1 @c Copyright (C) 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
2 @c 2000, 2001, 2002, 2003 Free Software Foundation, Inc.
3 @c This is part of the GCC manual.
4 @c For copying conditions, see the file gcc.texi.
8 Copyright @copyright{} 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997,
9 1998, 1999, 2000, 2001, 2002, 2003 Free Software Foundation, Inc.
11 Permission is granted to copy, distribute and/or modify this document
12 under the terms of the GNU Free Documentation License, Version 1.2 or
13 any later version published by the Free Software Foundation; with the
14 Invariant Sections being ``GNU General Public License'' and ``Funding
15 Free Software'', the Front-Cover texts being (a) (see below), and with
16 the Back-Cover Texts being (b) (see below). A copy of the license is
17 included in the gfdl(7) man page.
19 (a) The FSF's Front-Cover Text is:
23 (b) The FSF's Back-Cover Text is:
25 You have freedom to copy and modify this GNU Manual, like GNU
26 software. Copies published by the Free Software Foundation raise
27 funds for GNU development.
29 @c Set file name and title for the man page.
31 @settitle GNU project C and C++ compiler
33 gcc [@option{-c}|@option{-S}|@option{-E}] [@option{-std=}@var{standard}]
34 [@option{-g}] [@option{-pg}] [@option{-O}@var{level}]
35 [@option{-W}@var{warn}@dots{}] [@option{-pedantic}]
36 [@option{-I}@var{dir}@dots{}] [@option{-L}@var{dir}@dots{}]
37 [@option{-D}@var{macro}[=@var{defn}]@dots{}] [@option{-U}@var{macro}]
38 [@option{-f}@var{option}@dots{}] [@option{-m}@var{machine-option}@dots{}]
39 [@option{-o} @var{outfile}] @var{infile}@dots{}
41 Only the most useful options are listed here; see below for the
42 remainder. @samp{g++} accepts mostly the same options as @samp{gcc}.
45 gpl(7), gfdl(7), fsf-funding(7),
46 cpp(1), gcov(1), g77(1), as(1), ld(1), gdb(1), adb(1), dbx(1), sdb(1)
47 and the Info entries for @file{gcc}, @file{cpp}, @file{g77}, @file{as},
48 @file{ld}, @file{binutils} and @file{gdb}.
51 For instructions on reporting bugs, see
52 @w{@uref{http://gcc.gnu.org/bugs.html}}. Use of the @command{gccbug}
53 script to report bugs is recommended.
56 See the Info entry for @command{gcc}, or
57 @w{@uref{http://gcc.gnu.org/onlinedocs/gcc/Contributors.html}},
58 for contributors to GCC@.
63 @chapter GCC Command Options
64 @cindex GCC command options
65 @cindex command options
66 @cindex options, GCC command
68 @c man begin DESCRIPTION
69 When you invoke GCC, it normally does preprocessing, compilation,
70 assembly and linking. The ``overall options'' allow you to stop this
71 process at an intermediate stage. For example, the @option{-c} option
72 says not to run the linker. Then the output consists of object files
73 output by the assembler.
75 Other options are passed on to one stage of processing. Some options
76 control the preprocessor and others the compiler itself. Yet other
77 options control the assembler and linker; most of these are not
78 documented here, since you rarely need to use any of them.
80 @cindex C compilation options
81 Most of the command line options that you can use with GCC are useful
82 for C programs; when an option is only useful with another language
83 (usually C++), the explanation says so explicitly. If the description
84 for a particular option does not mention a source language, you can use
85 that option with all supported languages.
87 @cindex C++ compilation options
88 @xref{Invoking G++,,Compiling C++ Programs}, for a summary of special
89 options for compiling C++ programs.
91 @cindex grouping options
92 @cindex options, grouping
93 The @command{gcc} program accepts options and file names as operands. Many
94 options have multi-letter names; therefore multiple single-letter options
95 may @emph{not} be grouped: @option{-dr} is very different from @w{@samp{-d
98 @cindex order of options
99 @cindex options, order
100 You can mix options and other arguments. For the most part, the order
101 you use doesn't matter. Order does matter when you use several options
102 of the same kind; for example, if you specify @option{-L} more than once,
103 the directories are searched in the order specified.
105 Many options have long names starting with @samp{-f} or with
106 @samp{-W}---for example, @option{-fforce-mem},
107 @option{-fstrength-reduce}, @option{-Wformat} and so on. Most of
108 these have both positive and negative forms; the negative form of
109 @option{-ffoo} would be @option{-fno-foo}. This manual documents
110 only one of these two forms, whichever one is not the default.
114 @xref{Option Index}, for an index to GCC's options.
117 * Option Summary:: Brief list of all options, without explanations.
118 * Overall Options:: Controlling the kind of output:
119 an executable, object files, assembler files,
120 or preprocessed source.
121 * Invoking G++:: Compiling C++ programs.
122 * C Dialect Options:: Controlling the variant of C language compiled.
123 * C++ Dialect Options:: Variations on C++.
124 * Objective-C Dialect Options:: Variations on Objective-C.
125 * Language Independent Options:: Controlling how diagnostics should be
127 * Warning Options:: How picky should the compiler be?
128 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
129 * Optimize Options:: How much optimization?
130 * Preprocessor Options:: Controlling header files and macro definitions.
131 Also, getting dependency information for Make.
132 * Assembler Options:: Passing options to the assembler.
133 * Link Options:: Specifying libraries and so on.
134 * Directory Options:: Where to find header files and libraries.
135 Where to find the compiler executable files.
136 * Spec Files:: How to pass switches to sub-processes.
137 * Target Options:: Running a cross-compiler, or an old version of GCC.
138 * Submodel Options:: Specifying minor hardware or convention variations,
139 such as 68010 vs 68020.
140 * Code Gen Options:: Specifying conventions for function calls, data layout
142 * Environment Variables:: Env vars that affect GCC.
143 * Precompiled Headers:: Compiling a header once, and using it many times.
144 * Running Protoize:: Automatically adding or removing function prototypes.
150 @section Option Summary
152 Here is a summary of all the options, grouped by type. Explanations are
153 in the following sections.
156 @item Overall Options
157 @xref{Overall Options,,Options Controlling the Kind of Output}.
158 @gccoptlist{-c -S -E -o @var{file} -pipe -pass-exit-codes @gol
159 -x @var{language} -v -### --help --target-help --version}
161 @item C Language Options
162 @xref{C Dialect Options,,Options Controlling C Dialect}.
163 @gccoptlist{-ansi -std=@var{standard} -aux-info @var{filename} @gol
164 -fno-asm -fno-builtin -fno-builtin-@var{function} @gol
165 -fhosted -ffreestanding -fms-extensions @gol
166 -trigraphs -no-integrated-cpp -traditional -traditional-cpp @gol
167 -fallow-single-precision -fcond-mismatch @gol
168 -fsigned-bitfields -fsigned-char @gol
169 -funsigned-bitfields -funsigned-char @gol
172 @item C++ Language Options
173 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
174 @gccoptlist{-fabi-version=@var{n} -fno-access-control -fcheck-new @gol
175 -fconserve-space -fno-const-strings @gol
176 -fno-elide-constructors @gol
177 -fno-enforce-eh-specs -fexternal-templates @gol
178 -falt-external-templates @gol
179 -ffor-scope -fno-for-scope -fno-gnu-keywords @gol
180 -fno-implicit-templates @gol
181 -fno-implicit-inline-templates @gol
182 -fno-implement-inlines -fms-extensions @gol
183 -fno-nonansi-builtins -fno-operator-names @gol
184 -fno-optional-diags -fpermissive @gol
185 -frepo -fno-rtti -fstats -ftemplate-depth-@var{n} @gol
186 -fuse-cxa-atexit -fvtable-gc -fno-weak -nostdinc++ @gol
187 -fno-default-inline -Wabi -Wctor-dtor-privacy @gol
188 -Wnon-virtual-dtor -Wreorder @gol
189 -Weffc++ -Wno-deprecated @gol
190 -Wno-non-template-friend -Wold-style-cast @gol
191 -Woverloaded-virtual -Wno-pmf-conversions @gol
192 -Wsign-promo -Wsynth}
194 @item Objective-C Language Options
195 @xref{Objective-C Dialect Options,,Options Controlling Objective-C Dialect}.
196 @gccoptlist{-fconstant-string-class=@var{class-name} @gol
197 -fgnu-runtime -fnext-runtime -gen-decls @gol
198 -Wno-protocol -Wselector -Wundeclared-selector}
200 @item Language Independent Options
201 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
202 @gccoptlist{-fmessage-length=@var{n} @gol
203 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]}}
205 @item Warning Options
206 @xref{Warning Options,,Options to Request or Suppress Warnings}.
207 @gccoptlist{-fsyntax-only -pedantic -pedantic-errors @gol
208 -w -Wextra -Wall -Waggregate-return @gol
209 -Wcast-align -Wcast-qual -Wchar-subscripts -Wcomment @gol
210 -Wconversion -Wno-deprecated-declarations @gol
211 -Wdisabled-optimization -Wno-div-by-zero -Werror @gol
212 -Wfloat-equal -Wformat -Wformat=2 @gol
213 -Wformat-nonliteral -Wformat-security @gol
214 -Wimplicit -Wimplicit-int @gol
215 -Wimplicit-function-declaration @gol
216 -Werror-implicit-function-declaration @gol
217 -Wimport -Winline -Winvalid-pch -Wno-endif-labels @gol
218 -Wno-invalid-offsetof @gol
219 -Wlarger-than-@var{len} -Wlong-long @gol
220 -Wmain -Wmissing-braces @gol
221 -Wmissing-format-attribute -Wmissing-noreturn @gol
222 -Wno-multichar -Wno-format-extra-args -Wno-format-y2k @gol
223 -Wno-import -Wnonnull -Wpacked -Wpadded @gol
224 -Wparentheses -Wpointer-arith -Wredundant-decls @gol
225 -Wreturn-type -Wsequence-point -Wshadow @gol
226 -Wsign-compare -Wstrict-aliasing @gol
227 -Wswitch -Wswitch-default -Wswitch-enum @gol
228 -Wsystem-headers -Wtrigraphs -Wundef -Wuninitialized @gol
229 -Wunknown-pragmas -Wunreachable-code @gol
230 -Wunused -Wunused-function -Wunused-label -Wunused-parameter @gol
231 -Wunused-value -Wunused-variable -Wwrite-strings}
233 @item C-only Warning Options
234 @gccoptlist{-Wbad-function-cast -Wmissing-declarations @gol
235 -Wmissing-prototypes -Wnested-externs @gol
236 -Wstrict-prototypes -Wtraditional}
238 @item Debugging Options
239 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
240 @gccoptlist{-d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
241 -fdump-unnumbered -fdump-translation-unit@r{[}-@var{n}@r{]} @gol
242 -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
243 -fdump-tree-original@r{[}-@var{n}@r{]} @gol
244 -fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
245 -fdump-tree-inlined@r{[}-@var{n}@r{]} @gol
246 -feliminate-dwarf2-dups -feliminate-unused-debug-types @gol
247 -feliminate-unused-debug-symbols -fmem-report -fprofile-arcs @gol
248 -frandom-seed=@var{string} -fsched-verbose=@var{n} @gol
249 -ftest-coverage -ftime-report @gol
250 -g -g@var{level} -gcoff -gdwarf -gdwarf-1 -gdwarf-1+ -gdwarf-2 @gol
251 -ggdb -gstabs -gstabs+ -gvms -gxcoff -gxcoff+ @gol
252 -p -pg -print-file-name=@var{library} -print-libgcc-file-name @gol
253 -print-multi-directory -print-multi-lib @gol
254 -print-prog-name=@var{program} -print-search-dirs -Q @gol
257 @item Optimization Options
258 @xref{Optimize Options,,Options that Control Optimization}.
259 @gccoptlist{-falign-functions=@var{n} -falign-jumps=@var{n} @gol
260 -falign-labels=@var{n} -falign-loops=@var{n} @gol
261 -fbranch-probabilities -fprofile-values -fbranch-target-load-optimize @gol
262 -fbranch-target-load-optimize2 -fcaller-saves -fcprop-registers @gol
263 -fcse-follow-jumps -fcse-skip-blocks -fdata-sections @gol
264 -fdelayed-branch -fdelete-null-pointer-checks @gol
265 -fexpensive-optimizations -ffast-math -ffloat-store @gol
266 -fforce-addr -fforce-mem -ffunction-sections @gol
267 -fgcse -fgcse-lm -fgcse-sm -floop-optimize -fcrossjumping @gol
268 -fif-conversion -fif-conversion2 @gol
269 -finline-functions -finline-limit=@var{n} -fkeep-inline-functions @gol
270 -fkeep-static-consts -fmerge-constants -fmerge-all-constants @gol
271 -fmove-all-movables -fnew-ra -fno-branch-count-reg @gol
272 -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 -ffinite-math-only @gol
276 -fno-trapping-math -fno-zero-initialized-in-bss @gol
277 -fomit-frame-pointer -foptimize-register-move @gol
278 -foptimize-sibling-calls -fprefetch-loop-arrays @gol
279 -freduce-all-givs -fregmove -frename-registers @gol
280 -freorder-blocks -freorder-functions @gol
281 -frerun-cse-after-loop -frerun-loop-opt @gol
282 -fschedule-insns -fschedule-insns2 @gol
283 -fno-sched-interblock -fno-sched-spec -fsched-spec-load @gol
284 -fsched-spec-load-dangerous -fsched2-use-superblocks @gol
285 -fsched2-use-traces -fsignaling-nans @gol
286 -fsingle-precision-constant -fssa -fssa-ccp -fssa-dce @gol
287 -fstrength-reduce -fstrict-aliasing -ftracer -fthread-jumps @gol
288 -funroll-all-loops -funroll-loops -fpeel-loops @gol
289 -funswitch-loops -fold-unroll-loops -fold-unroll-all-loops @gol
290 --param @var{name}=@var{value}
291 -O -O0 -O1 -O2 -O3 -Os}
293 @item Preprocessor Options
294 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
295 @gccoptlist{-A@var{question}=@var{answer} @gol
296 -A-@var{question}@r{[}=@var{answer}@r{]} @gol
297 -C -dD -dI -dM -dN @gol
298 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
299 -idirafter @var{dir} @gol
300 -include @var{file} -imacros @var{file} @gol
301 -iprefix @var{file} -iwithprefix @var{dir} @gol
302 -iwithprefixbefore @var{dir} -isystem @var{dir} @gol
303 -M -MM -MF -MG -MP -MQ -MT -nostdinc -P -remap @gol
304 -trigraphs -undef -U@var{macro} -Wp,@var{option} @gol
305 -Xpreprocessor @var{option}}
307 @item Assembler Option
308 @xref{Assembler Options,,Passing Options to the Assembler}.
309 @gccoptlist{-Wa,@var{option} -Xassembler @var{option}}
312 @xref{Link Options,,Options for Linking}.
313 @gccoptlist{@var{object-file-name} -l@var{library} @gol
314 -nostartfiles -nodefaultlibs -nostdlib -pie @gol
315 -s -static -static-libgcc -shared -shared-libgcc -symbolic @gol
316 -Wl,@var{option} -Xlinker @var{option} @gol
319 @item Directory Options
320 @xref{Directory Options,,Options for Directory Search}.
321 @gccoptlist{-B@var{prefix} -I@var{dir} -I- -L@var{dir} -specs=@var{file}}
324 @c I wrote this xref this way to avoid overfull hbox. -- rms
325 @xref{Target Options}.
326 @gccoptlist{-V @var{version} -b @var{machine}}
328 @item Machine Dependent Options
329 @xref{Submodel Options,,Hardware Models and Configurations}.
331 @emph{M680x0 Options}
332 @gccoptlist{-m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
333 -m68060 -mcpu32 -m5200 -m68881 -mbitfield -mc68000 -mc68020 @gol
334 -mnobitfield -mrtd -mshort -msoft-float -mpcrel @gol
335 -malign-int -mstrict-align}
337 @emph{M68hc1x Options}
338 @gccoptlist{-m6811 -m6812 -m68hc11 -m68hc12 -m68hcs12 @gol
339 -mauto-incdec -minmax -mlong-calls -mshort @gol
340 -msoft-reg-count=@var{count}}
343 @gccoptlist{-mg -mgnu -munix}
346 @gccoptlist{-mcpu=@var{cpu-type} @gol
347 -mtune=@var{cpu-type} @gol
348 -mcmodel=@var{code-model} @gol
350 -mapp-regs -mbroken-saverestore -mcypress @gol
351 -mfaster-structs -mflat @gol
352 -mfpu -mhard-float -mhard-quad-float @gol
353 -mimpure-text -mlive-g0 -mno-app-regs @gol
354 -mno-faster-structs -mno-flat -mno-fpu @gol
355 -mno-impure-text -mno-stack-bias -mno-unaligned-doubles @gol
356 -msoft-float -msoft-quad-float -msparclite -mstack-bias @gol
357 -msupersparc -munaligned-doubles -mv8}
360 @gccoptlist{-mapcs-frame -mno-apcs-frame @gol
361 -mapcs-26 -mapcs-32 @gol
362 -mapcs-stack-check -mno-apcs-stack-check @gol
363 -mapcs-float -mno-apcs-float @gol
364 -mapcs-reentrant -mno-apcs-reentrant @gol
365 -msched-prolog -mno-sched-prolog @gol
366 -mlittle-endian -mbig-endian -mwords-little-endian @gol
367 -malignment-traps -mno-alignment-traps @gol
368 -msoft-float -mhard-float -mfpe @gol
369 -mthumb-interwork -mno-thumb-interwork @gol
370 -mcpu=@var{name} -march=@var{name} -mfpe=@var{name} @gol
371 -mstructure-size-boundary=@var{n} @gol
372 -mabort-on-noreturn @gol
373 -mlong-calls -mno-long-calls @gol
374 -msingle-pic-base -mno-single-pic-base @gol
375 -mpic-register=@var{reg} @gol
376 -mnop-fun-dllimport @gol
377 -mcirrus-fix-invalid-insns -mno-cirrus-fix-invalid-insns @gol
378 -mpoke-function-name @gol
380 -mtpcs-frame -mtpcs-leaf-frame @gol
381 -mcaller-super-interworking -mcallee-super-interworking}
383 @emph{MN10200 Options}
386 @emph{MN10300 Options}
387 @gccoptlist{-mmult-bug -mno-mult-bug @gol
388 -mam33 -mno-am33 @gol
391 @emph{M32R/D Options}
392 @gccoptlist{-m32rx -m32r -mcode-model=@var{model-type} @gol
393 -msdata=@var{sdata-type} -G @var{num}}
396 @gccoptlist{-m88000 -m88100 -m88110 -mbig-pic @gol
397 -mcheck-zero-division -mhandle-large-shift @gol
398 -midentify-revision -mno-check-zero-division @gol
399 -mno-ocs-debug-info -mno-ocs-frame-position @gol
400 -mno-optimize-arg-area -mno-serialize-volatile @gol
401 -mno-underscores -mocs-debug-info @gol
402 -mocs-frame-position -moptimize-arg-area @gol
403 -mserialize-volatile -mshort-data-@var{num} -msvr3 @gol
404 -msvr4 -mtrap-large-shift -muse-div-instruction @gol
405 -mversion-03.00 -mwarn-passed-structs}
407 @emph{RS/6000 and PowerPC Options}
408 @gccoptlist{-mcpu=@var{cpu-type} @gol
409 -mtune=@var{cpu-type} @gol
410 -mpower -mno-power -mpower2 -mno-power2 @gol
411 -mpowerpc -mpowerpc64 -mno-powerpc @gol
412 -maltivec -mno-altivec @gol
413 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
414 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
415 -mnew-mnemonics -mold-mnemonics @gol
416 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
417 -m64 -m32 -mxl-call -mno-xl-call -mpe @gol
418 -malign-power -malign-natural @gol
419 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
420 -mstring -mno-string -mupdate -mno-update @gol
421 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
422 -mstrict-align -mno-strict-align -mrelocatable @gol
423 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
424 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
425 -mdynamic-no-pic @gol
426 -mcall-sysv -mcall-netbsd @gol
427 -maix-struct-return -msvr4-struct-return @gol
428 -mabi=altivec -mabi=no-altivec @gol
429 -mabi=spe -mabi=no-spe @gol
430 -misel=yes -misel=no @gol
431 -mspe=yes -mspe=no @gol
432 -mfloat-gprs=yes -mfloat-gprs=no @gol
433 -mprototype -mno-prototype @gol
434 -msim -mmvme -mads -myellowknife -memb -msdata @gol
435 -msdata=@var{opt} -mvxworks -mwindiss -G @var{num} -pthread}
437 @emph{Darwin Options}
438 @gccoptlist{-all_load -allowable_client -arch -arch_errors_fatal @gol
439 -arch_only -bind_at_load -bundle -bundle_loader @gol
440 -client_name -compatibility_version -current_version @gol
441 -dependency-file -dylib_file -dylinker_install_name @gol
442 -dynamic -dynamiclib -exported_symbols_list @gol
443 -filelist -flat_namespace -force_cpusubtype_ALL @gol
444 -force_flat_namespace -headerpad_max_install_names @gol
445 -image_base -init -install_name -keep_private_externs @gol
446 -multi_module -multiply_defined -multiply_defined_unused @gol
447 -noall_load -nomultidefs -noprebind -noseglinkedit @gol
448 -pagezero_size -prebind -prebind_all_twolevel_modules @gol
449 -private_bundle -read_only_relocs -sectalign @gol
450 -sectobjectsymbols -whyload -seg1addr @gol
451 -sectcreate -sectobjectsymbols -sectorder @gol
452 -seg_addr_table -seg_addr_table_filename -seglinkedit @gol
453 -segprot -segs_read_only_addr -segs_read_write_addr @gol
454 -single_module -static -sub_library -sub_umbrella @gol
455 -twolevel_namespace -umbrella -undefined @gol
456 -unexported_symbols_list -weak_reference_mismatches @gol
460 @gccoptlist{-mcall-lib-mul -mfp-arg-in-fpregs -mfp-arg-in-gregs @gol
461 -mfull-fp-blocks -mhc-struct-return -min-line-mul @gol
462 -mminimum-fp-blocks -mnohc-struct-return}
465 @gccoptlist{-mabicalls -march=@var{cpu-type} -mtune=@var{cpu=type} @gol
466 -mcpu=@var{cpu-type} -membedded-data -muninit-const-in-rodata @gol
467 -membedded-pic -mfp32 -mfp64 -mfused-madd -mno-fused-madd @gol
468 -mgas -mgp32 -mgp64 @gol
469 -mgpopt -mhalf-pic -mhard-float -mint64 -mips1 @gol
470 -mips2 -mips3 -mips4 -mips32 -mips32r2 -mips64 @gol
471 -mlong64 -mlong32 -mlong-calls -mmemcpy @gol
472 -mmips-as -mmips-tfile -mno-abicalls @gol
473 -mno-embedded-data -mno-uninit-const-in-rodata @gol
474 -mno-embedded-pic -mno-gpopt -mno-long-calls @gol
475 -mno-memcpy -mno-mips-tfile -mno-rnames -mno-stats @gol
476 -mrnames -msoft-float @gol
477 -m4650 -msingle-float -mmad @gol
478 -EL -EB -G @var{num} -nocpp @gol
479 -mabi=32 -mabi=n32 -mabi=64 -mabi=eabi -mabi-fake-default @gol
480 -mfix7000 -mno-crt0 -mflush-func=@var{func} -mno-flush-func @gol
481 -mbranch-likely -mno-branch-likely}
483 @emph{i386 and x86-64 Options}
484 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
485 -mfpmath=@var{unit} @gol
486 -masm=@var{dialect} -mno-fancy-math-387 @gol
487 -mno-fp-ret-in-387 -msoft-float -msvr3-shlib @gol
488 -mno-wide-multiply -mrtd -malign-double @gol
489 -mpreferred-stack-boundary=@var{num} @gol
490 -mmmx -msse -msse2 -mpni -m3dnow @gol
491 -mthreads -mno-align-stringops -minline-all-stringops @gol
492 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
493 -m96bit-long-double -mregparm=@var{num} -momit-leaf-frame-pointer @gol
494 -mno-red-zone -mno-tls-direct-seg-refs @gol
495 -mcmodel=@var{code-model} @gol
499 @gccoptlist{-march=@var{architecture-type} @gol
500 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
501 -mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
502 -mjump-in-delay -mlinker-opt -mlong-calls @gol
503 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
504 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
505 -mno-jump-in-delay -mno-long-load-store @gol
506 -mno-portable-runtime -mno-soft-float @gol
507 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
508 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
509 -mschedule=@var{cpu-type} -mspace-regs -msio -mwsio @gol
510 -nolibdld -static -threads}
512 @emph{Intel 960 Options}
513 @gccoptlist{-m@var{cpu-type} -masm-compat -mclean-linkage @gol
514 -mcode-align -mcomplex-addr -mleaf-procedures @gol
515 -mic-compat -mic2.0-compat -mic3.0-compat @gol
516 -mintel-asm -mno-clean-linkage -mno-code-align @gol
517 -mno-complex-addr -mno-leaf-procedures @gol
518 -mno-old-align -mno-strict-align -mno-tail-call @gol
519 -mnumerics -mold-align -msoft-float -mstrict-align @gol
522 @emph{DEC Alpha Options}
523 @gccoptlist{-mno-fp-regs -msoft-float -malpha-as -mgas @gol
524 -mieee -mieee-with-inexact -mieee-conformant @gol
525 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
526 -mtrap-precision=@var{mode} -mbuild-constants @gol
527 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} @gol
528 -mbwx -mmax -mfix -mcix @gol
529 -mfloat-vax -mfloat-ieee @gol
530 -mexplicit-relocs -msmall-data -mlarge-data @gol
531 -msmall-text -mlarge-text @gol
532 -mmemory-latency=@var{time}}
534 @emph{DEC Alpha/VMS Options}
535 @gccoptlist{-mvms-return-codes}
537 @emph{H8/300 Options}
538 @gccoptlist{-mrelax -mh -ms -mn -mint32 -malign-300}
541 @gccoptlist{-m1 -m2 -m2e -m3 -m3e @gol
542 -m4-nofpu -m4-single-only -m4-single -m4 @gol
543 -m5-64media -m5-64media-nofpu @gol
544 -m5-32media -m5-32media-nofpu @gol
545 -m5-compact -m5-compact-nofpu @gol
546 -mb -ml -mdalign -mrelax @gol
547 -mbigtable -mfmovd -mhitachi -mnomacsave @gol
548 -mieee -misize -mpadstruct -mspace @gol
549 -mprefergot -musermode}
551 @emph{System V Options}
552 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
555 @gccoptlist{-EB -EL @gol
556 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
557 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
559 @emph{TMS320C3x/C4x Options}
560 @gccoptlist{-mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm @gol
561 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload @gol
562 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned @gol
563 -mparallel-insns -mparallel-mpy -mpreserve-float}
566 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
567 -mprolog-function -mno-prolog-function -mspace @gol
568 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
569 -mapp-regs -mno-app-regs @gol
570 -mdisable-callt -mno-disable-callt @gol
575 @gccoptlist{-m32032 -m32332 -m32532 -m32081 -m32381 @gol
576 -mmult-add -mnomult-add -msoft-float -mrtd -mnortd @gol
577 -mregparam -mnoregparam -msb -mnosb @gol
578 -mbitfield -mnobitfield -mhimem -mnohimem}
581 @gccoptlist{-mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts @gol
582 -mcall-prologues -mno-tablejump -mtiny-stack}
585 @gccoptlist{-mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
586 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
587 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
588 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
589 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
592 @gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
593 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
594 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
595 -mno-base-addresses -msingle-exit -mno-single-exit}
598 @gccoptlist{-mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
599 -mvolatile-asm-stop -mb-step -mregister-names -mno-sdata @gol
600 -mconstant-gp -mauto-pic -minline-float-divide-min-latency @gol
601 -minline-float-divide-max-throughput @gol
602 -minline-int-divide-min-latency @gol
603 -minline-int-divide-max-throughput -mno-dwarf2-asm @gol
604 -mfixed-range=@var{register-range}}
607 @gccoptlist{-mextmem -mextmemory -monchip -mno-asm-optimize @gol
608 -masm-optimize -mbranch-cost=@var{n} -mcond-exec=@var{n}}
610 @emph{S/390 and zSeries Options}
611 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
612 -mhard-float -msoft-float -mbackchain -mno-backchain @gol
613 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
614 -m64 -m31 -mdebug -mno-debug -mesa -mzarch}
617 @gccoptlist{-mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
618 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
619 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
620 -mstack-align -mdata-align -mconst-align @gol
621 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
622 -melf -maout -melinux -mlinux -sim -sim2}
624 @emph{PDP-11 Options}
625 @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
626 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
627 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
628 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
629 -mbranch-expensive -mbranch-cheap @gol
630 -msplit -mno-split -munix-asm -mdec-asm}
632 @emph{Xstormy16 Options}
635 @emph{Xtensa Options}
636 @gccoptlist{-mbig-endian -mlittle-endian @gol
637 -mdensity -mno-density @gol
638 -mconst16 -mno-const16 @gol
640 -maddx -mno-addx @gol
641 -mmac16 -mno-mac16 @gol
642 -mmul16 -mno-mul16 @gol
643 -mmul32 -mno-mul32 @gol
645 -mminmax -mno-minmax @gol
646 -msext -mno-sext @gol
647 -mbooleans -mno-booleans @gol
648 -mhard-float -msoft-float @gol
649 -mfused-madd -mno-fused-madd @gol
650 -mtext-section-literals -mno-text-section-literals @gol
651 -mtarget-align -mno-target-align @gol
652 -mlongcalls -mno-longcalls}
655 @gccoptlist{-mgpr-32 -mgpr-64 -mfpr-32 -mfpr-64 @gol
656 -mhard-float -msoft-float @gol
657 -malloc-cc -mfixed-cc -mdword -mno-dword @gol
658 -mdouble -mno-double @gol
659 -mmedia -mno-media -mmuladd -mno-muladd @gol
660 -mlibrary-pic -macc-4 -macc-8 @gol
661 -mpack -mno-pack -mno-eflags -mcond-move -mno-cond-move @gol
662 -mscc -mno-scc -mcond-exec -mno-cond-exec @gol
663 -mvliw-branch -mno-vliw-branch @gol
664 -mmulti-cond-exec -mno-multi-cond-exec -mnested-cond-exec @gol
665 -mno-nested-cond-exec -mtomcat-stats @gol
668 @item Code Generation Options
669 @xref{Code Gen Options,,Options for Code Generation Conventions}.
670 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
671 -ffixed-@var{reg} -fexceptions @gol
672 -fnon-call-exceptions -funwind-tables @gol
673 -fasynchronous-unwind-tables @gol
674 -finhibit-size-directive -finstrument-functions @gol
675 -fno-common -fno-ident -fno-gnu-linker @gol
676 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
677 -freg-struct-return -fshared-data -fshort-enums @gol
678 -fshort-double -fshort-wchar @gol
679 -fverbose-asm -fpack-struct -fstack-check @gol
680 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
681 -fargument-alias -fargument-noalias @gol
682 -fargument-noalias-global -fleading-underscore @gol
683 -ftls-model=@var{model} @gol
684 -ftrapv -fwrapv -fbounds-check}
688 * Overall Options:: Controlling the kind of output:
689 an executable, object files, assembler files,
690 or preprocessed source.
691 * C Dialect Options:: Controlling the variant of C language compiled.
692 * C++ Dialect Options:: Variations on C++.
693 * Objective-C Dialect Options:: Variations on Objective-C.
694 * Language Independent Options:: Controlling how diagnostics should be
696 * Warning Options:: How picky should the compiler be?
697 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
698 * Optimize Options:: How much optimization?
699 * Preprocessor Options:: Controlling header files and macro definitions.
700 Also, getting dependency information for Make.
701 * Assembler Options:: Passing options to the assembler.
702 * Link Options:: Specifying libraries and so on.
703 * Directory Options:: Where to find header files and libraries.
704 Where to find the compiler executable files.
705 * Spec Files:: How to pass switches to sub-processes.
706 * Target Options:: Running a cross-compiler, or an old version of GCC.
709 @node Overall Options
710 @section Options Controlling the Kind of Output
712 Compilation can involve up to four stages: preprocessing, compilation
713 proper, assembly and linking, always in that order. The first three
714 stages apply to an individual source file, and end by producing an
715 object file; linking combines all the object files (those newly
716 compiled, and those specified as input) into an executable file.
718 @cindex file name suffix
719 For any given input file, the file name suffix determines what kind of
724 C source code which must be preprocessed.
727 C source code which should not be preprocessed.
730 C++ source code which should not be preprocessed.
733 Objective-C source code. Note that you must link with the library
734 @file{libobjc.a} to make an Objective-C program work.
737 Objective-C source code which should not be preprocessed.
740 C or C++ header file to be turned into a precompiled header.
744 @itemx @var{file}.cxx
745 @itemx @var{file}.cpp
746 @itemx @var{file}.CPP
747 @itemx @var{file}.c++
749 C++ source code which must be preprocessed. Note that in @samp{.cxx},
750 the last two letters must both be literally @samp{x}. Likewise,
751 @samp{.C} refers to a literal capital C@.
755 C++ header file to be turned into a precompiled header.
758 @itemx @var{file}.for
759 @itemx @var{file}.FOR
760 Fortran source code which should not be preprocessed.
763 @itemx @var{file}.fpp
764 @itemx @var{file}.FPP
765 Fortran source code which must be preprocessed (with the traditional
769 Fortran source code which must be preprocessed with a RATFOR
770 preprocessor (not included with GCC)@.
772 @xref{Overall Options,,Options Controlling the Kind of Output, g77,
773 Using and Porting GNU Fortran}, for more details of the handling of
776 @c FIXME: Descriptions of Java file types.
783 Ada source code file which contains a library unit declaration (a
784 declaration of a package, subprogram, or generic, or a generic
785 instantiation), or a library unit renaming declaration (a package,
786 generic, or subprogram renaming declaration). Such files are also
789 @itemx @var{file}.adb
790 Ada source code file containing a library unit body (a subprogram or
791 package body). Such files are also called @dfn{bodies}.
793 @c GCC also knows about some suffixes for languages not yet included:
802 Assembler code which must be preprocessed.
805 An object file to be fed straight into linking.
806 Any file name with no recognized suffix is treated this way.
810 You can specify the input language explicitly with the @option{-x} option:
813 @item -x @var{language}
814 Specify explicitly the @var{language} for the following input files
815 (rather than letting the compiler choose a default based on the file
816 name suffix). This option applies to all following input files until
817 the next @option{-x} option. Possible values for @var{language} are:
819 c c-header cpp-output
820 c++ c++-header c++-cpp-output
821 objective-c objective-c-header objc-cpp-output
822 assembler assembler-with-cpp
824 f77 f77-cpp-input ratfor
830 Turn off any specification of a language, so that subsequent files are
831 handled according to their file name suffixes (as they are if @option{-x}
832 has not been used at all).
834 @item -pass-exit-codes
835 @opindex pass-exit-codes
836 Normally the @command{gcc} program will exit with the code of 1 if any
837 phase of the compiler returns a non-success return code. If you specify
838 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
839 numerically highest error produced by any phase that returned an error
843 If you only want some of the stages of compilation, you can use
844 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
845 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
846 @command{gcc} is to stop. Note that some combinations (for example,
847 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
852 Compile or assemble the source files, but do not link. The linking
853 stage simply is not done. The ultimate output is in the form of an
854 object file for each source file.
856 By default, the object file name for a source file is made by replacing
857 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
859 Unrecognized input files, not requiring compilation or assembly, are
864 Stop after the stage of compilation proper; do not assemble. The output
865 is in the form of an assembler code file for each non-assembler input
868 By default, the assembler file name for a source file is made by
869 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
871 Input files that don't require compilation are ignored.
875 Stop after the preprocessing stage; do not run the compiler proper. The
876 output is in the form of preprocessed source code, which is sent to the
879 Input files which don't require preprocessing are ignored.
881 @cindex output file option
884 Place output in file @var{file}. This applies regardless to whatever
885 sort of output is being produced, whether it be an executable file,
886 an object file, an assembler file or preprocessed C code.
888 Since only one output file can be specified, it does not make sense to
889 use @option{-o} when compiling more than one input file, unless you are
890 producing an executable file as output.
892 If @option{-o} is not specified, the default is to put an executable file
893 in @file{a.out}, the object file for @file{@var{source}.@var{suffix}} in
894 @file{@var{source}.o}, its assembler file in @file{@var{source}.s}, and
895 all preprocessed C source on standard output.
899 Print (on standard error output) the commands executed to run the stages
900 of compilation. Also print the version number of the compiler driver
901 program and of the preprocessor and the compiler proper.
905 Like @option{-v} except the commands are not executed and all command
906 arguments are quoted. This is useful for shell scripts to capture the
907 driver-generated command lines.
911 Use pipes rather than temporary files for communication between the
912 various stages of compilation. This fails to work on some systems where
913 the assembler is unable to read from a pipe; but the GNU assembler has
918 Print (on the standard output) a description of the command line options
919 understood by @command{gcc}. If the @option{-v} option is also specified
920 then @option{--help} will also be passed on to the various processes
921 invoked by @command{gcc}, so that they can display the command line options
922 they accept. If the @option{-Wextra} option is also specified then command
923 line options which have no documentation associated with them will also
928 Print (on the standard output) a description of target specific command
929 line options for each tool.
933 Display the version number and copyrights of the invoked GCC.
937 @section Compiling C++ Programs
939 @cindex suffixes for C++ source
940 @cindex C++ source file suffixes
941 C++ source files conventionally use one of the suffixes @samp{.C},
942 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
943 @samp{.cxx}; C++ header files often use @samp{.hh} or @samp{.H}; and
944 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
945 files with these names and compiles them as C++ programs even if you
946 call the compiler the same way as for compiling C programs (usually
947 with the name @command{gcc}).
951 However, C++ programs often require class libraries as well as a
952 compiler that understands the C++ language---and under some
953 circumstances, you might want to compile programs or header files from
954 standard input, or otherwise without a suffix that flags them as C++
955 programs. You might also like to precompile a C header file with a
956 @samp{.h} extension to be used in C++ compilations. @command{g++} is a
957 program that calls GCC with the default language set to C++, and
958 automatically specifies linking against the C++ library. On many
959 systems, @command{g++} is also installed with the name @command{c++}.
961 @cindex invoking @command{g++}
962 When you compile C++ programs, you may specify many of the same
963 command-line options that you use for compiling programs in any
964 language; or command-line options meaningful for C and related
965 languages; or options that are meaningful only for C++ programs.
966 @xref{C Dialect Options,,Options Controlling C Dialect}, for
967 explanations of options for languages related to C@.
968 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
969 explanations of options that are meaningful only for C++ programs.
971 @node C Dialect Options
972 @section Options Controlling C Dialect
973 @cindex dialect options
974 @cindex language dialect options
975 @cindex options, dialect
977 The following options control the dialect of C (or languages derived
978 from C, such as C++ and Objective-C) that the compiler accepts:
985 In C mode, support all ISO C90 programs. In C++ mode,
986 remove GNU extensions that conflict with ISO C++.
988 This turns off certain features of GCC that are incompatible with ISO
989 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
990 such as the @code{asm} and @code{typeof} keywords, and
991 predefined macros such as @code{unix} and @code{vax} that identify the
992 type of system you are using. It also enables the undesirable and
993 rarely used ISO trigraph feature. For the C compiler,
994 it disables recognition of C++ style @samp{//} comments as well as
995 the @code{inline} keyword.
997 The alternate keywords @code{__asm__}, @code{__extension__},
998 @code{__inline__} and @code{__typeof__} continue to work despite
999 @option{-ansi}. You would not want to use them in an ISO C program, of
1000 course, but it is useful to put them in header files that might be included
1001 in compilations done with @option{-ansi}. Alternate predefined macros
1002 such as @code{__unix__} and @code{__vax__} are also available, with or
1003 without @option{-ansi}.
1005 The @option{-ansi} option does not cause non-ISO programs to be
1006 rejected gratuitously. For that, @option{-pedantic} is required in
1007 addition to @option{-ansi}. @xref{Warning Options}.
1009 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1010 option is used. Some header files may notice this macro and refrain
1011 from declaring certain functions or defining certain macros that the
1012 ISO standard doesn't call for; this is to avoid interfering with any
1013 programs that might use these names for other things.
1015 Functions which would normally be built in but do not have semantics
1016 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1017 functions with @option{-ansi} is used. @xref{Other Builtins,,Other
1018 built-in functions provided by GCC}, for details of the functions
1023 Determine the language standard. This option is currently only
1024 supported when compiling C or C++. A value for this option must be
1025 provided; possible values are
1030 ISO C90 (same as @option{-ansi}).
1032 @item iso9899:199409
1033 ISO C90 as modified in amendment 1.
1039 ISO C99. Note that this standard is not yet fully supported; see
1040 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1041 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1044 Default, ISO C90 plus GNU extensions (including some C99 features).
1048 ISO C99 plus GNU extensions. When ISO C99 is fully implemented in GCC,
1049 this will become the default. The name @samp{gnu9x} is deprecated.
1052 The 1998 ISO C++ standard plus amendments.
1055 The same as @option{-std=c++98} plus GNU extensions. This is the
1056 default for C++ code.
1059 Even when this option is not specified, you can still use some of the
1060 features of newer standards in so far as they do not conflict with
1061 previous C standards. For example, you may use @code{__restrict__} even
1062 when @option{-std=c99} is not specified.
1064 The @option{-std} options specifying some version of ISO C have the same
1065 effects as @option{-ansi}, except that features that were not in ISO C90
1066 but are in the specified version (for example, @samp{//} comments and
1067 the @code{inline} keyword in ISO C99) are not disabled.
1069 @xref{Standards,,Language Standards Supported by GCC}, for details of
1070 these standard versions.
1072 @item -aux-info @var{filename}
1074 Output to the given filename prototyped declarations for all functions
1075 declared and/or defined in a translation unit, including those in header
1076 files. This option is silently ignored in any language other than C@.
1078 Besides declarations, the file indicates, in comments, the origin of
1079 each declaration (source file and line), whether the declaration was
1080 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1081 @samp{O} for old, respectively, in the first character after the line
1082 number and the colon), and whether it came from a declaration or a
1083 definition (@samp{C} or @samp{F}, respectively, in the following
1084 character). In the case of function definitions, a K&R-style list of
1085 arguments followed by their declarations is also provided, inside
1086 comments, after the declaration.
1090 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1091 keyword, so that code can use these words as identifiers. You can use
1092 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1093 instead. @option{-ansi} implies @option{-fno-asm}.
1095 In C++, this switch only affects the @code{typeof} keyword, since
1096 @code{asm} and @code{inline} are standard keywords. You may want to
1097 use the @option{-fno-gnu-keywords} flag instead, which has the same
1098 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1099 switch only affects the @code{asm} and @code{typeof} keywords, since
1100 @code{inline} is a standard keyword in ISO C99.
1103 @itemx -fno-builtin-@var{function}
1104 @opindex fno-builtin
1105 @cindex built-in functions
1106 Don't recognize built-in functions that do not begin with
1107 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1108 functions provided by GCC}, for details of the functions affected,
1109 including those which are not built-in functions when @option{-ansi} or
1110 @option{-std} options for strict ISO C conformance are used because they
1111 do not have an ISO standard meaning.
1113 GCC normally generates special code to handle certain built-in functions
1114 more efficiently; for instance, calls to @code{alloca} may become single
1115 instructions that adjust the stack directly, and calls to @code{memcpy}
1116 may become inline copy loops. The resulting code is often both smaller
1117 and faster, but since the function calls no longer appear as such, you
1118 cannot set a breakpoint on those calls, nor can you change the behavior
1119 of the functions by linking with a different library.
1121 With the @option{-fno-builtin-@var{function}} option
1122 only the built-in function @var{function} is
1123 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1124 function is named this is not built-in in this version of GCC, this
1125 option is ignored. There is no corresponding
1126 @option{-fbuiltin-@var{function}} option; if you wish to enable
1127 built-in functions selectively when using @option{-fno-builtin} or
1128 @option{-ffreestanding}, you may define macros such as:
1131 #define abs(n) __builtin_abs ((n))
1132 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1137 @cindex hosted environment
1139 Assert that compilation takes place in a hosted environment. This implies
1140 @option{-fbuiltin}. A hosted environment is one in which the
1141 entire standard library is available, and in which @code{main} has a return
1142 type of @code{int}. Examples are nearly everything except a kernel.
1143 This is equivalent to @option{-fno-freestanding}.
1145 @item -ffreestanding
1146 @opindex ffreestanding
1147 @cindex hosted environment
1149 Assert that compilation takes place in a freestanding environment. This
1150 implies @option{-fno-builtin}. A freestanding environment
1151 is one in which the standard library may not exist, and program startup may
1152 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1153 This is equivalent to @option{-fno-hosted}.
1155 @xref{Standards,,Language Standards Supported by GCC}, for details of
1156 freestanding and hosted environments.
1158 @item -fms-extensions
1159 @opindex fms-extensions
1160 Accept some non-standard constructs used in Microsoft header files.
1164 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1165 options for strict ISO C conformance) implies @option{-trigraphs}.
1167 @item -no-integrated-cpp
1168 @opindex no-integrated-cpp
1169 Performs a compilation in two passes: preprocessing and compiling. This
1170 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1171 @option{-B} option. The user supplied compilation step can then add in
1172 an additional preprocessing step after normal preprocessing but before
1173 compiling. The default is to use the integrated cpp (internal cpp)
1175 The semantics of this option will change if "cc1", "cc1plus", and
1176 "cc1obj" are merged.
1178 @cindex traditional C language
1179 @cindex C language, traditional
1181 @itemx -traditional-cpp
1182 @opindex traditional-cpp
1183 @opindex traditional
1184 Formerly, these options caused GCC to attempt to emulate a pre-standard
1185 C compiler. They are now only supported with the @option{-E} switch.
1186 The preprocessor continues to support a pre-standard mode. See the GNU
1187 CPP manual for details.
1189 @item -fcond-mismatch
1190 @opindex fcond-mismatch
1191 Allow conditional expressions with mismatched types in the second and
1192 third arguments. The value of such an expression is void. This option
1193 is not supported for C++.
1195 @item -funsigned-char
1196 @opindex funsigned-char
1197 Let the type @code{char} be unsigned, like @code{unsigned char}.
1199 Each kind of machine has a default for what @code{char} should
1200 be. It is either like @code{unsigned char} by default or like
1201 @code{signed char} by default.
1203 Ideally, a portable program should always use @code{signed char} or
1204 @code{unsigned char} when it depends on the signedness of an object.
1205 But many programs have been written to use plain @code{char} and
1206 expect it to be signed, or expect it to be unsigned, depending on the
1207 machines they were written for. This option, and its inverse, let you
1208 make such a program work with the opposite default.
1210 The type @code{char} is always a distinct type from each of
1211 @code{signed char} or @code{unsigned char}, even though its behavior
1212 is always just like one of those two.
1215 @opindex fsigned-char
1216 Let the type @code{char} be signed, like @code{signed char}.
1218 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1219 the negative form of @option{-funsigned-char}. Likewise, the option
1220 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1222 @item -fsigned-bitfields
1223 @itemx -funsigned-bitfields
1224 @itemx -fno-signed-bitfields
1225 @itemx -fno-unsigned-bitfields
1226 @opindex fsigned-bitfields
1227 @opindex funsigned-bitfields
1228 @opindex fno-signed-bitfields
1229 @opindex fno-unsigned-bitfields
1230 These options control whether a bit-field is signed or unsigned, when the
1231 declaration does not use either @code{signed} or @code{unsigned}. By
1232 default, such a bit-field is signed, because this is consistent: the
1233 basic integer types such as @code{int} are signed types.
1235 @item -fwritable-strings
1236 @opindex fwritable-strings
1237 Store string constants in the writable data segment and don't uniquize
1238 them. This is for compatibility with old programs which assume they can
1239 write into string constants.
1241 Writing into string constants is a very bad idea; ``constants'' should
1245 @node C++ Dialect Options
1246 @section Options Controlling C++ Dialect
1248 @cindex compiler options, C++
1249 @cindex C++ options, command line
1250 @cindex options, C++
1251 This section describes the command-line options that are only meaningful
1252 for C++ programs; but you can also use most of the GNU compiler options
1253 regardless of what language your program is in. For example, you
1254 might compile a file @code{firstClass.C} like this:
1257 g++ -g -frepo -O -c firstClass.C
1261 In this example, only @option{-frepo} is an option meant
1262 only for C++ programs; you can use the other options with any
1263 language supported by GCC@.
1265 Here is a list of options that are @emph{only} for compiling C++ programs:
1269 @item -fabi-version=@var{n}
1270 @opindex fabi-version
1271 Use version @var{n} of the C++ ABI. Version 1 is the version of the C++
1272 ABI that first appeared in G++ 3.2. Version 0 will always be the
1273 version that conforms most closely to the C++ ABI specification.
1274 Therefore, the ABI obtained using version 0 will change as ABI bugs are
1277 The default is version 1.
1279 @item -fno-access-control
1280 @opindex fno-access-control
1281 Turn off all access checking. This switch is mainly useful for working
1282 around bugs in the access control code.
1286 Check that the pointer returned by @code{operator new} is non-null
1287 before attempting to modify the storage allocated. This check is
1288 normally unnecessary because the C++ standard specifies that
1289 @code{operator new} will only return @code{0} if it is declared
1290 @samp{throw()}, in which case the compiler will always check the
1291 return value even without this option. In all other cases, when
1292 @code{operator new} has a non-empty exception specification, memory
1293 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1294 @samp{new (nothrow)}.
1296 @item -fconserve-space
1297 @opindex fconserve-space
1298 Put uninitialized or runtime-initialized global variables into the
1299 common segment, as C does. This saves space in the executable at the
1300 cost of not diagnosing duplicate definitions. If you compile with this
1301 flag and your program mysteriously crashes after @code{main()} has
1302 completed, you may have an object that is being destroyed twice because
1303 two definitions were merged.
1305 This option is no longer useful on most targets, now that support has
1306 been added for putting variables into BSS without making them common.
1308 @item -fno-const-strings
1309 @opindex fno-const-strings
1310 Give string constants type @code{char *} instead of type @code{const
1311 char *}. By default, G++ uses type @code{const char *} as required by
1312 the standard. Even if you use @option{-fno-const-strings}, you cannot
1313 actually modify the value of a string constant, unless you also use
1314 @option{-fwritable-strings}.
1316 This option might be removed in a future release of G++. For maximum
1317 portability, you should structure your code so that it works with
1318 string constants that have type @code{const char *}.
1320 @item -fno-elide-constructors
1321 @opindex fno-elide-constructors
1322 The C++ standard allows an implementation to omit creating a temporary
1323 which is only used to initialize another object of the same type.
1324 Specifying this option disables that optimization, and forces G++ to
1325 call the copy constructor in all cases.
1327 @item -fno-enforce-eh-specs
1328 @opindex fno-enforce-eh-specs
1329 Don't check for violation of exception specifications at runtime. This
1330 option violates the C++ standard, but may be useful for reducing code
1331 size in production builds, much like defining @samp{NDEBUG}. The compiler
1332 will still optimize based on the exception specifications.
1334 @item -fexternal-templates
1335 @opindex fexternal-templates
1337 Cause @samp{#pragma interface} and @samp{implementation} to apply to
1338 template instantiation; template instances are emitted or not according
1339 to the location of the template definition. @xref{Template
1340 Instantiation}, for more information.
1342 This option is deprecated.
1344 @item -falt-external-templates
1345 @opindex falt-external-templates
1346 Similar to @option{-fexternal-templates}, but template instances are
1347 emitted or not according to the place where they are first instantiated.
1348 @xref{Template Instantiation}, for more information.
1350 This option is deprecated.
1353 @itemx -fno-for-scope
1355 @opindex fno-for-scope
1356 If @option{-ffor-scope} is specified, the scope of variables declared in
1357 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1358 as specified by the C++ standard.
1359 If @option{-fno-for-scope} is specified, the scope of variables declared in
1360 a @i{for-init-statement} extends to the end of the enclosing scope,
1361 as was the case in old versions of G++, and other (traditional)
1362 implementations of C++.
1364 The default if neither flag is given to follow the standard,
1365 but to allow and give a warning for old-style code that would
1366 otherwise be invalid, or have different behavior.
1368 @item -fno-gnu-keywords
1369 @opindex fno-gnu-keywords
1370 Do not recognize @code{typeof} as a keyword, so that code can use this
1371 word as an identifier. You can use the keyword @code{__typeof__} instead.
1372 @option{-ansi} implies @option{-fno-gnu-keywords}.
1374 @item -fno-implicit-templates
1375 @opindex fno-implicit-templates
1376 Never emit code for non-inline templates which are instantiated
1377 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1378 @xref{Template Instantiation}, for more information.
1380 @item -fno-implicit-inline-templates
1381 @opindex fno-implicit-inline-templates
1382 Don't emit code for implicit instantiations of inline templates, either.
1383 The default is to handle inlines differently so that compiles with and
1384 without optimization will need the same set of explicit instantiations.
1386 @item -fno-implement-inlines
1387 @opindex fno-implement-inlines
1388 To save space, do not emit out-of-line copies of inline functions
1389 controlled by @samp{#pragma implementation}. This will cause linker
1390 errors if these functions are not inlined everywhere they are called.
1392 @item -fms-extensions
1393 @opindex fms-extensions
1394 Disable pedantic warnings about constructs used in MFC, such as implicit
1395 int and getting a pointer to member function via non-standard syntax.
1397 @item -fno-nonansi-builtins
1398 @opindex fno-nonansi-builtins
1399 Disable built-in declarations of functions that are not mandated by
1400 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1401 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1403 @item -fno-operator-names
1404 @opindex fno-operator-names
1405 Do not treat the operator name keywords @code{and}, @code{bitand},
1406 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1407 synonyms as keywords.
1409 @item -fno-optional-diags
1410 @opindex fno-optional-diags
1411 Disable diagnostics that the standard says a compiler does not need to
1412 issue. Currently, the only such diagnostic issued by G++ is the one for
1413 a name having multiple meanings within a class.
1416 @opindex fpermissive
1417 Downgrade messages about nonconformant code from errors to warnings. By
1418 default, G++ effectively sets @option{-pedantic-errors} without
1419 @option{-pedantic}; this option reverses that. This behavior and this
1420 option are superseded by @option{-pedantic}, which works as it does for GNU C@.
1424 Enable automatic template instantiation at link time. This option also
1425 implies @option{-fno-implicit-templates}. @xref{Template
1426 Instantiation}, for more information.
1430 Disable generation of information about every class with virtual
1431 functions for use by the C++ runtime type identification features
1432 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1433 of the language, you can save some space by using this flag. Note that
1434 exception handling uses the same information, but it will generate it as
1439 Emit statistics about front-end processing at the end of the compilation.
1440 This information is generally only useful to the G++ development team.
1442 @item -ftemplate-depth-@var{n}
1443 @opindex ftemplate-depth
1444 Set the maximum instantiation depth for template classes to @var{n}.
1445 A limit on the template instantiation depth is needed to detect
1446 endless recursions during template class instantiation. ANSI/ISO C++
1447 conforming programs must not rely on a maximum depth greater than 17.
1449 @item -fuse-cxa-atexit
1450 @opindex fuse-cxa-atexit
1451 Register destructors for objects with static storage duration with the
1452 @code{__cxa_atexit} function rather than the @code{atexit} function.
1453 This option is required for fully standards-compliant handling of static
1454 destructors, but will only work if your C library supports
1455 @code{__cxa_atexit}.
1459 Emit special relocations for vtables and virtual function references
1460 so that the linker can identify unused virtual functions and zero out
1461 vtable slots that refer to them. This is most useful with
1462 @option{-ffunction-sections} and @option{-Wl,--gc-sections}, in order to
1463 also discard the functions themselves.
1465 This optimization requires GNU as and GNU ld. Not all systems support
1466 this option. @option{-Wl,--gc-sections} is ignored without @option{-static}.
1470 Do not use weak symbol support, even if it is provided by the linker.
1471 By default, G++ will use weak symbols if they are available. This
1472 option exists only for testing, and should not be used by end-users;
1473 it will result in inferior code and has no benefits. This option may
1474 be removed in a future release of G++.
1478 Do not search for header files in the standard directories specific to
1479 C++, but do still search the other standard directories. (This option
1480 is used when building the C++ library.)
1483 In addition, these optimization, warning, and code generation options
1484 have meanings only for C++ programs:
1487 @item -fno-default-inline
1488 @opindex fno-default-inline
1489 Do not assume @samp{inline} for functions defined inside a class scope.
1490 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1491 functions will have linkage like inline functions; they just won't be
1494 @item -Wabi @r{(C++ only)}
1496 Warn when G++ generates code that is probably not compatible with the
1497 vendor-neutral C++ ABI. Although an effort has been made to warn about
1498 all such cases, there are probably some cases that are not warned about,
1499 even though G++ is generating incompatible code. There may also be
1500 cases where warnings are emitted even though the code that is generated
1503 You should rewrite your code to avoid these warnings if you are
1504 concerned about the fact that code generated by G++ may not be binary
1505 compatible with code generated by other compilers.
1507 The known incompatibilities at this point include:
1512 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
1513 pack data into the same byte as a base class. For example:
1516 struct A @{ virtual void f(); int f1 : 1; @};
1517 struct B : public A @{ int f2 : 1; @};
1521 In this case, G++ will place @code{B::f2} into the same byte
1522 as@code{A::f1}; other compilers will not. You can avoid this problem
1523 by explicitly padding @code{A} so that its size is a multiple of the
1524 byte size on your platform; that will cause G++ and other compilers to
1525 layout @code{B} identically.
1528 Incorrect handling of tail-padding for virtual bases. G++ does not use
1529 tail padding when laying out virtual bases. For example:
1532 struct A @{ virtual void f(); char c1; @};
1533 struct B @{ B(); char c2; @};
1534 struct C : public A, public virtual B @{@};
1538 In this case, G++ will not place @code{B} into the tail-padding for
1539 @code{A}; other compilers will. You can avoid this problem by
1540 explicitly padding @code{A} so that its size is a multiple of its
1541 alignment (ignoring virtual base classes); that will cause G++ and other
1542 compilers to layout @code{C} identically.
1545 Incorrect handling of bit-fields with declared widths greater than that
1546 of their underlying types, when the bit-fields appear in a union. For
1550 union U @{ int i : 4096; @};
1554 Assuming that an @code{int} does not have 4096 bits, G++ will make the
1555 union too small by the number of bits in an @code{int}.
1558 Empty classes can be placed at incorrect offsets. For example:
1568 struct C : public B, public A @{@};
1572 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
1573 it should be placed at offset zero. G++ mistakenly believes that the
1574 @code{A} data member of @code{B} is already at offset zero.
1577 Names of template functions whose types involve @code{typename} or
1578 template template parameters can be mangled incorrectly.
1581 template <typename Q>
1582 void f(typename Q::X) @{@}
1584 template <template <typename> class Q>
1585 void f(typename Q<int>::X) @{@}
1589 Instantiations of these templates may be mangled incorrectly.
1593 @item -Wctor-dtor-privacy @r{(C++ only)}
1594 @opindex Wctor-dtor-privacy
1595 Warn when a class seems unusable, because all the constructors or
1596 destructors in a class are private and the class has no friends or
1597 public static member functions.
1599 @item -Wnon-virtual-dtor @r{(C++ only)}
1600 @opindex Wnon-virtual-dtor
1601 Warn when a class declares a non-virtual destructor that should probably
1602 be virtual, because it looks like the class will be used polymorphically.
1603 This warning is enabled by @option{-Wall}.
1605 @item -Wreorder @r{(C++ only)}
1607 @cindex reordering, warning
1608 @cindex warning for reordering of member initializers
1609 Warn when the order of member initializers given in the code does not
1610 match the order in which they must be executed. For instance:
1616 A(): j (0), i (1) @{ @}
1620 Here the compiler will warn that the member initializers for @samp{i}
1621 and @samp{j} will be rearranged to match the declaration order of the
1622 members. This warning is enabled by @option{-Wall}.
1625 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
1628 @item -Weffc++ @r{(C++ only)}
1630 Warn about violations of the following style guidelines from Scott Meyers'
1631 @cite{Effective C++} book:
1635 Item 11: Define a copy constructor and an assignment operator for classes
1636 with dynamically allocated memory.
1639 Item 12: Prefer initialization to assignment in constructors.
1642 Item 14: Make destructors virtual in base classes.
1645 Item 15: Have @code{operator=} return a reference to @code{*this}.
1648 Item 23: Don't try to return a reference when you must return an object.
1652 and about violations of the following style guidelines from Scott Meyers'
1653 @cite{More Effective C++} book:
1657 Item 6: Distinguish between prefix and postfix forms of increment and
1658 decrement operators.
1661 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
1665 If you use this option, you should be aware that the standard library
1666 headers do not obey all of these guidelines; you can use @samp{grep -v}
1667 to filter out those warnings.
1669 @item -Wno-deprecated @r{(C++ only)}
1670 @opindex Wno-deprecated
1671 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1673 @item -Wno-non-template-friend @r{(C++ only)}
1674 @opindex Wno-non-template-friend
1675 Disable warnings when non-templatized friend functions are declared
1676 within a template. With the advent of explicit template specification
1677 support in G++, if the name of the friend is an unqualified-id (i.e.,
1678 @samp{friend foo(int)}), the C++ language specification demands that the
1679 friend declare or define an ordinary, nontemplate function. (Section
1680 14.5.3). Before G++ implemented explicit specification, unqualified-ids
1681 could be interpreted as a particular specialization of a templatized
1682 function. Because this non-conforming behavior is no longer the default
1683 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
1684 check existing code for potential trouble spots, and is on by default.
1685 This new compiler behavior can be turned off with
1686 @option{-Wno-non-template-friend} which keeps the conformant compiler code
1687 but disables the helpful warning.
1689 @item -Wold-style-cast @r{(C++ only)}
1690 @opindex Wold-style-cast
1691 Warn if an old-style (C-style) cast to a non-void type is used within
1692 a C++ program. The new-style casts (@samp{static_cast},
1693 @samp{reinterpret_cast}, and @samp{const_cast}) are less vulnerable to
1694 unintended effects, and much easier to grep for.
1696 @item -Woverloaded-virtual @r{(C++ only)}
1697 @opindex Woverloaded-virtual
1698 @cindex overloaded virtual fn, warning
1699 @cindex warning for overloaded virtual fn
1700 Warn when a function declaration hides virtual functions from a
1701 base class. For example, in:
1708 struct B: public A @{
1713 the @code{A} class version of @code{f} is hidden in @code{B}, and code
1721 will fail to compile.
1723 @item -Wno-pmf-conversions @r{(C++ only)}
1724 @opindex Wno-pmf-conversions
1725 Disable the diagnostic for converting a bound pointer to member function
1728 @item -Wsign-promo @r{(C++ only)}
1729 @opindex Wsign-promo
1730 Warn when overload resolution chooses a promotion from unsigned or
1731 enumeral type to a signed type over a conversion to an unsigned type of
1732 the same size. Previous versions of G++ would try to preserve
1733 unsignedness, but the standard mandates the current behavior.
1735 @item -Wsynth @r{(C++ only)}
1737 @cindex warning for synthesized methods
1738 @cindex synthesized methods, warning
1739 Warn when G++'s synthesis behavior does not match that of cfront. For
1745 A& operator = (int);
1755 In this example, G++ will synthesize a default @samp{A& operator =
1756 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1759 @node Objective-C Dialect Options
1760 @section Options Controlling Objective-C Dialect
1762 @cindex compiler options, Objective-C
1763 @cindex Objective-C options, command line
1764 @cindex options, Objective-C
1765 This section describes the command-line options that are only meaningful
1766 for Objective-C programs; but you can also use most of the GNU compiler
1767 options regardless of what language your program is in. For example,
1768 you might compile a file @code{some_class.m} like this:
1771 gcc -g -fgnu-runtime -O -c some_class.m
1775 In this example, only @option{-fgnu-runtime} is an option meant only for
1776 Objective-C programs; you can use the other options with any language
1779 Here is a list of options that are @emph{only} for compiling Objective-C
1783 @item -fconstant-string-class=@var{class-name}
1784 @opindex fconstant-string-class
1785 Use @var{class-name} as the name of the class to instantiate for each
1786 literal string specified with the syntax @code{@@"@dots{}"}. The default
1787 class name is @code{NXConstantString}.
1790 @opindex fgnu-runtime
1791 Generate object code compatible with the standard GNU Objective-C
1792 runtime. This is the default for most types of systems.
1794 @item -fnext-runtime
1795 @opindex fnext-runtime
1796 Generate output compatible with the NeXT runtime. This is the default
1797 for NeXT-based systems, including Darwin and Mac OS X@. The macro
1798 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
1803 Dump interface declarations for all classes seen in the source file to a
1804 file named @file{@var{sourcename}.decl}.
1807 @opindex Wno-protocol
1808 If a class is declared to implement a protocol, a warning is issued for
1809 every method in the protocol that is not implemented by the class. The
1810 default behavior is to issue a warning for every method not explicitly
1811 implemented in the class, even if a method implementation is inherited
1812 from the superclass. If you use the @code{-Wno-protocol} option, then
1813 methods inherited from the superclass are considered to be implemented,
1814 and no warning is issued for them.
1818 Warn if multiple methods of different types for the same selector are
1819 found during compilation. The check is performed on the list of methods
1820 in the final stage of compilation. Additionally, a check is performed
1821 that for each selector appearing in a @code{@@selector(@dots{})}
1822 expression, a corresponding method with that selector has been found
1823 during compilation. Because these checks scan the method table only at
1824 the end of compilation, these warnings are not produced if the final
1825 stage of compilation is not reached, for example because an error is
1826 found during compilation, or because the @code{-fsyntax-only} option is
1829 @item -Wundeclared-selector
1830 @opindex Wundeclared-selector
1831 Warn if a @code{@@selector(@dots{})} expression referring to an
1832 undeclared selector is found. A selector is considered undeclared if no
1833 method with that name has been declared (explicitly, in an
1834 @code{@@interface} or @code{@@protocol} declaration, or implicitly, in
1835 an @code{@@implementation} section) before the
1836 @code{@@selector(@dots{})} expression. This option always performs its
1837 checks as soon as a @code{@@selector(@dots{})} expression is found
1838 (while @code{-Wselector} only performs its checks in the final stage of
1839 compilation), and so additionally enforces the coding style convention
1840 that methods and selectors must be declared before being used.
1842 @c not documented because only avail via -Wp
1843 @c @item -print-objc-runtime-info
1847 @node Language Independent Options
1848 @section Options to Control Diagnostic Messages Formatting
1849 @cindex options to control diagnostics formatting
1850 @cindex diagnostic messages
1851 @cindex message formatting
1853 Traditionally, diagnostic messages have been formatted irrespective of
1854 the output device's aspect (e.g.@: its width, @dots{}). The options described
1855 below can be used to control the diagnostic messages formatting
1856 algorithm, e.g.@: how many characters per line, how often source location
1857 information should be reported. Right now, only the C++ front end can
1858 honor these options. However it is expected, in the near future, that
1859 the remaining front ends would be able to digest them correctly.
1862 @item -fmessage-length=@var{n}
1863 @opindex fmessage-length
1864 Try to format error messages so that they fit on lines of about @var{n}
1865 characters. The default is 72 characters for @command{g++} and 0 for the rest of
1866 the front ends supported by GCC@. If @var{n} is zero, then no
1867 line-wrapping will be done; each error message will appear on a single
1870 @opindex fdiagnostics-show-location
1871 @item -fdiagnostics-show-location=once
1872 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
1873 reporter to emit @emph{once} source location information; that is, in
1874 case the message is too long to fit on a single physical line and has to
1875 be wrapped, the source location won't be emitted (as prefix) again,
1876 over and over, in subsequent continuation lines. This is the default
1879 @item -fdiagnostics-show-location=every-line
1880 Only meaningful in line-wrapping mode. Instructs the diagnostic
1881 messages reporter to emit the same source location information (as
1882 prefix) for physical lines that result from the process of breaking
1883 a message which is too long to fit on a single line.
1887 @node Warning Options
1888 @section Options to Request or Suppress Warnings
1889 @cindex options to control warnings
1890 @cindex warning messages
1891 @cindex messages, warning
1892 @cindex suppressing warnings
1894 Warnings are diagnostic messages that report constructions which
1895 are not inherently erroneous but which are risky or suggest there
1896 may have been an error.
1898 You can request many specific warnings with options beginning @samp{-W},
1899 for example @option{-Wimplicit} to request warnings on implicit
1900 declarations. Each of these specific warning options also has a
1901 negative form beginning @samp{-Wno-} to turn off warnings;
1902 for example, @option{-Wno-implicit}. This manual lists only one of the
1903 two forms, whichever is not the default.
1905 The following options control the amount and kinds of warnings produced
1906 by GCC; for further, language-specific options also refer to
1907 @ref{C++ Dialect Options} and @ref{Objective-C Dialect Options}.
1910 @cindex syntax checking
1912 @opindex fsyntax-only
1913 Check the code for syntax errors, but don't do anything beyond that.
1917 Issue all the warnings demanded by strict ISO C and ISO C++;
1918 reject all programs that use forbidden extensions, and some other
1919 programs that do not follow ISO C and ISO C++. For ISO C, follows the
1920 version of the ISO C standard specified by any @option{-std} option used.
1922 Valid ISO C and ISO C++ programs should compile properly with or without
1923 this option (though a rare few will require @option{-ansi} or a
1924 @option{-std} option specifying the required version of ISO C)@. However,
1925 without this option, certain GNU extensions and traditional C and C++
1926 features are supported as well. With this option, they are rejected.
1928 @option{-pedantic} does not cause warning messages for use of the
1929 alternate keywords whose names begin and end with @samp{__}. Pedantic
1930 warnings are also disabled in the expression that follows
1931 @code{__extension__}. However, only system header files should use
1932 these escape routes; application programs should avoid them.
1933 @xref{Alternate Keywords}.
1935 Some users try to use @option{-pedantic} to check programs for strict ISO
1936 C conformance. They soon find that it does not do quite what they want:
1937 it finds some non-ISO practices, but not all---only those for which
1938 ISO C @emph{requires} a diagnostic, and some others for which
1939 diagnostics have been added.
1941 A feature to report any failure to conform to ISO C might be useful in
1942 some instances, but would require considerable additional work and would
1943 be quite different from @option{-pedantic}. We don't have plans to
1944 support such a feature in the near future.
1946 Where the standard specified with @option{-std} represents a GNU
1947 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
1948 corresponding @dfn{base standard}, the version of ISO C on which the GNU
1949 extended dialect is based. Warnings from @option{-pedantic} are given
1950 where they are required by the base standard. (It would not make sense
1951 for such warnings to be given only for features not in the specified GNU
1952 C dialect, since by definition the GNU dialects of C include all
1953 features the compiler supports with the given option, and there would be
1954 nothing to warn about.)
1956 @item -pedantic-errors
1957 @opindex pedantic-errors
1958 Like @option{-pedantic}, except that errors are produced rather than
1963 Inhibit all warning messages.
1967 Inhibit warning messages about the use of @samp{#import}.
1969 @item -Wchar-subscripts
1970 @opindex Wchar-subscripts
1971 Warn if an array subscript has type @code{char}. This is a common cause
1972 of error, as programmers often forget that this type is signed on some
1977 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
1978 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
1982 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
1983 the arguments supplied have types appropriate to the format string
1984 specified, and that the conversions specified in the format string make
1985 sense. This includes standard functions, and others specified by format
1986 attributes (@pxref{Function Attributes}), in the @code{printf},
1987 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
1988 not in the C standard) families.
1990 The formats are checked against the format features supported by GNU
1991 libc version 2.2. These include all ISO C90 and C99 features, as well
1992 as features from the Single Unix Specification and some BSD and GNU
1993 extensions. Other library implementations may not support all these
1994 features; GCC does not support warning about features that go beyond a
1995 particular library's limitations. However, if @option{-pedantic} is used
1996 with @option{-Wformat}, warnings will be given about format features not
1997 in the selected standard version (but not for @code{strfmon} formats,
1998 since those are not in any version of the C standard). @xref{C Dialect
1999 Options,,Options Controlling C Dialect}.
2001 Since @option{-Wformat} also checks for null format arguments for
2002 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
2004 @option{-Wformat} is included in @option{-Wall}. For more control over some
2005 aspects of format checking, the options @option{-Wno-format-y2k},
2006 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
2007 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
2008 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
2010 @item -Wno-format-y2k
2011 @opindex Wno-format-y2k
2012 If @option{-Wformat} is specified, do not warn about @code{strftime}
2013 formats which may yield only a two-digit year.
2015 @item -Wno-format-extra-args
2016 @opindex Wno-format-extra-args
2017 If @option{-Wformat} is specified, do not warn about excess arguments to a
2018 @code{printf} or @code{scanf} format function. The C standard specifies
2019 that such arguments are ignored.
2021 Where the unused arguments lie between used arguments that are
2022 specified with @samp{$} operand number specifications, normally
2023 warnings are still given, since the implementation could not know what
2024 type to pass to @code{va_arg} to skip the unused arguments. However,
2025 in the case of @code{scanf} formats, this option will suppress the
2026 warning if the unused arguments are all pointers, since the Single
2027 Unix Specification says that such unused arguments are allowed.
2029 @item -Wno-format-zero-length
2030 @opindex Wno-format-zero-length
2031 If @option{-Wformat} is specified, do not warn about zero-length formats.
2032 The C standard specifies that zero-length formats are allowed.
2034 @item -Wformat-nonliteral
2035 @opindex Wformat-nonliteral
2036 If @option{-Wformat} is specified, also warn if the format string is not a
2037 string literal and so cannot be checked, unless the format function
2038 takes its format arguments as a @code{va_list}.
2040 @item -Wformat-security
2041 @opindex Wformat-security
2042 If @option{-Wformat} is specified, also warn about uses of format
2043 functions that represent possible security problems. At present, this
2044 warns about calls to @code{printf} and @code{scanf} functions where the
2045 format string is not a string literal and there are no format arguments,
2046 as in @code{printf (foo);}. This may be a security hole if the format
2047 string came from untrusted input and contains @samp{%n}. (This is
2048 currently a subset of what @option{-Wformat-nonliteral} warns about, but
2049 in future warnings may be added to @option{-Wformat-security} that are not
2050 included in @option{-Wformat-nonliteral}.)
2054 Enable @option{-Wformat} plus format checks not included in
2055 @option{-Wformat}. Currently equivalent to @samp{-Wformat
2056 -Wformat-nonliteral -Wformat-security}.
2060 Enable warning about passing a null pointer for arguments marked as
2061 requiring a non-null value by the @code{nonnull} function attribute.
2063 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
2064 can be disabled with the @option{-Wno-nonnull} option.
2066 @item -Wimplicit-int
2067 @opindex Wimplicit-int
2068 Warn when a declaration does not specify a type.
2070 @item -Wimplicit-function-declaration
2071 @itemx -Werror-implicit-function-declaration
2072 @opindex Wimplicit-function-declaration
2073 @opindex Werror-implicit-function-declaration
2074 Give a warning (or error) whenever a function is used before being
2079 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
2083 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
2084 function with external linkage, returning int, taking either zero
2085 arguments, two, or three arguments of appropriate types.
2087 @item -Wmissing-braces
2088 @opindex Wmissing-braces
2089 Warn if an aggregate or union initializer is not fully bracketed. In
2090 the following example, the initializer for @samp{a} is not fully
2091 bracketed, but that for @samp{b} is fully bracketed.
2094 int a[2][2] = @{ 0, 1, 2, 3 @};
2095 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
2099 @opindex Wparentheses
2100 Warn if parentheses are omitted in certain contexts, such
2101 as when there is an assignment in a context where a truth value
2102 is expected, or when operators are nested whose precedence people
2103 often get confused about.
2105 Also warn about constructions where there may be confusion to which
2106 @code{if} statement an @code{else} branch belongs. Here is an example of
2121 In C, every @code{else} branch belongs to the innermost possible @code{if}
2122 statement, which in this example is @code{if (b)}. This is often not
2123 what the programmer expected, as illustrated in the above example by
2124 indentation the programmer chose. When there is the potential for this
2125 confusion, GCC will issue a warning when this flag is specified.
2126 To eliminate the warning, add explicit braces around the innermost
2127 @code{if} statement so there is no way the @code{else} could belong to
2128 the enclosing @code{if}. The resulting code would look like this:
2144 @item -Wsequence-point
2145 @opindex Wsequence-point
2146 Warn about code that may have undefined semantics because of violations
2147 of sequence point rules in the C standard.
2149 The C standard defines the order in which expressions in a C program are
2150 evaluated in terms of @dfn{sequence points}, which represent a partial
2151 ordering between the execution of parts of the program: those executed
2152 before the sequence point, and those executed after it. These occur
2153 after the evaluation of a full expression (one which is not part of a
2154 larger expression), after the evaluation of the first operand of a
2155 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
2156 function is called (but after the evaluation of its arguments and the
2157 expression denoting the called function), and in certain other places.
2158 Other than as expressed by the sequence point rules, the order of
2159 evaluation of subexpressions of an expression is not specified. All
2160 these rules describe only a partial order rather than a total order,
2161 since, for example, if two functions are called within one expression
2162 with no sequence point between them, the order in which the functions
2163 are called is not specified. However, the standards committee have
2164 ruled that function calls do not overlap.
2166 It is not specified when between sequence points modifications to the
2167 values of objects take effect. Programs whose behavior depends on this
2168 have undefined behavior; the C standard specifies that ``Between the
2169 previous and next sequence point an object shall have its stored value
2170 modified at most once by the evaluation of an expression. Furthermore,
2171 the prior value shall be read only to determine the value to be
2172 stored.''. If a program breaks these rules, the results on any
2173 particular implementation are entirely unpredictable.
2175 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
2176 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
2177 diagnosed by this option, and it may give an occasional false positive
2178 result, but in general it has been found fairly effective at detecting
2179 this sort of problem in programs.
2181 The present implementation of this option only works for C programs. A
2182 future implementation may also work for C++ programs.
2184 The C standard is worded confusingly, therefore there is some debate
2185 over the precise meaning of the sequence point rules in subtle cases.
2186 Links to discussions of the problem, including proposed formal
2187 definitions, may be found on our readings page, at
2188 @w{@uref{http://gcc.gnu.org/readings.html}}.
2191 @opindex Wreturn-type
2192 Warn whenever a function is defined with a return-type that defaults to
2193 @code{int}. Also warn about any @code{return} statement with no
2194 return-value in a function whose return-type is not @code{void}.
2196 For C++, a function without return type always produces a diagnostic
2197 message, even when @option{-Wno-return-type} is specified. The only
2198 exceptions are @samp{main} and functions defined in system headers.
2202 Warn whenever a @code{switch} statement has an index of enumeral type
2203 and lacks a @code{case} for one or more of the named codes of that
2204 enumeration. (The presence of a @code{default} label prevents this
2205 warning.) @code{case} labels outside the enumeration range also
2206 provoke warnings when this option is used.
2208 @item -Wswitch-default
2209 @opindex Wswitch-switch
2210 Warn whenever a @code{switch} statement does not have a @code{default}
2214 @opindex Wswitch-enum
2215 Warn whenever a @code{switch} statement has an index of enumeral type
2216 and lacks a @code{case} for one or more of the named codes of that
2217 enumeration. @code{case} labels outside the enumeration range also
2218 provoke warnings when this option is used.
2222 Warn if any trigraphs are encountered that might change the meaning of
2223 the program (trigraphs within comments are not warned about).
2225 @item -Wunused-function
2226 @opindex Wunused-function
2227 Warn whenever a static function is declared but not defined or a
2228 non\-inline static function is unused.
2230 @item -Wunused-label
2231 @opindex Wunused-label
2232 Warn whenever a label is declared but not used.
2234 To suppress this warning use the @samp{unused} attribute
2235 (@pxref{Variable Attributes}).
2237 @item -Wunused-parameter
2238 @opindex Wunused-parameter
2239 Warn whenever a function parameter is unused aside from its declaration.
2241 To suppress this warning use the @samp{unused} attribute
2242 (@pxref{Variable Attributes}).
2244 @item -Wunused-variable
2245 @opindex Wunused-variable
2246 Warn whenever a local variable or non-constant static variable is unused
2247 aside from its declaration
2249 To suppress this warning use the @samp{unused} attribute
2250 (@pxref{Variable Attributes}).
2252 @item -Wunused-value
2253 @opindex Wunused-value
2254 Warn whenever a statement computes a result that is explicitly not used.
2256 To suppress this warning cast the expression to @samp{void}.
2260 All the above @option{-Wunused} options combined.
2262 In order to get a warning about an unused function parameter, you must
2263 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
2264 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
2266 @item -Wuninitialized
2267 @opindex Wuninitialized
2268 Warn if an automatic variable is used without first being initialized or
2269 if a variable may be clobbered by a @code{setjmp} call.
2271 These warnings are possible only in optimizing compilation,
2272 because they require data flow information that is computed only
2273 when optimizing. If you don't specify @option{-O}, you simply won't
2276 These warnings occur only for variables that are candidates for
2277 register allocation. Therefore, they do not occur for a variable that
2278 is declared @code{volatile}, or whose address is taken, or whose size
2279 is other than 1, 2, 4 or 8 bytes. Also, they do not occur for
2280 structures, unions or arrays, even when they are in registers.
2282 Note that there may be no warning about a variable that is used only
2283 to compute a value that itself is never used, because such
2284 computations may be deleted by data flow analysis before the warnings
2287 These warnings are made optional because GCC is not smart
2288 enough to see all the reasons why the code might be correct
2289 despite appearing to have an error. Here is one example of how
2310 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
2311 always initialized, but GCC doesn't know this. Here is
2312 another common case:
2317 if (change_y) save_y = y, y = new_y;
2319 if (change_y) y = save_y;
2324 This has no bug because @code{save_y} is used only if it is set.
2326 @cindex @code{longjmp} warnings
2327 This option also warns when a non-volatile automatic variable might be
2328 changed by a call to @code{longjmp}. These warnings as well are possible
2329 only in optimizing compilation.
2331 The compiler sees only the calls to @code{setjmp}. It cannot know
2332 where @code{longjmp} will be called; in fact, a signal handler could
2333 call it at any point in the code. As a result, you may get a warning
2334 even when there is in fact no problem because @code{longjmp} cannot
2335 in fact be called at the place which would cause a problem.
2337 Some spurious warnings can be avoided if you declare all the functions
2338 you use that never return as @code{noreturn}. @xref{Function
2341 @item -Wunknown-pragmas
2342 @opindex Wunknown-pragmas
2343 @cindex warning for unknown pragmas
2344 @cindex unknown pragmas, warning
2345 @cindex pragmas, warning of unknown
2346 Warn when a #pragma directive is encountered which is not understood by
2347 GCC@. If this command line option is used, warnings will even be issued
2348 for unknown pragmas in system header files. This is not the case if
2349 the warnings were only enabled by the @option{-Wall} command line option.
2351 @item -Wstrict-aliasing
2352 @opindex Wstrict-aliasing
2353 This option is only active when @option{-fstrict-aliasing} is active.
2354 It warns about code which might break the strict aliasing rules that the
2355 compiler is using for optimization. The warning does not catch all
2356 cases, but does attempt to catch the more common pitfalls. It is
2357 included in @option{-Wall}.
2361 All of the above @samp{-W} options combined. This enables all the
2362 warnings about constructions that some users consider questionable, and
2363 that are easy to avoid (or modify to prevent the warning), even in
2364 conjunction with macros. This also enables some language-specific
2365 warnings described in @ref{C++ Dialect Options} and
2366 @ref{Objective-C Dialect Options}.
2369 The following @option{-W@dots{}} options are not implied by @option{-Wall}.
2370 Some of them warn about constructions that users generally do not
2371 consider questionable, but which occasionally you might wish to check
2372 for; others warn about constructions that are necessary or hard to avoid
2373 in some cases, and there is no simple way to modify the code to suppress
2380 (This option used to be called @option{-W}. The older name is still
2381 supported, but the newer name is more descriptive.) Print extra warning
2382 messages for these events:
2386 A function can return either with or without a value. (Falling
2387 off the end of the function body is considered returning without
2388 a value.) For example, this function would evoke such a
2402 An expression-statement or the left-hand side of a comma expression
2403 contains no side effects.
2404 To suppress the warning, cast the unused expression to void.
2405 For example, an expression such as @samp{x[i,j]} will cause a warning,
2406 but @samp{x[(void)i,j]} will not.
2409 An unsigned value is compared against zero with @samp{<} or @samp{>=}.
2412 A comparison like @samp{x<=y<=z} appears; this is equivalent to
2413 @samp{(x<=y ? 1 : 0) <= z}, which is a different interpretation from
2414 that of ordinary mathematical notation.
2417 Storage-class specifiers like @code{static} are not the first things in
2418 a declaration. According to the C Standard, this usage is obsolescent.
2421 The return type of a function has a type qualifier such as @code{const}.
2422 Such a type qualifier has no effect, since the value returned by a
2423 function is not an lvalue. (But don't warn about the GNU extension of
2424 @code{volatile void} return types. That extension will be warned about
2425 if @option{-pedantic} is specified.)
2428 If @option{-Wall} or @option{-Wunused} is also specified, warn about unused
2432 A comparison between signed and unsigned values could produce an
2433 incorrect result when the signed value is converted to unsigned.
2434 (But don't warn if @option{-Wno-sign-compare} is also specified.)
2437 An aggregate has an initializer which does not initialize all members.
2438 For example, the following code would cause such a warning, because
2439 @code{x.h} would be implicitly initialized to zero:
2442 struct s @{ int f, g, h; @};
2443 struct s x = @{ 3, 4 @};
2447 A function parameter is declared without a type specifier in K&R-style
2455 An empty body occurs in an @samp{if} or @samp{else} statement.
2458 A pointer is compared against integer zero with @samp{<}, @samp{<=},
2459 @samp{>}, or @samp{>=}.
2462 A variable might be changed by @samp{longjmp} or @samp{vfork}.
2465 Any of several floating-point events that often indicate errors, such as
2466 overflow, underflow, loss of precision, etc.
2468 @item @r{(C++ only)}
2469 An enumerator and a non-enumerator both appear in a conditional expression.
2471 @item @r{(C++ only)}
2472 A non-static reference or non-static @samp{const} member appears in a
2473 class without constructors.
2475 @item @r{(C++ only)}
2476 Ambiguous virtual bases.
2478 @item @r{(C++ only)}
2479 Subscripting an array which has been declared @samp{register}.
2481 @item @r{(C++ only)}
2482 Taking the address of a variable which has been declared @samp{register}.
2484 @item @r{(C++ only)}
2485 A base class is not initialized in a derived class' copy constructor.
2488 @item -Wno-div-by-zero
2489 @opindex Wno-div-by-zero
2490 @opindex Wdiv-by-zero
2491 Do not warn about compile-time integer division by zero. Floating point
2492 division by zero is not warned about, as it can be a legitimate way of
2493 obtaining infinities and NaNs.
2495 @item -Wsystem-headers
2496 @opindex Wsystem-headers
2497 @cindex warnings from system headers
2498 @cindex system headers, warnings from
2499 Print warning messages for constructs found in system header files.
2500 Warnings from system headers are normally suppressed, on the assumption
2501 that they usually do not indicate real problems and would only make the
2502 compiler output harder to read. Using this command line option tells
2503 GCC to emit warnings from system headers as if they occurred in user
2504 code. However, note that using @option{-Wall} in conjunction with this
2505 option will @emph{not} warn about unknown pragmas in system
2506 headers---for that, @option{-Wunknown-pragmas} must also be used.
2509 @opindex Wfloat-equal
2510 Warn if floating point values are used in equality comparisons.
2512 The idea behind this is that sometimes it is convenient (for the
2513 programmer) to consider floating-point values as approximations to
2514 infinitely precise real numbers. If you are doing this, then you need
2515 to compute (by analyzing the code, or in some other way) the maximum or
2516 likely maximum error that the computation introduces, and allow for it
2517 when performing comparisons (and when producing output, but that's a
2518 different problem). In particular, instead of testing for equality, you
2519 would check to see whether the two values have ranges that overlap; and
2520 this is done with the relational operators, so equality comparisons are
2523 @item -Wtraditional @r{(C only)}
2524 @opindex Wtraditional
2525 Warn about certain constructs that behave differently in traditional and
2526 ISO C@. Also warn about ISO C constructs that have no traditional C
2527 equivalent, and/or problematic constructs which should be avoided.
2531 Macro parameters that appear within string literals in the macro body.
2532 In traditional C macro replacement takes place within string literals,
2533 but does not in ISO C@.
2536 In traditional C, some preprocessor directives did not exist.
2537 Traditional preprocessors would only consider a line to be a directive
2538 if the @samp{#} appeared in column 1 on the line. Therefore
2539 @option{-Wtraditional} warns about directives that traditional C
2540 understands but would ignore because the @samp{#} does not appear as the
2541 first character on the line. It also suggests you hide directives like
2542 @samp{#pragma} not understood by traditional C by indenting them. Some
2543 traditional implementations would not recognize @samp{#elif}, so it
2544 suggests avoiding it altogether.
2547 A function-like macro that appears without arguments.
2550 The unary plus operator.
2553 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
2554 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
2555 constants.) Note, these suffixes appear in macros defined in the system
2556 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
2557 Use of these macros in user code might normally lead to spurious
2558 warnings, however gcc's integrated preprocessor has enough context to
2559 avoid warning in these cases.
2562 A function declared external in one block and then used after the end of
2566 A @code{switch} statement has an operand of type @code{long}.
2569 A non-@code{static} function declaration follows a @code{static} one.
2570 This construct is not accepted by some traditional C compilers.
2573 The ISO type of an integer constant has a different width or
2574 signedness from its traditional type. This warning is only issued if
2575 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
2576 typically represent bit patterns, are not warned about.
2579 Usage of ISO string concatenation is detected.
2582 Initialization of automatic aggregates.
2585 Identifier conflicts with labels. Traditional C lacks a separate
2586 namespace for labels.
2589 Initialization of unions. If the initializer is zero, the warning is
2590 omitted. This is done under the assumption that the zero initializer in
2591 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
2592 initializer warnings and relies on default initialization to zero in the
2596 Conversions by prototypes between fixed/floating point values and vice
2597 versa. The absence of these prototypes when compiling with traditional
2598 C would cause serious problems. This is a subset of the possible
2599 conversion warnings, for the full set use @option{-Wconversion}.
2602 Use of ISO C style function definitions. This warning intentionally is
2603 @emph{not} issued for prototype declarations or variadic functions
2604 because these ISO C features will appear in your code when using
2605 libiberty's traditional C compatibility macros, @code{PARAMS} and
2606 @code{VPARAMS}. This warning is also bypassed for nested functions
2607 because that feature is already a gcc extension and thus not relevant to
2608 traditional C compatibility.
2613 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
2615 @item -Wendif-labels
2616 @opindex Wendif-labels
2617 Warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
2621 Warn whenever a local variable shadows another local variable, parameter or
2622 global variable or whenever a built-in function is shadowed.
2624 @item -Wlarger-than-@var{len}
2625 @opindex Wlarger-than
2626 Warn whenever an object of larger than @var{len} bytes is defined.
2628 @item -Wpointer-arith
2629 @opindex Wpointer-arith
2630 Warn about anything that depends on the ``size of'' a function type or
2631 of @code{void}. GNU C assigns these types a size of 1, for
2632 convenience in calculations with @code{void *} pointers and pointers
2635 @item -Wbad-function-cast @r{(C only)}
2636 @opindex Wbad-function-cast
2637 Warn whenever a function call is cast to a non-matching type.
2638 For example, warn if @code{int malloc()} is cast to @code{anything *}.
2642 Warn whenever a pointer is cast so as to remove a type qualifier from
2643 the target type. For example, warn if a @code{const char *} is cast
2644 to an ordinary @code{char *}.
2647 @opindex Wcast-align
2648 Warn whenever a pointer is cast such that the required alignment of the
2649 target is increased. For example, warn if a @code{char *} is cast to
2650 an @code{int *} on machines where integers can only be accessed at
2651 two- or four-byte boundaries.
2653 @item -Wwrite-strings
2654 @opindex Wwrite-strings
2655 When compiling C, give string constants the type @code{const
2656 char[@var{length}]} so that
2657 copying the address of one into a non-@code{const} @code{char *}
2658 pointer will get a warning; when compiling C++, warn about the
2659 deprecated conversion from string constants to @code{char *}.
2660 These warnings will help you find at
2661 compile time code that can try to write into a string constant, but
2662 only if you have been very careful about using @code{const} in
2663 declarations and prototypes. Otherwise, it will just be a nuisance;
2664 this is why we did not make @option{-Wall} request these warnings.
2667 @opindex Wconversion
2668 Warn if a prototype causes a type conversion that is different from what
2669 would happen to the same argument in the absence of a prototype. This
2670 includes conversions of fixed point to floating and vice versa, and
2671 conversions changing the width or signedness of a fixed point argument
2672 except when the same as the default promotion.
2674 Also, warn if a negative integer constant expression is implicitly
2675 converted to an unsigned type. For example, warn about the assignment
2676 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
2677 casts like @code{(unsigned) -1}.
2679 @item -Wsign-compare
2680 @opindex Wsign-compare
2681 @cindex warning for comparison of signed and unsigned values
2682 @cindex comparison of signed and unsigned values, warning
2683 @cindex signed and unsigned values, comparison warning
2684 Warn when a comparison between signed and unsigned values could produce
2685 an incorrect result when the signed value is converted to unsigned.
2686 This warning is also enabled by @option{-Wextra}; to get the other warnings
2687 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
2689 @item -Waggregate-return
2690 @opindex Waggregate-return
2691 Warn if any functions that return structures or unions are defined or
2692 called. (In languages where you can return an array, this also elicits
2695 @item -Wstrict-prototypes @r{(C only)}
2696 @opindex Wstrict-prototypes
2697 Warn if a function is declared or defined without specifying the
2698 argument types. (An old-style function definition is permitted without
2699 a warning if preceded by a declaration which specifies the argument
2702 @item -Wmissing-prototypes @r{(C only)}
2703 @opindex Wmissing-prototypes
2704 Warn if a global function is defined without a previous prototype
2705 declaration. This warning is issued even if the definition itself
2706 provides a prototype. The aim is to detect global functions that fail
2707 to be declared in header files.
2709 @item -Wmissing-declarations @r{(C only)}
2710 @opindex Wmissing-declarations
2711 Warn if a global function is defined without a previous declaration.
2712 Do so even if the definition itself provides a prototype.
2713 Use this option to detect global functions that are not declared in
2716 @item -Wmissing-noreturn
2717 @opindex Wmissing-noreturn
2718 Warn about functions which might be candidates for attribute @code{noreturn}.
2719 Note these are only possible candidates, not absolute ones. Care should
2720 be taken to manually verify functions actually do not ever return before
2721 adding the @code{noreturn} attribute, otherwise subtle code generation
2722 bugs could be introduced. You will not get a warning for @code{main} in
2723 hosted C environments.
2725 @item -Wmissing-format-attribute
2726 @opindex Wmissing-format-attribute
2728 If @option{-Wformat} is enabled, also warn about functions which might be
2729 candidates for @code{format} attributes. Note these are only possible
2730 candidates, not absolute ones. GCC will guess that @code{format}
2731 attributes might be appropriate for any function that calls a function
2732 like @code{vprintf} or @code{vscanf}, but this might not always be the
2733 case, and some functions for which @code{format} attributes are
2734 appropriate may not be detected. This option has no effect unless
2735 @option{-Wformat} is enabled (possibly by @option{-Wall}).
2737 @item -Wno-multichar
2738 @opindex Wno-multichar
2740 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
2741 Usually they indicate a typo in the user's code, as they have
2742 implementation-defined values, and should not be used in portable code.
2744 @item -Wno-deprecated-declarations
2745 @opindex Wno-deprecated-declarations
2746 Do not warn about uses of functions, variables, and types marked as
2747 deprecated by using the @code{deprecated} attribute.
2748 (@pxref{Function Attributes}, @pxref{Variable Attributes},
2749 @pxref{Type Attributes}.)
2753 Warn if a structure is given the packed attribute, but the packed
2754 attribute has no effect on the layout or size of the structure.
2755 Such structures may be mis-aligned for little benefit. For
2756 instance, in this code, the variable @code{f.x} in @code{struct bar}
2757 will be misaligned even though @code{struct bar} does not itself
2758 have the packed attribute:
2765 @} __attribute__((packed));
2775 Warn if padding is included in a structure, either to align an element
2776 of the structure or to align the whole structure. Sometimes when this
2777 happens it is possible to rearrange the fields of the structure to
2778 reduce the padding and so make the structure smaller.
2780 @item -Wredundant-decls
2781 @opindex Wredundant-decls
2782 Warn if anything is declared more than once in the same scope, even in
2783 cases where multiple declaration is valid and changes nothing.
2785 @item -Wnested-externs @r{(C only)}
2786 @opindex Wnested-externs
2787 Warn if an @code{extern} declaration is encountered within a function.
2789 @item -Wunreachable-code
2790 @opindex Wunreachable-code
2791 Warn if the compiler detects that code will never be executed.
2793 This option is intended to warn when the compiler detects that at
2794 least a whole line of source code will never be executed, because
2795 some condition is never satisfied or because it is after a
2796 procedure that never returns.
2798 It is possible for this option to produce a warning even though there
2799 are circumstances under which part of the affected line can be executed,
2800 so care should be taken when removing apparently-unreachable code.
2802 For instance, when a function is inlined, a warning may mean that the
2803 line is unreachable in only one inlined copy of the function.
2805 This option is not made part of @option{-Wall} because in a debugging
2806 version of a program there is often substantial code which checks
2807 correct functioning of the program and is, hopefully, unreachable
2808 because the program does work. Another common use of unreachable
2809 code is to provide behavior which is selectable at compile-time.
2813 Warn if a function can not be inlined and it was declared as inline.
2814 Even with this option, the compiler will not warn about failures to
2815 inline functions declared in system headers.
2817 The compiler uses a variety of heuristics to determine whether or not
2818 to inline a function. For example, the compiler takes into account
2819 the size of the function being inlined and the the amount of inlining
2820 that has already been done in the current function. Therefore,
2821 seemingly insignificant changes in the source program can cause the
2822 warnings produced by @option{-Winline} to appear or disappear.
2824 @item -Wno-invalid-offsetof @r{(C++ only)}
2825 @opindex Wno-invalid-offsetof
2826 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
2827 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
2828 to a non-POD type is undefined. In existing C++ implementations,
2829 however, @samp{offsetof} typically gives meaningful results even when
2830 applied to certain kinds of non-POD types. (Such as a simple
2831 @samp{struct} that fails to be a POD type only by virtue of having a
2832 constructor.) This flag is for users who are aware that they are
2833 writing nonportable code and who have deliberately chosen to ignore the
2836 The restrictions on @samp{offsetof} may be relaxed in a future version
2837 of the C++ standard.
2840 @opindex Winvalid-pch
2841 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
2842 the search path but can't be used.
2846 @opindex Wno-long-long
2847 Warn if @samp{long long} type is used. This is default. To inhibit
2848 the warning messages, use @option{-Wno-long-long}. Flags
2849 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
2850 only when @option{-pedantic} flag is used.
2852 @item -Wdisabled-optimization
2853 @opindex Wdisabled-optimization
2854 Warn if a requested optimization pass is disabled. This warning does
2855 not generally indicate that there is anything wrong with your code; it
2856 merely indicates that GCC's optimizers were unable to handle the code
2857 effectively. Often, the problem is that your code is too big or too
2858 complex; GCC will refuse to optimize programs when the optimization
2859 itself is likely to take inordinate amounts of time.
2863 Make all warnings into errors.
2866 @node Debugging Options
2867 @section Options for Debugging Your Program or GCC
2868 @cindex options, debugging
2869 @cindex debugging information options
2871 GCC has various special options that are used for debugging
2872 either your program or GCC:
2877 Produce debugging information in the operating system's native format
2878 (stabs, COFF, XCOFF, or DWARF)@. GDB can work with this debugging
2881 On most systems that use stabs format, @option{-g} enables use of extra
2882 debugging information that only GDB can use; this extra information
2883 makes debugging work better in GDB but will probably make other debuggers
2885 refuse to read the program. If you want to control for certain whether
2886 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
2887 @option{-gxcoff+}, @option{-gxcoff}, @option{-gdwarf-1+}, @option{-gdwarf-1},
2888 or @option{-gvms} (see below).
2890 Unlike most other C compilers, GCC allows you to use @option{-g} with
2891 @option{-O}. The shortcuts taken by optimized code may occasionally
2892 produce surprising results: some variables you declared may not exist
2893 at all; flow of control may briefly move where you did not expect it;
2894 some statements may not be executed because they compute constant
2895 results or their values were already at hand; some statements may
2896 execute in different places because they were moved out of loops.
2898 Nevertheless it proves possible to debug optimized output. This makes
2899 it reasonable to use the optimizer for programs that might have bugs.
2901 The following options are useful when GCC is generated with the
2902 capability for more than one debugging format.
2906 Produce debugging information for use by GDB@. This means to use the
2907 most expressive format available (DWARF 2, stabs, or the native format
2908 if neither of those are supported), including GDB extensions if at all
2913 Produce debugging information in stabs format (if that is supported),
2914 without GDB extensions. This is the format used by DBX on most BSD
2915 systems. On MIPS, Alpha and System V Release 4 systems this option
2916 produces stabs debugging output which is not understood by DBX or SDB@.
2917 On System V Release 4 systems this option requires the GNU assembler.
2919 @item -feliminate-unused-debug-symbols
2920 @opindex feliminate-unused-debug-symbols
2921 Produce debugging information in stabs format (if that is supported),
2922 for only symbols that are actually used.
2926 Produce debugging information in stabs format (if that is supported),
2927 using GNU extensions understood only by the GNU debugger (GDB)@. The
2928 use of these extensions is likely to make other debuggers crash or
2929 refuse to read the program.
2933 Produce debugging information in COFF format (if that is supported).
2934 This is the format used by SDB on most System V systems prior to
2939 Produce debugging information in XCOFF format (if that is supported).
2940 This is the format used by the DBX debugger on IBM RS/6000 systems.
2944 Produce debugging information in XCOFF format (if that is supported),
2945 using GNU extensions understood only by the GNU debugger (GDB)@. The
2946 use of these extensions is likely to make other debuggers crash or
2947 refuse to read the program, and may cause assemblers other than the GNU
2948 assembler (GAS) to fail with an error.
2952 Produce debugging information in DWARF version 1 format (if that is
2953 supported). This is the format used by SDB on most System V Release 4
2956 This option is deprecated.
2960 Produce debugging information in DWARF version 1 format (if that is
2961 supported), using GNU extensions understood only by the GNU debugger
2962 (GDB)@. The use of these extensions is likely to make other debuggers
2963 crash or refuse to read the program.
2965 This option is deprecated.
2969 Produce debugging information in DWARF version 2 format (if that is
2970 supported). This is the format used by DBX on IRIX 6.
2974 Produce debugging information in VMS debug format (if that is
2975 supported). This is the format used by DEBUG on VMS systems.
2978 @itemx -ggdb@var{level}
2979 @itemx -gstabs@var{level}
2980 @itemx -gcoff@var{level}
2981 @itemx -gxcoff@var{level}
2982 @itemx -gvms@var{level}
2983 Request debugging information and also use @var{level} to specify how
2984 much information. The default level is 2.
2986 Level 1 produces minimal information, enough for making backtraces in
2987 parts of the program that you don't plan to debug. This includes
2988 descriptions of functions and external variables, but no information
2989 about local variables and no line numbers.
2991 Level 3 includes extra information, such as all the macro definitions
2992 present in the program. Some debuggers support macro expansion when
2993 you use @option{-g3}.
2995 Note that in order to avoid confusion between DWARF1 debug level 2,
2996 and DWARF2, neither @option{-gdwarf} nor @option{-gdwarf-2} accept
2997 a concatenated debug level. Instead use an additional @option{-g@var{level}}
2998 option to change the debug level for DWARF1 or DWARF2.
3000 @item -feliminate-dwarf2-dups
3001 @opindex feliminate-dwarf2-dups
3002 Compress DWARF2 debugging information by eliminating duplicated
3003 information about each symbol. This option only makes sense when
3004 generating DWARF2 debugging information with @option{-gdwarf-2}.
3006 @cindex @command{prof}
3009 Generate extra code to write profile information suitable for the
3010 analysis program @command{prof}. You must use this option when compiling
3011 the source files you want data about, and you must also use it when
3014 @cindex @command{gprof}
3017 Generate extra code to write profile information suitable for the
3018 analysis program @command{gprof}. You must use this option when compiling
3019 the source files you want data about, and you must also use it when
3024 Makes the compiler print out each function name as it is compiled, and
3025 print some statistics about each pass when it finishes.
3028 @opindex ftime-report
3029 Makes the compiler print some statistics about the time consumed by each
3030 pass when it finishes.
3033 @opindex fmem-report
3034 Makes the compiler print some statistics about permanent memory
3035 allocation when it finishes.
3037 @item -fprofile-arcs
3038 @opindex fprofile-arcs
3039 Add code so that program flow @dfn{arcs} are instrumented. During
3040 execution the program records how many times each branch and call is
3041 executed and how many times it is taken or returns. When the compiled
3042 program exits it saves this data to a file called
3043 @file{@var{auxname}.da} for each source file. The data may be used for
3044 profile-directed optimizations (@option{-fbranch-probabilities}), or for
3045 test coverage analysis (@option{-ftest-coverage}). Each object file's
3046 @var{auxname} is generated from the name of the output file, if
3047 explicitly specified and it is not the final executable, otherwise it is
3048 the basename of the source file. In both cases any suffix is removed
3049 (e.g. @file{foo.da} for input file @file{dir/foo.c}, or
3050 @file{dir/foo.da} for output file specified as @option{-o dir/foo.o}).
3055 Compile the source files with @option{-fprofile-arcs} plus optimization
3056 and code generation options. For test coverage analysis, use the
3057 additional @option{-ftest-coverage} option. You do not need to profile
3058 every source file in a program.
3061 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
3062 (the latter implies the former).
3065 Run the program on a representative workload to generate the arc profile
3066 information. This may be repeated any number of times. You can run
3067 concurrent instances of your program, and provided that the file system
3068 supports locking, the data files will be correctly updated. Also
3069 @code{fork} calls are detected and correctly handled (double counting
3073 For profile-directed optimizations, compile the source files again with
3074 the same optimization and code generation options plus
3075 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
3076 Control Optimization}).
3079 For test coverage analysis, use @command{gcov} to produce human readable
3080 information from the @file{.bbg} and @file{.da} files. Refer to the
3081 @command{gcov} documentation for further information.
3085 With @option{-fprofile-arcs}, for each function of your program GCC
3086 creates a program flow graph, then finds a spanning tree for the graph.
3087 Only arcs that are not on the spanning tree have to be instrumented: the
3088 compiler adds code to count the number of times that these arcs are
3089 executed. When an arc is the only exit or only entrance to a block, the
3090 instrumentation code can be added to the block; otherwise, a new basic
3091 block must be created to hold the instrumentation code.
3094 @item -ftest-coverage
3095 @opindex ftest-coverage
3096 Produce a graph file that the @command{gcov} code-coverage utility
3097 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
3098 show program coverage. Each source file's data file is called
3099 @file{@var{auxname}.bbg}. Refer to the @option{-fprofile-arcs} option
3100 above for a description of @var{auxname} and instructions on how to
3101 generate test coverage data. Coverage data will match the source files
3102 more closely, if you do not optimize.
3104 @item -d@var{letters}
3106 Says to make debugging dumps during compilation at times specified by
3107 @var{letters}. This is used for debugging the compiler. The file names
3108 for most of the dumps are made by appending a pass number and a word to
3109 the @var{dumpname}. @var{dumpname} is generated from the name of the
3110 output file, if explicitly specified and it is not an executable,
3111 otherwise it is the basename of the source file. In both cases any
3112 suffix is removed (e.g. @file{foo.00.rtl} or @file{foo.01.sibling}).
3113 Here are the possible letters for use in @var{letters}, and their
3119 Annotate the assembler output with miscellaneous debugging information.
3122 Dump after computing branch probabilities, to @file{@var{file}.15.bp}.
3125 Dump after block reordering, to @file{@var{file}.31.bbro}.
3128 Dump after instruction combination, to the file @file{@var{file}.21.combine}.
3131 Dump after the first if conversion, to the file @file{@var{file}.16.ce1}.
3132 Also dump after the second if conversion, to the file @file{@var{file}.22.ce2}.
3135 Dump after branch target load optimization, to to @file{@var{file}.33.btl}.
3136 Also dump after delayed branch scheduling, to @file{@var{file}.36.dbr}.
3139 Dump all macro definitions, at the end of preprocessing, in addition to
3143 Dump after SSA optimizations, to @file{@var{file}.04.ssa} and
3144 @file{@var{file}.07.ussa}.
3147 Dump after the second if conversion, to @file{@var{file}.32.ce3}.
3150 Dump after control and data flow analysis, to @file{@var{file}.14.cfg}.
3151 Also dump after life analysis, to @file{@var{file}.20.life}.
3154 Dump after purging @code{ADDRESSOF} codes, to @file{@var{file}.10.addressof}.
3157 Dump after global register allocation, to @file{@var{file}.26.greg}.
3160 Dump after GCSE, to @file{@var{file}.11.gcse}.
3161 Also dump after jump bypassing and control flow optimizations, to
3162 @file{@var{file}.13.bypass}.
3165 Dump after finalization of EH handling code, to @file{@var{file}.02.eh}.
3168 Dump after sibling call optimizations, to @file{@var{file}.01.sibling}.
3171 Dump after the first jump optimization, to @file{@var{file}.03.jump}.
3174 Dump after conversion from registers to stack, to @file{@var{file}.35.stack}.
3177 Dump after local register allocation, to @file{@var{file}.25.lreg}.
3180 Dump after loop optimization passes, to @file{@var{file}.12.loop} and
3181 @file{@var{file}.18.loop2}.
3184 Dump after performing the machine dependent reorganization pass, to
3185 @file{@var{file}.36.mach}.
3188 Dump after register renumbering, to @file{@var{file}.30.rnreg}.
3191 Dump after the register move pass, to @file{@var{file}.23.regmove}.
3194 Dump after post-reload optimizations, to @file{@var{file}.27.postreload}.
3197 Dump after RTL generation, to @file{@var{file}.00.rtl}.
3200 Dump after the second scheduling pass, to @file{@var{file}.34.sched2}.
3203 Dump after CSE (including the jump optimization that sometimes follows
3204 CSE), to @file{@var{file}.09.cse}.
3207 Dump after the first scheduling pass, to @file{@var{file}.24.sched}.
3210 Dump after the second CSE pass (including the jump optimization that
3211 sometimes follows CSE), to @file{@var{file}.19.cse2}.
3214 Dump after running tracer, to @file{@var{file}.17.tracer}.
3217 Dump after null pointer elimination pass to @file{@var{file}.08.null}.
3220 Dump after the second flow pass, to @file{@var{file}.28.flow2}.
3223 Dump after SSA conditional constant propagation, to
3224 @file{@var{file}.05.ssaccp}.
3227 Dump after SSA dead code elimination, to @file{@var{file}.06.ssadce}.
3230 Dump after the peephole pass, to @file{@var{file}.29.peephole2}.
3233 Produce all the dumps listed above.
3236 Produce a core dump whenever an error occurs.
3239 Print statistics on memory usage, at the end of the run, to
3243 Annotate the assembler output with a comment indicating which
3244 pattern and alternative was used. The length of each instruction is
3248 Dump the RTL in the assembler output as a comment before each instruction.
3249 Also turns on @option{-dp} annotation.
3252 For each of the other indicated dump files (except for
3253 @file{@var{file}.00.rtl}), dump a representation of the control flow graph
3254 suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
3257 Just generate RTL for a function instead of compiling it. Usually used
3261 Dump debugging information during parsing, to standard error.
3264 @item -fdump-unnumbered
3265 @opindex fdump-unnumbered
3266 When doing debugging dumps (see @option{-d} option above), suppress instruction
3267 numbers and line number note output. This makes it more feasible to
3268 use diff on debugging dumps for compiler invocations with different
3269 options, in particular with and without @option{-g}.
3271 @item -fdump-translation-unit @r{(C and C++ only)}
3272 @itemx -fdump-translation-unit-@var{options} @r{(C and C++ only)}
3273 @opindex fdump-translation-unit
3274 Dump a representation of the tree structure for the entire translation
3275 unit to a file. The file name is made by appending @file{.tu} to the
3276 source file name. If the @samp{-@var{options}} form is used, @var{options}
3277 controls the details of the dump as described for the
3278 @option{-fdump-tree} options.
3280 @item -fdump-class-hierarchy @r{(C++ only)}
3281 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
3282 @opindex fdump-class-hierarchy
3283 Dump a representation of each class's hierarchy and virtual function
3284 table layout to a file. The file name is made by appending @file{.class}
3285 to the source file name. If the @samp{-@var{options}} form is used,
3286 @var{options} controls the details of the dump as described for the
3287 @option{-fdump-tree} options.
3289 @item -fdump-tree-@var{switch} @r{(C++ only)}
3290 @itemx -fdump-tree-@var{switch}-@var{options} @r{(C++ only)}
3292 Control the dumping at various stages of processing the intermediate
3293 language tree to a file. The file name is generated by appending a switch
3294 specific suffix to the source file name. If the @samp{-@var{options}}
3295 form is used, @var{options} is a list of @samp{-} separated options that
3296 control the details of the dump. Not all options are applicable to all
3297 dumps, those which are not meaningful will be ignored. The following
3298 options are available
3302 Print the address of each node. Usually this is not meaningful as it
3303 changes according to the environment and source file. Its primary use
3304 is for tying up a dump file with a debug environment.
3306 Inhibit dumping of members of a scope or body of a function merely
3307 because that scope has been reached. Only dump such items when they
3308 are directly reachable by some other path.
3310 Turn on all options.
3313 The following tree dumps are possible:
3316 Dump before any tree based optimization, to @file{@var{file}.original}.
3318 Dump after all tree based optimization, to @file{@var{file}.optimized}.
3320 Dump after function inlining, to @file{@var{file}.inlined}.
3323 @item -frandom-seed=@var{string}
3324 @opindex frandom-string
3325 This option provides a seed that GCC uses when it would otherwise use
3326 random numbers. At present, this is used to generate certain symbol names
3327 that have to be different in every compiled file.
3329 The @var{string} should be different for every file you compile.
3331 @item -fsched-verbose=@var{n}
3332 @opindex fsched-verbose
3333 On targets that use instruction scheduling, this option controls the
3334 amount of debugging output the scheduler prints. This information is
3335 written to standard error, unless @option{-dS} or @option{-dR} is
3336 specified, in which case it is output to the usual dump
3337 listing file, @file{.sched} or @file{.sched2} respectively. However
3338 for @var{n} greater than nine, the output is always printed to standard
3341 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
3342 same information as @option{-dRS}. For @var{n} greater than one, it
3343 also output basic block probabilities, detailed ready list information
3344 and unit/insn info. For @var{n} greater than two, it includes RTL
3345 at abort point, control-flow and regions info. And for @var{n} over
3346 four, @option{-fsched-verbose} also includes dependence info.
3350 Store the usual ``temporary'' intermediate files permanently; place them
3351 in the current directory and name them based on the source file. Thus,
3352 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
3353 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
3354 preprocessed @file{foo.i} output file even though the compiler now
3355 normally uses an integrated preprocessor.
3359 Report the CPU time taken by each subprocess in the compilation
3360 sequence. For C source files, this is the compiler proper and assembler
3361 (plus the linker if linking is done). The output looks like this:
3368 The first number on each line is the ``user time,'' that is time spent
3369 executing the program itself. The second number is ``system time,''
3370 time spent executing operating system routines on behalf of the program.
3371 Both numbers are in seconds.
3373 @item -print-file-name=@var{library}
3374 @opindex print-file-name
3375 Print the full absolute name of the library file @var{library} that
3376 would be used when linking---and don't do anything else. With this
3377 option, GCC does not compile or link anything; it just prints the
3380 @item -print-multi-directory
3381 @opindex print-multi-directory
3382 Print the directory name corresponding to the multilib selected by any
3383 other switches present in the command line. This directory is supposed
3384 to exist in @env{GCC_EXEC_PREFIX}.
3386 @item -print-multi-lib
3387 @opindex print-multi-lib
3388 Print the mapping from multilib directory names to compiler switches
3389 that enable them. The directory name is separated from the switches by
3390 @samp{;}, and each switch starts with an @samp{@@} instead of the
3391 @samp{-}, without spaces between multiple switches. This is supposed to
3392 ease shell-processing.
3394 @item -print-prog-name=@var{program}
3395 @opindex print-prog-name
3396 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
3398 @item -print-libgcc-file-name
3399 @opindex print-libgcc-file-name
3400 Same as @option{-print-file-name=libgcc.a}.
3402 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
3403 but you do want to link with @file{libgcc.a}. You can do
3406 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
3409 @item -print-search-dirs
3410 @opindex print-search-dirs
3411 Print the name of the configured installation directory and a list of
3412 program and library directories gcc will search---and don't do anything else.
3414 This is useful when gcc prints the error message
3415 @samp{installation problem, cannot exec cpp0: No such file or directory}.
3416 To resolve this you either need to put @file{cpp0} and the other compiler
3417 components where gcc expects to find them, or you can set the environment
3418 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
3419 Don't forget the trailing '/'.
3420 @xref{Environment Variables}.
3423 @opindex dumpmachine
3424 Print the compiler's target machine (for example,
3425 @samp{i686-pc-linux-gnu})---and don't do anything else.
3428 @opindex dumpversion
3429 Print the compiler version (for example, @samp{3.0})---and don't do
3434 Print the compiler's built-in specs---and don't do anything else. (This
3435 is used when GCC itself is being built.) @xref{Spec Files}.
3437 @item -feliminate-unused-debug-types
3438 @opindex feliminate-unused-debug-types
3439 Normally, when producing DWARF2 output, GCC will emit debugging
3440 information for all types declared in a compilation
3441 unit, regardless of whether or not they are actually used
3442 in that compilation unit. Sometimes this is useful, such as
3443 if, in the debugger, you want to cast a value to a type that is
3444 not actually used in your program (but is declared). More often,
3445 however, this results in a significant amount of wasted space.
3446 With this option, GCC will avoid producing debug symbol output
3447 for types that are nowhere used in the source file being compiled.
3450 @node Optimize Options
3451 @section Options That Control Optimization
3452 @cindex optimize options
3453 @cindex options, optimization
3455 These options control various sorts of optimizations.
3457 Without any optimization option, the compiler's goal is to reduce the
3458 cost of compilation and to make debugging produce the expected
3459 results. Statements are independent: if you stop the program with a
3460 breakpoint between statements, you can then assign a new value to any
3461 variable or change the program counter to any other statement in the
3462 function and get exactly the results you would expect from the source
3465 Turning on optimization flags makes the compiler attempt to improve
3466 the performance and/or code size at the expense of compilation time
3467 and possibly the ability to debug the program.
3469 Not all optimizations are controlled directly by a flag. Only
3470 optimizations that have a flag are listed.
3477 Optimize. Optimizing compilation takes somewhat more time, and a lot
3478 more memory for a large function.
3480 With @option{-O}, the compiler tries to reduce code size and execution
3481 time, without performing any optimizations that take a great deal of
3484 @option{-O} turns on the following optimization flags:
3485 @gccoptlist{-fdefer-pop @gol
3486 -fmerge-constants @gol
3488 -floop-optimize @gol
3490 -fif-conversion @gol
3491 -fif-conversion2 @gol
3492 -fdelayed-branch @gol
3493 -fguess-branch-probability @gol
3496 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
3497 where doing so does not interfere with debugging.
3501 Optimize even more. GCC performs nearly all supported optimizations
3502 that do not involve a space-speed tradeoff. The compiler does not
3503 perform loop unrolling or function inlining when you specify @option{-O2}.
3504 As compared to @option{-O}, this option increases both compilation time
3505 and the performance of the generated code.
3507 @option{-O2} turns on all optimization flags specified by @option{-O}. It
3508 also turns on the following optimization flags:
3509 @gccoptlist{-fforce-mem @gol
3510 -foptimize-sibling-calls @gol
3511 -fstrength-reduce @gol
3512 -fcse-follow-jumps -fcse-skip-blocks @gol
3513 -frerun-cse-after-loop -frerun-loop-opt @gol
3514 -fgcse -fgcse-lm -fgcse-sm @gol
3515 -fdelete-null-pointer-checks @gol
3516 -fexpensive-optimizations @gol
3518 -fschedule-insns -fschedule-insns2 @gol
3519 -fsched-interblock -fsched-spec @gol
3522 -freorder-blocks -freorder-functions @gol
3523 -fstrict-aliasing @gol
3524 -falign-functions -falign-jumps @gol
3525 -falign-loops -falign-labels}
3527 Please note the warning under @option{-fgcse} about
3528 invoking @option{-O2} on programs that use computed gotos.
3532 Optimize yet more. @option{-O3} turns on all optimizations specified by
3533 @option{-O2} and also turns on the @option{-finline-functions},
3534 @option{-funit-at-a-time} and @option{-frename-registers} options.
3538 Do not optimize. This is the default.
3542 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
3543 do not typically increase code size. It also performs further
3544 optimizations designed to reduce code size.
3546 @option{-Os} disables the following optimization flags:
3547 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
3548 -falign-labels -freorder-blocks -fprefetch-loop-arrays}
3550 If you use multiple @option{-O} options, with or without level numbers,
3551 the last such option is the one that is effective.
3554 Options of the form @option{-f@var{flag}} specify machine-independent
3555 flags. Most flags have both positive and negative forms; the negative
3556 form of @option{-ffoo} would be @option{-fno-foo}. In the table
3557 below, only one of the forms is listed---the one you typically will
3558 use. You can figure out the other form by either removing @samp{no-}
3561 The following options control specific optimizations. They are either
3562 activated by @option{-O} options or are related to ones that are. You
3563 can use the following flags in the rare cases when ``fine-tuning'' of
3564 optimizations to be performed is desired.
3567 @item -fno-default-inline
3568 @opindex fno-default-inline
3569 Do not make member functions inline by default merely because they are
3570 defined inside the class scope (C++ only). Otherwise, when you specify
3571 @w{@option{-O}}, member functions defined inside class scope are compiled
3572 inline by default; i.e., you don't need to add @samp{inline} in front of
3573 the member function name.
3575 @item -fno-defer-pop
3576 @opindex fno-defer-pop
3577 Always pop the arguments to each function call as soon as that function
3578 returns. For machines which must pop arguments after a function call,
3579 the compiler normally lets arguments accumulate on the stack for several
3580 function calls and pops them all at once.
3582 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3586 Force memory operands to be copied into registers before doing
3587 arithmetic on them. This produces better code by making all memory
3588 references potential common subexpressions. When they are not common
3589 subexpressions, instruction combination should eliminate the separate
3592 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3595 @opindex fforce-addr
3596 Force memory address constants to be copied into registers before
3597 doing arithmetic on them. This may produce better code just as
3598 @option{-fforce-mem} may.
3600 @item -fomit-frame-pointer
3601 @opindex fomit-frame-pointer
3602 Don't keep the frame pointer in a register for functions that
3603 don't need one. This avoids the instructions to save, set up and
3604 restore frame pointers; it also makes an extra register available
3605 in many functions. @strong{It also makes debugging impossible on
3608 On some machines, such as the VAX, this flag has no effect, because
3609 the standard calling sequence automatically handles the frame pointer
3610 and nothing is saved by pretending it doesn't exist. The
3611 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
3612 whether a target machine supports this flag. @xref{Registers,,Register
3613 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
3615 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3617 @item -foptimize-sibling-calls
3618 @opindex foptimize-sibling-calls
3619 Optimize sibling and tail recursive calls.
3621 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3625 Don't pay attention to the @code{inline} keyword. Normally this option
3626 is used to keep the compiler from expanding any functions inline.
3627 Note that if you are not optimizing, no functions can be expanded inline.
3629 @item -finline-functions
3630 @opindex finline-functions
3631 Integrate all simple functions into their callers. The compiler
3632 heuristically decides which functions are simple enough to be worth
3633 integrating in this way.
3635 If all calls to a given function are integrated, and the function is
3636 declared @code{static}, then the function is normally not output as
3637 assembler code in its own right.
3639 Enabled at level @option{-O3}.
3641 @item -finline-limit=@var{n}
3642 @opindex finline-limit
3643 By default, gcc limits the size of functions that can be inlined. This flag
3644 allows the control of this limit for functions that are explicitly marked as
3645 inline (i.e., marked with the inline keyword or defined within the class
3646 definition in c++). @var{n} is the size of functions that can be inlined in
3647 number of pseudo instructions (not counting parameter handling). The default
3648 value of @var{n} is 600.
3649 Increasing this value can result in more inlined code at
3650 the cost of compilation time and memory consumption. Decreasing usually makes
3651 the compilation faster and less code will be inlined (which presumably
3652 means slower programs). This option is particularly useful for programs that
3653 use inlining heavily such as those based on recursive templates with C++.
3655 Inlining is actually controlled by a number of parameters, which may be
3656 specified individually by using @option{--param @var{name}=@var{value}}.
3657 The @option{-finline-limit=@var{n}} option sets some of these parameters
3661 @item max-inline-insns
3663 @item max-inline-insns-single
3664 is set to @var{n}/2.
3665 @item max-inline-insns-auto
3666 is set to @var{n}/2.
3667 @item min-inline-insns
3668 is set to 130 or @var{n}/4, whichever is smaller.
3669 @item max-inline-insns-rtl
3673 Using @option{-finline-limit=600} thus results in the default settings
3674 for these parameters. See below for a documentation of the individual
3675 parameters controlling inlining.
3677 @emph{Note:} pseudo instruction represents, in this particular context, an
3678 abstract measurement of function's size. In no way, it represents a count
3679 of assembly instructions and as such its exact meaning might change from one
3680 release to an another.
3682 @item -fkeep-inline-functions
3683 @opindex fkeep-inline-functions
3684 Even if all calls to a given function are integrated, and the function
3685 is declared @code{static}, nevertheless output a separate run-time
3686 callable version of the function. This switch does not affect
3687 @code{extern inline} functions.
3689 @item -fkeep-static-consts
3690 @opindex fkeep-static-consts
3691 Emit variables declared @code{static const} when optimization isn't turned
3692 on, even if the variables aren't referenced.
3694 GCC enables this option by default. If you want to force the compiler to
3695 check if the variable was referenced, regardless of whether or not
3696 optimization is turned on, use the @option{-fno-keep-static-consts} option.
3698 @item -fmerge-constants
3699 Attempt to merge identical constants (string constants and floating point
3700 constants) across compilation units.
3702 This option is the default for optimized compilation if the assembler and
3703 linker support it. Use @option{-fno-merge-constants} to inhibit this
3706 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3708 @item -fmerge-all-constants
3709 Attempt to merge identical constants and identical variables.
3711 This option implies @option{-fmerge-constants}. In addition to
3712 @option{-fmerge-constants} this considers e.g. even constant initialized
3713 arrays or initialized constant variables with integral or floating point
3714 types. Languages like C or C++ require each non-automatic variable to
3715 have distinct location, so using this option will result in non-conforming
3720 Use a graph coloring register allocator. Currently this option is meant
3721 for testing, so we are interested to hear about miscompilations with
3724 @item -fno-branch-count-reg
3725 @opindex fno-branch-count-reg
3726 Do not use ``decrement and branch'' instructions on a count register,
3727 but instead generate a sequence of instructions that decrement a
3728 register, compare it against zero, then branch based upon the result.
3729 This option is only meaningful on architectures that support such
3730 instructions, which include x86, PowerPC, IA-64 and S/390.
3732 The default is @option{-fbranch-count-reg}, enabled when
3733 @option{-fstrength-reduce} is enabled.
3735 @item -fno-function-cse
3736 @opindex fno-function-cse
3737 Do not put function addresses in registers; make each instruction that
3738 calls a constant function contain the function's address explicitly.
3740 This option results in less efficient code, but some strange hacks
3741 that alter the assembler output may be confused by the optimizations
3742 performed when this option is not used.
3744 The default is @option{-ffunction-cse}
3746 @item -fno-zero-initialized-in-bss
3747 @opindex fno-zero-initialized-in-bss
3748 If the target supports a BSS section, GCC by default puts variables that
3749 are initialized to zero into BSS@. This can save space in the resulting
3752 This option turns off this behavior because some programs explicitly
3753 rely on variables going to the data section. E.g., so that the
3754 resulting executable can find the beginning of that section and/or make
3755 assumptions based on that.
3757 The default is @option{-fzero-initialized-in-bss}.
3759 @item -fstrength-reduce
3760 @opindex fstrength-reduce
3761 Perform the optimizations of loop strength reduction and
3762 elimination of iteration variables.
3764 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3766 @item -fthread-jumps
3767 @opindex fthread-jumps
3768 Perform optimizations where we check to see if a jump branches to a
3769 location where another comparison subsumed by the first is found. If
3770 so, the first branch is redirected to either the destination of the
3771 second branch or a point immediately following it, depending on whether
3772 the condition is known to be true or false.
3774 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3776 @item -fcse-follow-jumps
3777 @opindex fcse-follow-jumps
3778 In common subexpression elimination, scan through jump instructions
3779 when the target of the jump is not reached by any other path. For
3780 example, when CSE encounters an @code{if} statement with an
3781 @code{else} clause, CSE will follow the jump when the condition
3784 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3786 @item -fcse-skip-blocks
3787 @opindex fcse-skip-blocks
3788 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
3789 follow jumps which conditionally skip over blocks. When CSE
3790 encounters a simple @code{if} statement with no else clause,
3791 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
3792 body of the @code{if}.
3794 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3796 @item -frerun-cse-after-loop
3797 @opindex frerun-cse-after-loop
3798 Re-run common subexpression elimination after loop optimizations has been
3801 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3803 @item -frerun-loop-opt
3804 @opindex frerun-loop-opt
3805 Run the loop optimizer twice.
3807 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3811 Perform a global common subexpression elimination pass.
3812 This pass also performs global constant and copy propagation.
3814 @emph{Note:} When compiling a program using computed gotos, a GCC
3815 extension, you may get better runtime performance if you disable
3816 the global common subexpression elimination pass by adding
3817 @option{-fno-gcse} to the command line.
3819 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3823 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
3824 attempt to move loads which are only killed by stores into themselves. This
3825 allows a loop containing a load/store sequence to be changed to a load outside
3826 the loop, and a copy/store within the loop.
3828 Enabled by default when gcse is enabled.
3832 When @option{-fgcse-sm} is enabled, A store motion pass is run after global common
3833 subexpression elimination. This pass will attempt to move stores out of loops.
3834 When used in conjunction with @option{-fgcse-lm}, loops containing a load/store sequence
3835 can be changed to a load before the loop and a store after the loop.
3837 Enabled by default when gcse is enabled.
3839 @item -floop-optimize
3840 @opindex floop-optimize
3841 Perform loop optimizations: move constant expressions out of loops, simplify
3842 exit test conditions and optionally do strength-reduction and loop unrolling as
3845 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3847 @item -fcrossjumping
3848 @opindex crossjumping
3849 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
3850 resulting code may or may not perform better than without cross-jumping.
3852 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3854 @item -fif-conversion
3855 @opindex if-conversion
3856 Attempt to transform conditional jumps into branch-less equivalents. This
3857 include use of conditional moves, min, max, set flags and abs instructions, and
3858 some tricks doable by standard arithmetics. The use of conditional execution
3859 on chips where it is available is controlled by @code{if-conversion2}.
3861 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3863 @item -fif-conversion2
3864 @opindex if-conversion2
3865 Use conditional execution (where available) to transform conditional jumps into
3866 branch-less equivalents.
3868 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3870 @item -fdelete-null-pointer-checks
3871 @opindex fdelete-null-pointer-checks
3872 Use global dataflow analysis to identify and eliminate useless checks
3873 for null pointers. The compiler assumes that dereferencing a null
3874 pointer would have halted the program. If a pointer is checked after
3875 it has already been dereferenced, it cannot be null.
3877 In some environments, this assumption is not true, and programs can
3878 safely dereference null pointers. Use
3879 @option{-fno-delete-null-pointer-checks} to disable this optimization
3880 for programs which depend on that behavior.
3882 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3884 @item -fexpensive-optimizations
3885 @opindex fexpensive-optimizations
3886 Perform a number of minor optimizations that are relatively expensive.
3888 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3890 @item -foptimize-register-move
3892 @opindex foptimize-register-move
3894 Attempt to reassign register numbers in move instructions and as
3895 operands of other simple instructions in order to maximize the amount of
3896 register tying. This is especially helpful on machines with two-operand
3899 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
3902 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3904 @item -fdelayed-branch
3905 @opindex fdelayed-branch
3906 If supported for the target machine, attempt to reorder instructions
3907 to exploit instruction slots available after delayed branch
3910 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3912 @item -fschedule-insns
3913 @opindex fschedule-insns
3914 If supported for the target machine, attempt to reorder instructions to
3915 eliminate execution stalls due to required data being unavailable. This
3916 helps machines that have slow floating point or memory load instructions
3917 by allowing other instructions to be issued until the result of the load
3918 or floating point instruction is required.
3920 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3922 @item -fschedule-insns2
3923 @opindex fschedule-insns2
3924 Similar to @option{-fschedule-insns}, but requests an additional pass of
3925 instruction scheduling after register allocation has been done. This is
3926 especially useful on machines with a relatively small number of
3927 registers and where memory load instructions take more than one cycle.
3929 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3931 @item -fno-sched-interblock
3932 @opindex fno-sched-interblock
3933 Don't schedule instructions across basic blocks. This is normally
3934 enabled by default when scheduling before register allocation, i.e.@:
3935 with @option{-fschedule-insns} or at @option{-O2} or higher.
3937 @item -fno-sched-spec
3938 @opindex fno-sched-spec
3939 Don't allow speculative motion of non-load instructions. This is normally
3940 enabled by default when scheduling before register allocation, i.e.@:
3941 with @option{-fschedule-insns} or at @option{-O2} or higher.
3943 @item -fsched-spec-load
3944 @opindex fsched-spec-load
3945 Allow speculative motion of some load instructions. This only makes
3946 sense when scheduling before register allocation, i.e.@: with
3947 @option{-fschedule-insns} or at @option{-O2} or higher.
3949 @item -fsched-spec-load-dangerous
3950 @opindex fsched-spec-load-dangerous
3951 Allow speculative motion of more load instructions. This only makes
3952 sense when scheduling before register allocation, i.e.@: with
3953 @option{-fschedule-insns} or at @option{-O2} or higher.
3955 @item -fsched2-use-superblocks
3956 @opindex fsched2-use-superblocks
3957 When schedulilng after register allocation, do use superblock scheduling
3958 algorithm. Superblock scheduling allows motion across basic block boundaries
3959 resulting on faster schedules. This option is experimental, as not all machine
3960 descriptions used by GCC model the CPU closely enough to avoid unreliable
3961 results from the algorithm.
3963 This only makes sense when scheduling after register allocation, i.e.@: with
3964 @option{-fschedule-insns2} or at @option{-O2} or higher.
3966 @item -fsched2-use-traces
3967 @opindex fsched2-use-traces
3968 Use @option{-fsched2-use-superblocks} algorithm when scheduling after register
3969 allocation and additionally perform code duplication in order to increase the
3970 size of superblocks using tracer pass. See @option{-ftracer} for details on
3973 This mode should produce faster but significantly longer programs. Also
3974 without @code{-fbranch-probabilities} the traces constructed may not match the
3975 reality and hurt the performance. This only makes
3976 sense when scheduling after register allocation, i.e.@: with
3977 @option{-fschedule-insns2} or at @option{-O2} or higher.
3979 @item -fcaller-saves
3980 @opindex fcaller-saves
3981 Enable values to be allocated in registers that will be clobbered by
3982 function calls, by emitting extra instructions to save and restore the
3983 registers around such calls. Such allocation is done only when it
3984 seems to result in better code than would otherwise be produced.
3986 This option is always enabled by default on certain machines, usually
3987 those which have no call-preserved registers to use instead.
3989 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3991 @item -fmove-all-movables
3992 @opindex fmove-all-movables
3993 Forces all invariant computations in loops to be moved
3996 @item -freduce-all-givs
3997 @opindex freduce-all-givs
3998 Forces all general-induction variables in loops to be
4001 @emph{Note:} When compiling programs written in Fortran,
4002 @option{-fmove-all-movables} and @option{-freduce-all-givs} are enabled
4003 by default when you use the optimizer.
4005 These options may generate better or worse code; results are highly
4006 dependent on the structure of loops within the source code.
4008 These two options are intended to be removed someday, once
4009 they have helped determine the efficacy of various
4010 approaches to improving loop optimizations.
4012 Please let us (@w{@email{gcc@@gcc.gnu.org}} and @w{@email{fortran@@gnu.org}})
4013 know how use of these options affects
4014 the performance of your production code.
4015 We're very interested in code that runs @emph{slower}
4016 when these options are @emph{enabled}.
4019 @itemx -fno-peephole2
4020 @opindex fno-peephole
4021 @opindex fno-peephole2
4022 Disable any machine-specific peephole optimizations. The difference
4023 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
4024 are implemented in the compiler; some targets use one, some use the
4025 other, a few use both.
4027 @option{-fpeephole} is enabled by default.
4028 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4030 @item -fno-guess-branch-probability
4031 @opindex fno-guess-branch-probability
4032 Do not guess branch probabilities using a randomized model.
4034 Sometimes gcc will opt to use a randomized model to guess branch
4035 probabilities, when none are available from either profiling feedback
4036 (@option{-fprofile-arcs}) or @samp{__builtin_expect}. This means that
4037 different runs of the compiler on the same program may produce different
4040 In a hard real-time system, people don't want different runs of the
4041 compiler to produce code that has different behavior; minimizing
4042 non-determinism is of paramount import. This switch allows users to
4043 reduce non-determinism, possibly at the expense of inferior
4046 The default is @option{-fguess-branch-probability} at levels
4047 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4049 @item -freorder-blocks
4050 @opindex freorder-blocks
4051 Reorder basic blocks in the compiled function in order to reduce number of
4052 taken branches and improve code locality.
4054 Enabled at levels @option{-O2}, @option{-O3}.
4056 @item -freorder-functions
4057 @opindex freorder-functions
4058 Reorder basic blocks in the compiled function in order to reduce number of
4059 taken branches and improve code locality. This is implemented by using special
4060 subsections @code{text.hot} for most frequently executed functions and
4061 @code{text.unlikely} for unlikely executed functions. Reordering is done by
4062 the linker so object file format must support named sections and linker must
4063 place them in a reasonable way.
4065 Also profile feedback must be available in to make this option effective. See
4066 @option{-fprofile-arcs} for details.
4068 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4070 @item -fstrict-aliasing
4071 @opindex fstrict-aliasing
4072 Allows the compiler to assume the strictest aliasing rules applicable to
4073 the language being compiled. For C (and C++), this activates
4074 optimizations based on the type of expressions. In particular, an
4075 object of one type is assumed never to reside at the same address as an
4076 object of a different type, unless the types are almost the same. For
4077 example, an @code{unsigned int} can alias an @code{int}, but not a
4078 @code{void*} or a @code{double}. A character type may alias any other
4081 Pay special attention to code like this:
4094 The practice of reading from a different union member than the one most
4095 recently written to (called ``type-punning'') is common. Even with
4096 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
4097 is accessed through the union type. So, the code above will work as
4098 expected. However, this code might not:
4109 Every language that wishes to perform language-specific alias analysis
4110 should define a function that computes, given an @code{tree}
4111 node, an alias set for the node. Nodes in different alias sets are not
4112 allowed to alias. For an example, see the C front-end function
4113 @code{c_get_alias_set}.
4115 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4117 @item -falign-functions
4118 @itemx -falign-functions=@var{n}
4119 @opindex falign-functions
4120 Align the start of functions to the next power-of-two greater than
4121 @var{n}, skipping up to @var{n} bytes. For instance,
4122 @option{-falign-functions=32} aligns functions to the next 32-byte
4123 boundary, but @option{-falign-functions=24} would align to the next
4124 32-byte boundary only if this can be done by skipping 23 bytes or less.
4126 @option{-fno-align-functions} and @option{-falign-functions=1} are
4127 equivalent and mean that functions will not be aligned.
4129 Some assemblers only support this flag when @var{n} is a power of two;
4130 in that case, it is rounded up.
4132 If @var{n} is not specified, use a machine-dependent default.
4134 Enabled at levels @option{-O2}, @option{-O3}.
4136 @item -falign-labels
4137 @itemx -falign-labels=@var{n}
4138 @opindex falign-labels
4139 Align all branch targets to a power-of-two boundary, skipping up to
4140 @var{n} bytes like @option{-falign-functions}. This option can easily
4141 make code slower, because it must insert dummy operations for when the
4142 branch target is reached in the usual flow of the code.
4144 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
4145 are greater than this value, then their values are used instead.
4147 If @var{n} is not specified, use a machine-dependent default which is
4148 very likely to be @samp{1}, meaning no alignment.
4150 Enabled at levels @option{-O2}, @option{-O3}.
4153 @itemx -falign-loops=@var{n}
4154 @opindex falign-loops
4155 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
4156 like @option{-falign-functions}. The hope is that the loop will be
4157 executed many times, which will make up for any execution of the dummy
4160 If @var{n} is not specified, use a machine-dependent default.
4162 Enabled at levels @option{-O2}, @option{-O3}.
4165 @itemx -falign-jumps=@var{n}
4166 @opindex falign-jumps
4167 Align branch targets to a power-of-two boundary, for branch targets
4168 where the targets can only be reached by jumping, skipping up to @var{n}
4169 bytes like @option{-falign-functions}. In this case, no dummy operations
4172 If @var{n} is not specified, use a machine-dependent default.
4174 Enabled at levels @option{-O2}, @option{-O3}.
4176 @item -frename-registers
4177 @opindex frename-registers
4178 Attempt to avoid false dependencies in scheduled code by making use
4179 of registers left over after register allocation. This optimization
4180 will most benefit processors with lots of registers. It can, however,
4181 make debugging impossible, since variables will no longer stay in
4182 a ``home register''.
4184 Enabled at levels @option{-O3}.
4186 @item -fno-cprop-registers
4187 @opindex fno-cprop-registers
4188 After register allocation and post-register allocation instruction splitting,
4189 we perform a copy-propagation pass to try to reduce scheduling dependencies
4190 and occasionally eliminate the copy.
4192 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4196 The following options control compiler behavior regarding floating
4197 point arithmetic. These options trade off between speed and
4198 correctness. All must be specifically enabled.
4202 @opindex ffloat-store
4203 Do not store floating point variables in registers, and inhibit other
4204 options that might change whether a floating point value is taken from a
4207 @cindex floating point precision
4208 This option prevents undesirable excess precision on machines such as
4209 the 68000 where the floating registers (of the 68881) keep more
4210 precision than a @code{double} is supposed to have. Similarly for the
4211 x86 architecture. For most programs, the excess precision does only
4212 good, but a few programs rely on the precise definition of IEEE floating
4213 point. Use @option{-ffloat-store} for such programs, after modifying
4214 them to store all pertinent intermediate computations into variables.
4218 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations}, @*
4219 @option{-fno-trapping-math}, @option{-ffinite-math-only} and @*
4220 @option{-fno-signaling-nans}.
4222 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
4224 This option should never be turned on by any @option{-O} option since
4225 it can result in incorrect output for programs which depend on
4226 an exact implementation of IEEE or ISO rules/specifications for
4229 @item -fno-math-errno
4230 @opindex fno-math-errno
4231 Do not set ERRNO after calling math functions that are executed
4232 with a single instruction, e.g., sqrt. A program that relies on
4233 IEEE exceptions for math error handling may want to use this flag
4234 for speed while maintaining IEEE arithmetic compatibility.
4236 This option should never be turned on by any @option{-O} option since
4237 it can result in incorrect output for programs which depend on
4238 an exact implementation of IEEE or ISO rules/specifications for
4241 The default is @option{-fmath-errno}.
4243 @item -funsafe-math-optimizations
4244 @opindex funsafe-math-optimizations
4245 Allow optimizations for floating-point arithmetic that (a) assume
4246 that arguments and results are valid and (b) may violate IEEE or
4247 ANSI standards. When used at link-time, it may include libraries
4248 or startup files that change the default FPU control word or other
4249 similar optimizations.
4251 This option should never be turned on by any @option{-O} option since
4252 it can result in incorrect output for programs which depend on
4253 an exact implementation of IEEE or ISO rules/specifications for
4256 The default is @option{-fno-unsafe-math-optimizations}.
4258 @item -ffinite-math-only
4259 @opindex ffinite-math-only
4260 Allow optimizations for floating-point arithmetic that assume
4261 that arguments and results are not NaNs or +-Infs.
4263 This option should never be turned on by any @option{-O} option since
4264 it can result in incorrect output for programs which depend on
4265 an exact implementation of IEEE or ISO rules/specifications.
4267 The default is @option{-fno-finite-math-only}.
4269 @item -fno-trapping-math
4270 @opindex fno-trapping-math
4271 Compile code assuming that floating-point operations cannot generate
4272 user-visible traps. These traps include division by zero, overflow,
4273 underflow, inexact result and invalid operation. This option implies
4274 @option{-fno-signaling-nans}. Setting this option may allow faster
4275 code if one relies on ``non-stop'' IEEE arithmetic, for example.
4277 This option should never be turned on by any @option{-O} option since
4278 it can result in incorrect output for programs which depend on
4279 an exact implementation of IEEE or ISO rules/specifications for
4282 The default is @option{-ftrapping-math}.
4284 @item -fsignaling-nans
4285 @opindex fsignaling-nans
4286 Compile code assuming that IEEE signaling NaNs may generate user-visible
4287 traps during floating-point operations. Setting this option disables
4288 optimizations that may change the number of exceptions visible with
4289 signaling NaNs. This option implies @option{-ftrapping-math}.
4291 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
4294 The default is @option{-fno-signaling-nans}.
4296 This option is experimental and does not currently guarantee to
4297 disable all GCC optimizations that affect signaling NaN behavior.
4299 @item -fsingle-precision-constant
4300 @opindex fsingle-precision-constant
4301 Treat floating point constant as single precision constant instead of
4302 implicitly converting it to double precision constant.
4307 The following options control optimizations that may improve
4308 performance, but are not enabled by any @option{-O} options. This
4309 section includes experimental options that may produce broken code.
4312 @item -fbranch-probabilities
4313 @opindex fbranch-probabilities
4314 After running a program compiled with @option{-fprofile-arcs}
4315 (@pxref{Debugging Options,, Options for Debugging Your Program or
4316 @command{gcc}}), you can compile it a second time using
4317 @option{-fbranch-probabilities}, to improve optimizations based on
4318 the number of times each branch was taken. When the program
4319 compiled with @option{-fprofile-arcs} exits it saves arc execution
4320 counts to a file called @file{@var{sourcename}.da} for each source
4321 file The information in this data file is very dependent on the
4322 structure of the generated code, so you must use the same source code
4323 and the same optimization options for both compilations.
4325 With @option{-fbranch-probabilities}, GCC puts a
4326 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
4327 These can be used to improve optimization. Currently, they are only
4328 used in one place: in @file{reorg.c}, instead of guessing which path a
4329 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
4330 exactly determine which path is taken more often.
4332 @item -fprofile-values
4333 @opindex fprofile-values
4334 If combined with @option{-fprofile-arcs}, it adds code so that some
4335 data about values of expressions in the program is gathered.
4339 Use a graph coloring register allocator. Currently this option is meant
4340 for testing, so we are interested to hear about miscompilations with
4345 Perform tail duplication to enlarge superblock size. This transformation
4346 simplifies the control flow of the function allowing other optimizations to do
4349 @item -funit-at-a-time
4350 @opindex funit-at-a-time
4351 Parse the whole compilation unit before starting to produce code. This allows some
4352 extra optimizations to take place but consumes more memory.
4354 @item -funroll-loops
4355 @opindex funroll-loops
4356 Unroll loops whose number of iterations can be determined at compile time or
4357 upon entry to the loop. @option{-funroll-loops} implies
4358 @option{-frerun-cse-after-loop}. It also turns on complete loop peeling
4359 (i.e. complete removal of loops with small constant number of iterations).
4360 This option makes code larger, and may or may not make it run faster.
4362 @item -funroll-all-loops
4363 @opindex funroll-all-loops
4364 Unroll all loops, even if their number of iterations is uncertain when
4365 the loop is entered. This usually makes programs run more slowly.
4366 @option{-funroll-all-loops} implies the same options as
4367 @option{-funroll-loops}.
4370 @opindex fpeel-loops
4371 Peels the loops for that there is enough information that they do not
4372 roll much (from profile feedback). It also turns on complete loop peeling
4373 (i.e. complete removal of loops with small constant number of iterations).
4375 @item -funswitch-loops
4376 @opindex funswitch-loops
4377 Move branches with loop invariant conditions out of the loop, with duplicates
4378 of the loop on both branches (modified according to result of the condition).
4380 @item -fold-unroll-loops
4381 @opindex fold-unroll-loops
4382 Unroll loops whose number of iterations can be determined at compile
4383 time or upon entry to the loop, using the old loop unroller whose loop
4384 recognition is based on notes from frontend. @option{-fold-unroll-loops} implies
4385 both @option{-fstrength-reduce} and @option{-frerun-cse-after-loop}. This
4386 option makes code larger, and may or may not make it run faster.
4388 @item -fold-unroll-all-loops
4389 @opindex fold-unroll-all-loops
4390 Unroll all loops, even if their number of iterations is uncertain when
4391 the loop is entered. This is done using the old loop unroller whose loop
4392 recognition is based on notes from frontend. This usually makes programs run more slowly.
4393 @option{-fold-unroll-all-loops} implies the same options as
4394 @option{-fold-unroll-loops}.
4396 @item -funswitch-loops
4397 @opindex funswitch-loops
4398 Move branches with loop invariant conditions out of the loop, with duplicates
4399 of the loop on both branches (modified according to result of the condition).
4401 @item -funswitch-loops
4402 @opindex funswitch-loops
4403 Move branches with loop invariant conditions out of the loop, with duplicates
4404 of the loop on both branches (modified according to result of the condition).
4406 @item -fprefetch-loop-arrays
4407 @opindex fprefetch-loop-arrays
4408 If supported by the target machine, generate instructions to prefetch
4409 memory to improve the performance of loops that access large arrays.
4411 Disabled at level @option{-Os}.
4413 @item -ffunction-sections
4414 @itemx -fdata-sections
4415 @opindex ffunction-sections
4416 @opindex fdata-sections
4417 Place each function or data item into its own section in the output
4418 file if the target supports arbitrary sections. The name of the
4419 function or the name of the data item determines the section's name
4422 Use these options on systems where the linker can perform optimizations
4423 to improve locality of reference in the instruction space. Most systems
4424 using the ELF object format and SPARC processors running Solaris 2 have
4425 linkers with such optimizations. AIX may have these optimizations in
4428 Only use these options when there are significant benefits from doing
4429 so. When you specify these options, the assembler and linker will
4430 create larger object and executable files and will also be slower.
4431 You will not be able to use @code{gprof} on all systems if you
4432 specify this option and you may have problems with debugging if
4433 you specify both this option and @option{-g}.
4437 Perform optimizations in static single assignment form. Each function's
4438 flow graph is translated into SSA form, optimizations are performed, and
4439 the flow graph is translated back from SSA form. Users should not
4440 specify this option, since it is not yet ready for production use.
4444 Perform Sparse Conditional Constant Propagation in SSA form. Requires
4445 @option{-fssa}. Like @option{-fssa}, this is an experimental feature.
4449 Perform aggressive dead-code elimination in SSA form. Requires @option{-fssa}.
4450 Like @option{-fssa}, this is an experimental feature.
4452 @item -fbranch-target-load-optimize
4453 @opindex fbranch-target-load-optimize
4454 Perform branch target register load optimization before prologue / epilogue
4456 The use of target registers can typically be exposed only during reload,
4457 thus hoisting loads out of loops and doing inter-block scheduling needs
4458 a separate optimization pass.
4460 @item -fbranch-target-load-optimize2
4461 @opindex fbranch-target-load-optimize2
4462 Perform branch target register load optimization after prologue / epilogue
4468 @item --param @var{name}=@var{value}
4470 In some places, GCC uses various constants to control the amount of
4471 optimization that is done. For example, GCC will not inline functions
4472 that contain more that a certain number of instructions. You can
4473 control some of these constants on the command-line using the
4474 @option{--param} option.
4476 In each case, the @var{value} is an integer. The allowable choices for
4477 @var{name} are given in the following table:
4480 @item max-crossjump-edges
4481 The maximum number of incoming edges to consider for crossjumping.
4482 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
4483 the number of edges incoming to each block. Increasing values mean
4484 more aggressive optimization, making the compile time increase with
4485 probably small improvement in executable size.
4487 @item max-delay-slot-insn-search
4488 The maximum number of instructions to consider when looking for an
4489 instruction to fill a delay slot. If more than this arbitrary number of
4490 instructions is searched, the time savings from filling the delay slot
4491 will be minimal so stop searching. Increasing values mean more
4492 aggressive optimization, making the compile time increase with probably
4493 small improvement in executable run time.
4495 @item max-delay-slot-live-search
4496 When trying to fill delay slots, the maximum number of instructions to
4497 consider when searching for a block with valid live register
4498 information. Increasing this arbitrarily chosen value means more
4499 aggressive optimization, increasing the compile time. This parameter
4500 should be removed when the delay slot code is rewritten to maintain the
4503 @item max-gcse-memory
4504 The approximate maximum amount of memory that will be allocated in
4505 order to perform the global common subexpression elimination
4506 optimization. If more memory than specified is required, the
4507 optimization will not be done.
4509 @item max-gcse-passes
4510 The maximum number of passes of GCSE to run.
4512 @item max-pending-list-length
4513 The maximum number of pending dependencies scheduling will allow
4514 before flushing the current state and starting over. Large functions
4515 with few branches or calls can create excessively large lists which
4516 needlessly consume memory and resources.
4518 @item max-inline-insns-single
4519 Several parameters control the tree inliner used in gcc.
4520 This number sets the maximum number of instructions (counted in gcc's
4521 internal representation) in a single function that the tree inliner
4522 will consider for inlining. This only affects functions declared
4523 inline and methods implemented in a class declaration (C++).
4524 The default value is 300.
4526 @item max-inline-insns-auto
4527 When you use @option{-finline-functions} (included in @option{-O3}),
4528 a lot of functions that would otherwise not be considered for inlining
4529 by the compiler will be investigated. To those functions, a different
4530 (more restrictive) limit compared to functions declared inline can
4532 The default value is 300.
4534 @item max-inline-insns
4535 The tree inliner does decrease the allowable size for single functions
4536 to be inlined after we already inlined the number of instructions
4537 given here by repeated inlining. This number should be a factor of
4538 two or more larger than the single function limit.
4539 Higher numbers result in better runtime performance, but incur higher
4540 compile-time resource (CPU time, memory) requirements and result in
4541 larger binaries. Very high values are not advisable, as too large
4542 binaries may adversely affect runtime performance.
4543 The default value is 600.
4545 @item max-inline-slope
4546 After exceeding the maximum number of inlined instructions by repeated
4547 inlining, a linear function is used to decrease the allowable size
4548 for single functions. The slope of that function is the negative
4549 reciprocal of the number specified here.
4550 The default value is 32.
4552 @item min-inline-insns
4553 The repeated inlining is throttled more and more by the linear function
4554 after exceeding the limit. To avoid too much throttling, a minimum for
4555 this function is specified here to allow repeated inlining for very small
4556 functions even when a lot of repeated inlining already has been done.
4557 The default value is 130.
4559 @item max-inline-insns-rtl
4560 For languages that use the RTL inliner (this happens at a later stage
4561 than tree inlining), you can set the maximum allowable size (counted
4562 in RTL instructions) for the RTL inliner with this parameter.
4563 The default value is 600.
4566 @item max-unrolled-insns
4567 The maximum number of instructions that a loop should have if that loop
4568 is unrolled, and if the loop is unrolled, it determines how many times
4569 the loop code is unrolled.
4571 @item max-average-unrolled-insns
4572 The maximum number of instructions biased by probabilities of their execution
4573 that a loop should have if that loop is unrolled, and if the loop is unrolled,
4574 it determines how many times the loop code is unrolled.
4576 @item max-unroll-times
4577 The maximum number of unrollings of a single loop.
4579 @item max-peeled-insns
4580 The maximum number of instructions that a loop should have if that loop
4581 is peeled, and if the loop is peeled, it determines how many times
4582 the loop code is peeled.
4584 @item max-peel-times
4585 The maximum number of peelings of a single loop.
4587 @item max-completely-peeled-insns
4588 The maximum number of insns of a completely peeled loop.
4590 @item max-completely-peel-times
4591 The maximum number of iterations of a loop to be suitable for complete peeling.
4593 @item max-unswitch-insns
4594 The maximum number of insns of an unswitched loop.
4596 @item max-unswitch-level
4597 The maximum number of branches unswitched in a single loop.
4599 @item hot-bb-count-fraction
4600 Select fraction of the maximal count of repetitions of basic block in program
4601 given basic block needs to have to be considered hot.
4603 @item hot-bb-frequency-fraction
4604 Select fraction of the maximal frequency of executions of basic block in
4605 function given basic block needs to have to be considered hot
4607 @item tracer-dynamic-coverage
4608 @itemx tracer-dynamic-coverage-feedback
4610 This value is used to limit superblock formation once the given percentage of
4611 executed instructions is covered. This limits unnecessary code size
4614 The @option{tracer-dynamic-coverage-feedback} is used only when profile
4615 feedback is available. The real profiles (as opposed to statically estimated
4616 ones) are much less balanced allowing the threshold to be larger value.
4618 @item tracer-max-code-growth
4619 Stop tail duplication once code growth has reached given percentage. This is
4620 rather hokey argument, as most of the duplicates will be eliminated later in
4621 cross jumping, so it may be set to much higher values than is the desired code
4624 @item tracer-min-branch-ratio
4626 Stop reverse growth when the reverse probability of best edge is less than this
4627 threshold (in percent).
4629 @item tracer-min-branch-ratio
4630 @itemx tracer-min-branch-ratio-feedback
4632 Stop forward growth if the best edge do have probability lower than this
4635 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
4636 compilation for profile feedback and one for compilation without. The value
4637 for compilation with profile feedback needs to be more conservative (higher) in
4638 order to make tracer effective.
4640 @item max-cse-path-length
4642 Maximum number of basic blocks on path that cse considers.
4644 @item ggc-min-expand
4646 GCC uses a garbage collector to manage its own memory allocation. This
4647 parameter specifies the minimum percentage by which the garbage
4648 collector's heap should be allowed to expand between collections.
4649 Tuning this may improve compilation speed; it has no effect on code
4652 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
4653 RAM >= 1GB. If @code{getrlimit} is available, the notion of "RAM" is
4654 the smallest of actual RAM, RLIMIT_RSS, RLIMIT_DATA and RLIMIT_AS. If
4655 GCC is not able to calculate RAM on a particular platform, the lower
4656 bound of 30% is used. Setting this parameter and
4657 @option{ggc-min-heapsize} to zero causes a full collection to occur at
4658 every opportunity. This is extremely slow, but can be useful for
4661 @item ggc-min-heapsize
4663 Minimum size of the garbage collector's heap before it begins bothering
4664 to collect garbage. The first collection occurs after the heap expands
4665 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
4666 tuning this may improve compilation speed, and has no effect on code
4669 The default is RAM/8, with a lower bound of 4096 (four megabytes) and an
4670 upper bound of 131072 (128 megabytes). If @code{getrlimit} is
4671 available, the notion of "RAM" is the smallest of actual RAM,
4672 RLIMIT_RSS, RLIMIT_DATA and RLIMIT_AS. If GCC is not able to calculate
4673 RAM on a particular platform, the lower bound is used. Setting this
4674 parameter very large effectively disables garbage collection. Setting
4675 this parameter and @option{ggc-min-expand} to zero causes a full
4676 collection to occur at every opportunity.
4678 @item reorder-blocks-duplicate
4679 @itemx reorder-blocks-duplicate-feedback
4681 Used by basic block reordering pass to decide whether to use unconditional
4682 branch or duplicate the code on it's destination. Code is duplicated when it's
4683 estimated size is smaller than this value multiplied by the estimated size of
4684 unconditional jump in the hot spots of the program.
4686 The @option{reorder-block-duplicate-feedback} is used only when profile
4687 feedback is available and may be set to higher values than
4688 @option{reorder-block-duplicate} since information about the hot spots is more
4693 @node Preprocessor Options
4694 @section Options Controlling the Preprocessor
4695 @cindex preprocessor options
4696 @cindex options, preprocessor
4698 These options control the C preprocessor, which is run on each C source
4699 file before actual compilation.
4701 If you use the @option{-E} option, nothing is done except preprocessing.
4702 Some of these options make sense only together with @option{-E} because
4703 they cause the preprocessor output to be unsuitable for actual
4708 You can use @option{-Wp,@var{option}} to bypass the compiler driver
4709 and pass @var{option} directly through to the preprocessor. If
4710 @var{option} contains commas, it is split into multiple options at the
4711 commas. However, many options are modified, translated or interpreted
4712 by the compiler driver before being passed to the preprocessor, and
4713 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
4714 interface is undocumented and subject to change, so whenever possible
4715 you should avoid using @option{-Wp} and let the driver handle the
4718 @item -Xpreprocessor @var{option}
4719 @opindex preprocessor
4720 Pass @var{option} as an option to the preprocessor. You can use this to
4721 supply system-specific preprocessor options which GCC does not know how to
4724 If you want to pass an option that takes an argument, you must use
4725 @option{-Xpreprocessor} twice, once for the option and once for the argument.
4728 @include cppopts.texi
4730 @node Assembler Options
4731 @section Passing Options to the Assembler
4733 @c prevent bad page break with this line
4734 You can pass options to the assembler.
4737 @item -Wa,@var{option}
4739 Pass @var{option} as an option to the assembler. If @var{option}
4740 contains commas, it is split into multiple options at the commas.
4742 @item -Xassembler @var{option}
4744 Pass @var{option} as an option to the assembler. You can use this to
4745 supply system-specific assembler options which GCC does not know how to
4748 If you want to pass an option that takes an argument, you must use
4749 @option{-Xassembler} twice, once for the option and once for the argument.
4754 @section Options for Linking
4755 @cindex link options
4756 @cindex options, linking
4758 These options come into play when the compiler links object files into
4759 an executable output file. They are meaningless if the compiler is
4760 not doing a link step.
4764 @item @var{object-file-name}
4765 A file name that does not end in a special recognized suffix is
4766 considered to name an object file or library. (Object files are
4767 distinguished from libraries by the linker according to the file
4768 contents.) If linking is done, these object files are used as input
4777 If any of these options is used, then the linker is not run, and
4778 object file names should not be used as arguments. @xref{Overall
4782 @item -l@var{library}
4783 @itemx -l @var{library}
4785 Search the library named @var{library} when linking. (The second
4786 alternative with the library as a separate argument is only for
4787 POSIX compliance and is not recommended.)
4789 It makes a difference where in the command you write this option; the
4790 linker searches and processes libraries and object files in the order they
4791 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
4792 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
4793 to functions in @samp{z}, those functions may not be loaded.
4795 The linker searches a standard list of directories for the library,
4796 which is actually a file named @file{lib@var{library}.a}. The linker
4797 then uses this file as if it had been specified precisely by name.
4799 The directories searched include several standard system directories
4800 plus any that you specify with @option{-L}.
4802 Normally the files found this way are library files---archive files
4803 whose members are object files. The linker handles an archive file by
4804 scanning through it for members which define symbols that have so far
4805 been referenced but not defined. But if the file that is found is an
4806 ordinary object file, it is linked in the usual fashion. The only
4807 difference between using an @option{-l} option and specifying a file name
4808 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
4809 and searches several directories.
4813 You need this special case of the @option{-l} option in order to
4814 link an Objective-C program.
4817 @opindex nostartfiles
4818 Do not use the standard system startup files when linking.
4819 The standard system libraries are used normally, unless @option{-nostdlib}
4820 or @option{-nodefaultlibs} is used.
4822 @item -nodefaultlibs
4823 @opindex nodefaultlibs
4824 Do not use the standard system libraries when linking.
4825 Only the libraries you specify will be passed to the linker.
4826 The standard startup files are used normally, unless @option{-nostartfiles}
4827 is used. The compiler may generate calls to memcmp, memset, and memcpy
4828 for System V (and ISO C) environments or to bcopy and bzero for
4829 BSD environments. These entries are usually resolved by entries in
4830 libc. These entry points should be supplied through some other
4831 mechanism when this option is specified.
4835 Do not use the standard system startup files or libraries when linking.
4836 No startup files and only the libraries you specify will be passed to
4837 the linker. The compiler may generate calls to memcmp, memset, and memcpy
4838 for System V (and ISO C) environments or to bcopy and bzero for
4839 BSD environments. These entries are usually resolved by entries in
4840 libc. These entry points should be supplied through some other
4841 mechanism when this option is specified.
4843 @cindex @option{-lgcc}, use with @option{-nostdlib}
4844 @cindex @option{-nostdlib} and unresolved references
4845 @cindex unresolved references and @option{-nostdlib}
4846 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
4847 @cindex @option{-nodefaultlibs} and unresolved references
4848 @cindex unresolved references and @option{-nodefaultlibs}
4849 One of the standard libraries bypassed by @option{-nostdlib} and
4850 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
4851 that GCC uses to overcome shortcomings of particular machines, or special
4852 needs for some languages.
4853 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
4854 Collection (GCC) Internals},
4855 for more discussion of @file{libgcc.a}.)
4856 In most cases, you need @file{libgcc.a} even when you want to avoid
4857 other standard libraries. In other words, when you specify @option{-nostdlib}
4858 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
4859 This ensures that you have no unresolved references to internal GCC
4860 library subroutines. (For example, @samp{__main}, used to ensure C++
4861 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
4862 GNU Compiler Collection (GCC) Internals}.)
4866 Produce a position independent executable on targets which support it.
4867 For predictable results, you must also specify the same set of options
4868 that were used to generate code (@option{-fpie}, @option{-fPIE},
4869 or model suboptions) when you specify this option.
4873 Remove all symbol table and relocation information from the executable.
4877 On systems that support dynamic linking, this prevents linking with the shared
4878 libraries. On other systems, this option has no effect.
4882 Produce a shared object which can then be linked with other objects to
4883 form an executable. Not all systems support this option. For predictable
4884 results, you must also specify the same set of options that were used to
4885 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
4886 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
4887 needs to build supplementary stub code for constructors to work. On
4888 multi-libbed systems, @samp{gcc -shared} must select the correct support
4889 libraries to link against. Failing to supply the correct flags may lead
4890 to subtle defects. Supplying them in cases where they are not necessary
4893 @item -shared-libgcc
4894 @itemx -static-libgcc
4895 @opindex shared-libgcc
4896 @opindex static-libgcc
4897 On systems that provide @file{libgcc} as a shared library, these options
4898 force the use of either the shared or static version respectively.
4899 If no shared version of @file{libgcc} was built when the compiler was
4900 configured, these options have no effect.
4902 There are several situations in which an application should use the
4903 shared @file{libgcc} instead of the static version. The most common
4904 of these is when the application wishes to throw and catch exceptions
4905 across different shared libraries. In that case, each of the libraries
4906 as well as the application itself should use the shared @file{libgcc}.
4908 Therefore, the G++ and GCJ drivers automatically add
4909 @option{-shared-libgcc} whenever you build a shared library or a main
4910 executable, because C++ and Java programs typically use exceptions, so
4911 this is the right thing to do.
4913 If, instead, you use the GCC driver to create shared libraries, you may
4914 find that they will not always be linked with the shared @file{libgcc}.
4915 If GCC finds, at its configuration time, that you have a GNU linker that
4916 does not support option @option{--eh-frame-hdr}, it will link the shared
4917 version of @file{libgcc} into shared libraries by default. Otherwise,
4918 it will take advantage of the linker and optimize away the linking with
4919 the shared version of @file{libgcc}, linking with the static version of
4920 libgcc by default. This allows exceptions to propagate through such
4921 shared libraries, without incurring relocation costs at library load
4924 However, if a library or main executable is supposed to throw or catch
4925 exceptions, you must link it using the G++ or GCJ driver, as appropriate
4926 for the languages used in the program, or using the option
4927 @option{-shared-libgcc}, such that it is linked with the shared
4932 Bind references to global symbols when building a shared object. Warn
4933 about any unresolved references (unless overridden by the link editor
4934 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
4937 @item -Xlinker @var{option}
4939 Pass @var{option} as an option to the linker. You can use this to
4940 supply system-specific linker options which GCC does not know how to
4943 If you want to pass an option that takes an argument, you must use
4944 @option{-Xlinker} twice, once for the option and once for the argument.
4945 For example, to pass @option{-assert definitions}, you must write
4946 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
4947 @option{-Xlinker "-assert definitions"}, because this passes the entire
4948 string as a single argument, which is not what the linker expects.
4950 @item -Wl,@var{option}
4952 Pass @var{option} as an option to the linker. If @var{option} contains
4953 commas, it is split into multiple options at the commas.
4955 @item -u @var{symbol}
4957 Pretend the symbol @var{symbol} is undefined, to force linking of
4958 library modules to define it. You can use @option{-u} multiple times with
4959 different symbols to force loading of additional library modules.
4962 @node Directory Options
4963 @section Options for Directory Search
4964 @cindex directory options
4965 @cindex options, directory search
4968 These options specify directories to search for header files, for
4969 libraries and for parts of the compiler:
4974 Add the directory @var{dir} to the head of the list of directories to be
4975 searched for header files. This can be used to override a system header
4976 file, substituting your own version, since these directories are
4977 searched before the system header file directories. However, you should
4978 not use this option to add directories that contain vendor-supplied
4979 system header files (use @option{-isystem} for that). If you use more than
4980 one @option{-I} option, the directories are scanned in left-to-right
4981 order; the standard system directories come after.
4983 If a standard system include directory, or a directory specified with
4984 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
4985 option will be ignored. The directory will still be searched but as a
4986 system directory at its normal position in the system include chain.
4987 This is to ensure that GCC's procedure to fix buggy system headers and
4988 the ordering for the include_next directive are not inadvertently changed.
4989 If you really need to change the search order for system directories,
4990 use the @option{-nostdinc} and/or @option{-isystem} options.
4994 Any directories you specify with @option{-I} options before the @option{-I-}
4995 option are searched only for the case of @samp{#include "@var{file}"};
4996 they are not searched for @samp{#include <@var{file}>}.
4998 If additional directories are specified with @option{-I} options after
4999 the @option{-I-}, these directories are searched for all @samp{#include}
5000 directives. (Ordinarily @emph{all} @option{-I} directories are used
5003 In addition, the @option{-I-} option inhibits the use of the current
5004 directory (where the current input file came from) as the first search
5005 directory for @samp{#include "@var{file}"}. There is no way to
5006 override this effect of @option{-I-}. With @option{-I.} you can specify
5007 searching the directory which was current when the compiler was
5008 invoked. That is not exactly the same as what the preprocessor does
5009 by default, but it is often satisfactory.
5011 @option{-I-} does not inhibit the use of the standard system directories
5012 for header files. Thus, @option{-I-} and @option{-nostdinc} are
5017 Add directory @var{dir} to the list of directories to be searched
5020 @item -B@var{prefix}
5022 This option specifies where to find the executables, libraries,
5023 include files, and data files of the compiler itself.
5025 The compiler driver program runs one or more of the subprograms
5026 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
5027 @var{prefix} as a prefix for each program it tries to run, both with and
5028 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
5030 For each subprogram to be run, the compiler driver first tries the
5031 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
5032 was not specified, the driver tries two standard prefixes, which are
5033 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc-lib/}. If neither of
5034 those results in a file name that is found, the unmodified program
5035 name is searched for using the directories specified in your
5036 @env{PATH} environment variable.
5038 The compiler will check to see if the path provided by the @option{-B}
5039 refers to a directory, and if necessary it will add a directory
5040 separator character at the end of the path.
5042 @option{-B} prefixes that effectively specify directory names also apply
5043 to libraries in the linker, because the compiler translates these
5044 options into @option{-L} options for the linker. They also apply to
5045 includes files in the preprocessor, because the compiler translates these
5046 options into @option{-isystem} options for the preprocessor. In this case,
5047 the compiler appends @samp{include} to the prefix.
5049 The run-time support file @file{libgcc.a} can also be searched for using
5050 the @option{-B} prefix, if needed. If it is not found there, the two
5051 standard prefixes above are tried, and that is all. The file is left
5052 out of the link if it is not found by those means.
5054 Another way to specify a prefix much like the @option{-B} prefix is to use
5055 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
5058 As a special kludge, if the path provided by @option{-B} is
5059 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
5060 9, then it will be replaced by @file{[dir/]include}. This is to help
5061 with boot-strapping the compiler.
5063 @item -specs=@var{file}
5065 Process @var{file} after the compiler reads in the standard @file{specs}
5066 file, in order to override the defaults that the @file{gcc} driver
5067 program uses when determining what switches to pass to @file{cc1},
5068 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
5069 @option{-specs=@var{file}} can be specified on the command line, and they
5070 are processed in order, from left to right.
5076 @section Specifying subprocesses and the switches to pass to them
5079 @command{gcc} is a driver program. It performs its job by invoking a
5080 sequence of other programs to do the work of compiling, assembling and
5081 linking. GCC interprets its command-line parameters and uses these to
5082 deduce which programs it should invoke, and which command-line options
5083 it ought to place on their command lines. This behavior is controlled
5084 by @dfn{spec strings}. In most cases there is one spec string for each
5085 program that GCC can invoke, but a few programs have multiple spec
5086 strings to control their behavior. The spec strings built into GCC can
5087 be overridden by using the @option{-specs=} command-line switch to specify
5090 @dfn{Spec files} are plaintext files that are used to construct spec
5091 strings. They consist of a sequence of directives separated by blank
5092 lines. The type of directive is determined by the first non-whitespace
5093 character on the line and it can be one of the following:
5096 @item %@var{command}
5097 Issues a @var{command} to the spec file processor. The commands that can
5101 @item %include <@var{file}>
5103 Search for @var{file} and insert its text at the current point in the
5106 @item %include_noerr <@var{file}>
5107 @cindex %include_noerr
5108 Just like @samp{%include}, but do not generate an error message if the include
5109 file cannot be found.
5111 @item %rename @var{old_name} @var{new_name}
5113 Rename the spec string @var{old_name} to @var{new_name}.
5117 @item *[@var{spec_name}]:
5118 This tells the compiler to create, override or delete the named spec
5119 string. All lines after this directive up to the next directive or
5120 blank line are considered to be the text for the spec string. If this
5121 results in an empty string then the spec will be deleted. (Or, if the
5122 spec did not exist, then nothing will happened.) Otherwise, if the spec
5123 does not currently exist a new spec will be created. If the spec does
5124 exist then its contents will be overridden by the text of this
5125 directive, unless the first character of that text is the @samp{+}
5126 character, in which case the text will be appended to the spec.
5128 @item [@var{suffix}]:
5129 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
5130 and up to the next directive or blank line are considered to make up the
5131 spec string for the indicated suffix. When the compiler encounters an
5132 input file with the named suffix, it will processes the spec string in
5133 order to work out how to compile that file. For example:
5140 This says that any input file whose name ends in @samp{.ZZ} should be
5141 passed to the program @samp{z-compile}, which should be invoked with the
5142 command-line switch @option{-input} and with the result of performing the
5143 @samp{%i} substitution. (See below.)
5145 As an alternative to providing a spec string, the text that follows a
5146 suffix directive can be one of the following:
5149 @item @@@var{language}
5150 This says that the suffix is an alias for a known @var{language}. This is
5151 similar to using the @option{-x} command-line switch to GCC to specify a
5152 language explicitly. For example:
5159 Says that .ZZ files are, in fact, C++ source files.
5162 This causes an error messages saying:
5165 @var{name} compiler not installed on this system.
5169 GCC already has an extensive list of suffixes built into it.
5170 This directive will add an entry to the end of the list of suffixes, but
5171 since the list is searched from the end backwards, it is effectively
5172 possible to override earlier entries using this technique.
5176 GCC has the following spec strings built into it. Spec files can
5177 override these strings or create their own. Note that individual
5178 targets can also add their own spec strings to this list.
5181 asm Options to pass to the assembler
5182 asm_final Options to pass to the assembler post-processor
5183 cpp Options to pass to the C preprocessor
5184 cc1 Options to pass to the C compiler
5185 cc1plus Options to pass to the C++ compiler
5186 endfile Object files to include at the end of the link
5187 link Options to pass to the linker
5188 lib Libraries to include on the command line to the linker
5189 libgcc Decides which GCC support library to pass to the linker
5190 linker Sets the name of the linker
5191 predefines Defines to be passed to the C preprocessor
5192 signed_char Defines to pass to CPP to say whether @code{char} is signed
5194 startfile Object files to include at the start of the link
5197 Here is a small example of a spec file:
5203 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
5206 This example renames the spec called @samp{lib} to @samp{old_lib} and
5207 then overrides the previous definition of @samp{lib} with a new one.
5208 The new definition adds in some extra command-line options before
5209 including the text of the old definition.
5211 @dfn{Spec strings} are a list of command-line options to be passed to their
5212 corresponding program. In addition, the spec strings can contain
5213 @samp{%}-prefixed sequences to substitute variable text or to
5214 conditionally insert text into the command line. Using these constructs
5215 it is possible to generate quite complex command lines.
5217 Here is a table of all defined @samp{%}-sequences for spec
5218 strings. Note that spaces are not generated automatically around the
5219 results of expanding these sequences. Therefore you can concatenate them
5220 together or combine them with constant text in a single argument.
5224 Substitute one @samp{%} into the program name or argument.
5227 Substitute the name of the input file being processed.
5230 Substitute the basename of the input file being processed.
5231 This is the substring up to (and not including) the last period
5232 and not including the directory.
5235 This is the same as @samp{%b}, but include the file suffix (text after
5239 Marks the argument containing or following the @samp{%d} as a
5240 temporary file name, so that that file will be deleted if GCC exits
5241 successfully. Unlike @samp{%g}, this contributes no text to the
5244 @item %g@var{suffix}
5245 Substitute a file name that has suffix @var{suffix} and is chosen
5246 once per compilation, and mark the argument in the same way as
5247 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
5248 name is now chosen in a way that is hard to predict even when previously
5249 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
5250 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
5251 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
5252 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
5253 was simply substituted with a file name chosen once per compilation,
5254 without regard to any appended suffix (which was therefore treated
5255 just like ordinary text), making such attacks more likely to succeed.
5257 @item %u@var{suffix}
5258 Like @samp{%g}, but generates a new temporary file name even if
5259 @samp{%u@var{suffix}} was already seen.
5261 @item %U@var{suffix}
5262 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
5263 new one if there is no such last file name. In the absence of any
5264 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
5265 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
5266 would involve the generation of two distinct file names, one
5267 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
5268 simply substituted with a file name chosen for the previous @samp{%u},
5269 without regard to any appended suffix.
5271 @item %j@var{suffix}
5272 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
5273 writable, and if save-temps is off; otherwise, substitute the name
5274 of a temporary file, just like @samp{%u}. This temporary file is not
5275 meant for communication between processes, but rather as a junk
5278 @item %|@var{suffix}
5279 @itemx %m@var{suffix}
5280 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
5281 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
5282 all. These are the two most common ways to instruct a program that it
5283 should read from standard input or write to standard output. If you
5284 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
5285 construct: see for example @file{f/lang-specs.h}.
5287 @item %.@var{SUFFIX}
5288 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
5289 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
5290 terminated by the next space or %.
5293 Marks the argument containing or following the @samp{%w} as the
5294 designated output file of this compilation. This puts the argument
5295 into the sequence of arguments that @samp{%o} will substitute later.
5298 Substitutes the names of all the output files, with spaces
5299 automatically placed around them. You should write spaces
5300 around the @samp{%o} as well or the results are undefined.
5301 @samp{%o} is for use in the specs for running the linker.
5302 Input files whose names have no recognized suffix are not compiled
5303 at all, but they are included among the output files, so they will
5307 Substitutes the suffix for object files. Note that this is
5308 handled specially when it immediately follows @samp{%g, %u, or %U},
5309 because of the need for those to form complete file names. The
5310 handling is such that @samp{%O} is treated exactly as if it had already
5311 been substituted, except that @samp{%g, %u, and %U} do not currently
5312 support additional @var{suffix} characters following @samp{%O} as they would
5313 following, for example, @samp{.o}.
5316 Substitutes the standard macro predefinitions for the
5317 current target machine. Use this when running @code{cpp}.
5320 Like @samp{%p}, but puts @samp{__} before and after the name of each
5321 predefined macro, except for macros that start with @samp{__} or with
5322 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
5326 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
5327 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}), and
5328 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
5332 Current argument is the name of a library or startup file of some sort.
5333 Search for that file in a standard list of directories and substitute
5334 the full name found.
5337 Print @var{str} as an error message. @var{str} is terminated by a newline.
5338 Use this when inconsistent options are detected.
5341 Substitute the contents of spec string @var{name} at this point.
5344 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
5346 @item %x@{@var{option}@}
5347 Accumulate an option for @samp{%X}.
5350 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
5354 Output the accumulated assembler options specified by @option{-Wa}.
5357 Output the accumulated preprocessor options specified by @option{-Wp}.
5360 Process the @code{asm} spec. This is used to compute the
5361 switches to be passed to the assembler.
5364 Process the @code{asm_final} spec. This is a spec string for
5365 passing switches to an assembler post-processor, if such a program is
5369 Process the @code{link} spec. This is the spec for computing the
5370 command line passed to the linker. Typically it will make use of the
5371 @samp{%L %G %S %D and %E} sequences.
5374 Dump out a @option{-L} option for each directory that GCC believes might
5375 contain startup files. If the target supports multilibs then the
5376 current multilib directory will be prepended to each of these paths.
5379 Output the multilib directory with directory separators replaced with
5380 @samp{_}. If multilib directories are not set, or the multilib directory is
5381 @file{.} then this option emits nothing.
5384 Process the @code{lib} spec. This is a spec string for deciding which
5385 libraries should be included on the command line to the linker.
5388 Process the @code{libgcc} spec. This is a spec string for deciding
5389 which GCC support library should be included on the command line to the linker.
5392 Process the @code{startfile} spec. This is a spec for deciding which
5393 object files should be the first ones passed to the linker. Typically
5394 this might be a file named @file{crt0.o}.
5397 Process the @code{endfile} spec. This is a spec string that specifies
5398 the last object files that will be passed to the linker.
5401 Process the @code{cpp} spec. This is used to construct the arguments
5402 to be passed to the C preprocessor.
5405 Process the @code{signed_char} spec. This is intended to be used
5406 to tell cpp whether a char is signed. It typically has the definition:
5408 %@{funsigned-char:-D__CHAR_UNSIGNED__@}
5412 Process the @code{cc1} spec. This is used to construct the options to be
5413 passed to the actual C compiler (@samp{cc1}).
5416 Process the @code{cc1plus} spec. This is used to construct the options to be
5417 passed to the actual C++ compiler (@samp{cc1plus}).
5420 Substitute the variable part of a matched option. See below.
5421 Note that each comma in the substituted string is replaced by
5425 Remove all occurrences of @code{-S} from the command line. Note---this
5426 command is position dependent. @samp{%} commands in the spec string
5427 before this one will see @code{-S}, @samp{%} commands in the spec string
5428 after this one will not.
5430 @item %:@var{function}(@var{args})
5431 Call the named function @var{function}, passing it @var{args}.
5432 @var{args} is first processed as a nested spec string, then split
5433 into an argument vector in the usual fashion. The function returns
5434 a string which is processed as if it had appeared literally as part
5435 of the current spec.
5437 The following built-in spec functions are provided:
5440 @item @code{if-exists}
5441 The @code{if-exists} spec function takes one argument, an absolute
5442 pathname to a file. If the file exists, @code{if-exists} returns the
5443 pathname. Here is a small example of its usage:
5447 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
5450 @item @code{if-exists-else}
5451 The @code{if-exists-else} spec function is similar to the @code{if-exists}
5452 spec function, except that it takes two arguments. The first argument is
5453 an absolute pathname to a file. If the file exists, @code{if-exists-else}
5454 returns the pathname. If it does not exist, it returns the second argument.
5455 This way, @code{if-exists-else} can be used to select one file or another,
5456 based on the existence of the first. Here is a small example of its usage:
5460 crt0%O%s %:if-exists(crti%O%s) \
5461 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
5466 Substitutes the @code{-S} switch, if that switch was given to GCC@.
5467 If that switch was not specified, this substitutes nothing. Note that
5468 the leading dash is omitted when specifying this option, and it is
5469 automatically inserted if the substitution is performed. Thus the spec
5470 string @samp{%@{foo@}} would match the command-line option @option{-foo}
5471 and would output the command line option @option{-foo}.
5473 @item %W@{@code{S}@}
5474 Like %@{@code{S}@} but mark last argument supplied within as a file to be
5477 @item %@{@code{S}*@}
5478 Substitutes all the switches specified to GCC whose names start
5479 with @code{-S}, but which also take an argument. This is used for
5480 switches like @option{-o}, @option{-D}, @option{-I}, etc.
5481 GCC considers @option{-o foo} as being
5482 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
5483 text, including the space. Thus two arguments would be generated.
5485 @item %@{@code{S}*&@code{T}*@}
5486 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
5487 (the order of @code{S} and @code{T} in the spec is not significant).
5488 There can be any number of ampersand-separated variables; for each the
5489 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
5491 @item %@{@code{S}:@code{X}@}
5492 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
5494 @item %@{!@code{S}:@code{X}@}
5495 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
5497 @item %@{@code{S}*:@code{X}@}
5498 Substitutes @code{X} if one or more switches whose names start with
5499 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
5500 once, no matter how many such switches appeared. However, if @code{%*}
5501 appears somewhere in @code{X}, then @code{X} will be substituted once
5502 for each matching switch, with the @code{%*} replaced by the part of
5503 that switch that matched the @code{*}.
5505 @item %@{.@code{S}:@code{X}@}
5506 Substitutes @code{X}, if processing a file with suffix @code{S}.
5508 @item %@{!.@code{S}:@code{X}@}
5509 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
5511 @item %@{@code{S}|@code{P}:@code{X}@}
5512 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC@.
5513 This may be combined with @samp{!}, @samp{.}, and @code{*} sequences as well,
5514 although they have a stronger binding than the @samp{|}. If @code{%*}
5515 appears in @code{X}, all of the alternatives must be starred, and only
5516 the first matching alternative is substituted.
5518 For example, a spec string like this:
5521 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
5524 will output the following command-line options from the following input
5525 command-line options:
5530 -d fred.c -foo -baz -boggle
5531 -d jim.d -bar -baz -boggle
5534 @item %@{S:X; T:Y; :D@}
5536 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
5537 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
5538 be as many clauses as you need. This may be combined with @code{.},
5539 @code{!}, @code{|}, and @code{*} as needed.
5544 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
5545 construct may contain other nested @samp{%} constructs or spaces, or
5546 even newlines. They are processed as usual, as described above.
5547 Trailing white space in @code{X} is ignored. White space may also
5548 appear anywhere on the left side of the colon in these constructs,
5549 except between @code{.} or @code{*} and the corresponding word.
5551 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
5552 handled specifically in these constructs. If another value of
5553 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
5554 @option{-W} switch is found later in the command line, the earlier
5555 switch value is ignored, except with @{@code{S}*@} where @code{S} is
5556 just one letter, which passes all matching options.
5558 The character @samp{|} at the beginning of the predicate text is used to
5559 indicate that a command should be piped to the following command, but
5560 only if @option{-pipe} is specified.
5562 It is built into GCC which switches take arguments and which do not.
5563 (You might think it would be useful to generalize this to allow each
5564 compiler's spec to say which switches take arguments. But this cannot
5565 be done in a consistent fashion. GCC cannot even decide which input
5566 files have been specified without knowing which switches take arguments,
5567 and it must know which input files to compile in order to tell which
5570 GCC also knows implicitly that arguments starting in @option{-l} are to be
5571 treated as compiler output files, and passed to the linker in their
5572 proper position among the other output files.
5574 @c man begin OPTIONS
5576 @node Target Options
5577 @section Specifying Target Machine and Compiler Version
5578 @cindex target options
5579 @cindex cross compiling
5580 @cindex specifying machine version
5581 @cindex specifying compiler version and target machine
5582 @cindex compiler version, specifying
5583 @cindex target machine, specifying
5585 The usual way to run GCC is to run the executable called @file{gcc}, or
5586 @file{<machine>-gcc} when cross-compiling, or
5587 @file{<machine>-gcc-<version>} to run a version other than the one that
5588 was installed last. Sometimes this is inconvenient, so GCC provides
5589 options that will switch to another cross-compiler or version.
5592 @item -b @var{machine}
5594 The argument @var{machine} specifies the target machine for compilation.
5596 The value to use for @var{machine} is the same as was specified as the
5597 machine type when configuring GCC as a cross-compiler. For
5598 example, if a cross-compiler was configured with @samp{configure
5599 i386v}, meaning to compile for an 80386 running System V, then you
5600 would specify @option{-b i386v} to run that cross compiler.
5602 @item -V @var{version}
5604 The argument @var{version} specifies which version of GCC to run.
5605 This is useful when multiple versions are installed. For example,
5606 @var{version} might be @samp{2.0}, meaning to run GCC version 2.0.
5609 The @option{-V} and @option{-b} options work by running the
5610 @file{<machine>-gcc-<version>} executable, so there's no real reason to
5611 use them if you can just run that directly.
5613 @node Submodel Options
5614 @section Hardware Models and Configurations
5615 @cindex submodel options
5616 @cindex specifying hardware config
5617 @cindex hardware models and configurations, specifying
5618 @cindex machine dependent options
5620 Earlier we discussed the standard option @option{-b} which chooses among
5621 different installed compilers for completely different target
5622 machines, such as VAX vs.@: 68000 vs.@: 80386.
5624 In addition, each of these target machine types can have its own
5625 special options, starting with @samp{-m}, to choose among various
5626 hardware models or configurations---for example, 68010 vs 68020,
5627 floating coprocessor or none. A single installed version of the
5628 compiler can compile for any model or configuration, according to the
5631 Some configurations of the compiler also support additional special
5632 options, usually for compatibility with other compilers on the same
5635 These options are defined by the macro @code{TARGET_SWITCHES} in the
5636 machine description. The default for the options is also defined by
5637 that macro, which enables you to change the defaults.
5649 * RS/6000 and PowerPC Options::
5653 * i386 and x86-64 Options::
5655 * Intel 960 Options::
5656 * DEC Alpha Options::
5657 * DEC Alpha/VMS Options::
5660 * System V Options::
5661 * TMS320C3x/C4x Options::
5669 * S/390 and zSeries Options::
5673 * Xstormy16 Options::
5678 @node M680x0 Options
5679 @subsection M680x0 Options
5680 @cindex M680x0 options
5682 These are the @samp{-m} options defined for the 68000 series. The default
5683 values for these options depends on which style of 68000 was selected when
5684 the compiler was configured; the defaults for the most common choices are
5692 Generate output for a 68000. This is the default
5693 when the compiler is configured for 68000-based systems.
5695 Use this option for microcontrollers with a 68000 or EC000 core,
5696 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
5702 Generate output for a 68020. This is the default
5703 when the compiler is configured for 68020-based systems.
5707 Generate output containing 68881 instructions for floating point.
5708 This is the default for most 68020 systems unless @option{--nfp} was
5709 specified when the compiler was configured.
5713 Generate output for a 68030. This is the default when the compiler is
5714 configured for 68030-based systems.
5718 Generate output for a 68040. This is the default when the compiler is
5719 configured for 68040-based systems.
5721 This option inhibits the use of 68881/68882 instructions that have to be
5722 emulated by software on the 68040. Use this option if your 68040 does not
5723 have code to emulate those instructions.
5727 Generate output for a 68060. This is the default when the compiler is
5728 configured for 68060-based systems.
5730 This option inhibits the use of 68020 and 68881/68882 instructions that
5731 have to be emulated by software on the 68060. Use this option if your 68060
5732 does not have code to emulate those instructions.
5736 Generate output for a CPU32. This is the default
5737 when the compiler is configured for CPU32-based systems.
5739 Use this option for microcontrollers with a
5740 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
5741 68336, 68340, 68341, 68349 and 68360.
5745 Generate output for a 520X ``coldfire'' family cpu. This is the default
5746 when the compiler is configured for 520X-based systems.
5748 Use this option for microcontroller with a 5200 core, including
5749 the MCF5202, MCF5203, MCF5204 and MCF5202.
5754 Generate output for a 68040, without using any of the new instructions.
5755 This results in code which can run relatively efficiently on either a
5756 68020/68881 or a 68030 or a 68040. The generated code does use the
5757 68881 instructions that are emulated on the 68040.
5761 Generate output for a 68060, without using any of the new instructions.
5762 This results in code which can run relatively efficiently on either a
5763 68020/68881 or a 68030 or a 68040. The generated code does use the
5764 68881 instructions that are emulated on the 68060.
5767 @opindex msoft-float
5768 Generate output containing library calls for floating point.
5769 @strong{Warning:} the requisite libraries are not available for all m68k
5770 targets. Normally the facilities of the machine's usual C compiler are
5771 used, but this can't be done directly in cross-compilation. You must
5772 make your own arrangements to provide suitable library functions for
5773 cross-compilation. The embedded targets @samp{m68k-*-aout} and
5774 @samp{m68k-*-coff} do provide software floating point support.
5778 Consider type @code{int} to be 16 bits wide, like @code{short int}.
5781 @opindex mnobitfield
5782 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
5783 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
5787 Do use the bit-field instructions. The @option{-m68020} option implies
5788 @option{-mbitfield}. This is the default if you use a configuration
5789 designed for a 68020.
5793 Use a different function-calling convention, in which functions
5794 that take a fixed number of arguments return with the @code{rtd}
5795 instruction, which pops their arguments while returning. This
5796 saves one instruction in the caller since there is no need to pop
5797 the arguments there.
5799 This calling convention is incompatible with the one normally
5800 used on Unix, so you cannot use it if you need to call libraries
5801 compiled with the Unix compiler.
5803 Also, you must provide function prototypes for all functions that
5804 take variable numbers of arguments (including @code{printf});
5805 otherwise incorrect code will be generated for calls to those
5808 In addition, seriously incorrect code will result if you call a
5809 function with too many arguments. (Normally, extra arguments are
5810 harmlessly ignored.)
5812 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
5813 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
5816 @itemx -mno-align-int
5818 @opindex mno-align-int
5819 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
5820 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
5821 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
5822 Aligning variables on 32-bit boundaries produces code that runs somewhat
5823 faster on processors with 32-bit busses at the expense of more memory.
5825 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
5826 align structures containing the above types differently than
5827 most published application binary interface specifications for the m68k.
5831 Use the pc-relative addressing mode of the 68000 directly, instead of
5832 using a global offset table. At present, this option implies @option{-fpic},
5833 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
5834 not presently supported with @option{-mpcrel}, though this could be supported for
5835 68020 and higher processors.
5837 @item -mno-strict-align
5838 @itemx -mstrict-align
5839 @opindex mno-strict-align
5840 @opindex mstrict-align
5841 Do not (do) assume that unaligned memory references will be handled by
5846 @node M68hc1x Options
5847 @subsection M68hc1x Options
5848 @cindex M68hc1x options
5850 These are the @samp{-m} options defined for the 68hc11 and 68hc12
5851 microcontrollers. The default values for these options depends on
5852 which style of microcontroller was selected when the compiler was configured;
5853 the defaults for the most common choices are given below.
5860 Generate output for a 68HC11. This is the default
5861 when the compiler is configured for 68HC11-based systems.
5867 Generate output for a 68HC12. This is the default
5868 when the compiler is configured for 68HC12-based systems.
5874 Generate output for a 68HCS12.
5877 @opindex mauto-incdec
5878 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
5885 Enable the use of 68HC12 min and max instructions.
5888 @itemx -mno-long-calls
5889 @opindex mlong-calls
5890 @opindex mno-long-calls
5891 Treat all calls as being far away (near). If calls are assumed to be
5892 far away, the compiler will use the @code{call} instruction to
5893 call a function and the @code{rtc} instruction for returning.
5897 Consider type @code{int} to be 16 bits wide, like @code{short int}.
5899 @item -msoft-reg-count=@var{count}
5900 @opindex msoft-reg-count
5901 Specify the number of pseudo-soft registers which are used for the
5902 code generation. The maximum number is 32. Using more pseudo-soft
5903 register may or may not result in better code depending on the program.
5904 The default is 4 for 68HC11 and 2 for 68HC12.
5909 @subsection VAX Options
5912 These @samp{-m} options are defined for the VAX:
5917 Do not output certain jump instructions (@code{aobleq} and so on)
5918 that the Unix assembler for the VAX cannot handle across long
5923 Do output those jump instructions, on the assumption that you
5924 will assemble with the GNU assembler.
5928 Output code for g-format floating point numbers instead of d-format.
5932 @subsection SPARC Options
5933 @cindex SPARC options
5935 These @samp{-m} switches are supported on the SPARC:
5940 @opindex mno-app-regs
5942 Specify @option{-mapp-regs} to generate output using the global registers
5943 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
5946 To be fully SVR4 ABI compliant at the cost of some performance loss,
5947 specify @option{-mno-app-regs}. You should compile libraries and system
5948 software with this option.
5953 @opindex mhard-float
5954 Generate output containing floating point instructions. This is the
5960 @opindex msoft-float
5961 Generate output containing library calls for floating point.
5962 @strong{Warning:} the requisite libraries are not available for all SPARC
5963 targets. Normally the facilities of the machine's usual C compiler are
5964 used, but this cannot be done directly in cross-compilation. You must make
5965 your own arrangements to provide suitable library functions for
5966 cross-compilation. The embedded targets @samp{sparc-*-aout} and
5967 @samp{sparclite-*-*} do provide software floating point support.
5969 @option{-msoft-float} changes the calling convention in the output file;
5970 therefore, it is only useful if you compile @emph{all} of a program with
5971 this option. In particular, you need to compile @file{libgcc.a}, the
5972 library that comes with GCC, with @option{-msoft-float} in order for
5975 @item -mhard-quad-float
5976 @opindex mhard-quad-float
5977 Generate output containing quad-word (long double) floating point
5981 @opindex mimpure-text
5982 @option{-mimpure-text}, used in addition to @option{-shared}, tells
5983 the compiler to not pass @option{-z text} to the linker when linking a
5984 shared object. Using this option, you can link position-dependent
5985 code into a shared object.
5987 @option{-mimpure-text} suppresses the ``relocations remain against
5988 allocatable but non-writable sections'' linker error message.
5989 However, the necessary relocations will trigger copy-on-write, and the
5990 shared object is not actually shared across processes. Instead of
5991 using @option{-mimpure-text}, you should compile all source code with
5992 @option{-fpic} or @option{-fPIC}.
5994 This option is only available on SunOS and Solaris.
5996 @item -msoft-quad-float
5997 @opindex msoft-quad-float
5998 Generate output containing library calls for quad-word (long double)
5999 floating point instructions. The functions called are those specified
6000 in the SPARC ABI@. This is the default.
6002 As of this writing, there are no sparc implementations that have hardware
6003 support for the quad-word floating point instructions. They all invoke
6004 a trap handler for one of these instructions, and then the trap handler
6005 emulates the effect of the instruction. Because of the trap handler overhead,
6006 this is much slower than calling the ABI library routines. Thus the
6007 @option{-msoft-quad-float} option is the default.
6013 With @option{-mflat}, the compiler does not generate save/restore instructions
6014 and will use a ``flat'' or single register window calling convention.
6015 This model uses %i7 as the frame pointer and is compatible with the normal
6016 register window model. Code from either may be intermixed.
6017 The local registers and the input registers (0--5) are still treated as
6018 ``call saved'' registers and will be saved on the stack as necessary.
6020 With @option{-mno-flat} (the default), the compiler emits save/restore
6021 instructions (except for leaf functions) and is the normal mode of operation.
6023 @item -mno-unaligned-doubles
6024 @itemx -munaligned-doubles
6025 @opindex mno-unaligned-doubles
6026 @opindex munaligned-doubles
6027 Assume that doubles have 8 byte alignment. This is the default.
6029 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
6030 alignment only if they are contained in another type, or if they have an
6031 absolute address. Otherwise, it assumes they have 4 byte alignment.
6032 Specifying this option avoids some rare compatibility problems with code
6033 generated by other compilers. It is not the default because it results
6034 in a performance loss, especially for floating point code.
6036 @item -mno-faster-structs
6037 @itemx -mfaster-structs
6038 @opindex mno-faster-structs
6039 @opindex mfaster-structs
6040 With @option{-mfaster-structs}, the compiler assumes that structures
6041 should have 8 byte alignment. This enables the use of pairs of
6042 @code{ldd} and @code{std} instructions for copies in structure
6043 assignment, in place of twice as many @code{ld} and @code{st} pairs.
6044 However, the use of this changed alignment directly violates the SPARC
6045 ABI@. Thus, it's intended only for use on targets where the developer
6046 acknowledges that their resulting code will not be directly in line with
6047 the rules of the ABI@.
6053 These two options select variations on the SPARC architecture.
6055 By default (unless specifically configured for the Fujitsu SPARClite),
6056 GCC generates code for the v7 variant of the SPARC architecture.
6058 @option{-mv8} will give you SPARC v8 code. The only difference from v7
6059 code is that the compiler emits the integer multiply and integer
6060 divide instructions which exist in SPARC v8 but not in SPARC v7.
6062 @option{-msparclite} will give you SPARClite code. This adds the integer
6063 multiply, integer divide step and scan (@code{ffs}) instructions which
6064 exist in SPARClite but not in SPARC v7.
6066 These options are deprecated and will be deleted in a future GCC release.
6067 They have been replaced with @option{-mcpu=xxx}.
6072 @opindex msupersparc
6073 These two options select the processor for which the code is optimized.
6075 With @option{-mcypress} (the default), the compiler optimizes code for the
6076 Cypress CY7C602 chip, as used in the SPARCStation/SPARCServer 3xx series.
6077 This is also appropriate for the older SPARCStation 1, 2, IPX etc.
6079 With @option{-msupersparc} the compiler optimizes code for the SuperSPARC cpu, as
6080 used in the SPARCStation 10, 1000 and 2000 series. This flag also enables use
6081 of the full SPARC v8 instruction set.
6083 These options are deprecated and will be deleted in a future GCC release.
6084 They have been replaced with @option{-mcpu=xxx}.
6086 @item -mcpu=@var{cpu_type}
6088 Set the instruction set, register set, and instruction scheduling parameters
6089 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
6090 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
6091 @samp{hypersparc}, @samp{sparclite86x}, @samp{f930}, @samp{f934},
6092 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc}, and
6095 Default instruction scheduling parameters are used for values that select
6096 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
6097 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
6099 Here is a list of each supported architecture and their supported
6104 v8: supersparc, hypersparc
6105 sparclite: f930, f934, sparclite86x
6107 v9: ultrasparc, ultrasparc3
6110 @item -mtune=@var{cpu_type}
6112 Set the instruction scheduling parameters for machine type
6113 @var{cpu_type}, but do not set the instruction set or register set that the
6114 option @option{-mcpu=@var{cpu_type}} would.
6116 The same values for @option{-mcpu=@var{cpu_type}} can be used for
6117 @option{-mtune=@var{cpu_type}}, but the only useful values are those
6118 that select a particular cpu implementation. Those are @samp{cypress},
6119 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
6120 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc}, and
6125 These @samp{-m} switches are supported in addition to the above
6126 on the SPARCLET processor.
6129 @item -mlittle-endian
6130 @opindex mlittle-endian
6131 Generate code for a processor running in little-endian mode.
6135 Treat register @code{%g0} as a normal register.
6136 GCC will continue to clobber it as necessary but will not assume
6137 it always reads as 0.
6139 @item -mbroken-saverestore
6140 @opindex mbroken-saverestore
6141 Generate code that does not use non-trivial forms of the @code{save} and
6142 @code{restore} instructions. Early versions of the SPARCLET processor do
6143 not correctly handle @code{save} and @code{restore} instructions used with
6144 arguments. They correctly handle them used without arguments. A @code{save}
6145 instruction used without arguments increments the current window pointer
6146 but does not allocate a new stack frame. It is assumed that the window
6147 overflow trap handler will properly handle this case as will interrupt
6151 These @samp{-m} switches are supported in addition to the above
6152 on SPARC V9 processors in 64-bit environments.
6155 @item -mlittle-endian
6156 @opindex mlittle-endian
6157 Generate code for a processor running in little-endian mode.
6163 Generate code for a 32-bit or 64-bit environment.
6164 The 32-bit environment sets int, long and pointer to 32 bits.
6165 The 64-bit environment sets int to 32 bits and long and pointer
6168 @item -mcmodel=medlow
6169 @opindex mcmodel=medlow
6170 Generate code for the Medium/Low code model: the program must be linked
6171 in the low 32 bits of the address space. Pointers are 64 bits.
6172 Programs can be statically or dynamically linked.
6174 @item -mcmodel=medmid
6175 @opindex mcmodel=medmid
6176 Generate code for the Medium/Middle code model: the program must be linked
6177 in the low 44 bits of the address space, the text segment must be less than
6178 2G bytes, and data segment must be within 2G of the text segment.
6179 Pointers are 64 bits.
6181 @item -mcmodel=medany
6182 @opindex mcmodel=medany
6183 Generate code for the Medium/Anywhere code model: the program may be linked
6184 anywhere in the address space, the text segment must be less than
6185 2G bytes, and data segment must be within 2G of the text segment.
6186 Pointers are 64 bits.
6188 @item -mcmodel=embmedany
6189 @opindex mcmodel=embmedany
6190 Generate code for the Medium/Anywhere code model for embedded systems:
6191 assume a 32-bit text and a 32-bit data segment, both starting anywhere
6192 (determined at link time). Register %g4 points to the base of the
6193 data segment. Pointers are still 64 bits.
6194 Programs are statically linked, PIC is not supported.
6197 @itemx -mno-stack-bias
6198 @opindex mstack-bias
6199 @opindex mno-stack-bias
6200 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
6201 frame pointer if present, are offset by @minus{}2047 which must be added back
6202 when making stack frame references.
6203 Otherwise, assume no such offset is present.
6207 @subsection ARM Options
6210 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
6215 @opindex mapcs-frame
6216 Generate a stack frame that is compliant with the ARM Procedure Call
6217 Standard for all functions, even if this is not strictly necessary for
6218 correct execution of the code. Specifying @option{-fomit-frame-pointer}
6219 with this option will cause the stack frames not to be generated for
6220 leaf functions. The default is @option{-mno-apcs-frame}.
6224 This is a synonym for @option{-mapcs-frame}.
6228 Generate code for a processor running with a 26-bit program counter,
6229 and conforming to the function calling standards for the APCS 26-bit
6230 option. This option replaces the @option{-m2} and @option{-m3} options
6231 of previous releases of the compiler.
6235 Generate code for a processor running with a 32-bit program counter,
6236 and conforming to the function calling standards for the APCS 32-bit
6237 option. This option replaces the @option{-m6} option of previous releases
6241 @c not currently implemented
6242 @item -mapcs-stack-check
6243 @opindex mapcs-stack-check
6244 Generate code to check the amount of stack space available upon entry to
6245 every function (that actually uses some stack space). If there is
6246 insufficient space available then either the function
6247 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
6248 called, depending upon the amount of stack space required. The run time
6249 system is required to provide these functions. The default is
6250 @option{-mno-apcs-stack-check}, since this produces smaller code.
6252 @c not currently implemented
6254 @opindex mapcs-float
6255 Pass floating point arguments using the float point registers. This is
6256 one of the variants of the APCS@. This option is recommended if the
6257 target hardware has a floating point unit or if a lot of floating point
6258 arithmetic is going to be performed by the code. The default is
6259 @option{-mno-apcs-float}, since integer only code is slightly increased in
6260 size if @option{-mapcs-float} is used.
6262 @c not currently implemented
6263 @item -mapcs-reentrant
6264 @opindex mapcs-reentrant
6265 Generate reentrant, position independent code. The default is
6266 @option{-mno-apcs-reentrant}.
6269 @item -mthumb-interwork
6270 @opindex mthumb-interwork
6271 Generate code which supports calling between the ARM and Thumb
6272 instruction sets. Without this option the two instruction sets cannot
6273 be reliably used inside one program. The default is
6274 @option{-mno-thumb-interwork}, since slightly larger code is generated
6275 when @option{-mthumb-interwork} is specified.
6277 @item -mno-sched-prolog
6278 @opindex mno-sched-prolog
6279 Prevent the reordering of instructions in the function prolog, or the
6280 merging of those instruction with the instructions in the function's
6281 body. This means that all functions will start with a recognizable set
6282 of instructions (or in fact one of a choice from a small set of
6283 different function prologues), and this information can be used to
6284 locate the start if functions inside an executable piece of code. The
6285 default is @option{-msched-prolog}.
6288 @opindex mhard-float
6289 Generate output containing floating point instructions. This is the
6293 @opindex msoft-float
6294 Generate output containing library calls for floating point.
6295 @strong{Warning:} the requisite libraries are not available for all ARM
6296 targets. Normally the facilities of the machine's usual C compiler are
6297 used, but this cannot be done directly in cross-compilation. You must make
6298 your own arrangements to provide suitable library functions for
6301 @option{-msoft-float} changes the calling convention in the output file;
6302 therefore, it is only useful if you compile @emph{all} of a program with
6303 this option. In particular, you need to compile @file{libgcc.a}, the
6304 library that comes with GCC, with @option{-msoft-float} in order for
6307 @item -mlittle-endian
6308 @opindex mlittle-endian
6309 Generate code for a processor running in little-endian mode. This is
6310 the default for all standard configurations.
6313 @opindex mbig-endian
6314 Generate code for a processor running in big-endian mode; the default is
6315 to compile code for a little-endian processor.
6317 @item -mwords-little-endian
6318 @opindex mwords-little-endian
6319 This option only applies when generating code for big-endian processors.
6320 Generate code for a little-endian word order but a big-endian byte
6321 order. That is, a byte order of the form @samp{32107654}. Note: this
6322 option should only be used if you require compatibility with code for
6323 big-endian ARM processors generated by versions of the compiler prior to
6326 @item -malignment-traps
6327 @opindex malignment-traps
6328 Generate code that will not trap if the MMU has alignment traps enabled.
6329 On ARM architectures prior to ARMv4, there were no instructions to
6330 access half-word objects stored in memory. However, when reading from
6331 memory a feature of the ARM architecture allows a word load to be used,
6332 even if the address is unaligned, and the processor core will rotate the
6333 data as it is being loaded. This option tells the compiler that such
6334 misaligned accesses will cause a MMU trap and that it should instead
6335 synthesize the access as a series of byte accesses. The compiler can
6336 still use word accesses to load half-word data if it knows that the
6337 address is aligned to a word boundary.
6339 This option is ignored when compiling for ARM architecture 4 or later,
6340 since these processors have instructions to directly access half-word
6343 @item -mno-alignment-traps
6344 @opindex mno-alignment-traps
6345 Generate code that assumes that the MMU will not trap unaligned
6346 accesses. This produces better code when the target instruction set
6347 does not have half-word memory operations (i.e.@: implementations prior to
6350 Note that you cannot use this option to access unaligned word objects,
6351 since the processor will only fetch one 32-bit aligned object from
6354 The default setting for most targets is @option{-mno-alignment-traps}, since
6355 this produces better code when there are no half-word memory
6356 instructions available.
6358 @item -mshort-load-bytes
6359 @itemx -mno-short-load-words
6360 @opindex mshort-load-bytes
6361 @opindex mno-short-load-words
6362 These are deprecated aliases for @option{-malignment-traps}.
6364 @item -mno-short-load-bytes
6365 @itemx -mshort-load-words
6366 @opindex mno-short-load-bytes
6367 @opindex mshort-load-words
6368 This are deprecated aliases for @option{-mno-alignment-traps}.
6370 @item -mcpu=@var{name}
6372 This specifies the name of the target ARM processor. GCC uses this name
6373 to determine what kind of instructions it can emit when generating
6374 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
6375 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
6376 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
6377 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
6378 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
6379 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm8},
6380 @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
6381 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
6382 @samp{arm920t}, @samp{arm940t}, @samp{arm9tdmi}, @samp{arm10tdmi},
6383 @samp{arm1020t}, @samp{xscale}, @samp{iwmmxt}, @samp{ep9312}.
6385 @itemx -mtune=@var{name}
6387 This option is very similar to the @option{-mcpu=} option, except that
6388 instead of specifying the actual target processor type, and hence
6389 restricting which instructions can be used, it specifies that GCC should
6390 tune the performance of the code as if the target were of the type
6391 specified in this option, but still choosing the instructions that it
6392 will generate based on the cpu specified by a @option{-mcpu=} option.
6393 For some ARM implementations better performance can be obtained by using
6396 @item -march=@var{name}
6398 This specifies the name of the target ARM architecture. GCC uses this
6399 name to determine what kind of instructions it can emit when generating
6400 assembly code. This option can be used in conjunction with or instead
6401 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
6402 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
6403 @samp{armv5}, @samp{armv5t}, @samp{armv5te}, @samp{iwmmxt}, @samp{ep9312}.
6405 @item -mfpe=@var{number}
6406 @itemx -mfp=@var{number}
6409 This specifies the version of the floating point emulation available on
6410 the target. Permissible values are 2 and 3. @option{-mfp=} is a synonym
6411 for @option{-mfpe=}, for compatibility with older versions of GCC@.
6413 @item -mstructure-size-boundary=@var{n}
6414 @opindex mstructure-size-boundary
6415 The size of all structures and unions will be rounded up to a multiple
6416 of the number of bits set by this option. Permissible values are 8 and
6417 32. The default value varies for different toolchains. For the COFF
6418 targeted toolchain the default value is 8. Specifying the larger number
6419 can produce faster, more efficient code, but can also increase the size
6420 of the program. The two values are potentially incompatible. Code
6421 compiled with one value cannot necessarily expect to work with code or
6422 libraries compiled with the other value, if they exchange information
6423 using structures or unions.
6425 @item -mabort-on-noreturn
6426 @opindex mabort-on-noreturn
6427 Generate a call to the function @code{abort} at the end of a
6428 @code{noreturn} function. It will be executed if the function tries to
6432 @itemx -mno-long-calls
6433 @opindex mlong-calls
6434 @opindex mno-long-calls
6435 Tells the compiler to perform function calls by first loading the
6436 address of the function into a register and then performing a subroutine
6437 call on this register. This switch is needed if the target function
6438 will lie outside of the 64 megabyte addressing range of the offset based
6439 version of subroutine call instruction.
6441 Even if this switch is enabled, not all function calls will be turned
6442 into long calls. The heuristic is that static functions, functions
6443 which have the @samp{short-call} attribute, functions that are inside
6444 the scope of a @samp{#pragma no_long_calls} directive and functions whose
6445 definitions have already been compiled within the current compilation
6446 unit, will not be turned into long calls. The exception to this rule is
6447 that weak function definitions, functions with the @samp{long-call}
6448 attribute or the @samp{section} attribute, and functions that are within
6449 the scope of a @samp{#pragma long_calls} directive, will always be
6450 turned into long calls.
6452 This feature is not enabled by default. Specifying
6453 @option{-mno-long-calls} will restore the default behavior, as will
6454 placing the function calls within the scope of a @samp{#pragma
6455 long_calls_off} directive. Note these switches have no effect on how
6456 the compiler generates code to handle function calls via function
6459 @item -mnop-fun-dllimport
6460 @opindex mnop-fun-dllimport
6461 Disable support for the @code{dllimport} attribute.
6463 @item -msingle-pic-base
6464 @opindex msingle-pic-base
6465 Treat the register used for PIC addressing as read-only, rather than
6466 loading it in the prologue for each function. The run-time system is
6467 responsible for initializing this register with an appropriate value
6468 before execution begins.
6470 @item -mpic-register=@var{reg}
6471 @opindex mpic-register
6472 Specify the register to be used for PIC addressing. The default is R10
6473 unless stack-checking is enabled, when R9 is used.
6475 @item -mcirrus-fix-invalid-insns
6476 @opindex mcirrus-fix-invalid-insns
6477 @opindex mno-cirrus-fix-invalid-insns
6478 Insert NOPs into the instruction stream to in order to work around
6479 problems with invalid Maverick instruction combinations. This option
6480 is only valid if the @option{-mcpu=ep9312} option has been used to
6481 enable generation of instructions for the Cirrus Maverick floating
6482 point co-processor. This option is not enabled by default, since the
6483 problem is only present in older Maverick implementations. The default
6484 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
6487 @item -mpoke-function-name
6488 @opindex mpoke-function-name
6489 Write the name of each function into the text section, directly
6490 preceding the function prologue. The generated code is similar to this:
6494 .ascii "arm_poke_function_name", 0
6497 .word 0xff000000 + (t1 - t0)
6498 arm_poke_function_name
6500 stmfd sp!, @{fp, ip, lr, pc@}
6504 When performing a stack backtrace, code can inspect the value of
6505 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
6506 location @code{pc - 12} and the top 8 bits are set, then we know that
6507 there is a function name embedded immediately preceding this location
6508 and has length @code{((pc[-3]) & 0xff000000)}.
6512 Generate code for the 16-bit Thumb instruction set. The default is to
6513 use the 32-bit ARM instruction set.
6516 @opindex mtpcs-frame
6517 Generate a stack frame that is compliant with the Thumb Procedure Call
6518 Standard for all non-leaf functions. (A leaf function is one that does
6519 not call any other functions.) The default is @option{-mno-tpcs-frame}.
6521 @item -mtpcs-leaf-frame
6522 @opindex mtpcs-leaf-frame
6523 Generate a stack frame that is compliant with the Thumb Procedure Call
6524 Standard for all leaf functions. (A leaf function is one that does
6525 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
6527 @item -mcallee-super-interworking
6528 @opindex mcallee-super-interworking
6529 Gives all externally visible functions in the file being compiled an ARM
6530 instruction set header which switches to Thumb mode before executing the
6531 rest of the function. This allows these functions to be called from
6532 non-interworking code.
6534 @item -mcaller-super-interworking
6535 @opindex mcaller-super-interworking
6536 Allows calls via function pointers (including virtual functions) to
6537 execute correctly regardless of whether the target code has been
6538 compiled for interworking or not. There is a small overhead in the cost
6539 of executing a function pointer if this option is enabled.
6543 @node MN10200 Options
6544 @subsection MN10200 Options
6545 @cindex MN10200 options
6547 These @option{-m} options are defined for Matsushita MN10200 architectures:
6552 Indicate to the linker that it should perform a relaxation optimization pass
6553 to shorten branches, calls and absolute memory addresses. This option only
6554 has an effect when used on the command line for the final link step.
6556 This option makes symbolic debugging impossible.
6559 @node MN10300 Options
6560 @subsection MN10300 Options
6561 @cindex MN10300 options
6563 These @option{-m} options are defined for Matsushita MN10300 architectures:
6568 Generate code to avoid bugs in the multiply instructions for the MN10300
6569 processors. This is the default.
6572 @opindex mno-mult-bug
6573 Do not generate code to avoid bugs in the multiply instructions for the
6578 Generate code which uses features specific to the AM33 processor.
6582 Do not generate code which uses features specific to the AM33 processor. This
6587 Do not link in the C run-time initialization object file.
6591 Indicate to the linker that it should perform a relaxation optimization pass
6592 to shorten branches, calls and absolute memory addresses. This option only
6593 has an effect when used on the command line for the final link step.
6595 This option makes symbolic debugging impossible.
6599 @node M32R/D Options
6600 @subsection M32R/D Options
6601 @cindex M32R/D options
6603 These @option{-m} options are defined for Mitsubishi M32R/D architectures:
6608 Generate code for the M32R/X@.
6612 Generate code for the M32R@. This is the default.
6614 @item -mcode-model=small
6615 @opindex mcode-model=small
6616 Assume all objects live in the lower 16MB of memory (so that their addresses
6617 can be loaded with the @code{ld24} instruction), and assume all subroutines
6618 are reachable with the @code{bl} instruction.
6619 This is the default.
6621 The addressability of a particular object can be set with the
6622 @code{model} attribute.
6624 @item -mcode-model=medium
6625 @opindex mcode-model=medium
6626 Assume objects may be anywhere in the 32-bit address space (the compiler
6627 will generate @code{seth/add3} instructions to load their addresses), and
6628 assume all subroutines are reachable with the @code{bl} instruction.
6630 @item -mcode-model=large
6631 @opindex mcode-model=large
6632 Assume objects may be anywhere in the 32-bit address space (the compiler
6633 will generate @code{seth/add3} instructions to load their addresses), and
6634 assume subroutines may not be reachable with the @code{bl} instruction
6635 (the compiler will generate the much slower @code{seth/add3/jl}
6636 instruction sequence).
6639 @opindex msdata=none
6640 Disable use of the small data area. Variables will be put into
6641 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
6642 @code{section} attribute has been specified).
6643 This is the default.
6645 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
6646 Objects may be explicitly put in the small data area with the
6647 @code{section} attribute using one of these sections.
6650 @opindex msdata=sdata
6651 Put small global and static data in the small data area, but do not
6652 generate special code to reference them.
6656 Put small global and static data in the small data area, and generate
6657 special instructions to reference them.
6661 @cindex smaller data references
6662 Put global and static objects less than or equal to @var{num} bytes
6663 into the small data or bss sections instead of the normal data or bss
6664 sections. The default value of @var{num} is 8.
6665 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
6666 for this option to have any effect.
6668 All modules should be compiled with the same @option{-G @var{num}} value.
6669 Compiling with different values of @var{num} may or may not work; if it
6670 doesn't the linker will give an error message---incorrect code will not be
6676 @subsection M88K Options
6677 @cindex M88k options
6679 These @samp{-m} options are defined for Motorola 88k architectures:
6684 Generate code that works well on both the m88100 and the
6689 Generate code that works best for the m88100, but that also
6694 Generate code that works best for the m88110, and may not run
6699 Obsolete option to be removed from the next revision.
6702 @item -midentify-revision
6703 @opindex midentify-revision
6704 @cindex identifying source, compiler (88k)
6705 Include an @code{ident} directive in the assembler output recording the
6706 source file name, compiler name and version, timestamp, and compilation
6709 @item -mno-underscores
6710 @opindex mno-underscores
6711 @cindex underscores, avoiding (88k)
6712 In assembler output, emit symbol names without adding an underscore
6713 character at the beginning of each name. The default is to use an
6714 underscore as prefix on each name.
6716 @item -mocs-debug-info
6717 @itemx -mno-ocs-debug-info
6718 @opindex mocs-debug-info
6719 @opindex mno-ocs-debug-info
6721 @cindex debugging, 88k OCS
6722 Include (or omit) additional debugging information (about registers used
6723 in each stack frame) as specified in the 88open Object Compatibility
6724 Standard, ``OCS''@. This extra information allows debugging of code that
6725 has had the frame pointer eliminated. The default for SVr4 and Delta 88
6726 SVr3.2 is to include this information; other 88k configurations omit this
6727 information by default.
6729 @item -mocs-frame-position
6730 @opindex mocs-frame-position
6731 @cindex register positions in frame (88k)
6732 When emitting COFF debugging information for automatic variables and
6733 parameters stored on the stack, use the offset from the canonical frame
6734 address, which is the stack pointer (register 31) on entry to the
6735 function. The SVr4 and Delta88 SVr3.2, and BCS configurations use
6736 @option{-mocs-frame-position}; other 88k configurations have the default
6737 @option{-mno-ocs-frame-position}.
6739 @item -mno-ocs-frame-position
6740 @opindex mno-ocs-frame-position
6741 @cindex register positions in frame (88k)
6742 When emitting COFF debugging information for automatic variables and
6743 parameters stored on the stack, use the offset from the frame pointer
6744 register (register 30). When this option is in effect, the frame
6745 pointer is not eliminated when debugging information is selected by the
6748 @item -moptimize-arg-area
6749 @opindex moptimize-arg-area
6750 @cindex arguments in frame (88k)
6751 Save space by reorganizing the stack frame. This option generates code
6752 that does not agree with the 88open specifications, but uses less
6755 @itemx -mno-optimize-arg-area
6756 @opindex mno-optimize-arg-area
6757 Do not reorganize the stack frame to save space. This is the default.
6758 The generated conforms to the specification, but uses more memory.
6760 @item -mshort-data-@var{num}
6761 @opindex mshort-data
6762 @cindex smaller data references (88k)
6763 @cindex r0-relative references (88k)
6764 Generate smaller data references by making them relative to @code{r0},
6765 which allows loading a value using a single instruction (rather than the
6766 usual two). You control which data references are affected by
6767 specifying @var{num} with this option. For example, if you specify
6768 @option{-mshort-data-512}, then the data references affected are those
6769 involving displacements of less than 512 bytes.
6770 @option{-mshort-data-@var{num}} is not effective for @var{num} greater
6773 @item -mserialize-volatile
6774 @opindex mserialize-volatile
6775 @itemx -mno-serialize-volatile
6776 @opindex mno-serialize-volatile
6777 @cindex sequential consistency on 88k
6778 Do, or don't, generate code to guarantee sequential consistency
6779 of volatile memory references. By default, consistency is
6782 The order of memory references made by the MC88110 processor does
6783 not always match the order of the instructions requesting those
6784 references. In particular, a load instruction may execute before
6785 a preceding store instruction. Such reordering violates
6786 sequential consistency of volatile memory references, when there
6787 are multiple processors. When consistency must be guaranteed,
6788 GCC generates special instructions, as needed, to force
6789 execution in the proper order.
6791 The MC88100 processor does not reorder memory references and so
6792 always provides sequential consistency. However, by default, GCC
6793 generates the special instructions to guarantee consistency
6794 even when you use @option{-m88100}, so that the code may be run on an
6795 MC88110 processor. If you intend to run your code only on the
6796 MC88100 processor, you may use @option{-mno-serialize-volatile}.
6798 The extra code generated to guarantee consistency may affect the
6799 performance of your application. If you know that you can safely
6800 forgo this guarantee, you may use @option{-mno-serialize-volatile}.
6806 @cindex assembler syntax, 88k
6808 Turn on (@option{-msvr4}) or off (@option{-msvr3}) compiler extensions
6809 related to System V release 4 (SVr4). This controls the following:
6813 Which variant of the assembler syntax to emit.
6815 @option{-msvr4} makes the C preprocessor recognize @samp{#pragma weak}
6816 that is used on System V release 4.
6818 @option{-msvr4} makes GCC issue additional declaration directives used in
6822 @option{-msvr4} is the default for the m88k-motorola-sysv4 configuration.
6823 @option{-msvr3} is the default for all other m88k configurations.
6825 @item -mversion-03.00
6826 @opindex mversion-03.00
6827 This option is obsolete, and is ignored.
6828 @c ??? which asm syntax better for GAS? option there too?
6830 @item -mno-check-zero-division
6831 @itemx -mcheck-zero-division
6832 @opindex mno-check-zero-division
6833 @opindex mcheck-zero-division
6834 @cindex zero division on 88k
6835 Do, or don't, generate code to guarantee that integer division by
6836 zero will be detected. By default, detection is guaranteed.
6838 Some models of the MC88100 processor fail to trap upon integer
6839 division by zero under certain conditions. By default, when
6840 compiling code that might be run on such a processor, GCC
6841 generates code that explicitly checks for zero-valued divisors
6842 and traps with exception number 503 when one is detected. Use of
6843 @option{-mno-check-zero-division} suppresses such checking for code
6844 generated to run on an MC88100 processor.
6846 GCC assumes that the MC88110 processor correctly detects all instances
6847 of integer division by zero. When @option{-m88110} is specified, no
6848 explicit checks for zero-valued divisors are generated, and both
6849 @option{-mcheck-zero-division} and @option{-mno-check-zero-division} are
6852 @item -muse-div-instruction
6853 @opindex muse-div-instruction
6854 @cindex divide instruction, 88k
6855 Use the div instruction for signed integer division on the
6856 MC88100 processor. By default, the div instruction is not used.
6858 On the MC88100 processor the signed integer division instruction
6859 div) traps to the operating system on a negative operand. The
6860 operating system transparently completes the operation, but at a
6861 large cost in execution time. By default, when compiling code
6862 that might be run on an MC88100 processor, GCC emulates signed
6863 integer division using the unsigned integer division instruction
6864 divu), thereby avoiding the large penalty of a trap to the
6865 operating system. Such emulation has its own, smaller, execution
6866 cost in both time and space. To the extent that your code's
6867 important signed integer division operations are performed on two
6868 nonnegative operands, it may be desirable to use the div
6869 instruction directly.
6871 On the MC88110 processor the div instruction (also known as the
6872 divs instruction) processes negative operands without trapping to
6873 the operating system. When @option{-m88110} is specified,
6874 @option{-muse-div-instruction} is ignored, and the div instruction is used
6875 for signed integer division.
6877 Note that the result of dividing @code{INT_MIN} by @minus{}1 is undefined. In
6878 particular, the behavior of such a division with and without
6879 @option{-muse-div-instruction} may differ.
6881 @item -mtrap-large-shift
6882 @itemx -mhandle-large-shift
6883 @opindex mtrap-large-shift
6884 @opindex mhandle-large-shift
6885 @cindex bit shift overflow (88k)
6886 @cindex large bit shifts (88k)
6887 Include code to detect bit-shifts of more than 31 bits; respectively,
6888 trap such shifts or emit code to handle them properly. By default GCC
6889 makes no special provision for large bit shifts.
6891 @item -mwarn-passed-structs
6892 @opindex mwarn-passed-structs
6893 @cindex structure passing (88k)
6894 Warn when a function passes a struct as an argument or result.
6895 Structure-passing conventions have changed during the evolution of the C
6896 language, and are often the source of portability problems. By default,
6897 GCC issues no such warning.
6900 @c break page here to avoid unsightly interparagraph stretch.
6904 @node RS/6000 and PowerPC Options
6905 @subsection IBM RS/6000 and PowerPC Options
6906 @cindex RS/6000 and PowerPC Options
6907 @cindex IBM RS/6000 and PowerPC Options
6909 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
6917 @itemx -mpowerpc-gpopt
6918 @itemx -mno-powerpc-gpopt
6919 @itemx -mpowerpc-gfxopt
6920 @itemx -mno-powerpc-gfxopt
6922 @itemx -mno-powerpc64
6928 @opindex mno-powerpc
6929 @opindex mpowerpc-gpopt
6930 @opindex mno-powerpc-gpopt
6931 @opindex mpowerpc-gfxopt
6932 @opindex mno-powerpc-gfxopt
6934 @opindex mno-powerpc64
6935 GCC supports two related instruction set architectures for the
6936 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
6937 instructions supported by the @samp{rios} chip set used in the original
6938 RS/6000 systems and the @dfn{PowerPC} instruction set is the
6939 architecture of the Motorola MPC5xx, MPC6xx, MPC8xx microprocessors, and
6940 the IBM 4xx microprocessors.
6942 Neither architecture is a subset of the other. However there is a
6943 large common subset of instructions supported by both. An MQ
6944 register is included in processors supporting the POWER architecture.
6946 You use these options to specify which instructions are available on the
6947 processor you are using. The default value of these options is
6948 determined when configuring GCC@. Specifying the
6949 @option{-mcpu=@var{cpu_type}} overrides the specification of these
6950 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
6951 rather than the options listed above.
6953 The @option{-mpower} option allows GCC to generate instructions that
6954 are found only in the POWER architecture and to use the MQ register.
6955 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
6956 to generate instructions that are present in the POWER2 architecture but
6957 not the original POWER architecture.
6959 The @option{-mpowerpc} option allows GCC to generate instructions that
6960 are found only in the 32-bit subset of the PowerPC architecture.
6961 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
6962 GCC to use the optional PowerPC architecture instructions in the
6963 General Purpose group, including floating-point square root. Specifying
6964 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
6965 use the optional PowerPC architecture instructions in the Graphics
6966 group, including floating-point select.
6968 The @option{-mpowerpc64} option allows GCC to generate the additional
6969 64-bit instructions that are found in the full PowerPC64 architecture
6970 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
6971 @option{-mno-powerpc64}.
6973 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
6974 will use only the instructions in the common subset of both
6975 architectures plus some special AIX common-mode calls, and will not use
6976 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
6977 permits GCC to use any instruction from either architecture and to
6978 allow use of the MQ register; specify this for the Motorola MPC601.
6980 @item -mnew-mnemonics
6981 @itemx -mold-mnemonics
6982 @opindex mnew-mnemonics
6983 @opindex mold-mnemonics
6984 Select which mnemonics to use in the generated assembler code. With
6985 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
6986 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
6987 assembler mnemonics defined for the POWER architecture. Instructions
6988 defined in only one architecture have only one mnemonic; GCC uses that
6989 mnemonic irrespective of which of these options is specified.
6991 GCC defaults to the mnemonics appropriate for the architecture in
6992 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
6993 value of these option. Unless you are building a cross-compiler, you
6994 should normally not specify either @option{-mnew-mnemonics} or
6995 @option{-mold-mnemonics}, but should instead accept the default.
6997 @item -mcpu=@var{cpu_type}
6999 Set architecture type, register usage, choice of mnemonics, and
7000 instruction scheduling parameters for machine type @var{cpu_type}.
7001 Supported values for @var{cpu_type} are @samp{rios}, @samp{rios1},
7002 @samp{rsc}, @samp{rios2}, @samp{rs64a}, @samp{601}, @samp{602},
7003 @samp{603}, @samp{603e}, @samp{604}, @samp{604e}, @samp{620},
7004 @samp{630}, @samp{740}, @samp{7400}, @samp{7450}, @samp{750},
7005 @samp{power}, @samp{power2}, @samp{powerpc}, @samp{403}, @samp{505},
7006 @samp{801}, @samp{821}, @samp{823}, and @samp{860} and @samp{common}.
7008 @option{-mcpu=common} selects a completely generic processor. Code
7009 generated under this option will run on any POWER or PowerPC processor.
7010 GCC will use only the instructions in the common subset of both
7011 architectures, and will not use the MQ register. GCC assumes a generic
7012 processor model for scheduling purposes.
7014 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
7015 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
7016 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
7017 types, with an appropriate, generic processor model assumed for
7018 scheduling purposes.
7020 The other options specify a specific processor. Code generated under
7021 those options will run best on that processor, and may not run at all on
7024 The @option{-mcpu} options automatically enable or disable other
7025 @option{-m} options as follows:
7029 @option{-mno-power}, @option{-mno-powerpc}
7036 @option{-mpower}, @option{-mno-powerpc}, @option{-mno-new-mnemonics}
7051 @option{-mno-power}, @option{-mpowerpc}, @option{-mnew-mnemonics}
7054 @option{-mpower}, @option{-mpowerpc}, @option{-mnew-mnemonics}
7059 @option{-mno-power}, @option{-mpowerpc}, @option{-mnew-mnemonics}, @option{-msoft-float}
7062 @item -mtune=@var{cpu_type}
7064 Set the instruction scheduling parameters for machine type
7065 @var{cpu_type}, but do not set the architecture type, register usage, or
7066 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
7067 values for @var{cpu_type} are used for @option{-mtune} as for
7068 @option{-mcpu}. If both are specified, the code generated will use the
7069 architecture, registers, and mnemonics set by @option{-mcpu}, but the
7070 scheduling parameters set by @option{-mtune}.
7075 @opindex mno-altivec
7076 These switches enable or disable the use of built-in functions that
7077 allow access to the AltiVec instruction set. You may also need to set
7078 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
7083 Extend the current ABI with SPE ABI extensions. This does not change
7084 the default ABI, instead it adds the SPE ABI extensions to the current
7088 @opindex mabi=no-spe
7089 Disable Booke SPE ABI extensions for the current ABI.
7091 @item -misel=@var{yes/no}
7094 This switch enables or disables the generation of ISEL instructions.
7096 @item -mspe=@var{yes/no}
7099 This switch enables or disables the generation of SPE simd
7102 @item -mfloat-gprs=@var{yes/no}
7104 @opindex mfloat-gprs
7105 This switch enables or disables the generation of floating point
7106 operations on the general purpose registers for architectures that
7107 support it. This option is currently only available on the MPC8540.
7110 @itemx -mno-fp-in-toc
7111 @itemx -mno-sum-in-toc
7112 @itemx -mminimal-toc
7114 @opindex mno-fp-in-toc
7115 @opindex mno-sum-in-toc
7116 @opindex mminimal-toc
7117 Modify generation of the TOC (Table Of Contents), which is created for
7118 every executable file. The @option{-mfull-toc} option is selected by
7119 default. In that case, GCC will allocate at least one TOC entry for
7120 each unique non-automatic variable reference in your program. GCC
7121 will also place floating-point constants in the TOC@. However, only
7122 16,384 entries are available in the TOC@.
7124 If you receive a linker error message that saying you have overflowed
7125 the available TOC space, you can reduce the amount of TOC space used
7126 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
7127 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
7128 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
7129 generate code to calculate the sum of an address and a constant at
7130 run-time instead of putting that sum into the TOC@. You may specify one
7131 or both of these options. Each causes GCC to produce very slightly
7132 slower and larger code at the expense of conserving TOC space.
7134 If you still run out of space in the TOC even when you specify both of
7135 these options, specify @option{-mminimal-toc} instead. This option causes
7136 GCC to make only one TOC entry for every file. When you specify this
7137 option, GCC will produce code that is slower and larger but which
7138 uses extremely little TOC space. You may wish to use this option
7139 only on files that contain less frequently executed code.
7145 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
7146 @code{long} type, and the infrastructure needed to support them.
7147 Specifying @option{-maix64} implies @option{-mpowerpc64} and
7148 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
7149 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
7154 @opindex mno-xl-call
7155 On AIX, pass floating-point arguments to prototyped functions beyond the
7156 register save area (RSA) on the stack in addition to argument FPRs. The
7157 AIX calling convention was extended but not initially documented to
7158 handle an obscure K&R C case of calling a function that takes the
7159 address of its arguments with fewer arguments than declared. AIX XL
7160 compilers access floating point arguments which do not fit in the
7161 RSA from the stack when a subroutine is compiled without
7162 optimization. Because always storing floating-point arguments on the
7163 stack is inefficient and rarely needed, this option is not enabled by
7164 default and only is necessary when calling subroutines compiled by AIX
7165 XL compilers without optimization.
7169 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
7170 application written to use message passing with special startup code to
7171 enable the application to run. The system must have PE installed in the
7172 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
7173 must be overridden with the @option{-specs=} option to specify the
7174 appropriate directory location. The Parallel Environment does not
7175 support threads, so the @option{-mpe} option and the @option{-pthread}
7176 option are incompatible.
7178 @item -malign-natural
7179 @itemx -malign-power
7180 @opindex malign-natural
7181 @opindex malign-power
7182 On AIX, Darwin, and 64-bit PowerPC Linux, the option
7183 @option{-malign-natural} overrides the ABI-defined alignment of larger
7184 types, such as floating-point doubles, on their natural size-based boundary.
7185 The option @option{-malign-power} instructs GCC to follow the ABI-specified
7186 alignment rules. GCC defaults to the standard alignment defined in the ABI.
7190 @opindex msoft-float
7191 @opindex mhard-float
7192 Generate code that does not use (uses) the floating-point register set.
7193 Software floating point emulation is provided if you use the
7194 @option{-msoft-float} option, and pass the option to GCC when linking.
7197 @itemx -mno-multiple
7199 @opindex mno-multiple
7200 Generate code that uses (does not use) the load multiple word
7201 instructions and the store multiple word instructions. These
7202 instructions are generated by default on POWER systems, and not
7203 generated on PowerPC systems. Do not use @option{-mmultiple} on little
7204 endian PowerPC systems, since those instructions do not work when the
7205 processor is in little endian mode. The exceptions are PPC740 and
7206 PPC750 which permit the instructions usage in little endian mode.
7212 Generate code that uses (does not use) the load string instructions
7213 and the store string word instructions to save multiple registers and
7214 do small block moves. These instructions are generated by default on
7215 POWER systems, and not generated on PowerPC systems. Do not use
7216 @option{-mstring} on little endian PowerPC systems, since those
7217 instructions do not work when the processor is in little endian mode.
7218 The exceptions are PPC740 and PPC750 which permit the instructions
7219 usage in little endian mode.
7225 Generate code that uses (does not use) the load or store instructions
7226 that update the base register to the address of the calculated memory
7227 location. These instructions are generated by default. If you use
7228 @option{-mno-update}, there is a small window between the time that the
7229 stack pointer is updated and the address of the previous frame is
7230 stored, which means code that walks the stack frame across interrupts or
7231 signals may get corrupted data.
7234 @itemx -mno-fused-madd
7235 @opindex mfused-madd
7236 @opindex mno-fused-madd
7237 Generate code that uses (does not use) the floating point multiply and
7238 accumulate instructions. These instructions are generated by default if
7239 hardware floating is used.
7241 @item -mno-bit-align
7243 @opindex mno-bit-align
7245 On System V.4 and embedded PowerPC systems do not (do) force structures
7246 and unions that contain bit-fields to be aligned to the base type of the
7249 For example, by default a structure containing nothing but 8
7250 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
7251 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
7252 the structure would be aligned to a 1 byte boundary and be one byte in
7255 @item -mno-strict-align
7256 @itemx -mstrict-align
7257 @opindex mno-strict-align
7258 @opindex mstrict-align
7259 On System V.4 and embedded PowerPC systems do not (do) assume that
7260 unaligned memory references will be handled by the system.
7263 @itemx -mno-relocatable
7264 @opindex mrelocatable
7265 @opindex mno-relocatable
7266 On embedded PowerPC systems generate code that allows (does not allow)
7267 the program to be relocated to a different address at runtime. If you
7268 use @option{-mrelocatable} on any module, all objects linked together must
7269 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
7271 @item -mrelocatable-lib
7272 @itemx -mno-relocatable-lib
7273 @opindex mrelocatable-lib
7274 @opindex mno-relocatable-lib
7275 On embedded PowerPC systems generate code that allows (does not allow)
7276 the program to be relocated to a different address at runtime. Modules
7277 compiled with @option{-mrelocatable-lib} can be linked with either modules
7278 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
7279 with modules compiled with the @option{-mrelocatable} options.
7285 On System V.4 and embedded PowerPC systems do not (do) assume that
7286 register 2 contains a pointer to a global area pointing to the addresses
7287 used in the program.
7290 @itemx -mlittle-endian
7292 @opindex mlittle-endian
7293 On System V.4 and embedded PowerPC systems compile code for the
7294 processor in little endian mode. The @option{-mlittle-endian} option is
7295 the same as @option{-mlittle}.
7300 @opindex mbig-endian
7301 On System V.4 and embedded PowerPC systems compile code for the
7302 processor in big endian mode. The @option{-mbig-endian} option is
7303 the same as @option{-mbig}.
7305 @item -mdynamic-no-pic
7306 @opindex mdynamic-no-pic
7307 On Darwin and Mac OS X systems, compile code so that it is not
7308 relocatable, but that its external references are relocatable. The
7309 resulting code is suitable for applications, but not shared
7314 On System V.4 and embedded PowerPC systems compile code using calling
7315 conventions that adheres to the March 1995 draft of the System V
7316 Application Binary Interface, PowerPC processor supplement. This is the
7317 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
7319 @item -mcall-sysv-eabi
7320 @opindex mcall-sysv-eabi
7321 Specify both @option{-mcall-sysv} and @option{-meabi} options.
7323 @item -mcall-sysv-noeabi
7324 @opindex mcall-sysv-noeabi
7325 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
7327 @item -mcall-solaris
7328 @opindex mcall-solaris
7329 On System V.4 and embedded PowerPC systems compile code for the Solaris
7333 @opindex mcall-linux
7334 On System V.4 and embedded PowerPC systems compile code for the
7335 Linux-based GNU system.
7339 On System V.4 and embedded PowerPC systems compile code for the
7340 Hurd-based GNU system.
7343 @opindex mcall-netbsd
7344 On System V.4 and embedded PowerPC systems compile code for the
7345 NetBSD operating system.
7347 @item -maix-struct-return
7348 @opindex maix-struct-return
7349 Return all structures in memory (as specified by the AIX ABI)@.
7351 @item -msvr4-struct-return
7352 @opindex msvr4-struct-return
7353 Return structures smaller than 8 bytes in registers (as specified by the
7357 @opindex mabi=altivec
7358 Extend the current ABI with AltiVec ABI extensions. This does not
7359 change the default ABI, instead it adds the AltiVec ABI extensions to
7362 @item -mabi=no-altivec
7363 @opindex mabi=no-altivec
7364 Disable AltiVec ABI extensions for the current ABI.
7367 @itemx -mno-prototype
7369 @opindex mno-prototype
7370 On System V.4 and embedded PowerPC systems assume that all calls to
7371 variable argument functions are properly prototyped. Otherwise, the
7372 compiler must insert an instruction before every non prototyped call to
7373 set or clear bit 6 of the condition code register (@var{CR}) to
7374 indicate whether floating point values were passed in the floating point
7375 registers in case the function takes a variable arguments. With
7376 @option{-mprototype}, only calls to prototyped variable argument functions
7377 will set or clear the bit.
7381 On embedded PowerPC systems, assume that the startup module is called
7382 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
7383 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
7388 On embedded PowerPC systems, assume that the startup module is called
7389 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
7394 On embedded PowerPC systems, assume that the startup module is called
7395 @file{crt0.o} and the standard C libraries are @file{libads.a} and
7399 @opindex myellowknife
7400 On embedded PowerPC systems, assume that the startup module is called
7401 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
7406 On System V.4 and embedded PowerPC systems, specify that you are
7407 compiling for a VxWorks system.
7411 Specify that you are compiling for the WindISS simulation environment.
7415 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
7416 header to indicate that @samp{eabi} extended relocations are used.
7422 On System V.4 and embedded PowerPC systems do (do not) adhere to the
7423 Embedded Applications Binary Interface (eabi) which is a set of
7424 modifications to the System V.4 specifications. Selecting @option{-meabi}
7425 means that the stack is aligned to an 8 byte boundary, a function
7426 @code{__eabi} is called to from @code{main} to set up the eabi
7427 environment, and the @option{-msdata} option can use both @code{r2} and
7428 @code{r13} to point to two separate small data areas. Selecting
7429 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
7430 do not call an initialization function from @code{main}, and the
7431 @option{-msdata} option will only use @code{r13} to point to a single
7432 small data area. The @option{-meabi} option is on by default if you
7433 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
7436 @opindex msdata=eabi
7437 On System V.4 and embedded PowerPC systems, put small initialized
7438 @code{const} global and static data in the @samp{.sdata2} section, which
7439 is pointed to by register @code{r2}. Put small initialized
7440 non-@code{const} global and static data in the @samp{.sdata} section,
7441 which is pointed to by register @code{r13}. Put small uninitialized
7442 global and static data in the @samp{.sbss} section, which is adjacent to
7443 the @samp{.sdata} section. The @option{-msdata=eabi} option is
7444 incompatible with the @option{-mrelocatable} option. The
7445 @option{-msdata=eabi} option also sets the @option{-memb} option.
7448 @opindex msdata=sysv
7449 On System V.4 and embedded PowerPC systems, put small global and static
7450 data in the @samp{.sdata} section, which is pointed to by register
7451 @code{r13}. Put small uninitialized global and static data in the
7452 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
7453 The @option{-msdata=sysv} option is incompatible with the
7454 @option{-mrelocatable} option.
7456 @item -msdata=default
7458 @opindex msdata=default
7460 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
7461 compile code the same as @option{-msdata=eabi}, otherwise compile code the
7462 same as @option{-msdata=sysv}.
7465 @opindex msdata-data
7466 On System V.4 and embedded PowerPC systems, put small global and static
7467 data in the @samp{.sdata} section. Put small uninitialized global and
7468 static data in the @samp{.sbss} section. Do not use register @code{r13}
7469 to address small data however. This is the default behavior unless
7470 other @option{-msdata} options are used.
7474 @opindex msdata=none
7476 On embedded PowerPC systems, put all initialized global and static data
7477 in the @samp{.data} section, and all uninitialized data in the
7478 @samp{.bss} section.
7482 @cindex smaller data references (PowerPC)
7483 @cindex .sdata/.sdata2 references (PowerPC)
7484 On embedded PowerPC systems, put global and static items less than or
7485 equal to @var{num} bytes into the small data or bss sections instead of
7486 the normal data or bss section. By default, @var{num} is 8. The
7487 @option{-G @var{num}} switch is also passed to the linker.
7488 All modules should be compiled with the same @option{-G @var{num}} value.
7491 @itemx -mno-regnames
7493 @opindex mno-regnames
7494 On System V.4 and embedded PowerPC systems do (do not) emit register
7495 names in the assembly language output using symbolic forms.
7498 @itemx -mno-longcall
7500 @opindex mno-longcall
7501 Default to making all function calls via pointers, so that functions
7502 which reside further than 64 megabytes (67,108,864 bytes) from the
7503 current location can be called. This setting can be overridden by the
7504 @code{shortcall} function attribute, or by @code{#pragma longcall(0)}.
7506 Some linkers are capable of detecting out-of-range calls and generating
7507 glue code on the fly. On these systems, long calls are unnecessary and
7508 generate slower code. As of this writing, the AIX linker can do this,
7509 as can the GNU linker for PowerPC/64. It is planned to add this feature
7510 to the GNU linker for 32-bit PowerPC systems as well.
7512 In the future, we may cause GCC to ignore all longcall specifications
7513 when the linker is known to generate glue.
7517 Adds support for multithreading with the @dfn{pthreads} library.
7518 This option sets flags for both the preprocessor and linker.
7522 @node Darwin Options
7523 @subsection Darwin Options
7524 @cindex Darwin options
7526 These options are defined for all architectures running the Darwin operating
7527 system. They are useful for compatibility with other Mac OS compilers.
7532 Loads all members of static archive libraries.
7533 See man ld(1) for more information.
7535 @item -arch_errors_fatal
7536 @opindex arch_errors_fatal
7537 Cause the errors having to do with files that have the wrong architecture
7541 @opindex bind_at_load
7542 Causes the output file to be marked such that the dynamic linker will
7543 bind all undefined references when the file is loaded or launched.
7547 Produce a Mach-o bundle format file.
7548 See man ld(1) for more information.
7550 @item -bundle_loader @var{executable}
7551 @opindex bundle_loader
7552 This specifies the @var{executable} that will be loading the build
7553 output file being linked. See man ld(1) for more information.
7555 @item -allowable_client @var{client_name}
7559 @item -compatibility_version
7560 @item -current_version
7561 @item -dependency-file
7563 @item -dylinker_install_name
7566 @item -exported_symbols_list
7568 @item -flat_namespace
7569 @item -force_cpusubtype_ALL
7570 @item -force_flat_namespace
7571 @item -headerpad_max_install_names
7575 @item -keep_private_externs
7577 @item -multiply_defined
7578 @item -multiply_defined_unused
7582 @item -noseglinkedit
7583 @item -pagezero_size
7585 @item -prebind_all_twolevel_modules
7586 @item -private_bundle
7587 @item -read_only_relocs
7589 @item -sectobjectsymbols
7593 @item -sectobjectsymbols
7595 @item -seg_addr_table
7596 @item -seg_addr_table_filename
7599 @item -segs_read_only_addr
7600 @item -segs_read_write_addr
7601 @item -single_module
7605 @item -twolevel_namespace
7608 @item -unexported_symbols_list
7609 @item -weak_reference_mismatches
7612 @opindex allowable_client
7614 @opindex client_name
7615 @opindex compatibility_version
7616 @opindex current_version
7617 @opindex dependency-file
7619 @opindex dylinker_install_name
7622 @opindex exported_symbols_list
7624 @opindex flat_namespace
7625 @opindex force_cpusubtype_ALL
7626 @opindex force_flat_namespace
7627 @opindex headerpad_max_install_names
7630 @opindex install_name
7631 @opindex keep_private_externs
7632 @opindex multi_module
7633 @opindex multiply_defined
7634 @opindex multiply_defined_unused
7636 @opindex nomultidefs
7638 @opindex noseglinkedit
7639 @opindex pagezero_size
7641 @opindex prebind_all_twolevel_modules
7642 @opindex private_bundle
7643 @opindex read_only_relocs
7645 @opindex sectobjectsymbols
7649 @opindex sectobjectsymbols
7651 @opindex seg_addr_table
7652 @opindex seg_addr_table_filename
7653 @opindex seglinkedit
7655 @opindex segs_read_only_addr
7656 @opindex segs_read_write_addr
7657 @opindex single_module
7659 @opindex sub_library
7660 @opindex sub_umbrella
7661 @opindex twolevel_namespace
7664 @opindex unexported_symbols_list
7665 @opindex weak_reference_mismatches
7666 @opindex whatsloaded
7668 This options are available for Darwin linker. Darwin linker man page
7669 describes them in detail.
7674 @subsection IBM RT Options
7676 @cindex IBM RT options
7678 These @samp{-m} options are defined for the IBM RT PC:
7682 @opindex min-line-mul
7683 Use an in-line code sequence for integer multiplies. This is the
7686 @item -mcall-lib-mul
7687 @opindex mcall-lib-mul
7688 Call @code{lmul$$} for integer multiples.
7690 @item -mfull-fp-blocks
7691 @opindex mfull-fp-blocks
7692 Generate full-size floating point data blocks, including the minimum
7693 amount of scratch space recommended by IBM@. This is the default.
7695 @item -mminimum-fp-blocks
7696 @opindex mminimum-fp-blocks
7697 Do not include extra scratch space in floating point data blocks. This
7698 results in smaller code, but slower execution, since scratch space must
7699 be allocated dynamically.
7701 @cindex @file{stdarg.h} and RT PC
7702 @item -mfp-arg-in-fpregs
7703 @opindex mfp-arg-in-fpregs
7704 Use a calling sequence incompatible with the IBM calling convention in
7705 which floating point arguments are passed in floating point registers.
7706 Note that @code{stdarg.h} will not work with floating point operands
7707 if this option is specified.
7709 @item -mfp-arg-in-gregs
7710 @opindex mfp-arg-in-gregs
7711 Use the normal calling convention for floating point arguments. This is
7714 @item -mhc-struct-return
7715 @opindex mhc-struct-return
7716 Return structures of more than one word in memory, rather than in a
7717 register. This provides compatibility with the MetaWare HighC (hc)
7718 compiler. Use the option @option{-fpcc-struct-return} for compatibility
7719 with the Portable C Compiler (pcc).
7721 @item -mnohc-struct-return
7722 @opindex mnohc-struct-return
7723 Return some structures of more than one word in registers, when
7724 convenient. This is the default. For compatibility with the
7725 IBM-supplied compilers, use the option @option{-fpcc-struct-return} or the
7726 option @option{-mhc-struct-return}.
7730 @subsection MIPS Options
7731 @cindex MIPS options
7733 These @samp{-m} options are defined for the MIPS family of computers:
7737 @item -march=@var{arch}
7739 Generate code that will run on @var{arch}, which can be the name of a
7740 generic MIPS ISA, or the name of a particular processor.
7742 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
7743 @samp{mips32}, @samp{mips32r2}, and @samp{mips64}.
7744 The processor names are:
7745 @samp{4kc}, @samp{4kp}, @samp{5kc}, @samp{20kc},
7747 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
7748 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000},
7751 @samp{vr4100}, @samp{vr4300}, and @samp{vr5000}.
7752 The special value @samp{from-abi} selects the
7753 most compatible architecture for the selected ABI (that is,
7754 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
7756 In processor names, a final @samp{000} can be abbreviated as @samp{k}
7757 (for example, @samp{-march=r2k}). Prefixes are optional, and
7758 @samp{vr} may be written @samp{r}.
7760 GCC defines two macros based on the value of this option. The first
7761 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
7762 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
7763 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
7764 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
7765 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
7767 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
7768 above. In other words, it will have the full prefix and will not
7769 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
7770 the macro names the resolved architecture (either @samp{"mips1"} or
7771 @samp{"mips3"}). It names the default architecture when no
7772 @option{-march} option is given.
7774 @item -mtune=@var{arch}
7776 Optimize for @var{arch}. Among other things, this option controls
7777 the way instructions are scheduled, and the perceived cost of arithmetic
7778 operations. The list of @var{arch} values is the same as for
7781 When this option is not used, GCC will optimize for the processor
7782 specified by @option{-march}. By using @option{-march} and
7783 @option{-mtune} together, it is possible to generate code that will
7784 run on a family of processors, but optimize the code for one
7785 particular member of that family.
7787 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
7788 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
7789 @samp{-march} ones described above.
7793 Equivalent to @samp{-march=mips1}.
7797 Equivalent to @samp{-march=mips2}.
7801 Equivalent to @samp{-march=mips3}.
7805 Equivalent to @samp{-march=mips4}.
7809 Equivalent to @samp{-march=mips32}.
7813 Equivalent to @samp{-march=mips32r2}.
7817 Equivalent to @samp{-march=mips64}.
7820 @itemx -mno-fused-madd
7821 @opindex mfused-madd
7822 @opindex mno-fused-madd
7823 Generate code that uses (does not use) the floating point multiply and
7824 accumulate instructions, when they are available. These instructions
7825 are generated by default if they are available, but this may be
7826 undesirable if the extra precision causes problems or on certain chips
7827 in the mode where denormals are rounded to zero where denormals
7828 generated by multiply and accumulate instructions cause exceptions
7833 Assume that floating point registers are 32 bits wide.
7837 Assume that floating point registers are 64 bits wide.
7841 Assume that general purpose registers are 32 bits wide.
7845 Assume that general purpose registers are 64 bits wide.
7849 Force int and long types to be 64 bits wide. See @option{-mlong32} for an
7850 explanation of the default, and the width of pointers.
7854 Force long types to be 64 bits wide. See @option{-mlong32} for an
7855 explanation of the default, and the width of pointers.
7859 Force long, int, and pointer types to be 32 bits wide.
7861 The default size of ints, longs and pointers depends on the ABI@. All
7862 the supported ABIs use 32-bit ints. The n64 ABI uses 64-bit longs, as
7863 does the 64-bit Cygnus EABI; the others use 32-bit longs. Pointers
7864 are the same size as longs, or the same size as integer registers,
7865 whichever is smaller.
7879 Generate code for the given ABI@.
7881 Note that there are two embedded ABIs: @option{-mabi=eabi}
7882 selects the one defined by Cygnus while @option{-meabi=meabi}
7883 selects the one defined by MIPS@. Both these ABIs have
7884 32-bit and 64-bit variants. Normally, GCC will generate
7885 64-bit code when you select a 64-bit architecture, but you
7886 can use @option{-mgp32} to get 32-bit code instead.
7888 @item -mabi-fake-default
7889 @opindex mabi-fake-default
7890 You don't want to know what this option does. No, really. I mean
7891 it. Move on to the next option.
7893 What? You're still here? Oh, well@enddots{} Ok, here's the deal. GCC
7894 wants the default set of options to get the root of the multilib tree,
7895 and the shared library SONAMEs without any multilib-indicating
7896 suffixes. This is not convenience for @samp{mips64-linux-gnu}, since
7897 we want to default to the N32 ABI, while still being binary-compatible
7898 with @samp{mips-linux-gnu} if you stick to the O32 ABI@. Being
7899 binary-compatible means shared libraries should have the same SONAMEs,
7900 and libraries should live in the same location. Having O32 libraries
7901 in a sub-directory named say @file{o32} is not acceptable.
7903 So we trick GCC into believing that O32 is the default ABI, except
7904 that we override the default with some internal command-line
7905 processing magic. Problem is, if we stopped at that, and you then
7906 created a multilib-aware package that used the output of @command{gcc
7907 -print-multi-lib} to decide which multilibs to build, and how, and
7908 you'd find yourself in an awkward situation when you found out that
7909 some of the options listed ended up mapping to the same multilib, and
7910 none of your libraries was actually built for the multilib that
7911 @option{-print-multi-lib} claims to be the default. So we added this
7912 option that disables the default switcher, falling back to GCC's
7913 original notion of the default library. Confused yet?
7915 For short: don't ever use this option, unless you find it in the list
7916 of additional options to be used when building for multilibs, in the
7917 output of @option{gcc -print-multi-lib}.
7921 Generate code for the MIPS assembler, and invoke @file{mips-tfile} to
7922 add normal debug information. This is the default for all
7923 platforms except for the OSF/1 reference platform, using the OSF/rose
7924 object format. If the either of the @option{-gstabs} or @option{-gstabs+}
7925 switches are used, the @file{mips-tfile} program will encapsulate the
7926 stabs within MIPS ECOFF@.
7930 Generate code for the GNU assembler. This is the default on the OSF/1
7931 reference platform, using the OSF/rose object format. Also, this is
7932 the default if the configure option @option{--with-gnu-as} is used.
7934 @item -msplit-addresses
7935 @itemx -mno-split-addresses
7936 @opindex msplit-addresses
7937 @opindex mno-split-addresses
7938 Generate code to load the high and low parts of address constants separately.
7939 This allows GCC to optimize away redundant loads of the high order
7940 bits of addresses. This optimization requires GNU as and GNU ld.
7941 This optimization is enabled by default for some embedded targets where
7942 GNU as and GNU ld are standard.
7948 The @option{-mrnames} switch says to output code using the MIPS software
7949 names for the registers, instead of the hardware names (ie, @var{a0}
7950 instead of @var{$4}). The only known assembler that supports this option
7951 is the Algorithmics assembler.
7957 The @option{-mgpopt} switch says to write all of the data declarations
7958 before the instructions in the text section, this allows the MIPS
7959 assembler to generate one word memory references instead of using two
7960 words for short global or static data items. This is on by default if
7961 optimization is selected.
7967 The @option{-mmemcpy} switch makes all block moves call the appropriate
7968 string function (@samp{memcpy} or @samp{bcopy}) instead of possibly
7969 generating inline code.
7972 @itemx -mno-mips-tfile
7973 @opindex mmips-tfile
7974 @opindex mno-mips-tfile
7975 The @option{-mno-mips-tfile} switch causes the compiler not
7976 postprocess the object file with the @file{mips-tfile} program,
7977 after the MIPS assembler has generated it to add debug support. If
7978 @file{mips-tfile} is not run, then no local variables will be
7979 available to the debugger. In addition, @file{stage2} and
7980 @file{stage3} objects will have the temporary file names passed to the
7981 assembler embedded in the object file, which means the objects will
7982 not compare the same. The @option{-mno-mips-tfile} switch should only
7983 be used when there are bugs in the @file{mips-tfile} program that
7984 prevents compilation.
7987 @opindex msoft-float
7988 Generate output containing library calls for floating point.
7989 @strong{Warning:} the requisite libraries are not part of GCC@.
7990 Normally the facilities of the machine's usual C compiler are used, but
7991 this can't be done directly in cross-compilation. You must make your
7992 own arrangements to provide suitable library functions for
7996 @opindex mhard-float
7997 Generate output containing floating point instructions. This is the
7998 default if you use the unmodified sources.
8001 @itemx -mno-abicalls
8003 @opindex mno-abicalls
8004 Emit (or do not emit) the pseudo operations @samp{.abicalls},
8005 @samp{.cpload}, and @samp{.cprestore} that some System V.4 ports use for
8006 position independent code.
8009 @itemx -mno-long-calls
8010 @opindex mlong-calls
8011 @opindex mno-long-calls
8012 Do all calls with the @samp{JALR} instruction, which requires
8013 loading up a function's address into a register before the call.
8014 You need to use this switch, if you call outside of the current
8015 512 megabyte segment to functions that are not through pointers.
8018 @itemx -mno-half-pic
8020 @opindex mno-half-pic
8021 Put pointers to extern references into the data section and load them
8022 up, rather than put the references in the text section.
8024 @item -membedded-pic
8025 @itemx -mno-embedded-pic
8026 @opindex membedded-pic
8027 @opindex mno-embedded-pic
8028 Generate PIC code suitable for some embedded systems. All calls are
8029 made using PC relative address, and all data is addressed using the $gp
8030 register. No more than 65536 bytes of global data may be used. This
8031 requires GNU as and GNU ld which do most of the work. This currently
8032 only works on targets which use ECOFF; it does not work with ELF@.
8034 @item -membedded-data
8035 @itemx -mno-embedded-data
8036 @opindex membedded-data
8037 @opindex mno-embedded-data
8038 Allocate variables to the read-only data section first if possible, then
8039 next in the small data section if possible, otherwise in data. This gives
8040 slightly slower code than the default, but reduces the amount of RAM required
8041 when executing, and thus may be preferred for some embedded systems.
8043 @item -muninit-const-in-rodata
8044 @itemx -mno-uninit-const-in-rodata
8045 @opindex muninit-const-in-rodata
8046 @opindex mno-uninit-const-in-rodata
8047 When used together with @option{-membedded-data}, it will always store uninitialized
8048 const variables in the read-only data section.
8050 @item -msingle-float
8051 @itemx -mdouble-float
8052 @opindex msingle-float
8053 @opindex mdouble-float
8054 The @option{-msingle-float} switch tells gcc to assume that the floating
8055 point coprocessor only supports single precision operations, as on the
8056 @samp{r4650} chip. The @option{-mdouble-float} switch permits gcc to use
8057 double precision operations. This is the default.
8063 Permit use of the @samp{mad}, @samp{madu} and @samp{mul} instructions,
8064 as on the @samp{r4650} chip.
8068 Turns on @option{-msingle-float}, @option{-mmad}, and, at least for now,
8069 @option{-mcpu=r4650}.
8075 Enable 16-bit instructions.
8079 Use the entry and exit pseudo ops. This option can only be used with
8084 Compile code for the processor in little endian mode.
8085 The requisite libraries are assumed to exist.
8089 Compile code for the processor in big endian mode.
8090 The requisite libraries are assumed to exist.
8094 @cindex smaller data references (MIPS)
8095 @cindex gp-relative references (MIPS)
8096 Put global and static items less than or equal to @var{num} bytes into
8097 the small data or bss sections instead of the normal data or bss
8098 section. This allows the assembler to emit one word memory reference
8099 instructions based on the global pointer (@var{gp} or @var{$28}),
8100 instead of the normal two words used. By default, @var{num} is 8 when
8101 the MIPS assembler is used, and 0 when the GNU assembler is used. The
8102 @option{-G @var{num}} switch is also passed to the assembler and linker.
8103 All modules should be compiled with the same @option{-G @var{num}}
8108 Tell the MIPS assembler to not run its preprocessor over user
8109 assembler files (with a @samp{.s} suffix) when assembling them.
8113 Pass an option to gas which will cause nops to be inserted if
8114 the read of the destination register of an mfhi or mflo instruction
8115 occurs in the following two instructions.
8119 Do not include the default crt0.
8121 @item -mflush-func=@var{func}
8122 @itemx -mno-flush-func
8123 @opindex mflush-func
8124 Specifies the function to call to flush the I and D caches, or to not
8125 call any such function. If called, the function must take the same
8126 arguments as the common @code{_flush_func()}, that is, the address of the
8127 memory range for which the cache is being flushed, the size of the
8128 memory range, and the number 3 (to flush both caches). The default
8129 depends on the target gcc was configured for, but commonly is either
8130 @samp{_flush_func} or @samp{__cpu_flush}.
8132 @item -mbranch-likely
8133 @itemx -mno-branch-likely
8134 @opindex mbranch-likely
8135 @opindex mno-branch-likely
8136 Enable or disable use of Branch Likely instructions, regardless of the
8137 default for the selected architecture. By default, Branch Likely
8138 instructions may be generated if they are supported by the selected
8139 architecture. An exception is for the MIPS32 and MIPS64 architectures
8140 and processors which implement those architectures; for those, Branch
8141 Likely instructions will not be generated by default because the MIPS32
8142 and MIPS64 architectures specifically deprecate their use.
8145 @node i386 and x86-64 Options
8146 @subsection Intel 386 and AMD x86-64 Options
8147 @cindex i386 Options
8148 @cindex x86-64 Options
8149 @cindex Intel 386 Options
8150 @cindex AMD x86-64 Options
8152 These @samp{-m} options are defined for the i386 and x86-64 family of
8156 @item -mtune=@var{cpu-type}
8158 Tune to @var{cpu-type} everything applicable about the generated code, except
8159 for the ABI and the set of available instructions. The choices for
8160 @var{cpu-type} are @samp{i386}, @samp{i486}, @samp{i586}, @samp{i686},
8161 @samp{pentium}, @samp{pentium-mmx}, @samp{pentiumpro}, @samp{pentium2},
8162 @samp{pentium3}, @samp{pentium4}, @samp{k6}, @samp{k6-2}, @samp{k6-3},
8163 @samp{athlon}, @samp{athlon-tbird}, @samp{athlon-4}, @samp{athlon-xp},
8164 @samp{athlon-mp}, @samp{winchip-c6}, @samp{winchip2}, @samp{k8}, @samp{c3}
8167 While picking a specific @var{cpu-type} will schedule things appropriately
8168 for that particular chip, the compiler will not generate any code that
8169 does not run on the i386 without the @option{-march=@var{cpu-type}} option
8170 being used. @samp{i586} is equivalent to @samp{pentium} and @samp{i686}
8171 is equivalent to @samp{pentiumpro}. @samp{k6} and @samp{athlon} are the
8172 AMD chips as opposed to the Intel ones.
8174 @item -march=@var{cpu-type}
8176 Generate instructions for the machine type @var{cpu-type}. The choices
8177 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
8178 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
8180 @item -mcpu=@var{cpu-type}
8182 A deprecated synonym for @option{-mtune}.
8191 @opindex mpentiumpro
8192 These options are synonyms for @option{-mtune=i386}, @option{-mtune=i486},
8193 @option{-mtune=pentium}, and @option{-mtune=pentiumpro} respectively.
8194 These synonyms are deprecated.
8196 @item -mfpmath=@var{unit}
8198 generate floating point arithmetics for selected unit @var{unit}. the choices
8203 Use the standard 387 floating point coprocessor present majority of chips and
8204 emulated otherwise. Code compiled with this option will run almost everywhere.
8205 The temporary results are computed in 80bit precision instead of precision
8206 specified by the type resulting in slightly different results compared to most
8207 of other chips. See @option{-ffloat-store} for more detailed description.
8209 This is the default choice for i386 compiler.
8212 Use scalar floating point instructions present in the SSE instruction set.
8213 This instruction set is supported by Pentium3 and newer chips, in the AMD line
8214 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
8215 instruction set supports only single precision arithmetics, thus the double and
8216 extended precision arithmetics is still done using 387. Later version, present
8217 only in Pentium4 and the future AMD x86-64 chips supports double precision
8220 For i387 you need to use @option{-march=@var{cpu-type}}, @option{-msse} or
8221 @option{-msse2} switches to enable SSE extensions and make this option
8222 effective. For x86-64 compiler, these extensions are enabled by default.
8224 The resulting code should be considerably faster in majority of cases and avoid
8225 the numerical instability problems of 387 code, but may break some existing
8226 code that expects temporaries to be 80bit.
8228 This is the default choice for x86-64 compiler.
8231 Use all SSE extensions enabled by @option{-msse2} as well as the new
8232 SSE extensions in Prescott New Instructions. @option{-mpni} also
8233 enables 2 builtin functions, @code{__builtin_ia32_monitor} and
8234 @code{__builtin_ia32_mwait}, for new instructions @code{monitor} and
8238 Attempt to utilize both instruction sets at once. This effectively double the
8239 amount of available registers and on chips with separate execution units for
8240 387 and SSE the execution resources too. Use this option with care, as it is
8241 still experimental, because gcc register allocator does not model separate
8242 functional units well resulting in instable performance.
8245 @item -masm=@var{dialect}
8246 @opindex masm=@var{dialect}
8247 Output asm instructions using selected @var{dialect}. Supported choices are
8248 @samp{intel} or @samp{att} (the default one).
8253 @opindex mno-ieee-fp
8254 Control whether or not the compiler uses IEEE floating point
8255 comparisons. These handle correctly the case where the result of a
8256 comparison is unordered.
8259 @opindex msoft-float
8260 Generate output containing library calls for floating point.
8261 @strong{Warning:} the requisite libraries are not part of GCC@.
8262 Normally the facilities of the machine's usual C compiler are used, but
8263 this can't be done directly in cross-compilation. You must make your
8264 own arrangements to provide suitable library functions for
8267 On machines where a function returns floating point results in the 80387
8268 register stack, some floating point opcodes may be emitted even if
8269 @option{-msoft-float} is used.
8271 @item -mno-fp-ret-in-387
8272 @opindex mno-fp-ret-in-387
8273 Do not use the FPU registers for return values of functions.
8275 The usual calling convention has functions return values of types
8276 @code{float} and @code{double} in an FPU register, even if there
8277 is no FPU@. The idea is that the operating system should emulate
8280 The option @option{-mno-fp-ret-in-387} causes such values to be returned
8281 in ordinary CPU registers instead.
8283 @item -mno-fancy-math-387
8284 @opindex mno-fancy-math-387
8285 Some 387 emulators do not support the @code{sin}, @code{cos} and
8286 @code{sqrt} instructions for the 387. Specify this option to avoid
8287 generating those instructions. This option is the default on FreeBSD,
8288 OpenBSD and NetBSD@. This option is overridden when @option{-march}
8289 indicates that the target cpu will always have an FPU and so the
8290 instruction will not need emulation. As of revision 2.6.1, these
8291 instructions are not generated unless you also use the
8292 @option{-funsafe-math-optimizations} switch.
8294 @item -malign-double
8295 @itemx -mno-align-double
8296 @opindex malign-double
8297 @opindex mno-align-double
8298 Control whether GCC aligns @code{double}, @code{long double}, and
8299 @code{long long} variables on a two word boundary or a one word
8300 boundary. Aligning @code{double} variables on a two word boundary will
8301 produce code that runs somewhat faster on a @samp{Pentium} at the
8302 expense of more memory.
8304 @strong{Warning:} if you use the @option{-malign-double} switch,
8305 structures containing the above types will be aligned differently than
8306 the published application binary interface specifications for the 386
8307 and will not be binary compatible with structures in code compiled
8308 without that switch.
8310 @item -m96bit-long-double
8311 @item -m128bit-long-double
8312 @opindex m96bit-long-double
8313 @opindex m128bit-long-double
8314 These switches control the size of @code{long double} type. The i386
8315 application binary interface specifies the size to be 96 bits,
8316 so @option{-m96bit-long-double} is the default in 32 bit mode.
8318 Modern architectures (Pentium and newer) would prefer @code{long double}
8319 to be aligned to an 8 or 16 byte boundary. In arrays or structures
8320 conforming to the ABI, this would not be possible. So specifying a
8321 @option{-m128bit-long-double} will align @code{long double}
8322 to a 16 byte boundary by padding the @code{long double} with an additional
8325 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
8326 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
8328 Notice that neither of these options enable any extra precision over the x87
8329 standard of 80 bits for a @code{long double}.
8331 @strong{Warning:} if you override the default value for your target ABI, the
8332 structures and arrays containing @code{long double} will change their size as
8333 well as function calling convention for function taking @code{long double}
8334 will be modified. Hence they will not be binary compatible with arrays or
8335 structures in code compiled without that switch.
8339 @itemx -mno-svr3-shlib
8340 @opindex msvr3-shlib
8341 @opindex mno-svr3-shlib
8342 Control whether GCC places uninitialized local variables into the
8343 @code{bss} or @code{data} segments. @option{-msvr3-shlib} places them
8344 into @code{bss}. These options are meaningful only on System V Release 3.
8348 Use a different function-calling convention, in which functions that
8349 take a fixed number of arguments return with the @code{ret} @var{num}
8350 instruction, which pops their arguments while returning. This saves one
8351 instruction in the caller since there is no need to pop the arguments
8354 You can specify that an individual function is called with this calling
8355 sequence with the function attribute @samp{stdcall}. You can also
8356 override the @option{-mrtd} option by using the function attribute
8357 @samp{cdecl}. @xref{Function Attributes}.
8359 @strong{Warning:} this calling convention is incompatible with the one
8360 normally used on Unix, so you cannot use it if you need to call
8361 libraries compiled with the Unix compiler.
8363 Also, you must provide function prototypes for all functions that
8364 take variable numbers of arguments (including @code{printf});
8365 otherwise incorrect code will be generated for calls to those
8368 In addition, seriously incorrect code will result if you call a
8369 function with too many arguments. (Normally, extra arguments are
8370 harmlessly ignored.)
8372 @item -mregparm=@var{num}
8374 Control how many registers are used to pass integer arguments. By
8375 default, no registers are used to pass arguments, and at most 3
8376 registers can be used. You can control this behavior for a specific
8377 function by using the function attribute @samp{regparm}.
8378 @xref{Function Attributes}.
8380 @strong{Warning:} if you use this switch, and
8381 @var{num} is nonzero, then you must build all modules with the same
8382 value, including any libraries. This includes the system libraries and
8385 @item -mpreferred-stack-boundary=@var{num}
8386 @opindex mpreferred-stack-boundary
8387 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
8388 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
8389 the default is 4 (16 bytes or 128 bits), except when optimizing for code
8390 size (@option{-Os}), in which case the default is the minimum correct
8391 alignment (4 bytes for x86, and 8 bytes for x86-64).
8393 On Pentium and PentiumPro, @code{double} and @code{long double} values
8394 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
8395 suffer significant run time performance penalties. On Pentium III, the
8396 Streaming SIMD Extension (SSE) data type @code{__m128} suffers similar
8397 penalties if it is not 16 byte aligned.
8399 To ensure proper alignment of this values on the stack, the stack boundary
8400 must be as aligned as that required by any value stored on the stack.
8401 Further, every function must be generated such that it keeps the stack
8402 aligned. Thus calling a function compiled with a higher preferred
8403 stack boundary from a function compiled with a lower preferred stack
8404 boundary will most likely misalign the stack. It is recommended that
8405 libraries that use callbacks always use the default setting.
8407 This extra alignment does consume extra stack space, and generally
8408 increases code size. Code that is sensitive to stack space usage, such
8409 as embedded systems and operating system kernels, may want to reduce the
8410 preferred alignment to @option{-mpreferred-stack-boundary=2}.
8428 These switches enable or disable the use of built-in functions that allow
8429 direct access to the MMX, SSE and 3Dnow extensions of the instruction set.
8431 @xref{X86 Built-in Functions}, for details of the functions enabled
8432 and disabled by these switches.
8434 To have SSE/SSE2 instructions generated automatically from floating-point
8435 code, see @option{-mfpmath=sse}.
8438 @itemx -mno-push-args
8440 @opindex mno-push-args
8441 Use PUSH operations to store outgoing parameters. This method is shorter
8442 and usually equally fast as method using SUB/MOV operations and is enabled
8443 by default. In some cases disabling it may improve performance because of
8444 improved scheduling and reduced dependencies.
8446 @item -maccumulate-outgoing-args
8447 @opindex maccumulate-outgoing-args
8448 If enabled, the maximum amount of space required for outgoing arguments will be
8449 computed in the function prologue. This is faster on most modern CPUs
8450 because of reduced dependencies, improved scheduling and reduced stack usage
8451 when preferred stack boundary is not equal to 2. The drawback is a notable
8452 increase in code size. This switch implies @option{-mno-push-args}.
8456 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
8457 on thread-safe exception handling must compile and link all code with the
8458 @option{-mthreads} option. When compiling, @option{-mthreads} defines
8459 @option{-D_MT}; when linking, it links in a special thread helper library
8460 @option{-lmingwthrd} which cleans up per thread exception handling data.
8462 @item -mno-align-stringops
8463 @opindex mno-align-stringops
8464 Do not align destination of inlined string operations. This switch reduces
8465 code size and improves performance in case the destination is already aligned,
8466 but gcc don't know about it.
8468 @item -minline-all-stringops
8469 @opindex minline-all-stringops
8470 By default GCC inlines string operations only when destination is known to be
8471 aligned at least to 4 byte boundary. This enables more inlining, increase code
8472 size, but may improve performance of code that depends on fast memcpy, strlen
8473 and memset for short lengths.
8475 @item -momit-leaf-frame-pointer
8476 @opindex momit-leaf-frame-pointer
8477 Don't keep the frame pointer in a register for leaf functions. This
8478 avoids the instructions to save, set up and restore frame pointers and
8479 makes an extra register available in leaf functions. The option
8480 @option{-fomit-frame-pointer} removes the frame pointer for all functions
8481 which might make debugging harder.
8483 @item -mtls-direct-seg-refs
8484 @itemx -mno-tls-direct-seg-refs
8485 @opindex mtls-direct-seg-refs
8486 Controls whether TLS variables may be accessed with offsets from the
8487 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
8488 or whether the thread base pointer must be added. Whether or not this
8489 is legal depends on the operating system, and whether it maps the
8490 segment to cover the entire TLS area.
8492 For systems that use GNU libc, the default is on.
8495 These @samp{-m} switches are supported in addition to the above
8496 on AMD x86-64 processors in 64-bit environments.
8503 Generate code for a 32-bit or 64-bit environment.
8504 The 32-bit environment sets int, long and pointer to 32 bits and
8505 generates code that runs on any i386 system.
8506 The 64-bit environment sets int to 32 bits and long and pointer
8507 to 64 bits and generates code for AMD's x86-64 architecture.
8510 @opindex no-red-zone
8511 Do not use a so called red zone for x86-64 code. The red zone is mandated
8512 by the x86-64 ABI, it is a 128-byte area beyond the location of the
8513 stack pointer that will not be modified by signal or interrupt handlers
8514 and therefore can be used for temporary data without adjusting the stack
8515 pointer. The flag @option{-mno-red-zone} disables this red zone.
8517 @item -mcmodel=small
8518 @opindex mcmodel=small
8519 Generate code for the small code model: the program and its symbols must
8520 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
8521 Programs can be statically or dynamically linked. This is the default
8524 @item -mcmodel=kernel
8525 @opindex mcmodel=kernel
8526 Generate code for the kernel code model. The kernel runs in the
8527 negative 2 GB of the address space.
8528 This model has to be used for Linux kernel code.
8530 @item -mcmodel=medium
8531 @opindex mcmodel=medium
8532 Generate code for the medium model: The program is linked in the lower 2
8533 GB of the address space but symbols can be located anywhere in the
8534 address space. Programs can be statically or dynamically linked, but
8535 building of shared libraries are not supported with the medium model.
8537 @item -mcmodel=large
8538 @opindex mcmodel=large
8539 Generate code for the large model: This model makes no assumptions
8540 about addresses and sizes of sections. Currently GCC does not implement
8545 @subsection HPPA Options
8546 @cindex HPPA Options
8548 These @samp{-m} options are defined for the HPPA family of computers:
8551 @item -march=@var{architecture-type}
8553 Generate code for the specified architecture. The choices for
8554 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
8555 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
8556 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
8557 architecture option for your machine. Code compiled for lower numbered
8558 architectures will run on higher numbered architectures, but not the
8561 PA 2.0 support currently requires gas snapshot 19990413 or later. The
8562 next release of binutils (current is 2.9.1) will probably contain PA 2.0
8566 @itemx -mpa-risc-1-1
8567 @itemx -mpa-risc-2-0
8568 @opindex mpa-risc-1-0
8569 @opindex mpa-risc-1-1
8570 @opindex mpa-risc-2-0
8571 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
8574 @opindex mbig-switch
8575 Generate code suitable for big switch tables. Use this option only if
8576 the assembler/linker complain about out of range branches within a switch
8579 @item -mjump-in-delay
8580 @opindex mjump-in-delay
8581 Fill delay slots of function calls with unconditional jump instructions
8582 by modifying the return pointer for the function call to be the target
8583 of the conditional jump.
8585 @item -mdisable-fpregs
8586 @opindex mdisable-fpregs
8587 Prevent floating point registers from being used in any manner. This is
8588 necessary for compiling kernels which perform lazy context switching of
8589 floating point registers. If you use this option and attempt to perform
8590 floating point operations, the compiler will abort.
8592 @item -mdisable-indexing
8593 @opindex mdisable-indexing
8594 Prevent the compiler from using indexing address modes. This avoids some
8595 rather obscure problems when compiling MIG generated code under MACH@.
8597 @item -mno-space-regs
8598 @opindex mno-space-regs
8599 Generate code that assumes the target has no space registers. This allows
8600 GCC to generate faster indirect calls and use unscaled index address modes.
8602 Such code is suitable for level 0 PA systems and kernels.
8604 @item -mfast-indirect-calls
8605 @opindex mfast-indirect-calls
8606 Generate code that assumes calls never cross space boundaries. This
8607 allows GCC to emit code which performs faster indirect calls.
8609 This option will not work in the presence of shared libraries or nested
8612 @item -mlong-load-store
8613 @opindex mlong-load-store
8614 Generate 3-instruction load and store sequences as sometimes required by
8615 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
8618 @item -mportable-runtime
8619 @opindex mportable-runtime
8620 Use the portable calling conventions proposed by HP for ELF systems.
8624 Enable the use of assembler directives only GAS understands.
8626 @item -mschedule=@var{cpu-type}
8628 Schedule code according to the constraints for the machine type
8629 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
8630 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
8631 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
8632 proper scheduling option for your machine. The default scheduling is
8636 @opindex mlinker-opt
8637 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
8638 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
8639 linkers in which they give bogus error messages when linking some programs.
8642 @opindex msoft-float
8643 Generate output containing library calls for floating point.
8644 @strong{Warning:} the requisite libraries are not available for all HPPA
8645 targets. Normally the facilities of the machine's usual C compiler are
8646 used, but this cannot be done directly in cross-compilation. You must make
8647 your own arrangements to provide suitable library functions for
8648 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
8649 does provide software floating point support.
8651 @option{-msoft-float} changes the calling convention in the output file;
8652 therefore, it is only useful if you compile @emph{all} of a program with
8653 this option. In particular, you need to compile @file{libgcc.a}, the
8654 library that comes with GCC, with @option{-msoft-float} in order for
8659 Generate the predefine, @code{_SIO}, for server IO. The default is
8660 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
8661 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO. These
8662 options are available under HP-UX and HI-UX.
8666 Use GNU ld specific options. This passes @option{-shared} to ld when
8667 building a shared library. It is the default when GCC is configured,
8668 explicitly or implicitly, with the GNU linker. This option does not
8669 have any affect on which ld is called, it only changes what parameters
8670 are passed to that ld. The ld that is called is determined by the
8671 @option{--with-ld} configure option, gcc's program search path, and
8672 finally by the user's @env{PATH}. The linker used by GCC can be printed
8673 using @samp{which `gcc -print-prog-name=ld`}.
8677 Use HP ld specific options. This passes @option{-b} to ld when building
8678 a shared library and passes @option{+Accept TypeMismatch} to ld on all
8679 links. It is the default when GCC is configured, explicitly or
8680 implicitly, with the HP linker. This option does not have any affect on
8681 which ld is called, it only changes what parameters are passed to that
8682 ld. The ld that is called is determined by the @option{--with-ld}
8683 configure option, gcc's program search path, and finally by the user's
8684 @env{PATH}. The linker used by GCC can be printed using @samp{which
8685 `gcc -print-prog-name=ld`}.
8688 @opindex mno-long-calls
8689 Generate code that uses long call sequences. This ensures that a call
8690 is always able to reach linker generated stubs. The default is to generate
8691 long calls only when the distance from the call site to the beginning
8692 of the function or translation unit, as the case may be, exceeds a
8693 predefined limit set by the branch type being used. The limits for
8694 normal calls are 7,600,000 and 240,000 bytes, respectively for the
8695 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
8698 Distances are measured from the beginning of functions when using the
8699 @option{-ffunction-sections} option, or when using the @option{-mgas}
8700 and @option{-mno-portable-runtime} options together under HP-UX with
8703 It is normally not desirable to use this option as it will degrade
8704 performance. However, it may be useful in large applications,
8705 particularly when partial linking is used to build the application.
8707 The types of long calls used depends on the capabilities of the
8708 assembler and linker, and the type of code being generated. The
8709 impact on systems that support long absolute calls, and long pic
8710 symbol-difference or pc-relative calls should be relatively small.
8711 However, an indirect call is used on 32-bit ELF systems in pic code
8712 and it is quite long.
8716 Suppress the generation of link options to search libdld.sl when the
8717 @option{-static} option is specified on HP-UX 10 and later.
8721 The HP-UX implementation of setlocale in libc has a dependency on
8722 libdld.sl. There isn't an archive version of libdld.sl. Thus,
8723 when the @option{-static} option is specified, special link options
8724 are needed to resolve this dependency.
8726 On HP-UX 10 and later, the GCC driver adds the necessary options to
8727 link with libdld.sl when the @option{-static} option is specified.
8728 This causes the resulting binary to be dynamic. On the 64-bit port,
8729 the linkers generate dynamic binaries by default in any case. The
8730 @option{-nolibdld} option can be used to prevent the GCC driver from
8731 adding these link options.
8735 Add support for multithreading with the @dfn{dce thread} library
8736 under HP-UX. This option sets flags for both the preprocessor and
8740 @node Intel 960 Options
8741 @subsection Intel 960 Options
8743 These @samp{-m} options are defined for the Intel 960 implementations:
8746 @item -m@var{cpu-type}
8754 Assume the defaults for the machine type @var{cpu-type} for some of
8755 the other options, including instruction scheduling, floating point
8756 support, and addressing modes. The choices for @var{cpu-type} are
8757 @samp{ka}, @samp{kb}, @samp{mc}, @samp{ca}, @samp{cf},
8758 @samp{sa}, and @samp{sb}.
8765 @opindex msoft-float
8766 The @option{-mnumerics} option indicates that the processor does support
8767 floating-point instructions. The @option{-msoft-float} option indicates
8768 that floating-point support should not be assumed.
8770 @item -mleaf-procedures
8771 @itemx -mno-leaf-procedures
8772 @opindex mleaf-procedures
8773 @opindex mno-leaf-procedures
8774 Do (or do not) attempt to alter leaf procedures to be callable with the
8775 @code{bal} instruction as well as @code{call}. This will result in more
8776 efficient code for explicit calls when the @code{bal} instruction can be
8777 substituted by the assembler or linker, but less efficient code in other
8778 cases, such as calls via function pointers, or using a linker that doesn't
8779 support this optimization.
8782 @itemx -mno-tail-call
8784 @opindex mno-tail-call
8785 Do (or do not) make additional attempts (beyond those of the
8786 machine-independent portions of the compiler) to optimize tail-recursive
8787 calls into branches. You may not want to do this because the detection of
8788 cases where this is not valid is not totally complete. The default is
8789 @option{-mno-tail-call}.
8791 @item -mcomplex-addr
8792 @itemx -mno-complex-addr
8793 @opindex mcomplex-addr
8794 @opindex mno-complex-addr
8795 Assume (or do not assume) that the use of a complex addressing mode is a
8796 win on this implementation of the i960. Complex addressing modes may not
8797 be worthwhile on the K-series, but they definitely are on the C-series.
8798 The default is currently @option{-mcomplex-addr} for all processors except
8802 @itemx -mno-code-align
8803 @opindex mcode-align
8804 @opindex mno-code-align
8805 Align code to 8-byte boundaries for faster fetching (or don't bother).
8806 Currently turned on by default for C-series implementations only.
8809 @item -mclean-linkage
8810 @itemx -mno-clean-linkage
8811 @opindex mclean-linkage
8812 @opindex mno-clean-linkage
8813 These options are not fully implemented.
8817 @itemx -mic2.0-compat
8818 @itemx -mic3.0-compat
8820 @opindex mic2.0-compat
8821 @opindex mic3.0-compat
8822 Enable compatibility with iC960 v2.0 or v3.0.
8826 @opindex masm-compat
8828 Enable compatibility with the iC960 assembler.
8830 @item -mstrict-align
8831 @itemx -mno-strict-align
8832 @opindex mstrict-align
8833 @opindex mno-strict-align
8834 Do not permit (do permit) unaligned accesses.
8838 Enable structure-alignment compatibility with Intel's gcc release version
8839 1.3 (based on gcc 1.37). This option implies @option{-mstrict-align}.
8841 @item -mlong-double-64
8842 @opindex mlong-double-64
8843 Implement type @samp{long double} as 64-bit floating point numbers.
8844 Without the option @samp{long double} is implemented by 80-bit
8845 floating point numbers. The only reason we have it because there is
8846 no 128-bit @samp{long double} support in @samp{fp-bit.c} yet. So it
8847 is only useful for people using soft-float targets. Otherwise, we
8848 should recommend against use of it.
8852 @node DEC Alpha Options
8853 @subsection DEC Alpha Options
8855 These @samp{-m} options are defined for the DEC Alpha implementations:
8858 @item -mno-soft-float
8860 @opindex mno-soft-float
8861 @opindex msoft-float
8862 Use (do not use) the hardware floating-point instructions for
8863 floating-point operations. When @option{-msoft-float} is specified,
8864 functions in @file{libgcc.a} will be used to perform floating-point
8865 operations. Unless they are replaced by routines that emulate the
8866 floating-point operations, or compiled in such a way as to call such
8867 emulations routines, these routines will issue floating-point
8868 operations. If you are compiling for an Alpha without floating-point
8869 operations, you must ensure that the library is built so as not to call
8872 Note that Alpha implementations without floating-point operations are
8873 required to have floating-point registers.
8878 @opindex mno-fp-regs
8879 Generate code that uses (does not use) the floating-point register set.
8880 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
8881 register set is not used, floating point operands are passed in integer
8882 registers as if they were integers and floating-point results are passed
8883 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
8884 so any function with a floating-point argument or return value called by code
8885 compiled with @option{-mno-fp-regs} must also be compiled with that
8888 A typical use of this option is building a kernel that does not use,
8889 and hence need not save and restore, any floating-point registers.
8893 The Alpha architecture implements floating-point hardware optimized for
8894 maximum performance. It is mostly compliant with the IEEE floating
8895 point standard. However, for full compliance, software assistance is
8896 required. This option generates code fully IEEE compliant code
8897 @emph{except} that the @var{inexact-flag} is not maintained (see below).
8898 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
8899 defined during compilation. The resulting code is less efficient but is
8900 able to correctly support denormalized numbers and exceptional IEEE
8901 values such as not-a-number and plus/minus infinity. Other Alpha
8902 compilers call this option @option{-ieee_with_no_inexact}.
8904 @item -mieee-with-inexact
8905 @opindex mieee-with-inexact
8906 This is like @option{-mieee} except the generated code also maintains
8907 the IEEE @var{inexact-flag}. Turning on this option causes the
8908 generated code to implement fully-compliant IEEE math. In addition to
8909 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
8910 macro. On some Alpha implementations the resulting code may execute
8911 significantly slower than the code generated by default. Since there is
8912 very little code that depends on the @var{inexact-flag}, you should
8913 normally not specify this option. Other Alpha compilers call this
8914 option @option{-ieee_with_inexact}.
8916 @item -mfp-trap-mode=@var{trap-mode}
8917 @opindex mfp-trap-mode
8918 This option controls what floating-point related traps are enabled.
8919 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
8920 The trap mode can be set to one of four values:
8924 This is the default (normal) setting. The only traps that are enabled
8925 are the ones that cannot be disabled in software (e.g., division by zero
8929 In addition to the traps enabled by @samp{n}, underflow traps are enabled
8933 Like @samp{su}, but the instructions are marked to be safe for software
8934 completion (see Alpha architecture manual for details).
8937 Like @samp{su}, but inexact traps are enabled as well.
8940 @item -mfp-rounding-mode=@var{rounding-mode}
8941 @opindex mfp-rounding-mode
8942 Selects the IEEE rounding mode. Other Alpha compilers call this option
8943 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
8948 Normal IEEE rounding mode. Floating point numbers are rounded towards
8949 the nearest machine number or towards the even machine number in case
8953 Round towards minus infinity.
8956 Chopped rounding mode. Floating point numbers are rounded towards zero.
8959 Dynamic rounding mode. A field in the floating point control register
8960 (@var{fpcr}, see Alpha architecture reference manual) controls the
8961 rounding mode in effect. The C library initializes this register for
8962 rounding towards plus infinity. Thus, unless your program modifies the
8963 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
8966 @item -mtrap-precision=@var{trap-precision}
8967 @opindex mtrap-precision
8968 In the Alpha architecture, floating point traps are imprecise. This
8969 means without software assistance it is impossible to recover from a
8970 floating trap and program execution normally needs to be terminated.
8971 GCC can generate code that can assist operating system trap handlers
8972 in determining the exact location that caused a floating point trap.
8973 Depending on the requirements of an application, different levels of
8974 precisions can be selected:
8978 Program precision. This option is the default and means a trap handler
8979 can only identify which program caused a floating point exception.
8982 Function precision. The trap handler can determine the function that
8983 caused a floating point exception.
8986 Instruction precision. The trap handler can determine the exact
8987 instruction that caused a floating point exception.
8990 Other Alpha compilers provide the equivalent options called
8991 @option{-scope_safe} and @option{-resumption_safe}.
8993 @item -mieee-conformant
8994 @opindex mieee-conformant
8995 This option marks the generated code as IEEE conformant. You must not
8996 use this option unless you also specify @option{-mtrap-precision=i} and either
8997 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
8998 is to emit the line @samp{.eflag 48} in the function prologue of the
8999 generated assembly file. Under DEC Unix, this has the effect that
9000 IEEE-conformant math library routines will be linked in.
9002 @item -mbuild-constants
9003 @opindex mbuild-constants
9004 Normally GCC examines a 32- or 64-bit integer constant to
9005 see if it can construct it from smaller constants in two or three
9006 instructions. If it cannot, it will output the constant as a literal and
9007 generate code to load it from the data segment at runtime.
9009 Use this option to require GCC to construct @emph{all} integer constants
9010 using code, even if it takes more instructions (the maximum is six).
9012 You would typically use this option to build a shared library dynamic
9013 loader. Itself a shared library, it must relocate itself in memory
9014 before it can find the variables and constants in its own data segment.
9020 Select whether to generate code to be assembled by the vendor-supplied
9021 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
9039 Indicate whether GCC should generate code to use the optional BWX,
9040 CIX, FIX and MAX instruction sets. The default is to use the instruction
9041 sets supported by the CPU type specified via @option{-mcpu=} option or that
9042 of the CPU on which GCC was built if none was specified.
9047 @opindex mfloat-ieee
9048 Generate code that uses (does not use) VAX F and G floating point
9049 arithmetic instead of IEEE single and double precision.
9051 @item -mexplicit-relocs
9052 @itemx -mno-explicit-relocs
9053 @opindex mexplicit-relocs
9054 @opindex mno-explicit-relocs
9055 Older Alpha assemblers provided no way to generate symbol relocations
9056 except via assembler macros. Use of these macros does not allow
9057 optimal instruction scheduling. GNU binutils as of version 2.12
9058 supports a new syntax that allows the compiler to explicitly mark
9059 which relocations should apply to which instructions. This option
9060 is mostly useful for debugging, as GCC detects the capabilities of
9061 the assembler when it is built and sets the default accordingly.
9065 @opindex msmall-data
9066 @opindex mlarge-data
9067 When @option{-mexplicit-relocs} is in effect, static data is
9068 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
9069 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
9070 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
9071 16-bit relocations off of the @code{$gp} register. This limits the
9072 size of the small data area to 64KB, but allows the variables to be
9073 directly accessed via a single instruction.
9075 The default is @option{-mlarge-data}. With this option the data area
9076 is limited to just below 2GB. Programs that require more than 2GB of
9077 data must use @code{malloc} or @code{mmap} to allocate the data in the
9078 heap instead of in the program's data segment.
9080 When generating code for shared libraries, @option{-fpic} implies
9081 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
9085 @opindex msmall-text
9086 @opindex mlarge-text
9087 When @option{-msmall-text} is used, the compiler assumes that the
9088 code of the entire program (or shared library) fits in 4MB, and is
9089 thus reachable with a branch instruction. When @option{-msmall-data}
9090 is used, the compiler can assume that all local symbols share the
9091 same @code{$gp} value, and thus reduce the number of instructions
9092 required for a function call from 4 to 1.
9094 The default is @option{-mlarge-text}.
9096 @item -mcpu=@var{cpu_type}
9098 Set the instruction set and instruction scheduling parameters for
9099 machine type @var{cpu_type}. You can specify either the @samp{EV}
9100 style name or the corresponding chip number. GCC supports scheduling
9101 parameters for the EV4, EV5 and EV6 family of processors and will
9102 choose the default values for the instruction set from the processor
9103 you specify. If you do not specify a processor type, GCC will default
9104 to the processor on which the compiler was built.
9106 Supported values for @var{cpu_type} are
9112 Schedules as an EV4 and has no instruction set extensions.
9116 Schedules as an EV5 and has no instruction set extensions.
9120 Schedules as an EV5 and supports the BWX extension.
9125 Schedules as an EV5 and supports the BWX and MAX extensions.
9129 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
9133 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
9136 @item -mtune=@var{cpu_type}
9138 Set only the instruction scheduling parameters for machine type
9139 @var{cpu_type}. The instruction set is not changed.
9141 @item -mmemory-latency=@var{time}
9142 @opindex mmemory-latency
9143 Sets the latency the scheduler should assume for typical memory
9144 references as seen by the application. This number is highly
9145 dependent on the memory access patterns used by the application
9146 and the size of the external cache on the machine.
9148 Valid options for @var{time} are
9152 A decimal number representing clock cycles.
9158 The compiler contains estimates of the number of clock cycles for
9159 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
9160 (also called Dcache, Scache, and Bcache), as well as to main memory.
9161 Note that L3 is only valid for EV5.
9166 @node DEC Alpha/VMS Options
9167 @subsection DEC Alpha/VMS Options
9169 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
9172 @item -mvms-return-codes
9173 @opindex mvms-return-codes
9174 Return VMS condition codes from main. The default is to return POSIX
9175 style condition (e.g.@ error) codes.
9178 @node H8/300 Options
9179 @subsection H8/300 Options
9181 These @samp{-m} options are defined for the H8/300 implementations:
9186 Shorten some address references at link time, when possible; uses the
9187 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
9188 ld.info, Using ld}, for a fuller description.
9192 Generate code for the H8/300H@.
9196 Generate code for the H8S@.
9200 Generate code for the H8S and H8/300H in the normal mode. This switch
9201 must be used either with -mh or -ms.
9205 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
9209 Make @code{int} data 32 bits by default.
9213 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
9214 The default for the H8/300H and H8S is to align longs and floats on 4
9216 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
9217 This option has no effect on the H8/300.
9221 @subsection SH Options
9223 These @samp{-m} options are defined for the SH implementations:
9228 Generate code for the SH1.
9232 Generate code for the SH2.
9235 Generate code for the SH2e.
9239 Generate code for the SH3.
9243 Generate code for the SH3e.
9247 Generate code for the SH4 without a floating-point unit.
9249 @item -m4-single-only
9250 @opindex m4-single-only
9251 Generate code for the SH4 with a floating-point unit that only
9252 supports single-precision arithmetic.
9256 Generate code for the SH4 assuming the floating-point unit is in
9257 single-precision mode by default.
9261 Generate code for the SH4.
9265 Compile code for the processor in big endian mode.
9269 Compile code for the processor in little endian mode.
9273 Align doubles at 64-bit boundaries. Note that this changes the calling
9274 conventions, and thus some functions from the standard C library will
9275 not work unless you recompile it first with @option{-mdalign}.
9279 Shorten some address references at link time, when possible; uses the
9280 linker option @option{-relax}.
9284 Use 32-bit offsets in @code{switch} tables. The default is to use
9289 Enable the use of the instruction @code{fmovd}.
9293 Comply with the calling conventions defined by Renesas.
9297 Mark the @code{MAC} register as call-clobbered, even if
9298 @option{-mhitachi} is given.
9302 Increase IEEE-compliance of floating-point code.
9306 Dump instruction size and location in the assembly code.
9310 This option is deprecated. It pads structures to multiple of 4 bytes,
9311 which is incompatible with the SH ABI@.
9315 Optimize for space instead of speed. Implied by @option{-Os}.
9319 When generating position-independent code, emit function calls using
9320 the Global Offset Table instead of the Procedure Linkage Table.
9324 Generate a library function call to invalidate instruction cache
9325 entries, after fixing up a trampoline. This library function call
9326 doesn't assume it can write to the whole memory address space. This
9327 is the default when the target is @code{sh-*-linux*}.
9330 @node System V Options
9331 @subsection Options for System V
9333 These additional options are available on System V Release 4 for
9334 compatibility with other compilers on those systems:
9339 Create a shared object.
9340 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
9344 Identify the versions of each tool used by the compiler, in a
9345 @code{.ident} assembler directive in the output.
9349 Refrain from adding @code{.ident} directives to the output file (this is
9352 @item -YP,@var{dirs}
9354 Search the directories @var{dirs}, and no others, for libraries
9355 specified with @option{-l}.
9359 Look in the directory @var{dir} to find the M4 preprocessor.
9360 The assembler uses this option.
9361 @c This is supposed to go with a -Yd for predefined M4 macro files, but
9362 @c the generic assembler that comes with Solaris takes just -Ym.
9365 @node TMS320C3x/C4x Options
9366 @subsection TMS320C3x/C4x Options
9367 @cindex TMS320C3x/C4x Options
9369 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
9373 @item -mcpu=@var{cpu_type}
9375 Set the instruction set, register set, and instruction scheduling
9376 parameters for machine type @var{cpu_type}. Supported values for
9377 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
9378 @samp{c44}. The default is @samp{c40} to generate code for the
9383 @itemx -msmall-memory
9385 @opindex mbig-memory
9387 @opindex msmall-memory
9389 Generates code for the big or small memory model. The small memory
9390 model assumed that all data fits into one 64K word page. At run-time
9391 the data page (DP) register must be set to point to the 64K page
9392 containing the .bss and .data program sections. The big memory model is
9393 the default and requires reloading of the DP register for every direct
9400 Allow (disallow) allocation of general integer operands into the block
9407 Enable (disable) generation of code using decrement and branch,
9408 DBcond(D), instructions. This is enabled by default for the C4x. To be
9409 on the safe side, this is disabled for the C3x, since the maximum
9410 iteration count on the C3x is @math{2^{23} + 1} (but who iterates loops more than
9411 @math{2^{23}} times on the C3x?). Note that GCC will try to reverse a loop so
9412 that it can utilize the decrement and branch instruction, but will give
9413 up if there is more than one memory reference in the loop. Thus a loop
9414 where the loop counter is decremented can generate slightly more
9415 efficient code, in cases where the RPTB instruction cannot be utilized.
9417 @item -mdp-isr-reload
9419 @opindex mdp-isr-reload
9421 Force the DP register to be saved on entry to an interrupt service
9422 routine (ISR), reloaded to point to the data section, and restored on
9423 exit from the ISR@. This should not be required unless someone has
9424 violated the small memory model by modifying the DP register, say within
9431 For the C3x use the 24-bit MPYI instruction for integer multiplies
9432 instead of a library call to guarantee 32-bit results. Note that if one
9433 of the operands is a constant, then the multiplication will be performed
9434 using shifts and adds. If the @option{-mmpyi} option is not specified for the C3x,
9435 then squaring operations are performed inline instead of a library call.
9438 @itemx -mno-fast-fix
9440 @opindex mno-fast-fix
9441 The C3x/C4x FIX instruction to convert a floating point value to an
9442 integer value chooses the nearest integer less than or equal to the
9443 floating point value rather than to the nearest integer. Thus if the
9444 floating point number is negative, the result will be incorrectly
9445 truncated an additional code is necessary to detect and correct this
9446 case. This option can be used to disable generation of the additional
9447 code required to correct the result.
9453 Enable (disable) generation of repeat block sequences using the RPTB
9454 instruction for zero overhead looping. The RPTB construct is only used
9455 for innermost loops that do not call functions or jump across the loop
9456 boundaries. There is no advantage having nested RPTB loops due to the
9457 overhead required to save and restore the RC, RS, and RE registers.
9458 This is enabled by default with @option{-O2}.
9460 @item -mrpts=@var{count}
9464 Enable (disable) the use of the single instruction repeat instruction
9465 RPTS@. If a repeat block contains a single instruction, and the loop
9466 count can be guaranteed to be less than the value @var{count}, GCC will
9467 emit a RPTS instruction instead of a RPTB@. If no value is specified,
9468 then a RPTS will be emitted even if the loop count cannot be determined
9469 at compile time. Note that the repeated instruction following RPTS does
9470 not have to be reloaded from memory each iteration, thus freeing up the
9471 CPU buses for operands. However, since interrupts are blocked by this
9472 instruction, it is disabled by default.
9474 @item -mloop-unsigned
9475 @itemx -mno-loop-unsigned
9476 @opindex mloop-unsigned
9477 @opindex mno-loop-unsigned
9478 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
9479 is @math{2^{31} + 1} since these instructions test if the iteration count is
9480 negative to terminate the loop. If the iteration count is unsigned
9481 there is a possibility than the @math{2^{31} + 1} maximum iteration count may be
9482 exceeded. This switch allows an unsigned iteration count.
9486 Try to emit an assembler syntax that the TI assembler (asm30) is happy
9487 with. This also enforces compatibility with the API employed by the TI
9488 C3x C compiler. For example, long doubles are passed as structures
9489 rather than in floating point registers.
9495 Generate code that uses registers (stack) for passing arguments to functions.
9496 By default, arguments are passed in registers where possible rather
9497 than by pushing arguments on to the stack.
9499 @item -mparallel-insns
9500 @itemx -mno-parallel-insns
9501 @opindex mparallel-insns
9502 @opindex mno-parallel-insns
9503 Allow the generation of parallel instructions. This is enabled by
9504 default with @option{-O2}.
9506 @item -mparallel-mpy
9507 @itemx -mno-parallel-mpy
9508 @opindex mparallel-mpy
9509 @opindex mno-parallel-mpy
9510 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
9511 provided @option{-mparallel-insns} is also specified. These instructions have
9512 tight register constraints which can pessimize the code generation
9518 @subsection V850 Options
9519 @cindex V850 Options
9521 These @samp{-m} options are defined for V850 implementations:
9525 @itemx -mno-long-calls
9526 @opindex mlong-calls
9527 @opindex mno-long-calls
9528 Treat all calls as being far away (near). If calls are assumed to be
9529 far away, the compiler will always load the functions address up into a
9530 register, and call indirect through the pointer.
9536 Do not optimize (do optimize) basic blocks that use the same index
9537 pointer 4 or more times to copy pointer into the @code{ep} register, and
9538 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
9539 option is on by default if you optimize.
9541 @item -mno-prolog-function
9542 @itemx -mprolog-function
9543 @opindex mno-prolog-function
9544 @opindex mprolog-function
9545 Do not use (do use) external functions to save and restore registers at
9546 the prolog and epilog of a function. The external functions are slower,
9547 but use less code space if more than one function saves the same number
9548 of registers. The @option{-mprolog-function} option is on by default if
9553 Try to make the code as small as possible. At present, this just turns
9554 on the @option{-mep} and @option{-mprolog-function} options.
9558 Put static or global variables whose size is @var{n} bytes or less into
9559 the tiny data area that register @code{ep} points to. The tiny data
9560 area can hold up to 256 bytes in total (128 bytes for byte references).
9564 Put static or global variables whose size is @var{n} bytes or less into
9565 the small data area that register @code{gp} points to. The small data
9566 area can hold up to 64 kilobytes.
9570 Put static or global variables whose size is @var{n} bytes or less into
9571 the first 32 kilobytes of memory.
9575 Specify that the target processor is the V850.
9578 @opindex mbig-switch
9579 Generate code suitable for big switch tables. Use this option only if
9580 the assembler/linker complain about out of range branches within a switch
9585 This option will cause r2 and r5 to be used in the code generated by
9586 the compiler. This setting is the default.
9589 @opindex mno-app-regs
9590 This option will cause r2 and r5 to be treated as fixed registers.
9594 Specify that the target processor is the V850E. The preprocessor
9595 constant @samp{__v850e__} will be defined if this option is used.
9597 If neither @option{-mv850} nor @option{-mv850e} are defined
9598 then a default target processor will be chosen and the relevant
9599 @samp{__v850*__} preprocessor constant will be defined.
9601 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
9602 defined, regardless of which processor variant is the target.
9604 @item -mdisable-callt
9605 @opindex mdisable-callt
9606 This option will suppress generation of the CALLT instruction for the
9607 v850e flavors of the v850 architecture. The default is
9608 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
9613 @subsection ARC Options
9616 These options are defined for ARC implementations:
9621 Compile code for little endian mode. This is the default.
9625 Compile code for big endian mode.
9628 @opindex mmangle-cpu
9629 Prepend the name of the cpu to all public symbol names.
9630 In multiple-processor systems, there are many ARC variants with different
9631 instruction and register set characteristics. This flag prevents code
9632 compiled for one cpu to be linked with code compiled for another.
9633 No facility exists for handling variants that are ``almost identical''.
9634 This is an all or nothing option.
9636 @item -mcpu=@var{cpu}
9638 Compile code for ARC variant @var{cpu}.
9639 Which variants are supported depend on the configuration.
9640 All variants support @option{-mcpu=base}, this is the default.
9642 @item -mtext=@var{text-section}
9643 @itemx -mdata=@var{data-section}
9644 @itemx -mrodata=@var{readonly-data-section}
9648 Put functions, data, and readonly data in @var{text-section},
9649 @var{data-section}, and @var{readonly-data-section} respectively
9650 by default. This can be overridden with the @code{section} attribute.
9651 @xref{Variable Attributes}.
9656 @subsection NS32K Options
9657 @cindex NS32K options
9659 These are the @samp{-m} options defined for the 32000 series. The default
9660 values for these options depends on which style of 32000 was selected when
9661 the compiler was configured; the defaults for the most common choices are
9669 Generate output for a 32032. This is the default
9670 when the compiler is configured for 32032 and 32016 based systems.
9676 Generate output for a 32332. This is the default
9677 when the compiler is configured for 32332-based systems.
9683 Generate output for a 32532. This is the default
9684 when the compiler is configured for 32532-based systems.
9688 Generate output containing 32081 instructions for floating point.
9689 This is the default for all systems.
9693 Generate output containing 32381 instructions for floating point. This
9694 also implies @option{-m32081}. The 32381 is only compatible with the 32332
9695 and 32532 cpus. This is the default for the pc532-netbsd configuration.
9699 Try and generate multiply-add floating point instructions @code{polyF}
9700 and @code{dotF}. This option is only available if the @option{-m32381}
9701 option is in effect. Using these instructions requires changes to
9702 register allocation which generally has a negative impact on
9703 performance. This option should only be enabled when compiling code
9704 particularly likely to make heavy use of multiply-add instructions.
9707 @opindex mnomulti-add
9708 Do not try and generate multiply-add floating point instructions
9709 @code{polyF} and @code{dotF}. This is the default on all platforms.
9712 @opindex msoft-float
9713 Generate output containing library calls for floating point.
9714 @strong{Warning:} the requisite libraries may not be available.
9716 @item -mieee-compare
9717 @itemx -mno-ieee-compare
9718 @opindex mieee-compare
9719 @opindex mno-ieee-compare
9720 Control whether or not the compiler uses IEEE floating point
9721 comparisons. These handle correctly the case where the result of a
9722 comparison is unordered.
9723 @strong{Warning:} the requisite kernel support may not be available.
9726 @opindex mnobitfield
9727 Do not use the bit-field instructions. On some machines it is faster to
9728 use shifting and masking operations. This is the default for the pc532.
9732 Do use the bit-field instructions. This is the default for all platforms
9737 Use a different function-calling convention, in which functions
9738 that take a fixed number of arguments return pop their
9739 arguments on return with the @code{ret} instruction.
9741 This calling convention is incompatible with the one normally
9742 used on Unix, so you cannot use it if you need to call libraries
9743 compiled with the Unix compiler.
9745 Also, you must provide function prototypes for all functions that
9746 take variable numbers of arguments (including @code{printf});
9747 otherwise incorrect code will be generated for calls to those
9750 In addition, seriously incorrect code will result if you call a
9751 function with too many arguments. (Normally, extra arguments are
9752 harmlessly ignored.)
9754 This option takes its name from the 680x0 @code{rtd} instruction.
9759 Use a different function-calling convention where the first two arguments
9760 are passed in registers.
9762 This calling convention is incompatible with the one normally
9763 used on Unix, so you cannot use it if you need to call libraries
9764 compiled with the Unix compiler.
9767 @opindex mnoregparam
9768 Do not pass any arguments in registers. This is the default for all
9773 It is OK to use the sb as an index register which is always loaded with
9774 zero. This is the default for the pc532-netbsd target.
9778 The sb register is not available for use or has not been initialized to
9779 zero by the run time system. This is the default for all targets except
9780 the pc532-netbsd. It is also implied whenever @option{-mhimem} or
9781 @option{-fpic} is set.
9785 Many ns32000 series addressing modes use displacements of up to 512MB@.
9786 If an address is above 512MB then displacements from zero can not be used.
9787 This option causes code to be generated which can be loaded above 512MB@.
9788 This may be useful for operating systems or ROM code.
9792 Assume code will be loaded in the first 512MB of virtual address space.
9793 This is the default for all platforms.
9799 @subsection AVR Options
9802 These options are defined for AVR implementations:
9805 @item -mmcu=@var{mcu}
9807 Specify ATMEL AVR instruction set or MCU type.
9809 Instruction set avr1 is for the minimal AVR core, not supported by the C
9810 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
9811 attiny11, attiny12, attiny15, attiny28).
9813 Instruction set avr2 (default) is for the classic AVR core with up to
9814 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
9815 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
9816 at90c8534, at90s8535).
9818 Instruction set avr3 is for the classic AVR core with up to 128K program
9819 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
9821 Instruction set avr4 is for the enhanced AVR core with up to 8K program
9822 memory space (MCU types: atmega8, atmega83, atmega85).
9824 Instruction set avr5 is for the enhanced AVR core with up to 128K program
9825 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
9826 atmega64, atmega128, at43usb355, at94k).
9830 Output instruction sizes to the asm file.
9832 @item -minit-stack=@var{N}
9833 @opindex minit-stack
9834 Specify the initial stack address, which may be a symbol or numeric value,
9835 @samp{__stack} is the default.
9837 @item -mno-interrupts
9838 @opindex mno-interrupts
9839 Generated code is not compatible with hardware interrupts.
9840 Code size will be smaller.
9842 @item -mcall-prologues
9843 @opindex mcall-prologues
9844 Functions prologues/epilogues expanded as call to appropriate
9845 subroutines. Code size will be smaller.
9847 @item -mno-tablejump
9848 @opindex mno-tablejump
9849 Do not generate tablejump insns which sometimes increase code size.
9852 @opindex mtiny-stack
9853 Change only the low 8 bits of the stack pointer.
9857 @subsection MCore Options
9858 @cindex MCore options
9860 These are the @samp{-m} options defined for the Motorola M*Core
9868 @opindex mno-hardlit
9869 Inline constants into the code stream if it can be done in two
9870 instructions or less.
9876 Use the divide instruction. (Enabled by default).
9878 @item -mrelax-immediate
9879 @itemx -mno-relax-immediate
9880 @opindex mrelax-immediate
9881 @opindex mno-relax-immediate
9882 Allow arbitrary sized immediates in bit operations.
9884 @item -mwide-bitfields
9885 @itemx -mno-wide-bitfields
9886 @opindex mwide-bitfields
9887 @opindex mno-wide-bitfields
9888 Always treat bit-fields as int-sized.
9890 @item -m4byte-functions
9891 @itemx -mno-4byte-functions
9892 @opindex m4byte-functions
9893 @opindex mno-4byte-functions
9894 Force all functions to be aligned to a four byte boundary.
9896 @item -mcallgraph-data
9897 @itemx -mno-callgraph-data
9898 @opindex mcallgraph-data
9899 @opindex mno-callgraph-data
9900 Emit callgraph information.
9903 @itemx -mno-slow-bytes
9904 @opindex mslow-bytes
9905 @opindex mno-slow-bytes
9906 Prefer word access when reading byte quantities.
9908 @item -mlittle-endian
9910 @opindex mlittle-endian
9911 @opindex mbig-endian
9912 Generate code for a little endian target.
9918 Generate code for the 210 processor.
9922 @subsection IA-64 Options
9923 @cindex IA-64 Options
9925 These are the @samp{-m} options defined for the Intel IA-64 architecture.
9929 @opindex mbig-endian
9930 Generate code for a big endian target. This is the default for HP-UX@.
9932 @item -mlittle-endian
9933 @opindex mlittle-endian
9934 Generate code for a little endian target. This is the default for AIX5
9941 Generate (or don't) code for the GNU assembler. This is the default.
9942 @c Also, this is the default if the configure option @option{--with-gnu-as}
9949 Generate (or don't) code for the GNU linker. This is the default.
9950 @c Also, this is the default if the configure option @option{--with-gnu-ld}
9955 Generate code that does not use a global pointer register. The result
9956 is not position independent code, and violates the IA-64 ABI@.
9958 @item -mvolatile-asm-stop
9959 @itemx -mno-volatile-asm-stop
9960 @opindex mvolatile-asm-stop
9961 @opindex mno-volatile-asm-stop
9962 Generate (or don't) a stop bit immediately before and after volatile asm
9967 Generate code that works around Itanium B step errata.
9969 @item -mregister-names
9970 @itemx -mno-register-names
9971 @opindex mregister-names
9972 @opindex mno-register-names
9973 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
9974 the stacked registers. This may make assembler output more readable.
9980 Disable (or enable) optimizations that use the small data section. This may
9981 be useful for working around optimizer bugs.
9984 @opindex mconstant-gp
9985 Generate code that uses a single constant global pointer value. This is
9986 useful when compiling kernel code.
9990 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
9991 This is useful when compiling firmware code.
9993 @item -minline-float-divide-min-latency
9994 @opindex minline-float-divide-min-latency
9995 Generate code for inline divides of floating point values
9996 using the minimum latency algorithm.
9998 @item -minline-float-divide-max-throughput
9999 @opindex minline-float-divide-max-throughput
10000 Generate code for inline divides of floating point values
10001 using the maximum throughput algorithm.
10003 @item -minline-int-divide-min-latency
10004 @opindex minline-int-divide-min-latency
10005 Generate code for inline divides of integer values
10006 using the minimum latency algorithm.
10008 @item -minline-int-divide-max-throughput
10009 @opindex minline-int-divide-max-throughput
10010 Generate code for inline divides of integer values
10011 using the maximum throughput algorithm.
10013 @item -mno-dwarf2-asm
10014 @itemx -mdwarf2-asm
10015 @opindex mno-dwarf2-asm
10016 @opindex mdwarf2-asm
10017 Don't (or do) generate assembler code for the DWARF2 line number debugging
10018 info. This may be useful when not using the GNU assembler.
10020 @item -mfixed-range=@var{register-range}
10021 @opindex mfixed-range
10022 Generate code treating the given register range as fixed registers.
10023 A fixed register is one that the register allocator can not use. This is
10024 useful when compiling kernel code. A register range is specified as
10025 two registers separated by a dash. Multiple register ranges can be
10026 specified separated by a comma.
10028 @item -mearly-stop-bits
10029 @itemx -mno-early-stop-bits
10030 @opindex mearly-stop-bits
10031 @opindex mno-early-stop-bits
10032 Allow stop bits to be placed earlier than immediately preceding the
10033 instruction that triggered the stop bit. This can improve instruction
10034 scheduling, but does not always do so.
10038 @subsection D30V Options
10039 @cindex D30V Options
10041 These @samp{-m} options are defined for D30V implementations:
10046 Link the @samp{.text}, @samp{.data}, @samp{.bss}, @samp{.strings},
10047 @samp{.rodata}, @samp{.rodata1}, @samp{.data1} sections into external
10048 memory, which starts at location @code{0x80000000}.
10051 @opindex mextmemory
10052 Same as the @option{-mextmem} switch.
10056 Link the @samp{.text} section into onchip text memory, which starts at
10057 location @code{0x0}. Also link @samp{.data}, @samp{.bss},
10058 @samp{.strings}, @samp{.rodata}, @samp{.rodata1}, @samp{.data1} sections
10059 into onchip data memory, which starts at location @code{0x20000000}.
10061 @item -mno-asm-optimize
10062 @itemx -masm-optimize
10063 @opindex mno-asm-optimize
10064 @opindex masm-optimize
10065 Disable (enable) passing @option{-O} to the assembler when optimizing.
10066 The assembler uses the @option{-O} option to automatically parallelize
10067 adjacent short instructions where possible.
10069 @item -mbranch-cost=@var{n}
10070 @opindex mbranch-cost
10071 Increase the internal costs of branches to @var{n}. Higher costs means
10072 that the compiler will issue more instructions to avoid doing a branch.
10075 @item -mcond-exec=@var{n}
10076 @opindex mcond-exec
10077 Specify the maximum number of conditionally executed instructions that
10078 replace a branch. The default is 4.
10081 @node S/390 and zSeries Options
10082 @subsection S/390 and zSeries Options
10083 @cindex S/390 and zSeries Options
10085 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
10089 @itemx -msoft-float
10090 @opindex mhard-float
10091 @opindex msoft-float
10092 Use (do not use) the hardware floating-point instructions and registers
10093 for floating-point operations. When @option{-msoft-float} is specified,
10094 functions in @file{libgcc.a} will be used to perform floating-point
10095 operations. When @option{-mhard-float} is specified, the compiler
10096 generates IEEE floating-point instructions. This is the default.
10099 @itemx -mno-backchain
10100 @opindex mbackchain
10101 @opindex mno-backchain
10102 Generate (or do not generate) code which maintains an explicit
10103 backchain within the stack frame that points to the caller's frame.
10104 This is currently needed to allow debugging. The default is to
10105 generate the backchain.
10108 @itemx -mno-small-exec
10109 @opindex msmall-exec
10110 @opindex mno-small-exec
10111 Generate (or do not generate) code using the @code{bras} instruction
10112 to do subroutine calls.
10113 This only works reliably if the total executable size does not
10114 exceed 64k. The default is to use the @code{basr} instruction instead,
10115 which does not have this limitation.
10121 When @option{-m31} is specified, generate code compliant to the
10122 Linux for S/390 ABI@. When @option{-m64} is specified, generate
10123 code compliant to the Linux for zSeries ABI@. This allows GCC in
10124 particular to generate 64-bit instructions. For the @samp{s390}
10125 targets, the default is @option{-m31}, while the @samp{s390x}
10126 targets default to @option{-m64}.
10132 When @option{-mzarch} is specified, generate code using the
10133 instructions available on z/Architecture.
10134 When @option{-mesa} is specified, generate code using the
10135 instructions available on ESA/390. Note that @option{-mesa} is
10136 not possible with @option{-m64}.
10137 When generating code compliant to the Linux for S/390 ABI,
10138 the default is @option{-mesa}. When generating code compliant
10139 to the Linux for zSeries ABI, the default is @option{-mzarch}.
10145 Generate (or do not generate) code using the @code{mvcle} instruction
10146 to perform block moves. When @option{-mno-mvcle} is specified,
10147 use a @code{mvc} loop instead. This is the default.
10153 Print (or do not print) additional debug information when compiling.
10154 The default is to not print debug information.
10156 @item -march=@var{cpu-type}
10158 Generate code that will run on @var{cpu-type}, which is the name of a system
10159 representing a certain processor type. Possible values for
10160 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, and @samp{z990}.
10161 When generating code using the instructions available on z/Architecture,
10162 the default is @option{-march=z900}. Otherwise, the default is
10163 @option{-march=g5}.
10165 @item -mtune=@var{cpu-type}
10167 Tune to @var{cpu-type} everything applicable about the generated code,
10168 except for the ABI and the set of available instructions.
10169 The list of @var{cpu-type} values is the same as for @option{-march}.
10170 The default is the value used for @option{-march}.
10175 @subsection CRIS Options
10176 @cindex CRIS Options
10178 These options are defined specifically for the CRIS ports.
10181 @item -march=@var{architecture-type}
10182 @itemx -mcpu=@var{architecture-type}
10185 Generate code for the specified architecture. The choices for
10186 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
10187 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX.
10188 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
10191 @item -mtune=@var{architecture-type}
10193 Tune to @var{architecture-type} everything applicable about the generated
10194 code, except for the ABI and the set of available instructions. The
10195 choices for @var{architecture-type} are the same as for
10196 @option{-march=@var{architecture-type}}.
10198 @item -mmax-stack-frame=@var{n}
10199 @opindex mmax-stack-frame
10200 Warn when the stack frame of a function exceeds @var{n} bytes.
10202 @item -melinux-stacksize=@var{n}
10203 @opindex melinux-stacksize
10204 Only available with the @samp{cris-axis-aout} target. Arranges for
10205 indications in the program to the kernel loader that the stack of the
10206 program should be set to @var{n} bytes.
10212 The options @option{-metrax4} and @option{-metrax100} are synonyms for
10213 @option{-march=v3} and @option{-march=v8} respectively.
10217 Enable CRIS-specific verbose debug-related information in the assembly
10218 code. This option also has the effect to turn off the @samp{#NO_APP}
10219 formatted-code indicator to the assembler at the beginning of the
10224 Do not use condition-code results from previous instruction; always emit
10225 compare and test instructions before use of condition codes.
10227 @item -mno-side-effects
10228 @opindex mno-side-effects
10229 Do not emit instructions with side-effects in addressing modes other than
10232 @item -mstack-align
10233 @itemx -mno-stack-align
10234 @itemx -mdata-align
10235 @itemx -mno-data-align
10236 @itemx -mconst-align
10237 @itemx -mno-const-align
10238 @opindex mstack-align
10239 @opindex mno-stack-align
10240 @opindex mdata-align
10241 @opindex mno-data-align
10242 @opindex mconst-align
10243 @opindex mno-const-align
10244 These options (no-options) arranges (eliminate arrangements) for the
10245 stack-frame, individual data and constants to be aligned for the maximum
10246 single data access size for the chosen CPU model. The default is to
10247 arrange for 32-bit alignment. ABI details such as structure layout are
10248 not affected by these options.
10256 Similar to the stack- data- and const-align options above, these options
10257 arrange for stack-frame, writable data and constants to all be 32-bit,
10258 16-bit or 8-bit aligned. The default is 32-bit alignment.
10260 @item -mno-prologue-epilogue
10261 @itemx -mprologue-epilogue
10262 @opindex mno-prologue-epilogue
10263 @opindex mprologue-epilogue
10264 With @option{-mno-prologue-epilogue}, the normal function prologue and
10265 epilogue that sets up the stack-frame are omitted and no return
10266 instructions or return sequences are generated in the code. Use this
10267 option only together with visual inspection of the compiled code: no
10268 warnings or errors are generated when call-saved registers must be saved,
10269 or storage for local variable needs to be allocated.
10273 @opindex mno-gotplt
10275 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
10276 instruction sequences that load addresses for functions from the PLT part
10277 of the GOT rather than (traditional on other architectures) calls to the
10278 PLT. The default is @option{-mgotplt}.
10282 Legacy no-op option only recognized with the cris-axis-aout target.
10286 Legacy no-op option only recognized with the cris-axis-elf and
10287 cris-axis-linux-gnu targets.
10291 Only recognized with the cris-axis-aout target, where it selects a
10292 GNU/linux-like multilib, include files and instruction set for
10293 @option{-march=v8}.
10297 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
10301 This option, recognized for the cris-axis-aout and cris-axis-elf arranges
10302 to link with input-output functions from a simulator library. Code,
10303 initialized data and zero-initialized data are allocated consecutively.
10307 Like @option{-sim}, but pass linker options to locate initialized data at
10308 0x40000000 and zero-initialized data at 0x80000000.
10312 @subsection MMIX Options
10313 @cindex MMIX Options
10315 These options are defined for the MMIX:
10319 @itemx -mno-libfuncs
10321 @opindex mno-libfuncs
10322 Specify that intrinsic library functions are being compiled, passing all
10323 values in registers, no matter the size.
10326 @itemx -mno-epsilon
10328 @opindex mno-epsilon
10329 Generate floating-point comparison instructions that compare with respect
10330 to the @code{rE} epsilon register.
10332 @item -mabi=mmixware
10334 @opindex mabi-mmixware
10336 Generate code that passes function parameters and return values that (in
10337 the called function) are seen as registers @code{$0} and up, as opposed to
10338 the GNU ABI which uses global registers @code{$231} and up.
10340 @item -mzero-extend
10341 @itemx -mno-zero-extend
10342 @opindex mzero-extend
10343 @opindex mno-zero-extend
10344 When reading data from memory in sizes shorter than 64 bits, use (do not
10345 use) zero-extending load instructions by default, rather than
10346 sign-extending ones.
10349 @itemx -mno-knuthdiv
10351 @opindex mno-knuthdiv
10352 Make the result of a division yielding a remainder have the same sign as
10353 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
10354 remainder follows the sign of the dividend. Both methods are
10355 arithmetically valid, the latter being almost exclusively used.
10357 @item -mtoplevel-symbols
10358 @itemx -mno-toplevel-symbols
10359 @opindex mtoplevel-symbols
10360 @opindex mno-toplevel-symbols
10361 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
10362 code can be used with the @code{PREFIX} assembly directive.
10366 Generate an executable in the ELF format, rather than the default
10367 @samp{mmo} format used by the @command{mmix} simulator.
10369 @item -mbranch-predict
10370 @itemx -mno-branch-predict
10371 @opindex mbranch-predict
10372 @opindex mno-branch-predict
10373 Use (do not use) the probable-branch instructions, when static branch
10374 prediction indicates a probable branch.
10376 @item -mbase-addresses
10377 @itemx -mno-base-addresses
10378 @opindex mbase-addresses
10379 @opindex mno-base-addresses
10380 Generate (do not generate) code that uses @emph{base addresses}. Using a
10381 base address automatically generates a request (handled by the assembler
10382 and the linker) for a constant to be set up in a global register. The
10383 register is used for one or more base address requests within the range 0
10384 to 255 from the value held in the register. The generally leads to short
10385 and fast code, but the number of different data items that can be
10386 addressed is limited. This means that a program that uses lots of static
10387 data may require @option{-mno-base-addresses}.
10389 @item -msingle-exit
10390 @itemx -mno-single-exit
10391 @opindex msingle-exit
10392 @opindex mno-single-exit
10393 Force (do not force) generated code to have a single exit point in each
10397 @node PDP-11 Options
10398 @subsection PDP-11 Options
10399 @cindex PDP-11 Options
10401 These options are defined for the PDP-11:
10406 Use hardware FPP floating point. This is the default. (FIS floating
10407 point on the PDP-11/40 is not supported.)
10410 @opindex msoft-float
10411 Do not use hardware floating point.
10415 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
10419 Return floating-point results in memory. This is the default.
10423 Generate code for a PDP-11/40.
10427 Generate code for a PDP-11/45. This is the default.
10431 Generate code for a PDP-11/10.
10433 @item -mbcopy-builtin
10434 @opindex bcopy-builtin
10435 Use inline @code{movstrhi} patterns for copying memory. This is the
10440 Do not use inline @code{movstrhi} patterns for copying memory.
10446 Use 16-bit @code{int}. This is the default.
10452 Use 32-bit @code{int}.
10455 @itemx -mno-float32
10457 @opindex mno-float32
10458 Use 64-bit @code{float}. This is the default.
10463 @opindex mno-float64
10464 Use 32-bit @code{float}.
10468 Use @code{abshi2} pattern. This is the default.
10472 Do not use @code{abshi2} pattern.
10474 @item -mbranch-expensive
10475 @opindex mbranch-expensive
10476 Pretend that branches are expensive. This is for experimenting with
10477 code generation only.
10479 @item -mbranch-cheap
10480 @opindex mbranch-cheap
10481 Do not pretend that branches are expensive. This is the default.
10485 Generate code for a system with split I&D.
10489 Generate code for a system without split I&D. This is the default.
10493 Use Unix assembler syntax. This is the default when configured for
10494 @samp{pdp11-*-bsd}.
10498 Use DEC assembler syntax. This is the default when configured for any
10499 PDP-11 target other than @samp{pdp11-*-bsd}.
10502 @node Xstormy16 Options
10503 @subsection Xstormy16 Options
10504 @cindex Xstormy16 Options
10506 These options are defined for Xstormy16:
10511 Choose startup files and linker script suitable for the simulator.
10515 @subsection FRV Options
10516 @cindex FRV Options
10522 Only use the first 32 general purpose registers.
10527 Use all 64 general purpose registers.
10532 Use only the first 32 floating point registers.
10537 Use all 64 floating point registers
10540 @opindex mhard-float
10542 Use hardware instructions for floating point operations.
10545 @opindex msoft-float
10547 Use library routines for floating point operations.
10552 Dynamically allocate condition code registers.
10557 Do not try to dynamically allocate condition code registers, only
10558 use @code{icc0} and @code{fcc0}.
10563 Change ABI to use double word insns.
10568 Do not use double word instructions.
10573 Use floating point double instructions.
10576 @opindex mno-double
10578 Do not use floating point double instructions.
10583 Use media instructions.
10588 Do not use media instructions.
10593 Use multiply and add/subtract instructions.
10596 @opindex mno-muladd
10598 Do not use multiply and add/subtract instructions.
10600 @item -mlibrary-pic
10601 @opindex mlibrary-pic
10603 Enable PIC support for building libraries
10608 Use only the first four media accumulator registers.
10613 Use all eight media accumulator registers.
10618 Pack VLIW instructions.
10623 Do not pack VLIW instructions.
10626 @opindex mno-eflags
10628 Do not mark ABI switches in e_flags.
10631 @opindex mcond-move
10633 Enable the use of conditional-move instructions (default).
10635 This switch is mainly for debugging the compiler and will likely be removed
10636 in a future version.
10638 @item -mno-cond-move
10639 @opindex mno-cond-move
10641 Disable the use of conditional-move instructions.
10643 This switch is mainly for debugging the compiler and will likely be removed
10644 in a future version.
10649 Enable the use of conditional set instructions (default).
10651 This switch is mainly for debugging the compiler and will likely be removed
10652 in a future version.
10657 Disable the use of conditional set instructions.
10659 This switch is mainly for debugging the compiler and will likely be removed
10660 in a future version.
10663 @opindex mcond-exec
10665 Enable the use of conditional execution (default).
10667 This switch is mainly for debugging the compiler and will likely be removed
10668 in a future version.
10670 @item -mno-cond-exec
10671 @opindex mno-cond-exec
10673 Disable the use of conditional execution.
10675 This switch is mainly for debugging the compiler and will likely be removed
10676 in a future version.
10678 @item -mvliw-branch
10679 @opindex mvliw-branch
10681 Run a pass to pack branches into VLIW instructions (default).
10683 This switch is mainly for debugging the compiler and will likely be removed
10684 in a future version.
10686 @item -mno-vliw-branch
10687 @opindex mno-vliw-branch
10689 Do not run a pass to pack branches into VLIW instructions.
10691 This switch is mainly for debugging the compiler and will likely be removed
10692 in a future version.
10694 @item -mmulti-cond-exec
10695 @opindex mmulti-cond-exec
10697 Enable optimization of @code{&&} and @code{||} in conditional execution
10700 This switch is mainly for debugging the compiler and will likely be removed
10701 in a future version.
10703 @item -mno-multi-cond-exec
10704 @opindex mno-multi-cond-exec
10706 Disable optimization of @code{&&} and @code{||} in conditional execution.
10708 This switch is mainly for debugging the compiler and will likely be removed
10709 in a future version.
10711 @item -mnested-cond-exec
10712 @opindex mnested-cond-exec
10714 Enable nested conditional execution optimizations (default).
10716 This switch is mainly for debugging the compiler and will likely be removed
10717 in a future version.
10719 @item -mno-nested-cond-exec
10720 @opindex mno-nested-cond-exec
10722 Disable nested conditional execution optimizations.
10724 This switch is mainly for debugging the compiler and will likely be removed
10725 in a future version.
10727 @item -mtomcat-stats
10728 @opindex mtomcat-stats
10730 Cause gas to print out tomcat statistics.
10732 @item -mcpu=@var{cpu}
10735 Select the processor type for which to generate code. Possible values are
10736 @samp{simple}, @samp{tomcat}, @samp{fr500}, @samp{fr400}, @samp{fr300},
10741 @node Xtensa Options
10742 @subsection Xtensa Options
10743 @cindex Xtensa Options
10745 The Xtensa architecture is designed to support many different
10746 configurations. The compiler's default options can be set to match a
10747 particular Xtensa configuration by copying a configuration file into the
10748 GCC sources when building GCC@. The options below may be used to
10749 override the default options.
10753 @itemx -mlittle-endian
10754 @opindex mbig-endian
10755 @opindex mlittle-endian
10756 Specify big-endian or little-endian byte ordering for the target Xtensa
10760 @itemx -mno-density
10762 @opindex mno-density
10763 Enable or disable use of the optional Xtensa code density instructions.
10766 @itemx -mno-const16
10768 @opindex mno-const16
10769 Enable or disable use of @code{CONST16} instructions for loading
10770 constant values. The @code{CONST16} instruction is currently not a
10771 standard option from Tensilica. When enabled, @code{CONST16}
10772 instructions are always used in place of the standard @code{L32R}
10773 instructions. The use of @code{CONST16} is enabled by default only if
10774 the @code{L32R} instruction is not available.
10780 Enable or disable use of the Xtensa @code{ABS} instruction for absolute
10787 Enable or disable use of the Xtensa @code{ADDX} and @code{SUBX}
10794 Enable or disable use of the Xtensa MAC16 option. When enabled, GCC
10795 will generate MAC16 instructions from standard C code, with the
10796 limitation that it will use neither the MR register file nor any
10797 instruction that operates on the MR registers. When this option is
10798 disabled, GCC will translate 16-bit multiply/accumulate operations to a
10799 combination of core instructions and library calls, depending on whether
10800 any other multiplier options are enabled.
10806 Enable or disable use of the 16-bit integer multiplier option. When
10807 enabled, the compiler will generate 16-bit multiply instructions for
10808 multiplications of 16 bits or smaller in standard C code. When this
10809 option is disabled, the compiler will either use 32-bit multiply or
10810 MAC16 instructions if they are available or generate library calls to
10811 perform the multiply operations using shifts and adds.
10817 Enable or disable use of the 32-bit integer multiplier option. When
10818 enabled, the compiler will generate 32-bit multiply instructions for
10819 multiplications of 32 bits or smaller in standard C code. When this
10820 option is disabled, the compiler will generate library calls to perform
10821 the multiply operations using either shifts and adds or 16-bit multiply
10822 instructions if they are available.
10828 Enable or disable use of the optional normalization shift amount
10829 (@code{NSA}) instructions to implement the built-in @code{ffs} function.
10834 @opindex mno-minmax
10835 Enable or disable use of the optional minimum and maximum value
10842 Enable or disable use of the optional sign extend (@code{SEXT})
10846 @itemx -mno-booleans
10848 @opindex mno-booleans
10849 Enable or disable support for the boolean register file used by Xtensa
10850 coprocessors. This is not typically useful by itself but may be
10851 required for other options that make use of the boolean registers (e.g.,
10852 the floating-point option).
10855 @itemx -msoft-float
10856 @opindex mhard-float
10857 @opindex msoft-float
10858 Enable or disable use of the floating-point option. When enabled, GCC
10859 generates floating-point instructions for 32-bit @code{float}
10860 operations. When this option is disabled, GCC generates library calls
10861 to emulate 32-bit floating-point operations using integer instructions.
10862 Regardless of this option, 64-bit @code{double} operations are always
10863 emulated with calls to library functions.
10866 @itemx -mno-fused-madd
10867 @opindex mfused-madd
10868 @opindex mno-fused-madd
10869 Enable or disable use of fused multiply/add and multiply/subtract
10870 instructions in the floating-point option. This has no effect if the
10871 floating-point option is not also enabled. Disabling fused multiply/add
10872 and multiply/subtract instructions forces the compiler to use separate
10873 instructions for the multiply and add/subtract operations. This may be
10874 desirable in some cases where strict IEEE 754-compliant results are
10875 required: the fused multiply add/subtract instructions do not round the
10876 intermediate result, thereby producing results with @emph{more} bits of
10877 precision than specified by the IEEE standard. Disabling fused multiply
10878 add/subtract instructions also ensures that the program output is not
10879 sensitive to the compiler's ability to combine multiply and add/subtract
10882 @item -mtext-section-literals
10883 @itemx -mno-text-section-literals
10884 @opindex mtext-section-literals
10885 @opindex mno-text-section-literals
10886 Control the treatment of literal pools. The default is
10887 @option{-mno-text-section-literals}, which places literals in a separate
10888 section in the output file. This allows the literal pool to be placed
10889 in a data RAM/ROM, and it also allows the linker to combine literal
10890 pools from separate object files to remove redundant literals and
10891 improve code size. With @option{-mtext-section-literals}, the literals
10892 are interspersed in the text section in order to keep them as close as
10893 possible to their references. This may be necessary for large assembly
10896 @item -mtarget-align
10897 @itemx -mno-target-align
10898 @opindex mtarget-align
10899 @opindex mno-target-align
10900 When this option is enabled, GCC instructs the assembler to
10901 automatically align instructions to reduce branch penalties at the
10902 expense of some code density. The assembler attempts to widen density
10903 instructions to align branch targets and the instructions following call
10904 instructions. If there are not enough preceding safe density
10905 instructions to align a target, no widening will be performed. The
10906 default is @option{-mtarget-align}. These options do not affect the
10907 treatment of auto-aligned instructions like @code{LOOP}, which the
10908 assembler will always align, either by widening density instructions or
10909 by inserting no-op instructions.
10912 @itemx -mno-longcalls
10913 @opindex mlongcalls
10914 @opindex mno-longcalls
10915 When this option is enabled, GCC instructs the assembler to translate
10916 direct calls to indirect calls unless it can determine that the target
10917 of a direct call is in the range allowed by the call instruction. This
10918 translation typically occurs for calls to functions in other source
10919 files. Specifically, the assembler translates a direct @code{CALL}
10920 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
10921 The default is @option{-mno-longcalls}. This option should be used in
10922 programs where the call target can potentially be out of range. This
10923 option is implemented in the assembler, not the compiler, so the
10924 assembly code generated by GCC will still show direct call
10925 instructions---look at the disassembled object code to see the actual
10926 instructions. Note that the assembler will use an indirect call for
10927 every cross-file call, not just those that really will be out of range.
10930 @node Code Gen Options
10931 @section Options for Code Generation Conventions
10932 @cindex code generation conventions
10933 @cindex options, code generation
10934 @cindex run-time options
10936 These machine-independent options control the interface conventions
10937 used in code generation.
10939 Most of them have both positive and negative forms; the negative form
10940 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
10941 one of the forms is listed---the one which is not the default. You
10942 can figure out the other form by either removing @samp{no-} or adding
10946 @item -fbounds-check
10947 @opindex fbounds-check
10948 For front-ends that support it, generate additional code to check that
10949 indices used to access arrays are within the declared range. This is
10950 currently only supported by the Java and Fortran 77 front-ends, where
10951 this option defaults to true and false respectively.
10955 This option generates traps for signed overflow on addition, subtraction,
10956 multiplication operations.
10960 This option instructs the compiler to assume that signed arithmetic
10961 overflow of addition, subtraction and multiplication wraps around
10962 using twos-complement representation. This flag enables some optimizations
10963 and disables other. This option is enabled by default for the Java
10964 front-end, as required by the Java language specification.
10967 @opindex fexceptions
10968 Enable exception handling. Generates extra code needed to propagate
10969 exceptions. For some targets, this implies GCC will generate frame
10970 unwind information for all functions, which can produce significant data
10971 size overhead, although it does not affect execution. If you do not
10972 specify this option, GCC will enable it by default for languages like
10973 C++ which normally require exception handling, and disable it for
10974 languages like C that do not normally require it. However, you may need
10975 to enable this option when compiling C code that needs to interoperate
10976 properly with exception handlers written in C++. You may also wish to
10977 disable this option if you are compiling older C++ programs that don't
10978 use exception handling.
10980 @item -fnon-call-exceptions
10981 @opindex fnon-call-exceptions
10982 Generate code that allows trapping instructions to throw exceptions.
10983 Note that this requires platform-specific runtime support that does
10984 not exist everywhere. Moreover, it only allows @emph{trapping}
10985 instructions to throw exceptions, i.e.@: memory references or floating
10986 point instructions. It does not allow exceptions to be thrown from
10987 arbitrary signal handlers such as @code{SIGALRM}.
10989 @item -funwind-tables
10990 @opindex funwind-tables
10991 Similar to @option{-fexceptions}, except that it will just generate any needed
10992 static data, but will not affect the generated code in any other way.
10993 You will normally not enable this option; instead, a language processor
10994 that needs this handling would enable it on your behalf.
10996 @item -fasynchronous-unwind-tables
10997 @opindex funwind-tables
10998 Generate unwind table in dwarf2 format, if supported by target machine. The
10999 table is exact at each instruction boundary, so it can be used for stack
11000 unwinding from asynchronous events (such as debugger or garbage collector).
11002 @item -fpcc-struct-return
11003 @opindex fpcc-struct-return
11004 Return ``short'' @code{struct} and @code{union} values in memory like
11005 longer ones, rather than in registers. This convention is less
11006 efficient, but it has the advantage of allowing intercallability between
11007 GCC-compiled files and files compiled with other compilers, particularly
11008 the Portable C Compiler (pcc).
11010 The precise convention for returning structures in memory depends
11011 on the target configuration macros.
11013 Short structures and unions are those whose size and alignment match
11014 that of some integer type.
11016 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
11017 switch is not binary compatible with code compiled with the
11018 @option{-freg-struct-return} switch.
11019 Use it to conform to a non-default application binary interface.
11021 @item -freg-struct-return
11022 @opindex freg-struct-return
11023 Return @code{struct} and @code{union} values in registers when possible.
11024 This is more efficient for small structures than
11025 @option{-fpcc-struct-return}.
11027 If you specify neither @option{-fpcc-struct-return} nor
11028 @option{-freg-struct-return}, GCC defaults to whichever convention is
11029 standard for the target. If there is no standard convention, GCC
11030 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
11031 the principal compiler. In those cases, we can choose the standard, and
11032 we chose the more efficient register return alternative.
11034 @strong{Warning:} code compiled with the @option{-freg-struct-return}
11035 switch is not binary compatible with code compiled with the
11036 @option{-fpcc-struct-return} switch.
11037 Use it to conform to a non-default application binary interface.
11039 @item -fshort-enums
11040 @opindex fshort-enums
11041 Allocate to an @code{enum} type only as many bytes as it needs for the
11042 declared range of possible values. Specifically, the @code{enum} type
11043 will be equivalent to the smallest integer type which has enough room.
11045 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
11046 code that is not binary compatible with code generated without that switch.
11047 Use it to conform to a non-default application binary interface.
11049 @item -fshort-double
11050 @opindex fshort-double
11051 Use the same size for @code{double} as for @code{float}.
11053 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
11054 code that is not binary compatible with code generated without that switch.
11055 Use it to conform to a non-default application binary interface.
11057 @item -fshort-wchar
11058 @opindex fshort-wchar
11059 Override the underlying type for @samp{wchar_t} to be @samp{short
11060 unsigned int} instead of the default for the target. This option is
11061 useful for building programs to run under WINE@.
11063 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
11064 code that is not binary compatible with code generated without that switch.
11065 Use it to conform to a non-default application binary interface.
11067 @item -fshared-data
11068 @opindex fshared-data
11069 Requests that the data and non-@code{const} variables of this
11070 compilation be shared data rather than private data. The distinction
11071 makes sense only on certain operating systems, where shared data is
11072 shared between processes running the same program, while private data
11073 exists in one copy per process.
11076 @opindex fno-common
11077 In C, allocate even uninitialized global variables in the data section of the
11078 object file, rather than generating them as common blocks. This has the
11079 effect that if the same variable is declared (without @code{extern}) in
11080 two different compilations, you will get an error when you link them.
11081 The only reason this might be useful is if you wish to verify that the
11082 program will work on other systems which always work this way.
11086 Ignore the @samp{#ident} directive.
11088 @item -fno-gnu-linker
11089 @opindex fno-gnu-linker
11090 Do not output global initializations (such as C++ constructors and
11091 destructors) in the form used by the GNU linker (on systems where the GNU
11092 linker is the standard method of handling them). Use this option when
11093 you want to use a non-GNU linker, which also requires using the
11094 @command{collect2} program to make sure the system linker includes
11095 constructors and destructors. (@command{collect2} is included in the GCC
11096 distribution.) For systems which @emph{must} use @command{collect2}, the
11097 compiler driver @command{gcc} is configured to do this automatically.
11099 @item -finhibit-size-directive
11100 @opindex finhibit-size-directive
11101 Don't output a @code{.size} assembler directive, or anything else that
11102 would cause trouble if the function is split in the middle, and the
11103 two halves are placed at locations far apart in memory. This option is
11104 used when compiling @file{crtstuff.c}; you should not need to use it
11107 @item -fverbose-asm
11108 @opindex fverbose-asm
11109 Put extra commentary information in the generated assembly code to
11110 make it more readable. This option is generally only of use to those
11111 who actually need to read the generated assembly code (perhaps while
11112 debugging the compiler itself).
11114 @option{-fno-verbose-asm}, the default, causes the
11115 extra information to be omitted and is useful when comparing two assembler
11120 @cindex global offset table
11122 Generate position-independent code (PIC) suitable for use in a shared
11123 library, if supported for the target machine. Such code accesses all
11124 constant addresses through a global offset table (GOT)@. The dynamic
11125 loader resolves the GOT entries when the program starts (the dynamic
11126 loader is not part of GCC; it is part of the operating system). If
11127 the GOT size for the linked executable exceeds a machine-specific
11128 maximum size, you get an error message from the linker indicating that
11129 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
11130 instead. (These maximums are 16k on the m88k, 8k on the SPARC, and 32k
11131 on the m68k and RS/6000. The 386 has no such limit.)
11133 Position-independent code requires special support, and therefore works
11134 only on certain machines. For the 386, GCC supports PIC for System V
11135 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
11136 position-independent.
11140 If supported for the target machine, emit position-independent code,
11141 suitable for dynamic linking and avoiding any limit on the size of the
11142 global offset table. This option makes a difference on the m68k, m88k,
11145 Position-independent code requires special support, and therefore works
11146 only on certain machines.
11152 These options are similar to @option{-fpic} and @option{-fPIC}, but
11153 generated position independent code can be only linked into executables.
11154 Usually these options are used when @option{-pie} GCC option will be
11155 used during linking.
11157 @item -ffixed-@var{reg}
11159 Treat the register named @var{reg} as a fixed register; generated code
11160 should never refer to it (except perhaps as a stack pointer, frame
11161 pointer or in some other fixed role).
11163 @var{reg} must be the name of a register. The register names accepted
11164 are machine-specific and are defined in the @code{REGISTER_NAMES}
11165 macro in the machine description macro file.
11167 This flag does not have a negative form, because it specifies a
11170 @item -fcall-used-@var{reg}
11171 @opindex fcall-used
11172 Treat the register named @var{reg} as an allocable register that is
11173 clobbered by function calls. It may be allocated for temporaries or
11174 variables that do not live across a call. Functions compiled this way
11175 will not save and restore the register @var{reg}.
11177 It is an error to used this flag with the frame pointer or stack pointer.
11178 Use of this flag for other registers that have fixed pervasive roles in
11179 the machine's execution model will produce disastrous results.
11181 This flag does not have a negative form, because it specifies a
11184 @item -fcall-saved-@var{reg}
11185 @opindex fcall-saved
11186 Treat the register named @var{reg} as an allocable register saved by
11187 functions. It may be allocated even for temporaries or variables that
11188 live across a call. Functions compiled this way will save and restore
11189 the register @var{reg} if they use it.
11191 It is an error to used this flag with the frame pointer or stack pointer.
11192 Use of this flag for other registers that have fixed pervasive roles in
11193 the machine's execution model will produce disastrous results.
11195 A different sort of disaster will result from the use of this flag for
11196 a register in which function values may be returned.
11198 This flag does not have a negative form, because it specifies a
11201 @item -fpack-struct
11202 @opindex fpack-struct
11203 Pack all structure members together without holes.
11205 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
11206 code that is not binary compatible with code generated without that switch.
11207 Additionally, it makes the code suboptimal.
11208 Use it to conform to a non-default application binary interface.
11210 @item -finstrument-functions
11211 @opindex finstrument-functions
11212 Generate instrumentation calls for entry and exit to functions. Just
11213 after function entry and just before function exit, the following
11214 profiling functions will be called with the address of the current
11215 function and its call site. (On some platforms,
11216 @code{__builtin_return_address} does not work beyond the current
11217 function, so the call site information may not be available to the
11218 profiling functions otherwise.)
11221 void __cyg_profile_func_enter (void *this_fn,
11223 void __cyg_profile_func_exit (void *this_fn,
11227 The first argument is the address of the start of the current function,
11228 which may be looked up exactly in the symbol table.
11230 This instrumentation is also done for functions expanded inline in other
11231 functions. The profiling calls will indicate where, conceptually, the
11232 inline function is entered and exited. This means that addressable
11233 versions of such functions must be available. If all your uses of a
11234 function are expanded inline, this may mean an additional expansion of
11235 code size. If you use @samp{extern inline} in your C code, an
11236 addressable version of such functions must be provided. (This is
11237 normally the case anyways, but if you get lucky and the optimizer always
11238 expands the functions inline, you might have gotten away without
11239 providing static copies.)
11241 A function may be given the attribute @code{no_instrument_function}, in
11242 which case this instrumentation will not be done. This can be used, for
11243 example, for the profiling functions listed above, high-priority
11244 interrupt routines, and any functions from which the profiling functions
11245 cannot safely be called (perhaps signal handlers, if the profiling
11246 routines generate output or allocate memory).
11248 @item -fstack-check
11249 @opindex fstack-check
11250 Generate code to verify that you do not go beyond the boundary of the
11251 stack. You should specify this flag if you are running in an
11252 environment with multiple threads, but only rarely need to specify it in
11253 a single-threaded environment since stack overflow is automatically
11254 detected on nearly all systems if there is only one stack.
11256 Note that this switch does not actually cause checking to be done; the
11257 operating system must do that. The switch causes generation of code
11258 to ensure that the operating system sees the stack being extended.
11260 @item -fstack-limit-register=@var{reg}
11261 @itemx -fstack-limit-symbol=@var{sym}
11262 @itemx -fno-stack-limit
11263 @opindex fstack-limit-register
11264 @opindex fstack-limit-symbol
11265 @opindex fno-stack-limit
11266 Generate code to ensure that the stack does not grow beyond a certain value,
11267 either the value of a register or the address of a symbol. If the stack
11268 would grow beyond the value, a signal is raised. For most targets,
11269 the signal is raised before the stack overruns the boundary, so
11270 it is possible to catch the signal without taking special precautions.
11272 For instance, if the stack starts at absolute address @samp{0x80000000}
11273 and grows downwards, you can use the flags
11274 @option{-fstack-limit-symbol=__stack_limit} and
11275 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
11276 of 128KB@. Note that this may only work with the GNU linker.
11278 @cindex aliasing of parameters
11279 @cindex parameters, aliased
11280 @item -fargument-alias
11281 @itemx -fargument-noalias
11282 @itemx -fargument-noalias-global
11283 @opindex fargument-alias
11284 @opindex fargument-noalias
11285 @opindex fargument-noalias-global
11286 Specify the possible relationships among parameters and between
11287 parameters and global data.
11289 @option{-fargument-alias} specifies that arguments (parameters) may
11290 alias each other and may alias global storage.@*
11291 @option{-fargument-noalias} specifies that arguments do not alias
11292 each other, but may alias global storage.@*
11293 @option{-fargument-noalias-global} specifies that arguments do not
11294 alias each other and do not alias global storage.
11296 Each language will automatically use whatever option is required by
11297 the language standard. You should not need to use these options yourself.
11299 @item -fleading-underscore
11300 @opindex fleading-underscore
11301 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
11302 change the way C symbols are represented in the object file. One use
11303 is to help link with legacy assembly code.
11305 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
11306 generate code that is not binary compatible with code generated without that
11307 switch. Use it to conform to a non-default application binary interface.
11308 Not all targets provide complete support for this switch.
11310 @item -ftls-model=@var{model}
11311 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
11312 The @var{model} argument should be one of @code{global-dynamic},
11313 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
11315 The default without @option{-fpic} is @code{initial-exec}; with
11316 @option{-fpic} the default is @code{global-dynamic}.
11321 @node Environment Variables
11322 @section Environment Variables Affecting GCC
11323 @cindex environment variables
11325 @c man begin ENVIRONMENT
11326 This section describes several environment variables that affect how GCC
11327 operates. Some of them work by specifying directories or prefixes to use
11328 when searching for various kinds of files. Some are used to specify other
11329 aspects of the compilation environment.
11331 Note that you can also specify places to search using options such as
11332 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
11333 take precedence over places specified using environment variables, which
11334 in turn take precedence over those specified by the configuration of GCC@.
11335 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
11336 GNU Compiler Collection (GCC) Internals}.
11341 @c @itemx LC_COLLATE
11343 @c @itemx LC_MONETARY
11344 @c @itemx LC_NUMERIC
11349 @c @findex LC_COLLATE
11350 @findex LC_MESSAGES
11351 @c @findex LC_MONETARY
11352 @c @findex LC_NUMERIC
11356 These environment variables control the way that GCC uses
11357 localization information that allow GCC to work with different
11358 national conventions. GCC inspects the locale categories
11359 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
11360 so. These locale categories can be set to any value supported by your
11361 installation. A typical value is @samp{en_UK} for English in the United
11364 The @env{LC_CTYPE} environment variable specifies character
11365 classification. GCC uses it to determine the character boundaries in
11366 a string; this is needed for some multibyte encodings that contain quote
11367 and escape characters that would otherwise be interpreted as a string
11370 The @env{LC_MESSAGES} environment variable specifies the language to
11371 use in diagnostic messages.
11373 If the @env{LC_ALL} environment variable is set, it overrides the value
11374 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
11375 and @env{LC_MESSAGES} default to the value of the @env{LANG}
11376 environment variable. If none of these variables are set, GCC
11377 defaults to traditional C English behavior.
11381 If @env{TMPDIR} is set, it specifies the directory to use for temporary
11382 files. GCC uses temporary files to hold the output of one stage of
11383 compilation which is to be used as input to the next stage: for example,
11384 the output of the preprocessor, which is the input to the compiler
11387 @item GCC_EXEC_PREFIX
11388 @findex GCC_EXEC_PREFIX
11389 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
11390 names of the subprograms executed by the compiler. No slash is added
11391 when this prefix is combined with the name of a subprogram, but you can
11392 specify a prefix that ends with a slash if you wish.
11394 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
11395 an appropriate prefix to use based on the pathname it was invoked with.
11397 If GCC cannot find the subprogram using the specified prefix, it
11398 tries looking in the usual places for the subprogram.
11400 The default value of @env{GCC_EXEC_PREFIX} is
11401 @file{@var{prefix}/lib/gcc-lib/} where @var{prefix} is the value
11402 of @code{prefix} when you ran the @file{configure} script.
11404 Other prefixes specified with @option{-B} take precedence over this prefix.
11406 This prefix is also used for finding files such as @file{crt0.o} that are
11409 In addition, the prefix is used in an unusual way in finding the
11410 directories to search for header files. For each of the standard
11411 directories whose name normally begins with @samp{/usr/local/lib/gcc-lib}
11412 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
11413 replacing that beginning with the specified prefix to produce an
11414 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
11415 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
11416 These alternate directories are searched first; the standard directories
11419 @item COMPILER_PATH
11420 @findex COMPILER_PATH
11421 The value of @env{COMPILER_PATH} is a colon-separated list of
11422 directories, much like @env{PATH}. GCC tries the directories thus
11423 specified when searching for subprograms, if it can't find the
11424 subprograms using @env{GCC_EXEC_PREFIX}.
11427 @findex LIBRARY_PATH
11428 The value of @env{LIBRARY_PATH} is a colon-separated list of
11429 directories, much like @env{PATH}. When configured as a native compiler,
11430 GCC tries the directories thus specified when searching for special
11431 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
11432 using GCC also uses these directories when searching for ordinary
11433 libraries for the @option{-l} option (but directories specified with
11434 @option{-L} come first).
11438 @cindex locale definition
11439 This variable is used to pass locale information to the compiler. One way in
11440 which this information is used is to determine the character set to be used
11441 when character literals, string literals and comments are parsed in C and C++.
11442 When the compiler is configured to allow multibyte characters,
11443 the following values for @env{LANG} are recognized:
11447 Recognize JIS characters.
11449 Recognize SJIS characters.
11451 Recognize EUCJP characters.
11454 If @env{LANG} is not defined, or if it has some other value, then the
11455 compiler will use mblen and mbtowc as defined by the default locale to
11456 recognize and translate multibyte characters.
11460 Some additional environments variables affect the behavior of the
11463 @include cppenv.texi
11467 @node Precompiled Headers
11468 @section Using Precompiled Headers
11469 @cindex precompiled headers
11470 @cindex speed of compilation
11472 Often large projects have many header files that are included in every
11473 source file. The time the compiler takes to process these header files
11474 over and over again can account for nearly all of the time required to
11475 build the project. To make builds faster, GCC allows users to
11476 `precompile' a header file; then, if builds can use the precompiled
11477 header file they will be much faster.
11479 To create a precompiled header file, simply compile it as you would any
11480 other file, if necessary using the @option{-x} option to make the driver
11481 treat it as a C or C++ header file. You will probably want to use a
11482 tool like @command{make} to keep the precompiled header up-to-date when
11483 the headers it contains change.
11485 A precompiled header file will be searched for when @code{#include} is
11486 seen in the compilation. As it searches for the included file
11487 (@pxref{Search Path,,Search Path,cpp.info,The C Preprocessor}) the
11488 compiler looks for a precompiled header in each directory just before it
11489 looks for the include file in that directory. The name searched for is
11490 the name specified in the @code{#include} with @samp{.gch} appended. If
11491 the precompiled header file can't be used, it is ignored.
11493 For instance, if you have @code{#include "all.h"}, and you have
11494 @file{all.h.gch} in the same directory as @file{all.h}, then the
11495 precompiled header file will be used if possible, and the original
11496 header will be used otherwise.
11498 Alternatively, you might decide to put the precompiled header file in a
11499 directory and use @option{-I} to ensure that directory is searched
11500 before (or instead of) the directory containing the original header.
11501 Then, if you want to check that the precompiled header file is always
11502 used, you can put a file of the same name as the original header in this
11503 directory containing an @code{#error} command.
11505 This also works with @option{-include}. So yet another way to use
11506 precompiled headers, good for projects not designed with precompiled
11507 header files in mind, is to simply take most of the header files used by
11508 a project, include them from another header file, precompile that header
11509 file, and @option{-include} the precompiled header. If the header files
11510 have guards against multiple inclusion, they will be skipped because
11511 they've already been included (in the precompiled header).
11513 If you need to precompile the same header file for different
11514 languages, targets, or compiler options, you can instead make a
11515 @emph{directory} named like @file{all.h.gch}, and put each precompiled
11516 header in the directory. (It doesn't matter what you call the files
11517 in the directory, every precompiled header in the directory will be
11518 considered.) The first precompiled header encountered in the
11519 directory that is valid for this compilation will be used; they're
11520 searched in no particular order.
11522 There are many other possibilities, limited only by your imagination,
11523 good sense, and the constraints of your build system.
11525 A precompiled header file can be used only when these conditions apply:
11529 Only one precompiled header can be used in a particular compilation.
11531 A precompiled header can't be used once the first C token is seen. You
11532 can have preprocessor directives before a precompiled header; you can
11533 even include a precompiled header from inside another header, so long as
11534 there are no C tokens before the @code{#include}.
11536 The precompiled header file must be produced for the same language as
11537 the current compilation. You can't use a C precompiled header for a C++
11540 The precompiled header file must be produced by the same compiler
11541 version and configuration as the current compilation is using.
11542 The easiest way to guarantee this is to use the same compiler binary
11543 for creating and using precompiled headers.
11545 Any macros defined before the precompiled header (including with
11546 @option{-D}) must either be defined in the same way as when the
11547 precompiled header was generated, or must not affect the precompiled
11548 header, which usually means that the they don't appear in the
11549 precompiled header at all.
11551 Certain command-line options must be defined in the same way as when the
11552 precompiled header was generated. At present, it's not clear which
11553 options are safe to change and which are not; the safest choice is to
11554 use exactly the same options when generating and using the precompiled
11558 For all of these but the last, the compiler will automatically ignore
11559 the precompiled header if the conditions aren't met. For the last item,
11560 some option changes will cause the precompiled header to be rejected,
11561 but not all incompatible option combinations have yet been found. If
11562 you find a new incompatible combination, please consider filing a bug
11563 report, see @ref{Bugs}.
11565 @node Running Protoize
11566 @section Running Protoize
11568 The program @code{protoize} is an optional part of GCC@. You can use
11569 it to add prototypes to a program, thus converting the program to ISO
11570 C in one respect. The companion program @code{unprotoize} does the
11571 reverse: it removes argument types from any prototypes that are found.
11573 When you run these programs, you must specify a set of source files as
11574 command line arguments. The conversion programs start out by compiling
11575 these files to see what functions they define. The information gathered
11576 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
11578 After scanning comes actual conversion. The specified files are all
11579 eligible to be converted; any files they include (whether sources or
11580 just headers) are eligible as well.
11582 But not all the eligible files are converted. By default,
11583 @code{protoize} and @code{unprotoize} convert only source and header
11584 files in the current directory. You can specify additional directories
11585 whose files should be converted with the @option{-d @var{directory}}
11586 option. You can also specify particular files to exclude with the
11587 @option{-x @var{file}} option. A file is converted if it is eligible, its
11588 directory name matches one of the specified directory names, and its
11589 name within the directory has not been excluded.
11591 Basic conversion with @code{protoize} consists of rewriting most
11592 function definitions and function declarations to specify the types of
11593 the arguments. The only ones not rewritten are those for varargs
11596 @code{protoize} optionally inserts prototype declarations at the
11597 beginning of the source file, to make them available for any calls that
11598 precede the function's definition. Or it can insert prototype
11599 declarations with block scope in the blocks where undeclared functions
11602 Basic conversion with @code{unprotoize} consists of rewriting most
11603 function declarations to remove any argument types, and rewriting
11604 function definitions to the old-style pre-ISO form.
11606 Both conversion programs print a warning for any function declaration or
11607 definition that they can't convert. You can suppress these warnings
11610 The output from @code{protoize} or @code{unprotoize} replaces the
11611 original source file. The original file is renamed to a name ending
11612 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
11613 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
11614 for DOS) file already exists, then the source file is simply discarded.
11616 @code{protoize} and @code{unprotoize} both depend on GCC itself to
11617 scan the program and collect information about the functions it uses.
11618 So neither of these programs will work until GCC is installed.
11620 Here is a table of the options you can use with @code{protoize} and
11621 @code{unprotoize}. Each option works with both programs unless
11625 @item -B @var{directory}
11626 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
11627 usual directory (normally @file{/usr/local/lib}). This file contains
11628 prototype information about standard system functions. This option
11629 applies only to @code{protoize}.
11631 @item -c @var{compilation-options}
11632 Use @var{compilation-options} as the options when running @command{gcc} to
11633 produce the @samp{.X} files. The special option @option{-aux-info} is
11634 always passed in addition, to tell @command{gcc} to write a @samp{.X} file.
11636 Note that the compilation options must be given as a single argument to
11637 @code{protoize} or @code{unprotoize}. If you want to specify several
11638 @command{gcc} options, you must quote the entire set of compilation options
11639 to make them a single word in the shell.
11641 There are certain @command{gcc} arguments that you cannot use, because they
11642 would produce the wrong kind of output. These include @option{-g},
11643 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
11644 the @var{compilation-options}, they are ignored.
11647 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
11648 systems) instead of @samp{.c}. This is convenient if you are converting
11649 a C program to C++. This option applies only to @code{protoize}.
11652 Add explicit global declarations. This means inserting explicit
11653 declarations at the beginning of each source file for each function
11654 that is called in the file and was not declared. These declarations
11655 precede the first function definition that contains a call to an
11656 undeclared function. This option applies only to @code{protoize}.
11658 @item -i @var{string}
11659 Indent old-style parameter declarations with the string @var{string}.
11660 This option applies only to @code{protoize}.
11662 @code{unprotoize} converts prototyped function definitions to old-style
11663 function definitions, where the arguments are declared between the
11664 argument list and the initial @samp{@{}. By default, @code{unprotoize}
11665 uses five spaces as the indentation. If you want to indent with just
11666 one space instead, use @option{-i " "}.
11669 Keep the @samp{.X} files. Normally, they are deleted after conversion
11673 Add explicit local declarations. @code{protoize} with @option{-l} inserts
11674 a prototype declaration for each function in each block which calls the
11675 function without any declaration. This option applies only to
11679 Make no real changes. This mode just prints information about the conversions
11680 that would have been done without @option{-n}.
11683 Make no @samp{.save} files. The original files are simply deleted.
11684 Use this option with caution.
11686 @item -p @var{program}
11687 Use the program @var{program} as the compiler. Normally, the name
11688 @file{gcc} is used.
11691 Work quietly. Most warnings are suppressed.
11694 Print the version number, just like @option{-v} for @command{gcc}.
11697 If you need special compiler options to compile one of your program's
11698 source files, then you should generate that file's @samp{.X} file
11699 specially, by running @command{gcc} on that source file with the
11700 appropriate options and the option @option{-aux-info}. Then run
11701 @code{protoize} on the entire set of files. @code{protoize} will use
11702 the existing @samp{.X} file because it is newer than the source file.
11706 gcc -Dfoo=bar file1.c -aux-info file1.X
11711 You need to include the special files along with the rest in the
11712 @code{protoize} command, even though their @samp{.X} files already
11713 exist, because otherwise they won't get converted.
11715 @xref{Protoize Caveats}, for more information on how to use
11716 @code{protoize} successfully.