 and supports the BWX extension.
9511 Schedules as an EV5 and supports the BWX and MAX extensions.
9515 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
9519 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
9522 @item -mtune=@var{cpu_type}
9524 Set only the instruction scheduling parameters for machine type
9525 @var{cpu_type}. The instruction set is not changed.
9527 @item -mmemory-latency=@var{time}
9528 @opindex mmemory-latency
9529 Sets the latency the scheduler should assume for typical memory
9530 references as seen by the application. This number is highly
9531 dependent on the memory access patterns used by the application
9532 and the size of the external cache on the machine.
9534 Valid options for @var{time} are
9538 A decimal number representing clock cycles.
9544 The compiler contains estimates of the number of clock cycles for
9545 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
9546 (also called Dcache, Scache, and Bcache), as well as to main memory.
9547 Note that L3 is only valid for EV5.
9552 @node DEC Alpha/VMS Options
9553 @subsection DEC Alpha/VMS Options
9555 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
9558 @item -mvms-return-codes
9559 @opindex mvms-return-codes
9560 Return VMS condition codes from main. The default is to return POSIX
9561 style condition (e.g.@ error) codes.
9564 @node H8/300 Options
9565 @subsection H8/300 Options
9567 These @samp{-m} options are defined for the H8/300 implementations:
9572 Shorten some address references at link time, when possible; uses the
9573 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
9574 ld, Using ld}, for a fuller description.
9578 Generate code for the H8/300H@.
9582 Generate code for the H8S@.
9586 Generate code for the H8S and H8/300H in the normal mode. This switch
9587 must be used either with -mh or -ms.
9591 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
9595 Make @code{int} data 32 bits by default.
9599 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
9600 The default for the H8/300H and H8S is to align longs and floats on 4
9602 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
9603 This option has no effect on the H8/300.
9607 @subsection SH Options
9609 These @samp{-m} options are defined for the SH implementations:
9614 Generate code for the SH1.
9618 Generate code for the SH2.
9621 Generate code for the SH2e.
9625 Generate code for the SH3.
9629 Generate code for the SH3e.
9633 Generate code for the SH4 without a floating-point unit.
9635 @item -m4-single-only
9636 @opindex m4-single-only
9637 Generate code for the SH4 with a floating-point unit that only
9638 supports single-precision arithmetic.
9642 Generate code for the SH4 assuming the floating-point unit is in
9643 single-precision mode by default.
9647 Generate code for the SH4.
9651 Compile code for the processor in big endian mode.
9655 Compile code for the processor in little endian mode.
9659 Align doubles at 64-bit boundaries. Note that this changes the calling
9660 conventions, and thus some functions from the standard C library will
9661 not work unless you recompile it first with @option{-mdalign}.
9665 Shorten some address references at link time, when possible; uses the
9666 linker option @option{-relax}.
9670 Use 32-bit offsets in @code{switch} tables. The default is to use
9675 Enable the use of the instruction @code{fmovd}.
9679 Comply with the calling conventions defined by Renesas.
9683 Mark the @code{MAC} register as call-clobbered, even if
9684 @option{-mhitachi} is given.
9688 Increase IEEE-compliance of floating-point code.
9692 Dump instruction size and location in the assembly code.
9696 This option is deprecated. It pads structures to multiple of 4 bytes,
9697 which is incompatible with the SH ABI@.
9701 Optimize for space instead of speed. Implied by @option{-Os}.
9705 When generating position-independent code, emit function calls using
9706 the Global Offset Table instead of the Procedure Linkage Table.
9710 Generate a library function call to invalidate instruction cache
9711 entries, after fixing up a trampoline. This library function call
9712 doesn't assume it can write to the whole memory address space. This
9713 is the default when the target is @code{sh-*-linux*}.
9716 @node System V Options
9717 @subsection Options for System V
9719 These additional options are available on System V Release 4 for
9720 compatibility with other compilers on those systems:
9725 Create a shared object.
9726 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
9730 Identify the versions of each tool used by the compiler, in a
9731 @code{.ident} assembler directive in the output.
9735 Refrain from adding @code{.ident} directives to the output file (this is
9738 @item -YP,@var{dirs}
9740 Search the directories @var{dirs}, and no others, for libraries
9741 specified with @option{-l}.
9745 Look in the directory @var{dir} to find the M4 preprocessor.
9746 The assembler uses this option.
9747 @c This is supposed to go with a -Yd for predefined M4 macro files, but
9748 @c the generic assembler that comes with Solaris takes just -Ym.
9751 @node TMS320C3x/C4x Options
9752 @subsection TMS320C3x/C4x Options
9753 @cindex TMS320C3x/C4x Options
9755 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
9759 @item -mcpu=@var{cpu_type}
9761 Set the instruction set, register set, and instruction scheduling
9762 parameters for machine type @var{cpu_type}. Supported values for
9763 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
9764 @samp{c44}. The default is @samp{c40} to generate code for the
9769 @itemx -msmall-memory
9771 @opindex mbig-memory
9773 @opindex msmall-memory
9775 Generates code for the big or small memory model. The small memory
9776 model assumed that all data fits into one 64K word page. At run-time
9777 the data page (DP) register must be set to point to the 64K page
9778 containing the .bss and .data program sections. The big memory model is
9779 the default and requires reloading of the DP register for every direct
9786 Allow (disallow) allocation of general integer operands into the block
9793 Enable (disable) generation of code using decrement and branch,
9794 DBcond(D), instructions. This is enabled by default for the C4x. To be
9795 on the safe side, this is disabled for the C3x, since the maximum
9796 iteration count on the C3x is @math{2^{23} + 1} (but who iterates loops more than
9797 @math{2^{23}} times on the C3x?). Note that GCC will try to reverse a loop so
9798 that it can utilize the decrement and branch instruction, but will give
9799 up if there is more than one memory reference in the loop. Thus a loop
9800 where the loop counter is decremented can generate slightly more
9801 efficient code, in cases where the RPTB instruction cannot be utilized.
9803 @item -mdp-isr-reload
9805 @opindex mdp-isr-reload
9807 Force the DP register to be saved on entry to an interrupt service
9808 routine (ISR), reloaded to point to the data section, and restored on
9809 exit from the ISR@. This should not be required unless someone has
9810 violated the small memory model by modifying the DP register, say within
9817 For the C3x use the 24-bit MPYI instruction for integer multiplies
9818 instead of a library call to guarantee 32-bit results. Note that if one
9819 of the operands is a constant, then the multiplication will be performed
9820 using shifts and adds. If the @option{-mmpyi} option is not specified for the C3x,
9821 then squaring operations are performed inline instead of a library call.
9824 @itemx -mno-fast-fix
9826 @opindex mno-fast-fix
9827 The C3x/C4x FIX instruction to convert a floating point value to an
9828 integer value chooses the nearest integer less than or equal to the
9829 floating point value rather than to the nearest integer. Thus if the
9830 floating point number is negative, the result will be incorrectly
9831 truncated an additional code is necessary to detect and correct this
9832 case. This option can be used to disable generation of the additional
9833 code required to correct the result.
9839 Enable (disable) generation of repeat block sequences using the RPTB
9840 instruction for zero overhead looping. The RPTB construct is only used
9841 for innermost loops that do not call functions or jump across the loop
9842 boundaries. There is no advantage having nested RPTB loops due to the
9843 overhead required to save and restore the RC, RS, and RE registers.
9844 This is enabled by default with @option{-O2}.
9846 @item -mrpts=@var{count}
9850 Enable (disable) the use of the single instruction repeat instruction
9851 RPTS@. If a repeat block contains a single instruction, and the loop
9852 count can be guaranteed to be less than the value @var{count}, GCC will
9853 emit a RPTS instruction instead of a RPTB@. If no value is specified,
9854 then a RPTS will be emitted even if the loop count cannot be determined
9855 at compile time. Note that the repeated instruction following RPTS does
9856 not have to be reloaded from memory each iteration, thus freeing up the
9857 CPU buses for operands. However, since interrupts are blocked by this
9858 instruction, it is disabled by default.
9860 @item -mloop-unsigned
9861 @itemx -mno-loop-unsigned
9862 @opindex mloop-unsigned
9863 @opindex mno-loop-unsigned
9864 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
9865 is @math{2^{31} + 1} since these instructions test if the iteration count is
9866 negative to terminate the loop. If the iteration count is unsigned
9867 there is a possibility than the @math{2^{31} + 1} maximum iteration count may be
9868 exceeded. This switch allows an unsigned iteration count.
9872 Try to emit an assembler syntax that the TI assembler (asm30) is happy
9873 with. This also enforces compatibility with the API employed by the TI
9874 C3x C compiler. For example, long doubles are passed as structures
9875 rather than in floating point registers.
9881 Generate code that uses registers (stack) for passing arguments to functions.
9882 By default, arguments are passed in registers where possible rather
9883 than by pushing arguments on to the stack.
9885 @item -mparallel-insns
9886 @itemx -mno-parallel-insns
9887 @opindex mparallel-insns
9888 @opindex mno-parallel-insns
9889 Allow the generation of parallel instructions. This is enabled by
9890 default with @option{-O2}.
9892 @item -mparallel-mpy
9893 @itemx -mno-parallel-mpy
9894 @opindex mparallel-mpy
9895 @opindex mno-parallel-mpy
9896 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
9897 provided @option{-mparallel-insns} is also specified. These instructions have
9898 tight register constraints which can pessimize the code generation
9904 @subsection V850 Options
9905 @cindex V850 Options
9907 These @samp{-m} options are defined for V850 implementations:
9911 @itemx -mno-long-calls
9912 @opindex mlong-calls
9913 @opindex mno-long-calls
9914 Treat all calls as being far away (near). If calls are assumed to be
9915 far away, the compiler will always load the functions address up into a
9916 register, and call indirect through the pointer.
9922 Do not optimize (do optimize) basic blocks that use the same index
9923 pointer 4 or more times to copy pointer into the @code{ep} register, and
9924 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
9925 option is on by default if you optimize.
9927 @item -mno-prolog-function
9928 @itemx -mprolog-function
9929 @opindex mno-prolog-function
9930 @opindex mprolog-function
9931 Do not use (do use) external functions to save and restore registers
9932 at the prologue and epilogue of a function. The external functions
9933 are slower, but use less code space if more than one function saves
9934 the same number of registers. The @option{-mprolog-function} option
9935 is on by default if you optimize.
9939 Try to make the code as small as possible. At present, this just turns
9940 on the @option{-mep} and @option{-mprolog-function} options.
9944 Put static or global variables whose size is @var{n} bytes or less into
9945 the tiny data area that register @code{ep} points to. The tiny data
9946 area can hold up to 256 bytes in total (128 bytes for byte references).
9950 Put static or global variables whose size is @var{n} bytes or less into
9951 the small data area that register @code{gp} points to. The small data
9952 area can hold up to 64 kilobytes.
9956 Put static or global variables whose size is @var{n} bytes or less into
9957 the first 32 kilobytes of memory.
9961 Specify that the target processor is the V850.
9964 @opindex mbig-switch
9965 Generate code suitable for big switch tables. Use this option only if
9966 the assembler/linker complain about out of range branches within a switch
9971 This option will cause r2 and r5 to be used in the code generated by
9972 the compiler. This setting is the default.
9975 @opindex mno-app-regs
9976 This option will cause r2 and r5 to be treated as fixed registers.
9980 Specify that the target processor is the V850E1. The preprocessor
9981 constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
9982 this option is used.
9986 Specify that the target processor is the V850E. The preprocessor
9987 constant @samp{__v850e__} will be defined if this option is used.
9989 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
9990 are defined then a default target processor will be chosen and the
9991 relevant @samp{__v850*__} preprocessor constant will be defined.
9993 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
9994 defined, regardless of which processor variant is the target.
9996 @item -mdisable-callt
9997 @opindex mdisable-callt
9998 This option will suppress generation of the CALLT instruction for the
9999 v850e and v850e1 flavors of the v850 architecture. The default is
10000 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
10005 @subsection ARC Options
10006 @cindex ARC Options
10008 These options are defined for ARC implementations:
10013 Compile code for little endian mode. This is the default.
10017 Compile code for big endian mode.
10020 @opindex mmangle-cpu
10021 Prepend the name of the cpu to all public symbol names.
10022 In multiple-processor systems, there are many ARC variants with different
10023 instruction and register set characteristics. This flag prevents code
10024 compiled for one cpu to be linked with code compiled for another.
10025 No facility exists for handling variants that are ``almost identical''.
10026 This is an all or nothing option.
10028 @item -mcpu=@var{cpu}
10030 Compile code for ARC variant @var{cpu}.
10031 Which variants are supported depend on the configuration.
10032 All variants support @option{-mcpu=base}, this is the default.
10034 @item -mtext=@var{text-section}
10035 @itemx -mdata=@var{data-section}
10036 @itemx -mrodata=@var{readonly-data-section}
10040 Put functions, data, and readonly data in @var{text-section},
10041 @var{data-section}, and @var{readonly-data-section} respectively
10042 by default. This can be overridden with the @code{section} attribute.
10043 @xref{Variable Attributes}.
10047 @node NS32K Options
10048 @subsection NS32K Options
10049 @cindex NS32K options
10051 These are the @samp{-m} options defined for the 32000 series. The default
10052 values for these options depends on which style of 32000 was selected when
10053 the compiler was configured; the defaults for the most common choices are
10061 Generate output for a 32032. This is the default
10062 when the compiler is configured for 32032 and 32016 based systems.
10068 Generate output for a 32332. This is the default
10069 when the compiler is configured for 32332-based systems.
10075 Generate output for a 32532. This is the default
10076 when the compiler is configured for 32532-based systems.
10080 Generate output containing 32081 instructions for floating point.
10081 This is the default for all systems.
10085 Generate output containing 32381 instructions for floating point. This
10086 also implies @option{-m32081}. The 32381 is only compatible with the 32332
10087 and 32532 cpus. This is the default for the pc532-netbsd configuration.
10090 @opindex mmulti-add
10091 Try and generate multiply-add floating point instructions @code{polyF}
10092 and @code{dotF}. This option is only available if the @option{-m32381}
10093 option is in effect. Using these instructions requires changes to
10094 register allocation which generally has a negative impact on
10095 performance. This option should only be enabled when compiling code
10096 particularly likely to make heavy use of multiply-add instructions.
10098 @item -mnomulti-add
10099 @opindex mnomulti-add
10100 Do not try and generate multiply-add floating point instructions
10101 @code{polyF} and @code{dotF}. This is the default on all platforms.
10104 @opindex msoft-float
10105 Generate output containing library calls for floating point.
10106 @strong{Warning:} the requisite libraries may not be available.
10108 @item -mieee-compare
10109 @itemx -mno-ieee-compare
10110 @opindex mieee-compare
10111 @opindex mno-ieee-compare
10112 Control whether or not the compiler uses IEEE floating point
10113 comparisons. These handle correctly the case where the result of a
10114 comparison is unordered.
10115 @strong{Warning:} the requisite kernel support may not be available.
10118 @opindex mnobitfield
10119 Do not use the bit-field instructions. On some machines it is faster to
10120 use shifting and masking operations. This is the default for the pc532.
10124 Do use the bit-field instructions. This is the default for all platforms
10129 Use a different function-calling convention, in which functions
10130 that take a fixed number of arguments return pop their
10131 arguments on return with the @code{ret} instruction.
10133 This calling convention is incompatible with the one normally
10134 used on Unix, so you cannot use it if you need to call libraries
10135 compiled with the Unix compiler.
10137 Also, you must provide function prototypes for all functions that
10138 take variable numbers of arguments (including @code{printf});
10139 otherwise incorrect code will be generated for calls to those
10142 In addition, seriously incorrect code will result if you call a
10143 function with too many arguments. (Normally, extra arguments are
10144 harmlessly ignored.)
10146 This option takes its name from the 680x0 @code{rtd} instruction.
10151 Use a different function-calling convention where the first two arguments
10152 are passed in registers.
10154 This calling convention is incompatible with the one normally
10155 used on Unix, so you cannot use it if you need to call libraries
10156 compiled with the Unix compiler.
10159 @opindex mnoregparam
10160 Do not pass any arguments in registers. This is the default for all
10165 It is OK to use the sb as an index register which is always loaded with
10166 zero. This is the default for the pc532-netbsd target.
10170 The sb register is not available for use or has not been initialized to
10171 zero by the run time system. This is the default for all targets except
10172 the pc532-netbsd. It is also implied whenever @option{-mhimem} or
10173 @option{-fpic} is set.
10177 Many ns32000 series addressing modes use displacements of up to 512MB@.
10178 If an address is above 512MB then displacements from zero can not be used.
10179 This option causes code to be generated which can be loaded above 512MB@.
10180 This may be useful for operating systems or ROM code.
10184 Assume code will be loaded in the first 512MB of virtual address space.
10185 This is the default for all platforms.
10191 @subsection AVR Options
10192 @cindex AVR Options
10194 These options are defined for AVR implementations:
10197 @item -mmcu=@var{mcu}
10199 Specify ATMEL AVR instruction set or MCU type.
10201 Instruction set avr1 is for the minimal AVR core, not supported by the C
10202 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
10203 attiny11, attiny12, attiny15, attiny28).
10205 Instruction set avr2 (default) is for the classic AVR core with up to
10206 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
10207 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
10208 at90c8534, at90s8535).
10210 Instruction set avr3 is for the classic AVR core with up to 128K program
10211 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
10213 Instruction set avr4 is for the enhanced AVR core with up to 8K program
10214 memory space (MCU types: atmega8, atmega83, atmega85).
10216 Instruction set avr5 is for the enhanced AVR core with up to 128K program
10217 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
10218 atmega64, atmega128, at43usb355, at94k).
10222 Output instruction sizes to the asm file.
10224 @item -minit-stack=@var{N}
10225 @opindex minit-stack
10226 Specify the initial stack address, which may be a symbol or numeric value,
10227 @samp{__stack} is the default.
10229 @item -mno-interrupts
10230 @opindex mno-interrupts
10231 Generated code is not compatible with hardware interrupts.
10232 Code size will be smaller.
10234 @item -mcall-prologues
10235 @opindex mcall-prologues
10236 Functions prologues/epilogues expanded as call to appropriate
10237 subroutines. Code size will be smaller.
10239 @item -mno-tablejump
10240 @opindex mno-tablejump
10241 Do not generate tablejump insns which sometimes increase code size.
10244 @opindex mtiny-stack
10245 Change only the low 8 bits of the stack pointer.
10248 @node MCore Options
10249 @subsection MCore Options
10250 @cindex MCore options
10252 These are the @samp{-m} options defined for the Motorola M*Core
10258 @itemx -mno-hardlit
10260 @opindex mno-hardlit
10261 Inline constants into the code stream if it can be done in two
10262 instructions or less.
10268 Use the divide instruction. (Enabled by default).
10270 @item -mrelax-immediate
10271 @itemx -mno-relax-immediate
10272 @opindex mrelax-immediate
10273 @opindex mno-relax-immediate
10274 Allow arbitrary sized immediates in bit operations.
10276 @item -mwide-bitfields
10277 @itemx -mno-wide-bitfields
10278 @opindex mwide-bitfields
10279 @opindex mno-wide-bitfields
10280 Always treat bit-fields as int-sized.
10282 @item -m4byte-functions
10283 @itemx -mno-4byte-functions
10284 @opindex m4byte-functions
10285 @opindex mno-4byte-functions
10286 Force all functions to be aligned to a four byte boundary.
10288 @item -mcallgraph-data
10289 @itemx -mno-callgraph-data
10290 @opindex mcallgraph-data
10291 @opindex mno-callgraph-data
10292 Emit callgraph information.
10295 @itemx -mno-slow-bytes
10296 @opindex mslow-bytes
10297 @opindex mno-slow-bytes
10298 Prefer word access when reading byte quantities.
10300 @item -mlittle-endian
10301 @itemx -mbig-endian
10302 @opindex mlittle-endian
10303 @opindex mbig-endian
10304 Generate code for a little endian target.
10310 Generate code for the 210 processor.
10313 @node IA-64 Options
10314 @subsection IA-64 Options
10315 @cindex IA-64 Options
10317 These are the @samp{-m} options defined for the Intel IA-64 architecture.
10321 @opindex mbig-endian
10322 Generate code for a big endian target. This is the default for HP-UX@.
10324 @item -mlittle-endian
10325 @opindex mlittle-endian
10326 Generate code for a little endian target. This is the default for AIX5
10332 @opindex mno-gnu-as
10333 Generate (or don't) code for the GNU assembler. This is the default.
10334 @c Also, this is the default if the configure option @option{--with-gnu-as}
10340 @opindex mno-gnu-ld
10341 Generate (or don't) code for the GNU linker. This is the default.
10342 @c Also, this is the default if the configure option @option{--with-gnu-ld}
10347 Generate code that does not use a global pointer register. The result
10348 is not position independent code, and violates the IA-64 ABI@.
10350 @item -mvolatile-asm-stop
10351 @itemx -mno-volatile-asm-stop
10352 @opindex mvolatile-asm-stop
10353 @opindex mno-volatile-asm-stop
10354 Generate (or don't) a stop bit immediately before and after volatile asm
10359 Generate code that works around Itanium B step errata.
10361 @item -mregister-names
10362 @itemx -mno-register-names
10363 @opindex mregister-names
10364 @opindex mno-register-names
10365 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
10366 the stacked registers. This may make assembler output more readable.
10372 Disable (or enable) optimizations that use the small data section. This may
10373 be useful for working around optimizer bugs.
10375 @item -mconstant-gp
10376 @opindex mconstant-gp
10377 Generate code that uses a single constant global pointer value. This is
10378 useful when compiling kernel code.
10382 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
10383 This is useful when compiling firmware code.
10385 @item -minline-float-divide-min-latency
10386 @opindex minline-float-divide-min-latency
10387 Generate code for inline divides of floating point values
10388 using the minimum latency algorithm.
10390 @item -minline-float-divide-max-throughput
10391 @opindex minline-float-divide-max-throughput
10392 Generate code for inline divides of floating point values
10393 using the maximum throughput algorithm.
10395 @item -minline-int-divide-min-latency
10396 @opindex minline-int-divide-min-latency
10397 Generate code for inline divides of integer values
10398 using the minimum latency algorithm.
10400 @item -minline-int-divide-max-throughput
10401 @opindex minline-int-divide-max-throughput
10402 Generate code for inline divides of integer values
10403 using the maximum throughput algorithm.
10405 @item -mno-dwarf2-asm
10406 @itemx -mdwarf2-asm
10407 @opindex mno-dwarf2-asm
10408 @opindex mdwarf2-asm
10409 Don't (or do) generate assembler code for the DWARF2 line number debugging
10410 info. This may be useful when not using the GNU assembler.
10412 @item -mfixed-range=@var{register-range}
10413 @opindex mfixed-range
10414 Generate code treating the given register range as fixed registers.
10415 A fixed register is one that the register allocator can not use. This is
10416 useful when compiling kernel code. A register range is specified as
10417 two registers separated by a dash. Multiple register ranges can be
10418 specified separated by a comma.
10420 @item -mearly-stop-bits
10421 @itemx -mno-early-stop-bits
10422 @opindex mearly-stop-bits
10423 @opindex mno-early-stop-bits
10424 Allow stop bits to be placed earlier than immediately preceding the
10425 instruction that triggered the stop bit. This can improve instruction
10426 scheduling, but does not always do so.
10429 @node S/390 and zSeries Options
10430 @subsection S/390 and zSeries Options
10431 @cindex S/390 and zSeries Options
10433 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
10437 @itemx -msoft-float
10438 @opindex mhard-float
10439 @opindex msoft-float
10440 Use (do not use) the hardware floating-point instructions and registers
10441 for floating-point operations. When @option{-msoft-float} is specified,
10442 functions in @file{libgcc.a} will be used to perform floating-point
10443 operations. When @option{-mhard-float} is specified, the compiler
10444 generates IEEE floating-point instructions. This is the default.
10447 @itemx -mno-backchain
10448 @opindex mbackchain
10449 @opindex mno-backchain
10450 Generate (or do not generate) code which maintains an explicit
10451 backchain within the stack frame that points to the caller's frame.
10452 This may be needed to allow debugging using tools that do not understand
10453 DWARF-2 call frame information. The default is not to generate the
10457 @itemx -mno-small-exec
10458 @opindex msmall-exec
10459 @opindex mno-small-exec
10460 Generate (or do not generate) code using the @code{bras} instruction
10461 to do subroutine calls.
10462 This only works reliably if the total executable size does not
10463 exceed 64k. The default is to use the @code{basr} instruction instead,
10464 which does not have this limitation.
10470 When @option{-m31} is specified, generate code compliant to the
10471 GNU/Linux for S/390 ABI@. When @option{-m64} is specified, generate
10472 code compliant to the GNU/Linux for zSeries ABI@. This allows GCC in
10473 particular to generate 64-bit instructions. For the @samp{s390}
10474 targets, the default is @option{-m31}, while the @samp{s390x}
10475 targets default to @option{-m64}.
10481 When @option{-mzarch} is specified, generate code using the
10482 instructions available on z/Architecture.
10483 When @option{-mesa} is specified, generate code using the
10484 instructions available on ESA/390. Note that @option{-mesa} is
10485 not possible with @option{-m64}.
10486 When generating code compliant to the GNU/Linux for S/390 ABI,
10487 the default is @option{-mesa}. When generating code compliant
10488 to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
10494 Generate (or do not generate) code using the @code{mvcle} instruction
10495 to perform block moves. When @option{-mno-mvcle} is specified,
10496 use a @code{mvc} loop instead. This is the default.
10502 Print (or do not print) additional debug information when compiling.
10503 The default is to not print debug information.
10505 @item -march=@var{cpu-type}
10507 Generate code that will run on @var{cpu-type}, which is the name of a system
10508 representing a certain processor type. Possible values for
10509 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, and @samp{z990}.
10510 When generating code using the instructions available on z/Architecture,
10511 the default is @option{-march=z900}. Otherwise, the default is
10512 @option{-march=g5}.
10514 @item -mtune=@var{cpu-type}
10516 Tune to @var{cpu-type} everything applicable about the generated code,
10517 except for the ABI and the set of available instructions.
10518 The list of @var{cpu-type} values is the same as for @option{-march}.
10519 The default is the value used for @option{-march}.
10522 @itemx -mno-fused-madd
10523 @opindex mfused-madd
10524 @opindex mno-fused-madd
10525 Generate code that uses (does not use) the floating point multiply and
10526 accumulate instructions. These instructions are generated by default if
10527 hardware floating point is used.
10531 @subsection CRIS Options
10532 @cindex CRIS Options
10534 These options are defined specifically for the CRIS ports.
10537 @item -march=@var{architecture-type}
10538 @itemx -mcpu=@var{architecture-type}
10541 Generate code for the specified architecture. The choices for
10542 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
10543 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX.
10544 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
10547 @item -mtune=@var{architecture-type}
10549 Tune to @var{architecture-type} everything applicable about the generated
10550 code, except for the ABI and the set of available instructions. The
10551 choices for @var{architecture-type} are the same as for
10552 @option{-march=@var{architecture-type}}.
10554 @item -mmax-stack-frame=@var{n}
10555 @opindex mmax-stack-frame
10556 Warn when the stack frame of a function exceeds @var{n} bytes.
10558 @item -melinux-stacksize=@var{n}
10559 @opindex melinux-stacksize
10560 Only available with the @samp{cris-axis-aout} target. Arranges for
10561 indications in the program to the kernel loader that the stack of the
10562 program should be set to @var{n} bytes.
10568 The options @option{-metrax4} and @option{-metrax100} are synonyms for
10569 @option{-march=v3} and @option{-march=v8} respectively.
10571 @item -mmul-bug-workaround
10572 @itemx -mno-mul-bug-workaround
10573 @opindex mmul-bug-workaround
10574 @opindex mno-mul-bug-workaround
10575 Work around a bug in the @code{muls} and @code{mulu} instructions for CPU
10576 models where it applies. This option is active by default.
10580 Enable CRIS-specific verbose debug-related information in the assembly
10581 code. This option also has the effect to turn off the @samp{#NO_APP}
10582 formatted-code indicator to the assembler at the beginning of the
10587 Do not use condition-code results from previous instruction; always emit
10588 compare and test instructions before use of condition codes.
10590 @item -mno-side-effects
10591 @opindex mno-side-effects
10592 Do not emit instructions with side-effects in addressing modes other than
10595 @item -mstack-align
10596 @itemx -mno-stack-align
10597 @itemx -mdata-align
10598 @itemx -mno-data-align
10599 @itemx -mconst-align
10600 @itemx -mno-const-align
10601 @opindex mstack-align
10602 @opindex mno-stack-align
10603 @opindex mdata-align
10604 @opindex mno-data-align
10605 @opindex mconst-align
10606 @opindex mno-const-align
10607 These options (no-options) arranges (eliminate arrangements) for the
10608 stack-frame, individual data and constants to be aligned for the maximum
10609 single data access size for the chosen CPU model. The default is to
10610 arrange for 32-bit alignment. ABI details such as structure layout are
10611 not affected by these options.
10619 Similar to the stack- data- and const-align options above, these options
10620 arrange for stack-frame, writable data and constants to all be 32-bit,
10621 16-bit or 8-bit aligned. The default is 32-bit alignment.
10623 @item -mno-prologue-epilogue
10624 @itemx -mprologue-epilogue
10625 @opindex mno-prologue-epilogue
10626 @opindex mprologue-epilogue
10627 With @option{-mno-prologue-epilogue}, the normal function prologue and
10628 epilogue that sets up the stack-frame are omitted and no return
10629 instructions or return sequences are generated in the code. Use this
10630 option only together with visual inspection of the compiled code: no
10631 warnings or errors are generated when call-saved registers must be saved,
10632 or storage for local variable needs to be allocated.
10636 @opindex mno-gotplt
10638 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
10639 instruction sequences that load addresses for functions from the PLT part
10640 of the GOT rather than (traditional on other architectures) calls to the
10641 PLT. The default is @option{-mgotplt}.
10645 Legacy no-op option only recognized with the cris-axis-aout target.
10649 Legacy no-op option only recognized with the cris-axis-elf and
10650 cris-axis-linux-gnu targets.
10654 Only recognized with the cris-axis-aout target, where it selects a
10655 GNU/linux-like multilib, include files and instruction set for
10656 @option{-march=v8}.
10660 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
10664 This option, recognized for the cris-axis-aout and cris-axis-elf arranges
10665 to link with input-output functions from a simulator library. Code,
10666 initialized data and zero-initialized data are allocated consecutively.
10670 Like @option{-sim}, but pass linker options to locate initialized data at
10671 0x40000000 and zero-initialized data at 0x80000000.
10675 @subsection MMIX Options
10676 @cindex MMIX Options
10678 These options are defined for the MMIX:
10682 @itemx -mno-libfuncs
10684 @opindex mno-libfuncs
10685 Specify that intrinsic library functions are being compiled, passing all
10686 values in registers, no matter the size.
10689 @itemx -mno-epsilon
10691 @opindex mno-epsilon
10692 Generate floating-point comparison instructions that compare with respect
10693 to the @code{rE} epsilon register.
10695 @item -mabi=mmixware
10697 @opindex mabi-mmixware
10699 Generate code that passes function parameters and return values that (in
10700 the called function) are seen as registers @code{$0} and up, as opposed to
10701 the GNU ABI which uses global registers @code{$231} and up.
10703 @item -mzero-extend
10704 @itemx -mno-zero-extend
10705 @opindex mzero-extend
10706 @opindex mno-zero-extend
10707 When reading data from memory in sizes shorter than 64 bits, use (do not
10708 use) zero-extending load instructions by default, rather than
10709 sign-extending ones.
10712 @itemx -mno-knuthdiv
10714 @opindex mno-knuthdiv
10715 Make the result of a division yielding a remainder have the same sign as
10716 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
10717 remainder follows the sign of the dividend. Both methods are
10718 arithmetically valid, the latter being almost exclusively used.
10720 @item -mtoplevel-symbols
10721 @itemx -mno-toplevel-symbols
10722 @opindex mtoplevel-symbols
10723 @opindex mno-toplevel-symbols
10724 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
10725 code can be used with the @code{PREFIX} assembly directive.
10729 Generate an executable in the ELF format, rather than the default
10730 @samp{mmo} format used by the @command{mmix} simulator.
10732 @item -mbranch-predict
10733 @itemx -mno-branch-predict
10734 @opindex mbranch-predict
10735 @opindex mno-branch-predict
10736 Use (do not use) the probable-branch instructions, when static branch
10737 prediction indicates a probable branch.
10739 @item -mbase-addresses
10740 @itemx -mno-base-addresses
10741 @opindex mbase-addresses
10742 @opindex mno-base-addresses
10743 Generate (do not generate) code that uses @emph{base addresses}. Using a
10744 base address automatically generates a request (handled by the assembler
10745 and the linker) for a constant to be set up in a global register. The
10746 register is used for one or more base address requests within the range 0
10747 to 255 from the value held in the register. The generally leads to short
10748 and fast code, but the number of different data items that can be
10749 addressed is limited. This means that a program that uses lots of static
10750 data may require @option{-mno-base-addresses}.
10752 @item -msingle-exit
10753 @itemx -mno-single-exit
10754 @opindex msingle-exit
10755 @opindex mno-single-exit
10756 Force (do not force) generated code to have a single exit point in each
10760 @node PDP-11 Options
10761 @subsection PDP-11 Options
10762 @cindex PDP-11 Options
10764 These options are defined for the PDP-11:
10769 Use hardware FPP floating point. This is the default. (FIS floating
10770 point on the PDP-11/40 is not supported.)
10773 @opindex msoft-float
10774 Do not use hardware floating point.
10778 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
10782 Return floating-point results in memory. This is the default.
10786 Generate code for a PDP-11/40.
10790 Generate code for a PDP-11/45. This is the default.
10794 Generate code for a PDP-11/10.
10796 @item -mbcopy-builtin
10797 @opindex bcopy-builtin
10798 Use inline @code{movstrhi} patterns for copying memory. This is the
10803 Do not use inline @code{movstrhi} patterns for copying memory.
10809 Use 16-bit @code{int}. This is the default.
10815 Use 32-bit @code{int}.
10818 @itemx -mno-float32
10820 @opindex mno-float32
10821 Use 64-bit @code{float}. This is the default.
10824 @itemx -mno-float64
10826 @opindex mno-float64
10827 Use 32-bit @code{float}.
10831 Use @code{abshi2} pattern. This is the default.
10835 Do not use @code{abshi2} pattern.
10837 @item -mbranch-expensive
10838 @opindex mbranch-expensive
10839 Pretend that branches are expensive. This is for experimenting with
10840 code generation only.
10842 @item -mbranch-cheap
10843 @opindex mbranch-cheap
10844 Do not pretend that branches are expensive. This is the default.
10848 Generate code for a system with split I&D.
10852 Generate code for a system without split I&D. This is the default.
10856 Use Unix assembler syntax. This is the default when configured for
10857 @samp{pdp11-*-bsd}.
10861 Use DEC assembler syntax. This is the default when configured for any
10862 PDP-11 target other than @samp{pdp11-*-bsd}.
10865 @node Xstormy16 Options
10866 @subsection Xstormy16 Options
10867 @cindex Xstormy16 Options
10869 These options are defined for Xstormy16:
10874 Choose startup files and linker script suitable for the simulator.
10878 @subsection FRV Options
10879 @cindex FRV Options
10885 Only use the first 32 general purpose registers.
10890 Use all 64 general purpose registers.
10895 Use only the first 32 floating point registers.
10900 Use all 64 floating point registers
10903 @opindex mhard-float
10905 Use hardware instructions for floating point operations.
10908 @opindex msoft-float
10910 Use library routines for floating point operations.
10915 Dynamically allocate condition code registers.
10920 Do not try to dynamically allocate condition code registers, only
10921 use @code{icc0} and @code{fcc0}.
10926 Change ABI to use double word insns.
10931 Do not use double word instructions.
10936 Use floating point double instructions.
10939 @opindex mno-double
10941 Do not use floating point double instructions.
10946 Use media instructions.
10951 Do not use media instructions.
10956 Use multiply and add/subtract instructions.
10959 @opindex mno-muladd
10961 Do not use multiply and add/subtract instructions.
10963 @item -mlibrary-pic
10964 @opindex mlibrary-pic
10966 Generate position-independent EABI code.
10971 Use only the first four media accumulator registers.
10976 Use all eight media accumulator registers.
10981 Pack VLIW instructions.
10986 Do not pack VLIW instructions.
10989 @opindex mno-eflags
10991 Do not mark ABI switches in e_flags.
10994 @opindex mcond-move
10996 Enable the use of conditional-move instructions (default).
10998 This switch is mainly for debugging the compiler and will likely be removed
10999 in a future version.
11001 @item -mno-cond-move
11002 @opindex mno-cond-move
11004 Disable the use of conditional-move instructions.
11006 This switch is mainly for debugging the compiler and will likely be removed
11007 in a future version.
11012 Enable the use of conditional set instructions (default).
11014 This switch is mainly for debugging the compiler and will likely be removed
11015 in a future version.
11020 Disable the use of conditional set instructions.
11022 This switch is mainly for debugging the compiler and will likely be removed
11023 in a future version.
11026 @opindex mcond-exec
11028 Enable the use of conditional execution (default).
11030 This switch is mainly for debugging the compiler and will likely be removed
11031 in a future version.
11033 @item -mno-cond-exec
11034 @opindex mno-cond-exec
11036 Disable the use of conditional execution.
11038 This switch is mainly for debugging the compiler and will likely be removed
11039 in a future version.
11041 @item -mvliw-branch
11042 @opindex mvliw-branch
11044 Run a pass to pack branches into VLIW instructions (default).
11046 This switch is mainly for debugging the compiler and will likely be removed
11047 in a future version.
11049 @item -mno-vliw-branch
11050 @opindex mno-vliw-branch
11052 Do not run a pass to pack branches into VLIW instructions.
11054 This switch is mainly for debugging the compiler and will likely be removed
11055 in a future version.
11057 @item -mmulti-cond-exec
11058 @opindex mmulti-cond-exec
11060 Enable optimization of @code{&&} and @code{||} in conditional execution
11063 This switch is mainly for debugging the compiler and will likely be removed
11064 in a future version.
11066 @item -mno-multi-cond-exec
11067 @opindex mno-multi-cond-exec
11069 Disable optimization of @code{&&} and @code{||} in conditional execution.
11071 This switch is mainly for debugging the compiler and will likely be removed
11072 in a future version.
11074 @item -mnested-cond-exec
11075 @opindex mnested-cond-exec
11077 Enable nested conditional execution optimizations (default).
11079 This switch is mainly for debugging the compiler and will likely be removed
11080 in a future version.
11082 @item -mno-nested-cond-exec
11083 @opindex mno-nested-cond-exec
11085 Disable nested conditional execution optimizations.
11087 This switch is mainly for debugging the compiler and will likely be removed
11088 in a future version.
11090 @item -mtomcat-stats
11091 @opindex mtomcat-stats
11093 Cause gas to print out tomcat statistics.
11095 @item -mcpu=@var{cpu}
11098 Select the processor type for which to generate code. Possible values are
11099 @samp{simple}, @samp{tomcat}, @samp{fr500}, @samp{fr400}, @samp{fr300},
11104 @node Xtensa Options
11105 @subsection Xtensa Options
11106 @cindex Xtensa Options
11108 These options are supported for Xtensa targets:
11112 @itemx -mno-const16
11114 @opindex mno-const16
11115 Enable or disable use of @code{CONST16} instructions for loading
11116 constant values. The @code{CONST16} instruction is currently not a
11117 standard option from Tensilica. When enabled, @code{CONST16}
11118 instructions are always used in place of the standard @code{L32R}
11119 instructions. The use of @code{CONST16} is enabled by default only if
11120 the @code{L32R} instruction is not available.
11123 @itemx -mno-fused-madd
11124 @opindex mfused-madd
11125 @opindex mno-fused-madd
11126 Enable or disable use of fused multiply/add and multiply/subtract
11127 instructions in the floating-point option. This has no effect if the
11128 floating-point option is not also enabled. Disabling fused multiply/add
11129 and multiply/subtract instructions forces the compiler to use separate
11130 instructions for the multiply and add/subtract operations. This may be
11131 desirable in some cases where strict IEEE 754-compliant results are
11132 required: the fused multiply add/subtract instructions do not round the
11133 intermediate result, thereby producing results with @emph{more} bits of
11134 precision than specified by the IEEE standard. Disabling fused multiply
11135 add/subtract instructions also ensures that the program output is not
11136 sensitive to the compiler's ability to combine multiply and add/subtract
11139 @item -mtext-section-literals
11140 @itemx -mno-text-section-literals
11141 @opindex mtext-section-literals
11142 @opindex mno-text-section-literals
11143 Control the treatment of literal pools. The default is
11144 @option{-mno-text-section-literals}, which places literals in a separate
11145 section in the output file. This allows the literal pool to be placed
11146 in a data RAM/ROM, and it also allows the linker to combine literal
11147 pools from separate object files to remove redundant literals and
11148 improve code size. With @option{-mtext-section-literals}, the literals
11149 are interspersed in the text section in order to keep them as close as
11150 possible to their references. This may be necessary for large assembly
11153 @item -mtarget-align
11154 @itemx -mno-target-align
11155 @opindex mtarget-align
11156 @opindex mno-target-align
11157 When this option is enabled, GCC instructs the assembler to
11158 automatically align instructions to reduce branch penalties at the
11159 expense of some code density. The assembler attempts to widen density
11160 instructions to align branch targets and the instructions following call
11161 instructions. If there are not enough preceding safe density
11162 instructions to align a target, no widening will be performed. The
11163 default is @option{-mtarget-align}. These options do not affect the
11164 treatment of auto-aligned instructions like @code{LOOP}, which the
11165 assembler will always align, either by widening density instructions or
11166 by inserting no-op instructions.
11169 @itemx -mno-longcalls
11170 @opindex mlongcalls
11171 @opindex mno-longcalls
11172 When this option is enabled, GCC instructs the assembler to translate
11173 direct calls to indirect calls unless it can determine that the target
11174 of a direct call is in the range allowed by the call instruction. This
11175 translation typically occurs for calls to functions in other source
11176 files. Specifically, the assembler translates a direct @code{CALL}
11177 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
11178 The default is @option{-mno-longcalls}. This option should be used in
11179 programs where the call target can potentially be out of range. This
11180 option is implemented in the assembler, not the compiler, so the
11181 assembly code generated by GCC will still show direct call
11182 instructions---look at the disassembled object code to see the actual
11183 instructions. Note that the assembler will use an indirect call for
11184 every cross-file call, not just those that really will be out of range.
11187 @node Code Gen Options
11188 @section Options for Code Generation Conventions
11189 @cindex code generation conventions
11190 @cindex options, code generation
11191 @cindex run-time options
11193 These machine-independent options control the interface conventions
11194 used in code generation.
11196 Most of them have both positive and negative forms; the negative form
11197 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
11198 one of the forms is listed---the one which is not the default. You
11199 can figure out the other form by either removing @samp{no-} or adding
11203 @item -fbounds-check
11204 @opindex fbounds-check
11205 For front-ends that support it, generate additional code to check that
11206 indices used to access arrays are within the declared range. This is
11207 currently only supported by the Java and Fortran 77 front-ends, where
11208 this option defaults to true and false respectively.
11212 This option generates traps for signed overflow on addition, subtraction,
11213 multiplication operations.
11217 This option instructs the compiler to assume that signed arithmetic
11218 overflow of addition, subtraction and multiplication wraps around
11219 using twos-complement representation. This flag enables some optimizations
11220 and disables other. This option is enabled by default for the Java
11221 front-end, as required by the Java language specification.
11224 @opindex fexceptions
11225 Enable exception handling. Generates extra code needed to propagate
11226 exceptions. For some targets, this implies GCC will generate frame
11227 unwind information for all functions, which can produce significant data
11228 size overhead, although it does not affect execution. If you do not
11229 specify this option, GCC will enable it by default for languages like
11230 C++ which normally require exception handling, and disable it for
11231 languages like C that do not normally require it. However, you may need
11232 to enable this option when compiling C code that needs to interoperate
11233 properly with exception handlers written in C++. You may also wish to
11234 disable this option if you are compiling older C++ programs that don't
11235 use exception handling.
11237 @item -fnon-call-exceptions
11238 @opindex fnon-call-exceptions
11239 Generate code that allows trapping instructions to throw exceptions.
11240 Note that this requires platform-specific runtime support that does
11241 not exist everywhere. Moreover, it only allows @emph{trapping}
11242 instructions to throw exceptions, i.e.@: memory references or floating
11243 point instructions. It does not allow exceptions to be thrown from
11244 arbitrary signal handlers such as @code{SIGALRM}.
11246 @item -funwind-tables
11247 @opindex funwind-tables
11248 Similar to @option{-fexceptions}, except that it will just generate any needed
11249 static data, but will not affect the generated code in any other way.
11250 You will normally not enable this option; instead, a language processor
11251 that needs this handling would enable it on your behalf.
11253 @item -fasynchronous-unwind-tables
11254 @opindex funwind-tables
11255 Generate unwind table in dwarf2 format, if supported by target machine. The
11256 table is exact at each instruction boundary, so it can be used for stack
11257 unwinding from asynchronous events (such as debugger or garbage collector).
11259 @item -fpcc-struct-return
11260 @opindex fpcc-struct-return
11261 Return ``short'' @code{struct} and @code{union} values in memory like
11262 longer ones, rather than in registers. This convention is less
11263 efficient, but it has the advantage of allowing intercallability between
11264 GCC-compiled files and files compiled with other compilers, particularly
11265 the Portable C Compiler (pcc).
11267 The precise convention for returning structures in memory depends
11268 on the target configuration macros.
11270 Short structures and unions are those whose size and alignment match
11271 that of some integer type.
11273 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
11274 switch is not binary compatible with code compiled with the
11275 @option{-freg-struct-return} switch.
11276 Use it to conform to a non-default application binary interface.
11278 @item -freg-struct-return
11279 @opindex freg-struct-return
11280 Return @code{struct} and @code{union} values in registers when possible.
11281 This is more efficient for small structures than
11282 @option{-fpcc-struct-return}.
11284 If you specify neither @option{-fpcc-struct-return} nor
11285 @option{-freg-struct-return}, GCC defaults to whichever convention is
11286 standard for the target. If there is no standard convention, GCC
11287 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
11288 the principal compiler. In those cases, we can choose the standard, and
11289 we chose the more efficient register return alternative.
11291 @strong{Warning:} code compiled with the @option{-freg-struct-return}
11292 switch is not binary compatible with code compiled with the
11293 @option{-fpcc-struct-return} switch.
11294 Use it to conform to a non-default application binary interface.
11296 @item -fshort-enums
11297 @opindex fshort-enums
11298 Allocate to an @code{enum} type only as many bytes as it needs for the
11299 declared range of possible values. Specifically, the @code{enum} type
11300 will be equivalent to the smallest integer type which has enough room.
11302 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
11303 code that is not binary compatible with code generated without that switch.
11304 Use it to conform to a non-default application binary interface.
11306 @item -fshort-double
11307 @opindex fshort-double
11308 Use the same size for @code{double} as for @code{float}.
11310 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
11311 code that is not binary compatible with code generated without that switch.
11312 Use it to conform to a non-default application binary interface.
11314 @item -fshort-wchar
11315 @opindex fshort-wchar
11316 Override the underlying type for @samp{wchar_t} to be @samp{short
11317 unsigned int} instead of the default for the target. This option is
11318 useful for building programs to run under WINE@.
11320 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
11321 code that is not binary compatible with code generated without that switch.
11322 Use it to conform to a non-default application binary interface.
11324 @item -fshared-data
11325 @opindex fshared-data
11326 Requests that the data and non-@code{const} variables of this
11327 compilation be shared data rather than private data. The distinction
11328 makes sense only on certain operating systems, where shared data is
11329 shared between processes running the same program, while private data
11330 exists in one copy per process.
11333 @opindex fno-common
11334 In C, allocate even uninitialized global variables in the data section of the
11335 object file, rather than generating them as common blocks. This has the
11336 effect that if the same variable is declared (without @code{extern}) in
11337 two different compilations, you will get an error when you link them.
11338 The only reason this might be useful is if you wish to verify that the
11339 program will work on other systems which always work this way.
11343 Ignore the @samp{#ident} directive.
11345 @item -finhibit-size-directive
11346 @opindex finhibit-size-directive
11347 Don't output a @code{.size} assembler directive, or anything else that
11348 would cause trouble if the function is split in the middle, and the
11349 two halves are placed at locations far apart in memory. This option is
11350 used when compiling @file{crtstuff.c}; you should not need to use it
11353 @item -fverbose-asm
11354 @opindex fverbose-asm
11355 Put extra commentary information in the generated assembly code to
11356 make it more readable. This option is generally only of use to those
11357 who actually need to read the generated assembly code (perhaps while
11358 debugging the compiler itself).
11360 @option{-fno-verbose-asm}, the default, causes the
11361 extra information to be omitted and is useful when comparing two assembler
11366 @cindex global offset table
11368 Generate position-independent code (PIC) suitable for use in a shared
11369 library, if supported for the target machine. Such code accesses all
11370 constant addresses through a global offset table (GOT)@. The dynamic
11371 loader resolves the GOT entries when the program starts (the dynamic
11372 loader is not part of GCC; it is part of the operating system). If
11373 the GOT size for the linked executable exceeds a machine-specific
11374 maximum size, you get an error message from the linker indicating that
11375 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
11376 instead. (These maximums are 8k on the SPARC and 32k
11377 on the m68k and RS/6000. The 386 has no such limit.)
11379 Position-independent code requires special support, and therefore works
11380 only on certain machines. For the 386, GCC supports PIC for System V
11381 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
11382 position-independent.
11386 If supported for the target machine, emit position-independent code,
11387 suitable for dynamic linking and avoiding any limit on the size of the
11388 global offset table. This option makes a difference on the m68k
11391 Position-independent code requires special support, and therefore works
11392 only on certain machines.
11398 These options are similar to @option{-fpic} and @option{-fPIC}, but
11399 generated position independent code can be only linked into executables.
11400 Usually these options are used when @option{-pie} GCC option will be
11401 used during linking.
11403 @item -ffixed-@var{reg}
11405 Treat the register named @var{reg} as a fixed register; generated code
11406 should never refer to it (except perhaps as a stack pointer, frame
11407 pointer or in some other fixed role).
11409 @var{reg} must be the name of a register. The register names accepted
11410 are machine-specific and are defined in the @code{REGISTER_NAMES}
11411 macro in the machine description macro file.
11413 This flag does not have a negative form, because it specifies a
11416 @item -fcall-used-@var{reg}
11417 @opindex fcall-used
11418 Treat the register named @var{reg} as an allocable register that is
11419 clobbered by function calls. It may be allocated for temporaries or
11420 variables that do not live across a call. Functions compiled this way
11421 will not save and restore the register @var{reg}.
11423 It is an error to used this flag with the frame pointer or stack pointer.
11424 Use of this flag for other registers that have fixed pervasive roles in
11425 the machine's execution model will produce disastrous results.
11427 This flag does not have a negative form, because it specifies a
11430 @item -fcall-saved-@var{reg}
11431 @opindex fcall-saved
11432 Treat the register named @var{reg} as an allocable register saved by
11433 functions. It may be allocated even for temporaries or variables that
11434 live across a call. Functions compiled this way will save and restore
11435 the register @var{reg} if they use it.
11437 It is an error to used this flag with the frame pointer or stack pointer.
11438 Use of this flag for other registers that have fixed pervasive roles in
11439 the machine's execution model will produce disastrous results.
11441 A different sort of disaster will result from the use of this flag for
11442 a register in which function values may be returned.
11444 This flag does not have a negative form, because it specifies a
11447 @item -fpack-struct
11448 @opindex fpack-struct
11449 Pack all structure members together without holes.
11451 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
11452 code that is not binary compatible with code generated without that switch.
11453 Additionally, it makes the code suboptimal.
11454 Use it to conform to a non-default application binary interface.
11456 @item -finstrument-functions
11457 @opindex finstrument-functions
11458 Generate instrumentation calls for entry and exit to functions. Just
11459 after function entry and just before function exit, the following
11460 profiling functions will be called with the address of the current
11461 function and its call site. (On some platforms,
11462 @code{__builtin_return_address} does not work beyond the current
11463 function, so the call site information may not be available to the
11464 profiling functions otherwise.)
11467 void __cyg_profile_func_enter (void *this_fn,
11469 void __cyg_profile_func_exit (void *this_fn,
11473 The first argument is the address of the start of the current function,
11474 which may be looked up exactly in the symbol table.
11476 This instrumentation is also done for functions expanded inline in other
11477 functions. The profiling calls will indicate where, conceptually, the
11478 inline function is entered and exited. This means that addressable
11479 versions of such functions must be available. If all your uses of a
11480 function are expanded inline, this may mean an additional expansion of
11481 code size. If you use @samp{extern inline} in your C code, an
11482 addressable version of such functions must be provided. (This is
11483 normally the case anyways, but if you get lucky and the optimizer always
11484 expands the functions inline, you might have gotten away without
11485 providing static copies.)
11487 A function may be given the attribute @code{no_instrument_function}, in
11488 which case this instrumentation will not be done. This can be used, for
11489 example, for the profiling functions listed above, high-priority
11490 interrupt routines, and any functions from which the profiling functions
11491 cannot safely be called (perhaps signal handlers, if the profiling
11492 routines generate output or allocate memory).
11494 @item -fstack-check
11495 @opindex fstack-check
11496 Generate code to verify that you do not go beyond the boundary of the
11497 stack. You should specify this flag if you are running in an
11498 environment with multiple threads, but only rarely need to specify it in
11499 a single-threaded environment since stack overflow is automatically
11500 detected on nearly all systems if there is only one stack.
11502 Note that this switch does not actually cause checking to be done; the
11503 operating system must do that. The switch causes generation of code
11504 to ensure that the operating system sees the stack being extended.
11506 @item -fstack-limit-register=@var{reg}
11507 @itemx -fstack-limit-symbol=@var{sym}
11508 @itemx -fno-stack-limit
11509 @opindex fstack-limit-register
11510 @opindex fstack-limit-symbol
11511 @opindex fno-stack-limit
11512 Generate code to ensure that the stack does not grow beyond a certain value,
11513 either the value of a register or the address of a symbol. If the stack
11514 would grow beyond the value, a signal is raised. For most targets,
11515 the signal is raised before the stack overruns the boundary, so
11516 it is possible to catch the signal without taking special precautions.
11518 For instance, if the stack starts at absolute address @samp{0x80000000}
11519 and grows downwards, you can use the flags
11520 @option{-fstack-limit-symbol=__stack_limit} and
11521 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
11522 of 128KB@. Note that this may only work with the GNU linker.
11524 @cindex aliasing of parameters
11525 @cindex parameters, aliased
11526 @item -fargument-alias
11527 @itemx -fargument-noalias
11528 @itemx -fargument-noalias-global
11529 @opindex fargument-alias
11530 @opindex fargument-noalias
11531 @opindex fargument-noalias-global
11532 Specify the possible relationships among parameters and between
11533 parameters and global data.
11535 @option{-fargument-alias} specifies that arguments (parameters) may
11536 alias each other and may alias global storage.@*
11537 @option{-fargument-noalias} specifies that arguments do not alias
11538 each other, but may alias global storage.@*
11539 @option{-fargument-noalias-global} specifies that arguments do not
11540 alias each other and do not alias global storage.
11542 Each language will automatically use whatever option is required by
11543 the language standard. You should not need to use these options yourself.
11545 @item -fleading-underscore
11546 @opindex fleading-underscore
11547 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
11548 change the way C symbols are represented in the object file. One use
11549 is to help link with legacy assembly code.
11551 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
11552 generate code that is not binary compatible with code generated without that
11553 switch. Use it to conform to a non-default application binary interface.
11554 Not all targets provide complete support for this switch.
11556 @item -ftls-model=@var{model}
11557 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
11558 The @var{model} argument should be one of @code{global-dynamic},
11559 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
11561 The default without @option{-fpic} is @code{initial-exec}; with
11562 @option{-fpic} the default is @code{global-dynamic}.
11567 @node Environment Variables
11568 @section Environment Variables Affecting GCC
11569 @cindex environment variables
11571 @c man begin ENVIRONMENT
11572 This section describes several environment variables that affect how GCC
11573 operates. Some of them work by specifying directories or prefixes to use
11574 when searching for various kinds of files. Some are used to specify other
11575 aspects of the compilation environment.
11577 Note that you can also specify places to search using options such as
11578 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
11579 take precedence over places specified using environment variables, which
11580 in turn take precedence over those specified by the configuration of GCC@.
11581 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
11582 GNU Compiler Collection (GCC) Internals}.
11587 @c @itemx LC_COLLATE
11589 @c @itemx LC_MONETARY
11590 @c @itemx LC_NUMERIC
11595 @c @findex LC_COLLATE
11596 @findex LC_MESSAGES
11597 @c @findex LC_MONETARY
11598 @c @findex LC_NUMERIC
11602 These environment variables control the way that GCC uses
11603 localization information that allow GCC to work with different
11604 national conventions. GCC inspects the locale categories
11605 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
11606 so. These locale categories can be set to any value supported by your
11607 installation. A typical value is @samp{en_UK} for English in the United
11610 The @env{LC_CTYPE} environment variable specifies character
11611 classification. GCC uses it to determine the character boundaries in
11612 a string; this is needed for some multibyte encodings that contain quote
11613 and escape characters that would otherwise be interpreted as a string
11616 The @env{LC_MESSAGES} environment variable specifies the language to
11617 use in diagnostic messages.
11619 If the @env{LC_ALL} environment variable is set, it overrides the value
11620 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
11621 and @env{LC_MESSAGES} default to the value of the @env{LANG}
11622 environment variable. If none of these variables are set, GCC
11623 defaults to traditional C English behavior.
11627 If @env{TMPDIR} is set, it specifies the directory to use for temporary
11628 files. GCC uses temporary files to hold the output of one stage of
11629 compilation which is to be used as input to the next stage: for example,
11630 the output of the preprocessor, which is the input to the compiler
11633 @item GCC_EXEC_PREFIX
11634 @findex GCC_EXEC_PREFIX
11635 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
11636 names of the subprograms executed by the compiler. No slash is added
11637 when this prefix is combined with the name of a subprogram, but you can
11638 specify a prefix that ends with a slash if you wish.
11640 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
11641 an appropriate prefix to use based on the pathname it was invoked with.
11643 If GCC cannot find the subprogram using the specified prefix, it
11644 tries looking in the usual places for the subprogram.
11646 The default value of @env{GCC_EXEC_PREFIX} is
11647 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the value
11648 of @code{prefix} when you ran the @file{configure} script.
11650 Other prefixes specified with @option{-B} take precedence over this prefix.
11652 This prefix is also used for finding files such as @file{crt0.o} that are
11655 In addition, the prefix is used in an unusual way in finding the
11656 directories to search for header files. For each of the standard
11657 directories whose name normally begins with @samp{/usr/local/lib/gcc}
11658 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
11659 replacing that beginning with the specified prefix to produce an
11660 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
11661 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
11662 These alternate directories are searched first; the standard directories
11665 @item COMPILER_PATH
11666 @findex COMPILER_PATH
11667 The value of @env{COMPILER_PATH} is a colon-separated list of
11668 directories, much like @env{PATH}. GCC tries the directories thus
11669 specified when searching for subprograms, if it can't find the
11670 subprograms using @env{GCC_EXEC_PREFIX}.
11673 @findex LIBRARY_PATH
11674 The value of @env{LIBRARY_PATH} is a colon-separated list of
11675 directories, much like @env{PATH}. When configured as a native compiler,
11676 GCC tries the directories thus specified when searching for special
11677 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
11678 using GCC also uses these directories when searching for ordinary
11679 libraries for the @option{-l} option (but directories specified with
11680 @option{-L} come first).
11684 @cindex locale definition
11685 This variable is used to pass locale information to the compiler. One way in
11686 which this information is used is to determine the character set to be used
11687 when character literals, string literals and comments are parsed in C and C++.
11688 When the compiler is configured to allow multibyte characters,
11689 the following values for @env{LANG} are recognized:
11693 Recognize JIS characters.
11695 Recognize SJIS characters.
11697 Recognize EUCJP characters.
11700 If @env{LANG} is not defined, or if it has some other value, then the
11701 compiler will use mblen and mbtowc as defined by the default locale to
11702 recognize and translate multibyte characters.
11706 Some additional environments variables affect the behavior of the
11709 @include cppenv.texi
11713 @node Precompiled Headers
11714 @section Using Precompiled Headers
11715 @cindex precompiled headers
11716 @cindex speed of compilation
11718 Often large projects have many header files that are included in every
11719 source file. The time the compiler takes to process these header files
11720 over and over again can account for nearly all of the time required to
11721 build the project. To make builds faster, GCC allows users to
11722 `precompile' a header file; then, if builds can use the precompiled
11723 header file they will be much faster.
11725 @strong{Caution:} There are a few known situations where GCC will
11726 crash when trying to use a precompiled header. If you have trouble
11727 with a precompiled header, you should remove the precompiled header
11728 and compile without it. In addition, please use GCC's on-line
11729 defect-tracking system to report any problems you encounter with
11730 precompiled headers. @xref{Bugs}.
11732 To create a precompiled header file, simply compile it as you would any
11733 other file, if necessary using the @option{-x} option to make the driver
11734 treat it as a C or C++ header file. You will probably want to use a
11735 tool like @command{make} to keep the precompiled header up-to-date when
11736 the headers it contains change.
11738 A precompiled header file will be searched for when @code{#include} is
11739 seen in the compilation. As it searches for the included file
11740 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
11741 compiler looks for a precompiled header in each directory just before it
11742 looks for the include file in that directory. The name searched for is
11743 the name specified in the @code{#include} with @samp{.gch} appended. If
11744 the precompiled header file can't be used, it is ignored.
11746 For instance, if you have @code{#include "all.h"}, and you have
11747 @file{all.h.gch} in the same directory as @file{all.h}, then the
11748 precompiled header file will be used if possible, and the original
11749 header will be used otherwise.
11751 Alternatively, you might decide to put the precompiled header file in a
11752 directory and use @option{-I} to ensure that directory is searched
11753 before (or instead of) the directory containing the original header.
11754 Then, if you want to check that the precompiled header file is always
11755 used, you can put a file of the same name as the original header in this
11756 directory containing an @code{#error} command.
11758 This also works with @option{-include}. So yet another way to use
11759 precompiled headers, good for projects not designed with precompiled
11760 header files in mind, is to simply take most of the header files used by
11761 a project, include them from another header file, precompile that header
11762 file, and @option{-include} the precompiled header. If the header files
11763 have guards against multiple inclusion, they will be skipped because
11764 they've already been included (in the precompiled header).
11766 If you need to precompile the same header file for different
11767 languages, targets, or compiler options, you can instead make a
11768 @emph{directory} named like @file{all.h.gch}, and put each precompiled
11769 header in the directory, perhaps using @option{-o}. It doesn't matter
11770 what you call the files in the directory, every precompiled header in
11771 the directory will be considered. The first precompiled header
11772 encountered in the directory that is valid for this compilation will
11773 be used; they're searched in no particular order.
11775 There are many other possibilities, limited only by your imagination,
11776 good sense, and the constraints of your build system.
11778 A precompiled header file can be used only when these conditions apply:
11782 Only one precompiled header can be used in a particular compilation.
11785 A precompiled header can't be used once the first C token is seen. You
11786 can have preprocessor directives before a precompiled header; you can
11787 even include a precompiled header from inside another header, so long as
11788 there are no C tokens before the @code{#include}.
11791 The precompiled header file must be produced for the same language as
11792 the current compilation. You can't use a C precompiled header for a C++
11796 The precompiled header file must be produced by the same compiler
11797 version and configuration as the current compilation is using.
11798 The easiest way to guarantee this is to use the same compiler binary
11799 for creating and using precompiled headers.
11802 Any macros defined before the precompiled header is included must
11803 either be defined in the same way as when the precompiled header was
11804 generated, or must not affect the precompiled header, which usually
11805 means that the they don't appear in the precompiled header at all.
11807 The @option{-D} option is one way to define a macro before a
11808 precompiled header is included; using a @code{#define} can also do it.
11809 There are also some options that define macros implicitly, like
11810 @option{-O} and @option{-Wdeprecated}; the same rule applies to macros
11813 @item If debugging information is output when using the precompiled
11814 header, using @option{-g} or similar, the same kind of debugging information
11815 must have been output when building the precompiled header. However,
11816 a precompiled header built using @option{-g} can be used in a compilation
11817 when no debugging information is being output.
11819 @item The same @option{-m} options must generally be used when building
11820 and using the precompiled header. @xref{Submodel Options},
11821 for any cases where this rule is relaxed.
11823 @item Each of the following options must be the same when building and using
11824 the precompiled header:
11826 @gccoptlist{-fexceptions -funit-at-a-time}
11829 Some other command-line options starting with @option{-f},
11830 @option{-p}, or @option{-O} must be defined in the same way as when
11831 the precompiled header was generated. At present, it's not clear
11832 which options are safe to change and which are not; the safest choice
11833 is to use exactly the same options when generating and using the
11834 precompiled header. The following are known to be safe:
11836 @gccoptlist{-pedantic-errors}
11840 For all of these except the last, the compiler will automatically
11841 ignore the precompiled header if the conditions aren't met. If you
11842 find an option combination that doesn't work and doesn't cause the
11843 precompiled header to be ignored, please consider filing a bug report,
11846 @node Running Protoize
11847 @section Running Protoize
11849 The program @code{protoize} is an optional part of GCC@. You can use
11850 it to add prototypes to a program, thus converting the program to ISO
11851 C in one respect. The companion program @code{unprotoize} does the
11852 reverse: it removes argument types from any prototypes that are found.
11854 When you run these programs, you must specify a set of source files as
11855 command line arguments. The conversion programs start out by compiling
11856 these files to see what functions they define. The information gathered
11857 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
11859 After scanning comes actual conversion. The specified files are all
11860 eligible to be converted; any files they include (whether sources or
11861 just headers) are eligible as well.
11863 But not all the eligible files are converted. By default,
11864 @code{protoize} and @code{unprotoize} convert only source and header
11865 files in the current directory. You can specify additional directories
11866 whose files should be converted with the @option{-d @var{directory}}
11867 option. You can also specify particular files to exclude with the
11868 @option{-x @var{file}} option. A file is converted if it is eligible, its
11869 directory name matches one of the specified directory names, and its
11870 name within the directory has not been excluded.
11872 Basic conversion with @code{protoize} consists of rewriting most
11873 function definitions and function declarations to specify the types of
11874 the arguments. The only ones not rewritten are those for varargs
11877 @code{protoize} optionally inserts prototype declarations at the
11878 beginning of the source file, to make them available for any calls that
11879 precede the function's definition. Or it can insert prototype
11880 declarations with block scope in the blocks where undeclared functions
11883 Basic conversion with @code{unprotoize} consists of rewriting most
11884 function declarations to remove any argument types, and rewriting
11885 function definitions to the old-style pre-ISO form.
11887 Both conversion programs print a warning for any function declaration or
11888 definition that they can't convert. You can suppress these warnings
11891 The output from @code{protoize} or @code{unprotoize} replaces the
11892 original source file. The original file is renamed to a name ending
11893 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
11894 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
11895 for DOS) file already exists, then the source file is simply discarded.
11897 @code{protoize} and @code{unprotoize} both depend on GCC itself to
11898 scan the program and collect information about the functions it uses.
11899 So neither of these programs will work until GCC is installed.
11901 Here is a table of the options you can use with @code{protoize} and
11902 @code{unprotoize}. Each option works with both programs unless
11906 @item -B @var{directory}
11907 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
11908 usual directory (normally @file{/usr/local/lib}). This file contains
11909 prototype information about standard system functions. This option
11910 applies only to @code{protoize}.
11912 @item -c @var{compilation-options}
11913 Use @var{compilation-options} as the options when running @command{gcc} to
11914 produce the @samp{.X} files. The special option @option{-aux-info} is
11915 always passed in addition, to tell @command{gcc} to write a @samp{.X} file.
11917 Note that the compilation options must be given as a single argument to
11918 @code{protoize} or @code{unprotoize}. If you want to specify several
11919 @command{gcc} options, you must quote the entire set of compilation options
11920 to make them a single word in the shell.
11922 There are certain @command{gcc} arguments that you cannot use, because they
11923 would produce the wrong kind of output. These include @option{-g},
11924 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
11925 the @var{compilation-options}, they are ignored.
11928 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
11929 systems) instead of @samp{.c}. This is convenient if you are converting
11930 a C program to C++. This option applies only to @code{protoize}.
11933 Add explicit global declarations. This means inserting explicit
11934 declarations at the beginning of each source file for each function
11935 that is called in the file and was not declared. These declarations
11936 precede the first function definition that contains a call to an
11937 undeclared function. This option applies only to @code{protoize}.
11939 @item -i @var{string}
11940 Indent old-style parameter declarations with the string @var{string}.
11941 This option applies only to @code{protoize}.
11943 @code{unprotoize} converts prototyped function definitions to old-style
11944 function definitions, where the arguments are declared between the
11945 argument list and the initial @samp{@{}. By default, @code{unprotoize}
11946 uses five spaces as the indentation. If you want to indent with just
11947 one space instead, use @option{-i " "}.
11950 Keep the @samp{.X} files. Normally, they are deleted after conversion
11954 Add explicit local declarations. @code{protoize} with @option{-l} inserts
11955 a prototype declaration for each function in each block which calls the
11956 function without any declaration. This option applies only to
11960 Make no real changes. This mode just prints information about the conversions
11961 that would have been done without @option{-n}.
11964 Make no @samp{.save} files. The original files are simply deleted.
11965 Use this option with caution.
11967 @item -p @var{program}
11968 Use the program @var{program} as the compiler. Normally, the name
11969 @file{gcc} is used.
11972 Work quietly. Most warnings are suppressed.
11975 Print the version number, just like @option{-v} for @command{gcc}.
11978 If you need special compiler options to compile one of your program's
11979 source files, then you should generate that file's @samp{.X} file
11980 specially, by running @command{gcc} on that source file with the
11981 appropriate options and the option @option{-aux-info}. Then run
11982 @code{protoize} on the entire set of files. @code{protoize} will use
11983 the existing @samp{.X} file because it is newer than the source file.
11987 gcc -Dfoo=bar file1.c -aux-info file1.X
11992 You need to include the special files along with the rest in the
11993 @code{protoize} command, even though their @samp{.X} files already
11994 exist, because otherwise they won't get converted.
11996 @xref{Protoize Caveats}, for more information on how to use
11997 @code{protoize} successfully.