1 @c Copyright (C) 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
2 @c 2000, 2001, 2002 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 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 1.1 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
70 When you invoke GCC, it normally does preprocessing, compilation,
71 assembly and linking. The ``overall options'' allow you to stop this
72 process at an intermediate stage. For example, the @option{-c} option
73 says not to run the linker. Then the output consists of object files
74 output by the assembler.
76 Other options are passed on to one stage of processing. Some options
77 control the preprocessor and others the compiler itself. Yet other
78 options control the assembler and linker; most of these are not
79 documented here, since you rarely need to use any of them.
81 @cindex C compilation options
82 Most of the command line options that you can use with GCC are useful
83 for C programs; when an option is only useful with another language
84 (usually C++), the explanation says so explicitly. If the description
85 for a particular option does not mention a source language, you can use
86 that option with all supported languages.
88 @cindex C++ compilation options
89 @xref{Invoking G++,,Compiling C++ Programs}, for a summary of special
90 options for compiling C++ programs.
92 @cindex grouping options
93 @cindex options, grouping
94 The @command{gcc} program accepts options and file names as operands. Many
95 options have multi-letter names; therefore multiple single-letter options
96 may @emph{not} be grouped: @option{-dr} is very different from @w{@samp{-d
99 @cindex order of options
100 @cindex options, order
101 You can mix options and other arguments. For the most part, the order
102 you use doesn't matter. Order does matter when you use several options
103 of the same kind; for example, if you specify @option{-L} more than once,
104 the directories are searched in the order specified.
106 Many options have long names starting with @samp{-f} or with
107 @samp{-W}---for example, @option{-fforce-mem},
108 @option{-fstrength-reduce}, @option{-Wformat} and so on. Most of
109 these have both positive and negative forms; the negative form of
110 @option{-ffoo} would be @option{-fno-foo}. This manual documents
111 only one of these two forms, whichever one is not the default.
115 @xref{Option Index}, for an index to GCC's options.
118 * Option Summary:: Brief list of all options, without explanations.
119 * Overall Options:: Controlling the kind of output:
120 an executable, object files, assembler files,
121 or preprocessed source.
122 * Invoking G++:: Compiling C++ programs.
123 * C Dialect Options:: Controlling the variant of C language compiled.
124 * C++ Dialect Options:: Variations on C++.
125 * Objective-C Dialect Options:: Variations on Objective-C.
126 * Language Independent Options:: Controlling how diagnostics should be
128 * Warning Options:: How picky should the compiler be?
129 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
130 * Optimize Options:: How much optimization?
131 * Preprocessor Options:: Controlling header files and macro definitions.
132 Also, getting dependency information for Make.
133 * Assembler Options:: Passing options to the assembler.
134 * Link Options:: Specifying libraries and so on.
135 * Directory Options:: Where to find header files and libraries.
136 Where to find the compiler executable files.
137 * Spec Files:: How to pass switches to sub-processes.
138 * Target Options:: Running a cross-compiler, or an old version of GCC.
139 * Submodel Options:: Specifying minor hardware or convention variations,
140 such as 68010 vs 68020.
141 * Code Gen Options:: Specifying conventions for function calls, data layout
143 * Environment Variables:: Env vars that affect GCC.
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}.
159 -c -S -E -o @var{file} -pipe -pass-exit-codes -x @var{language} @gol
160 -v --target-help --help}
162 @item C Language Options
163 @xref{C Dialect Options,,Options Controlling C Dialect}.
165 -ansi -std=@var{standard} -aux-info @var{filename} @gol
166 -fno-asm -fno-builtin -fno-builtin-@var{function} @gol
167 -fhosted -ffreestanding @gol
168 -trigraphs -traditional -traditional-cpp @gol
169 -fallow-single-precision -fcond-mismatch @gol
170 -fsigned-bitfields -fsigned-char @gol
171 -funsigned-bitfields -funsigned-char @gol
172 -fwritable-strings -fshort-wchar}
174 @item C++ Language Options
175 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
177 -fno-access-control -fcheck-new -fconserve-space @gol
178 -fno-const-strings -fdollars-in-identifiers @gol
179 -fno-elide-constructors @gol
180 -fno-enforce-eh-specs -fexternal-templates @gol
181 -falt-external-templates @gol
182 -ffor-scope -fno-for-scope -fno-gnu-keywords @gol
183 -fno-implicit-templates @gol
184 -fno-implicit-inline-templates @gol
185 -fno-implement-inlines -fms-extensions @gol
186 -fno-nonansi-builtins -fno-operator-names @gol
187 -fno-optional-diags -fpermissive @gol
188 -frepo -fno-rtti -fstats -ftemplate-depth-@var{n} @gol
189 -fuse-cxa-atexit -fvtable-gc -fno-weak -nostdinc++ @gol
190 -fno-default-inline -Wctor-dtor-privacy @gol
191 -Wnon-virtual-dtor -Wreorder @gol
192 -Weffc++ -Wno-deprecated @gol
193 -Wno-non-template-friend -Wold-style-cast @gol
194 -Woverloaded-virtual -Wno-pmf-conversions @gol
195 -Wsign-promo -Wsynth}
197 @item Objective-C Language Options
198 @xref{Objective-C Dialect Options,,Options Controlling Objective-C Dialect}.
200 -fconstant-string-class=@var{class-name} @gol
201 -fgnu-runtime -fnext-runtime -gen-decls @gol
202 -Wno-protocol -Wselector}
204 @item Language Independent Options
205 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
207 -fmessage-length=@var{n} @gol
208 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]}}
210 @item Warning Options
211 @xref{Warning Options,,Options to Request or Suppress Warnings}.
213 -fsyntax-only -pedantic -pedantic-errors @gol
214 -w -W -Wall -Waggregate-return @gol
215 -Wcast-align -Wcast-qual -Wchar-subscripts -Wcomment @gol
216 -Wconversion -Wno-deprecated-declarations @gol
217 -Wdisabled-optimization -Wdiv-by-zero -Werror @gol
218 -Wfloat-equal -Wformat -Wformat=2 @gol
219 -Wformat-nonliteral -Wformat-security @gol
220 -Wimplicit -Wimplicit-int @gol
221 -Wimplicit-function-declaration @gol
222 -Werror-implicit-function-declaration @gol
223 -Wimport -Winline @gol
224 -Wlarger-than-@var{len} -Wlong-long @gol
225 -Wmain -Wmissing-braces -Wmissing-declarations @gol
226 -Wmissing-format-attribute -Wmissing-noreturn @gol
227 -Wmultichar -Wno-format-extra-args -Wno-format-y2k @gol
228 -Wno-import -Wpacked -Wpadded @gol
229 -Wparentheses -Wpointer-arith -Wredundant-decls @gol
230 -Wreturn-type -Wsequence-point -Wshadow @gol
231 -Wsign-compare -Wswitch -Wsystem-headers @gol
232 -Wtrigraphs -Wundef -Wuninitialized @gol
233 -Wunknown-pragmas -Wunreachable-code @gol
234 -Wunused -Wunused-function -Wunused-label -Wunused-parameter @gol
235 -Wunused-value -Wunused-variable -Wwrite-strings}
237 @item C-only Warning Options
239 -Wbad-function-cast -Wmissing-prototypes -Wnested-externs @gol
240 -Wstrict-prototypes -Wtraditional}
242 @item Debugging Options
243 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
245 -d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
247 -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
248 -fdump-tree-original@r{[}-@var{n}@r{]} -fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
249 -fdump-tree-inlined@r{[}-@var{n}@r{]} @gol
250 -fmem-report -fpretend-float @gol
251 -fprofile-arcs -ftest-coverage -ftime-report @gol
252 -g -g@var{level} -gcoff -gdwarf -gdwarf-1 -gdwarf-1+ -gdwarf-2 @gol
253 -ggdb -gstabs -gstabs+ -gvms -gxcoff -gxcoff+ @gol
254 -p -pg -print-file-name=@var{library} -print-libgcc-file-name @gol
255 -print-multi-directory -print-multi-lib @gol
256 -print-prog-name=@var{program} -print-search-dirs -Q @gol
259 @item Optimization Options
260 @xref{Optimize Options,,Options that Control Optimization}.
262 -falign-functions=@var{n} -falign-jumps=@var{n} @gol
263 -falign-labels=@var{n} -falign-loops=@var{n} @gol
264 -fbranch-probabilities -fcaller-saves -fcprop-registers @gol
265 -fcse-follow-jumps -fcse-skip-blocks -fdata-sections @gol
266 -fdelayed-branch -fdelete-null-pointer-checks @gol
267 -fexpensive-optimizations -ffast-math -ffloat-store @gol
268 -fforce-addr -fforce-mem -ffunction-sections @gol
269 -fgcse -fgcse-lm -fgcse-sm @gol
270 -finline-functions -finline-limit=@var{n} -fkeep-inline-functions @gol
271 -fkeep-static-consts -fmerge-constants -fmerge-all-constants @gol
272 -fmove-all-movables -fno-default-inline -fno-defer-pop @gol
273 -fno-function-cse -fno-guess-branch-probability @gol
274 -fno-inline -fno-math-errno -fno-peephole -fno-peephole2 @gol
275 -funsafe-math-optimizations -fno-trapping-math @gol
276 -fomit-frame-pointer -foptimize-register-move @gol
277 -foptimize-sibling-calls -fprefetch-loop-arrays -freduce-all-givs @gol
278 -fregmove -frename-registers @gol
279 -frerun-cse-after-loop -frerun-loop-opt @gol
280 -fschedule-insns -fschedule-insns2 @gol
281 -fsingle-precision-constant -fssa -fssa-ccp -fssa-dce @gol
282 -fstrength-reduce -fstrict-aliasing -fthread-jumps -ftrapv @gol
283 -funroll-all-loops -funroll-loops @gol
284 --param @var{name}=@var{value}
285 -O -O0 -O1 -O2 -O3 -Os}
287 @item Preprocessor Options
288 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
290 -$ -A@var{question}=@var{answer} -A-@var{question}@r{[}=@var{answer}@r{]} @gol
291 -C -dD -dI -dM -dN @gol
292 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
293 -idirafter @var{dir} @gol
294 -include @var{file} -imacros @var{file} @gol
295 -iprefix @var{file} -iwithprefix @var{dir} @gol
296 -iwithprefixbefore @var{dir} -isystem @var{dir} @gol
297 -M -MM -MF -MG -MP -MQ -MT -nostdinc -P -remap @gol
298 -trigraphs -undef -U@var{macro} -Wp,@var{option}}
300 @item Assembler Option
301 @xref{Assembler Options,,Passing Options to the Assembler}.
306 @xref{Link Options,,Options for Linking}.
308 @var{object-file-name} -l@var{library} @gol
309 -nostartfiles -nodefaultlibs -nostdlib @gol
310 -s -static -static-libgcc -shared -shared-libgcc -symbolic @gol
311 -Wl,@var{option} -Xlinker @var{option} @gol
314 @item Directory Options
315 @xref{Directory Options,,Options for Directory Search}.
317 -B@var{prefix} -I@var{dir} -I- -L@var{dir} -specs=@var{file}}
320 @c I wrote this xref this way to avoid overfull hbox. -- rms
321 @xref{Target Options}.
323 -b @var{machine} -V @var{version}}
325 @item Machine Dependent Options
326 @xref{Submodel Options,,Hardware Models and Configurations}.
327 @emph{M680x0 Options}
329 -m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
330 -m68060 -mcpu32 -m5200 -m68881 -mbitfield -mc68000 -mc68020 @gol
331 -mfpa -mnobitfield -mrtd -mshort -msoft-float -mpcrel @gol
332 -malign-int -mstrict-align}
334 @emph{M68hc1x Options}
336 -m6811 -m6812 -m68hc11 -m68hc12 @gol
337 -mauto-incdec -mshort -msoft-reg-count=@var{count}}
345 -mcpu=@var{cpu-type} @gol
346 -mtune=@var{cpu-type} @gol
347 -mcmodel=@var{code-model} @gol
349 -mapp-regs -mbroken-saverestore -mcypress @gol
350 -mepilogue -mfaster-structs -mflat @gol
351 -mfpu -mhard-float -mhard-quad-float @gol
352 -mimpure-text -mlive-g0 -mno-app-regs @gol
353 -mno-epilogue -mno-faster-structs -mno-flat -mno-fpu @gol
354 -mno-impure-text -mno-stack-bias -mno-unaligned-doubles @gol
355 -msoft-float -msoft-quad-float -msparclite -mstack-bias @gol
356 -msupersparc -munaligned-doubles -mv8}
358 @emph{Convex Options}
360 -mc1 -mc2 -mc32 -mc34 -mc38 @gol
361 -margcount -mnoargcount @gol
362 -mlong32 -mlong64 @gol
363 -mvolatile-cache -mvolatile-nocache}
365 @emph{AMD29K Options}
367 -m29000 -m29050 -mbw -mnbw -mdw -mndw @gol
368 -mlarge -mnormal -msmall @gol
369 -mkernel-registers -mno-reuse-arg-regs @gol
370 -mno-stack-check -mno-storem-bug @gol
371 -mreuse-arg-regs -msoft-float -mstack-check @gol
372 -mstorem-bug -muser-registers}
376 -mapcs-frame -mno-apcs-frame @gol
377 -mapcs-26 -mapcs-32 @gol
378 -mapcs-stack-check -mno-apcs-stack-check @gol
379 -mapcs-float -mno-apcs-float @gol
380 -mapcs-reentrant -mno-apcs-reentrant @gol
381 -msched-prolog -mno-sched-prolog @gol
382 -mlittle-endian -mbig-endian -mwords-little-endian @gol
383 -malignment-traps -mno-alignment-traps @gol
384 -msoft-float -mhard-float -mfpe @gol
385 -mthumb-interwork -mno-thumb-interwork @gol
386 -mcpu=@var{name} -march=@var{name} -mfpe=@var{name} @gol
387 -mstructure-size-boundary=@var{n} @gol
388 -mbsd -mxopen -mno-symrename @gol
389 -mabort-on-noreturn @gol
390 -mlong-calls -mno-long-calls @gol
391 -msingle-pic-base -mno-single-pic-base @gol
392 -mpic-register=@var{reg} @gol
393 -mnop-fun-dllimport @gol
394 -mpoke-function-name @gol
396 -mtpcs-frame -mtpcs-leaf-frame @gol
397 -mcaller-super-interworking -mcallee-super-interworking }
399 @emph{MN10200 Options}
403 @emph{MN10300 Options}
405 -mmult-bug -mno-mult-bug @gol
406 -mam33 -mno-am33 @gol
409 @emph{M32R/D Options}
411 -m32rx -m32r -mcode-model=@var{model-type} -msdata=@var{sdata-type} @gol
416 -m88000 -m88100 -m88110 -mbig-pic @gol
417 -mcheck-zero-division -mhandle-large-shift @gol
418 -midentify-revision -mno-check-zero-division @gol
419 -mno-ocs-debug-info -mno-ocs-frame-position @gol
420 -mno-optimize-arg-area -mno-serialize-volatile @gol
421 -mno-underscores -mocs-debug-info @gol
422 -mocs-frame-position -moptimize-arg-area @gol
423 -mserialize-volatile -mshort-data-@var{num} -msvr3 @gol
424 -msvr4 -mtrap-large-shift -muse-div-instruction @gol
425 -mversion-03.00 -mwarn-passed-structs}
427 @emph{RS/6000 and PowerPC Options}
429 -mcpu=@var{cpu-type} @gol
430 -mtune=@var{cpu-type} @gol
431 -mpower -mno-power -mpower2 -mno-power2 @gol
432 -mpowerpc -mpowerpc64 -mno-powerpc @gol
433 -maltivec -mno-altivec @gol
434 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
435 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
436 -mnew-mnemonics -mold-mnemonics @gol
437 -mfull-toc -mminimal-toc -mno-fop-in-toc -mno-sum-in-toc @gol
438 -m64 -m32 -mxl-call -mno-xl-call -mpe @gol
439 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
440 -mstring -mno-string -mupdate -mno-update @gol
441 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
442 -mstrict-align -mno-strict-align -mrelocatable @gol
443 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
444 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
445 -mcall-aix -mcall-sysv -mcall-netbsd @gol
446 -maix-struct-return -msvr4-struct-return
448 -mprototype -mno-prototype @gol
449 -msim -mmvme -mads -myellowknife -memb -msdata @gol
450 -msdata=@var{opt} -mvxworks -G @var{num} -pthread}
454 -mcall-lib-mul -mfp-arg-in-fpregs -mfp-arg-in-gregs @gol
455 -mfull-fp-blocks -mhc-struct-return -min-line-mul @gol
456 -mminimum-fp-blocks -mnohc-struct-return}
460 -mabicalls -march=@var{cpu-type} -mtune=@var{cpu=type} @gol
461 -mcpu=@var{cpu-type} -membedded-data -muninit-const-in-rodata @gol
462 -membedded-pic -mfp32 -mfp64 -mfused-madd -mno-fused-madd @gol
463 -mgas -mgp32 -mgp64 @gol
464 -mgpopt -mhalf-pic -mhard-float -mint64 -mips1 @gol
465 -mips2 -mips3 -mips4 -mlong64 -mlong32 -mlong-calls -mmemcpy @gol
466 -mmips-as -mmips-tfile -mno-abicalls @gol
467 -mno-embedded-data -mno-uninit-const-in-rodata @gol
468 -mno-embedded-pic -mno-gpopt -mno-long-calls @gol
469 -mno-memcpy -mno-mips-tfile -mno-rnames -mno-stats @gol
470 -mrnames -msoft-float @gol
471 -m4650 -msingle-float -mmad @gol
472 -mstats -EL -EB -G @var{num} -nocpp @gol
473 -mabi=32 -mabi=n32 -mabi=64 -mabi=eabi @gol
474 -mfix7000 -mno-crt0 -mflush-func=@var{func} -mno-flush-func}
476 @emph{i386 and x86-64 Options}
478 -mcpu=@var{cpu-type} -march=@var{cpu-type} -mfpmath=@var{unit} @gol
479 -masm=@var{dialect} -mno-fancy-math-387 @gol
480 -mno-fp-ret-in-387 -msoft-float -msvr3-shlib @gol
481 -mno-wide-multiply -mrtd -malign-double @gol
482 -mpreferred-stack-boundary=@var{num} @gol
483 -mmmx -msse -msse2 -msse-math -m3dnow @gol
484 -mthreads -mno-align-stringops -minline-all-stringops @gol
485 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
486 -m96bit-long-double -mregparm=@var{num} -momit-leaf-frame-pointer @gol
492 -march=@var{architecture-type} @gol
493 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
494 -mfast-indirect-calls -mgas -mjump-in-delay @gol
495 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
496 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
497 -mno-jump-in-delay -mno-long-load-store @gol
498 -mno-portable-runtime -mno-soft-float @gol
499 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
500 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
501 -mschedule=@var{cpu-type} -mspace-regs}
503 @emph{Intel 960 Options}
505 -m@var{cpu-type} -masm-compat -mclean-linkage @gol
506 -mcode-align -mcomplex-addr -mleaf-procedures @gol
507 -mic-compat -mic2.0-compat -mic3.0-compat @gol
508 -mintel-asm -mno-clean-linkage -mno-code-align @gol
509 -mno-complex-addr -mno-leaf-procedures @gol
510 -mno-old-align -mno-strict-align -mno-tail-call @gol
511 -mnumerics -mold-align -msoft-float -mstrict-align @gol
514 @emph{DEC Alpha Options}
516 -mfp-regs -mno-fp-regs -mno-soft-float -msoft-float @gol
517 -malpha-as -mgas @gol
518 -mieee -mieee-with-inexact -mieee-conformant @gol
519 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
520 -mtrap-precision=@var{mode} -mbuild-constants @gol
521 -mcpu=@var{cpu-type} @gol
522 -mbwx -mno-bwx -mcix -mno-cix -mmax -mno-max @gol
523 -mmemory-latency=@var{time}}
525 @emph{DEC Alpha/VMS Options}
529 @emph{Clipper Options}
533 @emph{H8/300 Options}
535 -mrelax -mh -ms -mint32 -malign-300}
539 -m1 -m2 -m3 -m3e @gol
540 -m4-nofpu -m4-single-only -m4-single -m4 @gol
541 -mb -ml -mdalign -mrelax @gol
542 -mbigtable -mfmovd -mhitachi -mnomacsave @gol
543 -mieee -misize -mpadstruct -mspace @gol
544 -mprefergot -musermode}
546 @emph{System V Options}
548 -Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
553 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
554 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
556 @emph{TMS320C3x/C4x Options}
558 -mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm @gol
559 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload @gol
560 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned @gol
561 -mparallel-insns -mparallel-mpy -mpreserve-float}
565 -mlong-calls -mno-long-calls -mep -mno-ep @gol
566 -mprolog-function -mno-prolog-function -mspace @gol
567 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
572 -m32032 -m32332 -m32532 -m32081 -m32381 @gol
573 -mmult-add -mnomult-add -msoft-float -mrtd -mnortd @gol
574 -mregparam -mnoregparam -msb -mnosb @gol
575 -mbitfield -mnobitfield -mhimem -mnohimem}
579 -mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts @gol
580 -mcall-prologues -mno-tablejump -mtiny-stack}
584 -mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
585 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
586 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
587 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
588 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
592 -mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
593 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
594 -melf -mbranch-predict -mreg-stack-fill-bug-workaround @gol
595 -mno-branch-predict -mno-reg-stack-fill-bug-workaround}
599 -mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
600 -mvolatile-asm-stop -mb-step -mregister-names -mno-sdata @gol
601 -mconstant-gp -mauto-pic -minline-divide-min-latency @gol
602 -minline-divide-max-throughput -mno-dwarf2-asm @gol
603 -mfixed-range=@var{register-range}}
605 @emph{S/390 and zSeries Options}
607 -mhard-float -msoft-float -mbackchain -mno-backchain @gol
608 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
609 -m64 -m31 -mdebug -mno-debug}
613 -mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
614 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
615 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
616 -mstack-align -mdata-align -mconst-align @gol
617 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
618 -melf -maout -melinux -mlinux -sim -sim2}
620 @item Code Generation Options
621 @xref{Code Gen Options,,Options for Code Generation Conventions}.
623 -fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
624 -ffixed-@var{reg} -fexceptions @gol
625 -fnon-call-exceptions -funwind-tables -fasynchronous-unwind-tables @gol
626 -finhibit-size-directive -finstrument-functions @gol
627 -fno-common -fno-ident -fno-gnu-linker @gol
628 -fpcc-struct-return -fpic -fPIC @gol
629 -freg-struct-return -fshared-data -fshort-enums @gol
630 -fshort-double -fvolatile @gol
631 -fvolatile-global -fvolatile-static @gol
632 -fverbose-asm -fpack-struct -fstack-check @gol
633 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
634 -fargument-alias -fargument-noalias @gol
635 -fargument-noalias-global -fleading-underscore}
639 * Overall Options:: Controlling the kind of output:
640 an executable, object files, assembler files,
641 or preprocessed source.
642 * C Dialect Options:: Controlling the variant of C language compiled.
643 * C++ Dialect Options:: Variations on C++.
644 * Objective-C Dialect Options:: Variations on Objective-C.
645 * Language Independent Options:: Controlling how diagnostics should be
647 * Warning Options:: How picky should the compiler be?
648 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
649 * Optimize Options:: How much optimization?
650 * Preprocessor Options:: Controlling header files and macro definitions.
651 Also, getting dependency information for Make.
652 * Assembler Options:: Passing options to the assembler.
653 * Link Options:: Specifying libraries and so on.
654 * Directory Options:: Where to find header files and libraries.
655 Where to find the compiler executable files.
656 * Spec Files:: How to pass switches to sub-processes.
657 * Target Options:: Running a cross-compiler, or an old version of GCC.
660 @node Overall Options
661 @section Options Controlling the Kind of Output
663 Compilation can involve up to four stages: preprocessing, compilation
664 proper, assembly and linking, always in that order. The first three
665 stages apply to an individual source file, and end by producing an
666 object file; linking combines all the object files (those newly
667 compiled, and those specified as input) into an executable file.
669 @cindex file name suffix
670 For any given input file, the file name suffix determines what kind of
675 C source code which must be preprocessed.
678 C source code which should not be preprocessed.
681 C++ source code which should not be preprocessed.
684 Objective-C source code. Note that you must link with the library
685 @file{libobjc.a} to make an Objective-C program work.
688 Objective-C source code which should not be preprocessed.
691 C header file (not to be compiled or linked).
695 @itemx @var{file}.cxx
696 @itemx @var{file}.cpp
697 @itemx @var{file}.c++
699 C++ source code which must be preprocessed. Note that in @samp{.cxx},
700 the last two letters must both be literally @samp{x}. Likewise,
701 @samp{.C} refers to a literal capital C@.
704 @itemx @var{file}.for
705 @itemx @var{file}.FOR
706 Fortran source code which should not be preprocessed.
709 @itemx @var{file}.fpp
710 @itemx @var{file}.FPP
711 Fortran source code which must be preprocessed (with the traditional
715 Fortran source code which must be preprocessed with a RATFOR
716 preprocessor (not included with GCC)@.
718 @xref{Overall Options,,Options Controlling the Kind of Output, g77,
719 Using and Porting GNU Fortran}, for more details of the handling of
722 @c FIXME: Descriptions of Java file types.
729 Ada source code file which contains a library unit declaration (a
730 declaration of a package, subprogram, or generic, or a generic
731 instantiation), or a library unit renaming declaration (a package,
732 generic, or subprogram renaming declaration). Such files are also
735 @itemx @var{file}.adb
736 Ada source code file containing a library unit body (a subprogram or
737 package body). Such files are also called @dfn{bodies}.
739 @c GCC also knows about some suffixes for languages not yet included:
745 @itemx @var{file}.chi
746 CHILL source code (preprocessed with the traditional preprocessor).
752 Assembler code which must be preprocessed.
755 An object file to be fed straight into linking.
756 Any file name with no recognized suffix is treated this way.
760 You can specify the input language explicitly with the @option{-x} option:
763 @item -x @var{language}
764 Specify explicitly the @var{language} for the following input files
765 (rather than letting the compiler choose a default based on the file
766 name suffix). This option applies to all following input files until
767 the next @option{-x} option. Possible values for @var{language} are:
769 c c-header cpp-output
771 objective-c objc-cpp-output
772 assembler assembler-with-cpp
775 f77 f77-cpp-input ratfor
780 Turn off any specification of a language, so that subsequent files are
781 handled according to their file name suffixes (as they are if @option{-x}
782 has not been used at all).
784 @item -pass-exit-codes
785 @opindex pass-exit-codes
786 Normally the @command{gcc} program will exit with the code of 1 if any
787 phase of the compiler returns a non-success return code. If you specify
788 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
789 numerically highest error produced by any phase that returned an error
793 If you only want some of the stages of compilation, you can use
794 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
795 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
796 @command{gcc} is to stop. Note that some combinations (for example,
797 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
802 Compile or assemble the source files, but do not link. The linking
803 stage simply is not done. The ultimate output is in the form of an
804 object file for each source file.
806 By default, the object file name for a source file is made by replacing
807 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
809 Unrecognized input files, not requiring compilation or assembly, are
814 Stop after the stage of compilation proper; do not assemble. The output
815 is in the form of an assembler code file for each non-assembler input
818 By default, the assembler file name for a source file is made by
819 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
821 Input files that don't require compilation are ignored.
825 Stop after the preprocessing stage; do not run the compiler proper. The
826 output is in the form of preprocessed source code, which is sent to the
829 Input files which don't require preprocessing are ignored.
831 @cindex output file option
834 Place output in file @var{file}. This applies regardless to whatever
835 sort of output is being produced, whether it be an executable file,
836 an object file, an assembler file or preprocessed C code.
838 Since only one output file can be specified, it does not make sense to
839 use @option{-o} when compiling more than one input file, unless you are
840 producing an executable file as output.
842 If @option{-o} is not specified, the default is to put an executable file
843 in @file{a.out}, the object file for @file{@var{source}.@var{suffix}} in
844 @file{@var{source}.o}, its assembler file in @file{@var{source}.s}, and
845 all preprocessed C source on standard output.
849 Print (on standard error output) the commands executed to run the stages
850 of compilation. Also print the version number of the compiler driver
851 program and of the preprocessor and the compiler proper.
855 Use pipes rather than temporary files for communication between the
856 various stages of compilation. This fails to work on some systems where
857 the assembler is unable to read from a pipe; but the GNU assembler has
862 Print (on the standard output) a description of the command line options
863 understood by @command{gcc}. If the @option{-v} option is also specified
864 then @option{--help} will also be passed on to the various processes
865 invoked by @command{gcc}, so that they can display the command line options
866 they accept. If the @option{-W} option is also specified then command
867 line options which have no documentation associated with them will also
872 Print (on the standard output) a description of target specific command
873 line options for each tool.
877 @section Compiling C++ Programs
879 @cindex suffixes for C++ source
880 @cindex C++ source file suffixes
881 C++ source files conventionally use one of the suffixes @samp{.C},
882 @samp{.cc}, @samp{.cpp}, @samp{.c++}, @samp{.cp}, or @samp{.cxx};
883 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
884 files with these names and compiles them as C++ programs even if you
885 call the compiler the same way as for compiling C programs (usually with
886 the name @command{gcc}).
890 However, C++ programs often require class libraries as well as a
891 compiler that understands the C++ language---and under some
892 circumstances, you might want to compile programs from standard input,
893 or otherwise without a suffix that flags them as C++ programs.
894 @command{g++} is a program that calls GCC with the default language
895 set to C++, and automatically specifies linking against the C++
896 library. On many systems, @command{g++} is also
897 installed with the name @command{c++}.
899 @cindex invoking @command{g++}
900 When you compile C++ programs, you may specify many of the same
901 command-line options that you use for compiling programs in any
902 language; or command-line options meaningful for C and related
903 languages; or options that are meaningful only for C++ programs.
904 @xref{C Dialect Options,,Options Controlling C Dialect}, for
905 explanations of options for languages related to C@.
906 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
907 explanations of options that are meaningful only for C++ programs.
909 @node C Dialect Options
910 @section Options Controlling C Dialect
911 @cindex dialect options
912 @cindex language dialect options
913 @cindex options, dialect
915 The following options control the dialect of C (or languages derived
916 from C, such as C++ and Objective-C) that the compiler accepts:
923 In C mode, support all ISO C89 programs. In C++ mode,
924 remove GNU extensions that conflict with ISO C++.
926 This turns off certain features of GCC that are incompatible with ISO
927 C89 (when compiling C code), or of standard C++ (when compiling C++ code),
928 such as the @code{asm} and @code{typeof} keywords, and
929 predefined macros such as @code{unix} and @code{vax} that identify the
930 type of system you are using. It also enables the undesirable and
931 rarely used ISO trigraph feature. For the C compiler,
932 it disables recognition of C++ style @samp{//} comments as well as
933 the @code{inline} keyword.
935 The alternate keywords @code{__asm__}, @code{__extension__},
936 @code{__inline__} and @code{__typeof__} continue to work despite
937 @option{-ansi}. You would not want to use them in an ISO C program, of
938 course, but it is useful to put them in header files that might be included
939 in compilations done with @option{-ansi}. Alternate predefined macros
940 such as @code{__unix__} and @code{__vax__} are also available, with or
941 without @option{-ansi}.
943 The @option{-ansi} option does not cause non-ISO programs to be
944 rejected gratuitously. For that, @option{-pedantic} is required in
945 addition to @option{-ansi}. @xref{Warning Options}.
947 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
948 option is used. Some header files may notice this macro and refrain
949 from declaring certain functions or defining certain macros that the
950 ISO standard doesn't call for; this is to avoid interfering with any
951 programs that might use these names for other things.
953 Functions which would normally be built in but do not have semantics
954 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
955 functions with @option{-ansi} is used. @xref{Other Builtins,,Other
956 built-in functions provided by GCC}, for details of the functions
961 Determine the language standard. This option is currently only
962 supported when compiling C@. A value for this option must be provided;
968 ISO C89 (same as @option{-ansi}).
971 ISO C89 as modified in amendment 1.
977 ISO C99. Note that this standard is not yet fully supported; see
978 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
979 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
982 Default, ISO C89 plus GNU extensions (including some C99 features).
986 ISO C99 plus GNU extensions. When ISO C99 is fully implemented in GCC,
987 this will become the default. The name @samp{gnu9x} is deprecated.
991 Even when this option is not specified, you can still use some of the
992 features of newer standards in so far as they do not conflict with
993 previous C standards. For example, you may use @code{__restrict__} even
994 when @option{-std=c99} is not specified.
996 The @option{-std} options specifying some version of ISO C have the same
997 effects as @option{-ansi}, except that features that were not in ISO C89
998 but are in the specified version (for example, @samp{//} comments and
999 the @code{inline} keyword in ISO C99) are not disabled.
1001 @xref{Standards,,Language Standards Supported by GCC}, for details of
1002 these standard versions.
1004 @item -aux-info @var{filename}
1006 Output to the given filename prototyped declarations for all functions
1007 declared and/or defined in a translation unit, including those in header
1008 files. This option is silently ignored in any language other than C@.
1010 Besides declarations, the file indicates, in comments, the origin of
1011 each declaration (source file and line), whether the declaration was
1012 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1013 @samp{O} for old, respectively, in the first character after the line
1014 number and the colon), and whether it came from a declaration or a
1015 definition (@samp{C} or @samp{F}, respectively, in the following
1016 character). In the case of function definitions, a K&R-style list of
1017 arguments followed by their declarations is also provided, inside
1018 comments, after the declaration.
1022 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1023 keyword, so that code can use these words as identifiers. You can use
1024 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1025 instead. @option{-ansi} implies @option{-fno-asm}.
1027 In C++, this switch only affects the @code{typeof} keyword, since
1028 @code{asm} and @code{inline} are standard keywords. You may want to
1029 use the @option{-fno-gnu-keywords} flag instead, which has the same
1030 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1031 switch only affects the @code{asm} and @code{typeof} keywords, since
1032 @code{inline} is a standard keyword in ISO C99.
1035 @itemx -fno-builtin-@var{function} @r{(C and Objective-C only)}
1036 @opindex fno-builtin
1037 @cindex built-in functions
1038 Don't recognize built-in functions that do not begin with
1039 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1040 functions provided by GCC}, for details of the functions affected,
1041 including those which are not built-in functions when @option{-ansi} or
1042 @option{-std} options for strict ISO C conformance are used because they
1043 do not have an ISO standard meaning.
1045 GCC normally generates special code to handle certain built-in functions
1046 more efficiently; for instance, calls to @code{alloca} may become single
1047 instructions that adjust the stack directly, and calls to @code{memcpy}
1048 may become inline copy loops. The resulting code is often both smaller
1049 and faster, but since the function calls no longer appear as such, you
1050 cannot set a breakpoint on those calls, nor can you change the behavior
1051 of the functions by linking with a different library.
1053 In C++, @option{-fno-builtin} is always in effect. The @option{-fbuiltin}
1054 option has no effect. Therefore, in C++, the only way to get the
1055 optimization benefits of built-in functions is to call the function
1056 using the @samp{__builtin_} prefix. The GNU C++ Standard Library uses
1057 built-in functions to implement many functions (like
1058 @code{std::strchr}), so that you automatically get efficient code.
1060 With the @option{-fno-builtin-@var{function}} option, not available
1061 when compiling C++, only the built-in function @var{function} is
1062 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1063 function is named this is not built-in in this version of GCC, this
1064 option is ignored. There is no corresponding
1065 @option{-fbuiltin-@var{function}} option; if you wish to enable
1066 built-in functions selectively when using @option{-fno-builtin} or
1067 @option{-ffreestanding}, you may define macros such as:
1070 #define abs(n) __builtin_abs ((n))
1071 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1076 @cindex hosted environment
1078 Assert that compilation takes place in a hosted environment. This implies
1079 @option{-fbuiltin}. A hosted environment is one in which the
1080 entire standard library is available, and in which @code{main} has a return
1081 type of @code{int}. Examples are nearly everything except a kernel.
1082 This is equivalent to @option{-fno-freestanding}.
1084 @item -ffreestanding
1085 @opindex ffreestanding
1086 @cindex hosted environment
1088 Assert that compilation takes place in a freestanding environment. This
1089 implies @option{-fno-builtin}. A freestanding environment
1090 is one in which the standard library may not exist, and program startup may
1091 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1092 This is equivalent to @option{-fno-hosted}.
1094 @xref{Standards,,Language Standards Supported by GCC}, for details of
1095 freestanding and hosted environments.
1099 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1100 options for strict ISO C conformance) implies @option{-trigraphs}.
1102 @cindex traditional C language
1103 @cindex C language, traditional
1105 @opindex traditional
1106 Attempt to support some aspects of traditional C compilers.
1111 All @code{extern} declarations take effect globally even if they
1112 are written inside of a function definition. This includes implicit
1113 declarations of functions.
1116 The newer keywords @code{typeof}, @code{inline}, @code{signed}, @code{const}
1117 and @code{volatile} are not recognized. (You can still use the
1118 alternative keywords such as @code{__typeof__}, @code{__inline__}, and
1122 Comparisons between pointers and integers are always allowed.
1125 Integer types @code{unsigned short} and @code{unsigned char} promote
1126 to @code{unsigned int}.
1129 Out-of-range floating point literals are not an error.
1132 Certain constructs which ISO regards as a single invalid preprocessing
1133 number, such as @samp{0xe-0xd}, are treated as expressions instead.
1136 String ``constants'' are not necessarily constant; they are stored in
1137 writable space, and identical looking constants are allocated
1138 separately. (This is the same as the effect of
1139 @option{-fwritable-strings}.)
1141 @cindex @code{longjmp} and automatic variables
1143 All automatic variables not declared @code{register} are preserved by
1144 @code{longjmp}. Ordinarily, GNU C follows ISO C: automatic variables
1145 not declared @code{volatile} may be clobbered.
1150 @cindex escape sequences, traditional
1151 The character escape sequences @samp{\x} and @samp{\a} evaluate as the
1152 literal characters @samp{x} and @samp{a} respectively. Without
1153 @w{@option{-traditional}}, @samp{\x} is a prefix for the hexadecimal
1154 representation of a character, and @samp{\a} produces a bell.
1157 This option is deprecated and may be removed.
1159 You may wish to use @option{-fno-builtin} as well as @option{-traditional}
1160 if your program uses names that are normally GNU C built-in functions for
1161 other purposes of its own.
1163 You cannot use @option{-traditional} if you include any header files that
1164 rely on ISO C features. Some vendors are starting to ship systems with
1165 ISO C header files and you cannot use @option{-traditional} on such
1166 systems to compile files that include any system headers.
1168 The @option{-traditional} option also enables @option{-traditional-cpp}.
1170 @item -traditional-cpp
1171 @opindex traditional-cpp
1172 Attempt to support some aspects of traditional C preprocessors.
1173 See the GNU CPP manual for details.
1175 @item -fcond-mismatch
1176 @opindex fcond-mismatch
1177 Allow conditional expressions with mismatched types in the second and
1178 third arguments. The value of such an expression is void. This option
1179 is not supported for C++.
1181 @item -funsigned-char
1182 @opindex funsigned-char
1183 Let the type @code{char} be unsigned, like @code{unsigned char}.
1185 Each kind of machine has a default for what @code{char} should
1186 be. It is either like @code{unsigned char} by default or like
1187 @code{signed char} by default.
1189 Ideally, a portable program should always use @code{signed char} or
1190 @code{unsigned char} when it depends on the signedness of an object.
1191 But many programs have been written to use plain @code{char} and
1192 expect it to be signed, or expect it to be unsigned, depending on the
1193 machines they were written for. This option, and its inverse, let you
1194 make such a program work with the opposite default.
1196 The type @code{char} is always a distinct type from each of
1197 @code{signed char} or @code{unsigned char}, even though its behavior
1198 is always just like one of those two.
1201 @opindex fsigned-char
1202 Let the type @code{char} be signed, like @code{signed char}.
1204 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1205 the negative form of @option{-funsigned-char}. Likewise, the option
1206 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1208 @item -fsigned-bitfields
1209 @itemx -funsigned-bitfields
1210 @itemx -fno-signed-bitfields
1211 @itemx -fno-unsigned-bitfields
1212 @opindex fsigned-bitfields
1213 @opindex funsigned-bitfields
1214 @opindex fno-signed-bitfields
1215 @opindex fno-unsigned-bitfields
1216 These options control whether a bit-field is signed or unsigned, when the
1217 declaration does not use either @code{signed} or @code{unsigned}. By
1218 default, such a bit-field is signed, because this is consistent: the
1219 basic integer types such as @code{int} are signed types.
1221 However, when @option{-traditional} is used, bit-fields are all unsigned
1224 @item -fwritable-strings
1225 @opindex fwritable-strings
1226 Store string constants in the writable data segment and don't uniquize
1227 them. This is for compatibility with old programs which assume they can
1228 write into string constants. The option @option{-traditional} also has
1231 Writing into string constants is a very bad idea; ``constants'' should
1234 @item -fallow-single-precision
1235 @opindex fallow-single-precision
1236 Do not promote single precision math operations to double precision,
1237 even when compiling with @option{-traditional}.
1239 Traditional K&R C promotes all floating point operations to double
1240 precision, regardless of the sizes of the operands. On the
1241 architecture for which you are compiling, single precision may be faster
1242 than double precision. If you must use @option{-traditional}, but want
1243 to use single precision operations when the operands are single
1244 precision, use this option. This option has no effect when compiling
1245 with ISO or GNU C conventions (the default).
1248 @opindex fshort-wchar
1249 Override the underlying type for @samp{wchar_t} to be @samp{short
1250 unsigned int} instead of the default for the target. This option is
1251 useful for building programs to run under WINE@.
1254 @node C++ Dialect Options
1255 @section Options Controlling C++ Dialect
1257 @cindex compiler options, C++
1258 @cindex C++ options, command line
1259 @cindex options, C++
1260 This section describes the command-line options that are only meaningful
1261 for C++ programs; but you can also use most of the GNU compiler options
1262 regardless of what language your program is in. For example, you
1263 might compile a file @code{firstClass.C} like this:
1266 g++ -g -frepo -O -c firstClass.C
1270 In this example, only @option{-frepo} is an option meant
1271 only for C++ programs; you can use the other options with any
1272 language supported by GCC@.
1274 Here is a list of options that are @emph{only} for compiling C++ programs:
1277 @item -fno-access-control
1278 @opindex fno-access-control
1279 Turn off all access checking. This switch is mainly useful for working
1280 around bugs in the access control code.
1284 Check that the pointer returned by @code{operator new} is non-null
1285 before attempting to modify the storage allocated. The current Working
1286 Paper requires that @code{operator new} never return a null pointer, so
1287 this check is normally unnecessary.
1289 An alternative to using this option is to specify that your
1290 @code{operator new} does not throw any exceptions; if you declare it
1291 @samp{throw()}, G++ will check the return value. See also @samp{new
1294 @item -fconserve-space
1295 @opindex fconserve-space
1296 Put uninitialized or runtime-initialized global variables into the
1297 common segment, as C does. This saves space in the executable at the
1298 cost of not diagnosing duplicate definitions. If you compile with this
1299 flag and your program mysteriously crashes after @code{main()} has
1300 completed, you may have an object that is being destroyed twice because
1301 two definitions were merged.
1303 This option is no longer useful on most targets, now that support has
1304 been added for putting variables into BSS without making them common.
1306 @item -fno-const-strings
1307 @opindex fno-const-strings
1308 Give string constants type @code{char *} instead of type @code{const
1309 char *}. By default, G++ uses type @code{const char *} as required by
1310 the standard. Even if you use @option{-fno-const-strings}, you cannot
1311 actually modify the value of a string constant, unless you also use
1312 @option{-fwritable-strings}.
1314 This option might be removed in a future release of G++. For maximum
1315 portability, you should structure your code so that it works with
1316 string constants that have type @code{const char *}.
1318 @item -fdollars-in-identifiers
1319 @opindex fdollars-in-identifiers
1320 Accept @samp{$} in identifiers. You can also explicitly prohibit use of
1321 @samp{$} with the option @option{-fno-dollars-in-identifiers}. (GNU C allows
1322 @samp{$} by default on most target systems, but there are a few exceptions.)
1323 Traditional C allowed the character @samp{$} to form part of
1324 identifiers. However, ISO C and C++ forbid @samp{$} in identifiers.
1326 @item -fno-elide-constructors
1327 @opindex fno-elide-constructors
1328 The C++ standard allows an implementation to omit creating a temporary
1329 which is only used to initialize another object of the same type.
1330 Specifying this option disables that optimization, and forces G++ to
1331 call the copy constructor in all cases.
1333 @item -fno-enforce-eh-specs
1334 @opindex fno-enforce-eh-specs
1335 Don't check for violation of exception specifications at runtime. This
1336 option violates the C++ standard, but may be useful for reducing code
1337 size in production builds, much like defining @samp{NDEBUG}. The compiler
1338 will still optimize based on the exception specifications.
1340 @item -fexternal-templates
1341 @opindex fexternal-templates
1343 Cause @samp{#pragma interface} and @samp{implementation} to apply to
1344 template instantiation; template instances are emitted or not according
1345 to the location of the template definition. @xref{Template
1346 Instantiation}, for more information.
1348 This option is deprecated.
1350 @item -falt-external-templates
1351 @opindex falt-external-templates
1352 Similar to @option{-fexternal-templates}, but template instances are
1353 emitted or not according to the place where they are first instantiated.
1354 @xref{Template Instantiation}, for more information.
1356 This option is deprecated.
1359 @itemx -fno-for-scope
1361 @opindex fno-for-scope
1362 If @option{-ffor-scope} is specified, the scope of variables declared in
1363 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1364 as specified by the C++ standard.
1365 If @option{-fno-for-scope} is specified, the scope of variables declared in
1366 a @i{for-init-statement} extends to the end of the enclosing scope,
1367 as was the case in old versions of G++, and other (traditional)
1368 implementations of C++.
1370 The default if neither flag is given to follow the standard,
1371 but to allow and give a warning for old-style code that would
1372 otherwise be invalid, or have different behavior.
1374 @item -fno-gnu-keywords
1375 @opindex fno-gnu-keywords
1376 Do not recognize @code{typeof} as a keyword, so that code can use this
1377 word as an identifier. You can use the keyword @code{__typeof__} instead.
1378 @option{-ansi} implies @option{-fno-gnu-keywords}.
1380 @item -fno-implicit-templates
1381 @opindex fno-implicit-templates
1382 Never emit code for non-inline templates which are instantiated
1383 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1384 @xref{Template Instantiation}, for more information.
1386 @item -fno-implicit-inline-templates
1387 @opindex fno-implicit-inline-templates
1388 Don't emit code for implicit instantiations of inline templates, either.
1389 The default is to handle inlines differently so that compiles with and
1390 without optimization will need the same set of explicit instantiations.
1392 @item -fno-implement-inlines
1393 @opindex fno-implement-inlines
1394 To save space, do not emit out-of-line copies of inline functions
1395 controlled by @samp{#pragma implementation}. This will cause linker
1396 errors if these functions are not inlined everywhere they are called.
1398 @item -fms-extensions
1399 @opindex fms-extensions
1400 Disable pedantic warnings about constructs used in MFC, such as implicit
1401 int and getting a pointer to member function via non-standard syntax.
1403 @item -fno-nonansi-builtins
1404 @opindex fno-nonansi-builtins
1405 Disable built-in declarations of functions that are not mandated by
1406 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1407 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1409 @item -fno-operator-names
1410 @opindex fno-operator-names
1411 Do not treat the operator name keywords @code{and}, @code{bitand},
1412 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1413 synonyms as keywords.
1415 @item -fno-optional-diags
1416 @opindex fno-optional-diags
1417 Disable diagnostics that the standard says a compiler does not need to
1418 issue. Currently, the only such diagnostic issued by G++ is the one for
1419 a name having multiple meanings within a class.
1422 @opindex fpermissive
1423 Downgrade messages about nonconformant code from errors to warnings. By
1424 default, G++ effectively sets @option{-pedantic-errors} without
1425 @option{-pedantic}; this option reverses that. This behavior and this
1426 option are superseded by @option{-pedantic}, which works as it does for GNU C@.
1430 Enable automatic template instantiation at link time. This option also
1431 implies @option{-fno-implicit-templates}. @xref{Template
1432 Instantiation}, for more information.
1436 Disable generation of information about every class with virtual
1437 functions for use by the C++ runtime type identification features
1438 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1439 of the language, you can save some space by using this flag. Note that
1440 exception handling uses the same information, but it will generate it as
1445 Emit statistics about front-end processing at the end of the compilation.
1446 This information is generally only useful to the G++ development team.
1448 @item -ftemplate-depth-@var{n}
1449 @opindex ftemplate-depth
1450 Set the maximum instantiation depth for template classes to @var{n}.
1451 A limit on the template instantiation depth is needed to detect
1452 endless recursions during template class instantiation. ANSI/ISO C++
1453 conforming programs must not rely on a maximum depth greater than 17.
1455 @item -fuse-cxa-atexit
1456 @opindex fuse-cxa-atexit
1457 Register destructors for objects with static storage duration with the
1458 @code{__cxa_atexit} function rather than the @code{atexit} function.
1459 This option is required for fully standards-compliant handling of static
1460 destructors, but will only work if your C library supports
1461 @code{__cxa_atexit}.
1465 Emit special relocations for vtables and virtual function references
1466 so that the linker can identify unused virtual functions and zero out
1467 vtable slots that refer to them. This is most useful with
1468 @option{-ffunction-sections} and @option{-Wl,--gc-sections}, in order to
1469 also discard the functions themselves.
1471 This optimization requires GNU as and GNU ld. Not all systems support
1472 this option. @option{-Wl,--gc-sections} is ignored without @option{-static}.
1476 Do not use weak symbol support, even if it is provided by the linker.
1477 By default, G++ will use weak symbols if they are available. This
1478 option exists only for testing, and should not be used by end-users;
1479 it will result in inferior code and has no benefits. This option may
1480 be removed in a future release of G++.
1484 Do not search for header files in the standard directories specific to
1485 C++, but do still search the other standard directories. (This option
1486 is used when building the C++ library.)
1489 In addition, these optimization, warning, and code generation options
1490 have meanings only for C++ programs:
1493 @item -fno-default-inline
1494 @opindex fno-default-inline
1495 Do not assume @samp{inline} for functions defined inside a class scope.
1496 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1497 functions will have linkage like inline functions; they just won't be
1500 @item -Wctor-dtor-privacy @r{(C++ only)}
1501 @opindex Wctor-dtor-privacy
1502 Warn when a class seems unusable, because all the constructors or
1503 destructors in a class are private and the class has no friends or
1504 public static member functions.
1506 @item -Wnon-virtual-dtor @r{(C++ only)}
1507 @opindex Wnon-virtual-dtor
1508 Warn when a class declares a non-virtual destructor that should probably
1509 be virtual, because it looks like the class will be used polymorphically.
1511 @item -Wreorder @r{(C++ only)}
1513 @cindex reordering, warning
1514 @cindex warning for reordering of member initializers
1515 Warn when the order of member initializers given in the code does not
1516 match the order in which they must be executed. For instance:
1522 A(): j (0), i (1) @{ @}
1526 Here the compiler will warn that the member initializers for @samp{i}
1527 and @samp{j} will be rearranged to match the declaration order of the
1531 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
1534 @item -Weffc++ @r{(C++ only)}
1536 Warn about violations of various style guidelines from Scott Meyers'
1537 @cite{Effective C++} books. If you use this option, you should be aware
1538 that the standard library headers do not obey all of these guidelines;
1539 you can use @samp{grep -v} to filter out those warnings.
1541 @item -Wno-deprecated @r{(C++ only)}
1542 @opindex Wno-deprecated
1543 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1545 @item -Wno-non-template-friend @r{(C++ only)}
1546 @opindex Wno-non-template-friend
1547 Disable warnings when non-templatized friend functions are declared
1548 within a template. With the advent of explicit template specification
1549 support in G++, if the name of the friend is an unqualified-id (i.e.,
1550 @samp{friend foo(int)}), the C++ language specification demands that the
1551 friend declare or define an ordinary, nontemplate function. (Section
1552 14.5.3). Before G++ implemented explicit specification, unqualified-ids
1553 could be interpreted as a particular specialization of a templatized
1554 function. Because this non-conforming behavior is no longer the default
1555 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
1556 check existing code for potential trouble spots, and is on by default.
1557 This new compiler behavior can be turned off with
1558 @option{-Wno-non-template-friend} which keeps the conformant compiler code
1559 but disables the helpful warning.
1561 @item -Wold-style-cast @r{(C++ only)}
1562 @opindex Wold-style-cast
1563 Warn if an old-style (C-style) cast to a non-void type is used within
1564 a C++ program. The new-style casts (@samp{static_cast},
1565 @samp{reinterpret_cast}, and @samp{const_cast}) are less vulnerable to
1566 unintended effects, and much easier to grep for.
1568 @item -Woverloaded-virtual @r{(C++ only)}
1569 @opindex Woverloaded-virtual
1570 @cindex overloaded virtual fn, warning
1571 @cindex warning for overloaded virtual fn
1572 Warn when a function declaration hides virtual functions from a
1573 base class. For example, in:
1580 struct B: public A @{
1585 the @code{A} class version of @code{f} is hidden in @code{B}, and code
1593 will fail to compile.
1595 @item -Wno-pmf-conversions @r{(C++ only)}
1596 @opindex Wno-pmf-conversions
1597 Disable the diagnostic for converting a bound pointer to member function
1600 @item -Wsign-promo @r{(C++ only)}
1601 @opindex Wsign-promo
1602 Warn when overload resolution chooses a promotion from unsigned or
1603 enumeral type to a signed type over a conversion to an unsigned type of
1604 the same size. Previous versions of G++ would try to preserve
1605 unsignedness, but the standard mandates the current behavior.
1607 @item -Wsynth @r{(C++ only)}
1609 @cindex warning for synthesized methods
1610 @cindex synthesized methods, warning
1611 Warn when G++'s synthesis behavior does not match that of cfront. For
1617 A& operator = (int);
1627 In this example, G++ will synthesize a default @samp{A& operator =
1628 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1631 @node Objective-C Dialect Options
1632 @section Options Controlling Objective-C Dialect
1634 @cindex compiler options, Objective-C
1635 @cindex Objective-C options, command line
1636 @cindex options, Objective-C
1637 This section describes the command-line options that are only meaningful
1638 for Objective-C programs; but you can also use most of the GNU compiler
1639 options regardless of what language your program is in. For example,
1640 you might compile a file @code{some_class.m} like this:
1643 gcc -g -fgnu-runtime -O -c some_class.m
1647 In this example, only @option{-fgnu-runtime} is an option meant only for
1648 Objective-C programs; you can use the other options with any language
1651 Here is a list of options that are @emph{only} for compiling Objective-C
1655 @item -fconstant-string-class=@var{class-name}
1656 @opindex fconstant-string-class
1657 Use @var{class-name} as the name of the class to instantiate for each
1658 literal string specified with the syntax @code{@@"@dots{}"}. The default
1659 class name is @code{NXConstantString}.
1662 @opindex fgnu-runtime
1663 Generate object code compatible with the standard GNU Objective-C
1664 runtime. This is the default for most types of systems.
1666 @item -fnext-runtime
1667 @opindex fnext-runtime
1668 Generate output compatible with the NeXT runtime. This is the default
1669 for NeXT-based systems, including Darwin and Mac OS X@.
1673 Dump interface declarations for all classes seen in the source file to a
1674 file named @file{@var{sourcename}.decl}.
1677 @opindex Wno-protocol
1678 Do not warn if methods required by a protocol are not implemented
1679 in the class adopting it.
1683 Warn if a selector has multiple methods of different types defined.
1685 @c not documented because only avail via -Wp
1686 @c @item -print-objc-runtime-info
1690 @node Language Independent Options
1691 @section Options to Control Diagnostic Messages Formatting
1692 @cindex options to control diagnostics formatting
1693 @cindex diagnostic messages
1694 @cindex message formatting
1696 Traditionally, diagnostic messages have been formatted irrespective of
1697 the output device's aspect (e.g.@: its width, @dots{}). The options described
1698 below can be used to control the diagnostic messages formatting
1699 algorithm, e.g.@: how many characters per line, how often source location
1700 information should be reported. Right now, only the C++ front end can
1701 honor these options. However it is expected, in the near future, that
1702 the remaining front ends would be able to digest them correctly.
1705 @item -fmessage-length=@var{n}
1706 @opindex fmessage-length
1707 Try to format error messages so that they fit on lines of about @var{n}
1708 characters. The default is 72 characters for @command{g++} and 0 for the rest of
1709 the front ends supported by GCC@. If @var{n} is zero, then no
1710 line-wrapping will be done; each error message will appear on a single
1713 @opindex fdiagnostics-show-location
1714 @item -fdiagnostics-show-location=once
1715 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
1716 reporter to emit @emph{once} source location information; that is, in
1717 case the message is too long to fit on a single physical line and has to
1718 be wrapped, the source location won't be emitted (as prefix) again,
1719 over and over, in subsequent continuation lines. This is the default
1722 @item -fdiagnostics-show-location=every-line
1723 Only meaningful in line-wrapping mode. Instructs the diagnostic
1724 messages reporter to emit the same source location information (as
1725 prefix) for physical lines that result from the process of breaking
1726 a message which is too long to fit on a single line.
1730 @node Warning Options
1731 @section Options to Request or Suppress Warnings
1732 @cindex options to control warnings
1733 @cindex warning messages
1734 @cindex messages, warning
1735 @cindex suppressing warnings
1737 Warnings are diagnostic messages that report constructions which
1738 are not inherently erroneous but which are risky or suggest there
1739 may have been an error.
1741 You can request many specific warnings with options beginning @samp{-W},
1742 for example @option{-Wimplicit} to request warnings on implicit
1743 declarations. Each of these specific warning options also has a
1744 negative form beginning @samp{-Wno-} to turn off warnings;
1745 for example, @option{-Wno-implicit}. This manual lists only one of the
1746 two forms, whichever is not the default.
1748 These options control the amount and kinds of warnings produced by GCC:
1751 @cindex syntax checking
1753 @opindex fsyntax-only
1754 Check the code for syntax errors, but don't do anything beyond that.
1758 Issue all the warnings demanded by strict ISO C and ISO C++;
1759 reject all programs that use forbidden extensions, and some other
1760 programs that do not follow ISO C and ISO C++. For ISO C, follows the
1761 version of the ISO C standard specified by any @option{-std} option used.
1763 Valid ISO C and ISO C++ programs should compile properly with or without
1764 this option (though a rare few will require @option{-ansi} or a
1765 @option{-std} option specifying the required version of ISO C)@. However,
1766 without this option, certain GNU extensions and traditional C and C++
1767 features are supported as well. With this option, they are rejected.
1769 @option{-pedantic} does not cause warning messages for use of the
1770 alternate keywords whose names begin and end with @samp{__}. Pedantic
1771 warnings are also disabled in the expression that follows
1772 @code{__extension__}. However, only system header files should use
1773 these escape routes; application programs should avoid them.
1774 @xref{Alternate Keywords}.
1776 Some users try to use @option{-pedantic} to check programs for strict ISO
1777 C conformance. They soon find that it does not do quite what they want:
1778 it finds some non-ISO practices, but not all---only those for which
1779 ISO C @emph{requires} a diagnostic, and some others for which
1780 diagnostics have been added.
1782 A feature to report any failure to conform to ISO C might be useful in
1783 some instances, but would require considerable additional work and would
1784 be quite different from @option{-pedantic}. We don't have plans to
1785 support such a feature in the near future.
1787 Where the standard specified with @option{-std} represents a GNU
1788 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
1789 corresponding @dfn{base standard}, the version of ISO C on which the GNU
1790 extended dialect is based. Warnings from @option{-pedantic} are given
1791 where they are required by the base standard. (It would not make sense
1792 for such warnings to be given only for features not in the specified GNU
1793 C dialect, since by definition the GNU dialects of C include all
1794 features the compiler supports with the given option, and there would be
1795 nothing to warn about.)
1797 @item -pedantic-errors
1798 @opindex pedantic-errors
1799 Like @option{-pedantic}, except that errors are produced rather than
1804 Inhibit all warning messages.
1808 Inhibit warning messages about the use of @samp{#import}.
1810 @item -Wchar-subscripts
1811 @opindex Wchar-subscripts
1812 Warn if an array subscript has type @code{char}. This is a common cause
1813 of error, as programmers often forget that this type is signed on some
1818 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
1819 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
1823 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
1824 the arguments supplied have types appropriate to the format string
1825 specified, and that the conversions specified in the format string make
1826 sense. This includes standard functions, and others specified by format
1827 attributes (@pxref{Function Attributes}), in the @code{printf},
1828 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
1829 not in the C standard) families.
1831 The formats are checked against the format features supported by GNU
1832 libc version 2.2. These include all ISO C89 and C99 features, as well
1833 as features from the Single Unix Specification and some BSD and GNU
1834 extensions. Other library implementations may not support all these
1835 features; GCC does not support warning about features that go beyond a
1836 particular library's limitations. However, if @option{-pedantic} is used
1837 with @option{-Wformat}, warnings will be given about format features not
1838 in the selected standard version (but not for @code{strfmon} formats,
1839 since those are not in any version of the C standard). @xref{C Dialect
1840 Options,,Options Controlling C Dialect}.
1842 @option{-Wformat} is included in @option{-Wall}. For more control over some
1843 aspects of format checking, the options @option{-Wno-format-y2k},
1844 @option{-Wno-format-extra-args}, @option{-Wformat-nonliteral},
1845 @option{-Wformat-security} and @option{-Wformat=2} are available, but are
1846 not included in @option{-Wall}.
1848 @item -Wno-format-y2k
1849 @opindex Wno-format-y2k
1850 If @option{-Wformat} is specified, do not warn about @code{strftime}
1851 formats which may yield only a two-digit year.
1853 @item -Wno-format-extra-args
1854 @opindex Wno-format-extra-args
1855 If @option{-Wformat} is specified, do not warn about excess arguments to a
1856 @code{printf} or @code{scanf} format function. The C standard specifies
1857 that such arguments are ignored.
1859 Where the unused arguments lie between used arguments that are
1860 specified with @samp{$} operand number specifications, normally
1861 warnings are still given, since the implementation could not know what
1862 type to pass to @code{va_arg} to skip the unused arguments. However,
1863 in the case of @code{scanf} formats, this option will suppress the
1864 warning if the unused arguments are all pointers, since the Single
1865 Unix Specification says that such unused arguments are allowed.
1867 @item -Wformat-nonliteral
1868 @opindex Wformat-nonliteral
1869 If @option{-Wformat} is specified, also warn if the format string is not a
1870 string literal and so cannot be checked, unless the format function
1871 takes its format arguments as a @code{va_list}.
1873 @item -Wformat-security
1874 @opindex Wformat-security
1875 If @option{-Wformat} is specified, also warn about uses of format
1876 functions that represent possible security problems. At present, this
1877 warns about calls to @code{printf} and @code{scanf} functions where the
1878 format string is not a string literal and there are no format arguments,
1879 as in @code{printf (foo);}. This may be a security hole if the format
1880 string came from untrusted input and contains @samp{%n}. (This is
1881 currently a subset of what @option{-Wformat-nonliteral} warns about, but
1882 in future warnings may be added to @option{-Wformat-security} that are not
1883 included in @option{-Wformat-nonliteral}.)
1887 Enable @option{-Wformat} plus format checks not included in
1888 @option{-Wformat}. Currently equivalent to @samp{-Wformat
1889 -Wformat-nonliteral -Wformat-security}.
1891 @item -Wimplicit-int
1892 @opindex Wimplicit-int
1893 Warn when a declaration does not specify a type.
1895 @item -Wimplicit-function-declaration
1896 @itemx -Werror-implicit-function-declaration
1897 @opindex Wimplicit-function-declaration
1898 @opindex Werror-implicit-function-declaration
1899 Give a warning (or error) whenever a function is used before being
1904 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
1908 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
1909 function with external linkage, returning int, taking either zero
1910 arguments, two, or three arguments of appropriate types.
1912 @item -Wmissing-braces
1913 @opindex Wmissing-braces
1914 Warn if an aggregate or union initializer is not fully bracketed. In
1915 the following example, the initializer for @samp{a} is not fully
1916 bracketed, but that for @samp{b} is fully bracketed.
1919 int a[2][2] = @{ 0, 1, 2, 3 @};
1920 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
1924 @opindex Wparentheses
1925 Warn if parentheses are omitted in certain contexts, such
1926 as when there is an assignment in a context where a truth value
1927 is expected, or when operators are nested whose precedence people
1928 often get confused about.
1930 Also warn about constructions where there may be confusion to which
1931 @code{if} statement an @code{else} branch belongs. Here is an example of
1946 In C, every @code{else} branch belongs to the innermost possible @code{if}
1947 statement, which in this example is @code{if (b)}. This is often not
1948 what the programmer expected, as illustrated in the above example by
1949 indentation the programmer chose. When there is the potential for this
1950 confusion, GCC will issue a warning when this flag is specified.
1951 To eliminate the warning, add explicit braces around the innermost
1952 @code{if} statement so there is no way the @code{else} could belong to
1953 the enclosing @code{if}. The resulting code would look like this:
1969 @item -Wsequence-point
1970 @opindex Wsequence-point
1971 Warn about code that may have undefined semantics because of violations
1972 of sequence point rules in the C standard.
1974 The C standard defines the order in which expressions in a C program are
1975 evaluated in terms of @dfn{sequence points}, which represent a partial
1976 ordering between the execution of parts of the program: those executed
1977 before the sequence point, and those executed after it. These occur
1978 after the evaluation of a full expression (one which is not part of a
1979 larger expression), after the evaluation of the first operand of a
1980 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
1981 function is called (but after the evaluation of its arguments and the
1982 expression denoting the called function), and in certain other places.
1983 Other than as expressed by the sequence point rules, the order of
1984 evaluation of subexpressions of an expression is not specified. All
1985 these rules describe only a partial order rather than a total order,
1986 since, for example, if two functions are called within one expression
1987 with no sequence point between them, the order in which the functions
1988 are called is not specified. However, the standards committee have
1989 ruled that function calls do not overlap.
1991 It is not specified when between sequence points modifications to the
1992 values of objects take effect. Programs whose behavior depends on this
1993 have undefined behavior; the C standard specifies that ``Between the
1994 previous and next sequence point an object shall have its stored value
1995 modified at most once by the evaluation of an expression. Furthermore,
1996 the prior value shall be read only to determine the value to be
1997 stored.''. If a program breaks these rules, the results on any
1998 particular implementation are entirely unpredictable.
2000 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
2001 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
2002 diagnosed by this option, and it may give an occasional false positive
2003 result, but in general it has been found fairly effective at detecting
2004 this sort of problem in programs.
2006 The present implementation of this option only works for C programs. A
2007 future implementation may also work for C++ programs.
2009 The C standard is worded confusingly, therefore there is some debate
2010 over the precise meaning of the sequence point rules in subtle cases.
2011 Links to discussions of the problem, including proposed formal
2012 definitions, may be found on our readings page, at
2013 @w{@uref{http://gcc.gnu.org/readings.html}}.
2016 @opindex Wreturn-type
2017 Warn whenever a function is defined with a return-type that defaults to
2018 @code{int}. Also warn about any @code{return} statement with no
2019 return-value in a function whose return-type is not @code{void}.
2021 For C++, a function without return type always produces a diagnostic
2022 message, even when @option{-Wno-return-type} is specified. The only
2023 exceptions are @samp{main} and functions defined in system headers.
2027 Warn whenever a @code{switch} statement has an index of enumeral type
2028 and lacks a @code{case} for one or more of the named codes of that
2029 enumeration. (The presence of a @code{default} label prevents this
2030 warning.) @code{case} labels outside the enumeration range also
2031 provoke warnings when this option is used.
2035 Warn if any trigraphs are encountered that might change the meaning of
2036 the program (trigraphs within comments are not warned about).
2038 @item -Wunused-function
2039 @opindex Wunused-function
2040 Warn whenever a static function is declared but not defined or a
2041 non\-inline static function is unused.
2043 @item -Wunused-label
2044 @opindex Wunused-label
2045 Warn whenever a label is declared but not used.
2047 To suppress this warning use the @samp{unused} attribute
2048 (@pxref{Variable Attributes}).
2050 @item -Wunused-parameter
2051 @opindex Wunused-parameter
2052 Warn whenever a function parameter is unused aside from its declaration.
2054 To suppress this warning use the @samp{unused} attribute
2055 (@pxref{Variable Attributes}).
2057 @item -Wunused-variable
2058 @opindex Wunused-variable
2059 Warn whenever a local variable or non-constant static variable is unused
2060 aside from its declaration
2062 To suppress this warning use the @samp{unused} attribute
2063 (@pxref{Variable Attributes}).
2065 @item -Wunused-value
2066 @opindex Wunused-value
2067 Warn whenever a statement computes a result that is explicitly not used.
2069 To suppress this warning cast the expression to @samp{void}.
2073 All all the above @option{-Wunused} options combined.
2075 In order to get a warning about an unused function parameter, you must
2076 either specify @samp{-W -Wunused} or separately specify
2077 @option{-Wunused-parameter}.
2079 @item -Wuninitialized
2080 @opindex Wuninitialized
2081 Warn if an automatic variable is used without first being initialized or
2082 if a variable may be clobbered by a @code{setjmp} call.
2084 These warnings are possible only in optimizing compilation,
2085 because they require data flow information that is computed only
2086 when optimizing. If you don't specify @option{-O}, you simply won't
2089 These warnings occur only for variables that are candidates for
2090 register allocation. Therefore, they do not occur for a variable that
2091 is declared @code{volatile}, or whose address is taken, or whose size
2092 is other than 1, 2, 4 or 8 bytes. Also, they do not occur for
2093 structures, unions or arrays, even when they are in registers.
2095 Note that there may be no warning about a variable that is used only
2096 to compute a value that itself is never used, because such
2097 computations may be deleted by data flow analysis before the warnings
2100 These warnings are made optional because GCC is not smart
2101 enough to see all the reasons why the code might be correct
2102 despite appearing to have an error. Here is one example of how
2123 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
2124 always initialized, but GCC doesn't know this. Here is
2125 another common case:
2130 if (change_y) save_y = y, y = new_y;
2132 if (change_y) y = save_y;
2137 This has no bug because @code{save_y} is used only if it is set.
2139 @cindex @code{longjmp} warnings
2140 This option also warns when a non-volatile automatic variable might be
2141 changed by a call to @code{longjmp}. These warnings as well are possible
2142 only in optimizing compilation.
2144 The compiler sees only the calls to @code{setjmp}. It cannot know
2145 where @code{longjmp} will be called; in fact, a signal handler could
2146 call it at any point in the code. As a result, you may get a warning
2147 even when there is in fact no problem because @code{longjmp} cannot
2148 in fact be called at the place which would cause a problem.
2150 Some spurious warnings can be avoided if you declare all the functions
2151 you use that never return as @code{noreturn}. @xref{Function
2154 @item -Wreorder @r{(C++ only)}
2156 @cindex reordering, warning
2157 @cindex warning for reordering of member initializers
2158 Warn when the order of member initializers given in the code does not
2159 match the order in which they must be executed. For instance:
2161 @item -Wunknown-pragmas
2162 @opindex Wunknown-pragmas
2163 @cindex warning for unknown pragmas
2164 @cindex unknown pragmas, warning
2165 @cindex pragmas, warning of unknown
2166 Warn when a #pragma directive is encountered which is not understood by
2167 GCC@. If this command line option is used, warnings will even be issued
2168 for unknown pragmas in system header files. This is not the case if
2169 the warnings were only enabled by the @option{-Wall} command line option.
2173 All of the above @samp{-W} options combined. This enables all the
2174 warnings about constructions that some users consider questionable, and
2175 that are easy to avoid (or modify to prevent the warning), even in
2176 conjunction with macros.
2179 @opindex Wno-div-by-zero
2180 @opindex Wdiv-by-zero
2181 Warn about compile-time integer division by zero. This is default. To
2182 inhibit the warning messages, use @option{-Wno-div-by-zero}. Floating
2183 point division by zero is not warned about, as it can be a legitimate
2184 way of obtaining infinities and NaNs.
2187 @opindex Wno-multichar
2189 Warn if a multicharacter constant (@samp{'FOOF'}) is used. This is
2190 default. To inhibit the warning messages, use @option{-Wno-multichar}.
2191 Usually they indicate a typo in the user's code, as they have
2192 implementation-defined values, and should not be used in portable code.
2194 @item -Wsystem-headers
2195 @opindex Wsystem-headers
2196 @cindex warnings from system headers
2197 @cindex system headers, warnings from
2198 Print warning messages for constructs found in system header files.
2199 Warnings from system headers are normally suppressed, on the assumption
2200 that they usually do not indicate real problems and would only make the
2201 compiler output harder to read. Using this command line option tells
2202 GCC to emit warnings from system headers as if they occurred in user
2203 code. However, note that using @option{-Wall} in conjunction with this
2204 option will @emph{not} warn about unknown pragmas in system
2205 headers---for that, @option{-Wunknown-pragmas} must also be used.
2208 The following @option{-W@dots{}} options are not implied by @option{-Wall}.
2209 Some of them warn about constructions that users generally do not
2210 consider questionable, but which occasionally you might wish to check
2211 for; others warn about constructions that are necessary or hard to avoid
2212 in some cases, and there is no simple way to modify the code to suppress
2218 Print extra warning messages for these events:
2222 A function can return either with or without a value. (Falling
2223 off the end of the function body is considered returning without
2224 a value.) For example, this function would evoke such a
2238 An expression-statement or the left-hand side of a comma expression
2239 contains no side effects.
2240 To suppress the warning, cast the unused expression to void.
2241 For example, an expression such as @samp{x[i,j]} will cause a warning,
2242 but @samp{x[(void)i,j]} will not.
2245 An unsigned value is compared against zero with @samp{<} or @samp{<=}.
2248 A comparison like @samp{x<=y<=z} appears; this is equivalent to
2249 @samp{(x<=y ? 1 : 0) <= z}, which is a different interpretation from
2250 that of ordinary mathematical notation.
2253 Storage-class specifiers like @code{static} are not the first things in
2254 a declaration. According to the C Standard, this usage is obsolescent.
2257 The return type of a function has a type qualifier such as @code{const}.
2258 Such a type qualifier has no effect, since the value returned by a
2259 function is not an lvalue. (But don't warn about the GNU extension of
2260 @code{volatile void} return types. That extension will be warned about
2261 if @option{-pedantic} is specified.)
2264 If @option{-Wall} or @option{-Wunused} is also specified, warn about unused
2268 A comparison between signed and unsigned values could produce an
2269 incorrect result when the signed value is converted to unsigned.
2270 (But don't warn if @option{-Wno-sign-compare} is also specified.)
2273 An aggregate has a partly bracketed initializer.
2274 For example, the following code would evoke such a warning,
2275 because braces are missing around the initializer for @code{x.h}:
2278 struct s @{ int f, g; @};
2279 struct t @{ struct s h; int i; @};
2280 struct t x = @{ 1, 2, 3 @};
2284 An aggregate has an initializer which does not initialize all members.
2285 For example, the following code would cause such a warning, because
2286 @code{x.h} would be implicitly initialized to zero:
2289 struct s @{ int f, g, h; @};
2290 struct s x = @{ 3, 4 @};
2295 @opindex Wfloat-equal
2296 Warn if floating point values are used in equality comparisons.
2298 The idea behind this is that sometimes it is convenient (for the
2299 programmer) to consider floating-point values as approximations to
2300 infinitely precise real numbers. If you are doing this, then you need
2301 to compute (by analysing the code, or in some other way) the maximum or
2302 likely maximum error that the computation introduces, and allow for it
2303 when performing comparisons (and when producing output, but that's a
2304 different problem). In particular, instead of testing for equality, you
2305 would check to see whether the two values have ranges that overlap; and
2306 this is done with the relational operators, so equality comparisons are
2309 @item -Wtraditional @r{(C only)}
2310 @opindex Wtraditional
2311 Warn about certain constructs that behave differently in traditional and
2312 ISO C@. Also warn about ISO C constructs that have no traditional C
2313 equivalent, and/or problematic constructs which should be avoided.
2317 Macro parameters that appear within string literals in the macro body.
2318 In traditional C macro replacement takes place within string literals,
2319 but does not in ISO C@.
2322 In traditional C, some preprocessor directives did not exist.
2323 Traditional preprocessors would only consider a line to be a directive
2324 if the @samp{#} appeared in column 1 on the line. Therefore
2325 @option{-Wtraditional} warns about directives that traditional C
2326 understands but would ignore because the @samp{#} does not appear as the
2327 first character on the line. It also suggests you hide directives like
2328 @samp{#pragma} not understood by traditional C by indenting them. Some
2329 traditional implementations would not recognize @samp{#elif}, so it
2330 suggests avoiding it altogether.
2333 A function-like macro that appears without arguments.
2336 The unary plus operator.
2339 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
2340 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
2341 constants.) Note, these suffixes appear in macros defined in the system
2342 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
2343 Use of these macros in user code might normally lead to spurious
2344 warnings, however gcc's integrated preprocessor has enough context to
2345 avoid warning in these cases.
2348 A function declared external in one block and then used after the end of
2352 A @code{switch} statement has an operand of type @code{long}.
2355 A non-@code{static} function declaration follows a @code{static} one.
2356 This construct is not accepted by some traditional C compilers.
2359 The ISO type of an integer constant has a different width or
2360 signedness from its traditional type. This warning is only issued if
2361 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
2362 typically represent bit patterns, are not warned about.
2365 Usage of ISO string concatenation is detected.
2368 Initialization of automatic aggregates.
2371 Identifier conflicts with labels. Traditional C lacks a separate
2372 namespace for labels.
2375 Initialization of unions. If the initializer is zero, the warning is
2376 omitted. This is done under the assumption that the zero initializer in
2377 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
2378 initializer warnings and relies on default initialization to zero in the
2382 Conversions by prototypes between fixed/floating point values and vice
2383 versa. The absence of these prototypes when compiling with traditional
2384 C would cause serious problems. This is a subset of the possible
2385 conversion warnings, for the full set use @option{-Wconversion}.
2390 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
2394 Warn whenever a local variable shadows another local variable, parameter or
2395 global variable or whenever a built-in function is shadowed.
2397 @item -Wlarger-than-@var{len}
2398 @opindex Wlarger-than
2399 Warn whenever an object of larger than @var{len} bytes is defined.
2401 @item -Wpointer-arith
2402 @opindex Wpointer-arith
2403 Warn about anything that depends on the ``size of'' a function type or
2404 of @code{void}. GNU C assigns these types a size of 1, for
2405 convenience in calculations with @code{void *} pointers and pointers
2408 @item -Wbad-function-cast @r{(C only)}
2409 @opindex Wbad-function-cast
2410 Warn whenever a function call is cast to a non-matching type.
2411 For example, warn if @code{int malloc()} is cast to @code{anything *}.
2415 Warn whenever a pointer is cast so as to remove a type qualifier from
2416 the target type. For example, warn if a @code{const char *} is cast
2417 to an ordinary @code{char *}.
2420 @opindex Wcast-align
2421 Warn whenever a pointer is cast such that the required alignment of the
2422 target is increased. For example, warn if a @code{char *} is cast to
2423 an @code{int *} on machines where integers can only be accessed at
2424 two- or four-byte boundaries.
2426 @item -Wwrite-strings
2427 @opindex Wwrite-strings
2428 When compiling C, give string constants the type @code{const
2429 char[@var{length}]} so that
2430 copying the address of one into a non-@code{const} @code{char *}
2431 pointer will get a warning; when compiling C++, warn about the
2432 deprecated conversion from string constants to @code{char *}.
2433 These warnings will help you find at
2434 compile time code that can try to write into a string constant, but
2435 only if you have been very careful about using @code{const} in
2436 declarations and prototypes. Otherwise, it will just be a nuisance;
2437 this is why we did not make @option{-Wall} request these warnings.
2440 @opindex Wconversion
2441 Warn if a prototype causes a type conversion that is different from what
2442 would happen to the same argument in the absence of a prototype. This
2443 includes conversions of fixed point to floating and vice versa, and
2444 conversions changing the width or signedness of a fixed point argument
2445 except when the same as the default promotion.
2447 Also, warn if a negative integer constant expression is implicitly
2448 converted to an unsigned type. For example, warn about the assignment
2449 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
2450 casts like @code{(unsigned) -1}.
2452 @item -Wsign-compare
2453 @opindex Wsign-compare
2454 @cindex warning for comparison of signed and unsigned values
2455 @cindex comparison of signed and unsigned values, warning
2456 @cindex signed and unsigned values, comparison warning
2457 Warn when a comparison between signed and unsigned values could produce
2458 an incorrect result when the signed value is converted to unsigned.
2459 This warning is also enabled by @option{-W}; to get the other warnings
2460 of @option{-W} without this warning, use @samp{-W -Wno-sign-compare}.
2462 @item -Waggregate-return
2463 @opindex Waggregate-return
2464 Warn if any functions that return structures or unions are defined or
2465 called. (In languages where you can return an array, this also elicits
2468 @item -Wstrict-prototypes @r{(C only)}
2469 @opindex Wstrict-prototypes
2470 Warn if a function is declared or defined without specifying the
2471 argument types. (An old-style function definition is permitted without
2472 a warning if preceded by a declaration which specifies the argument
2475 @item -Wmissing-prototypes @r{(C only)}
2476 @opindex Wmissing-prototypes
2477 Warn if a global function is defined without a previous prototype
2478 declaration. This warning is issued even if the definition itself
2479 provides a prototype. The aim is to detect global functions that fail
2480 to be declared in header files.
2482 @item -Wmissing-declarations
2483 @opindex Wmissing-declarations
2484 Warn if a global function is defined without a previous declaration.
2485 Do so even if the definition itself provides a prototype.
2486 Use this option to detect global functions that are not declared in
2489 @item -Wmissing-noreturn
2490 @opindex Wmissing-noreturn
2491 Warn about functions which might be candidates for attribute @code{noreturn}.
2492 Note these are only possible candidates, not absolute ones. Care should
2493 be taken to manually verify functions actually do not ever return before
2494 adding the @code{noreturn} attribute, otherwise subtle code generation
2495 bugs could be introduced. You will not get a warning for @code{main} in
2496 hosted C environments.
2498 @item -Wmissing-format-attribute
2499 @opindex Wmissing-format-attribute
2501 If @option{-Wformat} is enabled, also warn about functions which might be
2502 candidates for @code{format} attributes. Note these are only possible
2503 candidates, not absolute ones. GCC will guess that @code{format}
2504 attributes might be appropriate for any function that calls a function
2505 like @code{vprintf} or @code{vscanf}, but this might not always be the
2506 case, and some functions for which @code{format} attributes are
2507 appropriate may not be detected. This option has no effect unless
2508 @option{-Wformat} is enabled (possibly by @option{-Wall}).
2510 @item -Wno-deprecated-declarations
2511 @opindex Wno-deprecated-declarations
2512 Do not warn about uses of functions, variables, and types marked as
2513 deprecated by using the @code{deprecated} attribute.
2514 (@pxref{Function Attributes}, @pxref{Variable Attributes},
2515 @pxref{Type Attributes}.)
2519 Warn if a structure is given the packed attribute, but the packed
2520 attribute has no effect on the layout or size of the structure.
2521 Such structures may be mis-aligned for little benefit. For
2522 instance, in this code, the variable @code{f.x} in @code{struct bar}
2523 will be misaligned even though @code{struct bar} does not itself
2524 have the packed attribute:
2531 @} __attribute__((packed));
2541 Warn if padding is included in a structure, either to align an element
2542 of the structure or to align the whole structure. Sometimes when this
2543 happens it is possible to rearrange the fields of the structure to
2544 reduce the padding and so make the structure smaller.
2546 @item -Wredundant-decls
2547 @opindex Wredundant-decls
2548 Warn if anything is declared more than once in the same scope, even in
2549 cases where multiple declaration is valid and changes nothing.
2551 @item -Wnested-externs @r{(C only)}
2552 @opindex Wnested-externs
2553 Warn if an @code{extern} declaration is encountered within a function.
2555 @item -Wunreachable-code
2556 @opindex Wunreachable-code
2557 Warn if the compiler detects that code will never be executed.
2559 This option is intended to warn when the compiler detects that at
2560 least a whole line of source code will never be executed, because
2561 some condition is never satisfied or because it is after a
2562 procedure that never returns.
2564 It is possible for this option to produce a warning even though there
2565 are circumstances under which part of the affected line can be executed,
2566 so care should be taken when removing apparently-unreachable code.
2568 For instance, when a function is inlined, a warning may mean that the
2569 line is unreachable in only one inlined copy of the function.
2571 This option is not made part of @option{-Wall} because in a debugging
2572 version of a program there is often substantial code which checks
2573 correct functioning of the program and is, hopefully, unreachable
2574 because the program does work. Another common use of unreachable
2575 code is to provide behavior which is selectable at compile-time.
2579 Warn if a function can not be inlined and it was declared as inline.
2583 @opindex Wno-long-long
2584 Warn if @samp{long long} type is used. This is default. To inhibit
2585 the warning messages, use @option{-Wno-long-long}. Flags
2586 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
2587 only when @option{-pedantic} flag is used.
2589 @item -Wdisabled-optimization
2590 @opindex Wdisabled-optimization
2591 Warn if a requested optimization pass is disabled. This warning does
2592 not generally indicate that there is anything wrong with your code; it
2593 merely indicates that GCC's optimizers were unable to handle the code
2594 effectively. Often, the problem is that your code is too big or too
2595 complex; GCC will refuse to optimize programs when the optimization
2596 itself is likely to take inordinate amounts of time.
2600 Make all warnings into errors.
2603 @node Debugging Options
2604 @section Options for Debugging Your Program or GCC
2605 @cindex options, debugging
2606 @cindex debugging information options
2608 GCC has various special options that are used for debugging
2609 either your program or GCC:
2614 Produce debugging information in the operating system's native format
2615 (stabs, COFF, XCOFF, or DWARF)@. GDB can work with this debugging
2618 On most systems that use stabs format, @option{-g} enables use of extra
2619 debugging information that only GDB can use; this extra information
2620 makes debugging work better in GDB but will probably make other debuggers
2622 refuse to read the program. If you want to control for certain whether
2623 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
2624 @option{-gxcoff+}, @option{-gxcoff}, @option{-gdwarf-1+}, @option{-gdwarf-1},
2625 or @option{-gvms} (see below).
2627 Unlike most other C compilers, GCC allows you to use @option{-g} with
2628 @option{-O}. The shortcuts taken by optimized code may occasionally
2629 produce surprising results: some variables you declared may not exist
2630 at all; flow of control may briefly move where you did not expect it;
2631 some statements may not be executed because they compute constant
2632 results or their values were already at hand; some statements may
2633 execute in different places because they were moved out of loops.
2635 Nevertheless it proves possible to debug optimized output. This makes
2636 it reasonable to use the optimizer for programs that might have bugs.
2638 The following options are useful when GCC is generated with the
2639 capability for more than one debugging format.
2643 Produce debugging information for use by GDB@. This means to use the
2644 most expressive format available (DWARF 2, stabs, or the native format
2645 if neither of those are supported), including GDB extensions if at all
2650 Produce debugging information in stabs format (if that is supported),
2651 without GDB extensions. This is the format used by DBX on most BSD
2652 systems. On MIPS, Alpha and System V Release 4 systems this option
2653 produces stabs debugging output which is not understood by DBX or SDB@.
2654 On System V Release 4 systems this option requires the GNU assembler.
2658 Produce debugging information in stabs format (if that is supported),
2659 using GNU extensions understood only by the GNU debugger (GDB)@. The
2660 use of these extensions is likely to make other debuggers crash or
2661 refuse to read the program.
2665 Produce debugging information in COFF format (if that is supported).
2666 This is the format used by SDB on most System V systems prior to
2671 Produce debugging information in XCOFF format (if that is supported).
2672 This is the format used by the DBX debugger on IBM RS/6000 systems.
2676 Produce debugging information in XCOFF format (if that is supported),
2677 using GNU extensions understood only by the GNU debugger (GDB)@. The
2678 use of these extensions is likely to make other debuggers crash or
2679 refuse to read the program, and may cause assemblers other than the GNU
2680 assembler (GAS) to fail with an error.
2684 Produce debugging information in DWARF version 1 format (if that is
2685 supported). This is the format used by SDB on most System V Release 4
2690 Produce debugging information in DWARF version 1 format (if that is
2691 supported), using GNU extensions understood only by the GNU debugger
2692 (GDB)@. The use of these extensions is likely to make other debuggers
2693 crash or refuse to read the program.
2697 Produce debugging information in DWARF version 2 format (if that is
2698 supported). This is the format used by DBX on IRIX 6.
2702 Produce debugging information in VMS debug format (if that is
2703 supported). This is the format used by DEBUG on VMS systems.
2706 @itemx -ggdb@var{level}
2707 @itemx -gstabs@var{level}
2708 @itemx -gcoff@var{level}
2709 @itemx -gxcoff@var{level}
2710 @itemx -gdwarf@var{level}
2711 @itemx -gdwarf-2@var{level}
2712 @itemx -gvms@var{level}
2713 Request debugging information and also use @var{level} to specify how
2714 much information. The default level is 2.
2716 Level 1 produces minimal information, enough for making backtraces in
2717 parts of the program that you don't plan to debug. This includes
2718 descriptions of functions and external variables, but no information
2719 about local variables and no line numbers.
2721 Level 3 includes extra information, such as all the macro definitions
2722 present in the program. Some debuggers support macro expansion when
2723 you use @option{-g3}.
2728 Generate extra code to write profile information suitable for the
2729 analysis program @code{prof}. You must use this option when compiling
2730 the source files you want data about, and you must also use it when
2733 @cindex @code{gprof}
2736 Generate extra code to write profile information suitable for the
2737 analysis program @code{gprof}. You must use this option when compiling
2738 the source files you want data about, and you must also use it when
2744 Generate extra code to write profile information for basic blocks, which will
2745 record the number of times each basic block is executed, the basic block start
2746 address, and the function name containing the basic block. If @option{-g} is
2747 used, the line number and filename of the start of the basic block will also be
2748 recorded. If not overridden by the machine description, the default action is
2749 to append to the text file @file{bb.out}.
2751 This data could be analyzed by a program like @code{tcov}. Note,
2752 however, that the format of the data is not what @code{tcov} expects.
2753 Eventually GNU @code{gprof} should be extended to process this data.
2757 Makes the compiler print out each function name as it is compiled, and
2758 print some statistics about each pass when it finishes.
2761 @opindex ftime-report
2762 Makes the compiler print some statistics about the time consumed by each
2763 pass when it finishes.
2766 @opindex fmem-report
2767 Makes the compiler print some statistics about permanent memory
2768 allocation when it finishes.
2770 @item -fprofile-arcs
2771 @opindex fprofile-arcs
2772 Instrument @dfn{arcs} during compilation to generate coverage data
2773 or for profile-directed block ordering. During execution the program
2774 records how many times each branch is executed and how many times it is
2775 taken. When the compiled program exits it saves this data to a file
2776 called @file{@var{sourcename}.da} for each source file.
2778 For profile-directed block ordering, compile the program with
2779 @option{-fprofile-arcs} plus optimization and code generation options,
2780 generate the arc profile information by running the program on a
2781 selected workload, and then compile the program again with the same
2782 optimization and code generation options plus
2783 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
2784 Control Optimization}).
2786 The other use of @option{-fprofile-arcs} is for use with @code{gcov},
2787 when it is used with the @option{-ftest-coverage} option. GCC
2788 supports two methods of determining code coverage: the options that
2789 support @code{gcov}, and options @option{-a} and @option{-ax}, which
2790 write information to text files. The options that support @code{gcov}
2791 do not need to instrument every arc in the program, so a program compiled
2792 with them runs faster than a program compiled with @option{-a}, which
2793 adds instrumentation code to every basic block in the program. The
2794 tradeoff: since @code{gcov} does not have execution counts for all
2795 branches, it must start with the execution counts for the instrumented
2796 branches, and then iterate over the program flow graph until the entire
2797 graph has been solved. Hence, @code{gcov} runs a little more slowly than
2798 a program which uses information from @option{-a} and @option{-ax}.
2800 With @option{-fprofile-arcs}, for each function of your program GCC
2801 creates a program flow graph, then finds a spanning tree for the graph.
2802 Only arcs that are not on the spanning tree have to be instrumented: the
2803 compiler adds code to count the number of times that these arcs are
2804 executed. When an arc is the only exit or only entrance to a block, the
2805 instrumentation code can be added to the block; otherwise, a new basic
2806 block must be created to hold the instrumentation code.
2808 This option makes it possible to estimate branch probabilities and to
2809 calculate basic block execution counts. In general, basic block
2810 execution counts as provided by @option{-a} do not give enough
2811 information to estimate all branch probabilities.
2814 @item -ftest-coverage
2815 @opindex ftest-coverage
2816 Create data files for the @code{gcov} code-coverage utility
2817 (@pxref{Gcov,, @code{gcov}: a GCC Test Coverage Program}).
2818 The data file names begin with the name of your source file:
2821 @item @var{sourcename}.bb
2822 A mapping from basic blocks to line numbers, which @code{gcov} uses to
2823 associate basic block execution counts with line numbers.
2825 @item @var{sourcename}.bbg
2826 A list of all arcs in the program flow graph. This allows @code{gcov}
2827 to reconstruct the program flow graph, so that it can compute all basic
2828 block and arc execution counts from the information in the
2829 @code{@var{sourcename}.da} file.
2832 Use @option{-ftest-coverage} with @option{-fprofile-arcs}; the latter
2833 option adds instrumentation to the program, which then writes
2834 execution counts to another data file:
2837 @item @var{sourcename}.da
2838 Runtime arc execution counts, used in conjunction with the arc
2839 information in the file @code{@var{sourcename}.bbg}.
2842 Coverage data will map better to the source files if
2843 @option{-ftest-coverage} is used without optimization.
2845 @item -d@var{letters}
2847 Says to make debugging dumps during compilation at times specified by
2848 @var{letters}. This is used for debugging the compiler. The file names
2849 for most of the dumps are made by appending a pass number and a word to
2850 the source file name (e.g. @file{foo.c.00.rtl} or @file{foo.c.01.sibling}).
2851 Here are the possible letters for use in @var{letters}, and their meanings:
2856 Annotate the assembler output with miscellaneous debugging information.
2859 Dump after computing branch probabilities, to @file{@var{file}.14.bp}.
2862 Dump after block reordering, to @file{@var{file}.28.bbro}.
2865 Dump after instruction combination, to the file @file{@var{file}.16.combine}.
2868 Dump after the first if conversion, to the file @file{@var{file}.17.ce}.
2871 Dump after delayed branch scheduling, to @file{@var{file}.31.dbr}.
2874 Dump all macro definitions, at the end of preprocessing, in addition to
2878 Dump after SSA optimizations, to @file{@var{file}.04.ssa} and
2879 @file{@var{file}.07.ussa}.
2882 Dump after the second if conversion, to @file{@var{file}.26.ce2}.
2885 Dump after life analysis, to @file{@var{file}.15.life}.
2888 Dump after purging @code{ADDRESSOF} codes, to @file{@var{file}.09.addressof}.
2891 Dump after global register allocation, to @file{@var{file}.21.greg}.
2894 Dump after finalization of EH handling code, to @file{@var{file}.02.eh}.
2897 Dump after post-reload optimizations, to @file{@var{file}.22.postreload}.
2900 Dump after GCSE, to @file{@var{file}.10.gcse}.
2903 Dump after sibling call optimizations, to @file{@var{file}.01.sibling}.
2906 Dump after the first jump optimization, to @file{@var{file}.03.jump}.
2909 Dump after conversion from registers to stack, to @file{@var{file}.32.stack}.
2912 Dump after local register allocation, to @file{@var{file}.20.lreg}.
2915 Dump after loop optimization, to @file{@var{file}.11.loop}.
2918 Dump after performing the machine dependent reorganisation pass, to
2919 @file{@var{file}.30.mach}.
2922 Dump after register renumbering, to @file{@var{file}.25.rnreg}.
2925 Dump after the register move pass, to @file{@var{file}.18.regmove}.
2928 Dump after RTL generation, to @file{@var{file}.00.rtl}.
2931 Dump after the second scheduling pass, to @file{@var{file}.27.sched2}.
2934 Dump after CSE (including the jump optimization that sometimes follows
2935 CSE), to @file{@var{file}.08.cse}.
2938 Dump after the first scheduling pass, to @file{@var{file}.19.sched}.
2941 Dump after the second CSE pass (including the jump optimization that
2942 sometimes follows CSE), to @file{@var{file}.12.cse2}.
2945 Dump after the second flow pass, to @file{@var{file}.23.flow2}.
2948 Dump after SSA dead code elimination, to @file{@var{file}.06.ssadce}.
2951 Dump after the peephole pass, to @file{@var{file}.24.peephole2}.
2954 Produce all the dumps listed above.
2957 Print statistics on memory usage, at the end of the run, to
2961 Annotate the assembler output with a comment indicating which
2962 pattern and alternative was used. The length of each instruction is
2966 Dump the RTL in the assembler output as a comment before each instruction.
2967 Also turns on @option{-dp} annotation.
2970 For each of the other indicated dump files (except for
2971 @file{@var{file}.00.rtl}), dump a representation of the control flow graph
2972 suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
2975 Just generate RTL for a function instead of compiling it. Usually used
2979 Dump debugging information during parsing, to standard error.
2982 @item -fdump-unnumbered
2983 @opindex fdump-unnumbered
2984 When doing debugging dumps (see @option{-d} option above), suppress instruction
2985 numbers and line number note output. This makes it more feasible to
2986 use diff on debugging dumps for compiler invocations with different
2987 options, in particular with and without @option{-g}.
2989 @item -fdump-class-hierarchy @r{(C++ only)}
2990 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
2991 @opindex fdump-class-hierarchy
2992 Dump a representation of each class's hierarchy and virtual function
2993 table layout to a file. The file name is made by appending @file{.class}
2994 to the source file name. If the @samp{-@var{options}} form is used,
2995 @var{options} controls the details of the dump as described for the
2996 @option{-fdump-tree} options.
2998 @item -fdump-tree-@var{switch} @r{(C++ only)}
2999 @itemx -fdump-tree-@var{switch}-@var{options} @r{(C++ only)}
3001 Control the dumping at various stages of processing the intermediate
3002 language tree to a file. The file name is generated by appending a switch
3003 specific suffix to the source file name. If the @samp{-@var{options}}
3004 form is used, @var{options} is a list of @samp{-} separated options that
3005 control the details of the dump. Not all options are applicable to all
3006 dumps, those which are not meaningful will be ignored. The following
3007 options are available
3011 Print the address of each node. Usually this is not meaningful as it
3012 changes according to the environment and source file. Its primary use
3013 is for tying up a dump file with a debug environment.
3015 Inhibit dumping of members of a scope or body of a function merely
3016 because that scope has been reached. Only dump such items when they
3017 are directly reachable by some other path.
3019 Turn on all options.
3022 The following tree dumps are possible:
3025 Dump before any tree based optimization, to @file{@var{file}.original}.
3027 Dump after all tree based optimization, to @file{@var{file}.optimized}.
3029 Dump after function inlining, to @file{@var{file}.inlined}.
3032 @item -fpretend-float
3033 @opindex fpretend-float
3034 When running a cross-compiler, pretend that the target machine uses the
3035 same floating point format as the host machine. This causes incorrect
3036 output of the actual floating constants, but the actual instruction
3037 sequence will probably be the same as GCC would make when running on
3042 Store the usual ``temporary'' intermediate files permanently; place them
3043 in the current directory and name them based on the source file. Thus,
3044 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
3045 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
3046 preprocessed @file{foo.i} output file even though the compiler now
3047 normally uses an integrated preprocessor.
3051 Report the CPU time taken by each subprocess in the compilation
3052 sequence. For C source files, this is the compiler proper and assembler
3053 (plus the linker if linking is done). The output looks like this:
3060 The first number on each line is the ``user time,'' that is time spent
3061 executing the program itself. The second number is ``system time,''
3062 time spent executing operating system routines on behalf of the program.
3063 Both numbers are in seconds.
3065 @item -print-file-name=@var{library}
3066 @opindex print-file-name
3067 Print the full absolute name of the library file @var{library} that
3068 would be used when linking---and don't do anything else. With this
3069 option, GCC does not compile or link anything; it just prints the
3072 @item -print-multi-directory
3073 @opindex print-multi-directory
3074 Print the directory name corresponding to the multilib selected by any
3075 other switches present in the command line. This directory is supposed
3076 to exist in @env{GCC_EXEC_PREFIX}.
3078 @item -print-multi-lib
3079 @opindex print-multi-lib
3080 Print the mapping from multilib directory names to compiler switches
3081 that enable them. The directory name is separated from the switches by
3082 @samp{;}, and each switch starts with an @samp{@@} instead of the
3083 @samp{-}, without spaces between multiple switches. This is supposed to
3084 ease shell-processing.
3086 @item -print-prog-name=@var{program}
3087 @opindex print-prog-name
3088 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
3090 @item -print-libgcc-file-name
3091 @opindex print-libgcc-file-name
3092 Same as @option{-print-file-name=libgcc.a}.
3094 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
3095 but you do want to link with @file{libgcc.a}. You can do
3098 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
3101 @item -print-search-dirs
3102 @opindex print-search-dirs
3103 Print the name of the configured installation directory and a list of
3104 program and library directories gcc will search---and don't do anything else.
3106 This is useful when gcc prints the error message
3107 @samp{installation problem, cannot exec cpp0: No such file or directory}.
3108 To resolve this you either need to put @file{cpp0} and the other compiler
3109 components where gcc expects to find them, or you can set the environment
3110 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
3111 Don't forget the trailing '/'.
3112 @xref{Environment Variables}.
3115 @opindex dumpmachine
3116 Print the compiler's target machine (for example,
3117 @samp{i686-pc-linux-gnu})---and don't do anything else.
3120 @opindex dumpversion
3121 Print the compiler version (for example, @samp{3.0})---and don't do
3126 Print the compiler's built-in specs---and don't do anything else. (This
3127 is used when GCC itself is being built.) @xref{Spec Files}.
3130 @node Optimize Options
3131 @section Options That Control Optimization
3132 @cindex optimize options
3133 @cindex options, optimization
3135 These options control various sorts of optimizations:
3142 Optimize. Optimizing compilation takes somewhat more time, and a lot
3143 more memory for a large function.
3145 Without @option{-O}, the compiler's goal is to reduce the cost of
3146 compilation and to make debugging produce the expected results.
3147 Statements are independent: if you stop the program with a breakpoint
3148 between statements, you can then assign a new value to any variable or
3149 change the program counter to any other statement in the function and
3150 get exactly the results you would expect from the source code.
3152 With @option{-O}, the compiler tries to reduce code size and execution
3153 time, without performing any optimizations that take a great deal of
3158 Optimize even more. GCC performs nearly all supported optimizations
3159 that do not involve a space-speed tradeoff. The compiler does not
3160 perform loop unrolling or function inlining when you specify @option{-O2}.
3161 As compared to @option{-O}, this option increases both compilation time
3162 and the performance of the generated code.
3164 @option{-O2} turns on all optional optimizations except for loop unrolling,
3165 function inlining, and register renaming. It also turns on the
3166 @option{-fforce-mem} option on all machines and frame pointer elimination
3167 on machines where doing so does not interfere with debugging.
3169 Please note the warning under @option{-fgcse} about
3170 invoking @option{-O2} on programs that use computed gotos.
3174 Optimize yet more. @option{-O3} turns on all optimizations specified by
3175 @option{-O2} and also turns on the @option{-finline-functions} and
3176 @option{-frename-registers} options.
3184 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
3185 do not typically increase code size. It also performs further
3186 optimizations designed to reduce code size.
3188 If you use multiple @option{-O} options, with or without level numbers,
3189 the last such option is the one that is effective.
3192 Options of the form @option{-f@var{flag}} specify machine-independent
3193 flags. Most flags have both positive and negative forms; the negative
3194 form of @option{-ffoo} would be @option{-fno-foo}. In the table below,
3195 only one of the forms is listed---the one which is not the default.
3196 You can figure out the other form by either removing @samp{no-} or
3201 @opindex ffloat-store
3202 Do not store floating point variables in registers, and inhibit other
3203 options that might change whether a floating point value is taken from a
3206 @cindex floating point precision
3207 This option prevents undesirable excess precision on machines such as
3208 the 68000 where the floating registers (of the 68881) keep more
3209 precision than a @code{double} is supposed to have. Similarly for the
3210 x86 architecture. For most programs, the excess precision does only
3211 good, but a few programs rely on the precise definition of IEEE floating
3212 point. Use @option{-ffloat-store} for such programs, after modifying
3213 them to store all pertinent intermediate computations into variables.
3215 @item -fno-default-inline
3216 @opindex fno-default-inline
3217 Do not make member functions inline by default merely because they are
3218 defined inside the class scope (C++ only). Otherwise, when you specify
3219 @w{@option{-O}}, member functions defined inside class scope are compiled
3220 inline by default; i.e., you don't need to add @samp{inline} in front of
3221 the member function name.
3223 @item -fno-defer-pop
3224 @opindex fno-defer-pop
3225 Always pop the arguments to each function call as soon as that function
3226 returns. For machines which must pop arguments after a function call,
3227 the compiler normally lets arguments accumulate on the stack for several
3228 function calls and pops them all at once.
3232 Force memory operands to be copied into registers before doing
3233 arithmetic on them. This produces better code by making all memory
3234 references potential common subexpressions. When they are not common
3235 subexpressions, instruction combination should eliminate the separate
3236 register-load. The @option{-O2} option turns on this option.
3239 @opindex fforce-addr
3240 Force memory address constants to be copied into registers before
3241 doing arithmetic on them. This may produce better code just as
3242 @option{-fforce-mem} may.
3244 @item -fomit-frame-pointer
3245 @opindex fomit-frame-pointer
3246 Don't keep the frame pointer in a register for functions that
3247 don't need one. This avoids the instructions to save, set up and
3248 restore frame pointers; it also makes an extra register available
3249 in many functions. @strong{It also makes debugging impossible on
3252 On some machines, such as the VAX, this flag has no effect, because
3253 the standard calling sequence automatically handles the frame pointer
3254 and nothing is saved by pretending it doesn't exist. The
3255 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
3256 whether a target machine supports this flag. @xref{Registers,,Register
3257 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
3259 @item -foptimize-sibling-calls
3260 @opindex foptimize-sibling-calls
3261 Optimize sibling and tail recursive calls.
3265 This option generates traps for signed overflow on addition, subtraction,
3266 multiplication operations.
3270 Don't pay attention to the @code{inline} keyword. Normally this option
3271 is used to keep the compiler from expanding any functions inline.
3272 Note that if you are not optimizing, no functions can be expanded inline.
3274 @item -finline-functions
3275 @opindex finline-functions
3276 Integrate all simple functions into their callers. The compiler
3277 heuristically decides which functions are simple enough to be worth
3278 integrating in this way.
3280 If all calls to a given function are integrated, and the function is
3281 declared @code{static}, then the function is normally not output as
3282 assembler code in its own right.
3284 @item -finline-limit=@var{n}
3285 @opindex finline-limit
3286 By default, gcc limits the size of functions that can be inlined. This flag
3287 allows the control of this limit for functions that are explicitly marked as
3288 inline (ie marked with the inline keyword or defined within the class
3289 definition in c++). @var{n} is the size of functions that can be inlined in
3290 number of pseudo instructions (not counting parameter handling). The default
3291 value of @var{n} is 600.
3292 Increasing this value can result in more inlined code at
3293 the cost of compilation time and memory consumption. Decreasing usually makes
3294 the compilation faster and less code will be inlined (which presumably
3295 means slower programs). This option is particularly useful for programs that
3296 use inlining heavily such as those based on recursive templates with C++.
3298 @emph{Note:} pseudo instruction represents, in this particular context, an
3299 abstract measurement of function's size. In no way, it represents a count
3300 of assembly instructions and as such its exact meaning might change from one
3301 release to an another.
3303 @item -fkeep-inline-functions
3304 @opindex fkeep-inline-functions
3305 Even if all calls to a given function are integrated, and the function
3306 is declared @code{static}, nevertheless output a separate run-time
3307 callable version of the function. This switch does not affect
3308 @code{extern inline} functions.
3310 @item -fkeep-static-consts
3311 @opindex fkeep-static-consts
3312 Emit variables declared @code{static const} when optimization isn't turned
3313 on, even if the variables aren't referenced.
3315 GCC enables this option by default. If you want to force the compiler to
3316 check if the variable was referenced, regardless of whether or not
3317 optimization is turned on, use the @option{-fno-keep-static-consts} option.
3319 @item -fmerge-constants
3320 Attempt to merge identical constants (string constants and floating point
3321 constants) accross compilation units.
3323 This option is default for optimized compilation if assembler and linker
3324 support it. Use @option{-fno-merge-constants} to inhibit this behavior.
3326 @item -fmerge-all-constants
3327 Attempt to merge identical constants and identical variables.
3329 This option implies @option{-fmerge-constants}. In addition to
3330 @option{-fmerge-constants} this considers e.g. even constant initialized
3331 arrays or initialized constant variables with integral or floating point
3332 types. Languages like C or C++ require each non-automatic variable to
3333 have distinct location, so using this option will result in non-conforming
3336 @item -fno-function-cse
3337 @opindex fno-function-cse
3338 Do not put function addresses in registers; make each instruction that
3339 calls a constant function contain the function's address explicitly.
3341 This option results in less efficient code, but some strange hacks
3342 that alter the assembler output may be confused by the optimizations
3343 performed when this option is not used.
3347 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations}, and @*
3348 @option{-fno-trapping-math}.
3350 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
3352 This option should never be turned on by any @option{-O} option since
3353 it can result in incorrect output for programs which depend on
3354 an exact implementation of IEEE or ISO rules/specifications for
3357 @item -fno-math-errno
3358 @opindex fno-math-errno
3359 Do not set ERRNO after calling math functions that are executed
3360 with a single instruction, e.g., sqrt. A program that relies on
3361 IEEE exceptions for math error handling may want to use this flag
3362 for speed while maintaining IEEE arithmetic compatibility.
3364 This option should never be turned on by any @option{-O} option since
3365 it can result in incorrect output for programs which depend on
3366 an exact implementation of IEEE or ISO rules/specifications for
3369 The default is @option{-fmath-errno}.
3371 @item -funsafe-math-optimizations
3372 @opindex funsafe-math-optimizations
3373 Allow optimizations for floating-point arithmetic that (a) assume
3374 that arguments and results are valid and (b) may violate IEEE or
3375 ANSI standards. When used at link-time, it may include libraries
3376 or startup files that change the default FPU control word or other
3377 similar optimizations.
3379 This option should never be turned on by any @option{-O} option since
3380 it can result in incorrect output for programs which depend on
3381 an exact implementation of IEEE or ISO rules/specifications for
3384 The default is @option{-fno-unsafe-math-optimizations}.
3386 @item -fno-trapping-math
3387 @opindex fno-trapping-math
3388 Compile code assuming that floating-point operations cannot generate
3389 user-visible traps. Setting this option may allow faster code
3390 if one relies on ``non-stop'' IEEE arithmetic, for example.
3392 This option should never be turned on by any @option{-O} option since
3393 it can result in incorrect output for programs which depend on
3394 an exact implementation of IEEE or ISO rules/specifications for
3397 The default is @option{-ftrapping-math}.
3400 The following options control specific optimizations. The @option{-O2}
3401 option turns on all of these optimizations except @option{-funroll-loops}
3402 and @option{-funroll-all-loops}. On most machines, the @option{-O} option
3403 turns on the @option{-fthread-jumps} and @option{-fdelayed-branch} options,
3404 but specific machines may handle it differently.
3406 You can use the following flags in the rare cases when ``fine-tuning''
3407 of optimizations to be performed is desired.
3409 Not all of the optimizations performed by GCC have @option{-f} options
3413 @item -fstrength-reduce
3414 @opindex fstrength-reduce
3415 Perform the optimizations of loop strength reduction and
3416 elimination of iteration variables.
3418 @item -fthread-jumps
3419 @opindex fthread-jumps
3420 Perform optimizations where we check to see if a jump branches to a
3421 location where another comparison subsumed by the first is found. If
3422 so, the first branch is redirected to either the destination of the
3423 second branch or a point immediately following it, depending on whether
3424 the condition is known to be true or false.
3426 @item -fcse-follow-jumps
3427 @opindex fcse-follow-jumps
3428 In common subexpression elimination, scan through jump instructions
3429 when the target of the jump is not reached by any other path. For
3430 example, when CSE encounters an @code{if} statement with an
3431 @code{else} clause, CSE will follow the jump when the condition
3434 @item -fcse-skip-blocks
3435 @opindex fcse-skip-blocks
3436 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
3437 follow jumps which conditionally skip over blocks. When CSE
3438 encounters a simple @code{if} statement with no else clause,
3439 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
3440 body of the @code{if}.
3442 @item -frerun-cse-after-loop
3443 @opindex frerun-cse-after-loop
3444 Re-run common subexpression elimination after loop optimizations has been
3447 @item -frerun-loop-opt
3448 @opindex frerun-loop-opt
3449 Run the loop optimizer twice.
3453 Perform a global common subexpression elimination pass.
3454 This pass also performs global constant and copy propagation.
3456 @emph{Note:} When compiling a program using computed gotos, a GCC
3457 extension, you may get better runtime performance if you disable
3458 the global common subexpression elmination pass by adding
3459 @option{-fno-gcse} to the command line.
3463 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
3464 attempt to move loads which are only killed by stores into themselves. This
3465 allows a loop containing a load/store sequence to be changed to a load outside
3466 the loop, and a copy/store within the loop.
3470 When @option{-fgcse-sm} is enabled, A store motion pass is run after global common
3471 subexpression elimination. This pass will attempt to move stores out of loops.
3472 When used in conjunction with @option{-fgcse-lm}, loops containing a load/store sequence
3473 can be changed to a load before the loop and a store after the loop.
3475 @item -fdelete-null-pointer-checks
3476 @opindex fdelete-null-pointer-checks
3477 Use global dataflow analysis to identify and eliminate useless checks
3478 for null pointers. The compiler assumes that dereferencing a null
3479 pointer would have halted the program. If a pointer is checked after
3480 it has already been dereferenced, it cannot be null.
3482 In some environments, this assumption is not true, and programs can
3483 safely dereference null pointers. Use
3484 @option{-fno-delete-null-pointer-checks} to disable this optimization
3485 for programs which depend on that behavior.
3487 @item -fexpensive-optimizations
3488 @opindex fexpensive-optimizations
3489 Perform a number of minor optimizations that are relatively expensive.
3491 @item -foptimize-register-move
3493 @opindex foptimize-register-move
3495 Attempt to reassign register numbers in move instructions and as
3496 operands of other simple instructions in order to maximize the amount of
3497 register tying. This is especially helpful on machines with two-operand
3498 instructions. GCC enables this optimization by default with @option{-O2}
3501 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
3504 @item -fdelayed-branch
3505 @opindex fdelayed-branch
3506 If supported for the target machine, attempt to reorder instructions
3507 to exploit instruction slots available after delayed branch
3510 @item -fschedule-insns
3511 @opindex fschedule-insns
3512 If supported for the target machine, attempt to reorder instructions to
3513 eliminate execution stalls due to required data being unavailable. This
3514 helps machines that have slow floating point or memory load instructions
3515 by allowing other instructions to be issued until the result of the load
3516 or floating point instruction is required.
3518 @item -fschedule-insns2
3519 @opindex fschedule-insns2
3520 Similar to @option{-fschedule-insns}, but requests an additional pass of
3521 instruction scheduling after register allocation has been done. This is
3522 especially useful on machines with a relatively small number of
3523 registers and where memory load instructions take more than one cycle.
3525 @item -ffunction-sections
3526 @itemx -fdata-sections
3527 @opindex ffunction-sections
3528 @opindex fdata-sections
3529 Place each function or data item into its own section in the output
3530 file if the target supports arbitrary sections. The name of the
3531 function or the name of the data item determines the section's name
3534 Use these options on systems where the linker can perform optimizations
3535 to improve locality of reference in the instruction space. HPPA
3536 processors running HP-UX and Sparc processors running Solaris 2 have
3537 linkers with such optimizations. Other systems using the ELF object format
3538 as well as AIX may have these optimizations in the future.
3540 Only use these options when there are significant benefits from doing
3541 so. When you specify these options, the assembler and linker will
3542 create larger object and executable files and will also be slower.
3543 You will not be able to use @code{gprof} on all systems if you
3544 specify this option and you may have problems with debugging if
3545 you specify both this option and @option{-g}.
3547 @item -fcaller-saves
3548 @opindex fcaller-saves
3549 Enable values to be allocated in registers that will be clobbered by
3550 function calls, by emitting extra instructions to save and restore the
3551 registers around such calls. Such allocation is done only when it
3552 seems to result in better code than would otherwise be produced.
3554 This option is always enabled by default on certain machines, usually
3555 those which have no call-preserved registers to use instead.
3557 For all machines, optimization level 2 and higher enables this flag by
3560 @item -funroll-loops
3561 @opindex funroll-loops
3562 Unroll loops whose number of iterations can be determined at compile
3563 time or upon entry to the loop. @option{-funroll-loops} implies both
3564 @option{-fstrength-reduce} and @option{-frerun-cse-after-loop}. This
3565 option makes code larger, and may or may not make it run faster.
3567 @item -funroll-all-loops
3568 @opindex funroll-all-loops
3569 Unroll all loops, even if their number of iterations is uncertain when
3570 the loop is entered. This usually makes programs run more slowly.
3571 @option{-funroll-all-loops} implies the same options as
3572 @option{-funroll-loops},
3574 @item -fprefetch-loop-arrays
3575 @opindex fprefetch-loop-arrays
3576 If supported by the target machine, generate instructions to prefetch
3577 memory to improve the performance of loops that access large arrays.
3579 @item -fmove-all-movables
3580 @opindex fmove-all-movables
3581 Forces all invariant computations in loops to be moved
3584 @item -freduce-all-givs
3585 @opindex freduce-all-givs
3586 Forces all general-induction variables in loops to be
3589 @emph{Note:} When compiling programs written in Fortran,
3590 @option{-fmove-all-movables} and @option{-freduce-all-givs} are enabled
3591 by default when you use the optimizer.
3593 These options may generate better or worse code; results are highly
3594 dependent on the structure of loops within the source code.
3596 These two options are intended to be removed someday, once
3597 they have helped determine the efficacy of various
3598 approaches to improving loop optimizations.
3600 Please let us (@w{@email{gcc@@gcc.gnu.org}} and @w{@email{fortran@@gnu.org}})
3601 know how use of these options affects
3602 the performance of your production code.
3603 We're very interested in code that runs @emph{slower}
3604 when these options are @emph{enabled}.
3607 @itemx -fno-peephole2
3608 @opindex fno-peephole
3609 @opindex fno-peephole2
3610 Disable any machine-specific peephole optimizations. The difference
3611 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
3612 are implemented in the compiler; some targets use one, some use the
3613 other, a few use both.
3615 @item -fbranch-probabilities
3616 @opindex fbranch-probabilities
3617 After running a program compiled with @option{-fprofile-arcs}
3618 (@pxref{Debugging Options,, Options for Debugging Your Program or
3619 @command{gcc}}), you can compile it a second time using
3620 @option{-fbranch-probabilities}, to improve optimizations based on
3621 the number of times each branch was taken. When the program
3622 compiled with @option{-fprofile-arcs} exits it saves arc execution
3623 counts to a file called @file{@var{sourcename}.da} for each source
3624 file The information in this data file is very dependent on the
3625 structure of the generated code, so you must use the same source code
3626 and the same optimization options for both compilations.
3628 With @option{-fbranch-probabilities}, GCC puts a @samp{REG_EXEC_COUNT}
3629 note on the first instruction of each basic block, and a
3630 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
3631 These can be used to improve optimization. Currently, they are only
3632 used in one place: in @file{reorg.c}, instead of guessing which path a
3633 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
3634 exactly determine which path is taken more often.
3636 @item -fno-guess-branch-probability
3637 @opindex fno-guess-branch-probability
3638 Do not guess branch probabilities using a randomized model.
3640 Sometimes gcc will opt to use a randomized model to guess branch
3641 probabilities, when none are available from either profiling feedback
3642 (@option{-fprofile-arcs}) or @samp{__builtin_expect}. This means that
3643 different runs of the compiler on the same program may produce different
3646 In a hard real-time system, people don't want different runs of the
3647 compiler to produce code that has different behavior; minimizing
3648 non-determinism is of paramount import. This switch allows users to
3649 reduce non-determinism, possibly at the expense of inferior
3652 @item -fstrict-aliasing
3653 @opindex fstrict-aliasing
3654 Allows the compiler to assume the strictest aliasing rules applicable to
3655 the language being compiled. For C (and C++), this activates
3656 optimizations based on the type of expressions. In particular, an
3657 object of one type is assumed never to reside at the same address as an
3658 object of a different type, unless the types are almost the same. For
3659 example, an @code{unsigned int} can alias an @code{int}, but not a
3660 @code{void*} or a @code{double}. A character type may alias any other
3663 Pay special attention to code like this:
3676 The practice of reading from a different union member than the one most
3677 recently written to (called ``type-punning'') is common. Even with
3678 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
3679 is accessed through the union type. So, the code above will work as
3680 expected. However, this code might not:
3691 Every language that wishes to perform language-specific alias analysis
3692 should define a function that computes, given an @code{tree}
3693 node, an alias set for the node. Nodes in different alias sets are not
3694 allowed to alias. For an example, see the C front-end function
3695 @code{c_get_alias_set}.
3697 @item -falign-functions
3698 @itemx -falign-functions=@var{n}
3699 @opindex falign-functions
3700 Align the start of functions to the next power-of-two greater than
3701 @var{n}, skipping up to @var{n} bytes. For instance,
3702 @option{-falign-functions=32} aligns functions to the next 32-byte
3703 boundary, but @option{-falign-functions=24} would align to the next
3704 32-byte boundary only if this can be done by skipping 23 bytes or less.
3706 @option{-fno-align-functions} and @option{-falign-functions=1} are
3707 equivalent and mean that functions will not be aligned.
3709 Some assemblers only support this flag when @var{n} is a power of two;
3710 in that case, it is rounded up.
3712 If @var{n} is not specified, use a machine-dependent default.
3714 @item -falign-labels
3715 @itemx -falign-labels=@var{n}
3716 @opindex falign-labels
3717 Align all branch targets to a power-of-two boundary, skipping up to
3718 @var{n} bytes like @option{-falign-functions}. This option can easily
3719 make code slower, because it must insert dummy operations for when the
3720 branch target is reached in the usual flow of the code.
3722 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
3723 are greater than this value, then their values are used instead.
3725 If @var{n} is not specified, use a machine-dependent default which is
3726 very likely to be @samp{1}, meaning no alignment.
3729 @itemx -falign-loops=@var{n}
3730 @opindex falign-loops
3731 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
3732 like @option{-falign-functions}. The hope is that the loop will be
3733 executed many times, which will make up for any execution of the dummy
3736 If @var{n} is not specified, use a machine-dependent default.
3739 @itemx -falign-jumps=@var{n}
3740 @opindex falign-jumps
3741 Align branch targets to a power-of-two boundary, for branch targets
3742 where the targets can only be reached by jumping, skipping up to @var{n}
3743 bytes like @option{-falign-functions}. In this case, no dummy operations
3746 If @var{n} is not specified, use a machine-dependent default.
3750 Perform optimizations in static single assignment form. Each function's
3751 flow graph is translated into SSA form, optimizations are performed, and
3752 the flow graph is translated back from SSA form. Users should not
3753 specify this option, since it is not yet ready for production use.
3757 Perform Sparse Conditional Constant Propagation in SSA form. Requires
3758 @option{-fssa}. Like @option{-fssa}, this is an experimental feature.
3762 Perform aggressive dead-code elimination in SSA form. Requires @option{-fssa}.
3763 Like @option{-fssa}, this is an experimental feature.
3765 @item -fsingle-precision-constant
3766 @opindex fsingle-precision-constant
3767 Treat floating point constant as single precision constant instead of
3768 implicitly converting it to double precision constant.
3770 @item -frename-registers
3771 @opindex frename-registers
3772 Attempt to avoid false dependencies in scheduled code by making use
3773 of registers left over after register allocation. This optimization
3774 will most benefit processors with lots of registers. It can, however,
3775 make debugging impossible, since variables will no longer stay in
3776 a ``home register''.
3778 @item -fno-cprop-registers
3779 @opindex fno-cprop-registers
3780 After register allocation and post-register allocation instruction splitting,
3781 we perform a copy-propagation pass to try to reduce scheduling dependencies
3782 and occasionally eliminate the copy.
3784 @item --param @var{name}=@var{value}
3786 In some places, GCC uses various constants to control the amount of
3787 optimization that is done. For example, GCC will not inline functions
3788 that contain more that a certain number of instructions. You can
3789 control some of these constants on the command-line using the
3790 @option{--param} option.
3792 In each case, the @var{value} is an integer. The allowable choices for
3793 @var{name} are given in the following table:
3796 @item max-delay-slot-insn-search
3797 The maximum number of instructions to consider when looking for an
3798 instruction to fill a delay slot. If more than this arbitrary number of
3799 instructions is searched, the time savings from filling the delay slot
3800 will be minimal so stop searching. Increasing values mean more
3801 aggressive optimization, making the compile time increase with probably
3802 small improvement in executable run time.
3804 @item max-delay-slot-live-search
3805 When trying to fill delay slots, the maximum number of instructions to
3806 consider when searching for a block with valid live register
3807 information. Increasing this arbitrarily chosen value means more
3808 aggressive optimization, increasing the compile time. This parameter
3809 should be removed when the delay slot code is rewritten to maintain the
3812 @item max-gcse-memory
3813 The approximate maximum amount of memory that will be allocated in
3814 order to perform the global common subexpression elimination
3815 optimization. If more memory than specified is required, the
3816 optimization will not be done.
3818 @item max-gcse-passes
3819 The maximum number of passes of GCSE to run.
3821 @item max-pending-list-length
3822 The maximum number of pending dependencies scheduling will allow
3823 before flushing the current state and starting over. Large functions
3824 with few branches or calls can create excessively large lists which
3825 needlessly consume memory and resources.
3827 @item max-inline-insns
3828 If an function contains more than this many instructions, it
3829 will not be inlined. This option is precisely equivalent to
3830 @option{-finline-limit}.
3835 @node Preprocessor Options
3836 @section Options Controlling the Preprocessor
3837 @cindex preprocessor options
3838 @cindex options, preprocessor
3840 These options control the C preprocessor, which is run on each C source
3841 file before actual compilation.
3843 If you use the @option{-E} option, nothing is done except preprocessing.
3844 Some of these options make sense only together with @option{-E} because
3845 they cause the preprocessor output to be unsuitable for actual
3849 @item -include @var{file}
3851 Process @var{file} as input before processing the regular input file.
3852 In effect, the contents of @var{file} are compiled first. Any @option{-D}
3853 and @option{-U} options on the command line are always processed before
3854 @option{-include @var{file}}, regardless of the order in which they are
3855 written. All the @option{-include} and @option{-imacros} options are
3856 processed in the order in which they are written.
3858 @item -imacros @var{file}
3860 Process @var{file} as input, discarding the resulting output, before
3861 processing the regular input file. Because the output generated from
3862 @var{file} is discarded, the only effect of @option{-imacros @var{file}}
3863 is to make the macros defined in @var{file} available for use in the
3864 main input. All the @option{-include} and @option{-imacros} options are
3865 processed in the order in which they are written.
3867 @item -idirafter @var{dir}
3869 @cindex second include path
3870 Add the directory @var{dir} to the second include path. The directories
3871 on the second include path are searched when a header file is not found
3872 in any of the directories in the main include path (the one that
3873 @option{-I} adds to).
3875 @item -iprefix @var{prefix}
3877 Specify @var{prefix} as the prefix for subsequent @option{-iwithprefix}
3880 @item -iwithprefix @var{dir}
3881 @opindex iwithprefix
3882 Add a directory to the second include path. The directory's name is
3883 made by concatenating @var{prefix} and @var{dir}, where @var{prefix} was
3884 specified previously with @option{-iprefix}. If you have not specified a
3885 prefix yet, the directory containing the installed passes of the
3886 compiler is used as the default.
3888 @item -iwithprefixbefore @var{dir}
3889 @opindex iwithprefixbefore
3890 Add a directory to the main include path. The directory's name is made
3891 by concatenating @var{prefix} and @var{dir}, as in the case of
3892 @option{-iwithprefix}.
3894 @item -isystem @var{dir}
3896 Add a directory to the beginning of the second include path, marking it
3897 as a system directory, so that it gets the same special treatment as
3898 is applied to the standard system directories.
3902 Do not search the standard system directories for header files. Only
3903 the directories you have specified with @option{-I} options (and the
3904 current directory, if appropriate) are searched. @xref{Directory
3905 Options}, for information on @option{-I}.
3907 By using both @option{-nostdinc} and @option{-I-}, you can limit the include-file
3908 search path to only those directories you specify explicitly.
3912 When searching for a header file in a directory, remap file names if a
3913 file named @file{header.gcc} exists in that directory. This can be used
3914 to work around limitations of file systems with file name restrictions.
3915 The @file{header.gcc} file should contain a series of lines with two
3916 tokens on each line: the first token is the name to map, and the second
3917 token is the actual name to use.
3921 Do not predefine any nonstandard macros. (Including architecture flags).
3925 Run only the C preprocessor. Preprocess all the C source files
3926 specified and output the results to standard output or to the
3927 specified output file.
3931 Tell the preprocessor not to discard comments. Used with the
3936 Tell the preprocessor not to generate @samp{#line} directives.
3937 Used with the @option{-E} option.
3940 @cindex dependencies, make
3943 Instead of outputting the result of preprocessing, output a rule
3944 suitable for @code{make} describing the dependencies of the main source
3945 file. The preprocessor outputs one @code{make} rule containing the
3946 object file name for that source file, a colon, and the names of all the
3947 included files. Unless overridden explicitly, the object file name
3948 consists of the basename of the source file with any suffix replaced with
3949 object file suffix. If there are many included files then the
3950 rule is split into several lines using @samp{\}-newline.
3952 @option{-M} implies @option{-E}.
3956 Like @option{-M}, but mention only the files included with @samp{#include
3957 "@var{file}"}. System header files included with @samp{#include
3958 <@var{file}>} are omitted.
3962 Like @option{-M} but the dependency information is written to a file
3963 rather than stdout. @code{gcc} will use the same file name and
3964 directory as the object file, but with the suffix @file{.d} instead.
3966 This is in addition to compiling the main file as specified---@option{-MD}
3967 does not inhibit ordinary compilation the way @option{-M} does,
3968 unless you also specify @option{-MG}.
3970 With Mach, you can use the utility @code{md} to merge multiple
3971 dependency files into a single dependency file suitable for using with
3972 the @samp{make} command.
3976 Like @option{-MD} except mention only user header files, not system
3979 @item -MF @var{file}
3981 When used with @option{-M} or @option{-MM}, specifies a file to write the
3982 dependencies to. This allows the preprocessor to write the preprocessed
3983 file to stdout normally. If no @option{-MF} switch is given, CPP sends
3984 the rules to stdout and suppresses normal preprocessed output.
3986 Another way to specify output of a @code{make} rule is by setting
3987 the environment variable @env{DEPENDENCIES_OUTPUT} (@pxref{Environment
3992 When used with @option{-M} or @option{-MM}, @option{-MG} says to treat missing
3993 header files as generated files and assume they live in the same
3994 directory as the source file. It suppresses preprocessed output, as a
3995 missing header file is ordinarily an error.
3997 This feature is used in automatic updating of makefiles.
4001 This option instructs CPP to add a phony target for each dependency
4002 other than the main file, causing each to depend on nothing. These
4003 dummy rules work around errors @code{make} gives if you remove header
4004 files without updating the @code{Makefile} to match.
4006 This is typical output:-
4009 /tmp/test.o: /tmp/test.c /tmp/test.h
4014 @item -MQ @var{target}
4015 @item -MT @var{target}
4018 By default CPP uses the main file name, including any path, and appends
4019 the object suffix, normally ``.o'', to it to obtain the name of the
4020 target for dependency generation. With @option{-MT} you can specify a
4021 target yourself, overriding the default one.
4023 If you want multiple targets, you can specify them as a single argument
4024 to @option{-MT}, or use multiple @option{-MT} options.
4026 The targets you specify are output in the order they appear on the
4027 command line. @option{-MQ} is identical to @option{-MT}, except that the
4028 target name is quoted for Make, but with @option{-MT} it isn't. For
4029 example, @option{-MT '$(objpfx)foo.o'} gives
4032 $(objpfx)foo.o: /tmp/foo.c
4035 but @option{-MQ '$(objpfx)foo.o'} gives
4038 $$(objpfx)foo.o: /tmp/foo.c
4041 The default target is automatically quoted, as if it were given with
4046 Print the name of each header file used, in addition to other normal
4049 @item -A@var{question}(@var{answer})
4051 Assert the answer @var{answer} for @var{question}, in case it is tested
4052 with a preprocessing conditional such as @samp{#if
4053 #@var{question}(@var{answer})}. @option{-A-} disables the standard
4054 assertions that normally describe the target machine.
4058 Define macro @var{macro} with the string @samp{1} as its definition.
4060 @item -D@var{macro}=@var{defn}
4061 Define macro @var{macro} as @var{defn}. All instances of @option{-D} on
4062 the command line are processed before any @option{-U} options.
4064 Any @option{-D} and @option{-U} options on the command line are processed in
4065 order, and always before @option{-imacros @var{file}}, regardless of the
4066 order in which they are written.
4070 Undefine macro @var{macro}. @option{-U} options are evaluated after all
4071 @option{-D} options, but before any @option{-include} and @option{-imacros}
4074 Any @option{-D} and @option{-U} options on the command line are processed in
4075 order, and always before @option{-imacros @var{file}}, regardless of the
4076 order in which they are written.
4080 Tell the preprocessor to output only a list of the macro definitions
4081 that are in effect at the end of preprocessing. Used with the @option{-E}
4086 Tell the preprocessing to pass all macro definitions into the output, in
4087 their proper sequence in the rest of the output.
4091 Like @option{-dD} except that the macro arguments and contents are omitted.
4092 Only @samp{#define @var{name}} is included in the output.
4096 Output @samp{#include} directives in addition to the result of
4099 @item -fpreprocessed
4100 @opindex fpreprocessed
4101 Indicate to the preprocessor that the input file has already been
4102 preprocessed. This suppresses things like macro expansion, trigraph
4103 conversion, escaped newline splicing, and processing of most directives.
4104 The preprocessor still recognizes and removes comments, so that you can
4105 pass a file preprocessed with @option{-C} to the compiler without
4106 problems. In this mode the integrated preprocessor is little more than
4107 a tokenizer for the front ends.
4109 @option{-fpreprocessed} is implicit if the input file has one of the
4110 extensions @samp{i}, @samp{ii} or @samp{mi}. These are the extensions
4111 that GCC uses for preprocessed files created by @option{-save-temps}.
4115 Process ISO standard trigraph sequences. These are three-character
4116 sequences, all starting with @samp{??}, that are defined by ISO C to
4117 stand for single characters. For example, @samp{??/} stands for
4118 @samp{\}, so @samp{'??/n'} is a character constant for a newline. By
4119 default, GCC ignores trigraphs, but in standard-conforming modes it
4120 converts them. See the @option{-std} and @option{-ansi} options.
4122 The nine trigraph sequences are
4125 @expansion{} @samp{[}
4128 @expansion{} @samp{]}
4131 @expansion{} @samp{@{}
4134 @expansion{} @samp{@}}
4137 @expansion{} @samp{#}
4140 @expansion{} @samp{\}
4143 @expansion{} @samp{^}
4146 @expansion{} @samp{|}
4149 @expansion{} @samp{~}
4153 Trigraph support is not popular, so many compilers do not implement it
4154 properly. Portable code should not rely on trigraphs being either
4155 converted or ignored.
4157 @item -Wp,@var{option}
4159 Pass @var{option} as an option to the preprocessor. If @var{option}
4160 contains commas, it is split into multiple options at the commas.
4163 @node Assembler Options
4164 @section Passing Options to the Assembler
4166 @c prevent bad page break with this line
4167 You can pass options to the assembler.
4170 @item -Wa,@var{option}
4172 Pass @var{option} as an option to the assembler. If @var{option}
4173 contains commas, it is split into multiple options at the commas.
4177 @section Options for Linking
4178 @cindex link options
4179 @cindex options, linking
4181 These options come into play when the compiler links object files into
4182 an executable output file. They are meaningless if the compiler is
4183 not doing a link step.
4187 @item @var{object-file-name}
4188 A file name that does not end in a special recognized suffix is
4189 considered to name an object file or library. (Object files are
4190 distinguished from libraries by the linker according to the file
4191 contents.) If linking is done, these object files are used as input
4200 If any of these options is used, then the linker is not run, and
4201 object file names should not be used as arguments. @xref{Overall
4205 @item -l@var{library}
4206 @itemx -l @var{library}
4208 Search the library named @var{library} when linking. (The second
4209 alternative with the library as a separate argument is only for
4210 POSIX compliance and is not recommended.)
4212 It makes a difference where in the command you write this option; the
4213 linker searches and processes libraries and object files in the order they
4214 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
4215 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
4216 to functions in @samp{z}, those functions may not be loaded.
4218 The linker searches a standard list of directories for the library,
4219 which is actually a file named @file{lib@var{library}.a}. The linker
4220 then uses this file as if it had been specified precisely by name.
4222 The directories searched include several standard system directories
4223 plus any that you specify with @option{-L}.
4225 Normally the files found this way are library files---archive files
4226 whose members are object files. The linker handles an archive file by
4227 scanning through it for members which define symbols that have so far
4228 been referenced but not defined. But if the file that is found is an
4229 ordinary object file, it is linked in the usual fashion. The only
4230 difference between using an @option{-l} option and specifying a file name
4231 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
4232 and searches several directories.
4236 You need this special case of the @option{-l} option in order to
4237 link an Objective-C program.
4240 @opindex nostartfiles
4241 Do not use the standard system startup files when linking.
4242 The standard system libraries are used normally, unless @option{-nostdlib}
4243 or @option{-nodefaultlibs} is used.
4245 @item -nodefaultlibs
4246 @opindex nodefaultlibs
4247 Do not use the standard system libraries when linking.
4248 Only the libraries you specify will be passed to the linker.
4249 The standard startup files are used normally, unless @option{-nostartfiles}
4250 is used. The compiler may generate calls to memcmp, memset, and memcpy
4251 for System V (and ISO C) environments or to bcopy and bzero for
4252 BSD environments. These entries are usually resolved by entries in
4253 libc. These entry points should be supplied through some other
4254 mechanism when this option is specified.
4258 Do not use the standard system startup files or libraries when linking.
4259 No startup files and only the libraries you specify will be passed to
4260 the linker. The compiler may generate calls to memcmp, memset, and memcpy
4261 for System V (and ISO C) environments or to bcopy and bzero for
4262 BSD environments. These entries are usually resolved by entries in
4263 libc. These entry points should be supplied through some other
4264 mechanism when this option is specified.
4266 @cindex @option{-lgcc}, use with @option{-nostdlib}
4267 @cindex @option{-nostdlib} and unresolved references
4268 @cindex unresolved references and @option{-nostdlib}
4269 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
4270 @cindex @option{-nodefaultlibs} and unresolved references
4271 @cindex unresolved references and @option{-nodefaultlibs}
4272 One of the standard libraries bypassed by @option{-nostdlib} and
4273 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
4274 that GCC uses to overcome shortcomings of particular machines, or special
4275 needs for some languages.
4276 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
4277 Collection (GCC) Internals},
4278 for more discussion of @file{libgcc.a}.)
4279 In most cases, you need @file{libgcc.a} even when you want to avoid
4280 other standard libraries. In other words, when you specify @option{-nostdlib}
4281 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
4282 This ensures that you have no unresolved references to internal GCC
4283 library subroutines. (For example, @samp{__main}, used to ensure C++
4284 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
4285 GNU Compiler Collection (GCC) Internals}.)
4289 Remove all symbol table and relocation information from the executable.
4293 On systems that support dynamic linking, this prevents linking with the shared
4294 libraries. On other systems, this option has no effect.
4298 Produce a shared object which can then be linked with other objects to
4299 form an executable. Not all systems support this option. For predictable
4300 results, you must also specify the same set of options that were used to
4301 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
4302 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
4303 needs to build supplementary stub code for constructors to work. On
4304 multi-libbed systems, @samp{gcc -shared} must select the correct support
4305 libraries to link against. Failing to supply the correct flags may lead
4306 to subtle defects. Supplying them in cases where they are not necessary
4309 @item -shared-libgcc
4310 @itemx -static-libgcc
4311 @opindex shared-libgcc
4312 @opindex static-libgcc
4313 On systems that provide @file{libgcc} as a shared library, these options
4314 force the use of either the shared or static version respectively.
4315 If no shared version of @file{libgcc} was built when the compiler was
4316 configured, these options have no effect.
4318 There are several situations in which an application should use the
4319 shared @file{libgcc} instead of the static version. The most common
4320 of these is when the application wishes to throw and catch exceptions
4321 across different shared libraries. In that case, each of the libraries
4322 as well as the application itself should use the shared @file{libgcc}.
4324 Therefore, whenever you specify the @option{-shared} option, the GCC
4325 driver automatically adds @option{-shared-libgcc}, unless you explicitly
4326 specify @option{-static-libgcc}. The G++ driver automatically adds
4327 @option{-shared-libgcc} when you build a main executable as well because
4328 for C++ programs that is typically the right thing to do.
4329 (Exception-handling will not work reliably otherwise.)
4331 However, when linking a main executable written in C, you must
4332 explicitly say @option{-shared-libgcc} if you want to use the shared
4337 Bind references to global symbols when building a shared object. Warn
4338 about any unresolved references (unless overridden by the link editor
4339 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
4342 @item -Xlinker @var{option}
4344 Pass @var{option} as an option to the linker. You can use this to
4345 supply system-specific linker options which GCC does not know how to
4348 If you want to pass an option that takes an argument, you must use
4349 @option{-Xlinker} twice, once for the option and once for the argument.
4350 For example, to pass @option{-assert definitions}, you must write
4351 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
4352 @option{-Xlinker "-assert definitions"}, because this passes the entire
4353 string as a single argument, which is not what the linker expects.
4355 @item -Wl,@var{option}
4357 Pass @var{option} as an option to the linker. If @var{option} contains
4358 commas, it is split into multiple options at the commas.
4360 @item -u @var{symbol}
4362 Pretend the symbol @var{symbol} is undefined, to force linking of
4363 library modules to define it. You can use @option{-u} multiple times with
4364 different symbols to force loading of additional library modules.
4367 @node Directory Options
4368 @section Options for Directory Search
4369 @cindex directory options
4370 @cindex options, directory search
4373 These options specify directories to search for header files, for
4374 libraries and for parts of the compiler:
4379 Add the directory @var{dir} to the head of the list of directories to be
4380 searched for header files. This can be used to override a system header
4381 file, substituting your own version, since these directories are
4382 searched before the system header file directories. However, you should
4383 not use this option to add directories that contain vendor-supplied
4384 system header files (use @option{-isystem} for that). If you use more than
4385 one @option{-I} option, the directories are scanned in left-to-right
4386 order; the standard system directories come after.
4388 If a standard system include directory, or a directory specified with
4389 @option{-isystem}, is also specified with @option{-I}, it will be
4390 searched only in the position requested by @option{-I}. Also, it will
4391 not be considered a system include directory. If that directory really
4392 does contain system headers, there is a good chance that they will
4393 break. For instance, if GCC's installation procedure edited the headers
4394 in @file{/usr/include} to fix bugs, @samp{-I/usr/include} will cause the
4395 original, buggy headers to be found instead of the corrected ones. GCC
4396 will issue a warning when a system include directory is hidden in this
4401 Any directories you specify with @option{-I} options before the @option{-I-}
4402 option are searched only for the case of @samp{#include "@var{file}"};
4403 they are not searched for @samp{#include <@var{file}>}.
4405 If additional directories are specified with @option{-I} options after
4406 the @option{-I-}, these directories are searched for all @samp{#include}
4407 directives. (Ordinarily @emph{all} @option{-I} directories are used
4410 In addition, the @option{-I-} option inhibits the use of the current
4411 directory (where the current input file came from) as the first search
4412 directory for @samp{#include "@var{file}"}. There is no way to
4413 override this effect of @option{-I-}. With @option{-I.} you can specify
4414 searching the directory which was current when the compiler was
4415 invoked. That is not exactly the same as what the preprocessor does
4416 by default, but it is often satisfactory.
4418 @option{-I-} does not inhibit the use of the standard system directories
4419 for header files. Thus, @option{-I-} and @option{-nostdinc} are
4424 Add directory @var{dir} to the list of directories to be searched
4427 @item -B@var{prefix}
4429 This option specifies where to find the executables, libraries,
4430 include files, and data files of the compiler itself.
4432 The compiler driver program runs one or more of the subprograms
4433 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
4434 @var{prefix} as a prefix for each program it tries to run, both with and
4435 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
4437 For each subprogram to be run, the compiler driver first tries the
4438 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
4439 was not specified, the driver tries two standard prefixes, which are
4440 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc-lib/}. If neither of
4441 those results in a file name that is found, the unmodified program
4442 name is searched for using the directories specified in your
4443 @env{PATH} environment variable.
4445 The compiler will check to see if the path provided by the @option{-B}
4446 refers to a directory, and if necessary it will add a directory
4447 separator character at the end of the path.
4449 @option{-B} prefixes that effectively specify directory names also apply
4450 to libraries in the linker, because the compiler translates these
4451 options into @option{-L} options for the linker. They also apply to
4452 includes files in the preprocessor, because the compiler translates these
4453 options into @option{-isystem} options for the preprocessor. In this case,
4454 the compiler appends @samp{include} to the prefix.
4456 The run-time support file @file{libgcc.a} can also be searched for using
4457 the @option{-B} prefix, if needed. If it is not found there, the two
4458 standard prefixes above are tried, and that is all. The file is left
4459 out of the link if it is not found by those means.
4461 Another way to specify a prefix much like the @option{-B} prefix is to use
4462 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
4465 As a special kludge, if the path provided by @option{-B} is
4466 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
4467 9, then it will be replaced by @file{[dir/]include}. This is to help
4468 with boot-strapping the compiler.
4470 @item -specs=@var{file}
4472 Process @var{file} after the compiler reads in the standard @file{specs}
4473 file, in order to override the defaults that the @file{gcc} driver
4474 program uses when determining what switches to pass to @file{cc1},
4475 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
4476 @option{-specs=@var{file}} can be specified on the command line, and they
4477 are processed in order, from left to right.
4483 @section Specifying subprocesses and the switches to pass to them
4485 @command{gcc} is a driver program. It performs its job by invoking a
4486 sequence of other programs to do the work of compiling, assembling and
4487 linking. GCC interprets its command-line parameters and uses these to
4488 deduce which programs it should invoke, and which command-line options
4489 it ought to place on their command lines. This behavior is controlled
4490 by @dfn{spec strings}. In most cases there is one spec string for each
4491 program that GCC can invoke, but a few programs have multiple spec
4492 strings to control their behavior. The spec strings built into GCC can
4493 be overridden by using the @option{-specs=} command-line switch to specify
4496 @dfn{Spec files} are plaintext files that are used to construct spec
4497 strings. They consist of a sequence of directives separated by blank
4498 lines. The type of directive is determined by the first non-whitespace
4499 character on the line and it can be one of the following:
4502 @item %@var{command}
4503 Issues a @var{command} to the spec file processor. The commands that can
4507 @item %include <@var{file}>
4509 Search for @var{file} and insert its text at the current point in the
4512 @item %include_noerr <@var{file}>
4513 @cindex %include_noerr
4514 Just like @samp{%include}, but do not generate an error message if the include
4515 file cannot be found.
4517 @item %rename @var{old_name} @var{new_name}
4519 Rename the spec string @var{old_name} to @var{new_name}.
4523 @item *[@var{spec_name}]:
4524 This tells the compiler to create, override or delete the named spec
4525 string. All lines after this directive up to the next directive or
4526 blank line are considered to be the text for the spec string. If this
4527 results in an empty string then the spec will be deleted. (Or, if the
4528 spec did not exist, then nothing will happened.) Otherwise, if the spec
4529 does not currently exist a new spec will be created. If the spec does
4530 exist then its contents will be overridden by the text of this
4531 directive, unless the first character of that text is the @samp{+}
4532 character, in which case the text will be appended to the spec.
4534 @item [@var{suffix}]:
4535 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
4536 and up to the next directive or blank line are considered to make up the
4537 spec string for the indicated suffix. When the compiler encounters an
4538 input file with the named suffix, it will processes the spec string in
4539 order to work out how to compile that file. For example:
4546 This says that any input file whose name ends in @samp{.ZZ} should be
4547 passed to the program @samp{z-compile}, which should be invoked with the
4548 command-line switch @option{-input} and with the result of performing the
4549 @samp{%i} substitution. (See below.)
4551 As an alternative to providing a spec string, the text that follows a
4552 suffix directive can be one of the following:
4555 @item @@@var{language}
4556 This says that the suffix is an alias for a known @var{language}. This is
4557 similar to using the @option{-x} command-line switch to GCC to specify a
4558 language explicitly. For example:
4565 Says that .ZZ files are, in fact, C++ source files.
4568 This causes an error messages saying:
4571 @var{name} compiler not installed on this system.
4575 GCC already has an extensive list of suffixes built into it.
4576 This directive will add an entry to the end of the list of suffixes, but
4577 since the list is searched from the end backwards, it is effectively
4578 possible to override earlier entries using this technique.
4582 GCC has the following spec strings built into it. Spec files can
4583 override these strings or create their own. Note that individual
4584 targets can also add their own spec strings to this list.
4587 asm Options to pass to the assembler
4588 asm_final Options to pass to the assembler post-processor
4589 cpp Options to pass to the C preprocessor
4590 cc1 Options to pass to the C compiler
4591 cc1plus Options to pass to the C++ compiler
4592 endfile Object files to include at the end of the link
4593 link Options to pass to the linker
4594 lib Libraries to include on the command line to the linker
4595 libgcc Decides which GCC support library to pass to the linker
4596 linker Sets the name of the linker
4597 predefines Defines to be passed to the C preprocessor
4598 signed_char Defines to pass to CPP to say whether @code{char} is signed
4600 startfile Object files to include at the start of the link
4603 Here is a small example of a spec file:
4609 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
4612 This example renames the spec called @samp{lib} to @samp{old_lib} and
4613 then overrides the previous definition of @samp{lib} with a new one.
4614 The new definition adds in some extra command-line options before
4615 including the text of the old definition.
4617 @dfn{Spec strings} are a list of command-line options to be passed to their
4618 corresponding program. In addition, the spec strings can contain
4619 @samp{%}-prefixed sequences to substitute variable text or to
4620 conditionally insert text into the command line. Using these constructs
4621 it is possible to generate quite complex command lines.
4623 Here is a table of all defined @samp{%}-sequences for spec
4624 strings. Note that spaces are not generated automatically around the
4625 results of expanding these sequences. Therefore you can concatenate them
4626 together or combine them with constant text in a single argument.
4630 Substitute one @samp{%} into the program name or argument.
4633 Substitute the name of the input file being processed.
4636 Substitute the basename of the input file being processed.
4637 This is the substring up to (and not including) the last period
4638 and not including the directory.
4641 This is the same as @samp{%b}, but include the file suffix (text after
4645 Marks the argument containing or following the @samp{%d} as a
4646 temporary file name, so that that file will be deleted if GCC exits
4647 successfully. Unlike @samp{%g}, this contributes no text to the
4650 @item %g@var{suffix}
4651 Substitute a file name that has suffix @var{suffix} and is chosen
4652 once per compilation, and mark the argument in the same way as
4653 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
4654 name is now chosen in a way that is hard to predict even when previously
4655 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
4656 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
4657 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
4658 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
4659 was simply substituted with a file name chosen once per compilation,
4660 without regard to any appended suffix (which was therefore treated
4661 just like ordinary text), making such attacks more likely to succeed.
4663 @item %u@var{suffix}
4664 Like @samp{%g}, but generates a new temporary file name even if
4665 @samp{%u@var{suffix}} was already seen.
4667 @item %U@var{suffix}
4668 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
4669 new one if there is no such last file name. In the absence of any
4670 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
4671 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
4672 would involve the generation of two distinct file names, one
4673 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
4674 simply substituted with a file name chosen for the previous @samp{%u},
4675 without regard to any appended suffix.
4677 @item %j@var{SUFFIX}
4678 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
4679 writable, and if save-temps is off; otherwise, substitute the name
4680 of a temporary file, just like @samp{%u}. This temporary file is not
4681 meant for communication between processes, but rather as a junk
4684 @item %.@var{SUFFIX}
4685 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
4686 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
4687 terminated by the next space or %.
4690 Marks the argument containing or following the @samp{%w} as the
4691 designated output file of this compilation. This puts the argument
4692 into the sequence of arguments that @samp{%o} will substitute later.
4695 Substitutes the names of all the output files, with spaces
4696 automatically placed around them. You should write spaces
4697 around the @samp{%o} as well or the results are undefined.
4698 @samp{%o} is for use in the specs for running the linker.
4699 Input files whose names have no recognized suffix are not compiled
4700 at all, but they are included among the output files, so they will
4704 Substitutes the suffix for object files. Note that this is
4705 handled specially when it immediately follows @samp{%g, %u, or %U},
4706 because of the need for those to form complete file names. The
4707 handling is such that @samp{%O} is treated exactly as if it had already
4708 been substituted, except that @samp{%g, %u, and %U} do not currently
4709 support additional @var{suffix} characters following @samp{%O} as they would
4710 following, for example, @samp{.o}.
4713 Substitutes the standard macro predefinitions for the
4714 current target machine. Use this when running @code{cpp}.
4717 Like @samp{%p}, but puts @samp{__} before and after the name of each
4718 predefined macro, except for macros that start with @samp{__} or with
4719 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
4723 Substitute a @option{-iprefix} option made from @env{GCC_EXEC_PREFIX}.
4726 Current argument is the name of a library or startup file of some sort.
4727 Search for that file in a standard list of directories and substitute
4728 the full name found.
4731 Print @var{str} as an error message. @var{str} is terminated by a newline.
4732 Use this when inconsistent options are detected.
4735 Output @samp{-} if the input for the current command is coming from a pipe.
4738 Substitute the contents of spec string @var{name} at this point.
4741 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
4743 @item %x@{@var{option}@}
4744 Accumulate an option for @samp{%X}.
4747 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
4751 Output the accumulated assembler options specified by @option{-Wa}.
4754 Output the accumulated preprocessor options specified by @option{-Wp}.
4757 Substitute the major version number of GCC@.
4758 (For version 2.9.5, this is 2.)
4761 Substitute the minor version number of GCC@.
4762 (For version 2.9.5, this is 9.)
4765 Substitute the patch level number of GCC@.
4766 (For version 2.9.5, this is 5.)
4769 Process the @code{asm} spec. This is used to compute the
4770 switches to be passed to the assembler.
4773 Process the @code{asm_final} spec. This is a spec string for
4774 passing switches to an assembler post-processor, if such a program is
4778 Process the @code{link} spec. This is the spec for computing the
4779 command line passed to the linker. Typically it will make use of the
4780 @samp{%L %G %S %D and %E} sequences.
4783 Dump out a @option{-L} option for each directory that GCC believes might
4784 contain startup files. If the target supports multilibs then the
4785 current multilib directory will be prepended to each of these paths.
4788 Output the multilib directory with directory separators replaced with
4789 @samp{_}. If multilib directories are not set, or the multilib directory is
4790 @file{.} then this option emits nothing.
4793 Process the @code{lib} spec. This is a spec string for deciding which
4794 libraries should be included on the command line to the linker.
4797 Process the @code{libgcc} spec. This is a spec string for deciding
4798 which GCC support library should be included on the command line to the linker.
4801 Process the @code{startfile} spec. This is a spec for deciding which
4802 object files should be the first ones passed to the linker. Typically
4803 this might be a file named @file{crt0.o}.
4806 Process the @code{endfile} spec. This is a spec string that specifies
4807 the last object files that will be passed to the linker.
4810 Process the @code{cpp} spec. This is used to construct the arguments
4811 to be passed to the C preprocessor.
4814 Process the @code{signed_char} spec. This is intended to be used
4815 to tell cpp whether a char is signed. It typically has the definition:
4817 %@{funsigned-char:-D__CHAR_UNSIGNED__@}
4821 Process the @code{cc1} spec. This is used to construct the options to be
4822 passed to the actual C compiler (@samp{cc1}).
4825 Process the @code{cc1plus} spec. This is used to construct the options to be
4826 passed to the actual C++ compiler (@samp{cc1plus}).
4829 Substitute the variable part of a matched option. See below.
4830 Note that each comma in the substituted string is replaced by
4834 Substitutes the @code{-S} switch, if that switch was given to GCC@.
4835 If that switch was not specified, this substitutes nothing. Note that
4836 the leading dash is omitted when specifying this option, and it is
4837 automatically inserted if the substitution is performed. Thus the spec
4838 string @samp{%@{foo@}} would match the command-line option @option{-foo}
4839 and would output the command line option @option{-foo}.
4841 @item %W@{@code{S}@}
4842 Like %@{@code{S}@} but mark last argument supplied within as a file to be
4845 @item %@{@code{S}*@}
4846 Substitutes all the switches specified to GCC whose names start
4847 with @code{-S}, but which also take an argument. This is used for
4848 switches like @option{-o}, @option{-D}, @option{-I}, etc.
4849 GCC considers @option{-o foo} as being
4850 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
4851 text, including the space. Thus two arguments would be generated.
4853 @item %@{^@code{S}*@}
4854 Like %@{@code{S}*@}, but don't put a blank between a switch and its
4855 argument. Thus %@{^o*@} would only generate one argument, not two.
4857 @item %@{@code{S}*&@code{T}*@}
4858 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
4859 (the order of @code{S} and @code{T} in the spec is not significant).
4860 There can be any number of ampersand-separated variables; for each the
4861 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
4863 @item %@{<@code{S}@}
4864 Remove all occurrences of @code{-S} from the command line. Note---this
4865 command is position dependent. @samp{%} commands in the spec string
4866 before this option will see @code{-S}, @samp{%} commands in the spec
4867 string after this option will not.
4869 @item %@{@code{S}*:@code{X}@}
4870 Substitutes @code{X} if one or more switches whose names start with
4871 @code{-S} are specified to GCC@. Note that the tail part of the
4872 @code{-S} option (i.e.@: the part matched by the @samp{*}) will be substituted
4873 for each occurrence of @samp{%*} within @code{X}.
4875 @item %@{@code{S}:@code{X}@}
4876 Substitutes @code{X}, but only if the @samp{-S} switch was given to GCC@.
4878 @item %@{!@code{S}:@code{X}@}
4879 Substitutes @code{X}, but only if the @samp{-S} switch was @emph{not} given to GCC@.
4881 @item %@{|@code{S}:@code{X}@}
4882 Like %@{@code{S}:@code{X}@}, but if no @code{S} switch, substitute @samp{-}.
4884 @item %@{|!@code{S}:@code{X}@}
4885 Like %@{!@code{S}:@code{X}@}, but if there is an @code{S} switch, substitute @samp{-}.
4887 @item %@{.@code{S}:@code{X}@}
4888 Substitutes @code{X}, but only if processing a file with suffix @code{S}.
4890 @item %@{!.@code{S}:@code{X}@}
4891 Substitutes @code{X}, but only if @emph{not} processing a file with suffix @code{S}.
4893 @item %@{@code{S}|@code{P}:@code{X}@}
4894 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC@. This may be
4895 combined with @samp{!} and @samp{.} sequences as well, although they
4896 have a stronger binding than the @samp{|}. For example a spec string
4900 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
4903 will output the following command-line options from the following input
4904 command-line options:
4909 -d fred.c -foo -baz -boggle
4910 -d jim.d -bar -baz -boggle
4915 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or
4916 %@{!@code{S}:@code{X}@} construct may contain other nested @samp{%} constructs
4917 or spaces, or even newlines. They are processed as usual, as described
4920 The @option{-O}, @option{-f}, @option{-m}, and @option{-W}
4921 switches are handled specifically in these
4922 constructs. If another value of @option{-O} or the negated form of a @option{-f}, @option{-m}, or
4923 @option{-W} switch is found later in the command line, the earlier switch
4924 value is ignored, except with @{@code{S}*@} where @code{S} is just one
4925 letter, which passes all matching options.
4927 The character @samp{|} at the beginning of the predicate text is used to indicate
4928 that a command should be piped to the following command, but only if @option{-pipe}
4931 It is built into GCC which switches take arguments and which do not.
4932 (You might think it would be useful to generalize this to allow each
4933 compiler's spec to say which switches take arguments. But this cannot
4934 be done in a consistent fashion. GCC cannot even decide which input
4935 files have been specified without knowing which switches take arguments,
4936 and it must know which input files to compile in order to tell which
4939 GCC also knows implicitly that arguments starting in @option{-l} are to be
4940 treated as compiler output files, and passed to the linker in their
4941 proper position among the other output files.
4943 @c man begin OPTIONS
4945 @node Target Options
4946 @section Specifying Target Machine and Compiler Version
4947 @cindex target options
4948 @cindex cross compiling
4949 @cindex specifying machine version
4950 @cindex specifying compiler version and target machine
4951 @cindex compiler version, specifying
4952 @cindex target machine, specifying
4954 By default, GCC compiles code for the same type of machine that you
4955 are using. However, it can also be installed as a cross-compiler, to
4956 compile for some other type of machine. In fact, several different
4957 configurations of GCC, for different target machines, can be
4958 installed side by side. Then you specify which one to use with the
4961 In addition, older and newer versions of GCC can be installed side
4962 by side. One of them (probably the newest) will be the default, but
4963 you may sometimes wish to use another.
4966 @item -b @var{machine}
4968 The argument @var{machine} specifies the target machine for compilation.
4969 This is useful when you have installed GCC as a cross-compiler.
4971 The value to use for @var{machine} is the same as was specified as the
4972 machine type when configuring GCC as a cross-compiler. For
4973 example, if a cross-compiler was configured with @samp{configure
4974 i386v}, meaning to compile for an 80386 running System V, then you
4975 would specify @option{-b i386v} to run that cross compiler.
4977 When you do not specify @option{-b}, it normally means to compile for
4978 the same type of machine that you are using.
4980 @item -V @var{version}
4982 The argument @var{version} specifies which version of GCC to run.
4983 This is useful when multiple versions are installed. For example,
4984 @var{version} might be @samp{2.0}, meaning to run GCC version 2.0.
4986 The default version, when you do not specify @option{-V}, is the last
4987 version of GCC that you installed.
4990 The @option{-b} and @option{-V} options actually work by controlling part of
4991 the file name used for the executable files and libraries used for
4992 compilation. A given version of GCC, for a given target machine, is
4993 normally kept in the directory @file{/usr/local/lib/gcc-lib/@var{machine}/@var{version}}.
4995 Thus, sites can customize the effect of @option{-b} or @option{-V} either by
4996 changing the names of these directories or adding alternate names (or
4997 symbolic links). If in directory @file{/usr/local/lib/gcc-lib/} the
4998 file @file{80386} is a link to the file @file{i386v}, then @option{-b
4999 80386} becomes an alias for @option{-b i386v}.
5001 In one respect, the @option{-b} or @option{-V} do not completely change
5002 to a different compiler: the top-level driver program @command{gcc}
5003 that you originally invoked continues to run and invoke the other
5004 executables (preprocessor, compiler per se, assembler and linker)
5005 that do the real work. However, since no real work is done in the
5006 driver program, it usually does not matter that the driver program
5007 in use is not the one for the specified target. It is common for the
5008 interface to the other executables to change incompatibly between
5009 compiler versions, so unless the version specified is very close to that
5010 of the driver (for example, @option{-V 3.0} with a driver program from GCC
5011 version 3.0.1), use of @option{-V} may not work; for example, using
5012 @option{-V 2.95.2} will not work with a driver program from GCC 3.0.
5014 The only way that the driver program depends on the target machine is
5015 in the parsing and handling of special machine-specific options.
5016 However, this is controlled by a file which is found, along with the
5017 other executables, in the directory for the specified version and
5018 target machine. As a result, a single installed driver program adapts
5019 to any specified target machine, and sufficiently similar compiler
5022 The driver program executable does control one significant thing,
5023 however: the default version and target machine. Therefore, you can
5024 install different instances of the driver program, compiled for
5025 different targets or versions, under different names.
5027 For example, if the driver for version 2.0 is installed as @command{ogcc}
5028 and that for version 2.1 is installed as @command{gcc}, then the command
5029 @command{gcc} will use version 2.1 by default, while @command{ogcc} will use
5030 2.0 by default. However, you can choose either version with either
5031 command with the @option{-V} option.
5033 @node Submodel Options
5034 @section Hardware Models and Configurations
5035 @cindex submodel options
5036 @cindex specifying hardware config
5037 @cindex hardware models and configurations, specifying
5038 @cindex machine dependent options
5040 Earlier we discussed the standard option @option{-b} which chooses among
5041 different installed compilers for completely different target
5042 machines, such as VAX vs.@: 68000 vs.@: 80386.
5044 In addition, each of these target machine types can have its own
5045 special options, starting with @samp{-m}, to choose among various
5046 hardware models or configurations---for example, 68010 vs 68020,
5047 floating coprocessor or none. A single installed version of the
5048 compiler can compile for any model or configuration, according to the
5051 Some configurations of the compiler also support additional special
5052 options, usually for compatibility with other compilers on the same
5055 These options are defined by the macro @code{TARGET_SWITCHES} in the
5056 machine description. The default for the options is also defined by
5057 that macro, which enables you to change the defaults.
5071 * RS/6000 and PowerPC Options::
5074 * i386 and x86-64 Options::
5076 * Intel 960 Options::
5077 * DEC Alpha Options::
5078 * DEC Alpha/VMS Options::
5082 * System V Options::
5083 * TMS320C3x/C4x Options::
5091 * S/390 and zSeries Options::
5096 @node M680x0 Options
5097 @subsection M680x0 Options
5098 @cindex M680x0 options
5100 These are the @samp{-m} options defined for the 68000 series. The default
5101 values for these options depends on which style of 68000 was selected when
5102 the compiler was configured; the defaults for the most common choices are
5110 Generate output for a 68000. This is the default
5111 when the compiler is configured for 68000-based systems.
5113 Use this option for microcontrollers with a 68000 or EC000 core,
5114 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
5120 Generate output for a 68020. This is the default
5121 when the compiler is configured for 68020-based systems.
5125 Generate output containing 68881 instructions for floating point.
5126 This is the default for most 68020 systems unless @option{--nfp} was
5127 specified when the compiler was configured.
5131 Generate output for a 68030. This is the default when the compiler is
5132 configured for 68030-based systems.
5136 Generate output for a 68040. This is the default when the compiler is
5137 configured for 68040-based systems.
5139 This option inhibits the use of 68881/68882 instructions that have to be
5140 emulated by software on the 68040. Use this option if your 68040 does not
5141 have code to emulate those instructions.
5145 Generate output for a 68060. This is the default when the compiler is
5146 configured for 68060-based systems.
5148 This option inhibits the use of 68020 and 68881/68882 instructions that
5149 have to be emulated by software on the 68060. Use this option if your 68060
5150 does not have code to emulate those instructions.
5154 Generate output for a CPU32. This is the default
5155 when the compiler is configured for CPU32-based systems.
5157 Use this option for microcontrollers with a
5158 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
5159 68336, 68340, 68341, 68349 and 68360.
5163 Generate output for a 520X ``coldfire'' family cpu. This is the default
5164 when the compiler is configured for 520X-based systems.
5166 Use this option for microcontroller with a 5200 core, including
5167 the MCF5202, MCF5203, MCF5204 and MCF5202.
5172 Generate output for a 68040, without using any of the new instructions.
5173 This results in code which can run relatively efficiently on either a
5174 68020/68881 or a 68030 or a 68040. The generated code does use the
5175 68881 instructions that are emulated on the 68040.
5179 Generate output for a 68060, without using any of the new instructions.
5180 This results in code which can run relatively efficiently on either a
5181 68020/68881 or a 68030 or a 68040. The generated code does use the
5182 68881 instructions that are emulated on the 68060.
5186 Generate output containing Sun FPA instructions for floating point.
5189 @opindex msoft-float
5190 Generate output containing library calls for floating point.
5191 @strong{Warning:} the requisite libraries are not available for all m68k
5192 targets. Normally the facilities of the machine's usual C compiler are
5193 used, but this can't be done directly in cross-compilation. You must
5194 make your own arrangements to provide suitable library functions for
5195 cross-compilation. The embedded targets @samp{m68k-*-aout} and
5196 @samp{m68k-*-coff} do provide software floating point support.
5200 Consider type @code{int} to be 16 bits wide, like @code{short int}.
5203 @opindex mnobitfield
5204 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
5205 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
5209 Do use the bit-field instructions. The @option{-m68020} option implies
5210 @option{-mbitfield}. This is the default if you use a configuration
5211 designed for a 68020.
5215 Use a different function-calling convention, in which functions
5216 that take a fixed number of arguments return with the @code{rtd}
5217 instruction, which pops their arguments while returning. This
5218 saves one instruction in the caller since there is no need to pop
5219 the arguments there.
5221 This calling convention is incompatible with the one normally
5222 used on Unix, so you cannot use it if you need to call libraries
5223 compiled with the Unix compiler.
5225 Also, you must provide function prototypes for all functions that
5226 take variable numbers of arguments (including @code{printf});
5227 otherwise incorrect code will be generated for calls to those
5230 In addition, seriously incorrect code will result if you call a
5231 function with too many arguments. (Normally, extra arguments are
5232 harmlessly ignored.)
5234 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
5235 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
5238 @itemx -mno-align-int
5240 @opindex mno-align-int
5241 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
5242 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
5243 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
5244 Aligning variables on 32-bit boundaries produces code that runs somewhat
5245 faster on processors with 32-bit busses at the expense of more memory.
5247 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
5248 align structures containing the above types differently than
5249 most published application binary interface specifications for the m68k.
5253 Use the pc-relative addressing mode of the 68000 directly, instead of
5254 using a global offset table. At present, this option implies @option{-fpic},
5255 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
5256 not presently supported with @option{-mpcrel}, though this could be supported for
5257 68020 and higher processors.
5259 @item -mno-strict-align
5260 @itemx -mstrict-align
5261 @opindex mno-strict-align
5262 @opindex mstrict-align
5263 Do not (do) assume that unaligned memory references will be handled by
5268 @node M68hc1x Options
5269 @subsection M68hc1x Options
5270 @cindex M68hc1x options
5272 These are the @samp{-m} options defined for the 68hc11 and 68hc12
5273 microcontrollers. The default values for these options depends on
5274 which style of microcontroller was selected when the compiler was configured;
5275 the defaults for the most common choices are given below.
5282 Generate output for a 68HC11. This is the default
5283 when the compiler is configured for 68HC11-based systems.
5289 Generate output for a 68HC12. This is the default
5290 when the compiler is configured for 68HC12-based systems.
5293 @opindex mauto-incdec
5294 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
5299 Consider type @code{int} to be 16 bits wide, like @code{short int}.
5301 @item -msoft-reg-count=@var{count}
5302 @opindex msoft-reg-count
5303 Specify the number of pseudo-soft registers which are used for the
5304 code generation. The maximum number is 32. Using more pseudo-soft
5305 register may or may not result in better code depending on the program.
5306 The default is 4 for 68HC11 and 2 for 68HC12.
5311 @subsection VAX Options
5314 These @samp{-m} options are defined for the VAX:
5319 Do not output certain jump instructions (@code{aobleq} and so on)
5320 that the Unix assembler for the VAX cannot handle across long
5325 Do output those jump instructions, on the assumption that you
5326 will assemble with the GNU assembler.
5330 Output code for g-format floating point numbers instead of d-format.
5334 @subsection SPARC Options
5335 @cindex SPARC options
5337 These @samp{-m} switches are supported on the SPARC:
5342 @opindex mno-app-regs
5344 Specify @option{-mapp-regs} to generate output using the global registers
5345 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
5348 To be fully SVR4 ABI compliant at the cost of some performance loss,
5349 specify @option{-mno-app-regs}. You should compile libraries and system
5350 software with this option.
5355 @opindex mhard-float
5356 Generate output containing floating point instructions. This is the
5362 @opindex msoft-float
5363 Generate output containing library calls for floating point.
5364 @strong{Warning:} the requisite libraries are not available for all SPARC
5365 targets. Normally the facilities of the machine's usual C compiler are
5366 used, but this cannot be done directly in cross-compilation. You must make
5367 your own arrangements to provide suitable library functions for
5368 cross-compilation. The embedded targets @samp{sparc-*-aout} and
5369 @samp{sparclite-*-*} do provide software floating point support.
5371 @option{-msoft-float} changes the calling convention in the output file;
5372 therefore, it is only useful if you compile @emph{all} of a program with
5373 this option. In particular, you need to compile @file{libgcc.a}, the
5374 library that comes with GCC, with @option{-msoft-float} in order for
5377 @item -mhard-quad-float
5378 @opindex mhard-quad-float
5379 Generate output containing quad-word (long double) floating point
5382 @item -msoft-quad-float
5383 @opindex msoft-quad-float
5384 Generate output containing library calls for quad-word (long double)
5385 floating point instructions. The functions called are those specified
5386 in the SPARC ABI@. This is the default.
5388 As of this writing, there are no sparc implementations that have hardware
5389 support for the quad-word floating point instructions. They all invoke
5390 a trap handler for one of these instructions, and then the trap handler
5391 emulates the effect of the instruction. Because of the trap handler overhead,
5392 this is much slower than calling the ABI library routines. Thus the
5393 @option{-msoft-quad-float} option is the default.
5397 @opindex mno-epilogue
5399 With @option{-mepilogue} (the default), the compiler always emits code for
5400 function exit at the end of each function. Any function exit in
5401 the middle of the function (such as a return statement in C) will
5402 generate a jump to the exit code at the end of the function.
5404 With @option{-mno-epilogue}, the compiler tries to emit exit code inline
5405 at every function exit.
5411 With @option{-mflat}, the compiler does not generate save/restore instructions
5412 and will use a ``flat'' or single register window calling convention.
5413 This model uses %i7 as the frame pointer and is compatible with the normal
5414 register window model. Code from either may be intermixed.
5415 The local registers and the input registers (0--5) are still treated as
5416 ``call saved'' registers and will be saved on the stack as necessary.
5418 With @option{-mno-flat} (the default), the compiler emits save/restore
5419 instructions (except for leaf functions) and is the normal mode of operation.
5421 @item -mno-unaligned-doubles
5422 @itemx -munaligned-doubles
5423 @opindex mno-unaligned-doubles
5424 @opindex munaligned-doubles
5425 Assume that doubles have 8 byte alignment. This is the default.
5427 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
5428 alignment only if they are contained in another type, or if they have an
5429 absolute address. Otherwise, it assumes they have 4 byte alignment.
5430 Specifying this option avoids some rare compatibility problems with code
5431 generated by other compilers. It is not the default because it results
5432 in a performance loss, especially for floating point code.
5434 @item -mno-faster-structs
5435 @itemx -mfaster-structs
5436 @opindex mno-faster-structs
5437 @opindex mfaster-structs
5438 With @option{-mfaster-structs}, the compiler assumes that structures
5439 should have 8 byte alignment. This enables the use of pairs of
5440 @code{ldd} and @code{std} instructions for copies in structure
5441 assignment, in place of twice as many @code{ld} and @code{st} pairs.
5442 However, the use of this changed alignment directly violates the Sparc
5443 ABI@. Thus, it's intended only for use on targets where the developer
5444 acknowledges that their resulting code will not be directly in line with
5445 the rules of the ABI@.
5451 These two options select variations on the SPARC architecture.
5453 By default (unless specifically configured for the Fujitsu SPARClite),
5454 GCC generates code for the v7 variant of the SPARC architecture.
5456 @option{-mv8} will give you SPARC v8 code. The only difference from v7
5457 code is that the compiler emits the integer multiply and integer
5458 divide instructions which exist in SPARC v8 but not in SPARC v7.
5460 @option{-msparclite} will give you SPARClite code. This adds the integer
5461 multiply, integer divide step and scan (@code{ffs}) instructions which
5462 exist in SPARClite but not in SPARC v7.
5464 These options are deprecated and will be deleted in a future GCC release.
5465 They have been replaced with @option{-mcpu=xxx}.
5470 @opindex msupersparc
5471 These two options select the processor for which the code is optimized.
5473 With @option{-mcypress} (the default), the compiler optimizes code for the
5474 Cypress CY7C602 chip, as used in the SparcStation/SparcServer 3xx series.
5475 This is also appropriate for the older SparcStation 1, 2, IPX etc.
5477 With @option{-msupersparc} the compiler optimizes code for the SuperSparc cpu, as
5478 used in the SparcStation 10, 1000 and 2000 series. This flag also enables use
5479 of the full SPARC v8 instruction set.
5481 These options are deprecated and will be deleted in a future GCC release.
5482 They have been replaced with @option{-mcpu=xxx}.
5484 @item -mcpu=@var{cpu_type}
5486 Set the instruction set, register set, and instruction scheduling parameters
5487 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
5488 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
5489 @samp{hypersparc}, @samp{sparclite86x}, @samp{f930}, @samp{f934},
5490 @samp{sparclet}, @samp{tsc701}, @samp{v9}, and @samp{ultrasparc}.
5492 Default instruction scheduling parameters are used for values that select
5493 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
5494 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
5496 Here is a list of each supported architecture and their supported
5501 v8: supersparc, hypersparc
5502 sparclite: f930, f934, sparclite86x
5507 @item -mtune=@var{cpu_type}
5509 Set the instruction scheduling parameters for machine type
5510 @var{cpu_type}, but do not set the instruction set or register set that the
5511 option @option{-mcpu=@var{cpu_type}} would.
5513 The same values for @option{-mcpu=@var{cpu_type}} can be used for
5514 @option{-mtune=@var{cpu_type}}, but the only useful values are those
5515 that select a particular cpu implementation. Those are @samp{cypress},
5516 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
5517 @samp{sparclite86x}, @samp{tsc701}, and @samp{ultrasparc}.
5521 These @samp{-m} switches are supported in addition to the above
5522 on the SPARCLET processor.
5525 @item -mlittle-endian
5526 @opindex mlittle-endian
5527 Generate code for a processor running in little-endian mode.
5531 Treat register @code{%g0} as a normal register.
5532 GCC will continue to clobber it as necessary but will not assume
5533 it always reads as 0.
5535 @item -mbroken-saverestore
5536 @opindex mbroken-saverestore
5537 Generate code that does not use non-trivial forms of the @code{save} and
5538 @code{restore} instructions. Early versions of the SPARCLET processor do
5539 not correctly handle @code{save} and @code{restore} instructions used with
5540 arguments. They correctly handle them used without arguments. A @code{save}
5541 instruction used without arguments increments the current window pointer
5542 but does not allocate a new stack frame. It is assumed that the window
5543 overflow trap handler will properly handle this case as will interrupt
5547 These @samp{-m} switches are supported in addition to the above
5548 on SPARC V9 processors in 64-bit environments.
5551 @item -mlittle-endian
5552 @opindex mlittle-endian
5553 Generate code for a processor running in little-endian mode.
5559 Generate code for a 32-bit or 64-bit environment.
5560 The 32-bit environment sets int, long and pointer to 32 bits.
5561 The 64-bit environment sets int to 32 bits and long and pointer
5564 @item -mcmodel=medlow
5565 @opindex mcmodel=medlow
5566 Generate code for the Medium/Low code model: the program must be linked
5567 in the low 32 bits of the address space. Pointers are 64 bits.
5568 Programs can be statically or dynamically linked.
5570 @item -mcmodel=medmid
5571 @opindex mcmodel=medmid
5572 Generate code for the Medium/Middle code model: the program must be linked
5573 in the low 44 bits of the address space, the text segment must be less than
5574 2G bytes, and data segment must be within 2G of the text segment.
5575 Pointers are 64 bits.
5577 @item -mcmodel=medany
5578 @opindex mcmodel=medany
5579 Generate code for the Medium/Anywhere code model: the program may be linked
5580 anywhere in the address space, the text segment must be less than
5581 2G bytes, and data segment must be within 2G of the text segment.
5582 Pointers are 64 bits.
5584 @item -mcmodel=embmedany
5585 @opindex mcmodel=embmedany
5586 Generate code for the Medium/Anywhere code model for embedded systems:
5587 assume a 32-bit text and a 32-bit data segment, both starting anywhere
5588 (determined at link time). Register %g4 points to the base of the
5589 data segment. Pointers are still 64 bits.
5590 Programs are statically linked, PIC is not supported.
5593 @itemx -mno-stack-bias
5594 @opindex mstack-bias
5595 @opindex mno-stack-bias
5596 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
5597 frame pointer if present, are offset by @minus{}2047 which must be added back
5598 when making stack frame references.
5599 Otherwise, assume no such offset is present.
5602 @node Convex Options
5603 @subsection Convex Options
5604 @cindex Convex options
5606 These @samp{-m} options are defined for Convex:
5611 Generate output for C1. The code will run on any Convex machine.
5612 The preprocessor symbol @code{__convex__c1__} is defined.
5616 Generate output for C2. Uses instructions not available on C1.
5617 Scheduling and other optimizations are chosen for max performance on C2.
5618 The preprocessor symbol @code{__convex_c2__} is defined.
5622 Generate output for C32xx. Uses instructions not available on C1.
5623 Scheduling and other optimizations are chosen for max performance on C32.
5624 The preprocessor symbol @code{__convex_c32__} is defined.
5628 Generate output for C34xx. Uses instructions not available on C1.
5629 Scheduling and other optimizations are chosen for max performance on C34.
5630 The preprocessor symbol @code{__convex_c34__} is defined.
5634 Generate output for C38xx. Uses instructions not available on C1.
5635 Scheduling and other optimizations are chosen for max performance on C38.
5636 The preprocessor symbol @code{__convex_c38__} is defined.
5640 Generate code which puts an argument count in the word preceding each
5641 argument list. This is compatible with regular CC, and a few programs
5642 may need the argument count word. GDB and other source-level debuggers
5643 do not need it; this info is in the symbol table.
5646 @opindex mnoargcount
5647 Omit the argument count word. This is the default.
5649 @item -mvolatile-cache
5650 @opindex mvolatile-cache
5651 Allow volatile references to be cached. This is the default.
5653 @item -mvolatile-nocache
5654 @opindex mvolatile-nocache
5655 Volatile references bypass the data cache, going all the way to memory.
5656 This is only needed for multi-processor code that does not use standard
5657 synchronization instructions. Making non-volatile references to volatile
5658 locations will not necessarily work.
5662 Type long is 32 bits, the same as type int. This is the default.
5666 Type long is 64 bits, the same as type long long. This option is useless,
5667 because no library support exists for it.
5670 @node AMD29K Options
5671 @subsection AMD29K Options
5672 @cindex AMD29K options
5674 These @samp{-m} options are defined for the AMD Am29000:
5679 @cindex DW bit (29k)
5680 Generate code that assumes the @code{DW} bit is set, i.e., that byte and
5681 halfword operations are directly supported by the hardware. This is the
5686 Generate code that assumes the @code{DW} bit is not set.
5690 @cindex byte writes (29k)
5691 Generate code that assumes the system supports byte and halfword write
5692 operations. This is the default.
5696 Generate code that assumes the systems does not support byte and
5697 halfword write operations. @option{-mnbw} implies @option{-mndw}.
5701 @cindex memory model (29k)
5702 Use a small memory model that assumes that all function addresses are
5703 either within a single 256 KB segment or at an absolute address of less
5704 than 256k. This allows the @code{call} instruction to be used instead
5705 of a @code{const}, @code{consth}, @code{calli} sequence.
5709 Use the normal memory model: Generate @code{call} instructions only when
5710 calling functions in the same file and @code{calli} instructions
5711 otherwise. This works if each file occupies less than 256 KB but allows
5712 the entire executable to be larger than 256 KB@. This is the default.
5716 Always use @code{calli} instructions. Specify this option if you expect
5717 a single file to compile into more than 256 KB of code.
5721 @cindex processor selection (29k)
5722 Generate code for the Am29050.
5726 Generate code for the Am29000. This is the default.
5728 @item -mkernel-registers
5729 @opindex mkernel-registers
5730 @cindex kernel and user registers (29k)
5731 Generate references to registers @code{gr64-gr95} instead of to
5732 registers @code{gr96-gr127}. This option can be used when compiling
5733 kernel code that wants a set of global registers disjoint from that used
5736 Note that when this option is used, register names in @samp{-f} flags
5737 must use the normal, user-mode, names.
5739 @item -muser-registers
5740 @opindex muser-registers
5741 Use the normal set of global registers, @code{gr96-gr127}. This is the
5745 @itemx -mno-stack-check
5746 @opindex mstack-check
5747 @opindex mno-stack-check
5748 @cindex stack checks (29k)
5749 Insert (or do not insert) a call to @code{__msp_check} after each stack
5750 adjustment. This is often used for kernel code.
5753 @itemx -mno-storem-bug
5754 @opindex mstorem-bug
5755 @opindex mno-storem-bug
5756 @cindex storem bug (29k)
5757 @option{-mstorem-bug} handles 29k processors which cannot handle the
5758 separation of a mtsrim insn and a storem instruction (most 29000 chips
5759 to date, but not the 29050).
5761 @item -mno-reuse-arg-regs
5762 @itemx -mreuse-arg-regs
5763 @opindex mno-reuse-arg-regs
5764 @opindex mreuse-arg-regs
5765 @option{-mno-reuse-arg-regs} tells the compiler to only use incoming argument
5766 registers for copying out arguments. This helps detect calling a function
5767 with fewer arguments than it was declared with.
5769 @item -mno-impure-text
5770 @itemx -mimpure-text
5771 @opindex mno-impure-text
5772 @opindex mimpure-text
5773 @option{-mimpure-text}, used in addition to @option{-shared}, tells the compiler to
5774 not pass @option{-assert pure-text} to the linker when linking a shared object.
5777 @opindex msoft-float
5778 Generate output containing library calls for floating point.
5779 @strong{Warning:} the requisite libraries are not part of GCC@.
5780 Normally the facilities of the machine's usual C compiler are used, but
5781 this can't be done directly in cross-compilation. You must make your
5782 own arrangements to provide suitable library functions for
5787 Do not generate multm or multmu instructions. This is useful for some embedded
5788 systems which do not have trap handlers for these instructions.
5792 @subsection ARM Options
5795 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
5800 @opindex mapcs-frame
5801 Generate a stack frame that is compliant with the ARM Procedure Call
5802 Standard for all functions, even if this is not strictly necessary for
5803 correct execution of the code. Specifying @option{-fomit-frame-pointer}
5804 with this option will cause the stack frames not to be generated for
5805 leaf functions. The default is @option{-mno-apcs-frame}.
5809 This is a synonym for @option{-mapcs-frame}.
5813 Generate code for a processor running with a 26-bit program counter,
5814 and conforming to the function calling standards for the APCS 26-bit
5815 option. This option replaces the @option{-m2} and @option{-m3} options
5816 of previous releases of the compiler.
5820 Generate code for a processor running with a 32-bit program counter,
5821 and conforming to the function calling standards for the APCS 32-bit
5822 option. This option replaces the @option{-m6} option of previous releases
5826 @c not currently implemented
5827 @item -mapcs-stack-check
5828 @opindex mapcs-stack-check
5829 Generate code to check the amount of stack space available upon entry to
5830 every function (that actually uses some stack space). If there is
5831 insufficient space available then either the function
5832 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
5833 called, depending upon the amount of stack space required. The run time
5834 system is required to provide these functions. The default is
5835 @option{-mno-apcs-stack-check}, since this produces smaller code.
5837 @c not currently implemented
5839 @opindex mapcs-float
5840 Pass floating point arguments using the float point registers. This is
5841 one of the variants of the APCS@. This option is recommended if the
5842 target hardware has a floating point unit or if a lot of floating point
5843 arithmetic is going to be performed by the code. The default is
5844 @option{-mno-apcs-float}, since integer only code is slightly increased in
5845 size if @option{-mapcs-float} is used.
5847 @c not currently implemented
5848 @item -mapcs-reentrant
5849 @opindex mapcs-reentrant
5850 Generate reentrant, position independent code. The default is
5851 @option{-mno-apcs-reentrant}.
5854 @item -mthumb-interwork
5855 @opindex mthumb-interwork
5856 Generate code which supports calling between the ARM and Thumb
5857 instruction sets. Without this option the two instruction sets cannot
5858 be reliably used inside one program. The default is
5859 @option{-mno-thumb-interwork}, since slightly larger code is generated
5860 when @option{-mthumb-interwork} is specified.
5862 @item -mno-sched-prolog
5863 @opindex mno-sched-prolog
5864 Prevent the reordering of instructions in the function prolog, or the
5865 merging of those instruction with the instructions in the function's
5866 body. This means that all functions will start with a recognizable set
5867 of instructions (or in fact one of a choice from a small set of
5868 different function prologues), and this information can be used to
5869 locate the start if functions inside an executable piece of code. The
5870 default is @option{-msched-prolog}.
5873 @opindex mhard-float
5874 Generate output containing floating point instructions. This is the
5878 @opindex msoft-float
5879 Generate output containing library calls for floating point.
5880 @strong{Warning:} the requisite libraries are not available for all ARM
5881 targets. Normally the facilities of the machine's usual C compiler are
5882 used, but this cannot be done directly in cross-compilation. You must make
5883 your own arrangements to provide suitable library functions for
5886 @option{-msoft-float} changes the calling convention in the output file;
5887 therefore, it is only useful if you compile @emph{all} of a program with
5888 this option. In particular, you need to compile @file{libgcc.a}, the
5889 library that comes with GCC, with @option{-msoft-float} in order for
5892 @item -mlittle-endian
5893 @opindex mlittle-endian
5894 Generate code for a processor running in little-endian mode. This is
5895 the default for all standard configurations.
5898 @opindex mbig-endian
5899 Generate code for a processor running in big-endian mode; the default is
5900 to compile code for a little-endian processor.
5902 @item -mwords-little-endian
5903 @opindex mwords-little-endian
5904 This option only applies when generating code for big-endian processors.
5905 Generate code for a little-endian word order but a big-endian byte
5906 order. That is, a byte order of the form @samp{32107654}. Note: this
5907 option should only be used if you require compatibility with code for
5908 big-endian ARM processors generated by versions of the compiler prior to
5911 @item -malignment-traps
5912 @opindex malignment-traps
5913 Generate code that will not trap if the MMU has alignment traps enabled.
5914 On ARM architectures prior to ARMv4, there were no instructions to
5915 access half-word objects stored in memory. However, when reading from
5916 memory a feature of the ARM architecture allows a word load to be used,
5917 even if the address is unaligned, and the processor core will rotate the
5918 data as it is being loaded. This option tells the compiler that such
5919 misaligned accesses will cause a MMU trap and that it should instead
5920 synthesise the access as a series of byte accesses. The compiler can
5921 still use word accesses to load half-word data if it knows that the
5922 address is aligned to a word boundary.
5924 This option is ignored when compiling for ARM architecture 4 or later,
5925 since these processors have instructions to directly access half-word
5928 @item -mno-alignment-traps
5929 @opindex mno-alignment-traps
5930 Generate code that assumes that the MMU will not trap unaligned
5931 accesses. This produces better code when the target instruction set
5932 does not have half-word memory operations (i.e.@: implementations prior to
5935 Note that you cannot use this option to access unaligned word objects,
5936 since the processor will only fetch one 32-bit aligned object from
5939 The default setting for most targets is @option{-mno-alignment-traps}, since
5940 this produces better code when there are no half-word memory
5941 instructions available.
5943 @item -mshort-load-bytes
5944 @itemx -mno-short-load-words
5945 @opindex mshort-load-bytes
5946 @opindex mno-short-load-words
5947 These are deprecated aliases for @option{-malignment-traps}.
5949 @item -mno-short-load-bytes
5950 @itemx -mshort-load-words
5951 @opindex mno-short-load-bytes
5952 @opindex mshort-load-words
5953 This are deprecated aliases for @option{-mno-alignment-traps}.
5957 This option only applies to RISC iX@. Emulate the native BSD-mode
5958 compiler. This is the default if @option{-ansi} is not specified.
5962 This option only applies to RISC iX@. Emulate the native X/Open-mode
5965 @item -mno-symrename
5966 @opindex mno-symrename
5967 This option only applies to RISC iX@. Do not run the assembler
5968 post-processor, @samp{symrename}, after code has been assembled.
5969 Normally it is necessary to modify some of the standard symbols in
5970 preparation for linking with the RISC iX C library; this option
5971 suppresses this pass. The post-processor is never run when the
5972 compiler is built for cross-compilation.
5974 @item -mcpu=@var{name}
5976 This specifies the name of the target ARM processor. GCC uses this name
5977 to determine what kind of instructions it can emit when generating
5978 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
5979 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
5980 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
5981 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
5982 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
5983 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm8},
5984 @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
5985 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
5986 @samp{arm920t}, @samp{arm940t}, @samp{arm9tdmi}, @samp{arm10tdmi},
5987 @samp{arm1020t}, @samp{xscale}.
5989 @itemx -mtune=@var{name}
5991 This option is very similar to the @option{-mcpu=} option, except that
5992 instead of specifying the actual target processor type, and hence
5993 restricting which instructions can be used, it specifies that GCC should
5994 tune the performance of the code as if the target were of the type
5995 specified in this option, but still choosing the instructions that it
5996 will generate based on the cpu specified by a @option{-mcpu=} option.
5997 For some ARM implementations better performance can be obtained by using
6000 @item -march=@var{name}
6002 This specifies the name of the target ARM architecture. GCC uses this
6003 name to determine what kind of instructions it can emit when generating
6004 assembly code. This option can be used in conjunction with or instead
6005 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
6006 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
6007 @samp{armv5}, @samp{armv5t}, @samp{armv5te}.
6009 @item -mfpe=@var{number}
6010 @itemx -mfp=@var{number}
6013 This specifies the version of the floating point emulation available on
6014 the target. Permissible values are 2 and 3. @option{-mfp=} is a synonym
6015 for @option{-mfpe=}, for compatibility with older versions of GCC@.
6017 @item -mstructure-size-boundary=@var{n}
6018 @opindex mstructure-size-boundary
6019 The size of all structures and unions will be rounded up to a multiple
6020 of the number of bits set by this option. Permissible values are 8 and
6021 32. The default value varies for different toolchains. For the COFF
6022 targeted toolchain the default value is 8. Specifying the larger number
6023 can produce faster, more efficient code, but can also increase the size
6024 of the program. The two values are potentially incompatible. Code
6025 compiled with one value cannot necessarily expect to work with code or
6026 libraries compiled with the other value, if they exchange information
6027 using structures or unions.
6029 @item -mabort-on-noreturn
6030 @opindex mabort-on-noreturn
6031 Generate a call to the function @code{abort} at the end of a
6032 @code{noreturn} function. It will be executed if the function tries to
6036 @itemx -mno-long-calls
6037 @opindex mlong-calls
6038 @opindex mno-long-calls
6039 Tells the compiler to perform function calls by first loading the
6040 address of the function into a register and then performing a subroutine
6041 call on this register. This switch is needed if the target function
6042 will lie outside of the 64 megabyte addressing range of the offset based
6043 version of subroutine call instruction.
6045 Even if this switch is enabled, not all function calls will be turned
6046 into long calls. The heuristic is that static functions, functions
6047 which have the @samp{short-call} attribute, functions that are inside
6048 the scope of a @samp{#pragma no_long_calls} directive and functions whose
6049 definitions have already been compiled within the current compilation
6050 unit, will not be turned into long calls. The exception to this rule is
6051 that weak function definitions, functions with the @samp{long-call}
6052 attribute or the @samp{section} attribute, and functions that are within
6053 the scope of a @samp{#pragma long_calls} directive, will always be
6054 turned into long calls.
6056 This feature is not enabled by default. Specifying
6057 @option{-mno-long-calls} will restore the default behavior, as will
6058 placing the function calls within the scope of a @samp{#pragma
6059 long_calls_off} directive. Note these switches have no effect on how
6060 the compiler generates code to handle function calls via function
6063 @item -mnop-fun-dllimport
6064 @opindex mnop-fun-dllimport
6065 Disable support for the @code{dllimport} attribute.
6067 @item -msingle-pic-base
6068 @opindex msingle-pic-base
6069 Treat the register used for PIC addressing as read-only, rather than
6070 loading it in the prologue for each function. The run-time system is
6071 responsible for initializing this register with an appropriate value
6072 before execution begins.
6074 @item -mpic-register=@var{reg}
6075 @opindex mpic-register
6076 Specify the register to be used for PIC addressing. The default is R10
6077 unless stack-checking is enabled, when R9 is used.
6079 @item -mpoke-function-name
6080 @opindex mpoke-function-name
6081 Write the name of each function into the text section, directly
6082 preceding the function prologue. The generated code is similar to this:
6086 .ascii "arm_poke_function_name", 0
6089 .word 0xff000000 + (t1 - t0)
6090 arm_poke_function_name
6092 stmfd sp!, @{fp, ip, lr, pc@}
6096 When performing a stack backtrace, code can inspect the value of
6097 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
6098 location @code{pc - 12} and the top 8 bits are set, then we know that
6099 there is a function name embedded immediately preceding this location
6100 and has length @code{((pc[-3]) & 0xff000000)}.
6104 Generate code for the 16-bit Thumb instruction set. The default is to
6105 use the 32-bit ARM instruction set.
6108 @opindex mtpcs-frame
6109 Generate a stack frame that is compliant with the Thumb Procedure Call
6110 Standard for all non-leaf functions. (A leaf function is one that does
6111 not call any other functions.) The default is @option{-mno-tpcs-frame}.
6113 @item -mtpcs-leaf-frame
6114 @opindex mtpcs-leaf-frame
6115 Generate a stack frame that is compliant with the Thumb Procedure Call
6116 Standard for all leaf functions. (A leaf function is one that does
6117 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
6119 @item -mcallee-super-interworking
6120 @opindex mcallee-super-interworking
6121 Gives all externally visible functions in the file being compiled an ARM
6122 instruction set header which switches to Thumb mode before executing the
6123 rest of the function. This allows these functions to be called from
6124 non-interworking code.
6126 @item -mcaller-super-interworking
6127 @opindex mcaller-super-interworking
6128 Allows calls via function pointers (including virtual functions) to
6129 execute correctly regardless of whether the target code has been
6130 compiled for interworking or not. There is a small overhead in the cost
6131 of executing a function pointer if this option is enabled.
6135 @node MN10200 Options
6136 @subsection MN10200 Options
6137 @cindex MN10200 options
6138 These @option{-m} options are defined for Matsushita MN10200 architectures:
6143 Indicate to the linker that it should perform a relaxation optimization pass
6144 to shorten branches, calls and absolute memory addresses. This option only
6145 has an effect when used on the command line for the final link step.
6147 This option makes symbolic debugging impossible.
6150 @node MN10300 Options
6151 @subsection MN10300 Options
6152 @cindex MN10300 options
6153 These @option{-m} options are defined for Matsushita MN10300 architectures:
6158 Generate code to avoid bugs in the multiply instructions for the MN10300
6159 processors. This is the default.
6162 @opindex mno-mult-bug
6163 Do not generate code to avoid bugs in the multiply instructions for the
6168 Generate code which uses features specific to the AM33 processor.
6172 Do not generate code which uses features specific to the AM33 processor. This
6177 Do not link in the C run-time initialization object file.
6181 Indicate to the linker that it should perform a relaxation optimization pass
6182 to shorten branches, calls and absolute memory addresses. This option only
6183 has an effect when used on the command line for the final link step.
6185 This option makes symbolic debugging impossible.
6189 @node M32R/D Options
6190 @subsection M32R/D Options
6191 @cindex M32R/D options
6193 These @option{-m} options are defined for Mitsubishi M32R/D architectures:
6198 Generate code for the M32R/X@.
6202 Generate code for the M32R@. This is the default.
6204 @item -mcode-model=small
6205 @opindex mcode-model=small
6206 Assume all objects live in the lower 16MB of memory (so that their addresses
6207 can be loaded with the @code{ld24} instruction), and assume all subroutines
6208 are reachable with the @code{bl} instruction.
6209 This is the default.
6211 The addressability of a particular object can be set with the
6212 @code{model} attribute.
6214 @item -mcode-model=medium
6215 @opindex mcode-model=medium
6216 Assume objects may be anywhere in the 32-bit address space (the compiler
6217 will generate @code{seth/add3} instructions to load their addresses), and
6218 assume all subroutines are reachable with the @code{bl} instruction.
6220 @item -mcode-model=large
6221 @opindex mcode-model=large
6222 Assume objects may be anywhere in the 32-bit address space (the compiler
6223 will generate @code{seth/add3} instructions to load their addresses), and
6224 assume subroutines may not be reachable with the @code{bl} instruction
6225 (the compiler will generate the much slower @code{seth/add3/jl}
6226 instruction sequence).
6229 @opindex msdata=none
6230 Disable use of the small data area. Variables will be put into
6231 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
6232 @code{section} attribute has been specified).
6233 This is the default.
6235 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
6236 Objects may be explicitly put in the small data area with the
6237 @code{section} attribute using one of these sections.
6240 @opindex msdata=sdata
6241 Put small global and static data in the small data area, but do not
6242 generate special code to reference them.
6246 Put small global and static data in the small data area, and generate
6247 special instructions to reference them.
6251 @cindex smaller data references
6252 Put global and static objects less than or equal to @var{num} bytes
6253 into the small data or bss sections instead of the normal data or bss
6254 sections. The default value of @var{num} is 8.
6255 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
6256 for this option to have any effect.
6258 All modules should be compiled with the same @option{-G @var{num}} value.
6259 Compiling with different values of @var{num} may or may not work; if it
6260 doesn't the linker will give an error message---incorrect code will not be
6266 @subsection M88K Options
6267 @cindex M88k options
6269 These @samp{-m} options are defined for Motorola 88k architectures:
6274 Generate code that works well on both the m88100 and the
6279 Generate code that works best for the m88100, but that also
6284 Generate code that works best for the m88110, and may not run
6289 Obsolete option to be removed from the next revision.
6292 @item -midentify-revision
6293 @opindex midentify-revision
6294 @cindex identifying source, compiler (88k)
6295 Include an @code{ident} directive in the assembler output recording the
6296 source file name, compiler name and version, timestamp, and compilation
6299 @item -mno-underscores
6300 @opindex mno-underscores
6301 @cindex underscores, avoiding (88k)
6302 In assembler output, emit symbol names without adding an underscore
6303 character at the beginning of each name. The default is to use an
6304 underscore as prefix on each name.
6306 @item -mocs-debug-info
6307 @itemx -mno-ocs-debug-info
6308 @opindex mocs-debug-info
6309 @opindex mno-ocs-debug-info
6311 @cindex debugging, 88k OCS
6312 Include (or omit) additional debugging information (about registers used
6313 in each stack frame) as specified in the 88open Object Compatibility
6314 Standard, ``OCS''@. This extra information allows debugging of code that
6315 has had the frame pointer eliminated. The default for DG/UX, SVr4, and
6316 Delta 88 SVr3.2 is to include this information; other 88k configurations
6317 omit this information by default.
6319 @item -mocs-frame-position
6320 @opindex mocs-frame-position
6321 @cindex register positions in frame (88k)
6322 When emitting COFF debugging information for automatic variables and
6323 parameters stored on the stack, use the offset from the canonical frame
6324 address, which is the stack pointer (register 31) on entry to the
6325 function. The DG/UX, SVr4, Delta88 SVr3.2, and BCS configurations use
6326 @option{-mocs-frame-position}; other 88k configurations have the default
6327 @option{-mno-ocs-frame-position}.
6329 @item -mno-ocs-frame-position
6330 @opindex mno-ocs-frame-position
6331 @cindex register positions in frame (88k)
6332 When emitting COFF debugging information for automatic variables and
6333 parameters stored on the stack, use the offset from the frame pointer
6334 register (register 30). When this option is in effect, the frame
6335 pointer is not eliminated when debugging information is selected by the
6338 @item -moptimize-arg-area
6339 @opindex moptimize-arg-area
6340 @cindex arguments in frame (88k)
6341 Save space by reorganizing the stack frame. This option generates code
6342 that does not agree with the 88open specifications, but uses less
6345 @itemx -mno-optimize-arg-area
6346 @opindex mno-optimize-arg-area
6347 Do not reorganize the stack frame to save space. This is the default.
6348 The generated conforms to the specification, but uses more memory.
6350 @item -mshort-data-@var{num}
6351 @opindex mshort-data
6352 @cindex smaller data references (88k)
6353 @cindex r0-relative references (88k)
6354 Generate smaller data references by making them relative to @code{r0},
6355 which allows loading a value using a single instruction (rather than the
6356 usual two). You control which data references are affected by
6357 specifying @var{num} with this option. For example, if you specify
6358 @option{-mshort-data-512}, then the data references affected are those
6359 involving displacements of less than 512 bytes.
6360 @option{-mshort-data-@var{num}} is not effective for @var{num} greater
6363 @item -mserialize-volatile
6364 @opindex mserialize-volatile
6365 @itemx -mno-serialize-volatile
6366 @opindex mno-serialize-volatile
6367 @cindex sequential consistency on 88k
6368 Do, or don't, generate code to guarantee sequential consistency
6369 of volatile memory references. By default, consistency is
6372 The order of memory references made by the MC88110 processor does
6373 not always match the order of the instructions requesting those
6374 references. In particular, a load instruction may execute before
6375 a preceding store instruction. Such reordering violates
6376 sequential consistency of volatile memory references, when there
6377 are multiple processors. When consistency must be guaranteed,
6378 GCC generates special instructions, as needed, to force
6379 execution in the proper order.
6381 The MC88100 processor does not reorder memory references and so
6382 always provides sequential consistency. However, by default, GCC
6383 generates the special instructions to guarantee consistency
6384 even when you use @option{-m88100}, so that the code may be run on an
6385 MC88110 processor. If you intend to run your code only on the
6386 MC88100 processor, you may use @option{-mno-serialize-volatile}.
6388 The extra code generated to guarantee consistency may affect the
6389 performance of your application. If you know that you can safely
6390 forgo this guarantee, you may use @option{-mno-serialize-volatile}.
6396 @cindex assembler syntax, 88k
6398 Turn on (@option{-msvr4}) or off (@option{-msvr3}) compiler extensions
6399 related to System V release 4 (SVr4). This controls the following:
6403 Which variant of the assembler syntax to emit.
6405 @option{-msvr4} makes the C preprocessor recognize @samp{#pragma weak}
6406 that is used on System V release 4.
6408 @option{-msvr4} makes GCC issue additional declaration directives used in
6412 @option{-msvr4} is the default for the m88k-motorola-sysv4 and
6413 m88k-dg-dgux m88k configurations. @option{-msvr3} is the default for all
6414 other m88k configurations.
6416 @item -mversion-03.00
6417 @opindex mversion-03.00
6418 This option is obsolete, and is ignored.
6419 @c ??? which asm syntax better for GAS? option there too?
6421 @item -mno-check-zero-division
6422 @itemx -mcheck-zero-division
6423 @opindex mno-check-zero-division
6424 @opindex mcheck-zero-division
6425 @cindex zero division on 88k
6426 Do, or don't, generate code to guarantee that integer division by
6427 zero will be detected. By default, detection is guaranteed.
6429 Some models of the MC88100 processor fail to trap upon integer
6430 division by zero under certain conditions. By default, when
6431 compiling code that might be run on such a processor, GCC
6432 generates code that explicitly checks for zero-valued divisors
6433 and traps with exception number 503 when one is detected. Use of
6434 @option{-mno-check-zero-division} suppresses such checking for code
6435 generated to run on an MC88100 processor.
6437 GCC assumes that the MC88110 processor correctly detects all instances
6438 of integer division by zero. When @option{-m88110} is specified, no
6439 explicit checks for zero-valued divisors are generated, and both
6440 @option{-mcheck-zero-division} and @option{-mno-check-zero-division} are
6443 @item -muse-div-instruction
6444 @opindex muse-div-instruction
6445 @cindex divide instruction, 88k
6446 Use the div instruction for signed integer division on the
6447 MC88100 processor. By default, the div instruction is not used.
6449 On the MC88100 processor the signed integer division instruction
6450 div) traps to the operating system on a negative operand. The
6451 operating system transparently completes the operation, but at a
6452 large cost in execution time. By default, when compiling code
6453 that might be run on an MC88100 processor, GCC emulates signed
6454 integer division using the unsigned integer division instruction
6455 divu), thereby avoiding the large penalty of a trap to the
6456 operating system. Such emulation has its own, smaller, execution
6457 cost in both time and space. To the extent that your code's
6458 important signed integer division operations are performed on two
6459 nonnegative operands, it may be desirable to use the div
6460 instruction directly.
6462 On the MC88110 processor the div instruction (also known as the
6463 divs instruction) processes negative operands without trapping to
6464 the operating system. When @option{-m88110} is specified,
6465 @option{-muse-div-instruction} is ignored, and the div instruction is used
6466 for signed integer division.
6468 Note that the result of dividing @code{INT_MIN} by @minus{}1 is undefined. In
6469 particular, the behavior of such a division with and without
6470 @option{-muse-div-instruction} may differ.
6472 @item -mtrap-large-shift
6473 @itemx -mhandle-large-shift
6474 @opindex mtrap-large-shift
6475 @opindex mhandle-large-shift
6476 @cindex bit shift overflow (88k)
6477 @cindex large bit shifts (88k)
6478 Include code to detect bit-shifts of more than 31 bits; respectively,
6479 trap such shifts or emit code to handle them properly. By default GCC
6480 makes no special provision for large bit shifts.
6482 @item -mwarn-passed-structs
6483 @opindex mwarn-passed-structs
6484 @cindex structure passing (88k)
6485 Warn when a function passes a struct as an argument or result.
6486 Structure-passing conventions have changed during the evolution of the C
6487 language, and are often the source of portability problems. By default,
6488 GCC issues no such warning.
6491 @c break page here to avoid unsightly interparagraph stretch.
6495 @node RS/6000 and PowerPC Options
6496 @subsection IBM RS/6000 and PowerPC Options
6497 @cindex RS/6000 and PowerPC Options
6498 @cindex IBM RS/6000 and PowerPC Options
6500 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
6508 @itemx -mpowerpc-gpopt
6509 @itemx -mno-powerpc-gpopt
6510 @itemx -mpowerpc-gfxopt
6511 @itemx -mno-powerpc-gfxopt
6513 @itemx -mno-powerpc64
6519 @opindex mno-powerpc
6520 @opindex mpowerpc-gpopt
6521 @opindex mno-powerpc-gpopt
6522 @opindex mpowerpc-gfxopt
6523 @opindex mno-powerpc-gfxopt
6525 @opindex mno-powerpc64
6526 GCC supports two related instruction set architectures for the
6527 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
6528 instructions supported by the @samp{rios} chip set used in the original
6529 RS/6000 systems and the @dfn{PowerPC} instruction set is the
6530 architecture of the Motorola MPC5xx, MPC6xx, MPC8xx microprocessors, and
6531 the IBM 4xx microprocessors.
6533 Neither architecture is a subset of the other. However there is a
6534 large common subset of instructions supported by both. An MQ
6535 register is included in processors supporting the POWER architecture.
6537 You use these options to specify which instructions are available on the
6538 processor you are using. The default value of these options is
6539 determined when configuring GCC@. Specifying the
6540 @option{-mcpu=@var{cpu_type}} overrides the specification of these
6541 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
6542 rather than the options listed above.
6544 The @option{-mpower} option allows GCC to generate instructions that
6545 are found only in the POWER architecture and to use the MQ register.
6546 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
6547 to generate instructions that are present in the POWER2 architecture but
6548 not the original POWER architecture.
6550 The @option{-mpowerpc} option allows GCC to generate instructions that
6551 are found only in the 32-bit subset of the PowerPC architecture.
6552 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
6553 GCC to use the optional PowerPC architecture instructions in the
6554 General Purpose group, including floating-point square root. Specifying
6555 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
6556 use the optional PowerPC architecture instructions in the Graphics
6557 group, including floating-point select.
6559 The @option{-mpowerpc64} option allows GCC to generate the additional
6560 64-bit instructions that are found in the full PowerPC64 architecture
6561 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
6562 @option{-mno-powerpc64}.
6564 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
6565 will use only the instructions in the common subset of both
6566 architectures plus some special AIX common-mode calls, and will not use
6567 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
6568 permits GCC to use any instruction from either architecture and to
6569 allow use of the MQ register; specify this for the Motorola MPC601.
6571 @item -mnew-mnemonics
6572 @itemx -mold-mnemonics
6573 @opindex mnew-mnemonics
6574 @opindex mold-mnemonics
6575 Select which mnemonics to use in the generated assembler code. With
6576 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
6577 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
6578 assembler mnemonics defined for the POWER architecture. Instructions
6579 defined in only one architecture have only one mnemonic; GCC uses that
6580 mnemonic irrespective of which of these options is specified.
6582 GCC defaults to the mnemonics appropriate for the architecture in
6583 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
6584 value of these option. Unless you are building a cross-compiler, you
6585 should normally not specify either @option{-mnew-mnemonics} or
6586 @option{-mold-mnemonics}, but should instead accept the default.
6588 @item -mcpu=@var{cpu_type}
6590 Set architecture type, register usage, choice of mnemonics, and
6591 instruction scheduling parameters for machine type @var{cpu_type}.
6592 Supported values for @var{cpu_type} are @samp{rios}, @samp{rios1},
6593 @samp{rsc}, @samp{rios2}, @samp{rs64a}, @samp{601}, @samp{602},
6594 @samp{603}, @samp{603e}, @samp{604}, @samp{604e}, @samp{620},
6595 @samp{630}, @samp{740}, @samp{7400}, @samp{7450}, @samp{750},
6596 @samp{power}, @samp{power2}, @samp{powerpc}, @samp{403}, @samp{505},
6597 @samp{801}, @samp{821}, @samp{823}, and @samp{860} and @samp{common}.
6599 @option{-mcpu=common} selects a completely generic processor. Code
6600 generated under this option will run on any POWER or PowerPC processor.
6601 GCC will use only the instructions in the common subset of both
6602 architectures, and will not use the MQ register. GCC assumes a generic
6603 processor model for scheduling purposes.
6605 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
6606 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
6607 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
6608 types, with an appropriate, generic processor model assumed for
6609 scheduling purposes.
6611 The other options specify a specific processor. Code generated under
6612 those options will run best on that processor, and may not run at all on
6615 The @option{-mcpu} options automatically enable or disable other
6616 @option{-m} options as follows:
6620 @option{-mno-power}, @option{-mno-powerc}
6627 @option{-mpower}, @option{-mno-powerpc}, @option{-mno-new-mnemonics}
6642 @option{-mno-power}, @option{-mpowerpc}, @option{-mnew-mnemonics}
6645 @option{-mpower}, @option{-mpowerpc}, @option{-mnew-mnemonics}
6650 @option{-mno-power}, @option{-mpowerpc}, @option{-mnew-mnemonics}, @option{-msoft-float}
6653 @item -mtune=@var{cpu_type}
6655 Set the instruction scheduling parameters for machine type
6656 @var{cpu_type}, but do not set the architecture type, register usage, or
6657 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
6658 values for @var{cpu_type} are used for @option{-mtune} as for
6659 @option{-mcpu}. If both are specified, the code generated will use the
6660 architecture, registers, and mnemonics set by @option{-mcpu}, but the
6661 scheduling parameters set by @option{-mtune}.
6666 @opindex mno-altivec
6667 These switches enable or disable the use of built-in functions that
6668 allow access to the AltiVec instruction set. You may also need to set
6669 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
6673 @itemx -mno-fp-in-toc
6674 @itemx -mno-sum-in-toc
6675 @itemx -mminimal-toc
6677 @opindex mno-fp-in-toc
6678 @opindex mno-sum-in-toc
6679 @opindex mminimal-toc
6680 Modify generation of the TOC (Table Of Contents), which is created for
6681 every executable file. The @option{-mfull-toc} option is selected by
6682 default. In that case, GCC will allocate at least one TOC entry for
6683 each unique non-automatic variable reference in your program. GCC
6684 will also place floating-point constants in the TOC@. However, only
6685 16,384 entries are available in the TOC@.
6687 If you receive a linker error message that saying you have overflowed
6688 the available TOC space, you can reduce the amount of TOC space used
6689 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
6690 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
6691 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
6692 generate code to calculate the sum of an address and a constant at
6693 run-time instead of putting that sum into the TOC@. You may specify one
6694 or both of these options. Each causes GCC to produce very slightly
6695 slower and larger code at the expense of conserving TOC space.
6697 If you still run out of space in the TOC even when you specify both of
6698 these options, specify @option{-mminimal-toc} instead. This option causes
6699 GCC to make only one TOC entry for every file. When you specify this
6700 option, GCC will produce code that is slower and larger but which
6701 uses extremely little TOC space. You may wish to use this option
6702 only on files that contain less frequently executed code.
6708 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
6709 @code{long} type, and the infrastructure needed to support them.
6710 Specifying @option{-maix64} implies @option{-mpowerpc64} and
6711 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
6712 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
6717 @opindex mno-xl-call
6718 On AIX, pass floating-point arguments to prototyped functions beyond the
6719 register save area (RSA) on the stack in addition to argument FPRs. The
6720 AIX calling convention was extended but not initially documented to
6721 handle an obscure K&R C case of calling a function that takes the
6722 address of its arguments with fewer arguments than declared. AIX XL
6723 compilers access floating point arguments which do not fit in the
6724 RSA from the stack when a subroutine is compiled without
6725 optimization. Because always storing floating-point arguments on the
6726 stack is inefficient and rarely needed, this option is not enabled by
6727 default and only is necessary when calling subroutines compiled by AIX
6728 XL compilers without optimization.
6732 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
6733 application written to use message passing with special startup code to
6734 enable the application to run. The system must have PE installed in the
6735 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
6736 must be overridden with the @option{-specs=} option to specify the
6737 appropriate directory location. The Parallel Environment does not
6738 support threads, so the @option{-mpe} option and the @option{-pthread}
6739 option are incompatible.
6743 @opindex msoft-float
6744 @opindex mhard-float
6745 Generate code that does not use (uses) the floating-point register set.
6746 Software floating point emulation is provided if you use the
6747 @option{-msoft-float} option, and pass the option to GCC when linking.
6750 @itemx -mno-multiple
6752 @opindex mno-multiple
6753 Generate code that uses (does not use) the load multiple word
6754 instructions and the store multiple word instructions. These
6755 instructions are generated by default on POWER systems, and not
6756 generated on PowerPC systems. Do not use @option{-mmultiple} on little
6757 endian PowerPC systems, since those instructions do not work when the
6758 processor is in little endian mode. The exceptions are PPC740 and
6759 PPC750 which permit the instructions usage in little endian mode.
6765 Generate code that uses (does not use) the load string instructions
6766 and the store string word instructions to save multiple registers and
6767 do small block moves. These instructions are generated by default on
6768 POWER systems, and not generated on PowerPC systems. Do not use
6769 @option{-mstring} on little endian PowerPC systems, since those
6770 instructions do not work when the processor is in little endian mode.
6771 The exceptions are PPC740 and PPC750 which permit the instructions
6772 usage in little endian mode.
6778 Generate code that uses (does not use) the load or store instructions
6779 that update the base register to the address of the calculated memory
6780 location. These instructions are generated by default. If you use
6781 @option{-mno-update}, there is a small window between the time that the
6782 stack pointer is updated and the address of the previous frame is
6783 stored, which means code that walks the stack frame across interrupts or
6784 signals may get corrupted data.
6787 @itemx -mno-fused-madd
6788 @opindex mfused-madd
6789 @opindex mno-fused-madd
6790 Generate code that uses (does not use) the floating point multiply and
6791 accumulate instructions. These instructions are generated by default if
6792 hardware floating is used.
6794 @item -mno-bit-align
6796 @opindex mno-bit-align
6798 On System V.4 and embedded PowerPC systems do not (do) force structures
6799 and unions that contain bit-fields to be aligned to the base type of the
6802 For example, by default a structure containing nothing but 8
6803 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
6804 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
6805 the structure would be aligned to a 1 byte boundary and be one byte in
6808 @item -mno-strict-align
6809 @itemx -mstrict-align
6810 @opindex mno-strict-align
6811 @opindex mstrict-align
6812 On System V.4 and embedded PowerPC systems do not (do) assume that
6813 unaligned memory references will be handled by the system.
6816 @itemx -mno-relocatable
6817 @opindex mrelocatable
6818 @opindex mno-relocatable
6819 On embedded PowerPC systems generate code that allows (does not allow)
6820 the program to be relocated to a different address at runtime. If you
6821 use @option{-mrelocatable} on any module, all objects linked together must
6822 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
6824 @item -mrelocatable-lib
6825 @itemx -mno-relocatable-lib
6826 @opindex mrelocatable-lib
6827 @opindex mno-relocatable-lib
6828 On embedded PowerPC systems generate code that allows (does not allow)
6829 the program to be relocated to a different address at runtime. Modules
6830 compiled with @option{-mrelocatable-lib} can be linked with either modules
6831 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
6832 with modules compiled with the @option{-mrelocatable} options.
6838 On System V.4 and embedded PowerPC systems do not (do) assume that
6839 register 2 contains a pointer to a global area pointing to the addresses
6840 used in the program.
6843 @itemx -mlittle-endian
6845 @opindex mlittle-endian
6846 On System V.4 and embedded PowerPC systems compile code for the
6847 processor in little endian mode. The @option{-mlittle-endian} option is
6848 the same as @option{-mlittle}.
6853 @opindex mbig-endian
6854 On System V.4 and embedded PowerPC systems compile code for the
6855 processor in big endian mode. The @option{-mbig-endian} option is
6856 the same as @option{-mbig}.
6860 On System V.4 and embedded PowerPC systems compile code using calling
6861 conventions that adheres to the March 1995 draft of the System V
6862 Application Binary Interface, PowerPC processor supplement. This is the
6863 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
6865 @item -mcall-sysv-eabi
6866 @opindex mcall-sysv-eabi
6867 Specify both @option{-mcall-sysv} and @option{-meabi} options.
6869 @item -mcall-sysv-noeabi
6870 @opindex mcall-sysv-noeabi
6871 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
6875 On System V.4 and embedded PowerPC systems compile code using calling
6876 conventions that are similar to those used on AIX@. This is the
6877 default if you configured GCC using @samp{powerpc-*-eabiaix}.
6879 @item -mcall-solaris
6880 @opindex mcall-solaris
6881 On System V.4 and embedded PowerPC systems compile code for the Solaris
6885 @opindex mcall-linux
6886 On System V.4 and embedded PowerPC systems compile code for the
6887 Linux-based GNU system.
6891 On System V.4 and embedded PowerPC systems compile code for the
6892 Hurd-based GNU system.
6895 @opindex mcall-netbsd
6896 On System V.4 and embedded PowerPC systems compile code for the
6897 NetBSD operating system.
6899 @item -maix-struct-return
6900 @opindex maix-struct-return
6901 Return all structures in memory (as specified by the AIX ABI)@.
6903 @item -msvr4-struct-return
6904 @opindex msvr4-struct-return
6905 Return structures smaller than 8 bytes in registers (as specified by the
6909 @opindex mabi=altivec
6910 Extend the current ABI with AltiVec ABI extensions. This does not
6911 change the default ABI, instead it adds the AltiVec ABI extensions to
6915 @itemx -mno-prototype
6917 @opindex mno-prototype
6918 On System V.4 and embedded PowerPC systems assume that all calls to
6919 variable argument functions are properly prototyped. Otherwise, the
6920 compiler must insert an instruction before every non prototyped call to
6921 set or clear bit 6 of the condition code register (@var{CR}) to
6922 indicate whether floating point values were passed in the floating point
6923 registers in case the function takes a variable arguments. With
6924 @option{-mprototype}, only calls to prototyped variable argument functions
6925 will set or clear the bit.
6929 On embedded PowerPC systems, assume that the startup module is called
6930 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
6931 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
6936 On embedded PowerPC systems, assume that the startup module is called
6937 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
6942 On embedded PowerPC systems, assume that the startup module is called
6943 @file{crt0.o} and the standard C libraries are @file{libads.a} and
6947 @opindex myellowknife
6948 On embedded PowerPC systems, assume that the startup module is called
6949 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
6954 On System V.4 and embedded PowerPC systems, specify that you are
6955 compiling for a VxWorks system.
6959 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
6960 header to indicate that @samp{eabi} extended relocations are used.
6966 On System V.4 and embedded PowerPC systems do (do not) adhere to the
6967 Embedded Applications Binary Interface (eabi) which is a set of
6968 modifications to the System V.4 specifications. Selecting @option{-meabi}
6969 means that the stack is aligned to an 8 byte boundary, a function
6970 @code{__eabi} is called to from @code{main} to set up the eabi
6971 environment, and the @option{-msdata} option can use both @code{r2} and
6972 @code{r13} to point to two separate small data areas. Selecting
6973 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
6974 do not call an initialization function from @code{main}, and the
6975 @option{-msdata} option will only use @code{r13} to point to a single
6976 small data area. The @option{-meabi} option is on by default if you
6977 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
6980 @opindex msdata=eabi
6981 On System V.4 and embedded PowerPC systems, put small initialized
6982 @code{const} global and static data in the @samp{.sdata2} section, which
6983 is pointed to by register @code{r2}. Put small initialized
6984 non-@code{const} global and static data in the @samp{.sdata} section,
6985 which is pointed to by register @code{r13}. Put small uninitialized
6986 global and static data in the @samp{.sbss} section, which is adjacent to
6987 the @samp{.sdata} section. The @option{-msdata=eabi} option is
6988 incompatible with the @option{-mrelocatable} option. The
6989 @option{-msdata=eabi} option also sets the @option{-memb} option.
6992 @opindex msdata=sysv
6993 On System V.4 and embedded PowerPC systems, put small global and static
6994 data in the @samp{.sdata} section, which is pointed to by register
6995 @code{r13}. Put small uninitialized global and static data in the
6996 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
6997 The @option{-msdata=sysv} option is incompatible with the
6998 @option{-mrelocatable} option.
7000 @item -msdata=default
7002 @opindex msdata=default
7004 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
7005 compile code the same as @option{-msdata=eabi}, otherwise compile code the
7006 same as @option{-msdata=sysv}.
7009 @opindex msdata-data
7010 On System V.4 and embedded PowerPC systems, put small global and static
7011 data in the @samp{.sdata} section. Put small uninitialized global and
7012 static data in the @samp{.sbss} section. Do not use register @code{r13}
7013 to address small data however. This is the default behavior unless
7014 other @option{-msdata} options are used.
7018 @opindex msdata=none
7020 On embedded PowerPC systems, put all initialized global and static data
7021 in the @samp{.data} section, and all uninitialized data in the
7022 @samp{.bss} section.
7026 @cindex smaller data references (PowerPC)
7027 @cindex .sdata/.sdata2 references (PowerPC)
7028 On embedded PowerPC systems, put global and static items less than or
7029 equal to @var{num} bytes into the small data or bss sections instead of
7030 the normal data or bss section. By default, @var{num} is 8. The
7031 @option{-G @var{num}} switch is also passed to the linker.
7032 All modules should be compiled with the same @option{-G @var{num}} value.
7035 @itemx -mno-regnames
7037 @opindex mno-regnames
7038 On System V.4 and embedded PowerPC systems do (do not) emit register
7039 names in the assembly language output using symbolic forms.
7043 Adds support for multithreading with the @dfn{pthreads} library.
7044 This option sets flags for both the preprocessor and linker.
7049 @subsection IBM RT Options
7051 @cindex IBM RT options
7053 These @samp{-m} options are defined for the IBM RT PC:
7057 @opindex min-line-mul
7058 Use an in-line code sequence for integer multiplies. This is the
7061 @item -mcall-lib-mul
7062 @opindex mcall-lib-mul
7063 Call @code{lmul$$} for integer multiples.
7065 @item -mfull-fp-blocks
7066 @opindex mfull-fp-blocks
7067 Generate full-size floating point data blocks, including the minimum
7068 amount of scratch space recommended by IBM@. This is the default.
7070 @item -mminimum-fp-blocks
7071 @opindex mminimum-fp-blocks
7072 Do not include extra scratch space in floating point data blocks. This
7073 results in smaller code, but slower execution, since scratch space must
7074 be allocated dynamically.
7076 @cindex @file{varargs.h} and RT PC
7077 @cindex @file{stdarg.h} and RT PC
7078 @item -mfp-arg-in-fpregs
7079 @opindex mfp-arg-in-fpregs
7080 Use a calling sequence incompatible with the IBM calling convention in
7081 which floating point arguments are passed in floating point registers.
7082 Note that @code{varargs.h} and @code{stdarg.h} will not work with
7083 floating point operands if this option is specified.
7085 @item -mfp-arg-in-gregs
7086 @opindex mfp-arg-in-gregs
7087 Use the normal calling convention for floating point arguments. This is
7090 @item -mhc-struct-return
7091 @opindex mhc-struct-return
7092 Return structures of more than one word in memory, rather than in a
7093 register. This provides compatibility with the MetaWare HighC (hc)
7094 compiler. Use the option @option{-fpcc-struct-return} for compatibility
7095 with the Portable C Compiler (pcc).
7097 @item -mnohc-struct-return
7098 @opindex mnohc-struct-return
7099 Return some structures of more than one word in registers, when
7100 convenient. This is the default. For compatibility with the
7101 IBM-supplied compilers, use the option @option{-fpcc-struct-return} or the
7102 option @option{-mhc-struct-return}.
7106 @subsection MIPS Options
7107 @cindex MIPS options
7109 These @samp{-m} options are defined for the MIPS family of computers:
7113 @item -march=@var{cpu-type}
7115 Assume the defaults for the machine type @var{cpu-type} when generating
7116 instructions. The choices for @var{cpu-type} are @samp{r2000}, @samp{r3000},
7117 @samp{r3900}, @samp{r4000}, @samp{r4100}, @samp{r4300}, @samp{r4400},
7118 @samp{r4600}, @samp{r4650}, @samp{r5000}, @samp{r6000}, @samp{r8000},
7119 and @samp{orion}. Additionally, the @samp{r2000}, @samp{r3000},
7120 @samp{r4000}, @samp{r5000}, and @samp{r6000} can be abbreviated as
7121 @samp{r2k} (or @samp{r2K}), @samp{r3k}, etc.
7123 @item -mtune=@var{cpu-type}
7125 Assume the defaults for the machine type @var{cpu-type} when scheduling
7126 instructions. The choices for @var{cpu-type} are @samp{r2000}, @samp{r3000},
7127 @samp{r3900}, @samp{r4000}, @samp{r4100}, @samp{r4300}, @samp{r4400},
7128 @samp{r4600}, @samp{r4650}, @samp{r5000}, @samp{r6000}, @samp{r8000},
7129 and @samp{orion}. Additionally, the @samp{r2000}, @samp{r3000},
7130 @samp{r4000}, @samp{r5000}, and @samp{r6000} can be abbreviated as
7131 @samp{r2k} (or @samp{r2K}), @samp{r3k}, etc. While picking a specific
7132 @var{cpu-type} will schedule things appropriately for that particular
7133 chip, the compiler will not generate any code that does not meet level 1
7134 of the MIPS ISA (instruction set architecture) without a @option{-mipsX}
7135 or @option{-mabi} switch being used.
7137 @item -mcpu=@var{cpu-type}
7139 This is identical to specifying both @option{-march} and @option{-mtune}.
7143 Issue instructions from level 1 of the MIPS ISA@. This is the default.
7144 @samp{r3000} is the default @var{cpu-type} at this ISA level.
7148 Issue instructions from level 2 of the MIPS ISA (branch likely, square
7149 root instructions). @samp{r6000} is the default @var{cpu-type} at this
7154 Issue instructions from level 3 of the MIPS ISA (64-bit instructions).
7155 @samp{r4000} is the default @var{cpu-type} at this ISA level.
7159 Issue instructions from level 4 of the MIPS ISA (conditional move,
7160 prefetch, enhanced FPU instructions). @samp{r8000} is the default
7161 @var{cpu-type} at this ISA level.
7165 Assume that 32 32-bit floating point registers are available. This is
7170 Assume that 32 64-bit floating point registers are available. This is
7171 the default when the @option{-mips3} option is used.
7174 @itemx -mno-fused-madd
7175 @opindex mfused-madd
7176 @opindex mno-fused-madd
7177 Generate code that uses (does not use) the floating point multiply and
7178 accumulate instructions, when they are available. These instructions
7179 are generated by default if they are available, but this may be
7180 undesirable if the extra precision causes problems or on certain chips
7181 in the mode where denormals are rounded to zero where denormals
7182 generated by multiply and accumulate instructions cause exceptions
7187 Assume that 32 32-bit general purpose registers are available. This is
7192 Assume that 32 64-bit general purpose registers are available. This is
7193 the default when the @option{-mips3} option is used.
7197 Force int and long types to be 64 bits wide. See @option{-mlong32} for an
7198 explanation of the default, and the width of pointers.
7202 Force long types to be 64 bits wide. See @option{-mlong32} for an
7203 explanation of the default, and the width of pointers.
7207 Force long, int, and pointer types to be 32 bits wide.
7209 If none of @option{-mlong32}, @option{-mlong64}, or @option{-mint64} are set,
7210 the size of ints, longs, and pointers depends on the ABI and ISA chosen.
7211 For @option{-mabi=32}, and @option{-mabi=n32}, ints and longs are 32 bits
7212 wide. For @option{-mabi=64}, ints are 32 bits, and longs are 64 bits wide.
7213 For @option{-mabi=eabi} and either @option{-mips1} or @option{-mips2}, ints
7214 and longs are 32 bits wide. For @option{-mabi=eabi} and higher ISAs, ints
7215 are 32 bits, and longs are 64 bits wide. The width of pointer types is
7216 the smaller of the width of longs or the width of general purpose
7217 registers (which in turn depends on the ISA)@.
7229 Generate code for the indicated ABI@. The default instruction level is
7230 @option{-mips1} for @samp{32}, @option{-mips3} for @samp{n32}, and
7231 @option{-mips4} otherwise. Conversely, with @option{-mips1} or
7232 @option{-mips2}, the default ABI is @samp{32}; otherwise, the default ABI
7237 Generate code for the MIPS assembler, and invoke @file{mips-tfile} to
7238 add normal debug information. This is the default for all
7239 platforms except for the OSF/1 reference platform, using the OSF/rose
7240 object format. If the either of the @option{-gstabs} or @option{-gstabs+}
7241 switches are used, the @file{mips-tfile} program will encapsulate the
7242 stabs within MIPS ECOFF@.
7246 Generate code for the GNU assembler. This is the default on the OSF/1
7247 reference platform, using the OSF/rose object format. Also, this is
7248 the default if the configure option @option{--with-gnu-as} is used.
7250 @item -msplit-addresses
7251 @itemx -mno-split-addresses
7252 @opindex msplit-addresses
7253 @opindex mno-split-addresses
7254 Generate code to load the high and low parts of address constants separately.
7255 This allows GCC to optimize away redundant loads of the high order
7256 bits of addresses. This optimization requires GNU as and GNU ld.
7257 This optimization is enabled by default for some embedded targets where
7258 GNU as and GNU ld are standard.
7264 The @option{-mrnames} switch says to output code using the MIPS software
7265 names for the registers, instead of the hardware names (ie, @var{a0}
7266 instead of @var{$4}). The only known assembler that supports this option
7267 is the Algorithmics assembler.
7273 The @option{-mgpopt} switch says to write all of the data declarations
7274 before the instructions in the text section, this allows the MIPS
7275 assembler to generate one word memory references instead of using two
7276 words for short global or static data items. This is on by default if
7277 optimization is selected.
7283 For each non-inline function processed, the @option{-mstats} switch
7284 causes the compiler to emit one line to the standard error file to
7285 print statistics about the program (number of registers saved, stack
7292 The @option{-mmemcpy} switch makes all block moves call the appropriate
7293 string function (@samp{memcpy} or @samp{bcopy}) instead of possibly
7294 generating inline code.
7297 @itemx -mno-mips-tfile
7298 @opindex mmips-tfile
7299 @opindex mno-mips-tfile
7300 The @option{-mno-mips-tfile} switch causes the compiler not
7301 postprocess the object file with the @file{mips-tfile} program,
7302 after the MIPS assembler has generated it to add debug support. If
7303 @file{mips-tfile} is not run, then no local variables will be
7304 available to the debugger. In addition, @file{stage2} and
7305 @file{stage3} objects will have the temporary file names passed to the
7306 assembler embedded in the object file, which means the objects will
7307 not compare the same. The @option{-mno-mips-tfile} switch should only
7308 be used when there are bugs in the @file{mips-tfile} program that
7309 prevents compilation.
7312 @opindex msoft-float
7313 Generate output containing library calls for floating point.
7314 @strong{Warning:} the requisite libraries are not part of GCC@.
7315 Normally the facilities of the machine's usual C compiler are used, but
7316 this can't be done directly in cross-compilation. You must make your
7317 own arrangements to provide suitable library functions for
7321 @opindex mhard-float
7322 Generate output containing floating point instructions. This is the
7323 default if you use the unmodified sources.
7326 @itemx -mno-abicalls
7328 @opindex mno-abicalls
7329 Emit (or do not emit) the pseudo operations @samp{.abicalls},
7330 @samp{.cpload}, and @samp{.cprestore} that some System V.4 ports use for
7331 position independent code.
7334 @itemx -mno-long-calls
7335 @opindex mlong-calls
7336 @opindex mno-long-calls
7337 Do all calls with the @samp{JALR} instruction, which requires
7338 loading up a function's address into a register before the call.
7339 You need to use this switch, if you call outside of the current
7340 512 megabyte segment to functions that are not through pointers.
7343 @itemx -mno-half-pic
7345 @opindex mno-half-pic
7346 Put pointers to extern references into the data section and load them
7347 up, rather than put the references in the text section.
7349 @item -membedded-pic
7350 @itemx -mno-embedded-pic
7351 @opindex membedded-pic
7352 @opindex mno-embedded-pic
7353 Generate PIC code suitable for some embedded systems. All calls are
7354 made using PC relative address, and all data is addressed using the $gp
7355 register. No more than 65536 bytes of global data may be used. This
7356 requires GNU as and GNU ld which do most of the work. This currently
7357 only works on targets which use ECOFF; it does not work with ELF@.
7359 @item -membedded-data
7360 @itemx -mno-embedded-data
7361 @opindex membedded-data
7362 @opindex mno-embedded-data
7363 Allocate variables to the read-only data section first if possible, then
7364 next in the small data section if possible, otherwise in data. This gives
7365 slightly slower code than the default, but reduces the amount of RAM required
7366 when executing, and thus may be preferred for some embedded systems.
7368 @item -muninit-const-in-rodata
7369 @itemx -mno-uninit-const-in-rodata
7370 @opindex muninit-const-in-rodata
7371 @opindex mno-uninit-const-in-rodata
7372 When used together with @option{-membedded-data}, it will always store uninitialized
7373 const variables in the read-only data section.
7375 @item -msingle-float
7376 @itemx -mdouble-float
7377 @opindex msingle-float
7378 @opindex mdouble-float
7379 The @option{-msingle-float} switch tells gcc to assume that the floating
7380 point coprocessor only supports single precision operations, as on the
7381 @samp{r4650} chip. The @option{-mdouble-float} switch permits gcc to use
7382 double precision operations. This is the default.
7388 Permit use of the @samp{mad}, @samp{madu} and @samp{mul} instructions,
7389 as on the @samp{r4650} chip.
7393 Turns on @option{-msingle-float}, @option{-mmad}, and, at least for now,
7394 @option{-mcpu=r4650}.
7400 Enable 16-bit instructions.
7404 Use the entry and exit pseudo ops. This option can only be used with
7409 Compile code for the processor in little endian mode.
7410 The requisite libraries are assumed to exist.
7414 Compile code for the processor in big endian mode.
7415 The requisite libraries are assumed to exist.
7419 @cindex smaller data references (MIPS)
7420 @cindex gp-relative references (MIPS)
7421 Put global and static items less than or equal to @var{num} bytes into
7422 the small data or bss sections instead of the normal data or bss
7423 section. This allows the assembler to emit one word memory reference
7424 instructions based on the global pointer (@var{gp} or @var{$28}),
7425 instead of the normal two words used. By default, @var{num} is 8 when
7426 the MIPS assembler is used, and 0 when the GNU assembler is used. The
7427 @option{-G @var{num}} switch is also passed to the assembler and linker.
7428 All modules should be compiled with the same @option{-G @var{num}}
7433 Tell the MIPS assembler to not run its preprocessor over user
7434 assembler files (with a @samp{.s} suffix) when assembling them.
7438 Pass an option to gas which will cause nops to be inserted if
7439 the read of the destination register of an mfhi or mflo instruction
7440 occurs in the following two instructions.
7444 Do not include the default crt0.
7446 @item -mflush-func=@var{func}
7447 @itemx -mno-flush-func
7448 @opindex mflush-func
7449 Specifies the function to call to flush the I and D caches, or to not
7450 call any such function. If called, the function must take the same
7451 arguments as the common @code{_flush_func()}, that is, the address of the
7452 memory range for which the cache is being flushed, the size of the
7453 memory range, and the number 3 (to flush both caches). The default
7454 depends on the target gcc was configured for, but commonly is either
7455 @samp{_flush_func} or @samp{__cpu_flush}.
7458 These options are defined by the macro
7459 @code{TARGET_SWITCHES} in the machine description. The default for the
7460 options is also defined by that macro, which enables you to change the
7463 @node i386 and x86-64 Options
7464 @subsection Intel 386 and AMD x86-64 Options
7465 @cindex i386 Options
7466 @cindex x86-64 Options
7467 @cindex Intel 386 Options
7468 @cindex AMD x86-64 Options
7470 These @samp{-m} options are defined for the i386 and x86-64 family of
7474 @item -mcpu=@var{cpu-type}
7476 Tune to @var{cpu-type} everything applicable about the generated code, except
7477 for the ABI and the set of available instructions. The choices for
7478 @var{cpu-type} are @samp{i386}, @samp{i486}, @samp{i586}, @samp{i686},
7479 @samp{pentium}, @samp{pentium-mmx}, @samp{pentiumpro}, @samp{pentium2},
7480 @samp{pentium3}, @samp{pentium4}, @samp{k6}, @samp{k6-2}, @samp{k6-3},
7481 @samp{athlon}, @samp{athlon-tbird}, @samp{athlon-4}, @samp{athlon-xp}
7482 and @samp{athlon-mp}.
7484 While picking a specific @var{cpu-type} will schedule things appropriately
7485 for that particular chip, the compiler will not generate any code that
7486 does not run on the i386 without the @option{-march=@var{cpu-type}} option
7487 being used. @samp{i586} is equivalent to @samp{pentium} and @samp{i686}
7488 is equivalent to @samp{pentiumpro}. @samp{k6} and @samp{athlon} are the
7489 AMD chips as opposed to the Intel ones.
7491 @item -march=@var{cpu-type}
7493 Generate instructions for the machine type @var{cpu-type}. The choices
7494 for @var{cpu-type} are the same as for @option{-mcpu}. Moreover,
7495 specifying @option{-march=@var{cpu-type}} implies @option{-mcpu=@var{cpu-type}}.
7504 @opindex mpentiumpro
7505 These options are synonyms for @option{-mcpu=i386}, @option{-mcpu=i486},
7506 @option{-mcpu=pentium}, and @option{-mcpu=pentiumpro} respectively.
7507 These synonyms are deprecated.
7509 @item -mfpmath=@var{unit}
7511 generate floating point arithmetics for selected unit @var{unit}. the choices
7516 Use the standard 387 floating point coprocessor present majority of chips and
7517 emulated otherwise. Code compiled with this option will run almost everywhere.
7518 The temporary results are computed in 80bit precesion instead of precision
7519 specified by the type resulting in slightly different results compared to most
7520 of other chips. See @option{-ffloat-store} for more detailed description.
7522 This is the default choice for i386 compiler.
7525 Use scalar floating point instructions present in the SSE instruction set.
7526 This instruction set is supported by Pentium3 and newer chips, in the AMD line
7527 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
7528 instruction set supports only single precision arithmetics, thus the double and
7529 extended precision arithmetics is still done using 387. Later version, present
7530 only in Pentium4 and the future AMD x86-64 chips supports double precision
7533 For i387 you need to use @option{-march=@var{cpu-type}}, @option{-msse} or
7534 @option{-msse2} switches to enable SSE extensions and make this option
7535 effective. For x86-64 compiler, these extensions are enabled by default.
7537 The resulting code should be considerably faster in majority of cases and avoid
7538 the numerical instability problems of 387 code, but may break some existing
7539 code that expects temporaries to be 80bit.
7541 This is the default choice for x86-64 compiler.
7544 Attempt to utilize both instruction sets at once. This effectivly double the
7545 amount of available registers and on chips with separate execution units for
7546 387 and SSE the execution resources too. Use this option with care, as it is
7547 still experimental, because gcc register allocator does not model separate
7548 functional units well resulting in instable performance.
7551 @item -masm=@var{dialect}
7552 @opindex masm=@var{dialect}
7553 Output asm instructions using selected @var{dialect}. Supported choices are
7554 @samp{intel} or @samp{att} (the default one).
7559 @opindex mno-ieee-fp
7560 Control whether or not the compiler uses IEEE floating point
7561 comparisons. These handle correctly the case where the result of a
7562 comparison is unordered.
7565 @opindex msoft-float
7566 Generate output containing library calls for floating point.
7567 @strong{Warning:} the requisite libraries are not part of GCC@.
7568 Normally the facilities of the machine's usual C compiler are used, but
7569 this can't be done directly in cross-compilation. You must make your
7570 own arrangements to provide suitable library functions for
7573 On machines where a function returns floating point results in the 80387
7574 register stack, some floating point opcodes may be emitted even if
7575 @option{-msoft-float} is used.
7577 @item -mno-fp-ret-in-387
7578 @opindex mno-fp-ret-in-387
7579 Do not use the FPU registers for return values of functions.
7581 The usual calling convention has functions return values of types
7582 @code{float} and @code{double} in an FPU register, even if there
7583 is no FPU@. The idea is that the operating system should emulate
7586 The option @option{-mno-fp-ret-in-387} causes such values to be returned
7587 in ordinary CPU registers instead.
7589 @item -mno-fancy-math-387
7590 @opindex mno-fancy-math-387
7591 Some 387 emulators do not support the @code{sin}, @code{cos} and
7592 @code{sqrt} instructions for the 387. Specify this option to avoid
7593 generating those instructions. This option is the default on FreeBSD@.
7594 As of revision 2.6.1, these instructions are not generated unless you
7595 also use the @option{-funsafe-math-optimizations} switch.
7597 @item -malign-double
7598 @itemx -mno-align-double
7599 @opindex malign-double
7600 @opindex mno-align-double
7601 Control whether GCC aligns @code{double}, @code{long double}, and
7602 @code{long long} variables on a two word boundary or a one word
7603 boundary. Aligning @code{double} variables on a two word boundary will
7604 produce code that runs somewhat faster on a @samp{Pentium} at the
7605 expense of more memory.
7607 @item -m128bit-long-double
7608 @opindex m128bit-long-double
7609 Control the size of @code{long double} type. i386 application binary interface
7610 specify the size to be 12 bytes, while modern architectures (Pentium and newer)
7611 prefer @code{long double} aligned to 8 or 16 byte boundary. This is
7612 impossible to reach with 12 byte long doubles in the array accesses.
7614 @strong{Warning:} if you use the @option{-m128bit-long-double} switch, the
7615 structures and arrays containing @code{long double} will change their size as
7616 well as function calling convention for function taking @code{long double}
7619 @item -m96bit-long-double
7620 @opindex m96bit-long-double
7621 Set the size of @code{long double} to 96 bits as required by the i386
7622 application binary interface. This is the default.
7625 @itemx -mno-svr3-shlib
7626 @opindex msvr3-shlib
7627 @opindex mno-svr3-shlib
7628 Control whether GCC places uninitialized local variables into the
7629 @code{bss} or @code{data} segments. @option{-msvr3-shlib} places them
7630 into @code{bss}. These options are meaningful only on System V Release 3.
7634 Use a different function-calling convention, in which functions that
7635 take a fixed number of arguments return with the @code{ret} @var{num}
7636 instruction, which pops their arguments while returning. This saves one
7637 instruction in the caller since there is no need to pop the arguments
7640 You can specify that an individual function is called with this calling
7641 sequence with the function attribute @samp{stdcall}. You can also
7642 override the @option{-mrtd} option by using the function attribute
7643 @samp{cdecl}. @xref{Function Attributes}.
7645 @strong{Warning:} this calling convention is incompatible with the one
7646 normally used on Unix, so you cannot use it if you need to call
7647 libraries compiled with the Unix compiler.
7649 Also, you must provide function prototypes for all functions that
7650 take variable numbers of arguments (including @code{printf});
7651 otherwise incorrect code will be generated for calls to those
7654 In addition, seriously incorrect code will result if you call a
7655 function with too many arguments. (Normally, extra arguments are
7656 harmlessly ignored.)
7658 @item -mregparm=@var{num}
7660 Control how many registers are used to pass integer arguments. By
7661 default, no registers are used to pass arguments, and at most 3
7662 registers can be used. You can control this behavior for a specific
7663 function by using the function attribute @samp{regparm}.
7664 @xref{Function Attributes}.
7666 @strong{Warning:} if you use this switch, and
7667 @var{num} is nonzero, then you must build all modules with the same
7668 value, including any libraries. This includes the system libraries and
7671 @item -mpreferred-stack-boundary=@var{num}
7672 @opindex mpreferred-stack-boundary
7673 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
7674 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
7675 the default is 4 (16 bytes or 128 bits), except when optimizing for code
7676 size (@option{-Os}), in which case the default is the minimum correct
7677 alignment (4 bytes for x86, and 8 bytes for x86-64).
7679 On Pentium and PentiumPro, @code{double} and @code{long double} values
7680 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
7681 suffer significant run time performance penalties. On Pentium III, the
7682 Streaming SIMD Extension (SSE) data type @code{__m128} suffers similar
7683 penalties if it is not 16 byte aligned.
7685 To ensure proper alignment of this values on the stack, the stack boundary
7686 must be as aligned as that required by any value stored on the stack.
7687 Further, every function must be generated such that it keeps the stack
7688 aligned. Thus calling a function compiled with a higher preferred
7689 stack boundary from a function compiled with a lower preferred stack
7690 boundary will most likely misalign the stack. It is recommended that
7691 libraries that use callbacks always use the default setting.
7693 This extra alignment does consume extra stack space, and generally
7694 increases code size. Code that is sensitive to stack space usage, such
7695 as embedded systems and operating system kernels, may want to reduce the
7696 preferred alignment to @option{-mpreferred-stack-boundary=2}.
7712 These switches enable or disable the use of built-in functions that allow
7713 direct access to the MMX, SSE and 3Dnow extensions of the instruction set.
7715 The following machine modes are available for use with MMX built-in functions
7716 (@pxref{Vector Extensions}): @code{V2SI} for a vector of two 32 bit integers,
7717 @code{V4HI} for a vector of four 16 bit integers, and @code{V8QI} for a
7718 vector of eight 8 bit integers. Some of the built-in functions operate on
7719 MMX registers as a whole 64 bit entity, these use @code{DI} as their mode.
7721 If 3Dnow extensions are enabled, @code{V2SF} is used as a mode for a vector
7722 of two 32 bit floating point values.
7724 If SSE extensions are enabled, @code{V4SF} is used for a vector of four 32 bit
7725 floating point values. Some instructions use a vector of four 32 bit
7726 integers, these use @code{V4SI}. Finally, some instructions operate on an
7727 entire vector register, interpreting it as a 128 bit integer, these use mode
7730 The following built-in functions are made available by @option{-mmmx}:
7732 @item v8qi __builtin_ia32_paddb (v8qi, v8qi)
7733 Generates the @code{paddb} machine instruction.
7734 @item v4hi __builtin_ia32_paddw (v4hi, v4hi)
7735 Generates the @code{paddw} machine instruction.
7736 @item v2si __builtin_ia32_paddd (v2si, v2si)
7737 Generates the @code{paddd} machine instruction.
7738 @item v8qi __builtin_ia32_psubb (v8qi, v8qi)
7739 Generates the @code{psubb} machine instruction.
7740 @item v4hi __builtin_ia32_psubw (v4hi, v4hi)
7741 Generates the @code{psubw} machine instruction.
7742 @item v2si __builtin_ia32_psubd (v2si, v2si)
7743 Generates the @code{psubd} machine instruction.
7745 @item v8qi __builtin_ia32_paddsb (v8qi, v8qi)
7746 Generates the @code{paddsb} machine instruction.
7747 @item v4hi __builtin_ia32_paddsw (v4hi, v4hi)
7748 Generates the @code{paddsw} machine instruction.
7749 @item v8qi __builtin_ia32_psubsb (v8qi, v8qi)
7750 Generates the @code{psubsb} machine instruction.
7751 @item v4hi __builtin_ia32_psubsw (v4hi, v4hi)
7752 Generates the @code{psubsw} machine instruction.
7754 @item v8qi __builtin_ia32_paddusb (v8qi, v8qi)
7755 Generates the @code{paddusb} machine instruction.
7756 @item v4hi __builtin_ia32_paddusw (v4hi, v4hi)
7757 Generates the @code{paddusw} machine instruction.
7758 @item v8qi __builtin_ia32_psubusb (v8qi, v8qi)
7759 Generates the @code{psubusb} machine instruction.
7760 @item v4hi __builtin_ia32_psubusw (v4hi, v4hi)
7761 Generates the @code{psubusw} machine instruction.
7763 @item v4hi __builtin_ia32_pmullw (v4hi, v4hi)
7764 Generates the @code{pmullw} machine instruction.
7765 @item v4hi __builtin_ia32_pmulhw (v4hi, v4hi)
7766 Generates the @code{pmulhw} machine instruction.
7768 @item di __builtin_ia32_pand (di, di)
7769 Generates the @code{pand} machine instruction.
7770 @item di __builtin_ia32_pandn (di,di)
7771 Generates the @code{pandn} machine instruction.
7772 @item di __builtin_ia32_por (di, di)
7773 Generates the @code{por} machine instruction.
7774 @item di __builtin_ia32_pxor (di, di)
7775 Generates the @code{pxor} machine instruction.
7777 @item v8qi __builtin_ia32_pcmpeqb (v8qi, v8qi)
7778 Generates the @code{pcmpeqb} machine instruction.
7779 @item v4hi __builtin_ia32_pcmpeqw (v4hi, v4hi)
7780 Generates the @code{pcmpeqw} machine instruction.
7781 @item v2si __builtin_ia32_pcmpeqd (v2si, v2si)
7782 Generates the @code{pcmpeqd} machine instruction.
7783 @item v8qi __builtin_ia32_pcmpgtb (v8qi, v8qi)
7784 Generates the @code{pcmpgtb} machine instruction.
7785 @item v4hi __builtin_ia32_pcmpgtw (v4hi, v4hi)
7786 Generates the @code{pcmpgtw} machine instruction.
7787 @item v2si __builtin_ia32_pcmpgtd (v2si, v2si)
7788 Generates the @code{pcmpgtd} machine instruction.
7790 @item v8qi __builtin_ia32_punpckhbw (v8qi, v8qi)
7791 Generates the @code{punpckhbw} machine instruction.
7792 @item v4hi __builtin_ia32_punpckhwd (v4hi, v4hi)
7793 Generates the @code{punpckhwd} machine instruction.
7794 @item v2si __builtin_ia32_punpckhdq (v2si, v2si)
7795 Generates the @code{punpckhdq} machine instruction.
7796 @item v8qi __builtin_ia32_punpcklbw (v8qi, v8qi)
7797 Generates the @code{punpcklbw} machine instruction.
7798 @item v4hi __builtin_ia32_punpcklwd (v4hi, v4hi)
7799 Generates the @code{punpcklwd} machine instruction.
7800 @item v2si __builtin_ia32_punpckldq (v2si, v2si)
7801 Generates the @code{punpckldq} machine instruction.
7803 @item v8qi __builtin_ia32_packsswb (v4hi, v4hi)
7804 Generates the @code{packsswb} machine instruction.
7805 @item v4hi __builtin_ia32_packssdw (v2si, v2si)
7806 Generates the @code{packssdw} machine instruction.
7807 @item v8qi __builtin_ia32_packuswb (v4hi, v4hi)
7808 Generates the @code{packuswb} machine instruction.
7812 The following built-in functions are made available either with @option{-msse}, or
7813 with a combination of @option{-m3dnow} and @option{-march=athlon}.
7816 @item v4hi __builtin_ia32_pmulhuw (v4hi, v4hi)
7817 Generates the @code{pmulhuw} machine instruction.
7819 @item v8qi __builtin_ia32_pavgb (v8qi, v8qi)
7820 Generates the @code{pavgb} machine instruction.
7821 @item v4hi __builtin_ia32_pavgw (v4hi, v4hi)
7822 Generates the @code{pavgw} machine instruction.
7823 @item v4hi __builtin_ia32_psadbw (v8qi, v8qi)
7824 Generates the @code{psadbw} machine instruction.
7826 @item v8qi __builtin_ia32_pmaxub (v8qi, v8qi)
7827 Generates the @code{pmaxub} machine instruction.
7828 @item v4hi __builtin_ia32_pmaxsw (v4hi, v4hi)
7829 Generates the @code{pmaxsw} machine instruction.
7830 @item v8qi __builtin_ia32_pminub (v8qi, v8qi)
7831 Generates the @code{pminub} machine instruction.
7832 @item v4hi __builtin_ia32_pminsw (v4hi, v4hi)
7833 Generates the @code{pminsw} machine instruction.
7835 @item int __builtin_ia32_pextrw (v4hi, int)
7836 Generates the @code{pextrw} machine instruction.
7837 @item v4hi __builtin_ia32_pinsrw (v4hi, int, int)
7838 Generates the @code{pinsrw} machine instruction.
7840 @item int __builtin_ia32_pmovmskb (v8qi)
7841 Generates the @code{pmovmskb} machine instruction.
7842 @item void __builtin_ia32_maskmovq (v8qi, v8qi, char *)
7843 Generates the @code{maskmovq} machine instruction.
7844 @item void __builtin_ia32_movntq (di *, di)
7845 Generates the @code{movntq} machine instruction.
7846 @item void __builtin_ia32_sfence (void)
7847 Generates the @code{sfence} machine instruction.
7848 @item void __builtin_ia32_prefetch (char *, int selector)
7849 Generates a prefetch machine instruction, depending on the value of
7850 selector. If @code{selector} is 0, it generates @code{prefetchnta}; for
7851 a value of 1, it generates @code{prefetcht0}; for a value of 2, it generates
7852 @code{prefetcht1}; and for a value of 3 it generates @code{prefetcht2}.
7856 The following built-in functions are available when @option{-msse} is used.
7859 @item int __builtin_ia32_comieq (v4sf, v4sf)
7860 Generates the @code{comiss} machine instruction and performs an equality
7861 comparison. The return value is the truth value of that comparison.
7862 @item int __builtin_ia32_comineq (v4sf, v4sf)
7863 Generates the @code{comiss} machine instruction and performs an inequality
7864 comparison. The return value is the truth value of that comparison.
7865 @item int __builtin_ia32_comilt (v4sf, v4sf)
7866 Generates the @code{comiss} machine instruction and performs a ``less than''
7867 comparison. The return value is the truth value of that comparison.
7868 @item int __builtin_ia32_comile (v4sf, v4sf)
7869 Generates the @code{comiss} machine instruction and performs a ``less or
7870 equal'' comparison. The return value is the truth value of that comparison.
7871 @item int __builtin_ia32_comigt (v4sf, v4sf)
7872 Generates the @code{comiss} machine instruction and performs a ``greater than''
7873 comparison. The return value is the truth value of that comparison.
7874 @item int __builtin_ia32_comige (v4sf, v4sf)
7875 Generates the @code{comiss} machine instruction and performs a ``greater or
7876 equal'' comparison. The return value is the truth value of that comparison.
7878 @item int __builtin_ia32_ucomieq (v4sf, v4sf)
7879 Generates the @code{ucomiss} machine instruction and performs an equality
7880 comparison. The return value is the truth value of that comparison.
7881 @item int __builtin_ia32_ucomineq (v4sf, v4sf)
7882 Generates the @code{ucomiss} machine instruction and performs an inequality
7883 comparison. The return value is the truth value of that comparison.
7884 @item int __builtin_ia32_ucomilt (v4sf, v4sf)
7885 Generates the @code{ucomiss} machine instruction and performs a ``less than''
7886 comparison. The return value is the truth value of that comparison.
7887 @item int __builtin_ia32_ucomile (v4sf, v4sf)
7888 Generates the @code{ucomiss} machine instruction and performs a ``less or
7889 equal'' comparison. The return value is the truth value of that comparison.
7890 @item int __builtin_ia32_ucomigt (v4sf, v4sf)
7891 Generates the @code{ucomiss} machine instruction and performs a ``greater than''
7892 comparison. The return value is the truth value of that comparison.
7893 @item int __builtin_ia32_ucomige (v4sf, v4sf)
7894 Generates the @code{ucomiss} machine instruction and performs a ``greater or
7895 equal'' comparison. The return value is the truth value of that comparison.
7897 @item v4sf __builtin_ia32_addps (v4sf, v4sf)
7898 Generates the @code{addps} machine instruction.
7899 @item v4sf __builtin_ia32_addss (v4sf, v4sf)
7900 Generates the @code{addss} machine instruction.
7901 @item v4sf __builtin_ia32_subps (v4sf, v4sf)
7902 Generates the @code{subps} machine instruction.
7903 @item v4sf __builtin_ia32_subss (v4sf, v4sf)
7904 Generates the @code{subss} machine instruction.
7905 @item v4sf __builtin_ia32_mulps (v4sf, v4sf)
7906 Generates the @code{mulps} machine instruction.
7907 @item v4sf __builtin_ia32_mulss (v4sf, v4sf)
7908 Generates the @code{mulss} machine instruction.
7909 @item v4sf __builtin_ia32_divps (v4sf, v4sf)
7910 Generates the @code{divps} machine instruction.
7911 @item v4sf __builtin_ia32_divss (v4sf, v4sf)
7912 Generates the @code{divss} machine instruction.
7914 @item v4si __builtin_ia32_cmpeqps (v4sf, v4sf)
7915 Generates the @code{cmpeqps} machine instruction.
7916 @item v4si __builtin_ia32_cmplts (v4sf, v4sf)
7917 Generates the @code{cmpltps} machine instruction.
7918 @item v4si __builtin_ia32_cmpleps (v4sf, v4sf)
7919 Generates the @code{cmpleps} machine instruction.
7920 @item v4si __builtin_ia32_cmpgtps (v4sf, v4sf)
7921 Generates the @code{cmpgtps} machine instruction.
7922 @item v4si __builtin_ia32_cmpgeps (v4sf, v4sf)
7923 Generates the @code{cmpgeps} machine instruction.
7924 @item v4si __builtin_ia32_cmpunordps (v4sf, v4sf)
7925 Generates the @code{cmpunodps} machine instruction.
7926 @item v4si __builtin_ia32_cmpneqps (v4sf, v4sf)
7927 Generates the @code{cmpeqps} machine instruction.
7928 @item v4si __builtin_ia32_cmpnltps (v4sf, v4sf)
7929 Generates the @code{cmpltps} machine instruction.
7930 @item v4si __builtin_ia32_cmpnleps (v4sf, v4sf)
7931 Generates the @code{cmpleps} machine instruction.
7932 @item v4si __builtin_ia32_cmpngtps (v4sf, v4sf)
7933 Generates the @code{cmpgtps} machine instruction.
7934 @item v4si __builtin_ia32_cmpngeps (v4sf, v4sf)
7935 Generates the @code{cmpgeps} machine instruction.
7936 @item v4si __builtin_ia32_cmpordps (v4sf, v4sf)
7937 Generates the @code{cmpunodps} machine instruction.
7939 @item v4si __builtin_ia32_cmpeqss (v4sf, v4sf)
7940 Generates the @code{cmpeqss} machine instruction.
7941 @item v4si __builtin_ia32_cmpltss (v4sf, v4sf)
7942 Generates the @code{cmpltss} machine instruction.
7943 @item v4si __builtin_ia32_cmpless (v4sf, v4sf)
7944 Generates the @code{cmpless} machine instruction.
7945 @item v4si __builtin_ia32_cmpgtss (v4sf, v4sf)
7946 Generates the @code{cmpgtss} machine instruction.
7947 @item v4si __builtin_ia32_cmpgess (v4sf, v4sf)
7948 Generates the @code{cmpgess} machine instruction.
7949 @item v4si __builtin_ia32_cmpunordss (v4sf, v4sf)
7950 Generates the @code{cmpunodss} machine instruction.
7951 @item v4si __builtin_ia32_cmpneqss (v4sf, v4sf)
7952 Generates the @code{cmpeqss} machine instruction.
7953 @item v4si __builtin_ia32_cmpnlts (v4sf, v4sf)
7954 Generates the @code{cmpltss} machine instruction.
7955 @item v4si __builtin_ia32_cmpnless (v4sf, v4sf)
7956 Generates the @code{cmpless} machine instruction.
7957 @item v4si __builtin_ia32_cmpngtss (v4sf, v4sf)
7958 Generates the @code{cmpgtss} machine instruction.
7959 @item v4si __builtin_ia32_cmpngess (v4sf, v4sf)
7960 Generates the @code{cmpgess} machine instruction.
7961 @item v4si __builtin_ia32_cmpordss (v4sf, v4sf)
7962 Generates the @code{cmpunodss} machine instruction.
7964 @item v4sf __builtin_ia32_maxps (v4sf, v4sf)
7965 Generates the @code{maxps} machine instruction.
7966 @item v4sf __builtin_ia32_maxsss (v4sf, v4sf)
7967 Generates the @code{maxss} machine instruction.
7968 @item v4sf __builtin_ia32_minps (v4sf, v4sf)
7969 Generates the @code{minps} machine instruction.
7970 @item v4sf __builtin_ia32_minsss (v4sf, v4sf)
7971 Generates the @code{minss} machine instruction.
7973 @item ti __builtin_ia32_andps (ti, ti)
7974 Generates the @code{andps} machine instruction.
7975 @item ti __builtin_ia32_andnps (ti, ti)
7976 Generates the @code{andnps} machine instruction.
7977 @item ti __builtin_ia32_orps (ti, ti)
7978 Generates the @code{orps} machine instruction.
7979 @item ti __builtin_ia32_xorps (ti, ti)
7980 Generates the @code{xorps} machine instruction.
7982 @item v4sf __builtin_ia32_movps (v4sf, v4sf)
7983 Generates the @code{movps} machine instruction.
7984 @item v4sf __builtin_ia32_movhlps (v4sf, v4sf)
7985 Generates the @code{movhlps} machine instruction.
7986 @item v4sf __builtin_ia32_movlhps (v4sf, v4sf)
7987 Generates the @code{movlhps} machine instruction.
7988 @item v4sf __builtin_ia32_unpckhps (v4sf, v4sf)
7989 Generates the @code{unpckhps} machine instruction.
7990 @item v4sf __builtin_ia32_unpcklps (v4sf, v4sf)
7991 Generates the @code{unpcklps} machine instruction.
7993 @item v4sf __builtin_ia32_cvtpi2ps (v4sf, v2si)
7994 Generates the @code{cvtpi2ps} machine instruction.
7995 @item v2si __builtin_ia32_cvtps2pi (v4sf)
7996 Generates the @code{cvtps2pi} machine instruction.
7997 @item v4sf __builtin_ia32_cvtsi2ss (v4sf, int)
7998 Generates the @code{cvtsi2ss} machine instruction.
7999 @item int __builtin_ia32_cvtss2si (v4sf)
8000 Generates the @code{cvtsi2ss} machine instruction.
8001 @item v2si __builtin_ia32_cvttps2pi (v4sf)
8002 Generates the @code{cvttps2pi} machine instruction.
8003 @item int __builtin_ia32_cvttss2si (v4sf)
8004 Generates the @code{cvttsi2ss} machine instruction.
8006 @item v4sf __builtin_ia32_rcpps (v4sf)
8007 Generates the @code{rcpps} machine instruction.
8008 @item v4sf __builtin_ia32_rsqrtps (v4sf)
8009 Generates the @code{rsqrtps} machine instruction.
8010 @item v4sf __builtin_ia32_sqrtps (v4sf)
8011 Generates the @code{sqrtps} machine instruction.
8012 @item v4sf __builtin_ia32_rcpss (v4sf)
8013 Generates the @code{rcpss} machine instruction.
8014 @item v4sf __builtin_ia32_rsqrtss (v4sf)
8015 Generates the @code{rsqrtss} machine instruction.
8016 @item v4sf __builtin_ia32_sqrtss (v4sf)
8017 Generates the @code{sqrtss} machine instruction.
8019 @item v4sf __builtin_ia32_shufps (v4sf, v4sf, int)
8020 Generates the @code{shufps} machine instruction.
8022 @item v4sf __builtin_ia32_loadaps (float *)
8023 Generates the @code{movaps} machine instruction as a load from memory.
8024 @item void __builtin_ia32_storeaps (float *, v4sf)
8025 Generates the @code{movaps} machine instruction as a store to memory.
8026 @item v4sf __builtin_ia32_loadups (float *)
8027 Generates the @code{movups} machine instruction as a load from memory.
8028 @item void __builtin_ia32_storeups (float *, v4sf)
8029 Generates the @code{movups} machine instruction as a store to memory.
8030 @item v4sf __builtin_ia32_loadsss (float *)
8031 Generates the @code{movss} machine instruction as a load from memory.
8032 @item void __builtin_ia32_storess (float *, v4sf)
8033 Generates the @code{movss} machine instruction as a store to memory.
8035 @item v4sf __builtin_ia32_loadhps (v4sf, v2si *)
8036 Generates the @code{movhps} machine instruction as a load from memory.
8037 @item v4sf __builtin_ia32_loadlps (v4sf, v2si *)
8038 Generates the @code{movlps} machine instruction as a load from memory
8039 @item void __builtin_ia32_storehps (v4sf, v2si *)
8040 Generates the @code{movhps} machine instruction as a store to memory.
8041 @item void __builtin_ia32_storelps (v4sf, v2si *)
8042 Generates the @code{movlps} machine instruction as a store to memory.
8044 @item void __builtin_ia32_movntps (float *, v4sf)
8045 Generates the @code{movntps} machine instruction.
8046 @item int __builtin_ia32_movmskps (v4sf)
8047 Generates the @code{movntps} machine instruction.
8049 @item void __builtin_ia32_storeps1 (float *, v4sf)
8050 Generates the @code{movaps} machine instruction as a store to memory.
8051 Before storing, the value is modified with a @code{shufps} instruction
8052 so that the lowest of the four floating point elements is replicated
8053 across the entire vector that is stored.
8054 @item void __builtin_ia32_storerps (float *, v4sf)
8055 Generates the @code{movaps} machine instruction as a store to memory.
8056 Before storing, the value is modified with a @code{shufps} instruction
8057 so that the order of the four floating point elements in the vector is
8059 @item v4sf __builtin_ia32_loadps1 (float *)
8060 Generates a @code{movss} machine instruction to load a floating point
8061 value from memory, and a @code{shufps} instruction to replicate the
8062 loaded value across all four elements of the result vector.
8063 @item v4sf __builtin_ia32_loadrps (float *)
8064 Generates a @code{movaps} machine instruction to load a vector from
8065 memory, and a @code{shufps} instruction to reverse the order of the
8066 four floating point elements in the result vector.
8067 @item v4sf __builtin_ia32_setps (float, float, float, float)
8068 Constructs a vector from four single floating point values. The return
8069 value is equal to the value that would result from storing the four
8070 arguments into consecutive memory locations and then executing a
8071 @code{movaps} to load the vector from memory.
8072 @item v4sf __builtin_ia32_setps1 (float)
8073 Constructs a vector from a single floating point value by replicating
8074 it across all four elements of the result vector.
8078 @itemx -mno-push-args
8080 @opindex mno-push-args
8081 Use PUSH operations to store outgoing parameters. This method is shorter
8082 and usually equally fast as method using SUB/MOV operations and is enabled
8083 by default. In some cases disabling it may improve performance because of
8084 improved scheduling and reduced dependencies.
8086 @item -maccumulate-outgoing-args
8087 @opindex maccumulate-outgoing-args
8088 If enabled, the maximum amount of space required for outgoing arguments will be
8089 computed in the function prologue. This is faster on most modern CPUs
8090 because of reduced dependencies, improved scheduling and reduced stack usage
8091 when preferred stack boundary is not equal to 2. The drawback is a notable
8092 increase in code size. This switch implies @option{-mno-push-args}.
8096 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
8097 on thread-safe exception handling must compile and link all code with the
8098 @option{-mthreads} option. When compiling, @option{-mthreads} defines
8099 @option{-D_MT}; when linking, it links in a special thread helper library
8100 @option{-lmingwthrd} which cleans up per thread exception handling data.
8102 @item -mno-align-stringops
8103 @opindex mno-align-stringops
8104 Do not align destination of inlined string operations. This switch reduces
8105 code size and improves performance in case the destination is already aligned,
8106 but gcc don't know about it.
8108 @item -minline-all-stringops
8109 @opindex minline-all-stringops
8110 By default GCC inlines string operations only when destination is known to be
8111 aligned at least to 4 byte boundary. This enables more inlining, increase code
8112 size, but may improve performance of code that depends on fast memcpy, strlen
8113 and memset for short lengths.
8115 @item -momit-leaf-frame-pointer
8116 @opindex momit-leaf-frame-pointer
8117 Don't keep the frame pointer in a register for leaf functions. This
8118 avoids the instructions to save, set up and restore frame pointers and
8119 makes an extra register available in leaf functions. The option
8120 @option{-fomit-frame-pointer} removes the frame pointer for all functions
8121 which might make debugging harder.
8124 These @samp{-m} switches are supported in addition to the above
8125 on AMD x86-64 processors in 64-bit environments.
8132 Generate code for a 32-bit or 64-bit environment.
8133 The 32-bit environment sets int, long and pointer to 32 bits and
8134 generates code that runs on any i386 system.
8135 The 64-bit environment sets int to 32 bits and long and pointer
8136 to 64 bits and generates code for AMD's x86-64 architecture.
8139 @opindex no-red-zone
8140 Do not use a so called red zone for x86-64 code. The red zone is mandated
8141 by the x86-64 ABI, it is a 128-byte area beyond the location of the
8142 stack pointer that will not be modified by signal or interrupt handlers
8143 and therefore can be used for temporary data without adjusting the stack
8144 pointer. The flag @option{-mno-red-zone} disables this red zone.
8148 @subsection HPPA Options
8149 @cindex HPPA Options
8151 These @samp{-m} options are defined for the HPPA family of computers:
8154 @item -march=@var{architecture-type}
8156 Generate code for the specified architecture. The choices for
8157 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
8158 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
8159 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
8160 architecture option for your machine. Code compiled for lower numbered
8161 architectures will run on higher numbered architectures, but not the
8164 PA 2.0 support currently requires gas snapshot 19990413 or later. The
8165 next release of binutils (current is 2.9.1) will probably contain PA 2.0
8169 @itemx -mpa-risc-1-1
8170 @itemx -mpa-risc-2-0
8171 @opindex mpa-risc-1-0
8172 @opindex mpa-risc-1-1
8173 @opindex mpa-risc-2-0
8174 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
8177 @opindex mbig-switch
8178 Generate code suitable for big switch tables. Use this option only if
8179 the assembler/linker complain about out of range branches within a switch
8182 @item -mjump-in-delay
8183 @opindex mjump-in-delay
8184 Fill delay slots of function calls with unconditional jump instructions
8185 by modifying the return pointer for the function call to be the target
8186 of the conditional jump.
8188 @item -mdisable-fpregs
8189 @opindex mdisable-fpregs
8190 Prevent floating point registers from being used in any manner. This is
8191 necessary for compiling kernels which perform lazy context switching of
8192 floating point registers. If you use this option and attempt to perform
8193 floating point operations, the compiler will abort.
8195 @item -mdisable-indexing
8196 @opindex mdisable-indexing
8197 Prevent the compiler from using indexing address modes. This avoids some
8198 rather obscure problems when compiling MIG generated code under MACH@.
8200 @item -mno-space-regs
8201 @opindex mno-space-regs
8202 Generate code that assumes the target has no space registers. This allows
8203 GCC to generate faster indirect calls and use unscaled index address modes.
8205 Such code is suitable for level 0 PA systems and kernels.
8207 @item -mfast-indirect-calls
8208 @opindex mfast-indirect-calls
8209 Generate code that assumes calls never cross space boundaries. This
8210 allows GCC to emit code which performs faster indirect calls.
8212 This option will not work in the presence of shared libraries or nested
8215 @item -mlong-load-store
8216 @opindex mlong-load-store
8217 Generate 3-instruction load and store sequences as sometimes required by
8218 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
8221 @item -mportable-runtime
8222 @opindex mportable-runtime
8223 Use the portable calling conventions proposed by HP for ELF systems.
8227 Enable the use of assembler directives only GAS understands.
8229 @item -mschedule=@var{cpu-type}
8231 Schedule code according to the constraints for the machine type
8232 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
8233 @samp{7100}, @samp{7100LC}, @samp{7200}, and @samp{8000}. Refer to
8234 @file{/usr/lib/sched.models} on an HP-UX system to determine the
8235 proper scheduling option for your machine.
8238 @opindex mlinker-opt
8239 Enable the optimization pass in the HPUX linker. Note this makes symbolic
8240 debugging impossible. It also triggers a bug in the HPUX 8 and HPUX 9 linkers
8241 in which they give bogus error messages when linking some programs.
8244 @opindex msoft-float
8245 Generate output containing library calls for floating point.
8246 @strong{Warning:} the requisite libraries are not available for all HPPA
8247 targets. Normally the facilities of the machine's usual C compiler are
8248 used, but this cannot be done directly in cross-compilation. You must make
8249 your own arrangements to provide suitable library functions for
8250 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
8251 does provide software floating point support.
8253 @option{-msoft-float} changes the calling convention in the output file;
8254 therefore, it is only useful if you compile @emph{all} of a program with
8255 this option. In particular, you need to compile @file{libgcc.a}, the
8256 library that comes with GCC, with @option{-msoft-float} in order for
8260 @node Intel 960 Options
8261 @subsection Intel 960 Options
8263 These @samp{-m} options are defined for the Intel 960 implementations:
8266 @item -m@var{cpu-type}
8274 Assume the defaults for the machine type @var{cpu-type} for some of
8275 the other options, including instruction scheduling, floating point
8276 support, and addressing modes. The choices for @var{cpu-type} are
8277 @samp{ka}, @samp{kb}, @samp{mc}, @samp{ca}, @samp{cf},
8278 @samp{sa}, and @samp{sb}.
8285 @opindex msoft-float
8286 The @option{-mnumerics} option indicates that the processor does support
8287 floating-point instructions. The @option{-msoft-float} option indicates
8288 that floating-point support should not be assumed.
8290 @item -mleaf-procedures
8291 @itemx -mno-leaf-procedures
8292 @opindex mleaf-procedures
8293 @opindex mno-leaf-procedures
8294 Do (or do not) attempt to alter leaf procedures to be callable with the
8295 @code{bal} instruction as well as @code{call}. This will result in more
8296 efficient code for explicit calls when the @code{bal} instruction can be
8297 substituted by the assembler or linker, but less efficient code in other
8298 cases, such as calls via function pointers, or using a linker that doesn't
8299 support this optimization.
8302 @itemx -mno-tail-call
8304 @opindex mno-tail-call
8305 Do (or do not) make additional attempts (beyond those of the
8306 machine-independent portions of the compiler) to optimize tail-recursive
8307 calls into branches. You may not want to do this because the detection of
8308 cases where this is not valid is not totally complete. The default is
8309 @option{-mno-tail-call}.
8311 @item -mcomplex-addr
8312 @itemx -mno-complex-addr
8313 @opindex mcomplex-addr
8314 @opindex mno-complex-addr
8315 Assume (or do not assume) that the use of a complex addressing mode is a
8316 win on this implementation of the i960. Complex addressing modes may not
8317 be worthwhile on the K-series, but they definitely are on the C-series.
8318 The default is currently @option{-mcomplex-addr} for all processors except
8322 @itemx -mno-code-align
8323 @opindex mcode-align
8324 @opindex mno-code-align
8325 Align code to 8-byte boundaries for faster fetching (or don't bother).
8326 Currently turned on by default for C-series implementations only.
8329 @item -mclean-linkage
8330 @itemx -mno-clean-linkage
8331 @opindex mclean-linkage
8332 @opindex mno-clean-linkage
8333 These options are not fully implemented.
8337 @itemx -mic2.0-compat
8338 @itemx -mic3.0-compat
8340 @opindex mic2.0-compat
8341 @opindex mic3.0-compat
8342 Enable compatibility with iC960 v2.0 or v3.0.
8346 @opindex masm-compat
8348 Enable compatibility with the iC960 assembler.
8350 @item -mstrict-align
8351 @itemx -mno-strict-align
8352 @opindex mstrict-align
8353 @opindex mno-strict-align
8354 Do not permit (do permit) unaligned accesses.
8358 Enable structure-alignment compatibility with Intel's gcc release version
8359 1.3 (based on gcc 1.37). This option implies @option{-mstrict-align}.
8361 @item -mlong-double-64
8362 @opindex mlong-double-64
8363 Implement type @samp{long double} as 64-bit floating point numbers.
8364 Without the option @samp{long double} is implemented by 80-bit
8365 floating point numbers. The only reason we have it because there is
8366 no 128-bit @samp{long double} support in @samp{fp-bit.c} yet. So it
8367 is only useful for people using soft-float targets. Otherwise, we
8368 should recommend against use of it.
8372 @node DEC Alpha Options
8373 @subsection DEC Alpha Options
8375 These @samp{-m} options are defined for the DEC Alpha implementations:
8378 @item -mno-soft-float
8380 @opindex mno-soft-float
8381 @opindex msoft-float
8382 Use (do not use) the hardware floating-point instructions for
8383 floating-point operations. When @option{-msoft-float} is specified,
8384 functions in @file{libgcc.a} will be used to perform floating-point
8385 operations. Unless they are replaced by routines that emulate the
8386 floating-point operations, or compiled in such a way as to call such
8387 emulations routines, these routines will issue floating-point
8388 operations. If you are compiling for an Alpha without floating-point
8389 operations, you must ensure that the library is built so as not to call
8392 Note that Alpha implementations without floating-point operations are
8393 required to have floating-point registers.
8398 @opindex mno-fp-regs
8399 Generate code that uses (does not use) the floating-point register set.
8400 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
8401 register set is not used, floating point operands are passed in integer
8402 registers as if they were integers and floating-point results are passed
8403 in $0 instead of $f0. This is a non-standard calling sequence, so any
8404 function with a floating-point argument or return value called by code
8405 compiled with @option{-mno-fp-regs} must also be compiled with that
8408 A typical use of this option is building a kernel that does not use,
8409 and hence need not save and restore, any floating-point registers.
8413 The Alpha architecture implements floating-point hardware optimized for
8414 maximum performance. It is mostly compliant with the IEEE floating
8415 point standard. However, for full compliance, software assistance is
8416 required. This option generates code fully IEEE compliant code
8417 @emph{except} that the @var{inexact-flag} is not maintained (see below).
8418 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
8419 defined during compilation. The resulting code is less efficient but is
8420 able to correctly support denormalized numbers and exceptional IEEE
8421 values such as not-a-number and plus/minus infinity. Other Alpha
8422 compilers call this option @option{-ieee_with_no_inexact}.
8424 @item -mieee-with-inexact
8425 @opindex mieee-with-inexact
8426 This is like @option{-mieee} except the generated code also maintains
8427 the IEEE @var{inexact-flag}. Turning on this option causes the
8428 generated code to implement fully-compliant IEEE math. In addition to
8429 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
8430 macro. On some Alpha implementations the resulting code may execute
8431 significantly slower than the code generated by default. Since there is
8432 very little code that depends on the @var{inexact-flag}, you should
8433 normally not specify this option. Other Alpha compilers call this
8434 option @option{-ieee_with_inexact}.
8436 @item -mfp-trap-mode=@var{trap-mode}
8437 @opindex mfp-trap-mode
8438 This option controls what floating-point related traps are enabled.
8439 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
8440 The trap mode can be set to one of four values:
8444 This is the default (normal) setting. The only traps that are enabled
8445 are the ones that cannot be disabled in software (e.g., division by zero
8449 In addition to the traps enabled by @samp{n}, underflow traps are enabled
8453 Like @samp{su}, but the instructions are marked to be safe for software
8454 completion (see Alpha architecture manual for details).
8457 Like @samp{su}, but inexact traps are enabled as well.
8460 @item -mfp-rounding-mode=@var{rounding-mode}
8461 @opindex mfp-rounding-mode
8462 Selects the IEEE rounding mode. Other Alpha compilers call this option
8463 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
8468 Normal IEEE rounding mode. Floating point numbers are rounded towards
8469 the nearest machine number or towards the even machine number in case
8473 Round towards minus infinity.
8476 Chopped rounding mode. Floating point numbers are rounded towards zero.
8479 Dynamic rounding mode. A field in the floating point control register
8480 (@var{fpcr}, see Alpha architecture reference manual) controls the
8481 rounding mode in effect. The C library initializes this register for
8482 rounding towards plus infinity. Thus, unless your program modifies the
8483 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
8486 @item -mtrap-precision=@var{trap-precision}
8487 @opindex mtrap-precision
8488 In the Alpha architecture, floating point traps are imprecise. This
8489 means without software assistance it is impossible to recover from a
8490 floating trap and program execution normally needs to be terminated.
8491 GCC can generate code that can assist operating system trap handlers
8492 in determining the exact location that caused a floating point trap.
8493 Depending on the requirements of an application, different levels of
8494 precisions can be selected:
8498 Program precision. This option is the default and means a trap handler
8499 can only identify which program caused a floating point exception.
8502 Function precision. The trap handler can determine the function that
8503 caused a floating point exception.
8506 Instruction precision. The trap handler can determine the exact
8507 instruction that caused a floating point exception.
8510 Other Alpha compilers provide the equivalent options called
8511 @option{-scope_safe} and @option{-resumption_safe}.
8513 @item -mieee-conformant
8514 @opindex mieee-conformant
8515 This option marks the generated code as IEEE conformant. You must not
8516 use this option unless you also specify @option{-mtrap-precision=i} and either
8517 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
8518 is to emit the line @samp{.eflag 48} in the function prologue of the
8519 generated assembly file. Under DEC Unix, this has the effect that
8520 IEEE-conformant math library routines will be linked in.
8522 @item -mbuild-constants
8523 @opindex mbuild-constants
8524 Normally GCC examines a 32- or 64-bit integer constant to
8525 see if it can construct it from smaller constants in two or three
8526 instructions. If it cannot, it will output the constant as a literal and
8527 generate code to load it from the data segment at runtime.
8529 Use this option to require GCC to construct @emph{all} integer constants
8530 using code, even if it takes more instructions (the maximum is six).
8532 You would typically use this option to build a shared library dynamic
8533 loader. Itself a shared library, it must relocate itself in memory
8534 before it can find the variables and constants in its own data segment.
8540 Select whether to generate code to be assembled by the vendor-supplied
8541 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
8555 Indicate whether GCC should generate code to use the optional BWX,
8556 CIX, and MAX instruction sets. The default is to use the instruction sets
8557 supported by the CPU type specified via @option{-mcpu=} option or that
8558 of the CPU on which GCC was built if none was specified.
8560 @item -mcpu=@var{cpu_type}
8562 Set the instruction set, register set, and instruction scheduling
8563 parameters for machine type @var{cpu_type}. You can specify either the
8564 @samp{EV} style name or the corresponding chip number. GCC
8565 supports scheduling parameters for the EV4 and EV5 family of processors
8566 and will choose the default values for the instruction set from
8567 the processor you specify. If you do not specify a processor type,
8568 GCC will default to the processor on which the compiler was built.
8570 Supported values for @var{cpu_type} are
8575 Schedules as an EV4 and has no instruction set extensions.
8579 Schedules as an EV5 and has no instruction set extensions.
8583 Schedules as an EV5 and supports the BWX extension.
8588 Schedules as an EV5 and supports the BWX and MAX extensions.
8592 Schedules as an EV5 (until Digital releases the scheduling parameters
8593 for the EV6) and supports the BWX, CIX, and MAX extensions.
8596 @item -mmemory-latency=@var{time}
8597 @opindex mmemory-latency
8598 Sets the latency the scheduler should assume for typical memory
8599 references as seen by the application. This number is highly
8600 dependent on the memory access patterns used by the application
8601 and the size of the external cache on the machine.
8603 Valid options for @var{time} are
8607 A decimal number representing clock cycles.
8613 The compiler contains estimates of the number of clock cycles for
8614 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
8615 (also called Dcache, Scache, and Bcache), as well as to main memory.
8616 Note that L3 is only valid for EV5.
8621 @node DEC Alpha/VMS Options
8622 @subsection DEC Alpha/VMS Options
8624 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
8627 @item -mvms-return-codes
8628 @opindex mvms-return-codes
8629 Return VMS condition codes from main. The default is to return POSIX
8630 style condition (e.g.@ error) codes.
8633 @node Clipper Options
8634 @subsection Clipper Options
8636 These @samp{-m} options are defined for the Clipper implementations:
8641 Produce code for a C300 Clipper processor. This is the default.
8645 Produce code for a C400 Clipper processor, i.e.@: use floating point
8649 @node H8/300 Options
8650 @subsection H8/300 Options
8652 These @samp{-m} options are defined for the H8/300 implementations:
8657 Shorten some address references at link time, when possible; uses the
8658 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
8659 ld.info, Using ld}, for a fuller description.
8663 Generate code for the H8/300H@.
8667 Generate code for the H8/S@.
8671 Generate code for the H8/S2600. This switch must be used with @option{-ms}.
8675 Make @code{int} data 32 bits by default.
8679 On the H8/300H and H8/S, use the same alignment rules as for the H8/300.
8680 The default for the H8/300H and H8/S is to align longs and floats on 4
8682 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
8683 This option has no effect on the H8/300.
8687 @subsection SH Options
8689 These @samp{-m} options are defined for the SH implementations:
8694 Generate code for the SH1.
8698 Generate code for the SH2.
8702 Generate code for the SH3.
8706 Generate code for the SH3e.
8710 Generate code for the SH4 without a floating-point unit.
8712 @item -m4-single-only
8713 @opindex m4-single-only
8714 Generate code for the SH4 with a floating-point unit that only
8715 supports single-precision arithmetic.
8719 Generate code for the SH4 assuming the floating-point unit is in
8720 single-precision mode by default.
8724 Generate code for the SH4.
8728 Compile code for the processor in big endian mode.
8732 Compile code for the processor in little endian mode.
8736 Align doubles at 64-bit boundaries. Note that this changes the calling
8737 conventions, and thus some functions from the standard C library will
8738 not work unless you recompile it first with @option{-mdalign}.
8742 Shorten some address references at link time, when possible; uses the
8743 linker option @option{-relax}.
8747 Use 32-bit offsets in @code{switch} tables. The default is to use
8752 Enable the use of the instruction @code{fmovd}.
8756 Comply with the calling conventions defined by Hitachi.
8760 Mark the @code{MAC} register as call-clobbered, even if
8761 @option{-mhitachi} is given.
8765 Increase IEEE-compliance of floating-point code.
8769 Dump instruction size and location in the assembly code.
8773 This option is deprecated. It pads structures to multiple of 4 bytes,
8774 which is incompatible with the SH ABI@.
8778 Optimize for space instead of speed. Implied by @option{-Os}.
8782 When generating position-independent code, emit function calls using
8783 the Global Offset Table instead of the Procedure Linkage Table.
8787 Generate a library function call to invalidate instruction cache
8788 entries, after fixing up a trampoline. This library function call
8789 doesn't assume it can write to the whole memory address space. This
8790 is the default when the target is @code{sh-*-linux*}.
8793 @node System V Options
8794 @subsection Options for System V
8796 These additional options are available on System V Release 4 for
8797 compatibility with other compilers on those systems:
8802 Create a shared object.
8803 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
8807 Identify the versions of each tool used by the compiler, in a
8808 @code{.ident} assembler directive in the output.
8812 Refrain from adding @code{.ident} directives to the output file (this is
8815 @item -YP,@var{dirs}
8817 Search the directories @var{dirs}, and no others, for libraries
8818 specified with @option{-l}.
8822 Look in the directory @var{dir} to find the M4 preprocessor.
8823 The assembler uses this option.
8824 @c This is supposed to go with a -Yd for predefined M4 macro files, but
8825 @c the generic assembler that comes with Solaris takes just -Ym.
8828 @node TMS320C3x/C4x Options
8829 @subsection TMS320C3x/C4x Options
8830 @cindex TMS320C3x/C4x Options
8832 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
8836 @item -mcpu=@var{cpu_type}
8838 Set the instruction set, register set, and instruction scheduling
8839 parameters for machine type @var{cpu_type}. Supported values for
8840 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
8841 @samp{c44}. The default is @samp{c40} to generate code for the
8846 @itemx -msmall-memory
8848 @opindex mbig-memory
8850 @opindex msmall-memory
8852 Generates code for the big or small memory model. The small memory
8853 model assumed that all data fits into one 64K word page. At run-time
8854 the data page (DP) register must be set to point to the 64K page
8855 containing the .bss and .data program sections. The big memory model is
8856 the default and requires reloading of the DP register for every direct
8863 Allow (disallow) allocation of general integer operands into the block
8870 Enable (disable) generation of code using decrement and branch,
8871 DBcond(D), instructions. This is enabled by default for the C4x. To be
8872 on the safe side, this is disabled for the C3x, since the maximum
8873 iteration count on the C3x is @math{2^23 + 1} (but who iterates loops more than
8874 @math{2^23} times on the C3x?). Note that GCC will try to reverse a loop so
8875 that it can utilise the decrement and branch instruction, but will give
8876 up if there is more than one memory reference in the loop. Thus a loop
8877 where the loop counter is decremented can generate slightly more
8878 efficient code, in cases where the RPTB instruction cannot be utilised.
8880 @item -mdp-isr-reload
8882 @opindex mdp-isr-reload
8884 Force the DP register to be saved on entry to an interrupt service
8885 routine (ISR), reloaded to point to the data section, and restored on
8886 exit from the ISR@. This should not be required unless someone has
8887 violated the small memory model by modifying the DP register, say within
8894 For the C3x use the 24-bit MPYI instruction for integer multiplies
8895 instead of a library call to guarantee 32-bit results. Note that if one
8896 of the operands is a constant, then the multiplication will be performed
8897 using shifts and adds. If the @option{-mmpyi} option is not specified for the C3x,
8898 then squaring operations are performed inline instead of a library call.
8901 @itemx -mno-fast-fix
8903 @opindex mno-fast-fix
8904 The C3x/C4x FIX instruction to convert a floating point value to an
8905 integer value chooses the nearest integer less than or equal to the
8906 floating point value rather than to the nearest integer. Thus if the
8907 floating point number is negative, the result will be incorrectly
8908 truncated an additional code is necessary to detect and correct this
8909 case. This option can be used to disable generation of the additional
8910 code required to correct the result.
8916 Enable (disable) generation of repeat block sequences using the RPTB
8917 instruction for zero overhead looping. The RPTB construct is only used
8918 for innermost loops that do not call functions or jump across the loop
8919 boundaries. There is no advantage having nested RPTB loops due to the
8920 overhead required to save and restore the RC, RS, and RE registers.
8921 This is enabled by default with @option{-O2}.
8923 @item -mrpts=@var{count}
8927 Enable (disable) the use of the single instruction repeat instruction
8928 RPTS@. If a repeat block contains a single instruction, and the loop
8929 count can be guaranteed to be less than the value @var{count}, GCC will
8930 emit a RPTS instruction instead of a RPTB@. If no value is specified,
8931 then a RPTS will be emitted even if the loop count cannot be determined
8932 at compile time. Note that the repeated instruction following RPTS does
8933 not have to be reloaded from memory each iteration, thus freeing up the
8934 CPU buses for operands. However, since interrupts are blocked by this
8935 instruction, it is disabled by default.
8937 @item -mloop-unsigned
8938 @itemx -mno-loop-unsigned
8939 @opindex mloop-unsigned
8940 @opindex mno-loop-unsigned
8941 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
8942 is @math{2^31 + 1} since these instructions test if the iteration count is
8943 negative to terminate the loop. If the iteration count is unsigned
8944 there is a possibility than the @math{2^31 + 1} maximum iteration count may be
8945 exceeded. This switch allows an unsigned iteration count.
8949 Try to emit an assembler syntax that the TI assembler (asm30) is happy
8950 with. This also enforces compatibility with the API employed by the TI
8951 C3x C compiler. For example, long doubles are passed as structures
8952 rather than in floating point registers.
8958 Generate code that uses registers (stack) for passing arguments to functions.
8959 By default, arguments are passed in registers where possible rather
8960 than by pushing arguments on to the stack.
8962 @item -mparallel-insns
8963 @itemx -mno-parallel-insns
8964 @opindex mparallel-insns
8965 @opindex mno-parallel-insns
8966 Allow the generation of parallel instructions. This is enabled by
8967 default with @option{-O2}.
8969 @item -mparallel-mpy
8970 @itemx -mno-parallel-mpy
8971 @opindex mparallel-mpy
8972 @opindex mno-parallel-mpy
8973 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
8974 provided @option{-mparallel-insns} is also specified. These instructions have
8975 tight register constraints which can pessimize the code generation
8981 @subsection V850 Options
8982 @cindex V850 Options
8984 These @samp{-m} options are defined for V850 implementations:
8988 @itemx -mno-long-calls
8989 @opindex mlong-calls
8990 @opindex mno-long-calls
8991 Treat all calls as being far away (near). If calls are assumed to be
8992 far away, the compiler will always load the functions address up into a
8993 register, and call indirect through the pointer.
8999 Do not optimize (do optimize) basic blocks that use the same index
9000 pointer 4 or more times to copy pointer into the @code{ep} register, and
9001 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
9002 option is on by default if you optimize.
9004 @item -mno-prolog-function
9005 @itemx -mprolog-function
9006 @opindex mno-prolog-function
9007 @opindex mprolog-function
9008 Do not use (do use) external functions to save and restore registers at
9009 the prolog and epilog of a function. The external functions are slower,
9010 but use less code space if more than one function saves the same number
9011 of registers. The @option{-mprolog-function} option is on by default if
9016 Try to make the code as small as possible. At present, this just turns
9017 on the @option{-mep} and @option{-mprolog-function} options.
9021 Put static or global variables whose size is @var{n} bytes or less into
9022 the tiny data area that register @code{ep} points to. The tiny data
9023 area can hold up to 256 bytes in total (128 bytes for byte references).
9027 Put static or global variables whose size is @var{n} bytes or less into
9028 the small data area that register @code{gp} points to. The small data
9029 area can hold up to 64 kilobytes.
9033 Put static or global variables whose size is @var{n} bytes or less into
9034 the first 32 kilobytes of memory.
9038 Specify that the target processor is the V850.
9041 @opindex mbig-switch
9042 Generate code suitable for big switch tables. Use this option only if
9043 the assembler/linker complain about out of range branches within a switch
9048 @subsection ARC Options
9051 These options are defined for ARC implementations:
9056 Compile code for little endian mode. This is the default.
9060 Compile code for big endian mode.
9063 @opindex mmangle-cpu
9064 Prepend the name of the cpu to all public symbol names.
9065 In multiple-processor systems, there are many ARC variants with different
9066 instruction and register set characteristics. This flag prevents code
9067 compiled for one cpu to be linked with code compiled for another.
9068 No facility exists for handling variants that are ``almost identical''.
9069 This is an all or nothing option.
9071 @item -mcpu=@var{cpu}
9073 Compile code for ARC variant @var{cpu}.
9074 Which variants are supported depend on the configuration.
9075 All variants support @option{-mcpu=base}, this is the default.
9077 @item -mtext=@var{text-section}
9078 @itemx -mdata=@var{data-section}
9079 @itemx -mrodata=@var{readonly-data-section}
9083 Put functions, data, and readonly data in @var{text-section},
9084 @var{data-section}, and @var{readonly-data-section} respectively
9085 by default. This can be overridden with the @code{section} attribute.
9086 @xref{Variable Attributes}.
9091 @subsection NS32K Options
9092 @cindex NS32K options
9094 These are the @samp{-m} options defined for the 32000 series. The default
9095 values for these options depends on which style of 32000 was selected when
9096 the compiler was configured; the defaults for the most common choices are
9104 Generate output for a 32032. This is the default
9105 when the compiler is configured for 32032 and 32016 based systems.
9111 Generate output for a 32332. This is the default
9112 when the compiler is configured for 32332-based systems.
9118 Generate output for a 32532. This is the default
9119 when the compiler is configured for 32532-based systems.
9123 Generate output containing 32081 instructions for floating point.
9124 This is the default for all systems.
9128 Generate output containing 32381 instructions for floating point. This
9129 also implies @option{-m32081}. The 32381 is only compatible with the 32332
9130 and 32532 cpus. This is the default for the pc532-netbsd configuration.
9134 Try and generate multiply-add floating point instructions @code{polyF}
9135 and @code{dotF}. This option is only available if the @option{-m32381}
9136 option is in effect. Using these instructions requires changes to
9137 register allocation which generally has a negative impact on
9138 performance. This option should only be enabled when compiling code
9139 particularly likely to make heavy use of multiply-add instructions.
9142 @opindex mnomulti-add
9143 Do not try and generate multiply-add floating point instructions
9144 @code{polyF} and @code{dotF}. This is the default on all platforms.
9147 @opindex msoft-float
9148 Generate output containing library calls for floating point.
9149 @strong{Warning:} the requisite libraries may not be available.
9152 @opindex mnobitfield
9153 Do not use the bit-field instructions. On some machines it is faster to
9154 use shifting and masking operations. This is the default for the pc532.
9158 Do use the bit-field instructions. This is the default for all platforms
9163 Use a different function-calling convention, in which functions
9164 that take a fixed number of arguments return pop their
9165 arguments on return with the @code{ret} instruction.
9167 This calling convention is incompatible with the one normally
9168 used on Unix, so you cannot use it if you need to call libraries
9169 compiled with the Unix compiler.
9171 Also, you must provide function prototypes for all functions that
9172 take variable numbers of arguments (including @code{printf});
9173 otherwise incorrect code will be generated for calls to those
9176 In addition, seriously incorrect code will result if you call a
9177 function with too many arguments. (Normally, extra arguments are
9178 harmlessly ignored.)
9180 This option takes its name from the 680x0 @code{rtd} instruction.
9185 Use a different function-calling convention where the first two arguments
9186 are passed in registers.
9188 This calling convention is incompatible with the one normally
9189 used on Unix, so you cannot use it if you need to call libraries
9190 compiled with the Unix compiler.
9193 @opindex mnoregparam
9194 Do not pass any arguments in registers. This is the default for all
9199 It is OK to use the sb as an index register which is always loaded with
9200 zero. This is the default for the pc532-netbsd target.
9204 The sb register is not available for use or has not been initialized to
9205 zero by the run time system. This is the default for all targets except
9206 the pc532-netbsd. It is also implied whenever @option{-mhimem} or
9207 @option{-fpic} is set.
9211 Many ns32000 series addressing modes use displacements of up to 512MB@.
9212 If an address is above 512MB then displacements from zero can not be used.
9213 This option causes code to be generated which can be loaded above 512MB@.
9214 This may be useful for operating systems or ROM code.
9218 Assume code will be loaded in the first 512MB of virtual address space.
9219 This is the default for all platforms.
9225 @subsection AVR Options
9228 These options are defined for AVR implementations:
9231 @item -mmcu=@var{mcu}
9233 Specify ATMEL AVR instruction set or MCU type.
9235 Instruction set avr1 is for the minimal AVR core, not supported by the C
9236 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
9237 attiny11, attiny12, attiny15, attiny28).
9239 Instruction set avr2 (default) is for the classic AVR core with up to
9240 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
9241 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
9242 at90c8534, at90s8535).
9244 Instruction set avr3 is for the classic AVR core with up to 128K program
9245 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
9247 Instruction set avr4 is for the enhanced AVR core with up to 8K program
9248 memory space (MCU types: atmega8, atmega83, atmega85).
9250 Instruction set avr5 is for the enhanced AVR core with up to 128K program
9251 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
9252 atmega64, atmega128, at43usb355, at94k).
9256 Output instruction sizes to the asm file.
9258 @item -minit-stack=@var{N}
9259 @opindex minit-stack
9260 Specify the initial stack address, which may be a symbol or numeric value,
9261 @samp{__stack} is the default.
9263 @item -mno-interrupts
9264 @opindex mno-interrupts
9265 Generated code is not compatible with hardware interrupts.
9266 Code size will be smaller.
9268 @item -mcall-prologues
9269 @opindex mcall-prologues
9270 Functions prologues/epilogues expanded as call to appropriate
9271 subroutines. Code size will be smaller.
9273 @item -mno-tablejump
9274 @opindex mno-tablejump
9275 Do not generate tablejump insns which sometimes increase code size.
9278 @opindex mtiny-stack
9279 Change only the low 8 bits of the stack pointer.
9283 @subsection MCore Options
9284 @cindex MCore options
9286 These are the @samp{-m} options defined for the Motorola M*Core
9296 @opindex mno-hardlit
9297 Inline constants into the code stream if it can be done in two
9298 instructions or less.
9306 Use the divide instruction. (Enabled by default).
9308 @item -mrelax-immediate
9309 @itemx -mrelax-immediate
9310 @itemx -mno-relax-immediate
9311 @opindex mrelax-immediate
9312 @opindex mrelax-immediate
9313 @opindex mno-relax-immediate
9314 Allow arbitrary sized immediates in bit operations.
9316 @item -mwide-bitfields
9317 @itemx -mwide-bitfields
9318 @itemx -mno-wide-bitfields
9319 @opindex mwide-bitfields
9320 @opindex mwide-bitfields
9321 @opindex mno-wide-bitfields
9322 Always treat bit-fields as int-sized.
9324 @item -m4byte-functions
9325 @itemx -m4byte-functions
9326 @itemx -mno-4byte-functions
9327 @opindex m4byte-functions
9328 @opindex m4byte-functions
9329 @opindex mno-4byte-functions
9330 Force all functions to be aligned to a four byte boundary.
9332 @item -mcallgraph-data
9333 @itemx -mcallgraph-data
9334 @itemx -mno-callgraph-data
9335 @opindex mcallgraph-data
9336 @opindex mcallgraph-data
9337 @opindex mno-callgraph-data
9338 Emit callgraph information.
9342 @itemx -mno-slow-bytes
9343 @opindex mslow-bytes
9344 @opindex mslow-bytes
9345 @opindex mno-slow-bytes
9346 Prefer word access when reading byte quantities.
9348 @item -mlittle-endian
9349 @itemx -mlittle-endian
9351 @opindex mlittle-endian
9352 @opindex mlittle-endian
9353 @opindex mbig-endian
9354 Generate code for a little endian target.
9362 Generate code for the 210 processor.
9366 @subsection IA-64 Options
9367 @cindex IA-64 Options
9369 These are the @samp{-m} options defined for the Intel IA-64 architecture.
9373 @opindex mbig-endian
9374 Generate code for a big endian target. This is the default for HPUX@.
9376 @item -mlittle-endian
9377 @opindex mlittle-endian
9378 Generate code for a little endian target. This is the default for AIX5
9385 Generate (or don't) code for the GNU assembler. This is the default.
9386 @c Also, this is the default if the configure option @option{--with-gnu-as}
9393 Generate (or don't) code for the GNU linker. This is the default.
9394 @c Also, this is the default if the configure option @option{--with-gnu-ld}
9399 Generate code that does not use a global pointer register. The result
9400 is not position independent code, and violates the IA-64 ABI@.
9402 @item -mvolatile-asm-stop
9403 @itemx -mno-volatile-asm-stop
9404 @opindex mvolatile-asm-stop
9405 @opindex mno-volatile-asm-stop
9406 Generate (or don't) a stop bit immediately before and after volatile asm
9411 Generate code that works around Itanium B step errata.
9413 @item -mregister-names
9414 @itemx -mno-register-names
9415 @opindex mregister-names
9416 @opindex mno-register-names
9417 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
9418 the stacked registers. This may make assembler output more readable.
9424 Disable (or enable) optimizations that use the small data section. This may
9425 be useful for working around optimizer bugs.
9428 @opindex mconstant-gp
9429 Generate code that uses a single constant global pointer value. This is
9430 useful when compiling kernel code.
9434 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
9435 This is useful when compiling firmware code.
9437 @item -minline-divide-min-latency
9438 @opindex minline-divide-min-latency
9439 Generate code for inline divides using the minimum latency algorithm.
9441 @item -minline-divide-max-throughput
9442 @opindex minline-divide-max-throughput
9443 Generate code for inline divides using the maximum throughput algorithm.
9445 @item -mno-dwarf2-asm
9447 @opindex mno-dwarf2-asm
9448 @opindex mdwarf2-asm
9449 Don't (or do) generate assembler code for the DWARF2 line number debugging
9450 info. This may be useful when not using the GNU assembler.
9452 @item -mfixed-range=@var{register-range}
9453 @opindex mfixed-range
9454 Generate code treating the given register range as fixed registers.
9455 A fixed register is one that the register allocator can not use. This is
9456 useful when compiling kernel code. A register range is specified as
9457 two registers separated by a dash. Multiple register ranges can be
9458 specified separated by a comma.
9462 @subsection D30V Options
9463 @cindex D30V Options
9465 These @samp{-m} options are defined for D30V implementations:
9470 Link the @samp{.text}, @samp{.data}, @samp{.bss}, @samp{.strings},
9471 @samp{.rodata}, @samp{.rodata1}, @samp{.data1} sections into external
9472 memory, which starts at location @code{0x80000000}.
9476 Same as the @option{-mextmem} switch.
9480 Link the @samp{.text} section into onchip text memory, which starts at
9481 location @code{0x0}. Also link @samp{.data}, @samp{.bss},
9482 @samp{.strings}, @samp{.rodata}, @samp{.rodata1}, @samp{.data1} sections
9483 into onchip data memory, which starts at location @code{0x20000000}.
9485 @item -mno-asm-optimize
9486 @itemx -masm-optimize
9487 @opindex mno-asm-optimize
9488 @opindex masm-optimize
9489 Disable (enable) passing @option{-O} to the assembler when optimizing.
9490 The assembler uses the @option{-O} option to automatically parallelize
9491 adjacent short instructions where possible.
9493 @item -mbranch-cost=@var{n}
9494 @opindex mbranch-cost
9495 Increase the internal costs of branches to @var{n}. Higher costs means
9496 that the compiler will issue more instructions to avoid doing a branch.
9499 @item -mcond-exec=@var{n}
9501 Specify the maximum number of conditionally executed instructions that
9502 replace a branch. The default is 4.
9505 @node S/390 and zSeries Options
9506 @subsection S/390 and zSeries Options
9507 @cindex S/390 and zSeries Options
9509 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
9514 @opindex mhard-float
9515 @opindex msoft-float
9516 Use (do not use) the hardware floating-point instructions and registers
9517 for floating-point operations. When @option{-msoft-float} is specified,
9518 functions in @file{libgcc.a} will be used to perform floating-point
9519 operations. When @option{-mhard-float} is specified, the compiler
9520 generates IEEE floating-point instructions. This is the default.
9523 @itemx -mno-backchain
9525 @opindex mno-backchain
9526 Generate (or do not generate) code which maintains an explicit
9527 backchain within the stack frame that points to the caller's frame.
9528 This is currently needed to allow debugging. The default is to
9529 generate the backchain.
9532 @itemx -mno-small-exec
9533 @opindex msmall-exec
9534 @opindex mno-small-exec
9535 Generate (or do not generate) code using the @code{bras} instruction
9536 to do subroutine calls.
9537 This only works reliably if the total executable size does not
9538 exceed 64k. The default is to use the @code{basr} instruction instead,
9539 which does not have this limitation.
9545 When @option{-m31} is specified, generate code compliant to the
9546 Linux for S/390 ABI@. When @option{-m64} is specified, generate
9547 code compliant to the Linux for zSeries ABI@. This allows GCC in
9548 particular to generate 64-bit instructions. For the @samp{s390}
9549 targets, the default is @option{-m31}, while the @samp{s390x}
9550 targets default to @option{-m64}.
9556 Generate (or do not generate) code using the @code{mvcle} instruction
9557 to perform block moves. When @option{-mno-mvcle} is specifed,
9558 use a @code{mvc} loop instead. This is the default.
9564 Print (or do not print) additional debug information when compiling.
9565 The default is to not print debug information.
9570 @subsection CRIS Options
9571 @cindex CRIS Options
9573 These options are defined specifically for the CRIS ports.
9576 @item -march=@var{architecture-type}
9577 @itemx -mcpu=@var{architecture-type}
9580 Generate code for the specified architecture. The choices for
9581 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
9582 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX.
9583 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
9586 @item -mtune=@var{architecture-type}
9588 Tune to @var{architecture-type} everything applicable about the generated
9589 code, except for the ABI and the set of available instructions. The
9590 choices for @var{architecture-type} are the same as for
9591 @option{-march=@var{architecture-type}}.
9593 @item -mmax-stack-frame=@var{n}
9594 @opindex mmax-stack-frame
9595 Warn when the stack frame of a function exceeds @var{n} bytes.
9597 @item -melinux-stacksize=@var{n}
9598 @opindex melinux-stacksize
9599 Only available with the @samp{cris-axis-aout} target. Arranges for
9600 indications in the program to the kernel loader that the stack of the
9601 program should be set to @var{n} bytes.
9607 The options @option{-metrax4} and @option{-metrax100} are synonyms for
9608 @option{-march=v3} and @option{-march=v8} respectively.
9612 Enable CRIS-specific verbose debug-related information in the assembly
9613 code. This option also has the effect to turn off the @samp{#NO_APP}
9614 formatted-code indicator to the assembler at the beginning of the
9619 Do not use condition-code results from previous instruction; always emit
9620 compare and test instructions before use of condition codes.
9622 @item -mno-side-effects
9623 @opindex mno-side-effects
9624 Do not emit instructions with side-effects in addressing modes other than
9628 @itemx -mno-stack-align
9630 @itemx -mno-data-align
9631 @itemx -mconst-align
9632 @itemx -mno-const-align
9633 @opindex mstack-align
9634 @opindex mno-stack-align
9635 @opindex mdata-align
9636 @opindex mno-data-align
9637 @opindex mconst-align
9638 @opindex mno-const-align
9639 These options (no-options) arranges (eliminate arrangements) for the
9640 stack-frame, individual data and constants to be aligned for the maximum
9641 single data access size for the chosen CPU model. The default is to
9642 arrange for 32-bit alignment. ABI details such as structure layout are
9643 not affected by these options.
9651 Similar to the stack- data- and const-align options above, these options
9652 arrange for stack-frame, writable data and constants to all be 32-bit,
9653 16-bit or 8-bit aligned. The default is 32-bit alignment.
9655 @item -mno-prologue-epilogue
9656 @itemx -mprologue-epilogue
9657 @opindex mno-prologue-epilogue
9658 @opindex mprologue-epilogue
9659 With @option{-mno-prologue-epilogue}, the normal function prologue and
9660 epilogue that sets up the stack-frame are omitted and no return
9661 instructions or return sequences are generated in the code. Use this
9662 option only together with visual inspection of the compiled code: no
9663 warnings or errors are generated when call-saved registers must be saved,
9664 or storage for local variable needs to be allocated.
9670 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
9671 instruction sequences that load addresses for functions from the PLT part
9672 of the GOT rather than (traditional on other architectures) calls to the
9673 PLT. The default is @option{-mgotplt}.
9677 Legacy no-op option only recognized with the cris-axis-aout target.
9681 Legacy no-op option only recognized with the cris-axis-elf and
9682 cris-axis-linux-gnu targets.
9686 Only recognized with the cris-axis-aout target, where it selects a
9687 GNU/linux-like multilib, include files and instruction set for
9692 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
9696 This option, recognized for the cris-axis-aout and cris-axis-elf arranges
9697 to link with input-output functions from a simulator library. Code,
9698 initialized data and zero-initialized data are allocated consecutively.
9702 Like @option{-sim}, but pass linker options to locate initialized data at
9703 0x40000000 and zero-initialized data at 0x80000000.
9707 @subsection MMIX Options
9708 @cindex MMIX Options
9710 These options are defined for the MMIX:
9714 @itemx -mno-libfuncs
9715 Specify that intrinsic library functions are being compiled, passing all
9716 values in registers, no matter the size.
9720 Generate floating-point comparison instructions that compare with respect
9721 to the @code{rE} epsilon register.
9723 @item -mabi=mmixware
9725 Generate code that passes function parameters and return values that (in
9726 the called function) are seen as registers @code{$0} and up, as opposed to
9727 the GNU ABI which uses global registers @code{$231} and up.
9730 @itemx -mno-zero-extend
9731 When reading data from memory in sizes shorter than 64 bits, use (do not
9732 use) zero-extending load instructions by default, rather than
9733 sign-extending ones.
9736 @itemx -mno-knuthdiv
9737 Make the result of a division yielding a remainder have the same sign as
9738 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
9739 remainder follows the sign of the dividend. Both methods are
9740 arithmetically valid, the latter being almost exclusively used.
9742 @item -mtoplevel-symbols
9743 @itemx -mno-toplevel-symbols
9744 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
9745 code can be used with the @code{PREFIX} assembly directive.
9748 Generate an executable in the ELF format, rather than the default
9749 @samp{mmo} format used by the @command{mmix} simulator.
9751 @item -mbranch-predict
9752 @itemx -mno-branch-predict
9753 Use (do not use) the probable-branch instructions, when static branch
9754 prediction indicates a probable branch.
9756 @item -mreg-stack-fill-bug-workaround
9757 @itemx -mno-reg-stack-fill-bug-workaround
9758 Work around (do not work around) an inconsistency in the circular
9759 register stack mechanism in the @command{mmix} simulator, which
9760 causes entries in the register stack to not be flushed to memory if
9761 the instruction causing the fill-up is @code{PUSHJ} or @code{PUSHGO}.
9764 @node Code Gen Options
9765 @section Options for Code Generation Conventions
9766 @cindex code generation conventions
9767 @cindex options, code generation
9768 @cindex run-time options
9770 These machine-independent options control the interface conventions
9771 used in code generation.
9773 Most of them have both positive and negative forms; the negative form
9774 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
9775 one of the forms is listed---the one which is not the default. You
9776 can figure out the other form by either removing @samp{no-} or adding
9781 @opindex fexceptions
9782 Enable exception handling. Generates extra code needed to propagate
9783 exceptions. For some targets, this implies GCC will generate frame
9784 unwind information for all functions, which can produce significant data
9785 size overhead, although it does not affect execution. If you do not
9786 specify this option, GCC will enable it by default for languages like
9787 C++ which normally require exception handling, and disable it for
9788 languages like C that do not normally require it. However, you may need
9789 to enable this option when compiling C code that needs to interoperate
9790 properly with exception handlers written in C++. You may also wish to
9791 disable this option if you are compiling older C++ programs that don't
9792 use exception handling.
9794 @item -fnon-call-exceptions
9795 @opindex fnon-call-exceptions
9796 Generate code that allows trapping instructions to throw exceptions.
9797 Note that this requires platform-specific runtime support that does
9798 not exist everywhere. Moreover, it only allows @emph{trapping}
9799 instructions to throw exceptions, i.e.@: memory references or floating
9800 point instructions. It does not allow exceptions to be thrown from
9801 arbitrary signal handlers such as @code{SIGALRM}.
9803 @item -funwind-tables
9804 @opindex funwind-tables
9805 Similar to @option{-fexceptions}, except that it will just generate any needed
9806 static data, but will not affect the generated code in any other way.
9807 You will normally not enable this option; instead, a language processor
9808 that needs this handling would enable it on your behalf.
9810 @item -fasynchronous-unwind-tables
9811 @opindex funwind-tables
9812 Generate unwind table in dwarf2 format, if supported by target machine. The
9813 table is exact at each instruction boundary, so it can be used for stack
9814 unwinding from asynchronous events (such as debugger or garbage collector).
9816 @item -fpcc-struct-return
9817 @opindex fpcc-struct-return
9818 Return ``short'' @code{struct} and @code{union} values in memory like
9819 longer ones, rather than in registers. This convention is less
9820 efficient, but it has the advantage of allowing intercallability between
9821 GCC-compiled files and files compiled with other compilers.
9823 The precise convention for returning structures in memory depends
9824 on the target configuration macros.
9826 Short structures and unions are those whose size and alignment match
9827 that of some integer type.
9829 @item -freg-struct-return
9830 @opindex freg-struct-return
9831 Return @code{struct} and @code{union} values in registers when possible.
9832 This is more efficient for small structures than
9833 @option{-fpcc-struct-return}.
9835 If you specify neither @option{-fpcc-struct-return} nor
9836 @option{-freg-struct-return}, GCC defaults to whichever convention is
9837 standard for the target. If there is no standard convention, GCC
9838 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
9839 the principal compiler. In those cases, we can choose the standard, and
9840 we chose the more efficient register return alternative.
9843 @opindex fshort-enums
9844 Allocate to an @code{enum} type only as many bytes as it needs for the
9845 declared range of possible values. Specifically, the @code{enum} type
9846 will be equivalent to the smallest integer type which has enough room.
9848 @item -fshort-double
9849 @opindex fshort-double
9850 Use the same size for @code{double} as for @code{float}.
9853 @opindex fshared-data
9854 Requests that the data and non-@code{const} variables of this
9855 compilation be shared data rather than private data. The distinction
9856 makes sense only on certain operating systems, where shared data is
9857 shared between processes running the same program, while private data
9858 exists in one copy per process.
9862 In C, allocate even uninitialized global variables in the data section of the
9863 object file, rather than generating them as common blocks. This has the
9864 effect that if the same variable is declared (without @code{extern}) in
9865 two different compilations, you will get an error when you link them.
9866 The only reason this might be useful is if you wish to verify that the
9867 program will work on other systems which always work this way.
9871 Ignore the @samp{#ident} directive.
9873 @item -fno-gnu-linker
9874 @opindex fno-gnu-linker
9875 Do not output global initializations (such as C++ constructors and
9876 destructors) in the form used by the GNU linker (on systems where the GNU
9877 linker is the standard method of handling them). Use this option when
9878 you want to use a non-GNU linker, which also requires using the
9879 @command{collect2} program to make sure the system linker includes
9880 constructors and destructors. (@command{collect2} is included in the GCC
9881 distribution.) For systems which @emph{must} use @command{collect2}, the
9882 compiler driver @command{gcc} is configured to do this automatically.
9884 @item -finhibit-size-directive
9885 @opindex finhibit-size-directive
9886 Don't output a @code{.size} assembler directive, or anything else that
9887 would cause trouble if the function is split in the middle, and the
9888 two halves are placed at locations far apart in memory. This option is
9889 used when compiling @file{crtstuff.c}; you should not need to use it
9893 @opindex fverbose-asm
9894 Put extra commentary information in the generated assembly code to
9895 make it more readable. This option is generally only of use to those
9896 who actually need to read the generated assembly code (perhaps while
9897 debugging the compiler itself).
9899 @option{-fno-verbose-asm}, the default, causes the
9900 extra information to be omitted and is useful when comparing two assembler
9905 Consider all memory references through pointers to be volatile.
9907 @item -fvolatile-global
9908 @opindex fvolatile-global
9909 Consider all memory references to extern and global data items to
9910 be volatile. GCC does not consider static data items to be volatile
9911 because of this switch.
9913 @item -fvolatile-static
9914 @opindex fvolatile-static
9915 Consider all memory references to static data to be volatile.
9919 @cindex global offset table
9921 Generate position-independent code (PIC) suitable for use in a shared
9922 library, if supported for the target machine. Such code accesses all
9923 constant addresses through a global offset table (GOT)@. The dynamic
9924 loader resolves the GOT entries when the program starts (the dynamic
9925 loader is not part of GCC; it is part of the operating system). If
9926 the GOT size for the linked executable exceeds a machine-specific
9927 maximum size, you get an error message from the linker indicating that
9928 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
9929 instead. (These maximums are 16k on the m88k, 8k on the Sparc, and 32k
9930 on the m68k and RS/6000. The 386 has no such limit.)
9932 Position-independent code requires special support, and therefore works
9933 only on certain machines. For the 386, GCC supports PIC for System V
9934 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
9935 position-independent.
9939 If supported for the target machine, emit position-independent code,
9940 suitable for dynamic linking and avoiding any limit on the size of the
9941 global offset table. This option makes a difference on the m68k, m88k,
9944 Position-independent code requires special support, and therefore works
9945 only on certain machines.
9947 @item -ffixed-@var{reg}
9949 Treat the register named @var{reg} as a fixed register; generated code
9950 should never refer to it (except perhaps as a stack pointer, frame
9951 pointer or in some other fixed role).
9953 @var{reg} must be the name of a register. The register names accepted
9954 are machine-specific and are defined in the @code{REGISTER_NAMES}
9955 macro in the machine description macro file.
9957 This flag does not have a negative form, because it specifies a
9960 @item -fcall-used-@var{reg}
9962 Treat the register named @var{reg} as an allocable register that is
9963 clobbered by function calls. It may be allocated for temporaries or
9964 variables that do not live across a call. Functions compiled this way
9965 will not save and restore the register @var{reg}.
9967 It is an error to used this flag with the frame pointer or stack pointer.
9968 Use of this flag for other registers that have fixed pervasive roles in
9969 the machine's execution model will produce disastrous results.
9971 This flag does not have a negative form, because it specifies a
9974 @item -fcall-saved-@var{reg}
9975 @opindex fcall-saved
9976 Treat the register named @var{reg} as an allocable register saved by
9977 functions. It may be allocated even for temporaries or variables that
9978 live across a call. Functions compiled this way will save and restore
9979 the register @var{reg} if they use it.
9981 It is an error to used this flag with the frame pointer or stack pointer.
9982 Use of this flag for other registers that have fixed pervasive roles in
9983 the machine's execution model will produce disastrous results.
9985 A different sort of disaster will result from the use of this flag for
9986 a register in which function values may be returned.
9988 This flag does not have a negative form, because it specifies a
9992 @opindex fpack-struct
9993 Pack all structure members together without holes. Usually you would
9994 not want to use this option, since it makes the code suboptimal, and
9995 the offsets of structure members won't agree with system libraries.
9997 @item -finstrument-functions
9998 @opindex finstrument-functions
9999 Generate instrumentation calls for entry and exit to functions. Just
10000 after function entry and just before function exit, the following
10001 profiling functions will be called with the address of the current
10002 function and its call site. (On some platforms,
10003 @code{__builtin_return_address} does not work beyond the current
10004 function, so the call site information may not be available to the
10005 profiling functions otherwise.)
10008 void __cyg_profile_func_enter (void *this_fn,
10010 void __cyg_profile_func_exit (void *this_fn,
10014 The first argument is the address of the start of the current function,
10015 which may be looked up exactly in the symbol table.
10017 This instrumentation is also done for functions expanded inline in other
10018 functions. The profiling calls will indicate where, conceptually, the
10019 inline function is entered and exited. This means that addressable
10020 versions of such functions must be available. If all your uses of a
10021 function are expanded inline, this may mean an additional expansion of
10022 code size. If you use @samp{extern inline} in your C code, an
10023 addressable version of such functions must be provided. (This is
10024 normally the case anyways, but if you get lucky and the optimizer always
10025 expands the functions inline, you might have gotten away without
10026 providing static copies.)
10028 A function may be given the attribute @code{no_instrument_function}, in
10029 which case this instrumentation will not be done. This can be used, for
10030 example, for the profiling functions listed above, high-priority
10031 interrupt routines, and any functions from which the profiling functions
10032 cannot safely be called (perhaps signal handlers, if the profiling
10033 routines generate output or allocate memory).
10035 @item -fstack-check
10036 @opindex fstack-check
10037 Generate code to verify that you do not go beyond the boundary of the
10038 stack. You should specify this flag if you are running in an
10039 environment with multiple threads, but only rarely need to specify it in
10040 a single-threaded environment since stack overflow is automatically
10041 detected on nearly all systems if there is only one stack.
10043 Note that this switch does not actually cause checking to be done; the
10044 operating system must do that. The switch causes generation of code
10045 to ensure that the operating system sees the stack being extended.
10047 @item -fstack-limit-register=@var{reg}
10048 @itemx -fstack-limit-symbol=@var{sym}
10049 @itemx -fno-stack-limit
10050 @opindex fstack-limit-register
10051 @opindex fstack-limit-symbol
10052 @opindex fno-stack-limit
10053 Generate code to ensure that the stack does not grow beyond a certain value,
10054 either the value of a register or the address of a symbol. If the stack
10055 would grow beyond the value, a signal is raised. For most targets,
10056 the signal is raised before the stack overruns the boundary, so
10057 it is possible to catch the signal without taking special precautions.
10059 For instance, if the stack starts at absolute address @samp{0x80000000}
10060 and grows downwards, you can use the flags
10061 @option{-fstack-limit-symbol=__stack_limit} and
10062 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
10063 of 128KB@. Note that this may only work with the GNU linker.
10065 @cindex aliasing of parameters
10066 @cindex parameters, aliased
10067 @item -fargument-alias
10068 @itemx -fargument-noalias
10069 @itemx -fargument-noalias-global
10070 @opindex fargument-alias
10071 @opindex fargument-noalias
10072 @opindex fargument-noalias-global
10073 Specify the possible relationships among parameters and between
10074 parameters and global data.
10076 @option{-fargument-alias} specifies that arguments (parameters) may
10077 alias each other and may alias global storage.@*
10078 @option{-fargument-noalias} specifies that arguments do not alias
10079 each other, but may alias global storage.@*
10080 @option{-fargument-noalias-global} specifies that arguments do not
10081 alias each other and do not alias global storage.
10083 Each language will automatically use whatever option is required by
10084 the language standard. You should not need to use these options yourself.
10086 @item -fleading-underscore
10087 @opindex fleading-underscore
10088 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
10089 change the way C symbols are represented in the object file. One use
10090 is to help link with legacy assembly code.
10092 Be warned that you should know what you are doing when invoking this
10093 option, and that not all targets provide complete support for it.
10098 @node Environment Variables
10099 @section Environment Variables Affecting GCC
10100 @cindex environment variables
10102 @c man begin ENVIRONMENT
10104 This section describes several environment variables that affect how GCC
10105 operates. Some of them work by specifying directories or prefixes to use
10106 when searching for various kinds of files. Some are used to specify other
10107 aspects of the compilation environment.
10109 Note that you can also specify places to search using options such as
10110 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
10111 take precedence over places specified using environment variables, which
10112 in turn take precedence over those specified by the configuration of GCC@.
10113 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
10114 GNU Compiler Collection (GCC) Internals}.
10119 @c @itemx LC_COLLATE
10121 @c @itemx LC_MONETARY
10122 @c @itemx LC_NUMERIC
10127 @c @findex LC_COLLATE
10128 @findex LC_MESSAGES
10129 @c @findex LC_MONETARY
10130 @c @findex LC_NUMERIC
10134 These environment variables control the way that GCC uses
10135 localization information that allow GCC to work with different
10136 national conventions. GCC inspects the locale categories
10137 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
10138 so. These locale categories can be set to any value supported by your
10139 installation. A typical value is @samp{en_UK} for English in the United
10142 The @env{LC_CTYPE} environment variable specifies character
10143 classification. GCC uses it to determine the character boundaries in
10144 a string; this is needed for some multibyte encodings that contain quote
10145 and escape characters that would otherwise be interpreted as a string
10148 The @env{LC_MESSAGES} environment variable specifies the language to
10149 use in diagnostic messages.
10151 If the @env{LC_ALL} environment variable is set, it overrides the value
10152 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
10153 and @env{LC_MESSAGES} default to the value of the @env{LANG}
10154 environment variable. If none of these variables are set, GCC
10155 defaults to traditional C English behavior.
10159 If @env{TMPDIR} is set, it specifies the directory to use for temporary
10160 files. GCC uses temporary files to hold the output of one stage of
10161 compilation which is to be used as input to the next stage: for example,
10162 the output of the preprocessor, which is the input to the compiler
10165 @item GCC_EXEC_PREFIX
10166 @findex GCC_EXEC_PREFIX
10167 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
10168 names of the subprograms executed by the compiler. No slash is added
10169 when this prefix is combined with the name of a subprogram, but you can
10170 specify a prefix that ends with a slash if you wish.
10172 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
10173 an appropriate prefix to use based on the pathname it was invoked with.
10175 If GCC cannot find the subprogram using the specified prefix, it
10176 tries looking in the usual places for the subprogram.
10178 The default value of @env{GCC_EXEC_PREFIX} is
10179 @file{@var{prefix}/lib/gcc-lib/} where @var{prefix} is the value
10180 of @code{prefix} when you ran the @file{configure} script.
10182 Other prefixes specified with @option{-B} take precedence over this prefix.
10184 This prefix is also used for finding files such as @file{crt0.o} that are
10187 In addition, the prefix is used in an unusual way in finding the
10188 directories to search for header files. For each of the standard
10189 directories whose name normally begins with @samp{/usr/local/lib/gcc-lib}
10190 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
10191 replacing that beginning with the specified prefix to produce an
10192 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
10193 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
10194 These alternate directories are searched first; the standard directories
10197 @item COMPILER_PATH
10198 @findex COMPILER_PATH
10199 The value of @env{COMPILER_PATH} is a colon-separated list of
10200 directories, much like @env{PATH}. GCC tries the directories thus
10201 specified when searching for subprograms, if it can't find the
10202 subprograms using @env{GCC_EXEC_PREFIX}.
10205 @findex LIBRARY_PATH
10206 The value of @env{LIBRARY_PATH} is a colon-separated list of
10207 directories, much like @env{PATH}. When configured as a native compiler,
10208 GCC tries the directories thus specified when searching for special
10209 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
10210 using GCC also uses these directories when searching for ordinary
10211 libraries for the @option{-l} option (but directories specified with
10212 @option{-L} come first).
10214 @item C_INCLUDE_PATH
10215 @itemx CPLUS_INCLUDE_PATH
10216 @itemx OBJC_INCLUDE_PATH
10217 @findex C_INCLUDE_PATH
10218 @findex CPLUS_INCLUDE_PATH
10219 @findex OBJC_INCLUDE_PATH
10220 @c @itemx OBJCPLUS_INCLUDE_PATH
10221 These environment variables pertain to particular languages. Each
10222 variable's value is a colon-separated list of directories, much like
10223 @env{PATH}. When GCC searches for header files, it tries the
10224 directories listed in the variable for the language you are using, after
10225 the directories specified with @option{-I} but before the standard header
10228 @item DEPENDENCIES_OUTPUT
10229 @findex DEPENDENCIES_OUTPUT
10230 @cindex dependencies for make as output
10231 If this variable is set, its value specifies how to output dependencies
10232 for Make based on the header files processed by the compiler. This
10233 output looks much like the output from the @option{-M} option
10234 (@pxref{Preprocessor Options}), but it goes to a separate file, and is
10235 in addition to the usual results of compilation.
10237 The value of @env{DEPENDENCIES_OUTPUT} can be just a file name, in
10238 which case the Make rules are written to that file, guessing the target
10239 name from the source file name. Or the value can have the form
10240 @samp{@var{file} @var{target}}, in which case the rules are written to
10241 file @var{file} using @var{target} as the target name.
10245 @cindex locale definition
10246 This variable is used to pass locale information to the compiler. One way in
10247 which this information is used is to determine the character set to be used
10248 when character literals, string literals and comments are parsed in C and C++.
10249 When the compiler is configured to allow multibyte characters,
10250 the following values for @env{LANG} are recognized:
10254 Recognize JIS characters.
10256 Recognize SJIS characters.
10258 Recognize EUCJP characters.
10261 If @env{LANG} is not defined, or if it has some other value, then the
10262 compiler will use mblen and mbtowc as defined by the default locale to
10263 recognize and translate multibyte characters.
10268 @node Running Protoize
10269 @section Running Protoize
10271 The program @code{protoize} is an optional part of GCC@. You can use
10272 it to add prototypes to a program, thus converting the program to ISO
10273 C in one respect. The companion program @code{unprotoize} does the
10274 reverse: it removes argument types from any prototypes that are found.
10276 When you run these programs, you must specify a set of source files as
10277 command line arguments. The conversion programs start out by compiling
10278 these files to see what functions they define. The information gathered
10279 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
10281 After scanning comes actual conversion. The specified files are all
10282 eligible to be converted; any files they include (whether sources or
10283 just headers) are eligible as well.
10285 But not all the eligible files are converted. By default,
10286 @code{protoize} and @code{unprotoize} convert only source and header
10287 files in the current directory. You can specify additional directories
10288 whose files should be converted with the @option{-d @var{directory}}
10289 option. You can also specify particular files to exclude with the
10290 @option{-x @var{file}} option. A file is converted if it is eligible, its
10291 directory name matches one of the specified directory names, and its
10292 name within the directory has not been excluded.
10294 Basic conversion with @code{protoize} consists of rewriting most
10295 function definitions and function declarations to specify the types of
10296 the arguments. The only ones not rewritten are those for varargs
10299 @code{protoize} optionally inserts prototype declarations at the
10300 beginning of the source file, to make them available for any calls that
10301 precede the function's definition. Or it can insert prototype
10302 declarations with block scope in the blocks where undeclared functions
10305 Basic conversion with @code{unprotoize} consists of rewriting most
10306 function declarations to remove any argument types, and rewriting
10307 function definitions to the old-style pre-ISO form.
10309 Both conversion programs print a warning for any function declaration or
10310 definition that they can't convert. You can suppress these warnings
10313 The output from @code{protoize} or @code{unprotoize} replaces the
10314 original source file. The original file is renamed to a name ending
10315 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
10316 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
10317 for DOS) file already exists, then the source file is simply discarded.
10319 @code{protoize} and @code{unprotoize} both depend on GCC itself to
10320 scan the program and collect information about the functions it uses.
10321 So neither of these programs will work until GCC is installed.
10323 Here is a table of the options you can use with @code{protoize} and
10324 @code{unprotoize}. Each option works with both programs unless
10328 @item -B @var{directory}
10329 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
10330 usual directory (normally @file{/usr/local/lib}). This file contains
10331 prototype information about standard system functions. This option
10332 applies only to @code{protoize}.
10334 @item -c @var{compilation-options}
10335 Use @var{compilation-options} as the options when running @code{gcc} to
10336 produce the @samp{.X} files. The special option @option{-aux-info} is
10337 always passed in addition, to tell @code{gcc} to write a @samp{.X} file.
10339 Note that the compilation options must be given as a single argument to
10340 @code{protoize} or @code{unprotoize}. If you want to specify several
10341 @code{gcc} options, you must quote the entire set of compilation options
10342 to make them a single word in the shell.
10344 There are certain @code{gcc} arguments that you cannot use, because they
10345 would produce the wrong kind of output. These include @option{-g},
10346 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
10347 the @var{compilation-options}, they are ignored.
10350 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
10351 systems) instead of @samp{.c}. This is convenient if you are converting
10352 a C program to C++. This option applies only to @code{protoize}.
10355 Add explicit global declarations. This means inserting explicit
10356 declarations at the beginning of each source file for each function
10357 that is called in the file and was not declared. These declarations
10358 precede the first function definition that contains a call to an
10359 undeclared function. This option applies only to @code{protoize}.
10361 @item -i @var{string}
10362 Indent old-style parameter declarations with the string @var{string}.
10363 This option applies only to @code{protoize}.
10365 @code{unprotoize} converts prototyped function definitions to old-style
10366 function definitions, where the arguments are declared between the
10367 argument list and the initial @samp{@{}. By default, @code{unprotoize}
10368 uses five spaces as the indentation. If you want to indent with just
10369 one space instead, use @option{-i " "}.
10372 Keep the @samp{.X} files. Normally, they are deleted after conversion
10376 Add explicit local declarations. @code{protoize} with @option{-l} inserts
10377 a prototype declaration for each function in each block which calls the
10378 function without any declaration. This option applies only to
10382 Make no real changes. This mode just prints information about the conversions
10383 that would have been done without @option{-n}.
10386 Make no @samp{.save} files. The original files are simply deleted.
10387 Use this option with caution.
10389 @item -p @var{program}
10390 Use the program @var{program} as the compiler. Normally, the name
10391 @file{gcc} is used.
10394 Work quietly. Most warnings are suppressed.
10397 Print the version number, just like @option{-v} for @code{gcc}.
10400 If you need special compiler options to compile one of your program's
10401 source files, then you should generate that file's @samp{.X} file
10402 specially, by running @code{gcc} on that source file with the
10403 appropriate options and the option @option{-aux-info}. Then run
10404 @code{protoize} on the entire set of files. @code{protoize} will use
10405 the existing @samp{.X} file because it is newer than the source file.
10409 gcc -Dfoo=bar file1.c -aux-info file1.X
10414 You need to include the special files along with the rest in the
10415 @code{protoize} command, even though their @samp{.X} files already
10416 exist, because otherwise they won't get converted.
10418 @xref{Protoize Caveats}, for more information on how to use
10419 @code{protoize} successfully.