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 -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}.
197 -fconstant-string-class=@var{class-name} @gol
198 -fgnu-runtime -fnext-runtime @gol
199 -fno-nil-receivers @gol
200 -fobjc-exceptions @gol
201 -freplace-objc-classes @gol
204 -Wno-protocol -Wselector -Wundeclared-selector}
206 @item Language Independent Options
207 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
208 @gccoptlist{-fmessage-length=@var{n} @gol
209 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]}}
211 @item Warning Options
212 @xref{Warning Options,,Options to Request or Suppress Warnings}.
213 @gccoptlist{-fsyntax-only -pedantic -pedantic-errors @gol
214 -w -Wextra -Wall -Waggregate-return @gol
215 -Wcast-align -Wcast-qual -Wchar-subscripts -Wcomment @gol
216 -Wconversion -Wno-deprecated-declarations @gol
217 -Wdisabled-optimization -Wno-div-by-zero -Wendif-labels @gol
218 -Werror -Werror-implicit-function-declaration @gol
219 -Wfloat-equal -Wformat -Wformat=2 @gol
220 -Wno-format-extra-args -Wformat-nonliteral @gol
221 -Wformat-security -Wno-format-y2k @gol
222 -Wimplicit -Wimplicit-function-declaration -Wimplicit-int @gol
223 -Wimport -Wno-import -Winit-self -Winline @gol
224 -Wno-invalid-offsetof -Winvalid-pch @gol
225 -Wlarger-than-@var{len} -Wlong-long @gol
226 -Wmain -Wmissing-braces @gol
227 -Wmissing-format-attribute -Wmissing-noreturn @gol
228 -Wno-multichar -Wnonnull -Wpacked -Wpadded @gol
229 -Wparentheses -Wpointer-arith -Wredundant-decls @gol
230 -Wreturn-type -Wsequence-point -Wshadow @gol
231 -Wsign-compare -Wstrict-aliasing @gol
232 -Wswitch -Wswitch-default -Wswitch-enum @gol
233 -Wsystem-headers -Wtrigraphs -Wundef -Wuninitialized @gol
234 -Wunknown-pragmas -Wunreachable-code @gol
235 -Wunused -Wunused-function -Wunused-label -Wunused-parameter @gol
236 -Wunused-value -Wunused-variable -Wwrite-strings}
238 @item C-only Warning Options
239 @gccoptlist{-Wbad-function-cast -Wmissing-declarations @gol
240 -Wmissing-prototypes -Wnested-externs -Wold-style-definition @gol
241 -Wstrict-prototypes -Wtraditional @gol
242 -Wdeclaration-after-statement}
244 @item Debugging Options
245 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
246 @gccoptlist{-d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
247 -fdump-unnumbered -fdump-translation-unit@r{[}-@var{n}@r{]} @gol
248 -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
249 -fdump-tree-original@r{[}-@var{n}@r{]} @gol
250 -fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
251 -fdump-tree-inlined@r{[}-@var{n}@r{]} @gol
252 -feliminate-dwarf2-dups -feliminate-unused-debug-types @gol
253 -feliminate-unused-debug-symbols -fmem-report -fprofile-arcs @gol
254 -frandom-seed=@var{string} -fsched-verbose=@var{n} @gol
255 -ftest-coverage -ftime-report @gol
256 -g -g@var{level} -gcoff -gdwarf -gdwarf-1 -gdwarf-1+ -gdwarf-2 @gol
257 -ggdb -gstabs -gstabs+ -gvms -gxcoff -gxcoff+ @gol
258 -p -pg -print-file-name=@var{library} -print-libgcc-file-name @gol
259 -print-multi-directory -print-multi-lib @gol
260 -print-prog-name=@var{program} -print-search-dirs -Q @gol
263 @item Optimization Options
264 @xref{Optimize Options,,Options that Control Optimization}.
265 @gccoptlist{-falign-functions=@var{n} -falign-jumps=@var{n} @gol
266 -falign-labels=@var{n} -falign-loops=@var{n} @gol
267 -fbranch-probabilities -fprofile-values -fbranch-target-load-optimize @gol
268 -fbranch-target-load-optimize2 -fcaller-saves -fcprop-registers @gol
269 -fcse-follow-jumps -fcse-skip-blocks -fdata-sections @gol
270 -fdelayed-branch -fdelete-null-pointer-checks @gol
271 -fexpensive-optimizations -ffast-math -ffloat-store @gol
272 -fforce-addr -fforce-mem -ffunction-sections @gol
273 -fgcse -fgcse-lm -fgcse-sm -fgcse-las -floop-optimize @gol
274 -fcrossjumping -fif-conversion -fif-conversion2 @gol
275 -finline-functions -finline-limit=@var{n} -fkeep-inline-functions @gol
276 -fkeep-static-consts -fmerge-constants -fmerge-all-constants @gol
277 -fmove-all-movables -fnew-ra -fno-branch-count-reg @gol
278 -fno-default-inline -fno-defer-pop @gol
279 -fno-function-cse -fno-guess-branch-probability @gol
280 -fno-inline -fno-math-errno -fno-peephole -fno-peephole2 @gol
281 -funsafe-math-optimizations -ffinite-math-only @gol
282 -fno-trapping-math -fno-zero-initialized-in-bss @gol
283 -fomit-frame-pointer -foptimize-register-move @gol
284 -foptimize-sibling-calls -fprefetch-loop-arrays @gol
285 -freduce-all-givs -fregmove -frename-registers @gol
286 -freorder-blocks -freorder-functions @gol
287 -frerun-cse-after-loop -frerun-loop-opt @gol
288 -frounding-math -fschedule-insns -fschedule-insns2 @gol
289 -fno-sched-interblock -fno-sched-spec -fsched-spec-load @gol
290 -fsched-spec-load-dangerous @gol
291 -fsched-stalled-insns=@var{n} -sched-stalled-insns-dep=@var{n} @gol
292 -fsched2-use-superblocks @gol
293 -fsched2-use-traces -fsignaling-nans @gol
294 -fsingle-precision-constant -fssa -fssa-ccp -fssa-dce @gol
295 -fstrength-reduce -fstrict-aliasing -ftracer -fthread-jumps @gol
296 -funroll-all-loops -funroll-loops -fpeel-loops @gol
297 -funswitch-loops -fold-unroll-loops -fold-unroll-all-loops @gol
298 --param @var{name}=@var{value}
299 -O -O0 -O1 -O2 -O3 -Os}
301 @item Preprocessor Options
302 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
303 @gccoptlist{-A@var{question}=@var{answer} @gol
304 -A-@var{question}@r{[}=@var{answer}@r{]} @gol
305 -C -dD -dI -dM -dN @gol
306 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
307 -idirafter @var{dir} @gol
308 -include @var{file} -imacros @var{file} @gol
309 -iprefix @var{file} -iwithprefix @var{dir} @gol
310 -iwithprefixbefore @var{dir} -isystem @var{dir} @gol
311 -M -MM -MF -MG -MP -MQ -MT -nostdinc @gol
312 -P -fworking-directory -remap @gol
313 -trigraphs -undef -U@var{macro} -Wp,@var{option} @gol
314 -Xpreprocessor @var{option}}
316 @item Assembler Option
317 @xref{Assembler Options,,Passing Options to the Assembler}.
318 @gccoptlist{-Wa,@var{option} -Xassembler @var{option}}
321 @xref{Link Options,,Options for Linking}.
322 @gccoptlist{@var{object-file-name} -l@var{library} @gol
323 -nostartfiles -nodefaultlibs -nostdlib -pie @gol
324 -s -static -static-libgcc -shared -shared-libgcc -symbolic @gol
325 -Wl,@var{option} -Xlinker @var{option} @gol
328 @item Directory Options
329 @xref{Directory Options,,Options for Directory Search}.
330 @gccoptlist{-B@var{prefix} -I@var{dir} -I- -L@var{dir} -specs=@var{file}}
333 @c I wrote this xref this way to avoid overfull hbox. -- rms
334 @xref{Target Options}.
335 @gccoptlist{-V @var{version} -b @var{machine}}
337 @item Machine Dependent Options
338 @xref{Submodel Options,,Hardware Models and Configurations}.
340 @emph{M680x0 Options}
341 @gccoptlist{-m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
342 -m68060 -mcpu32 -m5200 -m68881 -mbitfield -mc68000 -mc68020 @gol
343 -mnobitfield -mrtd -mshort -msoft-float -mpcrel @gol
344 -malign-int -mstrict-align}
346 @emph{M68hc1x Options}
347 @gccoptlist{-m6811 -m6812 -m68hc11 -m68hc12 -m68hcs12 @gol
348 -mauto-incdec -minmax -mlong-calls -mshort @gol
349 -msoft-reg-count=@var{count}}
352 @gccoptlist{-mg -mgnu -munix}
355 @gccoptlist{-mcpu=@var{cpu-type} @gol
356 -mtune=@var{cpu-type} @gol
357 -mcmodel=@var{code-model} @gol
359 -mapp-regs -mbroken-saverestore -mcypress @gol
360 -mfaster-structs -mflat @gol
361 -mfpu -mhard-float -mhard-quad-float @gol
362 -mimpure-text -mlittle-endian -mlive-g0 -mno-app-regs @gol
363 -mno-faster-structs -mno-flat -mno-fpu @gol
364 -mno-impure-text -mno-stack-bias -mno-unaligned-doubles @gol
365 -msoft-float -msoft-quad-float -msparclite -mstack-bias @gol
366 -msupersparc -munaligned-doubles -mv8}
369 @gccoptlist{-mapcs-frame -mno-apcs-frame @gol
370 -mapcs-26 -mapcs-32 @gol
371 -mapcs-stack-check -mno-apcs-stack-check @gol
372 -mapcs-float -mno-apcs-float @gol
373 -mapcs-reentrant -mno-apcs-reentrant @gol
374 -msched-prolog -mno-sched-prolog @gol
375 -mlittle-endian -mbig-endian -mwords-little-endian @gol
376 -malignment-traps -mno-alignment-traps @gol
377 -msoft-float -mhard-float -mfpe @gol
378 -mthumb-interwork -mno-thumb-interwork @gol
379 -mcpu=@var{name} -march=@var{name} -mfpe=@var{name} @gol
380 -mstructure-size-boundary=@var{n} @gol
381 -mabort-on-noreturn @gol
382 -mlong-calls -mno-long-calls @gol
383 -msingle-pic-base -mno-single-pic-base @gol
384 -mpic-register=@var{reg} @gol
385 -mnop-fun-dllimport @gol
386 -mcirrus-fix-invalid-insns -mno-cirrus-fix-invalid-insns @gol
387 -mpoke-function-name @gol
389 -mtpcs-frame -mtpcs-leaf-frame @gol
390 -mcaller-super-interworking -mcallee-super-interworking}
392 @emph{MN10200 Options}
395 @emph{MN10300 Options}
396 @gccoptlist{-mmult-bug -mno-mult-bug @gol
397 -mam33 -mno-am33 @gol
398 -mam33-2 -mno-am33-2 @gol
401 @emph{M32R/D Options}
402 @gccoptlist{-m32rx -m32r -mcode-model=@var{model-type} @gol
403 -msdata=@var{sdata-type} -G @var{num}}
406 @gccoptlist{-m88000 -m88100 -m88110 -mbig-pic @gol
407 -mcheck-zero-division -mhandle-large-shift @gol
408 -midentify-revision -mno-check-zero-division @gol
409 -mno-ocs-debug-info -mno-ocs-frame-position @gol
410 -mno-optimize-arg-area -mno-serialize-volatile @gol
411 -mno-underscores -mocs-debug-info @gol
412 -mocs-frame-position -moptimize-arg-area @gol
413 -mserialize-volatile -mshort-data-@var{num} -msvr3 @gol
414 -msvr4 -mtrap-large-shift -muse-div-instruction @gol
415 -mversion-03.00 -mwarn-passed-structs}
417 @emph{RS/6000 and PowerPC Options}
418 @gccoptlist{-mcpu=@var{cpu-type} @gol
419 -mtune=@var{cpu-type} @gol
420 -mpower -mno-power -mpower2 -mno-power2 @gol
421 -mpowerpc -mpowerpc64 -mno-powerpc @gol
422 -maltivec -mno-altivec @gol
423 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
424 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
425 -mnew-mnemonics -mold-mnemonics @gol
426 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
427 -m64 -m32 -mxl-call -mno-xl-call -mpe @gol
428 -malign-power -malign-natural @gol
429 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
430 -mstring -mno-string -mupdate -mno-update @gol
431 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
432 -mstrict-align -mno-strict-align -mrelocatable @gol
433 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
434 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
435 -mdynamic-no-pic @gol
436 -mprioritize-restricted-insns=@var{priority} @gol
437 -msched-costly-dep=@var{dependence_type} @gol
438 -mcall-sysv -mcall-netbsd @gol
439 -maix-struct-return -msvr4-struct-return @gol
440 -mabi=altivec -mabi=no-altivec @gol
441 -mabi=spe -mabi=no-spe @gol
442 -misel=yes -misel=no @gol
443 -mspe=yes -mspe=no @gol
444 -mfloat-gprs=yes -mfloat-gprs=no @gol
445 -mprototype -mno-prototype @gol
446 -msim -mmvme -mads -myellowknife -memb -msdata @gol
447 -msdata=@var{opt} -mvxworks -mwindiss -G @var{num} -pthread}
449 @emph{Darwin Options}
450 @gccoptlist{-all_load -allowable_client -arch -arch_errors_fatal @gol
451 -arch_only -bind_at_load -bundle -bundle_loader @gol
452 -client_name -compatibility_version -current_version @gol
453 -dependency-file -dylib_file -dylinker_install_name @gol
454 -dynamic -dynamiclib -exported_symbols_list @gol
455 -filelist -flat_namespace -force_cpusubtype_ALL @gol
456 -force_flat_namespace -headerpad_max_install_names @gol
457 -image_base -init -install_name -keep_private_externs @gol
458 -multi_module -multiply_defined -multiply_defined_unused @gol
459 -noall_load -nofixprebinding -nomultidefs -noprebind -noseglinkedit @gol
460 -pagezero_size -prebind -prebind_all_twolevel_modules @gol
461 -private_bundle -read_only_relocs -sectalign @gol
462 -sectobjectsymbols -whyload -seg1addr @gol
463 -sectcreate -sectobjectsymbols -sectorder @gol
464 -seg_addr_table -seg_addr_table_filename -seglinkedit @gol
465 -segprot -segs_read_only_addr -segs_read_write_addr @gol
466 -single_module -static -sub_library -sub_umbrella @gol
467 -twolevel_namespace -umbrella -undefined @gol
468 -unexported_symbols_list -weak_reference_mismatches @gol
472 @gccoptlist{-mcall-lib-mul -mfp-arg-in-fpregs -mfp-arg-in-gregs @gol
473 -mfull-fp-blocks -mhc-struct-return -min-line-mul @gol
474 -mminimum-fp-blocks -mnohc-struct-return}
477 @gccoptlist{-mabicalls -march=@var{cpu-type} -mtune=@var{cpu=type} @gol
478 -mcpu=@var{cpu-type} -membedded-data -muninit-const-in-rodata @gol
479 -membedded-pic -mfp32 -mfp64 -mfused-madd -mno-fused-madd @gol
480 -mgas -mgp32 -mgp64 -mhard-float -mint64 -mips1 @gol
481 -mips2 -mips3 -mips4 -mips32 -mips32r2 -mips64 @gol
482 -mlong64 -mlong32 -mlong-calls -mmemcpy @gol
483 -mmips-as -mmips-tfile -mno-abicalls -mxgot @gol
484 -mno-embedded-data -mno-uninit-const-in-rodata @gol
485 -mno-embedded-pic -mno-long-calls @gol
486 -mno-memcpy -mno-mips-tfile -mno-rnames @gol
487 -mrnames -msoft-float @gol
488 -m4650 -msingle-float -mmad @gol
489 -EL -EB -G @var{num} -nocpp @gol
490 -mabi=32 -mabi=n32 -mabi=64 -mabi=eabi -mabi-fake-default @gol
491 -mfix7000 -mfix-sb1 -mno-fix-sb1 @gol
492 -mno-crt0 -mflush-func=@var{func} -mno-flush-func @gol
493 -mbranch-likely -mno-branch-likely}
495 @emph{i386 and x86-64 Options}
496 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
497 -mfpmath=@var{unit} @gol
498 -masm=@var{dialect} -mno-fancy-math-387 @gol
499 -mno-fp-ret-in-387 -msoft-float -msvr3-shlib @gol
500 -mno-wide-multiply -mrtd -malign-double @gol
501 -mpreferred-stack-boundary=@var{num} @gol
502 -mmmx -msse -msse2 -mpni -m3dnow @gol
503 -mthreads -mno-align-stringops -minline-all-stringops @gol
504 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
505 -m96bit-long-double -mregparm=@var{num} -momit-leaf-frame-pointer @gol
506 -mno-red-zone -mno-tls-direct-seg-refs @gol
507 -mcmodel=@var{code-model} @gol
511 @gccoptlist{-march=@var{architecture-type} @gol
512 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
513 -mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
514 -mjump-in-delay -mlinker-opt -mlong-calls @gol
515 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
516 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
517 -mno-jump-in-delay -mno-long-load-store @gol
518 -mno-portable-runtime -mno-soft-float @gol
519 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
520 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
521 -mschedule=@var{cpu-type} -mspace-regs -msio -mwsio @gol
522 -nolibdld -static -threads}
524 @emph{Intel 960 Options}
525 @gccoptlist{-m@var{cpu-type} -masm-compat -mclean-linkage @gol
526 -mcode-align -mcomplex-addr -mleaf-procedures @gol
527 -mic-compat -mic2.0-compat -mic3.0-compat @gol
528 -mintel-asm -mno-clean-linkage -mno-code-align @gol
529 -mno-complex-addr -mno-leaf-procedures @gol
530 -mno-old-align -mno-strict-align -mno-tail-call @gol
531 -mnumerics -mold-align -msoft-float -mstrict-align @gol
534 @emph{DEC Alpha Options}
535 @gccoptlist{-mno-fp-regs -msoft-float -malpha-as -mgas @gol
536 -mieee -mieee-with-inexact -mieee-conformant @gol
537 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
538 -mtrap-precision=@var{mode} -mbuild-constants @gol
539 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} @gol
540 -mbwx -mmax -mfix -mcix @gol
541 -mfloat-vax -mfloat-ieee @gol
542 -mexplicit-relocs -msmall-data -mlarge-data @gol
543 -msmall-text -mlarge-text @gol
544 -mmemory-latency=@var{time}}
546 @emph{DEC Alpha/VMS Options}
547 @gccoptlist{-mvms-return-codes}
549 @emph{H8/300 Options}
550 @gccoptlist{-mrelax -mh -ms -mn -mint32 -malign-300}
553 @gccoptlist{-m1 -m2 -m2e -m3 -m3e @gol
554 -m4-nofpu -m4-single-only -m4-single -m4 @gol
555 -m5-64media -m5-64media-nofpu @gol
556 -m5-32media -m5-32media-nofpu @gol
557 -m5-compact -m5-compact-nofpu @gol
558 -mb -ml -mdalign -mrelax @gol
559 -mbigtable -mfmovd -mhitachi -mnomacsave @gol
560 -mieee -misize -mpadstruct -mspace @gol
561 -mprefergot -musermode}
563 @emph{System V Options}
564 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
567 @gccoptlist{-EB -EL @gol
568 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
569 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
571 @emph{TMS320C3x/C4x Options}
572 @gccoptlist{-mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm @gol
573 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload @gol
574 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned @gol
575 -mparallel-insns -mparallel-mpy -mpreserve-float}
578 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
579 -mprolog-function -mno-prolog-function -mspace @gol
580 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
581 -mapp-regs -mno-app-regs @gol
582 -mdisable-callt -mno-disable-callt @gol
588 @gccoptlist{-m32032 -m32332 -m32532 -m32081 -m32381 @gol
589 -mmult-add -mnomult-add -msoft-float -mrtd -mnortd @gol
590 -mregparam -mnoregparam -msb -mnosb @gol
591 -mbitfield -mnobitfield -mhimem -mnohimem}
594 @gccoptlist{-mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts @gol
595 -mcall-prologues -mno-tablejump -mtiny-stack}
598 @gccoptlist{-mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
599 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
600 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
601 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
602 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
605 @gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
606 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
607 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
608 -mno-base-addresses -msingle-exit -mno-single-exit}
611 @gccoptlist{-mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
612 -mvolatile-asm-stop -mb-step -mregister-names -mno-sdata @gol
613 -mconstant-gp -mauto-pic -minline-float-divide-min-latency @gol
614 -minline-float-divide-max-throughput @gol
615 -minline-int-divide-min-latency @gol
616 -minline-int-divide-max-throughput -mno-dwarf2-asm @gol
617 -mfixed-range=@var{register-range}}
620 @gccoptlist{-mextmem -mextmemory -monchip -mno-asm-optimize @gol
621 -masm-optimize -mbranch-cost=@var{n} -mcond-exec=@var{n}}
623 @emph{S/390 and zSeries Options}
624 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
625 -mhard-float -msoft-float -mbackchain -mno-backchain @gol
626 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
627 -m64 -m31 -mdebug -mno-debug -mesa -mzarch -mfused-madd -mno-fused-madd}
630 @gccoptlist{-mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
631 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
632 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
633 -mstack-align -mdata-align -mconst-align @gol
634 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
635 -melf -maout -melinux -mlinux -sim -sim2}
637 @emph{PDP-11 Options}
638 @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
639 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
640 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
641 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
642 -mbranch-expensive -mbranch-cheap @gol
643 -msplit -mno-split -munix-asm -mdec-asm}
645 @emph{Xstormy16 Options}
648 @emph{Xtensa Options}
649 @gccoptlist{-mconst16 -mno-const16 @gol
650 -mfused-madd -mno-fused-madd @gol
651 -mtext-section-literals -mno-text-section-literals @gol
652 -mtarget-align -mno-target-align @gol
653 -mlongcalls -mno-longcalls}
656 @gccoptlist{-mgpr-32 -mgpr-64 -mfpr-32 -mfpr-64 @gol
657 -mhard-float -msoft-float @gol
658 -malloc-cc -mfixed-cc -mdword -mno-dword @gol
659 -mdouble -mno-double @gol
660 -mmedia -mno-media -mmuladd -mno-muladd @gol
661 -mlibrary-pic -macc-4 -macc-8 @gol
662 -mpack -mno-pack -mno-eflags -mcond-move -mno-cond-move @gol
663 -mscc -mno-scc -mcond-exec -mno-cond-exec @gol
664 -mvliw-branch -mno-vliw-branch @gol
665 -mmulti-cond-exec -mno-multi-cond-exec -mnested-cond-exec @gol
666 -mno-nested-cond-exec -mtomcat-stats @gol
669 @item Code Generation Options
670 @xref{Code Gen Options,,Options for Code Generation Conventions}.
671 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
672 -ffixed-@var{reg} -fexceptions @gol
673 -fnon-call-exceptions -funwind-tables @gol
674 -fasynchronous-unwind-tables @gol
675 -finhibit-size-directive -finstrument-functions @gol
676 -fno-common -fno-ident -fno-gnu-linker @gol
677 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
678 -freg-struct-return -fshared-data -fshort-enums @gol
679 -fshort-double -fshort-wchar @gol
680 -fverbose-asm -fpack-struct -fstack-check @gol
681 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
682 -fargument-alias -fargument-noalias @gol
683 -fargument-noalias-global -fleading-underscore @gol
684 -ftls-model=@var{model} @gol
685 -ftrapv -fwrapv -fbounds-check}
689 * Overall Options:: Controlling the kind of output:
690 an executable, object files, assembler files,
691 or preprocessed source.
692 * C Dialect Options:: Controlling the variant of C language compiled.
693 * C++ Dialect Options:: Variations on C++.
694 * Objective-C Dialect Options:: Variations on Objective-C.
695 * Language Independent Options:: Controlling how diagnostics should be
697 * Warning Options:: How picky should the compiler be?
698 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
699 * Optimize Options:: How much optimization?
700 * Preprocessor Options:: Controlling header files and macro definitions.
701 Also, getting dependency information for Make.
702 * Assembler Options:: Passing options to the assembler.
703 * Link Options:: Specifying libraries and so on.
704 * Directory Options:: Where to find header files and libraries.
705 Where to find the compiler executable files.
706 * Spec Files:: How to pass switches to sub-processes.
707 * Target Options:: Running a cross-compiler, or an old version of GCC.
710 @node Overall Options
711 @section Options Controlling the Kind of Output
713 Compilation can involve up to four stages: preprocessing, compilation
714 proper, assembly and linking, always in that order. GCC is capable of
715 preprocessing and compiling several files either into several
716 assembler input files, or into one assembler input file; then each
717 assembler input file produces an object file, and linking combines all
718 the object files (those newly compiled, and those specified as input)
719 into an executable file.
721 @cindex file name suffix
722 For any given input file, the file name suffix determines what kind of
727 C source code which must be preprocessed.
730 C source code which should not be preprocessed.
733 C++ source code which should not be preprocessed.
736 Objective-C source code. Note that you must link with the library
737 @file{libobjc.a} to make an Objective-C program work.
740 Objective-C source code which should not be preprocessed.
743 C or C++ header file to be turned into a precompiled header.
747 @itemx @var{file}.cxx
748 @itemx @var{file}.cpp
749 @itemx @var{file}.CPP
750 @itemx @var{file}.c++
752 C++ source code which must be preprocessed. Note that in @samp{.cxx},
753 the last two letters must both be literally @samp{x}. Likewise,
754 @samp{.C} refers to a literal capital C@.
758 C++ header file to be turned into a precompiled header.
761 @itemx @var{file}.for
762 @itemx @var{file}.FOR
763 Fortran source code which should not be preprocessed.
766 @itemx @var{file}.fpp
767 @itemx @var{file}.FPP
768 Fortran source code which must be preprocessed (with the traditional
772 Fortran source code which must be preprocessed with a RATFOR
773 preprocessor (not included with GCC)@.
775 @xref{Overall Options,,Options Controlling the Kind of Output, g77,
776 Using and Porting GNU Fortran}, for more details of the handling of
779 @c FIXME: Descriptions of Java file types.
786 Ada source code file which contains a library unit declaration (a
787 declaration of a package, subprogram, or generic, or a generic
788 instantiation), or a library unit renaming declaration (a package,
789 generic, or subprogram renaming declaration). Such files are also
792 @itemx @var{file}.adb
793 Ada source code file containing a library unit body (a subprogram or
794 package body). Such files are also called @dfn{bodies}.
796 @c GCC also knows about some suffixes for languages not yet included:
805 Assembler code which must be preprocessed.
808 An object file to be fed straight into linking.
809 Any file name with no recognized suffix is treated this way.
813 You can specify the input language explicitly with the @option{-x} option:
816 @item -x @var{language}
817 Specify explicitly the @var{language} for the following input files
818 (rather than letting the compiler choose a default based on the file
819 name suffix). This option applies to all following input files until
820 the next @option{-x} option. Possible values for @var{language} are:
822 c c-header cpp-output
823 c++ c++-header c++-cpp-output
824 objective-c objective-c-header objc-cpp-output
825 assembler assembler-with-cpp
827 f77 f77-cpp-input ratfor
833 Turn off any specification of a language, so that subsequent files are
834 handled according to their file name suffixes (as they are if @option{-x}
835 has not been used at all).
837 @item -pass-exit-codes
838 @opindex pass-exit-codes
839 Normally the @command{gcc} program will exit with the code of 1 if any
840 phase of the compiler returns a non-success return code. If you specify
841 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
842 numerically highest error produced by any phase that returned an error
846 If you only want some of the stages of compilation, you can use
847 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
848 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
849 @command{gcc} is to stop. Note that some combinations (for example,
850 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
855 Compile or assemble the source files, but do not link. The linking
856 stage simply is not done. The ultimate output is in the form of an
857 object file for each source file.
859 By default, the object file name for a source file is made by replacing
860 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
862 Unrecognized input files, not requiring compilation or assembly, are
867 Stop after the stage of compilation proper; do not assemble. The output
868 is in the form of an assembler code file for each non-assembler input
871 By default, the assembler file name for a source file is made by
872 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
874 Input files that don't require compilation are ignored.
878 Stop after the preprocessing stage; do not run the compiler proper. The
879 output is in the form of preprocessed source code, which is sent to the
882 Input files which don't require preprocessing are ignored.
884 @cindex output file option
887 Place output in file @var{file}. This applies regardless to whatever
888 sort of output is being produced, whether it be an executable file,
889 an object file, an assembler file or preprocessed C code.
891 If you specify @option{-o} when compiling more than one input file, or
892 you are producing an executable file as output, all the source files
893 on the command line will be compiled at once.
895 If @option{-o} is not specified, the default is to put an executable file
896 in @file{a.out}, the object file for @file{@var{source}.@var{suffix}} in
897 @file{@var{source}.o}, its assembler file in @file{@var{source}.s}, and
898 all preprocessed C source on standard output.
902 Print (on standard error output) the commands executed to run the stages
903 of compilation. Also print the version number of the compiler driver
904 program and of the preprocessor and the compiler proper.
908 Like @option{-v} except the commands are not executed and all command
909 arguments are quoted. This is useful for shell scripts to capture the
910 driver-generated command lines.
914 Use pipes rather than temporary files for communication between the
915 various stages of compilation. This fails to work on some systems where
916 the assembler is unable to read from a pipe; but the GNU assembler has
921 Print (on the standard output) a description of the command line options
922 understood by @command{gcc}. If the @option{-v} option is also specified
923 then @option{--help} will also be passed on to the various processes
924 invoked by @command{gcc}, so that they can display the command line options
925 they accept. If the @option{-Wextra} option is also specified then command
926 line options which have no documentation associated with them will also
931 Print (on the standard output) a description of target specific command
932 line options for each tool.
936 Display the version number and copyrights of the invoked GCC.
940 @section Compiling C++ Programs
942 @cindex suffixes for C++ source
943 @cindex C++ source file suffixes
944 C++ source files conventionally use one of the suffixes @samp{.C},
945 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
946 @samp{.cxx}; C++ header files often use @samp{.hh} or @samp{.H}; and
947 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
948 files with these names and compiles them as C++ programs even if you
949 call the compiler the same way as for compiling C programs (usually
950 with the name @command{gcc}).
954 However, C++ programs often require class libraries as well as a
955 compiler that understands the C++ language---and under some
956 circumstances, you might want to compile programs or header files from
957 standard input, or otherwise without a suffix that flags them as C++
958 programs. You might also like to precompile a C header file with a
959 @samp{.h} extension to be used in C++ compilations. @command{g++} is a
960 program that calls GCC with the default language set to C++, and
961 automatically specifies linking against the C++ library. On many
962 systems, @command{g++} is also installed with the name @command{c++}.
964 @cindex invoking @command{g++}
965 When you compile C++ programs, you may specify many of the same
966 command-line options that you use for compiling programs in any
967 language; or command-line options meaningful for C and related
968 languages; or options that are meaningful only for C++ programs.
969 @xref{C Dialect Options,,Options Controlling C Dialect}, for
970 explanations of options for languages related to C@.
971 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
972 explanations of options that are meaningful only for C++ programs.
974 @node C Dialect Options
975 @section Options Controlling C Dialect
976 @cindex dialect options
977 @cindex language dialect options
978 @cindex options, dialect
980 The following options control the dialect of C (or languages derived
981 from C, such as C++ and Objective-C) that the compiler accepts:
988 In C mode, support all ISO C90 programs. In C++ mode,
989 remove GNU extensions that conflict with ISO C++.
991 This turns off certain features of GCC that are incompatible with ISO
992 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
993 such as the @code{asm} and @code{typeof} keywords, and
994 predefined macros such as @code{unix} and @code{vax} that identify the
995 type of system you are using. It also enables the undesirable and
996 rarely used ISO trigraph feature. For the C compiler,
997 it disables recognition of C++ style @samp{//} comments as well as
998 the @code{inline} keyword.
1000 The alternate keywords @code{__asm__}, @code{__extension__},
1001 @code{__inline__} and @code{__typeof__} continue to work despite
1002 @option{-ansi}. You would not want to use them in an ISO C program, of
1003 course, but it is useful to put them in header files that might be included
1004 in compilations done with @option{-ansi}. Alternate predefined macros
1005 such as @code{__unix__} and @code{__vax__} are also available, with or
1006 without @option{-ansi}.
1008 The @option{-ansi} option does not cause non-ISO programs to be
1009 rejected gratuitously. For that, @option{-pedantic} is required in
1010 addition to @option{-ansi}. @xref{Warning Options}.
1012 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1013 option is used. Some header files may notice this macro and refrain
1014 from declaring certain functions or defining certain macros that the
1015 ISO standard doesn't call for; this is to avoid interfering with any
1016 programs that might use these names for other things.
1018 Functions which would normally be built in but do not have semantics
1019 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1020 functions with @option{-ansi} is used. @xref{Other Builtins,,Other
1021 built-in functions provided by GCC}, for details of the functions
1026 Determine the language standard. This option is currently only
1027 supported when compiling C or C++. A value for this option must be
1028 provided; possible values are
1033 ISO C90 (same as @option{-ansi}).
1035 @item iso9899:199409
1036 ISO C90 as modified in amendment 1.
1042 ISO C99. Note that this standard is not yet fully supported; see
1043 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1044 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1047 Default, ISO C90 plus GNU extensions (including some C99 features).
1051 ISO C99 plus GNU extensions. When ISO C99 is fully implemented in GCC,
1052 this will become the default. The name @samp{gnu9x} is deprecated.
1055 The 1998 ISO C++ standard plus amendments.
1058 The same as @option{-std=c++98} plus GNU extensions. This is the
1059 default for C++ code.
1062 Even when this option is not specified, you can still use some of the
1063 features of newer standards in so far as they do not conflict with
1064 previous C standards. For example, you may use @code{__restrict__} even
1065 when @option{-std=c99} is not specified.
1067 The @option{-std} options specifying some version of ISO C have the same
1068 effects as @option{-ansi}, except that features that were not in ISO C90
1069 but are in the specified version (for example, @samp{//} comments and
1070 the @code{inline} keyword in ISO C99) are not disabled.
1072 @xref{Standards,,Language Standards Supported by GCC}, for details of
1073 these standard versions.
1075 @item -aux-info @var{filename}
1077 Output to the given filename prototyped declarations for all functions
1078 declared and/or defined in a translation unit, including those in header
1079 files. This option is silently ignored in any language other than C@.
1081 Besides declarations, the file indicates, in comments, the origin of
1082 each declaration (source file and line), whether the declaration was
1083 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1084 @samp{O} for old, respectively, in the first character after the line
1085 number and the colon), and whether it came from a declaration or a
1086 definition (@samp{C} or @samp{F}, respectively, in the following
1087 character). In the case of function definitions, a K&R-style list of
1088 arguments followed by their declarations is also provided, inside
1089 comments, after the declaration.
1093 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1094 keyword, so that code can use these words as identifiers. You can use
1095 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1096 instead. @option{-ansi} implies @option{-fno-asm}.
1098 In C++, this switch only affects the @code{typeof} keyword, since
1099 @code{asm} and @code{inline} are standard keywords. You may want to
1100 use the @option{-fno-gnu-keywords} flag instead, which has the same
1101 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1102 switch only affects the @code{asm} and @code{typeof} keywords, since
1103 @code{inline} is a standard keyword in ISO C99.
1106 @itemx -fno-builtin-@var{function}
1107 @opindex fno-builtin
1108 @cindex built-in functions
1109 Don't recognize built-in functions that do not begin with
1110 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1111 functions provided by GCC}, for details of the functions affected,
1112 including those which are not built-in functions when @option{-ansi} or
1113 @option{-std} options for strict ISO C conformance are used because they
1114 do not have an ISO standard meaning.
1116 GCC normally generates special code to handle certain built-in functions
1117 more efficiently; for instance, calls to @code{alloca} may become single
1118 instructions that adjust the stack directly, and calls to @code{memcpy}
1119 may become inline copy loops. The resulting code is often both smaller
1120 and faster, but since the function calls no longer appear as such, you
1121 cannot set a breakpoint on those calls, nor can you change the behavior
1122 of the functions by linking with a different library.
1124 With the @option{-fno-builtin-@var{function}} option
1125 only the built-in function @var{function} is
1126 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1127 function is named this is not built-in in this version of GCC, this
1128 option is ignored. There is no corresponding
1129 @option{-fbuiltin-@var{function}} option; if you wish to enable
1130 built-in functions selectively when using @option{-fno-builtin} or
1131 @option{-ffreestanding}, you may define macros such as:
1134 #define abs(n) __builtin_abs ((n))
1135 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1140 @cindex hosted environment
1142 Assert that compilation takes place in a hosted environment. This implies
1143 @option{-fbuiltin}. A hosted environment is one in which the
1144 entire standard library is available, and in which @code{main} has a return
1145 type of @code{int}. Examples are nearly everything except a kernel.
1146 This is equivalent to @option{-fno-freestanding}.
1148 @item -ffreestanding
1149 @opindex ffreestanding
1150 @cindex hosted environment
1152 Assert that compilation takes place in a freestanding environment. This
1153 implies @option{-fno-builtin}. A freestanding environment
1154 is one in which the standard library may not exist, and program startup may
1155 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1156 This is equivalent to @option{-fno-hosted}.
1158 @xref{Standards,,Language Standards Supported by GCC}, for details of
1159 freestanding and hosted environments.
1161 @item -fms-extensions
1162 @opindex fms-extensions
1163 Accept some non-standard constructs used in Microsoft header files.
1167 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1168 options for strict ISO C conformance) implies @option{-trigraphs}.
1170 @item -no-integrated-cpp
1171 @opindex no-integrated-cpp
1172 Performs a compilation in two passes: preprocessing and compiling. This
1173 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1174 @option{-B} option. The user supplied compilation step can then add in
1175 an additional preprocessing step after normal preprocessing but before
1176 compiling. The default is to use the integrated cpp (internal cpp)
1178 The semantics of this option will change if "cc1", "cc1plus", and
1179 "cc1obj" are merged.
1181 @cindex traditional C language
1182 @cindex C language, traditional
1184 @itemx -traditional-cpp
1185 @opindex traditional-cpp
1186 @opindex traditional
1187 Formerly, these options caused GCC to attempt to emulate a pre-standard
1188 C compiler. They are now only supported with the @option{-E} switch.
1189 The preprocessor continues to support a pre-standard mode. See the GNU
1190 CPP manual for details.
1192 @item -fcond-mismatch
1193 @opindex fcond-mismatch
1194 Allow conditional expressions with mismatched types in the second and
1195 third arguments. The value of such an expression is void. This option
1196 is not supported for C++.
1198 @item -funsigned-char
1199 @opindex funsigned-char
1200 Let the type @code{char} be unsigned, like @code{unsigned char}.
1202 Each kind of machine has a default for what @code{char} should
1203 be. It is either like @code{unsigned char} by default or like
1204 @code{signed char} by default.
1206 Ideally, a portable program should always use @code{signed char} or
1207 @code{unsigned char} when it depends on the signedness of an object.
1208 But many programs have been written to use plain @code{char} and
1209 expect it to be signed, or expect it to be unsigned, depending on the
1210 machines they were written for. This option, and its inverse, let you
1211 make such a program work with the opposite default.
1213 The type @code{char} is always a distinct type from each of
1214 @code{signed char} or @code{unsigned char}, even though its behavior
1215 is always just like one of those two.
1218 @opindex fsigned-char
1219 Let the type @code{char} be signed, like @code{signed char}.
1221 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1222 the negative form of @option{-funsigned-char}. Likewise, the option
1223 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1225 @item -fsigned-bitfields
1226 @itemx -funsigned-bitfields
1227 @itemx -fno-signed-bitfields
1228 @itemx -fno-unsigned-bitfields
1229 @opindex fsigned-bitfields
1230 @opindex funsigned-bitfields
1231 @opindex fno-signed-bitfields
1232 @opindex fno-unsigned-bitfields
1233 These options control whether a bit-field is signed or unsigned, when the
1234 declaration does not use either @code{signed} or @code{unsigned}. By
1235 default, such a bit-field is signed, because this is consistent: the
1236 basic integer types such as @code{int} are signed types.
1238 @item -fwritable-strings
1239 @opindex fwritable-strings
1240 Store string constants in the writable data segment and don't uniquize
1241 them. This is for compatibility with old programs which assume they can
1242 write into string constants.
1244 Writing into string constants is a very bad idea; ``constants'' should
1248 @node C++ Dialect Options
1249 @section Options Controlling C++ Dialect
1251 @cindex compiler options, C++
1252 @cindex C++ options, command line
1253 @cindex options, C++
1254 This section describes the command-line options that are only meaningful
1255 for C++ programs; but you can also use most of the GNU compiler options
1256 regardless of what language your program is in. For example, you
1257 might compile a file @code{firstClass.C} like this:
1260 g++ -g -frepo -O -c firstClass.C
1264 In this example, only @option{-frepo} is an option meant
1265 only for C++ programs; you can use the other options with any
1266 language supported by GCC@.
1268 Here is a list of options that are @emph{only} for compiling C++ programs:
1272 @item -fabi-version=@var{n}
1273 @opindex fabi-version
1274 Use version @var{n} of the C++ ABI. Version 1 is the version of the C++
1275 ABI that first appeared in G++ 3.2. Version 0 will always be the
1276 version that conforms most closely to the C++ ABI specification.
1277 Therefore, the ABI obtained using version 0 will change as ABI bugs are
1280 The default is version 1.
1282 @item -fno-access-control
1283 @opindex fno-access-control
1284 Turn off all access checking. This switch is mainly useful for working
1285 around bugs in the access control code.
1289 Check that the pointer returned by @code{operator new} is non-null
1290 before attempting to modify the storage allocated. This check is
1291 normally unnecessary because the C++ standard specifies that
1292 @code{operator new} will only return @code{0} if it is declared
1293 @samp{throw()}, in which case the compiler will always check the
1294 return value even without this option. In all other cases, when
1295 @code{operator new} has a non-empty exception specification, memory
1296 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1297 @samp{new (nothrow)}.
1299 @item -fconserve-space
1300 @opindex fconserve-space
1301 Put uninitialized or runtime-initialized global variables into the
1302 common segment, as C does. This saves space in the executable at the
1303 cost of not diagnosing duplicate definitions. If you compile with this
1304 flag and your program mysteriously crashes after @code{main()} has
1305 completed, you may have an object that is being destroyed twice because
1306 two definitions were merged.
1308 This option is no longer useful on most targets, now that support has
1309 been added for putting variables into BSS without making them common.
1311 @item -fno-const-strings
1312 @opindex fno-const-strings
1313 Give string constants type @code{char *} instead of type @code{const
1314 char *}. By default, G++ uses type @code{const char *} as required by
1315 the standard. Even if you use @option{-fno-const-strings}, you cannot
1316 actually modify the value of a string constant, unless you also use
1317 @option{-fwritable-strings}.
1319 This option might be removed in a future release of G++. For maximum
1320 portability, you should structure your code so that it works with
1321 string constants that have type @code{const char *}.
1323 @item -fno-elide-constructors
1324 @opindex fno-elide-constructors
1325 The C++ standard allows an implementation to omit creating a temporary
1326 which is only used to initialize another object of the same type.
1327 Specifying this option disables that optimization, and forces G++ to
1328 call the copy constructor in all cases.
1330 @item -fno-enforce-eh-specs
1331 @opindex fno-enforce-eh-specs
1332 Don't check for violation of exception specifications at runtime. This
1333 option violates the C++ standard, but may be useful for reducing code
1334 size in production builds, much like defining @samp{NDEBUG}. The compiler
1335 will still optimize based on the exception specifications.
1337 @item -fexternal-templates
1338 @opindex fexternal-templates
1340 Cause @samp{#pragma interface} and @samp{implementation} to apply to
1341 template instantiation; template instances are emitted or not according
1342 to the location of the template definition. @xref{Template
1343 Instantiation}, for more information.
1345 This option is deprecated.
1347 @item -falt-external-templates
1348 @opindex falt-external-templates
1349 Similar to @option{-fexternal-templates}, but template instances are
1350 emitted or not according to the place where they are first instantiated.
1351 @xref{Template Instantiation}, for more information.
1353 This option is deprecated.
1356 @itemx -fno-for-scope
1358 @opindex fno-for-scope
1359 If @option{-ffor-scope} is specified, the scope of variables declared in
1360 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1361 as specified by the C++ standard.
1362 If @option{-fno-for-scope} is specified, the scope of variables declared in
1363 a @i{for-init-statement} extends to the end of the enclosing scope,
1364 as was the case in old versions of G++, and other (traditional)
1365 implementations of C++.
1367 The default if neither flag is given to follow the standard,
1368 but to allow and give a warning for old-style code that would
1369 otherwise be invalid, or have different behavior.
1371 @item -fno-gnu-keywords
1372 @opindex fno-gnu-keywords
1373 Do not recognize @code{typeof} as a keyword, so that code can use this
1374 word as an identifier. You can use the keyword @code{__typeof__} instead.
1375 @option{-ansi} implies @option{-fno-gnu-keywords}.
1377 @item -fno-implicit-templates
1378 @opindex fno-implicit-templates
1379 Never emit code for non-inline templates which are instantiated
1380 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1381 @xref{Template Instantiation}, for more information.
1383 @item -fno-implicit-inline-templates
1384 @opindex fno-implicit-inline-templates
1385 Don't emit code for implicit instantiations of inline templates, either.
1386 The default is to handle inlines differently so that compiles with and
1387 without optimization will need the same set of explicit instantiations.
1389 @item -fno-implement-inlines
1390 @opindex fno-implement-inlines
1391 To save space, do not emit out-of-line copies of inline functions
1392 controlled by @samp{#pragma implementation}. This will cause linker
1393 errors if these functions are not inlined everywhere they are called.
1395 @item -fms-extensions
1396 @opindex fms-extensions
1397 Disable pedantic warnings about constructs used in MFC, such as implicit
1398 int and getting a pointer to member function via non-standard syntax.
1400 @item -fno-nonansi-builtins
1401 @opindex fno-nonansi-builtins
1402 Disable built-in declarations of functions that are not mandated by
1403 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1404 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1406 @item -fno-operator-names
1407 @opindex fno-operator-names
1408 Do not treat the operator name keywords @code{and}, @code{bitand},
1409 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1410 synonyms as keywords.
1412 @item -fno-optional-diags
1413 @opindex fno-optional-diags
1414 Disable diagnostics that the standard says a compiler does not need to
1415 issue. Currently, the only such diagnostic issued by G++ is the one for
1416 a name having multiple meanings within a class.
1419 @opindex fpermissive
1420 Downgrade some diagnostics about nonconformant code from errors to
1421 warnings. Thus, using @option{-fpermissive} will allow some
1422 nonconforming code to compile.
1426 Enable automatic template instantiation at link time. This option also
1427 implies @option{-fno-implicit-templates}. @xref{Template
1428 Instantiation}, for more information.
1432 Disable generation of information about every class with virtual
1433 functions for use by the C++ runtime type identification features
1434 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1435 of the language, you can save some space by using this flag. Note that
1436 exception handling uses the same information, but it will generate it as
1441 Emit statistics about front-end processing at the end of the compilation.
1442 This information is generally only useful to the G++ development team.
1444 @item -ftemplate-depth-@var{n}
1445 @opindex ftemplate-depth
1446 Set the maximum instantiation depth for template classes to @var{n}.
1447 A limit on the template instantiation depth is needed to detect
1448 endless recursions during template class instantiation. ANSI/ISO C++
1449 conforming programs must not rely on a maximum depth greater than 17.
1451 @item -fuse-cxa-atexit
1452 @opindex fuse-cxa-atexit
1453 Register destructors for objects with static storage duration with the
1454 @code{__cxa_atexit} function rather than the @code{atexit} function.
1455 This option is required for fully standards-compliant handling of static
1456 destructors, but will only work if your C library supports
1457 @code{__cxa_atexit}.
1461 Do not use weak symbol support, even if it is provided by the linker.
1462 By default, G++ will use weak symbols if they are available. This
1463 option exists only for testing, and should not be used by end-users;
1464 it will result in inferior code and has no benefits. This option may
1465 be removed in a future release of G++.
1469 Do not search for header files in the standard directories specific to
1470 C++, but do still search the other standard directories. (This option
1471 is used when building the C++ library.)
1474 In addition, these optimization, warning, and code generation options
1475 have meanings only for C++ programs:
1478 @item -fno-default-inline
1479 @opindex fno-default-inline
1480 Do not assume @samp{inline} for functions defined inside a class scope.
1481 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1482 functions will have linkage like inline functions; they just won't be
1485 @item -Wabi @r{(C++ only)}
1487 Warn when G++ generates code that is probably not compatible with the
1488 vendor-neutral C++ ABI. Although an effort has been made to warn about
1489 all such cases, there are probably some cases that are not warned about,
1490 even though G++ is generating incompatible code. There may also be
1491 cases where warnings are emitted even though the code that is generated
1494 You should rewrite your code to avoid these warnings if you are
1495 concerned about the fact that code generated by G++ may not be binary
1496 compatible with code generated by other compilers.
1498 The known incompatibilities at this point include:
1503 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
1504 pack data into the same byte as a base class. For example:
1507 struct A @{ virtual void f(); int f1 : 1; @};
1508 struct B : public A @{ int f2 : 1; @};
1512 In this case, G++ will place @code{B::f2} into the same byte
1513 as@code{A::f1}; other compilers will not. You can avoid this problem
1514 by explicitly padding @code{A} so that its size is a multiple of the
1515 byte size on your platform; that will cause G++ and other compilers to
1516 layout @code{B} identically.
1519 Incorrect handling of tail-padding for virtual bases. G++ does not use
1520 tail padding when laying out virtual bases. For example:
1523 struct A @{ virtual void f(); char c1; @};
1524 struct B @{ B(); char c2; @};
1525 struct C : public A, public virtual B @{@};
1529 In this case, G++ will not place @code{B} into the tail-padding for
1530 @code{A}; other compilers will. You can avoid this problem by
1531 explicitly padding @code{A} so that its size is a multiple of its
1532 alignment (ignoring virtual base classes); that will cause G++ and other
1533 compilers to layout @code{C} identically.
1536 Incorrect handling of bit-fields with declared widths greater than that
1537 of their underlying types, when the bit-fields appear in a union. For
1541 union U @{ int i : 4096; @};
1545 Assuming that an @code{int} does not have 4096 bits, G++ will make the
1546 union too small by the number of bits in an @code{int}.
1549 Empty classes can be placed at incorrect offsets. For example:
1559 struct C : public B, public A @{@};
1563 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
1564 it should be placed at offset zero. G++ mistakenly believes that the
1565 @code{A} data member of @code{B} is already at offset zero.
1568 Names of template functions whose types involve @code{typename} or
1569 template template parameters can be mangled incorrectly.
1572 template <typename Q>
1573 void f(typename Q::X) @{@}
1575 template <template <typename> class Q>
1576 void f(typename Q<int>::X) @{@}
1580 Instantiations of these templates may be mangled incorrectly.
1584 @item -Wctor-dtor-privacy @r{(C++ only)}
1585 @opindex Wctor-dtor-privacy
1586 Warn when a class seems unusable because all the constructors or
1587 destructors in that class are private, and it has neither friends nor
1588 public static member functions.
1590 @item -Wnon-virtual-dtor @r{(C++ only)}
1591 @opindex Wnon-virtual-dtor
1592 Warn when a class appears to be polymorphic, thereby requiring a virtual
1593 destructor, yet it declares a non-virtual one.
1594 This warning is enabled by @option{-Wall}.
1596 @item -Wreorder @r{(C++ only)}
1598 @cindex reordering, warning
1599 @cindex warning for reordering of member initializers
1600 Warn when the order of member initializers given in the code does not
1601 match the order in which they must be executed. For instance:
1607 A(): j (0), i (1) @{ @}
1611 The compiler will rearrange the member initializers for @samp{i}
1612 and @samp{j} to match the declaration order of the members, emitting
1613 a warning to that effect. This warning is enabled by @option{-Wall}.
1616 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
1619 @item -Weffc++ @r{(C++ only)}
1621 Warn about violations of the following style guidelines from Scott Meyers'
1622 @cite{Effective C++} book:
1626 Item 11: Define a copy constructor and an assignment operator for classes
1627 with dynamically allocated memory.
1630 Item 12: Prefer initialization to assignment in constructors.
1633 Item 14: Make destructors virtual in base classes.
1636 Item 15: Have @code{operator=} return a reference to @code{*this}.
1639 Item 23: Don't try to return a reference when you must return an object.
1643 Also warn about violations of the following style guidelines from
1644 Scott Meyers' @cite{More Effective C++} book:
1648 Item 6: Distinguish between prefix and postfix forms of increment and
1649 decrement operators.
1652 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
1656 When selecting this option, be aware that the standard library
1657 headers do not obey all of these guidelines; use @samp{grep -v}
1658 to filter out those warnings.
1660 @item -Wno-deprecated @r{(C++ only)}
1661 @opindex Wno-deprecated
1662 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1664 @item -Wno-non-template-friend @r{(C++ only)}
1665 @opindex Wno-non-template-friend
1666 Disable warnings when non-templatized friend functions are declared
1667 within a template. Since the advent of explicit template specification
1668 support in G++, if the name of the friend is an unqualified-id (i.e.,
1669 @samp{friend foo(int)}), the C++ language specification demands that the
1670 friend declare or define an ordinary, nontemplate function. (Section
1671 14.5.3). Before G++ implemented explicit specification, unqualified-ids
1672 could be interpreted as a particular specialization of a templatized
1673 function. Because this non-conforming behavior is no longer the default
1674 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
1675 check existing code for potential trouble spots and is on by default.
1676 This new compiler behavior can be turned off with
1677 @option{-Wno-non-template-friend} which keeps the conformant compiler code
1678 but disables the helpful warning.
1680 @item -Wold-style-cast @r{(C++ only)}
1681 @opindex Wold-style-cast
1682 Warn if an old-style (C-style) cast to a non-void type is used within
1683 a C++ program. The new-style casts (@samp{static_cast},
1684 @samp{reinterpret_cast}, and @samp{const_cast}) are less vulnerable to
1685 unintended effects and much easier to search for.
1687 @item -Woverloaded-virtual @r{(C++ only)}
1688 @opindex Woverloaded-virtual
1689 @cindex overloaded virtual fn, warning
1690 @cindex warning for overloaded virtual fn
1691 Warn when a function declaration hides virtual functions from a
1692 base class. For example, in:
1699 struct B: public A @{
1704 the @code{A} class version of @code{f} is hidden in @code{B}, and code
1712 will fail to compile.
1714 @item -Wno-pmf-conversions @r{(C++ only)}
1715 @opindex Wno-pmf-conversions
1716 Disable the diagnostic for converting a bound pointer to member function
1719 @item -Wsign-promo @r{(C++ only)}
1720 @opindex Wsign-promo
1721 Warn when overload resolution chooses a promotion from unsigned or
1722 enumeral type to a signed type, over a conversion to an unsigned type of
1723 the same size. Previous versions of G++ would try to preserve
1724 unsignedness, but the standard mandates the current behavior.
1726 @item -Wsynth @r{(C++ only)}
1728 @cindex warning for synthesized methods
1729 @cindex synthesized methods, warning
1730 Warn when G++'s synthesis behavior does not match that of cfront. For
1736 A& operator = (int);
1746 In this example, G++ will synthesize a default @samp{A& operator =
1747 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1750 @node Objective-C Dialect Options
1751 @section Options Controlling Objective-C Dialect
1753 @cindex compiler options, Objective-C
1754 @cindex Objective-C options, command line
1755 @cindex options, Objective-C
1756 (NOTE: This manual does not describe the Objective-C language itself. See
1757 @w{@uref{http://gcc.gnu.org/readings.html}} for references.)
1759 This section describes the command-line options that are only meaningful
1760 for Objective-C programs, but you can also use most of the GNU compiler
1761 options regardless of what language your program is in. For example,
1762 you might compile a file @code{some_class.m} like this:
1765 gcc -g -fgnu-runtime -O -c some_class.m
1769 In this example, @option{-fgnu-runtime} is an option meant only for
1770 Objective-C programs; you can use the other options with any language
1773 Here is a list of options that are @emph{only} for compiling Objective-C
1777 @item -fconstant-string-class=@var{class-name}
1778 @opindex fconstant-string-class
1779 Use @var{class-name} as the name of the class to instantiate for each
1780 literal string specified with the syntax @code{@@"@dots{}"}. The default
1781 class name is @code{NXConstantString} if the GNU runtime is being used, and
1782 @code{NSConstantString} if the NeXT runtime is being used (see below). The
1783 @option{-fconstant-cfstrings} option, if also present, will override the
1784 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
1785 to be laid out as constant CoreFoundation strings.
1788 @opindex fgnu-runtime
1789 Generate object code compatible with the standard GNU Objective-C
1790 runtime. This is the default for most types of systems.
1792 @item -fnext-runtime
1793 @opindex fnext-runtime
1794 Generate output compatible with the NeXT runtime. This is the default
1795 for NeXT-based systems, including Darwin and Mac OS X@. The macro
1796 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
1799 @item -fno-nil-receivers
1800 @opindex -fno-nil-receivers
1801 Assume that all Objective-C message dispatches (e.g.,
1802 @code{[receiver message:arg]}) in this translation unit ensure that the receiver
1803 is not @code{nil}. This allows for more efficient entry points in the runtime to be
1804 used. Currently, this option is only available in conjunction with
1805 the NeXT runtime on Mac OS X 10.3 and later.
1807 @item -fobjc-exceptions
1808 @opindex -fobjc-exceptions
1809 Enable syntactic support for structured exception handling in Objective-C,
1810 similar to what is offered by C++ and Java. Currently, this option is only
1811 available in conjunction with the NeXT runtime on Mac OS X 10.3 and later.
1819 @@catch (AnObjCClass *exc) @{
1826 @@catch (AnotherClass *exc) @{
1829 @@catch (id allOthers) @{
1839 The @code{@@throw} statement may appear anywhere in an Objective-C or
1840 Objective-C++ program; when used inside of a @code{@@catch} block, the
1841 @code{@@throw} may appear without an argument (as shown above), in which case
1842 the object caught by the @code{@@catch} will be rethrown.
1844 Note that only (pointers to) Objective-C objects may be thrown and
1845 caught using this scheme. When an object is thrown, it will be caught
1846 by the nearest @code{@@catch} clause capable of handling objects of that type,
1847 analogously to how @code{catch} blocks work in C++ and Java. A
1848 @code{@@catch(id @dots{})} clause (as shown above) may also be provided to catch
1849 any and all Objective-C exceptions not caught by previous @code{@@catch}
1852 The @code{@@finally} clause, if present, will be executed upon exit from the
1853 immediately preceding @code{@@try @dots{} @@catch} section. This will happen
1854 regardless of whether any exceptions are thrown, caught or rethrown
1855 inside the @code{@@try @dots{} @@catch} section, analogously to the behavior
1856 of the @code{finally} clause in Java.
1858 There are several caveats to using the new exception mechanism:
1862 Although currently designed to be binary compatible with @code{NS_HANDLER}-style
1863 idioms provided by the @code{NSException} class, the new
1864 exceptions can only be used on Mac OS X 10.3 (Panther) and later
1865 systems, due to additional functionality needed in the (NeXT) Objective-C
1869 As mentioned above, the new exceptions do not support handling
1870 types other than Objective-C objects. Furthermore, when used from
1871 Objective-C++, the Objective-C exception model does not interoperate with C++
1872 exceptions at this time. This means you cannot @code{@@throw} an exception
1873 from Objective-C and @code{catch} it in C++, or vice versa
1874 (i.e., @code{throw @dots{} @@catch}).
1877 The @option{-fobjc-exceptions} switch also enables the use of synchronization
1878 blocks for thread-safe execution:
1881 @@synchronized (ObjCClass *guard) @{
1886 Upon entering the @code{@@synchronized} block, a thread of execution shall
1887 first check whether a lock has been placed on the corresponding @code{guard}
1888 object by another thread. If it has, the current thread shall wait until
1889 the other thread relinquishes its lock. Once @code{guard} becomes available,
1890 the current thread will place its own lock on it, execute the code contained in
1891 the @code{@@synchronized} block, and finally relinquish the lock (thereby
1892 making @code{guard} available to other threads).
1894 Unlike Java, Objective-C does not allow for entire methods to be marked
1895 @code{@@synchronized}. Note that throwing exceptions out of
1896 @code{@@synchronized} blocks is allowed, and will cause the guarding object
1897 to be unlocked properly.
1899 @item -freplace-objc-classes
1900 @opindex -freplace-objc-classes
1901 Emit a special marker instructing @command{ld(1)} not to statically link in
1902 the resulting object file, and allow @command{dyld(1)} to load it in at
1903 run time instead. This is used in conjunction with the Fix-and-Continue
1904 debugging mode, where the object file in question may be recompiled and
1905 dynamically reloaded in the course of program execution, without the need
1906 to restart the program itself. Currently, Fix-and-Continue functionality
1907 is only available in conjunciton withthe NeXT runtime on Mac OS X 10.3
1911 @opindex -fzero-link
1912 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
1913 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
1914 compile time) with static class references that get initialized at load time,
1915 which improves run-time performance. Specifying the @option{-fzero-link} flag
1916 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
1917 to be retained. This is useful in Zero-Link debugging mode, since it allows
1918 for individual class implementations to be modified during program execution.
1922 Dump interface declarations for all classes seen in the source file to a
1923 file named @file{@var{sourcename}.decl}.
1926 @opindex Wno-protocol
1927 If a class is declared to implement a protocol, a warning is issued for
1928 every method in the protocol that is not implemented by the class. The
1929 default behavior is to issue a warning for every method not explicitly
1930 implemented in the class, even if a method implementation is inherited
1931 from the superclass. If you use the @code{-Wno-protocol} option, then
1932 methods inherited from the superclass are considered to be implemented,
1933 and no warning is issued for them.
1937 Warn if multiple methods of different types for the same selector are
1938 found during compilation. The check is performed on the list of methods
1939 in the final stage of compilation. Additionally, a check is performed
1940 for each selector appearing in a @code{@@selector(@dots{})}
1941 expression, and a corresponding method for that selector has been found
1942 during compilation. Because these checks scan the method table only at
1943 the end of compilation, these warnings are not produced if the final
1944 stage of compilation is not reached, for example because an error is
1945 found during compilation, or because the @code{-fsyntax-only} option is
1948 @item -Wundeclared-selector
1949 @opindex Wundeclared-selector
1950 Warn if a @code{@@selector(@dots{})} expression referring to an
1951 undeclared selector is found. A selector is considered undeclared if no
1952 method with that name has been declared before the
1953 @code{@@selector(@dots{})} expression, either explicitly in an
1954 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
1955 an @code{@@implementation} section. This option always performs its
1956 checks as soon as a @code{@@selector(@dots{})} expression is found,
1957 while @code{-Wselector} only performs its checks in the final stage of
1958 compilation. This also enforces the coding style convention
1959 that methods and selectors must be declared before being used.
1961 @c not documented because only avail via -Wp
1962 @c @item -print-objc-runtime-info
1966 @node Language Independent Options
1967 @section Options to Control Diagnostic Messages Formatting
1968 @cindex options to control diagnostics formatting
1969 @cindex diagnostic messages
1970 @cindex message formatting
1972 Traditionally, diagnostic messages have been formatted irrespective of
1973 the output device's aspect (e.g.@: its width, @dots{}). The options described
1974 below can be used to control the diagnostic messages formatting
1975 algorithm, e.g.@: how many characters per line, how often source location
1976 information should be reported. Right now, only the C++ front end can
1977 honor these options. However it is expected, in the near future, that
1978 the remaining front ends would be able to digest them correctly.
1981 @item -fmessage-length=@var{n}
1982 @opindex fmessage-length
1983 Try to format error messages so that they fit on lines of about @var{n}
1984 characters. The default is 72 characters for @command{g++} and 0 for the rest of
1985 the front ends supported by GCC@. If @var{n} is zero, then no
1986 line-wrapping will be done; each error message will appear on a single
1989 @opindex fdiagnostics-show-location
1990 @item -fdiagnostics-show-location=once
1991 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
1992 reporter to emit @emph{once} source location information; that is, in
1993 case the message is too long to fit on a single physical line and has to
1994 be wrapped, the source location won't be emitted (as prefix) again,
1995 over and over, in subsequent continuation lines. This is the default
1998 @item -fdiagnostics-show-location=every-line
1999 Only meaningful in line-wrapping mode. Instructs the diagnostic
2000 messages reporter to emit the same source location information (as
2001 prefix) for physical lines that result from the process of breaking
2002 a message which is too long to fit on a single line.
2006 @node Warning Options
2007 @section Options to Request or Suppress Warnings
2008 @cindex options to control warnings
2009 @cindex warning messages
2010 @cindex messages, warning
2011 @cindex suppressing warnings
2013 Warnings are diagnostic messages that report constructions which
2014 are not inherently erroneous but which are risky or suggest there
2015 may have been an error.
2017 You can request many specific warnings with options beginning @samp{-W},
2018 for example @option{-Wimplicit} to request warnings on implicit
2019 declarations. Each of these specific warning options also has a
2020 negative form beginning @samp{-Wno-} to turn off warnings;
2021 for example, @option{-Wno-implicit}. This manual lists only one of the
2022 two forms, whichever is not the default.
2024 The following options control the amount and kinds of warnings produced
2025 by GCC; for further, language-specific options also refer to
2026 @ref{C++ Dialect Options} and @ref{Objective-C Dialect Options}.
2029 @cindex syntax checking
2031 @opindex fsyntax-only
2032 Check the code for syntax errors, but don't do anything beyond that.
2036 Issue all the warnings demanded by strict ISO C and ISO C++;
2037 reject all programs that use forbidden extensions, and some other
2038 programs that do not follow ISO C and ISO C++. For ISO C, follows the
2039 version of the ISO C standard specified by any @option{-std} option used.
2041 Valid ISO C and ISO C++ programs should compile properly with or without
2042 this option (though a rare few will require @option{-ansi} or a
2043 @option{-std} option specifying the required version of ISO C)@. However,
2044 without this option, certain GNU extensions and traditional C and C++
2045 features are supported as well. With this option, they are rejected.
2047 @option{-pedantic} does not cause warning messages for use of the
2048 alternate keywords whose names begin and end with @samp{__}. Pedantic
2049 warnings are also disabled in the expression that follows
2050 @code{__extension__}. However, only system header files should use
2051 these escape routes; application programs should avoid them.
2052 @xref{Alternate Keywords}.
2054 Some users try to use @option{-pedantic} to check programs for strict ISO
2055 C conformance. They soon find that it does not do quite what they want:
2056 it finds some non-ISO practices, but not all---only those for which
2057 ISO C @emph{requires} a diagnostic, and some others for which
2058 diagnostics have been added.
2060 A feature to report any failure to conform to ISO C might be useful in
2061 some instances, but would require considerable additional work and would
2062 be quite different from @option{-pedantic}. We don't have plans to
2063 support such a feature in the near future.
2065 Where the standard specified with @option{-std} represents a GNU
2066 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
2067 corresponding @dfn{base standard}, the version of ISO C on which the GNU
2068 extended dialect is based. Warnings from @option{-pedantic} are given
2069 where they are required by the base standard. (It would not make sense
2070 for such warnings to be given only for features not in the specified GNU
2071 C dialect, since by definition the GNU dialects of C include all
2072 features the compiler supports with the given option, and there would be
2073 nothing to warn about.)
2075 @item -pedantic-errors
2076 @opindex pedantic-errors
2077 Like @option{-pedantic}, except that errors are produced rather than
2082 Inhibit all warning messages.
2086 Inhibit warning messages about the use of @samp{#import}.
2088 @item -Wchar-subscripts
2089 @opindex Wchar-subscripts
2090 Warn if an array subscript has type @code{char}. This is a common cause
2091 of error, as programmers often forget that this type is signed on some
2096 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
2097 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
2101 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
2102 the arguments supplied have types appropriate to the format string
2103 specified, and that the conversions specified in the format string make
2104 sense. This includes standard functions, and others specified by format
2105 attributes (@pxref{Function Attributes}), in the @code{printf},
2106 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
2107 not in the C standard) families.
2109 The formats are checked against the format features supported by GNU
2110 libc version 2.2. These include all ISO C90 and C99 features, as well
2111 as features from the Single Unix Specification and some BSD and GNU
2112 extensions. Other library implementations may not support all these
2113 features; GCC does not support warning about features that go beyond a
2114 particular library's limitations. However, if @option{-pedantic} is used
2115 with @option{-Wformat}, warnings will be given about format features not
2116 in the selected standard version (but not for @code{strfmon} formats,
2117 since those are not in any version of the C standard). @xref{C Dialect
2118 Options,,Options Controlling C Dialect}.
2120 Since @option{-Wformat} also checks for null format arguments for
2121 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
2123 @option{-Wformat} is included in @option{-Wall}. For more control over some
2124 aspects of format checking, the options @option{-Wno-format-y2k},
2125 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
2126 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
2127 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
2129 @item -Wno-format-y2k
2130 @opindex Wno-format-y2k
2131 If @option{-Wformat} is specified, do not warn about @code{strftime}
2132 formats which may yield only a two-digit year.
2134 @item -Wno-format-extra-args
2135 @opindex Wno-format-extra-args
2136 If @option{-Wformat} is specified, do not warn about excess arguments to a
2137 @code{printf} or @code{scanf} format function. The C standard specifies
2138 that such arguments are ignored.
2140 Where the unused arguments lie between used arguments that are
2141 specified with @samp{$} operand number specifications, normally
2142 warnings are still given, since the implementation could not know what
2143 type to pass to @code{va_arg} to skip the unused arguments. However,
2144 in the case of @code{scanf} formats, this option will suppress the
2145 warning if the unused arguments are all pointers, since the Single
2146 Unix Specification says that such unused arguments are allowed.
2148 @item -Wno-format-zero-length
2149 @opindex Wno-format-zero-length
2150 If @option{-Wformat} is specified, do not warn about zero-length formats.
2151 The C standard specifies that zero-length formats are allowed.
2153 @item -Wformat-nonliteral
2154 @opindex Wformat-nonliteral
2155 If @option{-Wformat} is specified, also warn if the format string is not a
2156 string literal and so cannot be checked, unless the format function
2157 takes its format arguments as a @code{va_list}.
2159 @item -Wformat-security
2160 @opindex Wformat-security
2161 If @option{-Wformat} is specified, also warn about uses of format
2162 functions that represent possible security problems. At present, this
2163 warns about calls to @code{printf} and @code{scanf} functions where the
2164 format string is not a string literal and there are no format arguments,
2165 as in @code{printf (foo);}. This may be a security hole if the format
2166 string came from untrusted input and contains @samp{%n}. (This is
2167 currently a subset of what @option{-Wformat-nonliteral} warns about, but
2168 in future warnings may be added to @option{-Wformat-security} that are not
2169 included in @option{-Wformat-nonliteral}.)
2173 Enable @option{-Wformat} plus format checks not included in
2174 @option{-Wformat}. Currently equivalent to @samp{-Wformat
2175 -Wformat-nonliteral -Wformat-security}.
2179 Warn about passing a null pointer for arguments marked as
2180 requiring a non-null value by the @code{nonnull} function attribute.
2182 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
2183 can be disabled with the @option{-Wno-nonnull} option.
2185 @item -Winit-self @r{(C, C++, and Objective-C only)}
2187 Warn about uninitialized variables which are initialized with themselves.
2188 Note this option can only be used with the @option{-Wuninitialized} option,
2189 which in turn only works with @option{-O1} and above.
2191 For example, GCC will warn about @code{i} being uninitialized in the
2192 following snippet only when @option{-Winit-self} has been specified:
2203 @item -Wimplicit-int
2204 @opindex Wimplicit-int
2205 Warn when a declaration does not specify a type.
2207 @item -Wimplicit-function-declaration
2208 @itemx -Werror-implicit-function-declaration
2209 @opindex Wimplicit-function-declaration
2210 @opindex Werror-implicit-function-declaration
2211 Give a warning (or error) whenever a function is used before being
2216 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
2220 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
2221 function with external linkage, returning int, taking either zero
2222 arguments, two, or three arguments of appropriate types.
2224 @item -Wmissing-braces
2225 @opindex Wmissing-braces
2226 Warn if an aggregate or union initializer is not fully bracketed. In
2227 the following example, the initializer for @samp{a} is not fully
2228 bracketed, but that for @samp{b} is fully bracketed.
2231 int a[2][2] = @{ 0, 1, 2, 3 @};
2232 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
2236 @opindex Wparentheses
2237 Warn if parentheses are omitted in certain contexts, such
2238 as when there is an assignment in a context where a truth value
2239 is expected, or when operators are nested whose precedence people
2240 often get confused about.
2242 Also warn about constructions where there may be confusion to which
2243 @code{if} statement an @code{else} branch belongs. Here is an example of
2258 In C, every @code{else} branch belongs to the innermost possible @code{if}
2259 statement, which in this example is @code{if (b)}. This is often not
2260 what the programmer expected, as illustrated in the above example by
2261 indentation the programmer chose. When there is the potential for this
2262 confusion, GCC will issue a warning when this flag is specified.
2263 To eliminate the warning, add explicit braces around the innermost
2264 @code{if} statement so there is no way the @code{else} could belong to
2265 the enclosing @code{if}. The resulting code would look like this:
2281 @item -Wsequence-point
2282 @opindex Wsequence-point
2283 Warn about code that may have undefined semantics because of violations
2284 of sequence point rules in the C standard.
2286 The C standard defines the order in which expressions in a C program are
2287 evaluated in terms of @dfn{sequence points}, which represent a partial
2288 ordering between the execution of parts of the program: those executed
2289 before the sequence point, and those executed after it. These occur
2290 after the evaluation of a full expression (one which is not part of a
2291 larger expression), after the evaluation of the first operand of a
2292 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
2293 function is called (but after the evaluation of its arguments and the
2294 expression denoting the called function), and in certain other places.
2295 Other than as expressed by the sequence point rules, the order of
2296 evaluation of subexpressions of an expression is not specified. All
2297 these rules describe only a partial order rather than a total order,
2298 since, for example, if two functions are called within one expression
2299 with no sequence point between them, the order in which the functions
2300 are called is not specified. However, the standards committee have
2301 ruled that function calls do not overlap.
2303 It is not specified when between sequence points modifications to the
2304 values of objects take effect. Programs whose behavior depends on this
2305 have undefined behavior; the C standard specifies that ``Between the
2306 previous and next sequence point an object shall have its stored value
2307 modified at most once by the evaluation of an expression. Furthermore,
2308 the prior value shall be read only to determine the value to be
2309 stored.''. If a program breaks these rules, the results on any
2310 particular implementation are entirely unpredictable.
2312 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
2313 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
2314 diagnosed by this option, and it may give an occasional false positive
2315 result, but in general it has been found fairly effective at detecting
2316 this sort of problem in programs.
2318 The present implementation of this option only works for C programs. A
2319 future implementation may also work for C++ programs.
2321 The C standard is worded confusingly, therefore there is some debate
2322 over the precise meaning of the sequence point rules in subtle cases.
2323 Links to discussions of the problem, including proposed formal
2324 definitions, may be found on our readings page, at
2325 @w{@uref{http://gcc.gnu.org/readings.html}}.
2328 @opindex Wreturn-type
2329 Warn whenever a function is defined with a return-type that defaults to
2330 @code{int}. Also warn about any @code{return} statement with no
2331 return-value in a function whose return-type is not @code{void}.
2333 For C++, a function without return type always produces a diagnostic
2334 message, even when @option{-Wno-return-type} is specified. The only
2335 exceptions are @samp{main} and functions defined in system headers.
2339 Warn whenever a @code{switch} statement has an index of enumeral type
2340 and lacks a @code{case} for one or more of the named codes of that
2341 enumeration. (The presence of a @code{default} label prevents this
2342 warning.) @code{case} labels outside the enumeration range also
2343 provoke warnings when this option is used.
2345 @item -Wswitch-default
2346 @opindex Wswitch-switch
2347 Warn whenever a @code{switch} statement does not have a @code{default}
2351 @opindex Wswitch-enum
2352 Warn whenever a @code{switch} statement has an index of enumeral type
2353 and lacks a @code{case} for one or more of the named codes of that
2354 enumeration. @code{case} labels outside the enumeration range also
2355 provoke warnings when this option is used.
2359 Warn if any trigraphs are encountered that might change the meaning of
2360 the program (trigraphs within comments are not warned about).
2362 @item -Wunused-function
2363 @opindex Wunused-function
2364 Warn whenever a static function is declared but not defined or a
2365 non\-inline static function is unused.
2367 @item -Wunused-label
2368 @opindex Wunused-label
2369 Warn whenever a label is declared but not used.
2371 To suppress this warning use the @samp{unused} attribute
2372 (@pxref{Variable Attributes}).
2374 @item -Wunused-parameter
2375 @opindex Wunused-parameter
2376 Warn whenever a function parameter is unused aside from its declaration.
2378 To suppress this warning use the @samp{unused} attribute
2379 (@pxref{Variable Attributes}).
2381 @item -Wunused-variable
2382 @opindex Wunused-variable
2383 Warn whenever a local variable or non-constant static variable is unused
2384 aside from its declaration
2386 To suppress this warning use the @samp{unused} attribute
2387 (@pxref{Variable Attributes}).
2389 @item -Wunused-value
2390 @opindex Wunused-value
2391 Warn whenever a statement computes a result that is explicitly not used.
2393 To suppress this warning cast the expression to @samp{void}.
2397 All the above @option{-Wunused} options combined.
2399 In order to get a warning about an unused function parameter, you must
2400 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
2401 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
2403 @item -Wuninitialized
2404 @opindex Wuninitialized
2405 Warn if an automatic variable is used without first being initialized or
2406 if a variable may be clobbered by a @code{setjmp} call.
2408 These warnings are possible only in optimizing compilation,
2409 because they require data flow information that is computed only
2410 when optimizing. If you don't specify @option{-O}, you simply won't
2413 If you want to warn about code which uses the uninitialized value of the
2414 variable in its own initializer, use the @option{-Winit-self} option.
2416 These warnings occur only for variables that are candidates for
2417 register allocation. Therefore, they do not occur for a variable that
2418 is declared @code{volatile}, or whose address is taken, or whose size
2419 is other than 1, 2, 4 or 8 bytes. Also, they do not occur for
2420 structures, unions or arrays, even when they are in registers.
2422 Note that there may be no warning about a variable that is used only
2423 to compute a value that itself is never used, because such
2424 computations may be deleted by data flow analysis before the warnings
2427 These warnings are made optional because GCC is not smart
2428 enough to see all the reasons why the code might be correct
2429 despite appearing to have an error. Here is one example of how
2450 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
2451 always initialized, but GCC doesn't know this. Here is
2452 another common case:
2457 if (change_y) save_y = y, y = new_y;
2459 if (change_y) y = save_y;
2464 This has no bug because @code{save_y} is used only if it is set.
2466 @cindex @code{longjmp} warnings
2467 This option also warns when a non-volatile automatic variable might be
2468 changed by a call to @code{longjmp}. These warnings as well are possible
2469 only in optimizing compilation.
2471 The compiler sees only the calls to @code{setjmp}. It cannot know
2472 where @code{longjmp} will be called; in fact, a signal handler could
2473 call it at any point in the code. As a result, you may get a warning
2474 even when there is in fact no problem because @code{longjmp} cannot
2475 in fact be called at the place which would cause a problem.
2477 Some spurious warnings can be avoided if you declare all the functions
2478 you use that never return as @code{noreturn}. @xref{Function
2481 @item -Wunknown-pragmas
2482 @opindex Wunknown-pragmas
2483 @cindex warning for unknown pragmas
2484 @cindex unknown pragmas, warning
2485 @cindex pragmas, warning of unknown
2486 Warn when a #pragma directive is encountered which is not understood by
2487 GCC@. If this command line option is used, warnings will even be issued
2488 for unknown pragmas in system header files. This is not the case if
2489 the warnings were only enabled by the @option{-Wall} command line option.
2491 @item -Wstrict-aliasing
2492 @opindex Wstrict-aliasing
2493 This option is only active when @option{-fstrict-aliasing} is active.
2494 It warns about code which might break the strict aliasing rules that the
2495 compiler is using for optimization. The warning does not catch all
2496 cases, but does attempt to catch the more common pitfalls. It is
2497 included in @option{-Wall}.
2501 All of the above @samp{-W} options combined. This enables all the
2502 warnings about constructions that some users consider questionable, and
2503 that are easy to avoid (or modify to prevent the warning), even in
2504 conjunction with macros. This also enables some language-specific
2505 warnings described in @ref{C++ Dialect Options} and
2506 @ref{Objective-C Dialect Options}.
2509 The following @option{-W@dots{}} options are not implied by @option{-Wall}.
2510 Some of them warn about constructions that users generally do not
2511 consider questionable, but which occasionally you might wish to check
2512 for; others warn about constructions that are necessary or hard to avoid
2513 in some cases, and there is no simple way to modify the code to suppress
2520 (This option used to be called @option{-W}. The older name is still
2521 supported, but the newer name is more descriptive.) Print extra warning
2522 messages for these events:
2526 A function can return either with or without a value. (Falling
2527 off the end of the function body is considered returning without
2528 a value.) For example, this function would evoke such a
2542 An expression-statement or the left-hand side of a comma expression
2543 contains no side effects.
2544 To suppress the warning, cast the unused expression to void.
2545 For example, an expression such as @samp{x[i,j]} will cause a warning,
2546 but @samp{x[(void)i,j]} will not.
2549 An unsigned value is compared against zero with @samp{<} or @samp{>=}.
2552 A comparison like @samp{x<=y<=z} appears; this is equivalent to
2553 @samp{(x<=y ? 1 : 0) <= z}, which is a different interpretation from
2554 that of ordinary mathematical notation.
2557 Storage-class specifiers like @code{static} are not the first things in
2558 a declaration. According to the C Standard, this usage is obsolescent.
2561 The return type of a function has a type qualifier such as @code{const}.
2562 Such a type qualifier has no effect, since the value returned by a
2563 function is not an lvalue. (But don't warn about the GNU extension of
2564 @code{volatile void} return types. That extension will be warned about
2565 if @option{-pedantic} is specified.)
2568 If @option{-Wall} or @option{-Wunused} is also specified, warn about unused
2572 A comparison between signed and unsigned values could produce an
2573 incorrect result when the signed value is converted to unsigned.
2574 (But don't warn if @option{-Wno-sign-compare} is also specified.)
2577 An aggregate has an initializer which does not initialize all members.
2578 For example, the following code would cause such a warning, because
2579 @code{x.h} would be implicitly initialized to zero:
2582 struct s @{ int f, g, h; @};
2583 struct s x = @{ 3, 4 @};
2587 A function parameter is declared without a type specifier in K&R-style
2595 An empty body occurs in an @samp{if} or @samp{else} statement.
2598 A pointer is compared against integer zero with @samp{<}, @samp{<=},
2599 @samp{>}, or @samp{>=}.
2602 A variable might be changed by @samp{longjmp} or @samp{vfork}.
2605 Any of several floating-point events that often indicate errors, such as
2606 overflow, underflow, loss of precision, etc.
2608 @item @r{(C++ only)}
2609 An enumerator and a non-enumerator both appear in a conditional expression.
2611 @item @r{(C++ only)}
2612 A non-static reference or non-static @samp{const} member appears in a
2613 class without constructors.
2615 @item @r{(C++ only)}
2616 Ambiguous virtual bases.
2618 @item @r{(C++ only)}
2619 Subscripting an array which has been declared @samp{register}.
2621 @item @r{(C++ only)}
2622 Taking the address of a variable which has been declared @samp{register}.
2624 @item @r{(C++ only)}
2625 A base class is not initialized in a derived class' copy constructor.
2628 @item -Wno-div-by-zero
2629 @opindex Wno-div-by-zero
2630 @opindex Wdiv-by-zero
2631 Do not warn about compile-time integer division by zero. Floating point
2632 division by zero is not warned about, as it can be a legitimate way of
2633 obtaining infinities and NaNs.
2635 @item -Wsystem-headers
2636 @opindex Wsystem-headers
2637 @cindex warnings from system headers
2638 @cindex system headers, warnings from
2639 Print warning messages for constructs found in system header files.
2640 Warnings from system headers are normally suppressed, on the assumption
2641 that they usually do not indicate real problems and would only make the
2642 compiler output harder to read. Using this command line option tells
2643 GCC to emit warnings from system headers as if they occurred in user
2644 code. However, note that using @option{-Wall} in conjunction with this
2645 option will @emph{not} warn about unknown pragmas in system
2646 headers---for that, @option{-Wunknown-pragmas} must also be used.
2649 @opindex Wfloat-equal
2650 Warn if floating point values are used in equality comparisons.
2652 The idea behind this is that sometimes it is convenient (for the
2653 programmer) to consider floating-point values as approximations to
2654 infinitely precise real numbers. If you are doing this, then you need
2655 to compute (by analyzing the code, or in some other way) the maximum or
2656 likely maximum error that the computation introduces, and allow for it
2657 when performing comparisons (and when producing output, but that's a
2658 different problem). In particular, instead of testing for equality, you
2659 would check to see whether the two values have ranges that overlap; and
2660 this is done with the relational operators, so equality comparisons are
2663 @item -Wtraditional @r{(C only)}
2664 @opindex Wtraditional
2665 Warn about certain constructs that behave differently in traditional and
2666 ISO C@. Also warn about ISO C constructs that have no traditional C
2667 equivalent, and/or problematic constructs which should be avoided.
2671 Macro parameters that appear within string literals in the macro body.
2672 In traditional C macro replacement takes place within string literals,
2673 but does not in ISO C@.
2676 In traditional C, some preprocessor directives did not exist.
2677 Traditional preprocessors would only consider a line to be a directive
2678 if the @samp{#} appeared in column 1 on the line. Therefore
2679 @option{-Wtraditional} warns about directives that traditional C
2680 understands but would ignore because the @samp{#} does not appear as the
2681 first character on the line. It also suggests you hide directives like
2682 @samp{#pragma} not understood by traditional C by indenting them. Some
2683 traditional implementations would not recognize @samp{#elif}, so it
2684 suggests avoiding it altogether.
2687 A function-like macro that appears without arguments.
2690 The unary plus operator.
2693 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
2694 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
2695 constants.) Note, these suffixes appear in macros defined in the system
2696 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
2697 Use of these macros in user code might normally lead to spurious
2698 warnings, however gcc's integrated preprocessor has enough context to
2699 avoid warning in these cases.
2702 A function declared external in one block and then used after the end of
2706 A @code{switch} statement has an operand of type @code{long}.
2709 A non-@code{static} function declaration follows a @code{static} one.
2710 This construct is not accepted by some traditional C compilers.
2713 The ISO type of an integer constant has a different width or
2714 signedness from its traditional type. This warning is only issued if
2715 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
2716 typically represent bit patterns, are not warned about.
2719 Usage of ISO string concatenation is detected.
2722 Initialization of automatic aggregates.
2725 Identifier conflicts with labels. Traditional C lacks a separate
2726 namespace for labels.
2729 Initialization of unions. If the initializer is zero, the warning is
2730 omitted. This is done under the assumption that the zero initializer in
2731 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
2732 initializer warnings and relies on default initialization to zero in the
2736 Conversions by prototypes between fixed/floating point values and vice
2737 versa. The absence of these prototypes when compiling with traditional
2738 C would cause serious problems. This is a subset of the possible
2739 conversion warnings, for the full set use @option{-Wconversion}.
2742 Use of ISO C style function definitions. This warning intentionally is
2743 @emph{not} issued for prototype declarations or variadic functions
2744 because these ISO C features will appear in your code when using
2745 libiberty's traditional C compatibility macros, @code{PARAMS} and
2746 @code{VPARAMS}. This warning is also bypassed for nested functions
2747 because that feature is already a gcc extension and thus not relevant to
2748 traditional C compatibility.
2751 @item -Wdeclaration-after-statement @r{(C only)}
2752 @opindex Wdeclaration-after-statement
2753 Warn when a declaration is found after a statement in a block. This
2754 construct, known from C++, was introduced with ISO C99 and is by default
2755 allowed in GCC@. It is not supported by ISO C90 and was not supported by
2756 GCC versions before GCC 3.0. @xref{Mixed Declarations}.
2760 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
2762 @item -Wendif-labels
2763 @opindex Wendif-labels
2764 Warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
2768 Warn whenever a local variable shadows another local variable, parameter or
2769 global variable or whenever a built-in function is shadowed.
2771 @item -Wlarger-than-@var{len}
2772 @opindex Wlarger-than
2773 Warn whenever an object of larger than @var{len} bytes is defined.
2775 @item -Wpointer-arith
2776 @opindex Wpointer-arith
2777 Warn about anything that depends on the ``size of'' a function type or
2778 of @code{void}. GNU C assigns these types a size of 1, for
2779 convenience in calculations with @code{void *} pointers and pointers
2782 @item -Wbad-function-cast @r{(C only)}
2783 @opindex Wbad-function-cast
2784 Warn whenever a function call is cast to a non-matching type.
2785 For example, warn if @code{int malloc()} is cast to @code{anything *}.
2789 Warn whenever a pointer is cast so as to remove a type qualifier from
2790 the target type. For example, warn if a @code{const char *} is cast
2791 to an ordinary @code{char *}.
2794 @opindex Wcast-align
2795 Warn whenever a pointer is cast such that the required alignment of the
2796 target is increased. For example, warn if a @code{char *} is cast to
2797 an @code{int *} on machines where integers can only be accessed at
2798 two- or four-byte boundaries.
2800 @item -Wwrite-strings
2801 @opindex Wwrite-strings
2802 When compiling C, give string constants the type @code{const
2803 char[@var{length}]} so that
2804 copying the address of one into a non-@code{const} @code{char *}
2805 pointer will get a warning; when compiling C++, warn about the
2806 deprecated conversion from string constants to @code{char *}.
2807 These warnings will help you find at
2808 compile time code that can try to write into a string constant, but
2809 only if you have been very careful about using @code{const} in
2810 declarations and prototypes. Otherwise, it will just be a nuisance;
2811 this is why we did not make @option{-Wall} request these warnings.
2814 @opindex Wconversion
2815 Warn if a prototype causes a type conversion that is different from what
2816 would happen to the same argument in the absence of a prototype. This
2817 includes conversions of fixed point to floating and vice versa, and
2818 conversions changing the width or signedness of a fixed point argument
2819 except when the same as the default promotion.
2821 Also, warn if a negative integer constant expression is implicitly
2822 converted to an unsigned type. For example, warn about the assignment
2823 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
2824 casts like @code{(unsigned) -1}.
2826 @item -Wsign-compare
2827 @opindex Wsign-compare
2828 @cindex warning for comparison of signed and unsigned values
2829 @cindex comparison of signed and unsigned values, warning
2830 @cindex signed and unsigned values, comparison warning
2831 Warn when a comparison between signed and unsigned values could produce
2832 an incorrect result when the signed value is converted to unsigned.
2833 This warning is also enabled by @option{-Wextra}; to get the other warnings
2834 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
2836 @item -Waggregate-return
2837 @opindex Waggregate-return
2838 Warn if any functions that return structures or unions are defined or
2839 called. (In languages where you can return an array, this also elicits
2842 @item -Wstrict-prototypes @r{(C only)}
2843 @opindex Wstrict-prototypes
2844 Warn if a function is declared or defined without specifying the
2845 argument types. (An old-style function definition is permitted without
2846 a warning if preceded by a declaration which specifies the argument
2849 @item -Wold-style-definition @r{(C only)}
2850 @opindex Wold-style-definition
2851 Warn if an old-style function definition is used. A warning is given
2852 even if there is a previous prototype.
2854 @item -Wmissing-prototypes @r{(C only)}
2855 @opindex Wmissing-prototypes
2856 Warn if a global function is defined without a previous prototype
2857 declaration. This warning is issued even if the definition itself
2858 provides a prototype. The aim is to detect global functions that fail
2859 to be declared in header files.
2861 @item -Wmissing-declarations @r{(C only)}
2862 @opindex Wmissing-declarations
2863 Warn if a global function is defined without a previous declaration.
2864 Do so even if the definition itself provides a prototype.
2865 Use this option to detect global functions that are not declared in
2868 @item -Wmissing-noreturn
2869 @opindex Wmissing-noreturn
2870 Warn about functions which might be candidates for attribute @code{noreturn}.
2871 Note these are only possible candidates, not absolute ones. Care should
2872 be taken to manually verify functions actually do not ever return before
2873 adding the @code{noreturn} attribute, otherwise subtle code generation
2874 bugs could be introduced. You will not get a warning for @code{main} in
2875 hosted C environments.
2877 @item -Wmissing-format-attribute
2878 @opindex Wmissing-format-attribute
2880 If @option{-Wformat} is enabled, also warn about functions which might be
2881 candidates for @code{format} attributes. Note these are only possible
2882 candidates, not absolute ones. GCC will guess that @code{format}
2883 attributes might be appropriate for any function that calls a function
2884 like @code{vprintf} or @code{vscanf}, but this might not always be the
2885 case, and some functions for which @code{format} attributes are
2886 appropriate may not be detected. This option has no effect unless
2887 @option{-Wformat} is enabled (possibly by @option{-Wall}).
2889 @item -Wno-multichar
2890 @opindex Wno-multichar
2892 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
2893 Usually they indicate a typo in the user's code, as they have
2894 implementation-defined values, and should not be used in portable code.
2896 @item -Wno-deprecated-declarations
2897 @opindex Wno-deprecated-declarations
2898 Do not warn about uses of functions, variables, and types marked as
2899 deprecated by using the @code{deprecated} attribute.
2900 (@pxref{Function Attributes}, @pxref{Variable Attributes},
2901 @pxref{Type Attributes}.)
2905 Warn if a structure is given the packed attribute, but the packed
2906 attribute has no effect on the layout or size of the structure.
2907 Such structures may be mis-aligned for little benefit. For
2908 instance, in this code, the variable @code{f.x} in @code{struct bar}
2909 will be misaligned even though @code{struct bar} does not itself
2910 have the packed attribute:
2917 @} __attribute__((packed));
2927 Warn if padding is included in a structure, either to align an element
2928 of the structure or to align the whole structure. Sometimes when this
2929 happens it is possible to rearrange the fields of the structure to
2930 reduce the padding and so make the structure smaller.
2932 @item -Wredundant-decls
2933 @opindex Wredundant-decls
2934 Warn if anything is declared more than once in the same scope, even in
2935 cases where multiple declaration is valid and changes nothing.
2937 @item -Wnested-externs @r{(C only)}
2938 @opindex Wnested-externs
2939 Warn if an @code{extern} declaration is encountered within a function.
2941 @item -Wunreachable-code
2942 @opindex Wunreachable-code
2943 Warn if the compiler detects that code will never be executed.
2945 This option is intended to warn when the compiler detects that at
2946 least a whole line of source code will never be executed, because
2947 some condition is never satisfied or because it is after a
2948 procedure that never returns.
2950 It is possible for this option to produce a warning even though there
2951 are circumstances under which part of the affected line can be executed,
2952 so care should be taken when removing apparently-unreachable code.
2954 For instance, when a function is inlined, a warning may mean that the
2955 line is unreachable in only one inlined copy of the function.
2957 This option is not made part of @option{-Wall} because in a debugging
2958 version of a program there is often substantial code which checks
2959 correct functioning of the program and is, hopefully, unreachable
2960 because the program does work. Another common use of unreachable
2961 code is to provide behavior which is selectable at compile-time.
2965 Warn if a function can not be inlined and it was declared as inline.
2966 Even with this option, the compiler will not warn about failures to
2967 inline functions declared in system headers.
2969 The compiler uses a variety of heuristics to determine whether or not
2970 to inline a function. For example, the compiler takes into account
2971 the size of the function being inlined and the the amount of inlining
2972 that has already been done in the current function. Therefore,
2973 seemingly insignificant changes in the source program can cause the
2974 warnings produced by @option{-Winline} to appear or disappear.
2976 @item -Wno-invalid-offsetof @r{(C++ only)}
2977 @opindex Wno-invalid-offsetof
2978 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
2979 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
2980 to a non-POD type is undefined. In existing C++ implementations,
2981 however, @samp{offsetof} typically gives meaningful results even when
2982 applied to certain kinds of non-POD types. (Such as a simple
2983 @samp{struct} that fails to be a POD type only by virtue of having a
2984 constructor.) This flag is for users who are aware that they are
2985 writing nonportable code and who have deliberately chosen to ignore the
2988 The restrictions on @samp{offsetof} may be relaxed in a future version
2989 of the C++ standard.
2992 @opindex Winvalid-pch
2993 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
2994 the search path but can't be used.
2998 @opindex Wno-long-long
2999 Warn if @samp{long long} type is used. This is default. To inhibit
3000 the warning messages, use @option{-Wno-long-long}. Flags
3001 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
3002 only when @option{-pedantic} flag is used.
3004 @item -Wdisabled-optimization
3005 @opindex Wdisabled-optimization
3006 Warn if a requested optimization pass is disabled. This warning does
3007 not generally indicate that there is anything wrong with your code; it
3008 merely indicates that GCC's optimizers were unable to handle the code
3009 effectively. Often, the problem is that your code is too big or too
3010 complex; GCC will refuse to optimize programs when the optimization
3011 itself is likely to take inordinate amounts of time.
3015 Make all warnings into errors.
3018 @node Debugging Options
3019 @section Options for Debugging Your Program or GCC
3020 @cindex options, debugging
3021 @cindex debugging information options
3023 GCC has various special options that are used for debugging
3024 either your program or GCC:
3029 Produce debugging information in the operating system's native format
3030 (stabs, COFF, XCOFF, or DWARF)@. GDB can work with this debugging
3033 On most systems that use stabs format, @option{-g} enables use of extra
3034 debugging information that only GDB can use; this extra information
3035 makes debugging work better in GDB but will probably make other debuggers
3037 refuse to read the program. If you want to control for certain whether
3038 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
3039 @option{-gxcoff+}, @option{-gxcoff}, @option{-gdwarf-1+}, @option{-gdwarf-1},
3040 or @option{-gvms} (see below).
3042 Unlike most other C compilers, GCC allows you to use @option{-g} with
3043 @option{-O}. The shortcuts taken by optimized code may occasionally
3044 produce surprising results: some variables you declared may not exist
3045 at all; flow of control may briefly move where you did not expect it;
3046 some statements may not be executed because they compute constant
3047 results or their values were already at hand; some statements may
3048 execute in different places because they were moved out of loops.
3050 Nevertheless it proves possible to debug optimized output. This makes
3051 it reasonable to use the optimizer for programs that might have bugs.
3053 The following options are useful when GCC is generated with the
3054 capability for more than one debugging format.
3058 Produce debugging information for use by GDB@. This means to use the
3059 most expressive format available (DWARF 2, stabs, or the native format
3060 if neither of those are supported), including GDB extensions if at all
3065 Produce debugging information in stabs format (if that is supported),
3066 without GDB extensions. This is the format used by DBX on most BSD
3067 systems. On MIPS, Alpha and System V Release 4 systems this option
3068 produces stabs debugging output which is not understood by DBX or SDB@.
3069 On System V Release 4 systems this option requires the GNU assembler.
3071 @item -feliminate-unused-debug-symbols
3072 @opindex feliminate-unused-debug-symbols
3073 Produce debugging information in stabs format (if that is supported),
3074 for only symbols that are actually used.
3078 Produce debugging information in stabs format (if that is supported),
3079 using GNU extensions understood only by the GNU debugger (GDB)@. The
3080 use of these extensions is likely to make other debuggers crash or
3081 refuse to read the program.
3085 Produce debugging information in COFF format (if that is supported).
3086 This is the format used by SDB on most System V systems prior to
3091 Produce debugging information in XCOFF format (if that is supported).
3092 This is the format used by the DBX debugger on IBM RS/6000 systems.
3096 Produce debugging information in XCOFF format (if that is supported),
3097 using GNU extensions understood only by the GNU debugger (GDB)@. The
3098 use of these extensions is likely to make other debuggers crash or
3099 refuse to read the program, and may cause assemblers other than the GNU
3100 assembler (GAS) to fail with an error.
3104 Produce debugging information in DWARF version 1 format (if that is
3105 supported). This is the format used by SDB on most System V Release 4
3108 This option is deprecated.
3112 Produce debugging information in DWARF version 1 format (if that is
3113 supported), using GNU extensions understood only by the GNU debugger
3114 (GDB)@. The use of these extensions is likely to make other debuggers
3115 crash or refuse to read the program.
3117 This option is deprecated.
3121 Produce debugging information in DWARF version 2 format (if that is
3122 supported). This is the format used by DBX on IRIX 6.
3126 Produce debugging information in VMS debug format (if that is
3127 supported). This is the format used by DEBUG on VMS systems.
3130 @itemx -ggdb@var{level}
3131 @itemx -gstabs@var{level}
3132 @itemx -gcoff@var{level}
3133 @itemx -gxcoff@var{level}
3134 @itemx -gvms@var{level}
3135 Request debugging information and also use @var{level} to specify how
3136 much information. The default level is 2.
3138 Level 1 produces minimal information, enough for making backtraces in
3139 parts of the program that you don't plan to debug. This includes
3140 descriptions of functions and external variables, but no information
3141 about local variables and no line numbers.
3143 Level 3 includes extra information, such as all the macro definitions
3144 present in the program. Some debuggers support macro expansion when
3145 you use @option{-g3}.
3147 Note that in order to avoid confusion between DWARF1 debug level 2,
3148 and DWARF2, neither @option{-gdwarf} nor @option{-gdwarf-2} accept
3149 a concatenated debug level. Instead use an additional @option{-g@var{level}}
3150 option to change the debug level for DWARF1 or DWARF2.
3152 @item -feliminate-dwarf2-dups
3153 @opindex feliminate-dwarf2-dups
3154 Compress DWARF2 debugging information by eliminating duplicated
3155 information about each symbol. This option only makes sense when
3156 generating DWARF2 debugging information with @option{-gdwarf-2}.
3158 @cindex @command{prof}
3161 Generate extra code to write profile information suitable for the
3162 analysis program @command{prof}. You must use this option when compiling
3163 the source files you want data about, and you must also use it when
3166 @cindex @command{gprof}
3169 Generate extra code to write profile information suitable for the
3170 analysis program @command{gprof}. You must use this option when compiling
3171 the source files you want data about, and you must also use it when
3176 Makes the compiler print out each function name as it is compiled, and
3177 print some statistics about each pass when it finishes.
3180 @opindex ftime-report
3181 Makes the compiler print some statistics about the time consumed by each
3182 pass when it finishes.
3185 @opindex fmem-report
3186 Makes the compiler print some statistics about permanent memory
3187 allocation when it finishes.
3189 @item -fprofile-arcs
3190 @opindex fprofile-arcs
3191 Add code so that program flow @dfn{arcs} are instrumented. During
3192 execution the program records how many times each branch and call is
3193 executed and how many times it is taken or returns. When the compiled
3194 program exits it saves this data to a file called
3195 @file{@var{auxname}.gcda} for each source file. The data may be used for
3196 profile-directed optimizations (@option{-fbranch-probabilities}), or for
3197 test coverage analysis (@option{-ftest-coverage}). Each object file's
3198 @var{auxname} is generated from the name of the output file, if
3199 explicitly specified and it is not the final executable, otherwise it is
3200 the basename of the source file. In both cases any suffix is removed
3201 (e.g. @file{foo.gcda} for input file @file{dir/foo.c}, or
3202 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
3207 Compile the source files with @option{-fprofile-arcs} plus optimization
3208 and code generation options. For test coverage analysis, use the
3209 additional @option{-ftest-coverage} option. You do not need to profile
3210 every source file in a program.
3213 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
3214 (the latter implies the former).
3217 Run the program on a representative workload to generate the arc profile
3218 information. This may be repeated any number of times. You can run
3219 concurrent instances of your program, and provided that the file system
3220 supports locking, the data files will be correctly updated. Also
3221 @code{fork} calls are detected and correctly handled (double counting
3225 For profile-directed optimizations, compile the source files again with
3226 the same optimization and code generation options plus
3227 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
3228 Control Optimization}).
3231 For test coverage analysis, use @command{gcov} to produce human readable
3232 information from the @file{.gcno} and @file{.gcda} files. Refer to the
3233 @command{gcov} documentation for further information.
3237 With @option{-fprofile-arcs}, for each function of your program GCC
3238 creates a program flow graph, then finds a spanning tree for the graph.
3239 Only arcs that are not on the spanning tree have to be instrumented: the
3240 compiler adds code to count the number of times that these arcs are
3241 executed. When an arc is the only exit or only entrance to a block, the
3242 instrumentation code can be added to the block; otherwise, a new basic
3243 block must be created to hold the instrumentation code.
3246 @item -ftest-coverage
3247 @opindex ftest-coverage
3248 Produce a notes file that the @command{gcov} code-coverage utility
3249 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
3250 show program coverage. Each source file's note file is called
3251 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
3252 above for a description of @var{auxname} and instructions on how to
3253 generate test coverage data. Coverage data will match the source files
3254 more closely, if you do not optimize.
3256 @item -d@var{letters}
3258 Says to make debugging dumps during compilation at times specified by
3259 @var{letters}. This is used for debugging the compiler. The file names
3260 for most of the dumps are made by appending a pass number and a word to
3261 the @var{dumpname}. @var{dumpname} is generated from the name of the
3262 output file, if explicitly specified and it is not an executable,
3263 otherwise it is the basename of the source file. In both cases any
3264 suffix is removed (e.g. @file{foo.00.rtl} or @file{foo.01.sibling}).
3265 Here are the possible letters for use in @var{letters}, and their
3271 Annotate the assembler output with miscellaneous debugging information.
3274 Dump after computing branch probabilities, to @file{@var{file}.16.bp}.
3277 Dump after block reordering, to @file{@var{file}.32.bbro}.
3280 Dump after instruction combination, to the file @file{@var{file}.22.combine}.
3283 Dump after the first if conversion, to the file @file{@var{file}.17.ce1}.
3284 Also dump after the second if conversion, to the file @file{@var{file}.23.ce2}.
3287 Dump after branch target load optimization, to to @file{@var{file}.34.btl}.
3288 Also dump after delayed branch scheduling, to @file{@var{file}.37.dbr}.
3291 Dump all macro definitions, at the end of preprocessing, in addition to
3295 Dump after SSA optimizations, to @file{@var{file}.05.ssa} and
3296 @file{@var{file}.010.ussa}.
3299 Dump after the second if conversion, to @file{@var{file}.33.ce3}.
3302 Dump after control and data flow analysis, to @file{@var{file}.15.cfg}.
3303 Also dump after life analysis, to @file{@var{file}.21.life}.
3306 Dump after purging @code{ADDRESSOF} codes, to @file{@var{file}.11.addressof}.
3309 Dump after global register allocation, to @file{@var{file}.27.greg}.
3312 Dump after GCSE, to @file{@var{file}.12.gcse}.
3313 Also dump after jump bypassing and control flow optimizations, to
3314 @file{@var{file}.14.bypass}.
3317 Dump after finalization of EH handling code, to @file{@var{file}.03.eh}.
3320 Dump after sibling call optimizations, to @file{@var{file}.02.sibling}.
3323 Dump after the first jump optimization, to @file{@var{file}.04.jump}.
3326 Dump after conversion from registers to stack, to @file{@var{file}.36.stack}.
3329 Dump after local register allocation, to @file{@var{file}.26.lreg}.
3332 Dump after loop optimization passes, to @file{@var{file}.13.loop} and
3333 @file{@var{file}.19.loop2}.
3336 Dump after performing the machine dependent reorganization pass, to
3337 @file{@var{file}.37.mach}.
3340 Dump after register renumbering, to @file{@var{file}.31.rnreg}.
3343 Dump after the register move pass, to @file{@var{file}.24.regmove}.
3346 Dump after post-reload optimizations, to @file{@var{file}.28.postreload}.
3349 Dump after RTL generation, to @file{@var{file}.01.rtl}.
3352 Dump after the second scheduling pass, to @file{@var{file}.35.sched2}.
3355 Dump after CSE (including the jump optimization that sometimes follows
3356 CSE), to @file{@var{file}.019.cse}.
3359 Dump after the first scheduling pass, to @file{@var{file}.25.sched}.
3362 Dump after the second CSE pass (including the jump optimization that
3363 sometimes follows CSE), to @file{@var{file}.20.cse2}.
3366 Dump after running tracer, to @file{@var{file}.18.tracer}.
3369 Dump after null pointer elimination pass to @file{@var{file}.018.null}.
3372 Dump callgraph and unit-at-a-time optimization @file{@var{file}.00.unit}.
3375 Dump after the second flow pass, to @file{@var{file}.29.flow2}.
3378 Dump after SSA conditional constant propagation, to
3379 @file{@var{file}.06.ssaccp}.
3382 Dump after SSA dead code elimination, to @file{@var{file}.07.ssadce}.
3385 Dump after the peephole pass, to @file{@var{file}.30.peephole2}.
3388 Produce all the dumps listed above.
3391 Produce a core dump whenever an error occurs.
3394 Print statistics on memory usage, at the end of the run, to
3398 Annotate the assembler output with a comment indicating which
3399 pattern and alternative was used. The length of each instruction is
3403 Dump the RTL in the assembler output as a comment before each instruction.
3404 Also turns on @option{-dp} annotation.
3407 For each of the other indicated dump files (except for
3408 @file{@var{file}.01.rtl}), dump a representation of the control flow graph
3409 suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
3412 Just generate RTL for a function instead of compiling it. Usually used
3416 Dump debugging information during parsing, to standard error.
3419 @item -fdump-unnumbered
3420 @opindex fdump-unnumbered
3421 When doing debugging dumps (see @option{-d} option above), suppress instruction
3422 numbers and line number note output. This makes it more feasible to
3423 use diff on debugging dumps for compiler invocations with different
3424 options, in particular with and without @option{-g}.
3426 @item -fdump-translation-unit @r{(C and C++ only)}
3427 @itemx -fdump-translation-unit-@var{options} @r{(C and C++ only)}
3428 @opindex fdump-translation-unit
3429 Dump a representation of the tree structure for the entire translation
3430 unit to a file. The file name is made by appending @file{.tu} to the
3431 source file name. If the @samp{-@var{options}} form is used, @var{options}
3432 controls the details of the dump as described for the
3433 @option{-fdump-tree} options.
3435 @item -fdump-class-hierarchy @r{(C++ only)}
3436 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
3437 @opindex fdump-class-hierarchy
3438 Dump a representation of each class's hierarchy and virtual function
3439 table layout to a file. The file name is made by appending @file{.class}
3440 to the source file name. If the @samp{-@var{options}} form is used,
3441 @var{options} controls the details of the dump as described for the
3442 @option{-fdump-tree} options.
3444 @item -fdump-tree-@var{switch} @r{(C++ only)}
3445 @itemx -fdump-tree-@var{switch}-@var{options} @r{(C++ only)}
3447 Control the dumping at various stages of processing the intermediate
3448 language tree to a file. The file name is generated by appending a switch
3449 specific suffix to the source file name. If the @samp{-@var{options}}
3450 form is used, @var{options} is a list of @samp{-} separated options that
3451 control the details of the dump. Not all options are applicable to all
3452 dumps, those which are not meaningful will be ignored. The following
3453 options are available
3457 Print the address of each node. Usually this is not meaningful as it
3458 changes according to the environment and source file. Its primary use
3459 is for tying up a dump file with a debug environment.
3461 Inhibit dumping of members of a scope or body of a function merely
3462 because that scope has been reached. Only dump such items when they
3463 are directly reachable by some other path.
3465 Turn on all options.
3468 The following tree dumps are possible:
3471 Dump before any tree based optimization, to @file{@var{file}.original}.
3473 Dump after all tree based optimization, to @file{@var{file}.optimized}.
3475 Dump after function inlining, to @file{@var{file}.inlined}.
3478 @item -frandom-seed=@var{string}
3479 @opindex frandom-string
3480 This option provides a seed that GCC uses when it would otherwise use
3481 random numbers. It is used to generate certain symbol names
3482 that have to be different in every compiled file. It is also used to
3483 place unique stamps in coverage data files and the object files that
3484 produce them. You can use the @option{-frandom-seed} option to produce
3485 reproducibly identical object files.
3487 The @var{string} should be different for every file you compile.
3489 @item -fsched-verbose=@var{n}
3490 @opindex fsched-verbose
3491 On targets that use instruction scheduling, this option controls the
3492 amount of debugging output the scheduler prints. This information is
3493 written to standard error, unless @option{-dS} or @option{-dR} is
3494 specified, in which case it is output to the usual dump
3495 listing file, @file{.sched} or @file{.sched2} respectively. However
3496 for @var{n} greater than nine, the output is always printed to standard
3499 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
3500 same information as @option{-dRS}. For @var{n} greater than one, it
3501 also output basic block probabilities, detailed ready list information
3502 and unit/insn info. For @var{n} greater than two, it includes RTL
3503 at abort point, control-flow and regions info. And for @var{n} over
3504 four, @option{-fsched-verbose} also includes dependence info.
3508 Store the usual ``temporary'' intermediate files permanently; place them
3509 in the current directory and name them based on the source file. Thus,
3510 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
3511 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
3512 preprocessed @file{foo.i} output file even though the compiler now
3513 normally uses an integrated preprocessor.
3517 Report the CPU time taken by each subprocess in the compilation
3518 sequence. For C source files, this is the compiler proper and assembler
3519 (plus the linker if linking is done). The output looks like this:
3526 The first number on each line is the ``user time,'' that is time spent
3527 executing the program itself. The second number is ``system time,''
3528 time spent executing operating system routines on behalf of the program.
3529 Both numbers are in seconds.
3531 @item -print-file-name=@var{library}
3532 @opindex print-file-name
3533 Print the full absolute name of the library file @var{library} that
3534 would be used when linking---and don't do anything else. With this
3535 option, GCC does not compile or link anything; it just prints the
3538 @item -print-multi-directory
3539 @opindex print-multi-directory
3540 Print the directory name corresponding to the multilib selected by any
3541 other switches present in the command line. This directory is supposed
3542 to exist in @env{GCC_EXEC_PREFIX}.
3544 @item -print-multi-lib
3545 @opindex print-multi-lib
3546 Print the mapping from multilib directory names to compiler switches
3547 that enable them. The directory name is separated from the switches by
3548 @samp{;}, and each switch starts with an @samp{@@} instead of the
3549 @samp{-}, without spaces between multiple switches. This is supposed to
3550 ease shell-processing.
3552 @item -print-prog-name=@var{program}
3553 @opindex print-prog-name
3554 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
3556 @item -print-libgcc-file-name
3557 @opindex print-libgcc-file-name
3558 Same as @option{-print-file-name=libgcc.a}.
3560 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
3561 but you do want to link with @file{libgcc.a}. You can do
3564 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
3567 @item -print-search-dirs
3568 @opindex print-search-dirs
3569 Print the name of the configured installation directory and a list of
3570 program and library directories gcc will search---and don't do anything else.
3572 This is useful when gcc prints the error message
3573 @samp{installation problem, cannot exec cpp0: No such file or directory}.
3574 To resolve this you either need to put @file{cpp0} and the other compiler
3575 components where gcc expects to find them, or you can set the environment
3576 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
3577 Don't forget the trailing '/'.
3578 @xref{Environment Variables}.
3581 @opindex dumpmachine
3582 Print the compiler's target machine (for example,
3583 @samp{i686-pc-linux-gnu})---and don't do anything else.
3586 @opindex dumpversion
3587 Print the compiler version (for example, @samp{3.0})---and don't do
3592 Print the compiler's built-in specs---and don't do anything else. (This
3593 is used when GCC itself is being built.) @xref{Spec Files}.
3595 @item -feliminate-unused-debug-types
3596 @opindex feliminate-unused-debug-types
3597 Normally, when producing DWARF2 output, GCC will emit debugging
3598 information for all types declared in a compilation
3599 unit, regardless of whether or not they are actually used
3600 in that compilation unit. Sometimes this is useful, such as
3601 if, in the debugger, you want to cast a value to a type that is
3602 not actually used in your program (but is declared). More often,
3603 however, this results in a significant amount of wasted space.
3604 With this option, GCC will avoid producing debug symbol output
3605 for types that are nowhere used in the source file being compiled.
3608 @node Optimize Options
3609 @section Options That Control Optimization
3610 @cindex optimize options
3611 @cindex options, optimization
3613 These options control various sorts of optimizations.
3615 Without any optimization option, the compiler's goal is to reduce the
3616 cost of compilation and to make debugging produce the expected
3617 results. Statements are independent: if you stop the program with a
3618 breakpoint between statements, you can then assign a new value to any
3619 variable or change the program counter to any other statement in the
3620 function and get exactly the results you would expect from the source
3623 Turning on optimization flags makes the compiler attempt to improve
3624 the performance and/or code size at the expense of compilation time
3625 and possibly the ability to debug the program.
3627 The compiler performs optimisation based on the knowledge it has of
3628 the program. Using the @option{-funit-at-a-time} flag will allow the
3629 compiler to consider information gained from later functions in the
3630 file when compiling a function. Compiling multiple files at once to a
3631 single output file (and using @option{-funit-at-a-time}) will allow
3632 the compiler to use information gained from all of the files when
3633 compiling each of them.
3635 Not all optimizations are controlled directly by a flag. Only
3636 optimizations that have a flag are listed.
3643 Optimize. Optimizing compilation takes somewhat more time, and a lot
3644 more memory for a large function.
3646 With @option{-O}, the compiler tries to reduce code size and execution
3647 time, without performing any optimizations that take a great deal of
3650 @option{-O} turns on the following optimization flags:
3651 @gccoptlist{-fdefer-pop @gol
3652 -fmerge-constants @gol
3654 -floop-optimize @gol
3656 -fif-conversion @gol
3657 -fif-conversion2 @gol
3658 -fdelayed-branch @gol
3659 -fguess-branch-probability @gol
3662 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
3663 where doing so does not interfere with debugging.
3667 Optimize even more. GCC performs nearly all supported optimizations
3668 that do not involve a space-speed tradeoff. The compiler does not
3669 perform loop unrolling or function inlining when you specify @option{-O2}.
3670 As compared to @option{-O}, this option increases both compilation time
3671 and the performance of the generated code.
3673 @option{-O2} turns on all optimization flags specified by @option{-O}. It
3674 also turns on the following optimization flags:
3675 @gccoptlist{-fforce-mem @gol
3676 -foptimize-sibling-calls @gol
3677 -fstrength-reduce @gol
3678 -fcse-follow-jumps -fcse-skip-blocks @gol
3679 -frerun-cse-after-loop -frerun-loop-opt @gol
3680 -fgcse -fgcse-lm -fgcse-sm -fgcse-las @gol
3681 -fdelete-null-pointer-checks @gol
3682 -fexpensive-optimizations @gol
3684 -fschedule-insns -fschedule-insns2 @gol
3685 -fsched-interblock -fsched-spec @gol
3688 -freorder-blocks -freorder-functions @gol
3689 -fstrict-aliasing @gol
3690 -funit-at-a-time @gol
3691 -falign-functions -falign-jumps @gol
3692 -falign-loops -falign-labels}
3694 Please note the warning under @option{-fgcse} about
3695 invoking @option{-O2} on programs that use computed gotos.
3699 Optimize yet more. @option{-O3} turns on all optimizations specified by
3700 @option{-O2} and also turns on the @option{-finline-functions},
3701 @option{-fweb} and @option{-frename-registers} options.
3705 Do not optimize. This is the default.
3709 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
3710 do not typically increase code size. It also performs further
3711 optimizations designed to reduce code size.
3713 @option{-Os} disables the following optimization flags:
3714 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
3715 -falign-labels -freorder-blocks -fprefetch-loop-arrays}
3717 If you use multiple @option{-O} options, with or without level numbers,
3718 the last such option is the one that is effective.
3721 Options of the form @option{-f@var{flag}} specify machine-independent
3722 flags. Most flags have both positive and negative forms; the negative
3723 form of @option{-ffoo} would be @option{-fno-foo}. In the table
3724 below, only one of the forms is listed---the one you typically will
3725 use. You can figure out the other form by either removing @samp{no-}
3728 The following options control specific optimizations. They are either
3729 activated by @option{-O} options or are related to ones that are. You
3730 can use the following flags in the rare cases when ``fine-tuning'' of
3731 optimizations to be performed is desired.
3734 @item -fno-default-inline
3735 @opindex fno-default-inline
3736 Do not make member functions inline by default merely because they are
3737 defined inside the class scope (C++ only). Otherwise, when you specify
3738 @w{@option{-O}}, member functions defined inside class scope are compiled
3739 inline by default; i.e., you don't need to add @samp{inline} in front of
3740 the member function name.
3742 @item -fno-defer-pop
3743 @opindex fno-defer-pop
3744 Always pop the arguments to each function call as soon as that function
3745 returns. For machines which must pop arguments after a function call,
3746 the compiler normally lets arguments accumulate on the stack for several
3747 function calls and pops them all at once.
3749 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3753 Force memory operands to be copied into registers before doing
3754 arithmetic on them. This produces better code by making all memory
3755 references potential common subexpressions. When they are not common
3756 subexpressions, instruction combination should eliminate the separate
3759 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3762 @opindex fforce-addr
3763 Force memory address constants to be copied into registers before
3764 doing arithmetic on them. This may produce better code just as
3765 @option{-fforce-mem} may.
3767 @item -fomit-frame-pointer
3768 @opindex fomit-frame-pointer
3769 Don't keep the frame pointer in a register for functions that
3770 don't need one. This avoids the instructions to save, set up and
3771 restore frame pointers; it also makes an extra register available
3772 in many functions. @strong{It also makes debugging impossible on
3775 On some machines, such as the VAX, this flag has no effect, because
3776 the standard calling sequence automatically handles the frame pointer
3777 and nothing is saved by pretending it doesn't exist. The
3778 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
3779 whether a target machine supports this flag. @xref{Registers,,Register
3780 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
3782 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3784 @item -foptimize-sibling-calls
3785 @opindex foptimize-sibling-calls
3786 Optimize sibling and tail recursive calls.
3788 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3792 Don't pay attention to the @code{inline} keyword. Normally this option
3793 is used to keep the compiler from expanding any functions inline.
3794 Note that if you are not optimizing, no functions can be expanded inline.
3796 @item -finline-functions
3797 @opindex finline-functions
3798 Integrate all simple functions into their callers. The compiler
3799 heuristically decides which functions are simple enough to be worth
3800 integrating in this way.
3802 If all calls to a given function are integrated, and the function is
3803 declared @code{static}, then the function is normally not output as
3804 assembler code in its own right.
3806 Enabled at level @option{-O3}.
3808 @item -finline-limit=@var{n}
3809 @opindex finline-limit
3810 By default, gcc limits the size of functions that can be inlined. This flag
3811 allows the control of this limit for functions that are explicitly marked as
3812 inline (i.e., marked with the inline keyword or defined within the class
3813 definition in c++). @var{n} is the size of functions that can be inlined in
3814 number of pseudo instructions (not counting parameter handling). The default
3815 value of @var{n} is 600.
3816 Increasing this value can result in more inlined code at
3817 the cost of compilation time and memory consumption. Decreasing usually makes
3818 the compilation faster and less code will be inlined (which presumably
3819 means slower programs). This option is particularly useful for programs that
3820 use inlining heavily such as those based on recursive templates with C++.
3822 Inlining is actually controlled by a number of parameters, which may be
3823 specified individually by using @option{--param @var{name}=@var{value}}.
3824 The @option{-finline-limit=@var{n}} option sets some of these parameters
3828 @item max-inline-insns
3830 @item max-inline-insns-single
3831 is set to @var{n}/2.
3832 @item max-inline-insns-auto
3833 is set to @var{n}/2.
3834 @item min-inline-insns
3835 is set to 130 or @var{n}/4, whichever is smaller.
3836 @item max-inline-insns-rtl
3840 Using @option{-finline-limit=600} thus results in the default settings
3841 for these parameters. See below for a documentation of the individual
3842 parameters controlling inlining.
3844 @emph{Note:} pseudo instruction represents, in this particular context, an
3845 abstract measurement of function's size. In no way, it represents a count
3846 of assembly instructions and as such its exact meaning might change from one
3847 release to an another.
3849 @item -fkeep-inline-functions
3850 @opindex fkeep-inline-functions
3851 Even if all calls to a given function are integrated, and the function
3852 is declared @code{static}, nevertheless output a separate run-time
3853 callable version of the function. This switch does not affect
3854 @code{extern inline} functions.
3856 @item -fkeep-static-consts
3857 @opindex fkeep-static-consts
3858 Emit variables declared @code{static const} when optimization isn't turned
3859 on, even if the variables aren't referenced.
3861 GCC enables this option by default. If you want to force the compiler to
3862 check if the variable was referenced, regardless of whether or not
3863 optimization is turned on, use the @option{-fno-keep-static-consts} option.
3865 @item -fmerge-constants
3866 Attempt to merge identical constants (string constants and floating point
3867 constants) across compilation units.
3869 This option is the default for optimized compilation if the assembler and
3870 linker support it. Use @option{-fno-merge-constants} to inhibit this
3873 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3875 @item -fmerge-all-constants
3876 Attempt to merge identical constants and identical variables.
3878 This option implies @option{-fmerge-constants}. In addition to
3879 @option{-fmerge-constants} this considers e.g. even constant initialized
3880 arrays or initialized constant variables with integral or floating point
3881 types. Languages like C or C++ require each non-automatic variable to
3882 have distinct location, so using this option will result in non-conforming
3887 Use a graph coloring register allocator. Currently this option is meant
3888 for testing, so we are interested to hear about miscompilations with
3891 @item -fno-branch-count-reg
3892 @opindex fno-branch-count-reg
3893 Do not use ``decrement and branch'' instructions on a count register,
3894 but instead generate a sequence of instructions that decrement a
3895 register, compare it against zero, then branch based upon the result.
3896 This option is only meaningful on architectures that support such
3897 instructions, which include x86, PowerPC, IA-64 and S/390.
3899 The default is @option{-fbranch-count-reg}, enabled when
3900 @option{-fstrength-reduce} is enabled.
3902 @item -fno-function-cse
3903 @opindex fno-function-cse
3904 Do not put function addresses in registers; make each instruction that
3905 calls a constant function contain the function's address explicitly.
3907 This option results in less efficient code, but some strange hacks
3908 that alter the assembler output may be confused by the optimizations
3909 performed when this option is not used.
3911 The default is @option{-ffunction-cse}
3913 @item -fno-zero-initialized-in-bss
3914 @opindex fno-zero-initialized-in-bss
3915 If the target supports a BSS section, GCC by default puts variables that
3916 are initialized to zero into BSS@. This can save space in the resulting
3919 This option turns off this behavior because some programs explicitly
3920 rely on variables going to the data section. E.g., so that the
3921 resulting executable can find the beginning of that section and/or make
3922 assumptions based on that.
3924 The default is @option{-fzero-initialized-in-bss}.
3926 @item -fstrength-reduce
3927 @opindex fstrength-reduce
3928 Perform the optimizations of loop strength reduction and
3929 elimination of iteration variables.
3931 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3933 @item -fthread-jumps
3934 @opindex fthread-jumps
3935 Perform optimizations where we check to see if a jump branches to a
3936 location where another comparison subsumed by the first is found. If
3937 so, the first branch is redirected to either the destination of the
3938 second branch or a point immediately following it, depending on whether
3939 the condition is known to be true or false.
3941 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3943 @item -fcse-follow-jumps
3944 @opindex fcse-follow-jumps
3945 In common subexpression elimination, scan through jump instructions
3946 when the target of the jump is not reached by any other path. For
3947 example, when CSE encounters an @code{if} statement with an
3948 @code{else} clause, CSE will follow the jump when the condition
3951 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3953 @item -fcse-skip-blocks
3954 @opindex fcse-skip-blocks
3955 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
3956 follow jumps which conditionally skip over blocks. When CSE
3957 encounters a simple @code{if} statement with no else clause,
3958 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
3959 body of the @code{if}.
3961 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3963 @item -frerun-cse-after-loop
3964 @opindex frerun-cse-after-loop
3965 Re-run common subexpression elimination after loop optimizations has been
3968 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3970 @item -frerun-loop-opt
3971 @opindex frerun-loop-opt
3972 Run the loop optimizer twice.
3974 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3978 Perform a global common subexpression elimination pass.
3979 This pass also performs global constant and copy propagation.
3981 @emph{Note:} When compiling a program using computed gotos, a GCC
3982 extension, you may get better runtime performance if you disable
3983 the global common subexpression elimination pass by adding
3984 @option{-fno-gcse} to the command line.
3986 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3990 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
3991 attempt to move loads which are only killed by stores into themselves. This
3992 allows a loop containing a load/store sequence to be changed to a load outside
3993 the loop, and a copy/store within the loop.
3995 Enabled by default when gcse is enabled.
3999 When @option{-fgcse-sm} is enabled, a store motion pass is run after
4000 global common subexpression elimination. This pass will attempt to move
4001 stores out of loops. When used in conjunction with @option{-fgcse-lm},
4002 loops containing a load/store sequence can be changed to a load before
4003 the loop and a store after the loop.
4005 Enabled by default when gcse is enabled.
4009 When @option{-fgcse-las} is enabled, the global common subexpression
4010 elimination pass eliminates redundant loads that come after stores to the
4011 same memory location (both partial and full redundacies).
4013 Enabled by default when gcse is enabled.
4015 @item -floop-optimize
4016 @opindex floop-optimize
4017 Perform loop optimizations: move constant expressions out of loops, simplify
4018 exit test conditions and optionally do strength-reduction and loop unrolling as
4021 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4023 @item -fcrossjumping
4024 @opindex crossjumping
4025 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
4026 resulting code may or may not perform better than without cross-jumping.
4028 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4030 @item -fif-conversion
4031 @opindex if-conversion
4032 Attempt to transform conditional jumps into branch-less equivalents. This
4033 include use of conditional moves, min, max, set flags and abs instructions, and
4034 some tricks doable by standard arithmetics. The use of conditional execution
4035 on chips where it is available is controlled by @code{if-conversion2}.
4037 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4039 @item -fif-conversion2
4040 @opindex if-conversion2
4041 Use conditional execution (where available) to transform conditional jumps into
4042 branch-less equivalents.
4044 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4046 @item -fdelete-null-pointer-checks
4047 @opindex fdelete-null-pointer-checks
4048 Use global dataflow analysis to identify and eliminate useless checks
4049 for null pointers. The compiler assumes that dereferencing a null
4050 pointer would have halted the program. If a pointer is checked after
4051 it has already been dereferenced, it cannot be null.
4053 In some environments, this assumption is not true, and programs can
4054 safely dereference null pointers. Use
4055 @option{-fno-delete-null-pointer-checks} to disable this optimization
4056 for programs which depend on that behavior.
4058 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4060 @item -fexpensive-optimizations
4061 @opindex fexpensive-optimizations
4062 Perform a number of minor optimizations that are relatively expensive.
4064 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4066 @item -foptimize-register-move
4068 @opindex foptimize-register-move
4070 Attempt to reassign register numbers in move instructions and as
4071 operands of other simple instructions in order to maximize the amount of
4072 register tying. This is especially helpful on machines with two-operand
4075 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
4078 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4080 @item -fdelayed-branch
4081 @opindex fdelayed-branch
4082 If supported for the target machine, attempt to reorder instructions
4083 to exploit instruction slots available after delayed branch
4086 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4088 @item -fschedule-insns
4089 @opindex fschedule-insns
4090 If supported for the target machine, attempt to reorder instructions to
4091 eliminate execution stalls due to required data being unavailable. This
4092 helps machines that have slow floating point or memory load instructions
4093 by allowing other instructions to be issued until the result of the load
4094 or floating point instruction is required.
4096 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4098 @item -fschedule-insns2
4099 @opindex fschedule-insns2
4100 Similar to @option{-fschedule-insns}, but requests an additional pass of
4101 instruction scheduling after register allocation has been done. This is
4102 especially useful on machines with a relatively small number of
4103 registers and where memory load instructions take more than one cycle.
4105 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4107 @item -fno-sched-interblock
4108 @opindex fno-sched-interblock
4109 Don't schedule instructions across basic blocks. This is normally
4110 enabled by default when scheduling before register allocation, i.e.@:
4111 with @option{-fschedule-insns} or at @option{-O2} or higher.
4113 @item -fno-sched-spec
4114 @opindex fno-sched-spec
4115 Don't allow speculative motion of non-load instructions. This is normally
4116 enabled by default when scheduling before register allocation, i.e.@:
4117 with @option{-fschedule-insns} or at @option{-O2} or higher.
4119 @item -fsched-spec-load
4120 @opindex fsched-spec-load
4121 Allow speculative motion of some load instructions. This only makes
4122 sense when scheduling before register allocation, i.e.@: with
4123 @option{-fschedule-insns} or at @option{-O2} or higher.
4125 @item -fsched-spec-load-dangerous
4126 @opindex fsched-spec-load-dangerous
4127 Allow speculative motion of more load instructions. This only makes
4128 sense when scheduling before register allocation, i.e.@: with
4129 @option{-fschedule-insns} or at @option{-O2} or higher.
4131 @item -fsched-stalled-insns=@var{n}
4132 @opindex fsched-stalled-insns
4133 Define how many insns (if any) can be moved prematurely from the queue
4134 of stalled insns into the ready list, during the second scheduling pass.
4136 @item -fsched-stalled-insns-dep=@var{n}
4137 @opindex fsched-stalled-insns-dep
4138 Define how many insn groups (cycles) will be examined for a dependency
4139 on a stalled insn that is candidate for premature removal from the queue
4140 of stalled insns. Has an effect only during the second scheduling pass,
4141 and only if @option{-fsched-stalled-insns} is used and its value is not zero.
4143 @item -fsched2-use-superblocks
4144 @opindex fsched2-use-superblocks
4145 When scheduling after register allocation, do use superblock scheduling
4146 algorithm. Superblock scheduling allows motion across basic block boundaries
4147 resulting on faster schedules. This option is experimental, as not all machine
4148 descriptions used by GCC model the CPU closely enough to avoid unreliable
4149 results from the algorithm.
4151 This only makes sense when scheduling after register allocation, i.e.@: with
4152 @option{-fschedule-insns2} or at @option{-O2} or higher.
4154 @item -fsched2-use-traces
4155 @opindex fsched2-use-traces
4156 Use @option{-fsched2-use-superblocks} algorithm when scheduling after register
4157 allocation and additionally perform code duplication in order to increase the
4158 size of superblocks using tracer pass. See @option{-ftracer} for details on
4161 This mode should produce faster but significantly longer programs. Also
4162 without @code{-fbranch-probabilities} the traces constructed may not match the
4163 reality and hurt the performance. This only makes
4164 sense when scheduling after register allocation, i.e.@: with
4165 @option{-fschedule-insns2} or at @option{-O2} or higher.
4167 @item -fcaller-saves
4168 @opindex fcaller-saves
4169 Enable values to be allocated in registers that will be clobbered by
4170 function calls, by emitting extra instructions to save and restore the
4171 registers around such calls. Such allocation is done only when it
4172 seems to result in better code than would otherwise be produced.
4174 This option is always enabled by default on certain machines, usually
4175 those which have no call-preserved registers to use instead.
4177 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4179 @item -fmove-all-movables
4180 @opindex fmove-all-movables
4181 Forces all invariant computations in loops to be moved
4184 @item -freduce-all-givs
4185 @opindex freduce-all-givs
4186 Forces all general-induction variables in loops to be
4189 @emph{Note:} When compiling programs written in Fortran,
4190 @option{-fmove-all-movables} and @option{-freduce-all-givs} are enabled
4191 by default when you use the optimizer.
4193 These options may generate better or worse code; results are highly
4194 dependent on the structure of loops within the source code.
4196 These two options are intended to be removed someday, once
4197 they have helped determine the efficacy of various
4198 approaches to improving loop optimizations.
4200 Please let us (@w{@email{gcc@@gcc.gnu.org}} and @w{@email{fortran@@gnu.org}})
4201 know how use of these options affects
4202 the performance of your production code.
4203 We're very interested in code that runs @emph{slower}
4204 when these options are @emph{enabled}.
4207 @itemx -fno-peephole2
4208 @opindex fno-peephole
4209 @opindex fno-peephole2
4210 Disable any machine-specific peephole optimizations. The difference
4211 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
4212 are implemented in the compiler; some targets use one, some use the
4213 other, a few use both.
4215 @option{-fpeephole} is enabled by default.
4216 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4218 @item -fno-guess-branch-probability
4219 @opindex fno-guess-branch-probability
4220 Do not guess branch probabilities using a randomized model.
4222 Sometimes gcc will opt to use a randomized model to guess branch
4223 probabilities, when none are available from either profiling feedback
4224 (@option{-fprofile-arcs}) or @samp{__builtin_expect}. This means that
4225 different runs of the compiler on the same program may produce different
4228 In a hard real-time system, people don't want different runs of the
4229 compiler to produce code that has different behavior; minimizing
4230 non-determinism is of paramount import. This switch allows users to
4231 reduce non-determinism, possibly at the expense of inferior
4234 The default is @option{-fguess-branch-probability} at levels
4235 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4237 @item -freorder-blocks
4238 @opindex freorder-blocks
4239 Reorder basic blocks in the compiled function in order to reduce number of
4240 taken branches and improve code locality.
4242 Enabled at levels @option{-O2}, @option{-O3}.
4244 @item -freorder-functions
4245 @opindex freorder-functions
4246 Reorder basic blocks in the compiled function in order to reduce number of
4247 taken branches and improve code locality. This is implemented by using special
4248 subsections @code{text.hot} for most frequently executed functions and
4249 @code{text.unlikely} for unlikely executed functions. Reordering is done by
4250 the linker so object file format must support named sections and linker must
4251 place them in a reasonable way.
4253 Also profile feedback must be available in to make this option effective. See
4254 @option{-fprofile-arcs} for details.
4256 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4258 @item -fstrict-aliasing
4259 @opindex fstrict-aliasing
4260 Allows the compiler to assume the strictest aliasing rules applicable to
4261 the language being compiled. For C (and C++), this activates
4262 optimizations based on the type of expressions. In particular, an
4263 object of one type is assumed never to reside at the same address as an
4264 object of a different type, unless the types are almost the same. For
4265 example, an @code{unsigned int} can alias an @code{int}, but not a
4266 @code{void*} or a @code{double}. A character type may alias any other
4269 Pay special attention to code like this:
4282 The practice of reading from a different union member than the one most
4283 recently written to (called ``type-punning'') is common. Even with
4284 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
4285 is accessed through the union type. So, the code above will work as
4286 expected. However, this code might not:
4297 Every language that wishes to perform language-specific alias analysis
4298 should define a function that computes, given an @code{tree}
4299 node, an alias set for the node. Nodes in different alias sets are not
4300 allowed to alias. For an example, see the C front-end function
4301 @code{c_get_alias_set}.
4303 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4305 @item -falign-functions
4306 @itemx -falign-functions=@var{n}
4307 @opindex falign-functions
4308 Align the start of functions to the next power-of-two greater than
4309 @var{n}, skipping up to @var{n} bytes. For instance,
4310 @option{-falign-functions=32} aligns functions to the next 32-byte
4311 boundary, but @option{-falign-functions=24} would align to the next
4312 32-byte boundary only if this can be done by skipping 23 bytes or less.
4314 @option{-fno-align-functions} and @option{-falign-functions=1} are
4315 equivalent and mean that functions will not be aligned.
4317 Some assemblers only support this flag when @var{n} is a power of two;
4318 in that case, it is rounded up.
4320 If @var{n} is not specified or is zero, use a machine-dependent default.
4322 Enabled at levels @option{-O2}, @option{-O3}.
4324 @item -falign-labels
4325 @itemx -falign-labels=@var{n}
4326 @opindex falign-labels
4327 Align all branch targets to a power-of-two boundary, skipping up to
4328 @var{n} bytes like @option{-falign-functions}. This option can easily
4329 make code slower, because it must insert dummy operations for when the
4330 branch target is reached in the usual flow of the code.
4332 @option{-fno-align-labels} and @option{-falign-labels=1} are
4333 equivalent and mean that labels will not be aligned.
4335 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
4336 are greater than this value, then their values are used instead.
4338 If @var{n} is not specified or is zero, use a machine-dependent default
4339 which is very likely to be @samp{1}, meaning no alignment.
4341 Enabled at levels @option{-O2}, @option{-O3}.
4344 @itemx -falign-loops=@var{n}
4345 @opindex falign-loops
4346 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
4347 like @option{-falign-functions}. The hope is that the loop will be
4348 executed many times, which will make up for any execution of the dummy
4351 @option{-fno-align-loops} and @option{-falign-loops=1} are
4352 equivalent and mean that loops will not be aligned.
4354 If @var{n} is not specified or is zero, use a machine-dependent default.
4356 Enabled at levels @option{-O2}, @option{-O3}.
4359 @itemx -falign-jumps=@var{n}
4360 @opindex falign-jumps
4361 Align branch targets to a power-of-two boundary, for branch targets
4362 where the targets can only be reached by jumping, skipping up to @var{n}
4363 bytes like @option{-falign-functions}. In this case, no dummy operations
4366 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
4367 equivalent and mean that loops will not be aligned.
4369 If @var{n} is not specified or is zero, use a machine-dependent default.
4371 Enabled at levels @option{-O2}, @option{-O3}.
4373 @item -frename-registers
4374 @opindex frename-registers
4375 Attempt to avoid false dependencies in scheduled code by making use
4376 of registers left over after register allocation. This optimization
4377 will most benefit processors with lots of registers. It can, however,
4378 make debugging impossible, since variables will no longer stay in
4379 a ``home register''.
4381 Enabled at levels @option{-O3}.
4383 @item -fno-cprop-registers
4384 @opindex fno-cprop-registers
4385 After register allocation and post-register allocation instruction splitting,
4386 we perform a copy-propagation pass to try to reduce scheduling dependencies
4387 and occasionally eliminate the copy.
4389 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4393 The following options control compiler behavior regarding floating
4394 point arithmetic. These options trade off between speed and
4395 correctness. All must be specifically enabled.
4399 @opindex ffloat-store
4400 Do not store floating point variables in registers, and inhibit other
4401 options that might change whether a floating point value is taken from a
4404 @cindex floating point precision
4405 This option prevents undesirable excess precision on machines such as
4406 the 68000 where the floating registers (of the 68881) keep more
4407 precision than a @code{double} is supposed to have. Similarly for the
4408 x86 architecture. For most programs, the excess precision does only
4409 good, but a few programs rely on the precise definition of IEEE floating
4410 point. Use @option{-ffloat-store} for such programs, after modifying
4411 them to store all pertinent intermediate computations into variables.
4415 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations}, @*
4416 @option{-fno-trapping-math}, @option{-ffinite-math-only},
4417 @option{-fno-rounding-math} and @option{-fno-signaling-nans}.
4419 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
4421 This option should never be turned on by any @option{-O} option since
4422 it can result in incorrect output for programs which depend on
4423 an exact implementation of IEEE or ISO rules/specifications for
4426 @item -fno-math-errno
4427 @opindex fno-math-errno
4428 Do not set ERRNO after calling math functions that are executed
4429 with a single instruction, e.g., sqrt. A program that relies on
4430 IEEE exceptions for math error handling may want to use this flag
4431 for speed while maintaining IEEE arithmetic compatibility.
4433 This option should never be turned on by any @option{-O} option since
4434 it can result in incorrect output for programs which depend on
4435 an exact implementation of IEEE or ISO rules/specifications for
4438 The default is @option{-fmath-errno}.
4440 @item -funsafe-math-optimizations
4441 @opindex funsafe-math-optimizations
4442 Allow optimizations for floating-point arithmetic that (a) assume
4443 that arguments and results are valid and (b) may violate IEEE or
4444 ANSI standards. When used at link-time, it may include libraries
4445 or startup files that change the default FPU control word or other
4446 similar optimizations.
4448 This option should never be turned on by any @option{-O} option since
4449 it can result in incorrect output for programs which depend on
4450 an exact implementation of IEEE or ISO rules/specifications for
4453 The default is @option{-fno-unsafe-math-optimizations}.
4455 @item -ffinite-math-only
4456 @opindex ffinite-math-only
4457 Allow optimizations for floating-point arithmetic that assume
4458 that arguments and results are not NaNs or +-Infs.
4460 This option should never be turned on by any @option{-O} option since
4461 it can result in incorrect output for programs which depend on
4462 an exact implementation of IEEE or ISO rules/specifications.
4464 The default is @option{-fno-finite-math-only}.
4466 @item -fno-trapping-math
4467 @opindex fno-trapping-math
4468 Compile code assuming that floating-point operations cannot generate
4469 user-visible traps. These traps include division by zero, overflow,
4470 underflow, inexact result and invalid operation. This option implies
4471 @option{-fno-signaling-nans}. Setting this option may allow faster
4472 code if one relies on ``non-stop'' IEEE arithmetic, for example.
4474 This option should never be turned on by any @option{-O} option since
4475 it can result in incorrect output for programs which depend on
4476 an exact implementation of IEEE or ISO rules/specifications for
4479 The default is @option{-ftrapping-math}.
4481 @item -frounding-math
4482 @opindex frounding-math
4483 Disable transformations and optimizations that assume default floating
4484 point rounding behavior. This is round-to-zero for all floating point
4485 to integer conversions, and round-to-nearest for all other arithmetic
4486 truncations. This option should be specified for programs that change
4487 the FP rounding mode dynamically, or that may be executed with a
4488 non-default rounding mode. This option disables constant folding of
4489 floating point expressions at compile-time (which may be affected by
4490 rounding mode) and arithmetic transformations that are unsafe in the
4491 presence of sign-dependent rounding modes.
4493 The default is @option{-fno-rounding-math}.
4495 This option is experimental and does not currently guarantee to
4496 disable all GCC optimizations that are affected by rounding mode.
4497 Future versions of gcc may provide finer control of this setting
4498 using C99's @code{FENV_ACCESS} pragma. This command line option
4499 will be used to specify the default state for @code{FENV_ACCESS}.
4501 @item -fsignaling-nans
4502 @opindex fsignaling-nans
4503 Compile code assuming that IEEE signaling NaNs may generate user-visible
4504 traps during floating-point operations. Setting this option disables
4505 optimizations that may change the number of exceptions visible with
4506 signaling NaNs. This option implies @option{-ftrapping-math}.
4508 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
4511 The default is @option{-fno-signaling-nans}.
4513 This option is experimental and does not currently guarantee to
4514 disable all GCC optimizations that affect signaling NaN behavior.
4516 @item -fsingle-precision-constant
4517 @opindex fsingle-precision-constant
4518 Treat floating point constant as single precision constant instead of
4519 implicitly converting it to double precision constant.
4524 The following options control optimizations that may improve
4525 performance, but are not enabled by any @option{-O} options. This
4526 section includes experimental options that may produce broken code.
4529 @item -fbranch-probabilities
4530 @opindex fbranch-probabilities
4531 After running a program compiled with @option{-fprofile-arcs}
4532 (@pxref{Debugging Options,, Options for Debugging Your Program or
4533 @command{gcc}}), you can compile it a second time using
4534 @option{-fbranch-probabilities}, to improve optimizations based on
4535 the number of times each branch was taken. When the program
4536 compiled with @option{-fprofile-arcs} exits it saves arc execution
4537 counts to a file called @file{@var{sourcename}.gcda} for each source
4538 file The information in this data file is very dependent on the
4539 structure of the generated code, so you must use the same source code
4540 and the same optimization options for both compilations.
4542 With @option{-fbranch-probabilities}, GCC puts a
4543 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
4544 These can be used to improve optimization. Currently, they are only
4545 used in one place: in @file{reorg.c}, instead of guessing which path a
4546 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
4547 exactly determine which path is taken more often.
4549 @item -fprofile-values
4550 @opindex fprofile-values
4551 If combined with @option{-fprofile-arcs}, it adds code so that some
4552 data about values of expressions in the program is gathered.
4554 With @option{-fbranch-probabilities}, it reads back the data gathered
4555 from profiling values of expressions and adds @samp{REG_VALUE_PROFILE}
4556 notes to instructions for their later usage in optimizations.
4560 Use a graph coloring register allocator. Currently this option is meant
4561 for testing, so we are interested to hear about miscompilations with
4566 Perform tail duplication to enlarge superblock size. This transformation
4567 simplifies the control flow of the function allowing other optimizations to do
4570 @item -funit-at-a-time
4571 @opindex funit-at-a-time
4572 Parse the whole compilation unit before starting to produce code.
4573 This allows some extra optimizations to take place but consumes more
4576 @item -funroll-loops
4577 @opindex funroll-loops
4578 Unroll loops whose number of iterations can be determined at compile time or
4579 upon entry to the loop. @option{-funroll-loops} implies
4580 @option{-frerun-cse-after-loop}. It also turns on complete loop peeling
4581 (i.e. complete removal of loops with small constant number of iterations).
4582 This option makes code larger, and may or may not make it run faster.
4584 @item -funroll-all-loops
4585 @opindex funroll-all-loops
4586 Unroll all loops, even if their number of iterations is uncertain when
4587 the loop is entered. This usually makes programs run more slowly.
4588 @option{-funroll-all-loops} implies the same options as
4589 @option{-funroll-loops}.
4592 @opindex fpeel-loops
4593 Peels the loops for that there is enough information that they do not
4594 roll much (from profile feedback). It also turns on complete loop peeling
4595 (i.e. complete removal of loops with small constant number of iterations).
4597 @item -funswitch-loops
4598 @opindex funswitch-loops
4599 Move branches with loop invariant conditions out of the loop, with duplicates
4600 of the loop on both branches (modified according to result of the condition).
4602 @item -fold-unroll-loops
4603 @opindex fold-unroll-loops
4604 Unroll loops whose number of iterations can be determined at compile
4605 time or upon entry to the loop, using the old loop unroller whose loop
4606 recognition is based on notes from frontend. @option{-fold-unroll-loops} implies
4607 both @option{-fstrength-reduce} and @option{-frerun-cse-after-loop}. This
4608 option makes code larger, and may or may not make it run faster.
4610 @item -fold-unroll-all-loops
4611 @opindex fold-unroll-all-loops
4612 Unroll all loops, even if their number of iterations is uncertain when
4613 the loop is entered. This is done using the old loop unroller whose loop
4614 recognition is based on notes from frontend. This usually makes programs run more slowly.
4615 @option{-fold-unroll-all-loops} implies the same options as
4616 @option{-fold-unroll-loops}.
4618 @item -funswitch-loops
4619 @opindex funswitch-loops
4620 Move branches with loop invariant conditions out of the loop, with duplicates
4621 of the loop on both branches (modified according to result of the condition).
4623 @item -funswitch-loops
4624 @opindex funswitch-loops
4625 Move branches with loop invariant conditions out of the loop, with duplicates
4626 of the loop on both branches (modified according to result of the condition).
4628 @item -fprefetch-loop-arrays
4629 @opindex fprefetch-loop-arrays
4630 If supported by the target machine, generate instructions to prefetch
4631 memory to improve the performance of loops that access large arrays.
4633 Disabled at level @option{-Os}.
4635 @item -ffunction-sections
4636 @itemx -fdata-sections
4637 @opindex ffunction-sections
4638 @opindex fdata-sections
4639 Place each function or data item into its own section in the output
4640 file if the target supports arbitrary sections. The name of the
4641 function or the name of the data item determines the section's name
4644 Use these options on systems where the linker can perform optimizations
4645 to improve locality of reference in the instruction space. Most systems
4646 using the ELF object format and SPARC processors running Solaris 2 have
4647 linkers with such optimizations. AIX may have these optimizations in
4650 Only use these options when there are significant benefits from doing
4651 so. When you specify these options, the assembler and linker will
4652 create larger object and executable files and will also be slower.
4653 You will not be able to use @code{gprof} on all systems if you
4654 specify this option and you may have problems with debugging if
4655 you specify both this option and @option{-g}.
4659 Perform optimizations in static single assignment form. Each function's
4660 flow graph is translated into SSA form, optimizations are performed, and
4661 the flow graph is translated back from SSA form. Users should not
4662 specify this option, since it is not yet ready for production use.
4666 Perform Sparse Conditional Constant Propagation in SSA form. Requires
4667 @option{-fssa}. Like @option{-fssa}, this is an experimental feature.
4671 Perform aggressive dead-code elimination in SSA form. Requires @option{-fssa}.
4672 Like @option{-fssa}, this is an experimental feature.
4674 @item -fbranch-target-load-optimize
4675 @opindex fbranch-target-load-optimize
4676 Perform branch target register load optimization before prologue / epilogue
4678 The use of target registers can typically be exposed only during reload,
4679 thus hoisting loads out of loops and doing inter-block scheduling needs
4680 a separate optimization pass.
4682 @item -fbranch-target-load-optimize2
4683 @opindex fbranch-target-load-optimize2
4684 Perform branch target register load optimization after prologue / epilogue
4690 @item --param @var{name}=@var{value}
4692 In some places, GCC uses various constants to control the amount of
4693 optimization that is done. For example, GCC will not inline functions
4694 that contain more that a certain number of instructions. You can
4695 control some of these constants on the command-line using the
4696 @option{--param} option.
4698 In each case, the @var{value} is an integer. The allowable choices for
4699 @var{name} are given in the following table:
4702 @item max-crossjump-edges
4703 The maximum number of incoming edges to consider for crossjumping.
4704 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
4705 the number of edges incoming to each block. Increasing values mean
4706 more aggressive optimization, making the compile time increase with
4707 probably small improvement in executable size.
4709 @item max-delay-slot-insn-search
4710 The maximum number of instructions to consider when looking for an
4711 instruction to fill a delay slot. If more than this arbitrary number of
4712 instructions is searched, the time savings from filling the delay slot
4713 will be minimal so stop searching. Increasing values mean more
4714 aggressive optimization, making the compile time increase with probably
4715 small improvement in executable run time.
4717 @item max-delay-slot-live-search
4718 When trying to fill delay slots, the maximum number of instructions to
4719 consider when searching for a block with valid live register
4720 information. Increasing this arbitrarily chosen value means more
4721 aggressive optimization, increasing the compile time. This parameter
4722 should be removed when the delay slot code is rewritten to maintain the
4725 @item max-gcse-memory
4726 The approximate maximum amount of memory that will be allocated in
4727 order to perform the global common subexpression elimination
4728 optimization. If more memory than specified is required, the
4729 optimization will not be done.
4731 @item max-gcse-passes
4732 The maximum number of passes of GCSE to run.
4734 @item max-pending-list-length
4735 The maximum number of pending dependencies scheduling will allow
4736 before flushing the current state and starting over. Large functions
4737 with few branches or calls can create excessively large lists which
4738 needlessly consume memory and resources.
4740 @item max-inline-insns-single
4741 Several parameters control the tree inliner used in gcc.
4742 This number sets the maximum number of instructions (counted in gcc's
4743 internal representation) in a single function that the tree inliner
4744 will consider for inlining. This only affects functions declared
4745 inline and methods implemented in a class declaration (C++).
4746 The default value is 500.
4748 @item max-inline-insns-auto
4749 When you use @option{-finline-functions} (included in @option{-O3}),
4750 a lot of functions that would otherwise not be considered for inlining
4751 by the compiler will be investigated. To those functions, a different
4752 (more restrictive) limit compared to functions declared inline can
4754 The default value is 150.
4756 @item max-inline-insns
4757 The tree inliner does decrease the allowable size for single functions
4758 to be inlined after we already inlined the number of instructions
4759 given here by repeated inlining. This number should be a factor of
4760 two or more larger than the single function limit.
4761 Higher numbers result in better runtime performance, but incur higher
4762 compile-time resource (CPU time, memory) requirements and result in
4763 larger binaries. Very high values are not advisable, as too large
4764 binaries may adversely affect runtime performance.
4765 The default value is 200.
4767 @item max-inline-slope
4768 After exceeding the maximum number of inlined instructions by repeated
4769 inlining, a linear function is used to decrease the allowable size
4770 for single functions. The slope of that function is the negative
4771 reciprocal of the number specified here.
4772 This parameter is ignored when @option{-funit-at-a-time} is used.
4773 The default value is 32.
4775 @item min-inline-insns
4776 The repeated inlining is throttled more and more by the linear function
4777 after exceeding the limit. To avoid too much throttling, a minimum for
4778 this function is specified here to allow repeated inlining for very small
4779 functions even when a lot of repeated inlining already has been done.
4780 This parameter is ignored when @option{-funit-at-a-time} is used.
4781 The default value is 10.
4783 @item large-function-insns
4784 The limit specifying really large functions. For functions greater than this
4785 limit inlining is constrained by @option{--param large-function-growth}.
4786 This parameter is useful primarily to avoid extreme compilation time caused by non-linear
4787 algorithms used by the backend.
4788 This parameter is ignored when @option{-funit-at-a-time} is not used.
4789 The default value is 30000.
4791 @item large-function-growth
4792 Specifies maximal growth of large functtion caused by inlining in percents.
4793 This parameter is ignored when @option{-funit-at-a-time} is not used.
4794 The default value is 200.
4796 @item inline-unit-growth
4797 Specifies maximal overall growth of the compilation unit caused by inlining.
4798 This parameter is ignored when @option{-funit-at-a-time} is not used.
4799 The default value is 150.
4801 @item max-inline-insns-rtl
4802 For languages that use the RTL inliner (this happens at a later stage
4803 than tree inlining), you can set the maximum allowable size (counted
4804 in RTL instructions) for the RTL inliner with this parameter.
4805 The default value is 600.
4808 @item max-unrolled-insns
4809 The maximum number of instructions that a loop should have if that loop
4810 is unrolled, and if the loop is unrolled, it determines how many times
4811 the loop code is unrolled.
4813 @item max-average-unrolled-insns
4814 The maximum number of instructions biased by probabilities of their execution
4815 that a loop should have if that loop is unrolled, and if the loop is unrolled,
4816 it determines how many times the loop code is unrolled.
4818 @item max-unroll-times
4819 The maximum number of unrollings of a single loop.
4821 @item max-peeled-insns
4822 The maximum number of instructions that a loop should have if that loop
4823 is peeled, and if the loop is peeled, it determines how many times
4824 the loop code is peeled.
4826 @item max-peel-times
4827 The maximum number of peelings of a single loop.
4829 @item max-completely-peeled-insns
4830 The maximum number of insns of a completely peeled loop.
4832 @item max-completely-peel-times
4833 The maximum number of iterations of a loop to be suitable for complete peeling.
4835 @item max-unswitch-insns
4836 The maximum number of insns of an unswitched loop.
4838 @item max-unswitch-level
4839 The maximum number of branches unswitched in a single loop.
4841 @item hot-bb-count-fraction
4842 Select fraction of the maximal count of repetitions of basic block in program
4843 given basic block needs to have to be considered hot.
4845 @item hot-bb-frequency-fraction
4846 Select fraction of the maximal frequency of executions of basic block in
4847 function given basic block needs to have to be considered hot
4849 @item tracer-dynamic-coverage
4850 @itemx tracer-dynamic-coverage-feedback
4852 This value is used to limit superblock formation once the given percentage of
4853 executed instructions is covered. This limits unnecessary code size
4856 The @option{tracer-dynamic-coverage-feedback} is used only when profile
4857 feedback is available. The real profiles (as opposed to statically estimated
4858 ones) are much less balanced allowing the threshold to be larger value.
4860 @item tracer-max-code-growth
4861 Stop tail duplication once code growth has reached given percentage. This is
4862 rather hokey argument, as most of the duplicates will be eliminated later in
4863 cross jumping, so it may be set to much higher values than is the desired code
4866 @item tracer-min-branch-ratio
4868 Stop reverse growth when the reverse probability of best edge is less than this
4869 threshold (in percent).
4871 @item tracer-min-branch-ratio
4872 @itemx tracer-min-branch-ratio-feedback
4874 Stop forward growth if the best edge do have probability lower than this
4877 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
4878 compilation for profile feedback and one for compilation without. The value
4879 for compilation with profile feedback needs to be more conservative (higher) in
4880 order to make tracer effective.
4882 @item max-cse-path-length
4884 Maximum number of basic blocks on path that cse considers.
4886 @item ggc-min-expand
4888 GCC uses a garbage collector to manage its own memory allocation. This
4889 parameter specifies the minimum percentage by which the garbage
4890 collector's heap should be allowed to expand between collections.
4891 Tuning this may improve compilation speed; it has no effect on code
4894 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
4895 RAM >= 1GB. If @code{getrlimit} is available, the notion of "RAM" is
4896 the smallest of actual RAM, RLIMIT_RSS, RLIMIT_DATA and RLIMIT_AS. If
4897 GCC is not able to calculate RAM on a particular platform, the lower
4898 bound of 30% is used. Setting this parameter and
4899 @option{ggc-min-heapsize} to zero causes a full collection to occur at
4900 every opportunity. This is extremely slow, but can be useful for
4903 @item ggc-min-heapsize
4905 Minimum size of the garbage collector's heap before it begins bothering
4906 to collect garbage. The first collection occurs after the heap expands
4907 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
4908 tuning this may improve compilation speed, and has no effect on code
4911 The default is RAM/8, with a lower bound of 4096 (four megabytes) and an
4912 upper bound of 131072 (128 megabytes). If @code{getrlimit} is
4913 available, the notion of "RAM" is the smallest of actual RAM,
4914 RLIMIT_RSS, RLIMIT_DATA and RLIMIT_AS. If GCC is not able to calculate
4915 RAM on a particular platform, the lower bound is used. Setting this
4916 parameter very large effectively disables garbage collection. Setting
4917 this parameter and @option{ggc-min-expand} to zero causes a full
4918 collection to occur at every opportunity.
4920 @item reorder-blocks-duplicate
4921 @itemx reorder-blocks-duplicate-feedback
4923 Used by basic block reordering pass to decide whether to use unconditional
4924 branch or duplicate the code on its destination. Code is duplicated when its
4925 estimated size is smaller than this value multiplied by the estimated size of
4926 unconditional jump in the hot spots of the program.
4928 The @option{reorder-block-duplicate-feedback} is used only when profile
4929 feedback is available and may be set to higher values than
4930 @option{reorder-block-duplicate} since information about the hot spots is more
4935 @node Preprocessor Options
4936 @section Options Controlling the Preprocessor
4937 @cindex preprocessor options
4938 @cindex options, preprocessor
4940 These options control the C preprocessor, which is run on each C source
4941 file before actual compilation.
4943 If you use the @option{-E} option, nothing is done except preprocessing.
4944 Some of these options make sense only together with @option{-E} because
4945 they cause the preprocessor output to be unsuitable for actual
4950 You can use @option{-Wp,@var{option}} to bypass the compiler driver
4951 and pass @var{option} directly through to the preprocessor. If
4952 @var{option} contains commas, it is split into multiple options at the
4953 commas. However, many options are modified, translated or interpreted
4954 by the compiler driver before being passed to the preprocessor, and
4955 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
4956 interface is undocumented and subject to change, so whenever possible
4957 you should avoid using @option{-Wp} and let the driver handle the
4960 @item -Xpreprocessor @var{option}
4961 @opindex preprocessor
4962 Pass @var{option} as an option to the preprocessor. You can use this to
4963 supply system-specific preprocessor options which GCC does not know how to
4966 If you want to pass an option that takes an argument, you must use
4967 @option{-Xpreprocessor} twice, once for the option and once for the argument.
4970 @include cppopts.texi
4972 @node Assembler Options
4973 @section Passing Options to the Assembler
4975 @c prevent bad page break with this line
4976 You can pass options to the assembler.
4979 @item -Wa,@var{option}
4981 Pass @var{option} as an option to the assembler. If @var{option}
4982 contains commas, it is split into multiple options at the commas.
4984 @item -Xassembler @var{option}
4986 Pass @var{option} as an option to the assembler. You can use this to
4987 supply system-specific assembler options which GCC does not know how to
4990 If you want to pass an option that takes an argument, you must use
4991 @option{-Xassembler} twice, once for the option and once for the argument.
4996 @section Options for Linking
4997 @cindex link options
4998 @cindex options, linking
5000 These options come into play when the compiler links object files into
5001 an executable output file. They are meaningless if the compiler is
5002 not doing a link step.
5006 @item @var{object-file-name}
5007 A file name that does not end in a special recognized suffix is
5008 considered to name an object file or library. (Object files are
5009 distinguished from libraries by the linker according to the file
5010 contents.) If linking is done, these object files are used as input
5019 If any of these options is used, then the linker is not run, and
5020 object file names should not be used as arguments. @xref{Overall
5024 @item -l@var{library}
5025 @itemx -l @var{library}
5027 Search the library named @var{library} when linking. (The second
5028 alternative with the library as a separate argument is only for
5029 POSIX compliance and is not recommended.)
5031 It makes a difference where in the command you write this option; the
5032 linker searches and processes libraries and object files in the order they
5033 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
5034 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
5035 to functions in @samp{z}, those functions may not be loaded.
5037 The linker searches a standard list of directories for the library,
5038 which is actually a file named @file{lib@var{library}.a}. The linker
5039 then uses this file as if it had been specified precisely by name.
5041 The directories searched include several standard system directories
5042 plus any that you specify with @option{-L}.
5044 Normally the files found this way are library files---archive files
5045 whose members are object files. The linker handles an archive file by
5046 scanning through it for members which define symbols that have so far
5047 been referenced but not defined. But if the file that is found is an
5048 ordinary object file, it is linked in the usual fashion. The only
5049 difference between using an @option{-l} option and specifying a file name
5050 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
5051 and searches several directories.
5055 You need this special case of the @option{-l} option in order to
5056 link an Objective-C program.
5059 @opindex nostartfiles
5060 Do not use the standard system startup files when linking.
5061 The standard system libraries are used normally, unless @option{-nostdlib}
5062 or @option{-nodefaultlibs} is used.
5064 @item -nodefaultlibs
5065 @opindex nodefaultlibs
5066 Do not use the standard system libraries when linking.
5067 Only the libraries you specify will be passed to the linker.
5068 The standard startup files are used normally, unless @option{-nostartfiles}
5069 is used. The compiler may generate calls to memcmp, memset, and memcpy
5070 for System V (and ISO C) environments or to bcopy and bzero for
5071 BSD environments. These entries are usually resolved by entries in
5072 libc. These entry points should be supplied through some other
5073 mechanism when this option is specified.
5077 Do not use the standard system startup files or libraries when linking.
5078 No startup files and only the libraries you specify will be passed to
5079 the linker. The compiler may generate calls to memcmp, memset, and memcpy
5080 for System V (and ISO C) environments or to bcopy and bzero for
5081 BSD environments. These entries are usually resolved by entries in
5082 libc. These entry points should be supplied through some other
5083 mechanism when this option is specified.
5085 @cindex @option{-lgcc}, use with @option{-nostdlib}
5086 @cindex @option{-nostdlib} and unresolved references
5087 @cindex unresolved references and @option{-nostdlib}
5088 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
5089 @cindex @option{-nodefaultlibs} and unresolved references
5090 @cindex unresolved references and @option{-nodefaultlibs}
5091 One of the standard libraries bypassed by @option{-nostdlib} and
5092 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
5093 that GCC uses to overcome shortcomings of particular machines, or special
5094 needs for some languages.
5095 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
5096 Collection (GCC) Internals},
5097 for more discussion of @file{libgcc.a}.)
5098 In most cases, you need @file{libgcc.a} even when you want to avoid
5099 other standard libraries. In other words, when you specify @option{-nostdlib}
5100 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
5101 This ensures that you have no unresolved references to internal GCC
5102 library subroutines. (For example, @samp{__main}, used to ensure C++
5103 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
5104 GNU Compiler Collection (GCC) Internals}.)
5108 Produce a position independent executable on targets which support it.
5109 For predictable results, you must also specify the same set of options
5110 that were used to generate code (@option{-fpie}, @option{-fPIE},
5111 or model suboptions) when you specify this option.
5115 Remove all symbol table and relocation information from the executable.
5119 On systems that support dynamic linking, this prevents linking with the shared
5120 libraries. On other systems, this option has no effect.
5124 Produce a shared object which can then be linked with other objects to
5125 form an executable. Not all systems support this option. For predictable
5126 results, you must also specify the same set of options that were used to
5127 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
5128 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
5129 needs to build supplementary stub code for constructors to work. On
5130 multi-libbed systems, @samp{gcc -shared} must select the correct support
5131 libraries to link against. Failing to supply the correct flags may lead
5132 to subtle defects. Supplying them in cases where they are not necessary
5135 @item -shared-libgcc
5136 @itemx -static-libgcc
5137 @opindex shared-libgcc
5138 @opindex static-libgcc
5139 On systems that provide @file{libgcc} as a shared library, these options
5140 force the use of either the shared or static version respectively.
5141 If no shared version of @file{libgcc} was built when the compiler was
5142 configured, these options have no effect.
5144 There are several situations in which an application should use the
5145 shared @file{libgcc} instead of the static version. The most common
5146 of these is when the application wishes to throw and catch exceptions
5147 across different shared libraries. In that case, each of the libraries
5148 as well as the application itself should use the shared @file{libgcc}.
5150 Therefore, the G++ and GCJ drivers automatically add
5151 @option{-shared-libgcc} whenever you build a shared library or a main
5152 executable, because C++ and Java programs typically use exceptions, so
5153 this is the right thing to do.
5155 If, instead, you use the GCC driver to create shared libraries, you may
5156 find that they will not always be linked with the shared @file{libgcc}.
5157 If GCC finds, at its configuration time, that you have a GNU linker that
5158 does not support option @option{--eh-frame-hdr}, it will link the shared
5159 version of @file{libgcc} into shared libraries by default. Otherwise,
5160 it will take advantage of the linker and optimize away the linking with
5161 the shared version of @file{libgcc}, linking with the static version of
5162 libgcc by default. This allows exceptions to propagate through such
5163 shared libraries, without incurring relocation costs at library load
5166 However, if a library or main executable is supposed to throw or catch
5167 exceptions, you must link it using the G++ or GCJ driver, as appropriate
5168 for the languages used in the program, or using the option
5169 @option{-shared-libgcc}, such that it is linked with the shared
5174 Bind references to global symbols when building a shared object. Warn
5175 about any unresolved references (unless overridden by the link editor
5176 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
5179 @item -Xlinker @var{option}
5181 Pass @var{option} as an option to the linker. You can use this to
5182 supply system-specific linker options which GCC does not know how to
5185 If you want to pass an option that takes an argument, you must use
5186 @option{-Xlinker} twice, once for the option and once for the argument.
5187 For example, to pass @option{-assert definitions}, you must write
5188 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
5189 @option{-Xlinker "-assert definitions"}, because this passes the entire
5190 string as a single argument, which is not what the linker expects.
5192 @item -Wl,@var{option}
5194 Pass @var{option} as an option to the linker. If @var{option} contains
5195 commas, it is split into multiple options at the commas.
5197 @item -u @var{symbol}
5199 Pretend the symbol @var{symbol} is undefined, to force linking of
5200 library modules to define it. You can use @option{-u} multiple times with
5201 different symbols to force loading of additional library modules.
5204 @node Directory Options
5205 @section Options for Directory Search
5206 @cindex directory options
5207 @cindex options, directory search
5210 These options specify directories to search for header files, for
5211 libraries and for parts of the compiler:
5216 Add the directory @var{dir} to the head of the list of directories to be
5217 searched for header files. This can be used to override a system header
5218 file, substituting your own version, since these directories are
5219 searched before the system header file directories. However, you should
5220 not use this option to add directories that contain vendor-supplied
5221 system header files (use @option{-isystem} for that). If you use more than
5222 one @option{-I} option, the directories are scanned in left-to-right
5223 order; the standard system directories come after.
5225 If a standard system include directory, or a directory specified with
5226 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
5227 option will be ignored. The directory will still be searched but as a
5228 system directory at its normal position in the system include chain.
5229 This is to ensure that GCC's procedure to fix buggy system headers and
5230 the ordering for the include_next directive are not inadvertently changed.
5231 If you really need to change the search order for system directories,
5232 use the @option{-nostdinc} and/or @option{-isystem} options.
5236 Any directories you specify with @option{-I} options before the @option{-I-}
5237 option are searched only for the case of @samp{#include "@var{file}"};
5238 they are not searched for @samp{#include <@var{file}>}.
5240 If additional directories are specified with @option{-I} options after
5241 the @option{-I-}, these directories are searched for all @samp{#include}
5242 directives. (Ordinarily @emph{all} @option{-I} directories are used
5245 In addition, the @option{-I-} option inhibits the use of the current
5246 directory (where the current input file came from) as the first search
5247 directory for @samp{#include "@var{file}"}. There is no way to
5248 override this effect of @option{-I-}. With @option{-I.} you can specify
5249 searching the directory which was current when the compiler was
5250 invoked. That is not exactly the same as what the preprocessor does
5251 by default, but it is often satisfactory.
5253 @option{-I-} does not inhibit the use of the standard system directories
5254 for header files. Thus, @option{-I-} and @option{-nostdinc} are
5259 Add directory @var{dir} to the list of directories to be searched
5262 @item -B@var{prefix}
5264 This option specifies where to find the executables, libraries,
5265 include files, and data files of the compiler itself.
5267 The compiler driver program runs one or more of the subprograms
5268 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
5269 @var{prefix} as a prefix for each program it tries to run, both with and
5270 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
5272 For each subprogram to be run, the compiler driver first tries the
5273 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
5274 was not specified, the driver tries two standard prefixes, which are
5275 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
5276 those results in a file name that is found, the unmodified program
5277 name is searched for using the directories specified in your
5278 @env{PATH} environment variable.
5280 The compiler will check to see if the path provided by the @option{-B}
5281 refers to a directory, and if necessary it will add a directory
5282 separator character at the end of the path.
5284 @option{-B} prefixes that effectively specify directory names also apply
5285 to libraries in the linker, because the compiler translates these
5286 options into @option{-L} options for the linker. They also apply to
5287 includes files in the preprocessor, because the compiler translates these
5288 options into @option{-isystem} options for the preprocessor. In this case,
5289 the compiler appends @samp{include} to the prefix.
5291 The run-time support file @file{libgcc.a} can also be searched for using
5292 the @option{-B} prefix, if needed. If it is not found there, the two
5293 standard prefixes above are tried, and that is all. The file is left
5294 out of the link if it is not found by those means.
5296 Another way to specify a prefix much like the @option{-B} prefix is to use
5297 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
5300 As a special kludge, if the path provided by @option{-B} is
5301 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
5302 9, then it will be replaced by @file{[dir/]include}. This is to help
5303 with boot-strapping the compiler.
5305 @item -specs=@var{file}
5307 Process @var{file} after the compiler reads in the standard @file{specs}
5308 file, in order to override the defaults that the @file{gcc} driver
5309 program uses when determining what switches to pass to @file{cc1},
5310 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
5311 @option{-specs=@var{file}} can be specified on the command line, and they
5312 are processed in order, from left to right.
5318 @section Specifying subprocesses and the switches to pass to them
5321 @command{gcc} is a driver program. It performs its job by invoking a
5322 sequence of other programs to do the work of compiling, assembling and
5323 linking. GCC interprets its command-line parameters and uses these to
5324 deduce which programs it should invoke, and which command-line options
5325 it ought to place on their command lines. This behavior is controlled
5326 by @dfn{spec strings}. In most cases there is one spec string for each
5327 program that GCC can invoke, but a few programs have multiple spec
5328 strings to control their behavior. The spec strings built into GCC can
5329 be overridden by using the @option{-specs=} command-line switch to specify
5332 @dfn{Spec files} are plaintext files that are used to construct spec
5333 strings. They consist of a sequence of directives separated by blank
5334 lines. The type of directive is determined by the first non-whitespace
5335 character on the line and it can be one of the following:
5338 @item %@var{command}
5339 Issues a @var{command} to the spec file processor. The commands that can
5343 @item %include <@var{file}>
5345 Search for @var{file} and insert its text at the current point in the
5348 @item %include_noerr <@var{file}>
5349 @cindex %include_noerr
5350 Just like @samp{%include}, but do not generate an error message if the include
5351 file cannot be found.
5353 @item %rename @var{old_name} @var{new_name}
5355 Rename the spec string @var{old_name} to @var{new_name}.
5359 @item *[@var{spec_name}]:
5360 This tells the compiler to create, override or delete the named spec
5361 string. All lines after this directive up to the next directive or
5362 blank line are considered to be the text for the spec string. If this
5363 results in an empty string then the spec will be deleted. (Or, if the
5364 spec did not exist, then nothing will happened.) Otherwise, if the spec
5365 does not currently exist a new spec will be created. If the spec does
5366 exist then its contents will be overridden by the text of this
5367 directive, unless the first character of that text is the @samp{+}
5368 character, in which case the text will be appended to the spec.
5370 @item [@var{suffix}]:
5371 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
5372 and up to the next directive or blank line are considered to make up the
5373 spec string for the indicated suffix. When the compiler encounters an
5374 input file with the named suffix, it will processes the spec string in
5375 order to work out how to compile that file. For example:
5382 This says that any input file whose name ends in @samp{.ZZ} should be
5383 passed to the program @samp{z-compile}, which should be invoked with the
5384 command-line switch @option{-input} and with the result of performing the
5385 @samp{%i} substitution. (See below.)
5387 As an alternative to providing a spec string, the text that follows a
5388 suffix directive can be one of the following:
5391 @item @@@var{language}
5392 This says that the suffix is an alias for a known @var{language}. This is
5393 similar to using the @option{-x} command-line switch to GCC to specify a
5394 language explicitly. For example:
5401 Says that .ZZ files are, in fact, C++ source files.
5404 This causes an error messages saying:
5407 @var{name} compiler not installed on this system.
5411 GCC already has an extensive list of suffixes built into it.
5412 This directive will add an entry to the end of the list of suffixes, but
5413 since the list is searched from the end backwards, it is effectively
5414 possible to override earlier entries using this technique.
5418 GCC has the following spec strings built into it. Spec files can
5419 override these strings or create their own. Note that individual
5420 targets can also add their own spec strings to this list.
5423 asm Options to pass to the assembler
5424 asm_final Options to pass to the assembler post-processor
5425 cpp Options to pass to the C preprocessor
5426 cc1 Options to pass to the C compiler
5427 cc1plus Options to pass to the C++ compiler
5428 endfile Object files to include at the end of the link
5429 link Options to pass to the linker
5430 lib Libraries to include on the command line to the linker
5431 libgcc Decides which GCC support library to pass to the linker
5432 linker Sets the name of the linker
5433 predefines Defines to be passed to the C preprocessor
5434 signed_char Defines to pass to CPP to say whether @code{char} is signed
5436 startfile Object files to include at the start of the link
5439 Here is a small example of a spec file:
5445 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
5448 This example renames the spec called @samp{lib} to @samp{old_lib} and
5449 then overrides the previous definition of @samp{lib} with a new one.
5450 The new definition adds in some extra command-line options before
5451 including the text of the old definition.
5453 @dfn{Spec strings} are a list of command-line options to be passed to their
5454 corresponding program. In addition, the spec strings can contain
5455 @samp{%}-prefixed sequences to substitute variable text or to
5456 conditionally insert text into the command line. Using these constructs
5457 it is possible to generate quite complex command lines.
5459 Here is a table of all defined @samp{%}-sequences for spec
5460 strings. Note that spaces are not generated automatically around the
5461 results of expanding these sequences. Therefore you can concatenate them
5462 together or combine them with constant text in a single argument.
5466 Substitute one @samp{%} into the program name or argument.
5469 Substitute the name of the input file being processed.
5472 Substitute the basename of the input file being processed.
5473 This is the substring up to (and not including) the last period
5474 and not including the directory.
5477 This is the same as @samp{%b}, but include the file suffix (text after
5481 Marks the argument containing or following the @samp{%d} as a
5482 temporary file name, so that that file will be deleted if GCC exits
5483 successfully. Unlike @samp{%g}, this contributes no text to the
5486 @item %g@var{suffix}
5487 Substitute a file name that has suffix @var{suffix} and is chosen
5488 once per compilation, and mark the argument in the same way as
5489 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
5490 name is now chosen in a way that is hard to predict even when previously
5491 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
5492 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
5493 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
5494 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
5495 was simply substituted with a file name chosen once per compilation,
5496 without regard to any appended suffix (which was therefore treated
5497 just like ordinary text), making such attacks more likely to succeed.
5499 @item %u@var{suffix}
5500 Like @samp{%g}, but generates a new temporary file name even if
5501 @samp{%u@var{suffix}} was already seen.
5503 @item %U@var{suffix}
5504 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
5505 new one if there is no such last file name. In the absence of any
5506 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
5507 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
5508 would involve the generation of two distinct file names, one
5509 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
5510 simply substituted with a file name chosen for the previous @samp{%u},
5511 without regard to any appended suffix.
5513 @item %j@var{suffix}
5514 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
5515 writable, and if save-temps is off; otherwise, substitute the name
5516 of a temporary file, just like @samp{%u}. This temporary file is not
5517 meant for communication between processes, but rather as a junk
5520 @item %|@var{suffix}
5521 @itemx %m@var{suffix}
5522 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
5523 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
5524 all. These are the two most common ways to instruct a program that it
5525 should read from standard input or write to standard output. If you
5526 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
5527 construct: see for example @file{f/lang-specs.h}.
5529 @item %.@var{SUFFIX}
5530 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
5531 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
5532 terminated by the next space or %.
5535 Marks the argument containing or following the @samp{%w} as the
5536 designated output file of this compilation. This puts the argument
5537 into the sequence of arguments that @samp{%o} will substitute later.
5540 Substitutes the names of all the output files, with spaces
5541 automatically placed around them. You should write spaces
5542 around the @samp{%o} as well or the results are undefined.
5543 @samp{%o} is for use in the specs for running the linker.
5544 Input files whose names have no recognized suffix are not compiled
5545 at all, but they are included among the output files, so they will
5549 Substitutes the suffix for object files. Note that this is
5550 handled specially when it immediately follows @samp{%g, %u, or %U},
5551 because of the need for those to form complete file names. The
5552 handling is such that @samp{%O} is treated exactly as if it had already
5553 been substituted, except that @samp{%g, %u, and %U} do not currently
5554 support additional @var{suffix} characters following @samp{%O} as they would
5555 following, for example, @samp{.o}.
5558 Substitutes the standard macro predefinitions for the
5559 current target machine. Use this when running @code{cpp}.
5562 Like @samp{%p}, but puts @samp{__} before and after the name of each
5563 predefined macro, except for macros that start with @samp{__} or with
5564 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
5568 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
5569 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}), and
5570 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
5574 Current argument is the name of a library or startup file of some sort.
5575 Search for that file in a standard list of directories and substitute
5576 the full name found.
5579 Print @var{str} as an error message. @var{str} is terminated by a newline.
5580 Use this when inconsistent options are detected.
5583 Substitute the contents of spec string @var{name} at this point.
5586 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
5588 @item %x@{@var{option}@}
5589 Accumulate an option for @samp{%X}.
5592 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
5596 Output the accumulated assembler options specified by @option{-Wa}.
5599 Output the accumulated preprocessor options specified by @option{-Wp}.
5602 Process the @code{asm} spec. This is used to compute the
5603 switches to be passed to the assembler.
5606 Process the @code{asm_final} spec. This is a spec string for
5607 passing switches to an assembler post-processor, if such a program is
5611 Process the @code{link} spec. This is the spec for computing the
5612 command line passed to the linker. Typically it will make use of the
5613 @samp{%L %G %S %D and %E} sequences.
5616 Dump out a @option{-L} option for each directory that GCC believes might
5617 contain startup files. If the target supports multilibs then the
5618 current multilib directory will be prepended to each of these paths.
5621 Output the multilib directory with directory separators replaced with
5622 @samp{_}. If multilib directories are not set, or the multilib directory is
5623 @file{.} then this option emits nothing.
5626 Process the @code{lib} spec. This is a spec string for deciding which
5627 libraries should be included on the command line to the linker.
5630 Process the @code{libgcc} spec. This is a spec string for deciding
5631 which GCC support library should be included on the command line to the linker.
5634 Process the @code{startfile} spec. This is a spec for deciding which
5635 object files should be the first ones passed to the linker. Typically
5636 this might be a file named @file{crt0.o}.
5639 Process the @code{endfile} spec. This is a spec string that specifies
5640 the last object files that will be passed to the linker.
5643 Process the @code{cpp} spec. This is used to construct the arguments
5644 to be passed to the C preprocessor.
5647 Process the @code{signed_char} spec. This is intended to be used
5648 to tell cpp whether a char is signed. It typically has the definition:
5650 %@{funsigned-char:-D__CHAR_UNSIGNED__@}
5654 Process the @code{cc1} spec. This is used to construct the options to be
5655 passed to the actual C compiler (@samp{cc1}).
5658 Process the @code{cc1plus} spec. This is used to construct the options to be
5659 passed to the actual C++ compiler (@samp{cc1plus}).
5662 Substitute the variable part of a matched option. See below.
5663 Note that each comma in the substituted string is replaced by
5667 Remove all occurrences of @code{-S} from the command line. Note---this
5668 command is position dependent. @samp{%} commands in the spec string
5669 before this one will see @code{-S}, @samp{%} commands in the spec string
5670 after this one will not.
5672 @item %:@var{function}(@var{args})
5673 Call the named function @var{function}, passing it @var{args}.
5674 @var{args} is first processed as a nested spec string, then split
5675 into an argument vector in the usual fashion. The function returns
5676 a string which is processed as if it had appeared literally as part
5677 of the current spec.
5679 The following built-in spec functions are provided:
5682 @item @code{if-exists}
5683 The @code{if-exists} spec function takes one argument, an absolute
5684 pathname to a file. If the file exists, @code{if-exists} returns the
5685 pathname. Here is a small example of its usage:
5689 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
5692 @item @code{if-exists-else}
5693 The @code{if-exists-else} spec function is similar to the @code{if-exists}
5694 spec function, except that it takes two arguments. The first argument is
5695 an absolute pathname to a file. If the file exists, @code{if-exists-else}
5696 returns the pathname. If it does not exist, it returns the second argument.
5697 This way, @code{if-exists-else} can be used to select one file or another,
5698 based on the existence of the first. Here is a small example of its usage:
5702 crt0%O%s %:if-exists(crti%O%s) \
5703 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
5708 Substitutes the @code{-S} switch, if that switch was given to GCC@.
5709 If that switch was not specified, this substitutes nothing. Note that
5710 the leading dash is omitted when specifying this option, and it is
5711 automatically inserted if the substitution is performed. Thus the spec
5712 string @samp{%@{foo@}} would match the command-line option @option{-foo}
5713 and would output the command line option @option{-foo}.
5715 @item %W@{@code{S}@}
5716 Like %@{@code{S}@} but mark last argument supplied within as a file to be
5719 @item %@{@code{S}*@}
5720 Substitutes all the switches specified to GCC whose names start
5721 with @code{-S}, but which also take an argument. This is used for
5722 switches like @option{-o}, @option{-D}, @option{-I}, etc.
5723 GCC considers @option{-o foo} as being
5724 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
5725 text, including the space. Thus two arguments would be generated.
5727 @item %@{@code{S}*&@code{T}*@}
5728 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
5729 (the order of @code{S} and @code{T} in the spec is not significant).
5730 There can be any number of ampersand-separated variables; for each the
5731 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
5733 @item %@{@code{S}:@code{X}@}
5734 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
5736 @item %@{!@code{S}:@code{X}@}
5737 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
5739 @item %@{@code{S}*:@code{X}@}
5740 Substitutes @code{X} if one or more switches whose names start with
5741 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
5742 once, no matter how many such switches appeared. However, if @code{%*}
5743 appears somewhere in @code{X}, then @code{X} will be substituted once
5744 for each matching switch, with the @code{%*} replaced by the part of
5745 that switch that matched the @code{*}.
5747 @item %@{.@code{S}:@code{X}@}
5748 Substitutes @code{X}, if processing a file with suffix @code{S}.
5750 @item %@{!.@code{S}:@code{X}@}
5751 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
5753 @item %@{@code{S}|@code{P}:@code{X}@}
5754 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC@.
5755 This may be combined with @samp{!}, @samp{.}, and @code{*} sequences as well,
5756 although they have a stronger binding than the @samp{|}. If @code{%*}
5757 appears in @code{X}, all of the alternatives must be starred, and only
5758 the first matching alternative is substituted.
5760 For example, a spec string like this:
5763 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
5766 will output the following command-line options from the following input
5767 command-line options:
5772 -d fred.c -foo -baz -boggle
5773 -d jim.d -bar -baz -boggle
5776 @item %@{S:X; T:Y; :D@}
5778 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
5779 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
5780 be as many clauses as you need. This may be combined with @code{.},
5781 @code{!}, @code{|}, and @code{*} as needed.
5786 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
5787 construct may contain other nested @samp{%} constructs or spaces, or
5788 even newlines. They are processed as usual, as described above.
5789 Trailing white space in @code{X} is ignored. White space may also
5790 appear anywhere on the left side of the colon in these constructs,
5791 except between @code{.} or @code{*} and the corresponding word.
5793 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
5794 handled specifically in these constructs. If another value of
5795 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
5796 @option{-W} switch is found later in the command line, the earlier
5797 switch value is ignored, except with @{@code{S}*@} where @code{S} is
5798 just one letter, which passes all matching options.
5800 The character @samp{|} at the beginning of the predicate text is used to
5801 indicate that a command should be piped to the following command, but
5802 only if @option{-pipe} is specified.
5804 It is built into GCC which switches take arguments and which do not.
5805 (You might think it would be useful to generalize this to allow each
5806 compiler's spec to say which switches take arguments. But this cannot
5807 be done in a consistent fashion. GCC cannot even decide which input
5808 files have been specified without knowing which switches take arguments,
5809 and it must know which input files to compile in order to tell which
5812 GCC also knows implicitly that arguments starting in @option{-l} are to be
5813 treated as compiler output files, and passed to the linker in their
5814 proper position among the other output files.
5816 @c man begin OPTIONS
5818 @node Target Options
5819 @section Specifying Target Machine and Compiler Version
5820 @cindex target options
5821 @cindex cross compiling
5822 @cindex specifying machine version
5823 @cindex specifying compiler version and target machine
5824 @cindex compiler version, specifying
5825 @cindex target machine, specifying
5827 The usual way to run GCC is to run the executable called @file{gcc}, or
5828 @file{<machine>-gcc} when cross-compiling, or
5829 @file{<machine>-gcc-<version>} to run a version other than the one that
5830 was installed last. Sometimes this is inconvenient, so GCC provides
5831 options that will switch to another cross-compiler or version.
5834 @item -b @var{machine}
5836 The argument @var{machine} specifies the target machine for compilation.
5838 The value to use for @var{machine} is the same as was specified as the
5839 machine type when configuring GCC as a cross-compiler. For
5840 example, if a cross-compiler was configured with @samp{configure
5841 i386v}, meaning to compile for an 80386 running System V, then you
5842 would specify @option{-b i386v} to run that cross compiler.
5844 @item -V @var{version}
5846 The argument @var{version} specifies which version of GCC to run.
5847 This is useful when multiple versions are installed. For example,
5848 @var{version} might be @samp{2.0}, meaning to run GCC version 2.0.
5851 The @option{-V} and @option{-b} options work by running the
5852 @file{<machine>-gcc-<version>} executable, so there's no real reason to
5853 use them if you can just run that directly.
5855 @node Submodel Options
5856 @section Hardware Models and Configurations
5857 @cindex submodel options
5858 @cindex specifying hardware config
5859 @cindex hardware models and configurations, specifying
5860 @cindex machine dependent options
5862 Earlier we discussed the standard option @option{-b} which chooses among
5863 different installed compilers for completely different target
5864 machines, such as VAX vs.@: 68000 vs.@: 80386.
5866 In addition, each of these target machine types can have its own
5867 special options, starting with @samp{-m}, to choose among various
5868 hardware models or configurations---for example, 68010 vs 68020,
5869 floating coprocessor or none. A single installed version of the
5870 compiler can compile for any model or configuration, according to the
5873 Some configurations of the compiler also support additional special
5874 options, usually for compatibility with other compilers on the same
5877 These options are defined by the macro @code{TARGET_SWITCHES} in the
5878 machine description. The default for the options is also defined by
5879 that macro, which enables you to change the defaults.
5891 * RS/6000 and PowerPC Options::
5895 * i386 and x86-64 Options::
5897 * Intel 960 Options::
5898 * DEC Alpha Options::
5899 * DEC Alpha/VMS Options::
5902 * System V Options::
5903 * TMS320C3x/C4x Options::
5911 * S/390 and zSeries Options::
5915 * Xstormy16 Options::
5920 @node M680x0 Options
5921 @subsection M680x0 Options
5922 @cindex M680x0 options
5924 These are the @samp{-m} options defined for the 68000 series. The default
5925 values for these options depends on which style of 68000 was selected when
5926 the compiler was configured; the defaults for the most common choices are
5934 Generate output for a 68000. This is the default
5935 when the compiler is configured for 68000-based systems.
5937 Use this option for microcontrollers with a 68000 or EC000 core,
5938 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
5944 Generate output for a 68020. This is the default
5945 when the compiler is configured for 68020-based systems.
5949 Generate output containing 68881 instructions for floating point.
5950 This is the default for most 68020 systems unless @option{--nfp} was
5951 specified when the compiler was configured.
5955 Generate output for a 68030. This is the default when the compiler is
5956 configured for 68030-based systems.
5960 Generate output for a 68040. This is the default when the compiler is
5961 configured for 68040-based systems.
5963 This option inhibits the use of 68881/68882 instructions that have to be
5964 emulated by software on the 68040. Use this option if your 68040 does not
5965 have code to emulate those instructions.
5969 Generate output for a 68060. This is the default when the compiler is
5970 configured for 68060-based systems.
5972 This option inhibits the use of 68020 and 68881/68882 instructions that
5973 have to be emulated by software on the 68060. Use this option if your 68060
5974 does not have code to emulate those instructions.
5978 Generate output for a CPU32. This is the default
5979 when the compiler is configured for CPU32-based systems.
5981 Use this option for microcontrollers with a
5982 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
5983 68336, 68340, 68341, 68349 and 68360.
5987 Generate output for a 520X ``coldfire'' family cpu. This is the default
5988 when the compiler is configured for 520X-based systems.
5990 Use this option for microcontroller with a 5200 core, including
5991 the MCF5202, MCF5203, MCF5204 and MCF5202.
5996 Generate output for a 68040, without using any of the new instructions.
5997 This results in code which can run relatively efficiently on either a
5998 68020/68881 or a 68030 or a 68040. The generated code does use the
5999 68881 instructions that are emulated on the 68040.
6003 Generate output for a 68060, without using any of the new instructions.
6004 This results in code which can run relatively efficiently on either a
6005 68020/68881 or a 68030 or a 68040. The generated code does use the
6006 68881 instructions that are emulated on the 68060.
6009 @opindex msoft-float
6010 Generate output containing library calls for floating point.
6011 @strong{Warning:} the requisite libraries are not available for all m68k
6012 targets. Normally the facilities of the machine's usual C compiler are
6013 used, but this can't be done directly in cross-compilation. You must
6014 make your own arrangements to provide suitable library functions for
6015 cross-compilation. The embedded targets @samp{m68k-*-aout} and
6016 @samp{m68k-*-coff} do provide software floating point support.
6020 Consider type @code{int} to be 16 bits wide, like @code{short int}.
6023 @opindex mnobitfield
6024 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
6025 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
6029 Do use the bit-field instructions. The @option{-m68020} option implies
6030 @option{-mbitfield}. This is the default if you use a configuration
6031 designed for a 68020.
6035 Use a different function-calling convention, in which functions
6036 that take a fixed number of arguments return with the @code{rtd}
6037 instruction, which pops their arguments while returning. This
6038 saves one instruction in the caller since there is no need to pop
6039 the arguments there.
6041 This calling convention is incompatible with the one normally
6042 used on Unix, so you cannot use it if you need to call libraries
6043 compiled with the Unix compiler.
6045 Also, you must provide function prototypes for all functions that
6046 take variable numbers of arguments (including @code{printf});
6047 otherwise incorrect code will be generated for calls to those
6050 In addition, seriously incorrect code will result if you call a
6051 function with too many arguments. (Normally, extra arguments are
6052 harmlessly ignored.)
6054 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
6055 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
6058 @itemx -mno-align-int
6060 @opindex mno-align-int
6061 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
6062 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
6063 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
6064 Aligning variables on 32-bit boundaries produces code that runs somewhat
6065 faster on processors with 32-bit busses at the expense of more memory.
6067 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
6068 align structures containing the above types differently than
6069 most published application binary interface specifications for the m68k.
6073 Use the pc-relative addressing mode of the 68000 directly, instead of
6074 using a global offset table. At present, this option implies @option{-fpic},
6075 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
6076 not presently supported with @option{-mpcrel}, though this could be supported for
6077 68020 and higher processors.
6079 @item -mno-strict-align
6080 @itemx -mstrict-align
6081 @opindex mno-strict-align
6082 @opindex mstrict-align
6083 Do not (do) assume that unaligned memory references will be handled by
6088 @node M68hc1x Options
6089 @subsection M68hc1x Options
6090 @cindex M68hc1x options
6092 These are the @samp{-m} options defined for the 68hc11 and 68hc12
6093 microcontrollers. The default values for these options depends on
6094 which style of microcontroller was selected when the compiler was configured;
6095 the defaults for the most common choices are given below.
6102 Generate output for a 68HC11. This is the default
6103 when the compiler is configured for 68HC11-based systems.
6109 Generate output for a 68HC12. This is the default
6110 when the compiler is configured for 68HC12-based systems.
6116 Generate output for a 68HCS12.
6119 @opindex mauto-incdec
6120 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
6127 Enable the use of 68HC12 min and max instructions.
6130 @itemx -mno-long-calls
6131 @opindex mlong-calls
6132 @opindex mno-long-calls
6133 Treat all calls as being far away (near). If calls are assumed to be
6134 far away, the compiler will use the @code{call} instruction to
6135 call a function and the @code{rtc} instruction for returning.
6139 Consider type @code{int} to be 16 bits wide, like @code{short int}.
6141 @item -msoft-reg-count=@var{count}
6142 @opindex msoft-reg-count
6143 Specify the number of pseudo-soft registers which are used for the
6144 code generation. The maximum number is 32. Using more pseudo-soft
6145 register may or may not result in better code depending on the program.
6146 The default is 4 for 68HC11 and 2 for 68HC12.
6151 @subsection VAX Options
6154 These @samp{-m} options are defined for the VAX:
6159 Do not output certain jump instructions (@code{aobleq} and so on)
6160 that the Unix assembler for the VAX cannot handle across long
6165 Do output those jump instructions, on the assumption that you
6166 will assemble with the GNU assembler.
6170 Output code for g-format floating point numbers instead of d-format.
6174 @subsection SPARC Options
6175 @cindex SPARC options
6177 These @samp{-m} switches are supported on the SPARC:
6182 @opindex mno-app-regs
6184 Specify @option{-mapp-regs} to generate output using the global registers
6185 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
6188 To be fully SVR4 ABI compliant at the cost of some performance loss,
6189 specify @option{-mno-app-regs}. You should compile libraries and system
6190 software with this option.
6195 @opindex mhard-float
6196 Generate output containing floating point instructions. This is the
6202 @opindex msoft-float
6203 Generate output containing library calls for floating point.
6204 @strong{Warning:} the requisite libraries are not available for all SPARC
6205 targets. Normally the facilities of the machine's usual C compiler are
6206 used, but this cannot be done directly in cross-compilation. You must make
6207 your own arrangements to provide suitable library functions for
6208 cross-compilation. The embedded targets @samp{sparc-*-aout} and
6209 @samp{sparclite-*-*} do provide software floating point support.
6211 @option{-msoft-float} changes the calling convention in the output file;
6212 therefore, it is only useful if you compile @emph{all} of a program with
6213 this option. In particular, you need to compile @file{libgcc.a}, the
6214 library that comes with GCC, with @option{-msoft-float} in order for
6217 @item -mhard-quad-float
6218 @opindex mhard-quad-float
6219 Generate output containing quad-word (long double) floating point
6222 @item -msoft-quad-float
6223 @opindex msoft-quad-float
6224 Generate output containing library calls for quad-word (long double)
6225 floating point instructions. The functions called are those specified
6226 in the SPARC ABI@. This is the default.
6228 As of this writing, there are no sparc implementations that have hardware
6229 support for the quad-word floating point instructions. They all invoke
6230 a trap handler for one of these instructions, and then the trap handler
6231 emulates the effect of the instruction. Because of the trap handler overhead,
6232 this is much slower than calling the ABI library routines. Thus the
6233 @option{-msoft-quad-float} option is the default.
6239 With @option{-mflat}, the compiler does not generate save/restore instructions
6240 and will use a ``flat'' or single register window calling convention.
6241 This model uses %i7 as the frame pointer and is compatible with the normal
6242 register window model. Code from either may be intermixed.
6243 The local registers and the input registers (0--5) are still treated as
6244 ``call saved'' registers and will be saved on the stack as necessary.
6246 With @option{-mno-flat} (the default), the compiler emits save/restore
6247 instructions (except for leaf functions) and is the normal mode of operation.
6249 @item -mno-unaligned-doubles
6250 @itemx -munaligned-doubles
6251 @opindex mno-unaligned-doubles
6252 @opindex munaligned-doubles
6253 Assume that doubles have 8 byte alignment. This is the default.
6255 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
6256 alignment only if they are contained in another type, or if they have an
6257 absolute address. Otherwise, it assumes they have 4 byte alignment.
6258 Specifying this option avoids some rare compatibility problems with code
6259 generated by other compilers. It is not the default because it results
6260 in a performance loss, especially for floating point code.
6262 @item -mno-faster-structs
6263 @itemx -mfaster-structs
6264 @opindex mno-faster-structs
6265 @opindex mfaster-structs
6266 With @option{-mfaster-structs}, the compiler assumes that structures
6267 should have 8 byte alignment. This enables the use of pairs of
6268 @code{ldd} and @code{std} instructions for copies in structure
6269 assignment, in place of twice as many @code{ld} and @code{st} pairs.
6270 However, the use of this changed alignment directly violates the SPARC
6271 ABI@. Thus, it's intended only for use on targets where the developer
6272 acknowledges that their resulting code will not be directly in line with
6273 the rules of the ABI@.
6276 @opindex mimpure-text
6277 @option{-mimpure-text}, used in addition to @option{-shared}, tells
6278 the compiler to not pass @option{-z text} to the linker when linking a
6279 shared object. Using this option, you can link position-dependent
6280 code into a shared object.
6282 @option{-mimpure-text} suppresses the ``relocations remain against
6283 allocatable but non-writable sections'' linker error message.
6284 However, the necessary relocations will trigger copy-on-write, and the
6285 shared object is not actually shared across processes. Instead of
6286 using @option{-mimpure-text}, you should compile all source code with
6287 @option{-fpic} or @option{-fPIC}.
6289 This option is only available on SunOS and Solaris.
6295 These two options select variations on the SPARC architecture.
6297 By default (unless specifically configured for the Fujitsu SPARClite),
6298 GCC generates code for the v7 variant of the SPARC architecture.
6300 @option{-mv8} will give you SPARC v8 code. The only difference from v7
6301 code is that the compiler emits the integer multiply and integer
6302 divide instructions which exist in SPARC v8 but not in SPARC v7.
6304 @option{-msparclite} will give you SPARClite code. This adds the integer
6305 multiply, integer divide step and scan (@code{ffs}) instructions which
6306 exist in SPARClite but not in SPARC v7.
6308 These options are deprecated and will be deleted in a future GCC release.
6309 They have been replaced with @option{-mcpu=xxx}.
6314 @opindex msupersparc
6315 These two options select the processor for which the code is optimized.
6317 With @option{-mcypress} (the default), the compiler optimizes code for the
6318 Cypress CY7C602 chip, as used in the SPARCStation/SPARCServer 3xx series.
6319 This is also appropriate for the older SPARCStation 1, 2, IPX etc.
6321 With @option{-msupersparc} the compiler optimizes code for the SuperSPARC cpu, as
6322 used in the SPARCStation 10, 1000 and 2000 series. This flag also enables use
6323 of the full SPARC v8 instruction set.
6325 These options are deprecated and will be deleted in a future GCC release.
6326 They have been replaced with @option{-mcpu=xxx}.
6328 @item -mcpu=@var{cpu_type}
6330 Set the instruction set, register set, and instruction scheduling parameters
6331 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
6332 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
6333 @samp{hypersparc}, @samp{sparclite86x}, @samp{f930}, @samp{f934},
6334 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc}, and
6337 Default instruction scheduling parameters are used for values that select
6338 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
6339 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
6341 Here is a list of each supported architecture and their supported
6346 v8: supersparc, hypersparc
6347 sparclite: f930, f934, sparclite86x
6349 v9: ultrasparc, ultrasparc3
6352 @item -mtune=@var{cpu_type}
6354 Set the instruction scheduling parameters for machine type
6355 @var{cpu_type}, but do not set the instruction set or register set that the
6356 option @option{-mcpu=@var{cpu_type}} would.
6358 The same values for @option{-mcpu=@var{cpu_type}} can be used for
6359 @option{-mtune=@var{cpu_type}}, but the only useful values are those
6360 that select a particular cpu implementation. Those are @samp{cypress},
6361 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
6362 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc}, and
6367 These @samp{-m} switches are supported in addition to the above
6368 on the SPARCLET processor.
6371 @item -mlittle-endian
6372 @opindex mlittle-endian
6373 Generate code for a processor running in little-endian mode.
6377 Treat register @code{%g0} as a normal register.
6378 GCC will continue to clobber it as necessary but will not assume
6379 it always reads as 0.
6381 @item -mbroken-saverestore
6382 @opindex mbroken-saverestore
6383 Generate code that does not use non-trivial forms of the @code{save} and
6384 @code{restore} instructions. Early versions of the SPARCLET processor do
6385 not correctly handle @code{save} and @code{restore} instructions used with
6386 arguments. They correctly handle them used without arguments. A @code{save}
6387 instruction used without arguments increments the current window pointer
6388 but does not allocate a new stack frame. It is assumed that the window
6389 overflow trap handler will properly handle this case as will interrupt
6393 These @samp{-m} switches are supported in addition to the above
6394 on SPARC V9 processors in 64-bit environments.
6397 @item -mlittle-endian
6398 @opindex mlittle-endian
6399 Generate code for a processor running in little-endian mode. It is only
6400 available for a few configurations and most notably not on Solaris.
6406 Generate code for a 32-bit or 64-bit environment.
6407 The 32-bit environment sets int, long and pointer to 32 bits.
6408 The 64-bit environment sets int to 32 bits and long and pointer
6411 @item -mcmodel=medlow
6412 @opindex mcmodel=medlow
6413 Generate code for the Medium/Low code model: the program must be linked
6414 in the low 32 bits of the address space. Pointers are 64 bits.
6415 Programs can be statically or dynamically linked.
6417 @item -mcmodel=medmid
6418 @opindex mcmodel=medmid
6419 Generate code for the Medium/Middle code model: the program must be linked
6420 in the low 44 bits of the address space, the text segment must be less than
6421 2G bytes, and data segment must be within 2G of the text segment.
6422 Pointers are 64 bits.
6424 @item -mcmodel=medany
6425 @opindex mcmodel=medany
6426 Generate code for the Medium/Anywhere code model: the program may be linked
6427 anywhere in the address space, the text segment must be less than
6428 2G bytes, and data segment must be within 2G of the text segment.
6429 Pointers are 64 bits.
6431 @item -mcmodel=embmedany
6432 @opindex mcmodel=embmedany
6433 Generate code for the Medium/Anywhere code model for embedded systems:
6434 assume a 32-bit text and a 32-bit data segment, both starting anywhere
6435 (determined at link time). Register %g4 points to the base of the
6436 data segment. Pointers are still 64 bits.
6437 Programs are statically linked, PIC is not supported.
6440 @itemx -mno-stack-bias
6441 @opindex mstack-bias
6442 @opindex mno-stack-bias
6443 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
6444 frame pointer if present, are offset by @minus{}2047 which must be added back
6445 when making stack frame references.
6446 Otherwise, assume no such offset is present.
6450 @subsection ARM Options
6453 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
6458 @opindex mapcs-frame
6459 Generate a stack frame that is compliant with the ARM Procedure Call
6460 Standard for all functions, even if this is not strictly necessary for
6461 correct execution of the code. Specifying @option{-fomit-frame-pointer}
6462 with this option will cause the stack frames not to be generated for
6463 leaf functions. The default is @option{-mno-apcs-frame}.
6467 This is a synonym for @option{-mapcs-frame}.
6471 Generate code for a processor running with a 26-bit program counter,
6472 and conforming to the function calling standards for the APCS 26-bit
6473 option. This option replaces the @option{-m2} and @option{-m3} options
6474 of previous releases of the compiler.
6478 Generate code for a processor running with a 32-bit program counter,
6479 and conforming to the function calling standards for the APCS 32-bit
6480 option. This option replaces the @option{-m6} option of previous releases
6484 @c not currently implemented
6485 @item -mapcs-stack-check
6486 @opindex mapcs-stack-check
6487 Generate code to check the amount of stack space available upon entry to
6488 every function (that actually uses some stack space). If there is
6489 insufficient space available then either the function
6490 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
6491 called, depending upon the amount of stack space required. The run time
6492 system is required to provide these functions. The default is
6493 @option{-mno-apcs-stack-check}, since this produces smaller code.
6495 @c not currently implemented
6497 @opindex mapcs-float
6498 Pass floating point arguments using the float point registers. This is
6499 one of the variants of the APCS@. This option is recommended if the
6500 target hardware has a floating point unit or if a lot of floating point
6501 arithmetic is going to be performed by the code. The default is
6502 @option{-mno-apcs-float}, since integer only code is slightly increased in
6503 size if @option{-mapcs-float} is used.
6505 @c not currently implemented
6506 @item -mapcs-reentrant
6507 @opindex mapcs-reentrant
6508 Generate reentrant, position independent code. The default is
6509 @option{-mno-apcs-reentrant}.
6512 @item -mthumb-interwork
6513 @opindex mthumb-interwork
6514 Generate code which supports calling between the ARM and Thumb
6515 instruction sets. Without this option the two instruction sets cannot
6516 be reliably used inside one program. The default is
6517 @option{-mno-thumb-interwork}, since slightly larger code is generated
6518 when @option{-mthumb-interwork} is specified.
6520 @item -mno-sched-prolog
6521 @opindex mno-sched-prolog
6522 Prevent the reordering of instructions in the function prolog, or the
6523 merging of those instruction with the instructions in the function's
6524 body. This means that all functions will start with a recognizable set
6525 of instructions (or in fact one of a choice from a small set of
6526 different function prologues), and this information can be used to
6527 locate the start if functions inside an executable piece of code. The
6528 default is @option{-msched-prolog}.
6531 @opindex mhard-float
6532 Generate output containing floating point instructions. This is the
6536 @opindex msoft-float
6537 Generate output containing library calls for floating point.
6538 @strong{Warning:} the requisite libraries are not available for all ARM
6539 targets. Normally the facilities of the machine's usual C compiler are
6540 used, but this cannot be done directly in cross-compilation. You must make
6541 your own arrangements to provide suitable library functions for
6544 @option{-msoft-float} changes the calling convention in the output file;
6545 therefore, it is only useful if you compile @emph{all} of a program with
6546 this option. In particular, you need to compile @file{libgcc.a}, the
6547 library that comes with GCC, with @option{-msoft-float} in order for
6550 @item -mlittle-endian
6551 @opindex mlittle-endian
6552 Generate code for a processor running in little-endian mode. This is
6553 the default for all standard configurations.
6556 @opindex mbig-endian
6557 Generate code for a processor running in big-endian mode; the default is
6558 to compile code for a little-endian processor.
6560 @item -mwords-little-endian
6561 @opindex mwords-little-endian
6562 This option only applies when generating code for big-endian processors.
6563 Generate code for a little-endian word order but a big-endian byte
6564 order. That is, a byte order of the form @samp{32107654}. Note: this
6565 option should only be used if you require compatibility with code for
6566 big-endian ARM processors generated by versions of the compiler prior to
6569 @item -malignment-traps
6570 @opindex malignment-traps
6571 Generate code that will not trap if the MMU has alignment traps enabled.
6572 On ARM architectures prior to ARMv4, there were no instructions to
6573 access half-word objects stored in memory. However, when reading from
6574 memory a feature of the ARM architecture allows a word load to be used,
6575 even if the address is unaligned, and the processor core will rotate the
6576 data as it is being loaded. This option tells the compiler that such
6577 misaligned accesses will cause a MMU trap and that it should instead
6578 synthesize the access as a series of byte accesses. The compiler can
6579 still use word accesses to load half-word data if it knows that the
6580 address is aligned to a word boundary.
6582 This option is ignored when compiling for ARM architecture 4 or later,
6583 since these processors have instructions to directly access half-word
6586 @item -mno-alignment-traps
6587 @opindex mno-alignment-traps
6588 Generate code that assumes that the MMU will not trap unaligned
6589 accesses. This produces better code when the target instruction set
6590 does not have half-word memory operations (i.e.@: implementations prior to
6593 Note that you cannot use this option to access unaligned word objects,
6594 since the processor will only fetch one 32-bit aligned object from
6597 The default setting for most targets is @option{-mno-alignment-traps}, since
6598 this produces better code when there are no half-word memory
6599 instructions available.
6601 @item -mshort-load-bytes
6602 @itemx -mno-short-load-words
6603 @opindex mshort-load-bytes
6604 @opindex mno-short-load-words
6605 These are deprecated aliases for @option{-malignment-traps}.
6607 @item -mno-short-load-bytes
6608 @itemx -mshort-load-words
6609 @opindex mno-short-load-bytes
6610 @opindex mshort-load-words
6611 This are deprecated aliases for @option{-mno-alignment-traps}.
6613 @item -mcpu=@var{name}
6615 This specifies the name of the target ARM processor. GCC uses this name
6616 to determine what kind of instructions it can emit when generating
6617 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
6618 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
6619 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
6620 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
6621 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
6622 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm8},
6623 @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
6624 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
6625 @samp{arm920t}, @samp{arm940t}, @samp{arm9tdmi}, @samp{arm10tdmi},
6626 @samp{arm1020t}, @samp{xscale}, @samp{iwmmxt}, @samp{ep9312}.
6628 @itemx -mtune=@var{name}
6630 This option is very similar to the @option{-mcpu=} option, except that
6631 instead of specifying the actual target processor type, and hence
6632 restricting which instructions can be used, it specifies that GCC should
6633 tune the performance of the code as if the target were of the type
6634 specified in this option, but still choosing the instructions that it
6635 will generate based on the cpu specified by a @option{-mcpu=} option.
6636 For some ARM implementations better performance can be obtained by using
6639 @item -march=@var{name}
6641 This specifies the name of the target ARM architecture. GCC uses this
6642 name to determine what kind of instructions it can emit when generating
6643 assembly code. This option can be used in conjunction with or instead
6644 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
6645 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
6646 @samp{armv5}, @samp{armv5t}, @samp{armv5te}, @samp{iwmmxt}, @samp{ep9312}.
6648 @item -mfpe=@var{number}
6649 @itemx -mfp=@var{number}
6652 This specifies the version of the floating point emulation available on
6653 the target. Permissible values are 2 and 3. @option{-mfp=} is a synonym
6654 for @option{-mfpe=}, for compatibility with older versions of GCC@.
6656 @item -mstructure-size-boundary=@var{n}
6657 @opindex mstructure-size-boundary
6658 The size of all structures and unions will be rounded up to a multiple
6659 of the number of bits set by this option. Permissible values are 8 and
6660 32. The default value varies for different toolchains. For the COFF
6661 targeted toolchain the default value is 8. Specifying the larger number
6662 can produce faster, more efficient code, but can also increase the size
6663 of the program. The two values are potentially incompatible. Code
6664 compiled with one value cannot necessarily expect to work with code or
6665 libraries compiled with the other value, if they exchange information
6666 using structures or unions.
6668 @item -mabort-on-noreturn
6669 @opindex mabort-on-noreturn
6670 Generate a call to the function @code{abort} at the end of a
6671 @code{noreturn} function. It will be executed if the function tries to
6675 @itemx -mno-long-calls
6676 @opindex mlong-calls
6677 @opindex mno-long-calls
6678 Tells the compiler to perform function calls by first loading the
6679 address of the function into a register and then performing a subroutine
6680 call on this register. This switch is needed if the target function
6681 will lie outside of the 64 megabyte addressing range of the offset based
6682 version of subroutine call instruction.
6684 Even if this switch is enabled, not all function calls will be turned
6685 into long calls. The heuristic is that static functions, functions
6686 which have the @samp{short-call} attribute, functions that are inside
6687 the scope of a @samp{#pragma no_long_calls} directive and functions whose
6688 definitions have already been compiled within the current compilation
6689 unit, will not be turned into long calls. The exception to this rule is
6690 that weak function definitions, functions with the @samp{long-call}
6691 attribute or the @samp{section} attribute, and functions that are within
6692 the scope of a @samp{#pragma long_calls} directive, will always be
6693 turned into long calls.
6695 This feature is not enabled by default. Specifying
6696 @option{-mno-long-calls} will restore the default behavior, as will
6697 placing the function calls within the scope of a @samp{#pragma
6698 long_calls_off} directive. Note these switches have no effect on how
6699 the compiler generates code to handle function calls via function
6702 @item -mnop-fun-dllimport
6703 @opindex mnop-fun-dllimport
6704 Disable support for the @code{dllimport} attribute.
6706 @item -msingle-pic-base
6707 @opindex msingle-pic-base
6708 Treat the register used for PIC addressing as read-only, rather than
6709 loading it in the prologue for each function. The run-time system is
6710 responsible for initializing this register with an appropriate value
6711 before execution begins.
6713 @item -mpic-register=@var{reg}
6714 @opindex mpic-register
6715 Specify the register to be used for PIC addressing. The default is R10
6716 unless stack-checking is enabled, when R9 is used.
6718 @item -mcirrus-fix-invalid-insns
6719 @opindex mcirrus-fix-invalid-insns
6720 @opindex mno-cirrus-fix-invalid-insns
6721 Insert NOPs into the instruction stream to in order to work around
6722 problems with invalid Maverick instruction combinations. This option
6723 is only valid if the @option{-mcpu=ep9312} option has been used to
6724 enable generation of instructions for the Cirrus Maverick floating
6725 point co-processor. This option is not enabled by default, since the
6726 problem is only present in older Maverick implementations. The default
6727 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
6730 @item -mpoke-function-name
6731 @opindex mpoke-function-name
6732 Write the name of each function into the text section, directly
6733 preceding the function prologue. The generated code is similar to this:
6737 .ascii "arm_poke_function_name", 0
6740 .word 0xff000000 + (t1 - t0)
6741 arm_poke_function_name
6743 stmfd sp!, @{fp, ip, lr, pc@}
6747 When performing a stack backtrace, code can inspect the value of
6748 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
6749 location @code{pc - 12} and the top 8 bits are set, then we know that
6750 there is a function name embedded immediately preceding this location
6751 and has length @code{((pc[-3]) & 0xff000000)}.
6755 Generate code for the 16-bit Thumb instruction set. The default is to
6756 use the 32-bit ARM instruction set.
6759 @opindex mtpcs-frame
6760 Generate a stack frame that is compliant with the Thumb Procedure Call
6761 Standard for all non-leaf functions. (A leaf function is one that does
6762 not call any other functions.) The default is @option{-mno-tpcs-frame}.
6764 @item -mtpcs-leaf-frame
6765 @opindex mtpcs-leaf-frame
6766 Generate a stack frame that is compliant with the Thumb Procedure Call
6767 Standard for all leaf functions. (A leaf function is one that does
6768 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
6770 @item -mcallee-super-interworking
6771 @opindex mcallee-super-interworking
6772 Gives all externally visible functions in the file being compiled an ARM
6773 instruction set header which switches to Thumb mode before executing the
6774 rest of the function. This allows these functions to be called from
6775 non-interworking code.
6777 @item -mcaller-super-interworking
6778 @opindex mcaller-super-interworking
6779 Allows calls via function pointers (including virtual functions) to
6780 execute correctly regardless of whether the target code has been
6781 compiled for interworking or not. There is a small overhead in the cost
6782 of executing a function pointer if this option is enabled.
6786 @node MN10200 Options
6787 @subsection MN10200 Options
6788 @cindex MN10200 options
6790 These @option{-m} options are defined for Matsushita MN10200 architectures:
6795 Indicate to the linker that it should perform a relaxation optimization pass
6796 to shorten branches, calls and absolute memory addresses. This option only
6797 has an effect when used on the command line for the final link step.
6799 This option makes symbolic debugging impossible.
6802 @node MN10300 Options
6803 @subsection MN10300 Options
6804 @cindex MN10300 options
6806 These @option{-m} options are defined for Matsushita MN10300 architectures:
6811 Generate code to avoid bugs in the multiply instructions for the MN10300
6812 processors. This is the default.
6815 @opindex mno-mult-bug
6816 Do not generate code to avoid bugs in the multiply instructions for the
6821 Generate code which uses features specific to the AM33 processor.
6825 Do not generate code which uses features specific to the AM33 processor. This
6830 Do not link in the C run-time initialization object file.
6834 Indicate to the linker that it should perform a relaxation optimization pass
6835 to shorten branches, calls and absolute memory addresses. This option only
6836 has an effect when used on the command line for the final link step.
6838 This option makes symbolic debugging impossible.
6842 @node M32R/D Options
6843 @subsection M32R/D Options
6844 @cindex M32R/D options
6846 These @option{-m} options are defined for Mitsubishi M32R/D architectures:
6851 Generate code for the M32R/X@.
6855 Generate code for the M32R@. This is the default.
6857 @item -mcode-model=small
6858 @opindex mcode-model=small
6859 Assume all objects live in the lower 16MB of memory (so that their addresses
6860 can be loaded with the @code{ld24} instruction), and assume all subroutines
6861 are reachable with the @code{bl} instruction.
6862 This is the default.
6864 The addressability of a particular object can be set with the
6865 @code{model} attribute.
6867 @item -mcode-model=medium
6868 @opindex mcode-model=medium
6869 Assume objects may be anywhere in the 32-bit address space (the compiler
6870 will generate @code{seth/add3} instructions to load their addresses), and
6871 assume all subroutines are reachable with the @code{bl} instruction.
6873 @item -mcode-model=large
6874 @opindex mcode-model=large
6875 Assume objects may be anywhere in the 32-bit address space (the compiler
6876 will generate @code{seth/add3} instructions to load their addresses), and
6877 assume subroutines may not be reachable with the @code{bl} instruction
6878 (the compiler will generate the much slower @code{seth/add3/jl}
6879 instruction sequence).
6882 @opindex msdata=none
6883 Disable use of the small data area. Variables will be put into
6884 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
6885 @code{section} attribute has been specified).
6886 This is the default.
6888 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
6889 Objects may be explicitly put in the small data area with the
6890 @code{section} attribute using one of these sections.
6893 @opindex msdata=sdata
6894 Put small global and static data in the small data area, but do not
6895 generate special code to reference them.
6899 Put small global and static data in the small data area, and generate
6900 special instructions to reference them.
6904 @cindex smaller data references
6905 Put global and static objects less than or equal to @var{num} bytes
6906 into the small data or bss sections instead of the normal data or bss
6907 sections. The default value of @var{num} is 8.
6908 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
6909 for this option to have any effect.
6911 All modules should be compiled with the same @option{-G @var{num}} value.
6912 Compiling with different values of @var{num} may or may not work; if it
6913 doesn't the linker will give an error message---incorrect code will not be
6919 @subsection M88K Options
6920 @cindex M88k options
6922 These @samp{-m} options are defined for Motorola 88k architectures:
6927 Generate code that works well on both the m88100 and the
6932 Generate code that works best for the m88100, but that also
6937 Generate code that works best for the m88110, and may not run
6942 Obsolete option to be removed from the next revision.
6945 @item -midentify-revision
6946 @opindex midentify-revision
6947 @cindex identifying source, compiler (88k)
6948 Include an @code{ident} directive in the assembler output recording the
6949 source file name, compiler name and version, timestamp, and compilation
6952 @item -mno-underscores
6953 @opindex mno-underscores
6954 @cindex underscores, avoiding (88k)
6955 In assembler output, emit symbol names without adding an underscore
6956 character at the beginning of each name. The default is to use an
6957 underscore as prefix on each name.
6959 @item -mocs-debug-info
6960 @itemx -mno-ocs-debug-info
6961 @opindex mocs-debug-info
6962 @opindex mno-ocs-debug-info
6964 @cindex debugging, 88k OCS
6965 Include (or omit) additional debugging information (about registers used
6966 in each stack frame) as specified in the 88open Object Compatibility
6967 Standard, ``OCS''@. This extra information allows debugging of code that
6968 has had the frame pointer eliminated. The default for SVr4 and Delta 88
6969 SVr3.2 is to include this information; other 88k configurations omit this
6970 information by default.
6972 @item -mocs-frame-position
6973 @opindex mocs-frame-position
6974 @cindex register positions in frame (88k)
6975 When emitting COFF debugging information for automatic variables and
6976 parameters stored on the stack, use the offset from the canonical frame
6977 address, which is the stack pointer (register 31) on entry to the
6978 function. The SVr4 and Delta88 SVr3.2, and BCS configurations use
6979 @option{-mocs-frame-position}; other 88k configurations have the default
6980 @option{-mno-ocs-frame-position}.
6982 @item -mno-ocs-frame-position
6983 @opindex mno-ocs-frame-position
6984 @cindex register positions in frame (88k)
6985 When emitting COFF debugging information for automatic variables and
6986 parameters stored on the stack, use the offset from the frame pointer
6987 register (register 30). When this option is in effect, the frame
6988 pointer is not eliminated when debugging information is selected by the
6991 @item -moptimize-arg-area
6992 @opindex moptimize-arg-area
6993 @cindex arguments in frame (88k)
6994 Save space by reorganizing the stack frame. This option generates code
6995 that does not agree with the 88open specifications, but uses less
6998 @itemx -mno-optimize-arg-area
6999 @opindex mno-optimize-arg-area
7000 Do not reorganize the stack frame to save space. This is the default.
7001 The generated conforms to the specification, but uses more memory.
7003 @item -mshort-data-@var{num}
7004 @opindex mshort-data
7005 @cindex smaller data references (88k)
7006 @cindex r0-relative references (88k)
7007 Generate smaller data references by making them relative to @code{r0},
7008 which allows loading a value using a single instruction (rather than the
7009 usual two). You control which data references are affected by
7010 specifying @var{num} with this option. For example, if you specify
7011 @option{-mshort-data-512}, then the data references affected are those
7012 involving displacements of less than 512 bytes.
7013 @option{-mshort-data-@var{num}} is not effective for @var{num} greater
7016 @item -mserialize-volatile
7017 @opindex mserialize-volatile
7018 @itemx -mno-serialize-volatile
7019 @opindex mno-serialize-volatile
7020 @cindex sequential consistency on 88k
7021 Do, or don't, generate code to guarantee sequential consistency
7022 of volatile memory references. By default, consistency is
7025 The order of memory references made by the MC88110 processor does
7026 not always match the order of the instructions requesting those
7027 references. In particular, a load instruction may execute before
7028 a preceding store instruction. Such reordering violates
7029 sequential consistency of volatile memory references, when there
7030 are multiple processors. When consistency must be guaranteed,
7031 GCC generates special instructions, as needed, to force
7032 execution in the proper order.
7034 The MC88100 processor does not reorder memory references and so
7035 always provides sequential consistency. However, by default, GCC
7036 generates the special instructions to guarantee consistency
7037 even when you use @option{-m88100}, so that the code may be run on an
7038 MC88110 processor. If you intend to run your code only on the
7039 MC88100 processor, you may use @option{-mno-serialize-volatile}.
7041 The extra code generated to guarantee consistency may affect the
7042 performance of your application. If you know that you can safely
7043 forgo this guarantee, you may use @option{-mno-serialize-volatile}.
7049 @cindex assembler syntax, 88k
7051 Turn on (@option{-msvr4}) or off (@option{-msvr3}) compiler extensions
7052 related to System V release 4 (SVr4). This controls the following:
7056 Which variant of the assembler syntax to emit.
7058 @option{-msvr4} makes the C preprocessor recognize @samp{#pragma weak}
7059 that is used on System V release 4.
7061 @option{-msvr4} makes GCC issue additional declaration directives used in
7065 @option{-msvr4} is the default for the m88k-motorola-sysv4 configuration.
7066 @option{-msvr3} is the default for all other m88k configurations.
7068 @item -mversion-03.00
7069 @opindex mversion-03.00
7070 This option is obsolete, and is ignored.
7071 @c ??? which asm syntax better for GAS? option there too?
7073 @item -mno-check-zero-division
7074 @itemx -mcheck-zero-division
7075 @opindex mno-check-zero-division
7076 @opindex mcheck-zero-division
7077 @cindex zero division on 88k
7078 Do, or don't, generate code to guarantee that integer division by
7079 zero will be detected. By default, detection is guaranteed.
7081 Some models of the MC88100 processor fail to trap upon integer
7082 division by zero under certain conditions. By default, when
7083 compiling code that might be run on such a processor, GCC
7084 generates code that explicitly checks for zero-valued divisors
7085 and traps with exception number 503 when one is detected. Use of
7086 @option{-mno-check-zero-division} suppresses such checking for code
7087 generated to run on an MC88100 processor.
7089 GCC assumes that the MC88110 processor correctly detects all instances
7090 of integer division by zero. When @option{-m88110} is specified, no
7091 explicit checks for zero-valued divisors are generated, and both
7092 @option{-mcheck-zero-division} and @option{-mno-check-zero-division} are
7095 @item -muse-div-instruction
7096 @opindex muse-div-instruction
7097 @cindex divide instruction, 88k
7098 Use the div instruction for signed integer division on the
7099 MC88100 processor. By default, the div instruction is not used.
7101 On the MC88100 processor the signed integer division instruction
7102 div) traps to the operating system on a negative operand. The
7103 operating system transparently completes the operation, but at a
7104 large cost in execution time. By default, when compiling code
7105 that might be run on an MC88100 processor, GCC emulates signed
7106 integer division using the unsigned integer division instruction
7107 divu), thereby avoiding the large penalty of a trap to the
7108 operating system. Such emulation has its own, smaller, execution
7109 cost in both time and space. To the extent that your code's
7110 important signed integer division operations are performed on two
7111 nonnegative operands, it may be desirable to use the div
7112 instruction directly.
7114 On the MC88110 processor the div instruction (also known as the
7115 divs instruction) processes negative operands without trapping to
7116 the operating system. When @option{-m88110} is specified,
7117 @option{-muse-div-instruction} is ignored, and the div instruction is used
7118 for signed integer division.
7120 Note that the result of dividing @code{INT_MIN} by @minus{}1 is undefined. In
7121 particular, the behavior of such a division with and without
7122 @option{-muse-div-instruction} may differ.
7124 @item -mtrap-large-shift
7125 @itemx -mhandle-large-shift
7126 @opindex mtrap-large-shift
7127 @opindex mhandle-large-shift
7128 @cindex bit shift overflow (88k)
7129 @cindex large bit shifts (88k)
7130 Include code to detect bit-shifts of more than 31 bits; respectively,
7131 trap such shifts or emit code to handle them properly. By default GCC
7132 makes no special provision for large bit shifts.
7134 @item -mwarn-passed-structs
7135 @opindex mwarn-passed-structs
7136 @cindex structure passing (88k)
7137 Warn when a function passes a struct as an argument or result.
7138 Structure-passing conventions have changed during the evolution of the C
7139 language, and are often the source of portability problems. By default,
7140 GCC issues no such warning.
7143 @c break page here to avoid unsightly interparagraph stretch.
7147 @node RS/6000 and PowerPC Options
7148 @subsection IBM RS/6000 and PowerPC Options
7149 @cindex RS/6000 and PowerPC Options
7150 @cindex IBM RS/6000 and PowerPC Options
7152 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
7160 @itemx -mpowerpc-gpopt
7161 @itemx -mno-powerpc-gpopt
7162 @itemx -mpowerpc-gfxopt
7163 @itemx -mno-powerpc-gfxopt
7165 @itemx -mno-powerpc64
7171 @opindex mno-powerpc
7172 @opindex mpowerpc-gpopt
7173 @opindex mno-powerpc-gpopt
7174 @opindex mpowerpc-gfxopt
7175 @opindex mno-powerpc-gfxopt
7177 @opindex mno-powerpc64
7178 GCC supports two related instruction set architectures for the
7179 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
7180 instructions supported by the @samp{rios} chip set used in the original
7181 RS/6000 systems and the @dfn{PowerPC} instruction set is the
7182 architecture of the Motorola MPC5xx, MPC6xx, MPC8xx microprocessors, and
7183 the IBM 4xx microprocessors.
7185 Neither architecture is a subset of the other. However there is a
7186 large common subset of instructions supported by both. An MQ
7187 register is included in processors supporting the POWER architecture.
7189 You use these options to specify which instructions are available on the
7190 processor you are using. The default value of these options is
7191 determined when configuring GCC@. Specifying the
7192 @option{-mcpu=@var{cpu_type}} overrides the specification of these
7193 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
7194 rather than the options listed above.
7196 The @option{-mpower} option allows GCC to generate instructions that
7197 are found only in the POWER architecture and to use the MQ register.
7198 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
7199 to generate instructions that are present in the POWER2 architecture but
7200 not the original POWER architecture.
7202 The @option{-mpowerpc} option allows GCC to generate instructions that
7203 are found only in the 32-bit subset of the PowerPC architecture.
7204 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
7205 GCC to use the optional PowerPC architecture instructions in the
7206 General Purpose group, including floating-point square root. Specifying
7207 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
7208 use the optional PowerPC architecture instructions in the Graphics
7209 group, including floating-point select.
7211 The @option{-mpowerpc64} option allows GCC to generate the additional
7212 64-bit instructions that are found in the full PowerPC64 architecture
7213 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
7214 @option{-mno-powerpc64}.
7216 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
7217 will use only the instructions in the common subset of both
7218 architectures plus some special AIX common-mode calls, and will not use
7219 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
7220 permits GCC to use any instruction from either architecture and to
7221 allow use of the MQ register; specify this for the Motorola MPC601.
7223 @item -mnew-mnemonics
7224 @itemx -mold-mnemonics
7225 @opindex mnew-mnemonics
7226 @opindex mold-mnemonics
7227 Select which mnemonics to use in the generated assembler code. With
7228 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
7229 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
7230 assembler mnemonics defined for the POWER architecture. Instructions
7231 defined in only one architecture have only one mnemonic; GCC uses that
7232 mnemonic irrespective of which of these options is specified.
7234 GCC defaults to the mnemonics appropriate for the architecture in
7235 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
7236 value of these option. Unless you are building a cross-compiler, you
7237 should normally not specify either @option{-mnew-mnemonics} or
7238 @option{-mold-mnemonics}, but should instead accept the default.
7240 @item -mcpu=@var{cpu_type}
7242 Set architecture type, register usage, choice of mnemonics, and
7243 instruction scheduling parameters for machine type @var{cpu_type}.
7244 Supported values for @var{cpu_type} are @samp{rios}, @samp{rios1},
7245 @samp{rsc}, @samp{rios2}, @samp{rs64a}, @samp{601}, @samp{602},
7246 @samp{603}, @samp{603e}, @samp{604}, @samp{604e}, @samp{620},
7247 @samp{630}, @samp{740}, @samp{7400}, @samp{7450}, @samp{750},
7248 @samp{power}, @samp{power2}, @samp{powerpc}, @samp{403}, @samp{505},
7249 @samp{801}, @samp{821}, @samp{823}, and @samp{860} and @samp{common}.
7251 @option{-mcpu=common} selects a completely generic processor. Code
7252 generated under this option will run on any POWER or PowerPC processor.
7253 GCC will use only the instructions in the common subset of both
7254 architectures, and will not use the MQ register. GCC assumes a generic
7255 processor model for scheduling purposes.
7257 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
7258 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
7259 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
7260 types, with an appropriate, generic processor model assumed for
7261 scheduling purposes.
7263 The other options specify a specific processor. Code generated under
7264 those options will run best on that processor, and may not run at all on
7267 The @option{-mcpu} options automatically enable or disable other
7268 @option{-m} options as follows:
7272 @option{-mno-power}, @option{-mno-powerpc}
7279 @option{-mpower}, @option{-mno-powerpc}, @option{-mno-new-mnemonics}
7294 @option{-mno-power}, @option{-mpowerpc}, @option{-mnew-mnemonics}
7297 @option{-mpower}, @option{-mpowerpc}, @option{-mnew-mnemonics}
7302 @option{-mno-power}, @option{-mpowerpc}, @option{-mnew-mnemonics}, @option{-msoft-float}
7305 @item -mtune=@var{cpu_type}
7307 Set the instruction scheduling parameters for machine type
7308 @var{cpu_type}, but do not set the architecture type, register usage, or
7309 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
7310 values for @var{cpu_type} are used for @option{-mtune} as for
7311 @option{-mcpu}. If both are specified, the code generated will use the
7312 architecture, registers, and mnemonics set by @option{-mcpu}, but the
7313 scheduling parameters set by @option{-mtune}.
7318 @opindex mno-altivec
7319 These switches enable or disable the use of built-in functions that
7320 allow access to the AltiVec instruction set. You may also need to set
7321 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
7326 Extend the current ABI with SPE ABI extensions. This does not change
7327 the default ABI, instead it adds the SPE ABI extensions to the current
7331 @opindex mabi=no-spe
7332 Disable Booke SPE ABI extensions for the current ABI.
7334 @item -misel=@var{yes/no}
7337 This switch enables or disables the generation of ISEL instructions.
7339 @item -mspe=@var{yes/no}
7342 This switch enables or disables the generation of SPE simd
7345 @item -mfloat-gprs=@var{yes/no}
7347 @opindex mfloat-gprs
7348 This switch enables or disables the generation of floating point
7349 operations on the general purpose registers for architectures that
7350 support it. This option is currently only available on the MPC8540.
7353 @itemx -mno-fp-in-toc
7354 @itemx -mno-sum-in-toc
7355 @itemx -mminimal-toc
7357 @opindex mno-fp-in-toc
7358 @opindex mno-sum-in-toc
7359 @opindex mminimal-toc
7360 Modify generation of the TOC (Table Of Contents), which is created for
7361 every executable file. The @option{-mfull-toc} option is selected by
7362 default. In that case, GCC will allocate at least one TOC entry for
7363 each unique non-automatic variable reference in your program. GCC
7364 will also place floating-point constants in the TOC@. However, only
7365 16,384 entries are available in the TOC@.
7367 If you receive a linker error message that saying you have overflowed
7368 the available TOC space, you can reduce the amount of TOC space used
7369 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
7370 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
7371 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
7372 generate code to calculate the sum of an address and a constant at
7373 run-time instead of putting that sum into the TOC@. You may specify one
7374 or both of these options. Each causes GCC to produce very slightly
7375 slower and larger code at the expense of conserving TOC space.
7377 If you still run out of space in the TOC even when you specify both of
7378 these options, specify @option{-mminimal-toc} instead. This option causes
7379 GCC to make only one TOC entry for every file. When you specify this
7380 option, GCC will produce code that is slower and larger but which
7381 uses extremely little TOC space. You may wish to use this option
7382 only on files that contain less frequently executed code.
7388 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
7389 @code{long} type, and the infrastructure needed to support them.
7390 Specifying @option{-maix64} implies @option{-mpowerpc64} and
7391 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
7392 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
7397 @opindex mno-xl-call
7398 On AIX, pass floating-point arguments to prototyped functions beyond the
7399 register save area (RSA) on the stack in addition to argument FPRs. The
7400 AIX calling convention was extended but not initially documented to
7401 handle an obscure K&R C case of calling a function that takes the
7402 address of its arguments with fewer arguments than declared. AIX XL
7403 compilers access floating point arguments which do not fit in the
7404 RSA from the stack when a subroutine is compiled without
7405 optimization. Because always storing floating-point arguments on the
7406 stack is inefficient and rarely needed, this option is not enabled by
7407 default and only is necessary when calling subroutines compiled by AIX
7408 XL compilers without optimization.
7412 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
7413 application written to use message passing with special startup code to
7414 enable the application to run. The system must have PE installed in the
7415 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
7416 must be overridden with the @option{-specs=} option to specify the
7417 appropriate directory location. The Parallel Environment does not
7418 support threads, so the @option{-mpe} option and the @option{-pthread}
7419 option are incompatible.
7421 @item -malign-natural
7422 @itemx -malign-power
7423 @opindex malign-natural
7424 @opindex malign-power
7425 On AIX, Darwin, and 64-bit PowerPC Linux, the option
7426 @option{-malign-natural} overrides the ABI-defined alignment of larger
7427 types, such as floating-point doubles, on their natural size-based boundary.
7428 The option @option{-malign-power} instructs GCC to follow the ABI-specified
7429 alignment rules. GCC defaults to the standard alignment defined in the ABI.
7433 @opindex msoft-float
7434 @opindex mhard-float
7435 Generate code that does not use (uses) the floating-point register set.
7436 Software floating point emulation is provided if you use the
7437 @option{-msoft-float} option, and pass the option to GCC when linking.
7440 @itemx -mno-multiple
7442 @opindex mno-multiple
7443 Generate code that uses (does not use) the load multiple word
7444 instructions and the store multiple word instructions. These
7445 instructions are generated by default on POWER systems, and not
7446 generated on PowerPC systems. Do not use @option{-mmultiple} on little
7447 endian PowerPC systems, since those instructions do not work when the
7448 processor is in little endian mode. The exceptions are PPC740 and
7449 PPC750 which permit the instructions usage in little endian mode.
7455 Generate code that uses (does not use) the load string instructions
7456 and the store string word instructions to save multiple registers and
7457 do small block moves. These instructions are generated by default on
7458 POWER systems, and not generated on PowerPC systems. Do not use
7459 @option{-mstring} on little endian PowerPC systems, since those
7460 instructions do not work when the processor is in little endian mode.
7461 The exceptions are PPC740 and PPC750 which permit the instructions
7462 usage in little endian mode.
7468 Generate code that uses (does not use) the load or store instructions
7469 that update the base register to the address of the calculated memory
7470 location. These instructions are generated by default. If you use
7471 @option{-mno-update}, there is a small window between the time that the
7472 stack pointer is updated and the address of the previous frame is
7473 stored, which means code that walks the stack frame across interrupts or
7474 signals may get corrupted data.
7477 @itemx -mno-fused-madd
7478 @opindex mfused-madd
7479 @opindex mno-fused-madd
7480 Generate code that uses (does not use) the floating point multiply and
7481 accumulate instructions. These instructions are generated by default if
7482 hardware floating is used.
7484 @item -mno-bit-align
7486 @opindex mno-bit-align
7488 On System V.4 and embedded PowerPC systems do not (do) force structures
7489 and unions that contain bit-fields to be aligned to the base type of the
7492 For example, by default a structure containing nothing but 8
7493 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
7494 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
7495 the structure would be aligned to a 1 byte boundary and be one byte in
7498 @item -mno-strict-align
7499 @itemx -mstrict-align
7500 @opindex mno-strict-align
7501 @opindex mstrict-align
7502 On System V.4 and embedded PowerPC systems do not (do) assume that
7503 unaligned memory references will be handled by the system.
7506 @itemx -mno-relocatable
7507 @opindex mrelocatable
7508 @opindex mno-relocatable
7509 On embedded PowerPC systems generate code that allows (does not allow)
7510 the program to be relocated to a different address at runtime. If you
7511 use @option{-mrelocatable} on any module, all objects linked together must
7512 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
7514 @item -mrelocatable-lib
7515 @itemx -mno-relocatable-lib
7516 @opindex mrelocatable-lib
7517 @opindex mno-relocatable-lib
7518 On embedded PowerPC systems generate code that allows (does not allow)
7519 the program to be relocated to a different address at runtime. Modules
7520 compiled with @option{-mrelocatable-lib} can be linked with either modules
7521 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
7522 with modules compiled with the @option{-mrelocatable} options.
7528 On System V.4 and embedded PowerPC systems do not (do) assume that
7529 register 2 contains a pointer to a global area pointing to the addresses
7530 used in the program.
7533 @itemx -mlittle-endian
7535 @opindex mlittle-endian
7536 On System V.4 and embedded PowerPC systems compile code for the
7537 processor in little endian mode. The @option{-mlittle-endian} option is
7538 the same as @option{-mlittle}.
7543 @opindex mbig-endian
7544 On System V.4 and embedded PowerPC systems compile code for the
7545 processor in big endian mode. The @option{-mbig-endian} option is
7546 the same as @option{-mbig}.
7548 @item -mdynamic-no-pic
7549 @opindex mdynamic-no-pic
7550 On Darwin and Mac OS X systems, compile code so that it is not
7551 relocatable, but that its external references are relocatable. The
7552 resulting code is suitable for applications, but not shared
7555 @item -mprioritize-restricted-insns=@var{priority}
7556 @opindex mprioritize-restricted-insns
7557 This option controls the priority that is assigned to
7558 dispatch-slot restricted instructions during the second scheduling
7559 pass. The argument @var{priority} takes the value @var{0/1/2} to assign
7560 @var{no/highest/second-highest} priority to dispatch slot restricted
7563 @item -msched-costly-dep=@var{dependence_type}
7564 @opindex msched-costly-dep
7565 This option controls which dependences are considered costly
7566 by the target during instruction scheduling. The argument
7567 @var{dependence_type} takes one of the following values:
7568 @var{no}: no dependence is costly,
7569 @var{all}: all dependences are costly,
7570 @var{true_store_to_load}: a true dependence from store to load is costly,
7571 @var{store_to_load}: any dependence from store to load is costly,
7572 @var{number}: any dependence which latency >= @var{number} is costly.
7576 On System V.4 and embedded PowerPC systems compile code using calling
7577 conventions that adheres to the March 1995 draft of the System V
7578 Application Binary Interface, PowerPC processor supplement. This is the
7579 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
7581 @item -mcall-sysv-eabi
7582 @opindex mcall-sysv-eabi
7583 Specify both @option{-mcall-sysv} and @option{-meabi} options.
7585 @item -mcall-sysv-noeabi
7586 @opindex mcall-sysv-noeabi
7587 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
7589 @item -mcall-solaris
7590 @opindex mcall-solaris
7591 On System V.4 and embedded PowerPC systems compile code for the Solaris
7595 @opindex mcall-linux
7596 On System V.4 and embedded PowerPC systems compile code for the
7597 Linux-based GNU system.
7601 On System V.4 and embedded PowerPC systems compile code for the
7602 Hurd-based GNU system.
7605 @opindex mcall-netbsd
7606 On System V.4 and embedded PowerPC systems compile code for the
7607 NetBSD operating system.
7609 @item -maix-struct-return
7610 @opindex maix-struct-return
7611 Return all structures in memory (as specified by the AIX ABI)@.
7613 @item -msvr4-struct-return
7614 @opindex msvr4-struct-return
7615 Return structures smaller than 8 bytes in registers (as specified by the
7619 @opindex mabi=altivec
7620 Extend the current ABI with AltiVec ABI extensions. This does not
7621 change the default ABI, instead it adds the AltiVec ABI extensions to
7624 @item -mabi=no-altivec
7625 @opindex mabi=no-altivec
7626 Disable AltiVec ABI extensions for the current ABI.
7629 @itemx -mno-prototype
7631 @opindex mno-prototype
7632 On System V.4 and embedded PowerPC systems assume that all calls to
7633 variable argument functions are properly prototyped. Otherwise, the
7634 compiler must insert an instruction before every non prototyped call to
7635 set or clear bit 6 of the condition code register (@var{CR}) to
7636 indicate whether floating point values were passed in the floating point
7637 registers in case the function takes a variable arguments. With
7638 @option{-mprototype}, only calls to prototyped variable argument functions
7639 will set or clear the bit.
7643 On embedded PowerPC systems, assume that the startup module is called
7644 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
7645 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
7650 On embedded PowerPC systems, assume that the startup module is called
7651 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
7656 On embedded PowerPC systems, assume that the startup module is called
7657 @file{crt0.o} and the standard C libraries are @file{libads.a} and
7661 @opindex myellowknife
7662 On embedded PowerPC systems, assume that the startup module is called
7663 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
7668 On System V.4 and embedded PowerPC systems, specify that you are
7669 compiling for a VxWorks system.
7673 Specify that you are compiling for the WindISS simulation environment.
7677 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
7678 header to indicate that @samp{eabi} extended relocations are used.
7684 On System V.4 and embedded PowerPC systems do (do not) adhere to the
7685 Embedded Applications Binary Interface (eabi) which is a set of
7686 modifications to the System V.4 specifications. Selecting @option{-meabi}
7687 means that the stack is aligned to an 8 byte boundary, a function
7688 @code{__eabi} is called to from @code{main} to set up the eabi
7689 environment, and the @option{-msdata} option can use both @code{r2} and
7690 @code{r13} to point to two separate small data areas. Selecting
7691 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
7692 do not call an initialization function from @code{main}, and the
7693 @option{-msdata} option will only use @code{r13} to point to a single
7694 small data area. The @option{-meabi} option is on by default if you
7695 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
7698 @opindex msdata=eabi
7699 On System V.4 and embedded PowerPC systems, put small initialized
7700 @code{const} global and static data in the @samp{.sdata2} section, which
7701 is pointed to by register @code{r2}. Put small initialized
7702 non-@code{const} global and static data in the @samp{.sdata} section,
7703 which is pointed to by register @code{r13}. Put small uninitialized
7704 global and static data in the @samp{.sbss} section, which is adjacent to
7705 the @samp{.sdata} section. The @option{-msdata=eabi} option is
7706 incompatible with the @option{-mrelocatable} option. The
7707 @option{-msdata=eabi} option also sets the @option{-memb} option.
7710 @opindex msdata=sysv
7711 On System V.4 and embedded PowerPC systems, put small global and static
7712 data in the @samp{.sdata} section, which is pointed to by register
7713 @code{r13}. Put small uninitialized global and static data in the
7714 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
7715 The @option{-msdata=sysv} option is incompatible with the
7716 @option{-mrelocatable} option.
7718 @item -msdata=default
7720 @opindex msdata=default
7722 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
7723 compile code the same as @option{-msdata=eabi}, otherwise compile code the
7724 same as @option{-msdata=sysv}.
7727 @opindex msdata-data
7728 On System V.4 and embedded PowerPC systems, put small global and static
7729 data in the @samp{.sdata} section. Put small uninitialized global and
7730 static data in the @samp{.sbss} section. Do not use register @code{r13}
7731 to address small data however. This is the default behavior unless
7732 other @option{-msdata} options are used.
7736 @opindex msdata=none
7738 On embedded PowerPC systems, put all initialized global and static data
7739 in the @samp{.data} section, and all uninitialized data in the
7740 @samp{.bss} section.
7744 @cindex smaller data references (PowerPC)
7745 @cindex .sdata/.sdata2 references (PowerPC)
7746 On embedded PowerPC systems, put global and static items less than or
7747 equal to @var{num} bytes into the small data or bss sections instead of
7748 the normal data or bss section. By default, @var{num} is 8. The
7749 @option{-G @var{num}} switch is also passed to the linker.
7750 All modules should be compiled with the same @option{-G @var{num}} value.
7753 @itemx -mno-regnames
7755 @opindex mno-regnames
7756 On System V.4 and embedded PowerPC systems do (do not) emit register
7757 names in the assembly language output using symbolic forms.
7760 @itemx -mno-longcall
7762 @opindex mno-longcall
7763 Default to making all function calls via pointers, so that functions
7764 which reside further than 64 megabytes (67,108,864 bytes) from the
7765 current location can be called. This setting can be overridden by the
7766 @code{shortcall} function attribute, or by @code{#pragma longcall(0)}.
7768 Some linkers are capable of detecting out-of-range calls and generating
7769 glue code on the fly. On these systems, long calls are unnecessary and
7770 generate slower code. As of this writing, the AIX linker can do this,
7771 as can the GNU linker for PowerPC/64. It is planned to add this feature
7772 to the GNU linker for 32-bit PowerPC systems as well.
7774 In the future, we may cause GCC to ignore all longcall specifications
7775 when the linker is known to generate glue.
7779 Adds support for multithreading with the @dfn{pthreads} library.
7780 This option sets flags for both the preprocessor and linker.
7784 @node Darwin Options
7785 @subsection Darwin Options
7786 @cindex Darwin options
7788 These options are defined for all architectures running the Darwin operating
7789 system. They are useful for compatibility with other Mac OS compilers.
7794 Loads all members of static archive libraries.
7795 See man ld(1) for more information.
7797 @item -arch_errors_fatal
7798 @opindex arch_errors_fatal
7799 Cause the errors having to do with files that have the wrong architecture
7803 @opindex bind_at_load
7804 Causes the output file to be marked such that the dynamic linker will
7805 bind all undefined references when the file is loaded or launched.
7809 Produce a Mach-o bundle format file.
7810 See man ld(1) for more information.
7812 @item -bundle_loader @var{executable}
7813 @opindex bundle_loader
7814 This specifies the @var{executable} that will be loading the build
7815 output file being linked. See man ld(1) for more information.
7817 @item -allowable_client @var{client_name}
7821 @item -compatibility_version
7822 @item -current_version
7823 @item -dependency-file
7825 @item -dylinker_install_name
7828 @item -exported_symbols_list
7830 @item -flat_namespace
7831 @item -force_cpusubtype_ALL
7832 @item -force_flat_namespace
7833 @item -headerpad_max_install_names
7837 @item -keep_private_externs
7839 @item -multiply_defined
7840 @item -multiply_defined_unused
7842 @item -nofixprebinding
7845 @item -noseglinkedit
7846 @item -pagezero_size
7848 @item -prebind_all_twolevel_modules
7849 @item -private_bundle
7850 @item -read_only_relocs
7852 @item -sectobjectsymbols
7856 @item -sectobjectsymbols
7858 @item -seg_addr_table
7859 @item -seg_addr_table_filename
7862 @item -segs_read_only_addr
7863 @item -segs_read_write_addr
7864 @item -single_module
7868 @item -twolevel_namespace
7871 @item -unexported_symbols_list
7872 @item -weak_reference_mismatches
7875 @opindex allowable_client
7877 @opindex client_name
7878 @opindex compatibility_version
7879 @opindex current_version
7880 @opindex dependency-file
7882 @opindex dylinker_install_name
7885 @opindex exported_symbols_list
7887 @opindex flat_namespace
7888 @opindex force_cpusubtype_ALL
7889 @opindex force_flat_namespace
7890 @opindex headerpad_max_install_names
7893 @opindex install_name
7894 @opindex keep_private_externs
7895 @opindex multi_module
7896 @opindex multiply_defined
7897 @opindex multiply_defined_unused
7899 @opindex nofixprebinding
7900 @opindex nomultidefs
7902 @opindex noseglinkedit
7903 @opindex pagezero_size
7905 @opindex prebind_all_twolevel_modules
7906 @opindex private_bundle
7907 @opindex read_only_relocs
7909 @opindex sectobjectsymbols
7913 @opindex sectobjectsymbols
7915 @opindex seg_addr_table
7916 @opindex seg_addr_table_filename
7917 @opindex seglinkedit
7919 @opindex segs_read_only_addr
7920 @opindex segs_read_write_addr
7921 @opindex single_module
7923 @opindex sub_library
7924 @opindex sub_umbrella
7925 @opindex twolevel_namespace
7928 @opindex unexported_symbols_list
7929 @opindex weak_reference_mismatches
7930 @opindex whatsloaded
7932 This options are available for Darwin linker. Darwin linker man page
7933 describes them in detail.
7938 @subsection IBM RT Options
7940 @cindex IBM RT options
7942 These @samp{-m} options are defined for the IBM RT PC:
7946 @opindex min-line-mul
7947 Use an in-line code sequence for integer multiplies. This is the
7950 @item -mcall-lib-mul
7951 @opindex mcall-lib-mul
7952 Call @code{lmul$$} for integer multiples.
7954 @item -mfull-fp-blocks
7955 @opindex mfull-fp-blocks
7956 Generate full-size floating point data blocks, including the minimum
7957 amount of scratch space recommended by IBM@. This is the default.
7959 @item -mminimum-fp-blocks
7960 @opindex mminimum-fp-blocks
7961 Do not include extra scratch space in floating point data blocks. This
7962 results in smaller code, but slower execution, since scratch space must
7963 be allocated dynamically.
7965 @cindex @file{stdarg.h} and RT PC
7966 @item -mfp-arg-in-fpregs
7967 @opindex mfp-arg-in-fpregs
7968 Use a calling sequence incompatible with the IBM calling convention in
7969 which floating point arguments are passed in floating point registers.
7970 Note that @code{stdarg.h} will not work with floating point operands
7971 if this option is specified.
7973 @item -mfp-arg-in-gregs
7974 @opindex mfp-arg-in-gregs
7975 Use the normal calling convention for floating point arguments. This is
7978 @item -mhc-struct-return
7979 @opindex mhc-struct-return
7980 Return structures of more than one word in memory, rather than in a
7981 register. This provides compatibility with the MetaWare HighC (hc)
7982 compiler. Use the option @option{-fpcc-struct-return} for compatibility
7983 with the Portable C Compiler (pcc).
7985 @item -mnohc-struct-return
7986 @opindex mnohc-struct-return
7987 Return some structures of more than one word in registers, when
7988 convenient. This is the default. For compatibility with the
7989 IBM-supplied compilers, use the option @option{-fpcc-struct-return} or the
7990 option @option{-mhc-struct-return}.
7994 @subsection MIPS Options
7995 @cindex MIPS options
7997 These @samp{-m} options are defined for the MIPS family of computers:
8001 @item -march=@var{arch}
8003 Generate code that will run on @var{arch}, which can be the name of a
8004 generic MIPS ISA, or the name of a particular processor.
8006 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
8007 @samp{mips32}, @samp{mips32r2}, and @samp{mips64}.
8008 The processor names are:
8009 @samp{4kc}, @samp{4kp}, @samp{5kc}, @samp{20kc},
8011 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
8012 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000}, @samp{rm7000},
8016 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4300},
8017 @samp{vr5000}, @samp{vr5400} and @samp{vr5500}.
8018 The special value @samp{from-abi} selects the
8019 most compatible architecture for the selected ABI (that is,
8020 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
8022 In processor names, a final @samp{000} can be abbreviated as @samp{k}
8023 (for example, @samp{-march=r2k}). Prefixes are optional, and
8024 @samp{vr} may be written @samp{r}.
8026 GCC defines two macros based on the value of this option. The first
8027 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
8028 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
8029 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
8030 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
8031 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
8033 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
8034 above. In other words, it will have the full prefix and will not
8035 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
8036 the macro names the resolved architecture (either @samp{"mips1"} or
8037 @samp{"mips3"}). It names the default architecture when no
8038 @option{-march} option is given.
8040 @item -mtune=@var{arch}
8042 Optimize for @var{arch}. Among other things, this option controls
8043 the way instructions are scheduled, and the perceived cost of arithmetic
8044 operations. The list of @var{arch} values is the same as for
8047 When this option is not used, GCC will optimize for the processor
8048 specified by @option{-march}. By using @option{-march} and
8049 @option{-mtune} together, it is possible to generate code that will
8050 run on a family of processors, but optimize the code for one
8051 particular member of that family.
8053 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
8054 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
8055 @samp{-march} ones described above.
8059 Equivalent to @samp{-march=mips1}.
8063 Equivalent to @samp{-march=mips2}.
8067 Equivalent to @samp{-march=mips3}.
8071 Equivalent to @samp{-march=mips4}.
8075 Equivalent to @samp{-march=mips32}.
8079 Equivalent to @samp{-march=mips32r2}.
8083 Equivalent to @samp{-march=mips64}.
8086 @itemx -mno-fused-madd
8087 @opindex mfused-madd
8088 @opindex mno-fused-madd
8089 Generate code that uses (does not use) the floating point multiply and
8090 accumulate instructions, when they are available. These instructions
8091 are generated by default if they are available, but this may be
8092 undesirable if the extra precision causes problems or on certain chips
8093 in the mode where denormals are rounded to zero where denormals
8094 generated by multiply and accumulate instructions cause exceptions
8099 Assume that floating point registers are 32 bits wide.
8103 Assume that floating point registers are 64 bits wide.
8107 Assume that general purpose registers are 32 bits wide.
8111 Assume that general purpose registers are 64 bits wide.
8115 Force int and long types to be 64 bits wide. See @option{-mlong32} for an
8116 explanation of the default, and the width of pointers.
8120 Force long types to be 64 bits wide. See @option{-mlong32} for an
8121 explanation of the default, and the width of pointers.
8125 Force long, int, and pointer types to be 32 bits wide.
8127 The default size of ints, longs and pointers depends on the ABI@. All
8128 the supported ABIs use 32-bit ints. The n64 ABI uses 64-bit longs, as
8129 does the 64-bit Cygnus EABI; the others use 32-bit longs. Pointers
8130 are the same size as longs, or the same size as integer registers,
8131 whichever is smaller.
8145 Generate code for the given ABI@.
8147 Note that there are two embedded ABIs: @option{-mabi=eabi}
8148 selects the one defined by Cygnus while @option{-meabi=meabi}
8149 selects the one defined by MIPS@. Both these ABIs have
8150 32-bit and 64-bit variants. Normally, GCC will generate
8151 64-bit code when you select a 64-bit architecture, but you
8152 can use @option{-mgp32} to get 32-bit code instead.
8154 @item -mabi-fake-default
8155 @opindex mabi-fake-default
8156 You don't want to know what this option does. No, really. I mean
8157 it. Move on to the next option.
8159 What? You're still here? Oh, well@enddots{} Ok, here's the deal. GCC
8160 wants the default set of options to get the root of the multilib tree,
8161 and the shared library SONAMEs without any multilib-indicating
8162 suffixes. This is not convenience for @samp{mips64-linux-gnu}, since
8163 we want to default to the N32 ABI, while still being binary-compatible
8164 with @samp{mips-linux-gnu} if you stick to the O32 ABI@. Being
8165 binary-compatible means shared libraries should have the same SONAMEs,
8166 and libraries should live in the same location. Having O32 libraries
8167 in a sub-directory named say @file{o32} is not acceptable.
8169 So we trick GCC into believing that O32 is the default ABI, except
8170 that we override the default with some internal command-line
8171 processing magic. Problem is, if we stopped at that, and you then
8172 created a multilib-aware package that used the output of @command{gcc
8173 -print-multi-lib} to decide which multilibs to build, and how, and
8174 you'd find yourself in an awkward situation when you found out that
8175 some of the options listed ended up mapping to the same multilib, and
8176 none of your libraries was actually built for the multilib that
8177 @option{-print-multi-lib} claims to be the default. So we added this
8178 option that disables the default switcher, falling back to GCC's
8179 original notion of the default library. Confused yet?
8181 For short: don't ever use this option, unless you find it in the list
8182 of additional options to be used when building for multilibs, in the
8183 output of @option{gcc -print-multi-lib}.
8187 Generate code for the MIPS assembler, and invoke @file{mips-tfile} to
8188 add normal debug information. This is the default for all
8189 platforms except for the OSF/1 reference platform, using the OSF/rose
8190 object format. If the either of the @option{-gstabs} or @option{-gstabs+}
8191 switches are used, the @file{mips-tfile} program will encapsulate the
8192 stabs within MIPS ECOFF@.
8196 Generate code for the GNU assembler. This is the default on the OSF/1
8197 reference platform, using the OSF/rose object format. Also, this is
8198 the default if the configure option @option{--with-gnu-as} is used.
8200 @item -msplit-addresses
8201 @itemx -mno-split-addresses
8202 @opindex msplit-addresses
8203 @opindex mno-split-addresses
8204 Generate code to load the high and low parts of address constants separately.
8205 This allows GCC to optimize away redundant loads of the high order
8206 bits of addresses. This optimization requires GNU as and GNU ld.
8207 This optimization is enabled by default for some embedded targets where
8208 GNU as and GNU ld are standard.
8214 The @option{-mrnames} switch says to output code using the MIPS software
8215 names for the registers, instead of the hardware names (ie, @var{a0}
8216 instead of @var{$4}). The only known assembler that supports this option
8217 is the Algorithmics assembler.
8223 The @option{-mmemcpy} switch makes all block moves call the appropriate
8224 string function (@samp{memcpy} or @samp{bcopy}) instead of possibly
8225 generating inline code.
8228 @itemx -mno-mips-tfile
8229 @opindex mmips-tfile
8230 @opindex mno-mips-tfile
8231 The @option{-mno-mips-tfile} switch causes the compiler not
8232 postprocess the object file with the @file{mips-tfile} program,
8233 after the MIPS assembler has generated it to add debug support. If
8234 @file{mips-tfile} is not run, then no local variables will be
8235 available to the debugger. In addition, @file{stage2} and
8236 @file{stage3} objects will have the temporary file names passed to the
8237 assembler embedded in the object file, which means the objects will
8238 not compare the same. The @option{-mno-mips-tfile} switch should only
8239 be used when there are bugs in the @file{mips-tfile} program that
8240 prevents compilation.
8243 @opindex msoft-float
8244 Generate output containing library calls for floating point.
8245 @strong{Warning:} the requisite libraries are not part of GCC@.
8246 Normally the facilities of the machine's usual C compiler are used, but
8247 this can't be done directly in cross-compilation. You must make your
8248 own arrangements to provide suitable library functions for
8252 @opindex mhard-float
8253 Generate output containing floating point instructions. This is the
8254 default if you use the unmodified sources.
8257 @itemx -mno-abicalls
8259 @opindex mno-abicalls
8260 Emit (or do not emit) the pseudo operations @samp{.abicalls},
8261 @samp{.cpload}, and @samp{.cprestore} that some System V.4 ports use for
8262 position independent code.
8268 Lift (or do not lift) the usual restrictions on the size of the global
8271 GCC normally uses a single instruction to load values from the GOT.
8272 While this is relatively efficient, it will only work if the GOT
8273 is smaller than about 64k. Anything larger will cause the linker
8274 to report an error such as:
8276 @cindex relocation truncated to fit (MIPS)
8278 relocation truncated to fit: R_MIPS_GOT16 foobar
8281 If this happens, you should recompile your code with @option{-mxgot}.
8282 It should then work with very large GOTs, although it will also be
8283 less efficient, since it will take three instructions to fetch the
8284 value of a global symbol.
8286 Note that some linkers can create multiple GOTs. If you have such a
8287 linker, you should only need to use @option{-mxgot} when a single object
8288 file accesses more than 64k's worth of GOT entries. Very few do.
8290 These options have no effect unless GCC is generating position
8294 @itemx -mno-long-calls
8295 @opindex mlong-calls
8296 @opindex mno-long-calls
8297 Do all calls with the @samp{JALR} instruction, which requires
8298 loading up a function's address into a register before the call.
8299 You need to use this switch, if you call outside of the current
8300 512 megabyte segment to functions that are not through pointers.
8302 @item -membedded-pic
8303 @itemx -mno-embedded-pic
8304 @opindex membedded-pic
8305 @opindex mno-embedded-pic
8306 Generate PIC code suitable for some embedded systems. All calls are
8307 made using PC relative address, and all data is addressed using the $gp
8308 register. No more than 65536 bytes of global data may be used. This
8309 requires GNU as and GNU ld which do most of the work. This currently
8310 only works on targets which use ECOFF; it does not work with ELF@.
8312 @item -membedded-data
8313 @itemx -mno-embedded-data
8314 @opindex membedded-data
8315 @opindex mno-embedded-data
8316 Allocate variables to the read-only data section first if possible, then
8317 next in the small data section if possible, otherwise in data. This gives
8318 slightly slower code than the default, but reduces the amount of RAM required
8319 when executing, and thus may be preferred for some embedded systems.
8321 @item -muninit-const-in-rodata
8322 @itemx -mno-uninit-const-in-rodata
8323 @opindex muninit-const-in-rodata
8324 @opindex mno-uninit-const-in-rodata
8325 When used together with @option{-membedded-data}, it will always store uninitialized
8326 const variables in the read-only data section.
8328 @item -msingle-float
8329 @itemx -mdouble-float
8330 @opindex msingle-float
8331 @opindex mdouble-float
8332 The @option{-msingle-float} switch tells gcc to assume that the floating
8333 point coprocessor only supports single precision operations, as on the
8334 @samp{r4650} chip. The @option{-mdouble-float} switch permits gcc to use
8335 double precision operations. This is the default.
8341 Permit use of the @samp{mad}, @samp{madu} and @samp{mul} instructions,
8342 as on the @samp{r4650} chip.
8346 Turns on @option{-msingle-float}, @option{-mmad}, and, at least for now,
8347 @option{-mcpu=r4650}.
8353 Enable 16-bit instructions.
8357 Compile code for the processor in little endian mode.
8358 The requisite libraries are assumed to exist.
8362 Compile code for the processor in big endian mode.
8363 The requisite libraries are assumed to exist.
8367 @cindex smaller data references (MIPS)
8368 @cindex gp-relative references (MIPS)
8369 Put global and static items less than or equal to @var{num} bytes into
8370 the small data or bss sections instead of the normal data or bss
8371 section. This allows the assembler to emit one word memory reference
8372 instructions based on the global pointer (@var{gp} or @var{$28}),
8373 instead of the normal two words used. By default, @var{num} is 8 when
8374 the MIPS assembler is used, and 0 when the GNU assembler is used. The
8375 @option{-G @var{num}} switch is also passed to the assembler and linker.
8376 All modules should be compiled with the same @option{-G @var{num}}
8381 Tell the MIPS assembler to not run its preprocessor over user
8382 assembler files (with a @samp{.s} suffix) when assembling them.
8386 Pass an option to gas which will cause nops to be inserted if
8387 the read of the destination register of an mfhi or mflo instruction
8388 occurs in the following two instructions.
8393 Work around certain SB-1 CPU core errata.
8394 (This flag currently works around the SB-1 revision 2
8395 ``F1'' and ``F2'' floating point errata.)
8399 Do not include the default crt0.
8401 @item -mflush-func=@var{func}
8402 @itemx -mno-flush-func
8403 @opindex mflush-func
8404 Specifies the function to call to flush the I and D caches, or to not
8405 call any such function. If called, the function must take the same
8406 arguments as the common @code{_flush_func()}, that is, the address of the
8407 memory range for which the cache is being flushed, the size of the
8408 memory range, and the number 3 (to flush both caches). The default
8409 depends on the target gcc was configured for, but commonly is either
8410 @samp{_flush_func} or @samp{__cpu_flush}.
8412 @item -mbranch-likely
8413 @itemx -mno-branch-likely
8414 @opindex mbranch-likely
8415 @opindex mno-branch-likely
8416 Enable or disable use of Branch Likely instructions, regardless of the
8417 default for the selected architecture. By default, Branch Likely
8418 instructions may be generated if they are supported by the selected
8419 architecture. An exception is for the MIPS32 and MIPS64 architectures
8420 and processors which implement those architectures; for those, Branch
8421 Likely instructions will not be generated by default because the MIPS32
8422 and MIPS64 architectures specifically deprecate their use.
8425 @node i386 and x86-64 Options
8426 @subsection Intel 386 and AMD x86-64 Options
8427 @cindex i386 Options
8428 @cindex x86-64 Options
8429 @cindex Intel 386 Options
8430 @cindex AMD x86-64 Options
8432 These @samp{-m} options are defined for the i386 and x86-64 family of
8436 @item -mtune=@var{cpu-type}
8438 Tune to @var{cpu-type} everything applicable about the generated code, except
8439 for the ABI and the set of available instructions. The choices for
8440 @var{cpu-type} are @samp{i386}, @samp{i486}, @samp{i586}, @samp{i686},
8441 @samp{pentium}, @samp{pentium-mmx}, @samp{pentiumpro}, @samp{pentium2},
8442 @samp{pentium3}, @samp{pentium4}, @samp{k6}, @samp{k6-2}, @samp{k6-3},
8443 @samp{athlon}, @samp{athlon-tbird}, @samp{athlon-4}, @samp{athlon-xp},
8444 @samp{athlon-mp}, @samp{winchip-c6}, @samp{winchip2}, @samp{k8}, @samp{c3}
8447 While picking a specific @var{cpu-type} will schedule things appropriately
8448 for that particular chip, the compiler will not generate any code that
8449 does not run on the i386 without the @option{-march=@var{cpu-type}} option
8450 being used. @samp{i586} is equivalent to @samp{pentium} and @samp{i686}
8451 is equivalent to @samp{pentiumpro}. @samp{k6} and @samp{athlon} are the
8452 AMD chips as opposed to the Intel ones.
8454 @item -march=@var{cpu-type}
8456 Generate instructions for the machine type @var{cpu-type}. The choices
8457 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
8458 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
8460 @item -mcpu=@var{cpu-type}
8462 A deprecated synonym for @option{-mtune}.
8471 @opindex mpentiumpro
8472 These options are synonyms for @option{-mtune=i386}, @option{-mtune=i486},
8473 @option{-mtune=pentium}, and @option{-mtune=pentiumpro} respectively.
8474 These synonyms are deprecated.
8476 @item -mfpmath=@var{unit}
8478 generate floating point arithmetics for selected unit @var{unit}. the choices
8483 Use the standard 387 floating point coprocessor present majority of chips and
8484 emulated otherwise. Code compiled with this option will run almost everywhere.
8485 The temporary results are computed in 80bit precision instead of precision
8486 specified by the type resulting in slightly different results compared to most
8487 of other chips. See @option{-ffloat-store} for more detailed description.
8489 This is the default choice for i386 compiler.
8492 Use scalar floating point instructions present in the SSE instruction set.
8493 This instruction set is supported by Pentium3 and newer chips, in the AMD line
8494 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
8495 instruction set supports only single precision arithmetics, thus the double and
8496 extended precision arithmetics is still done using 387. Later version, present
8497 only in Pentium4 and the future AMD x86-64 chips supports double precision
8500 For i387 you need to use @option{-march=@var{cpu-type}}, @option{-msse} or
8501 @option{-msse2} switches to enable SSE extensions and make this option
8502 effective. For x86-64 compiler, these extensions are enabled by default.
8504 The resulting code should be considerably faster in majority of cases and avoid
8505 the numerical instability problems of 387 code, but may break some existing
8506 code that expects temporaries to be 80bit.
8508 This is the default choice for x86-64 compiler.
8511 Use all SSE extensions enabled by @option{-msse2} as well as the new
8512 SSE extensions in Prescott New Instructions. @option{-mpni} also
8513 enables 2 builtin functions, @code{__builtin_ia32_monitor} and
8514 @code{__builtin_ia32_mwait}, for new instructions @code{monitor} and
8518 Attempt to utilize both instruction sets at once. This effectively double the
8519 amount of available registers and on chips with separate execution units for
8520 387 and SSE the execution resources too. Use this option with care, as it is
8521 still experimental, because gcc register allocator does not model separate
8522 functional units well resulting in instable performance.
8525 @item -masm=@var{dialect}
8526 @opindex masm=@var{dialect}
8527 Output asm instructions using selected @var{dialect}. Supported choices are
8528 @samp{intel} or @samp{att} (the default one).
8533 @opindex mno-ieee-fp
8534 Control whether or not the compiler uses IEEE floating point
8535 comparisons. These handle correctly the case where the result of a
8536 comparison is unordered.
8539 @opindex msoft-float
8540 Generate output containing library calls for floating point.
8541 @strong{Warning:} the requisite libraries are not part of GCC@.
8542 Normally the facilities of the machine's usual C compiler are used, but
8543 this can't be done directly in cross-compilation. You must make your
8544 own arrangements to provide suitable library functions for
8547 On machines where a function returns floating point results in the 80387
8548 register stack, some floating point opcodes may be emitted even if
8549 @option{-msoft-float} is used.
8551 @item -mno-fp-ret-in-387
8552 @opindex mno-fp-ret-in-387
8553 Do not use the FPU registers for return values of functions.
8555 The usual calling convention has functions return values of types
8556 @code{float} and @code{double} in an FPU register, even if there
8557 is no FPU@. The idea is that the operating system should emulate
8560 The option @option{-mno-fp-ret-in-387} causes such values to be returned
8561 in ordinary CPU registers instead.
8563 @item -mno-fancy-math-387
8564 @opindex mno-fancy-math-387
8565 Some 387 emulators do not support the @code{sin}, @code{cos} and
8566 @code{sqrt} instructions for the 387. Specify this option to avoid
8567 generating those instructions. This option is the default on FreeBSD,
8568 OpenBSD and NetBSD@. This option is overridden when @option{-march}
8569 indicates that the target cpu will always have an FPU and so the
8570 instruction will not need emulation. As of revision 2.6.1, these
8571 instructions are not generated unless you also use the
8572 @option{-funsafe-math-optimizations} switch.
8574 @item -malign-double
8575 @itemx -mno-align-double
8576 @opindex malign-double
8577 @opindex mno-align-double
8578 Control whether GCC aligns @code{double}, @code{long double}, and
8579 @code{long long} variables on a two word boundary or a one word
8580 boundary. Aligning @code{double} variables on a two word boundary will
8581 produce code that runs somewhat faster on a @samp{Pentium} at the
8582 expense of more memory.
8584 @strong{Warning:} if you use the @option{-malign-double} switch,
8585 structures containing the above types will be aligned differently than
8586 the published application binary interface specifications for the 386
8587 and will not be binary compatible with structures in code compiled
8588 without that switch.
8590 @item -m96bit-long-double
8591 @item -m128bit-long-double
8592 @opindex m96bit-long-double
8593 @opindex m128bit-long-double
8594 These switches control the size of @code{long double} type. The i386
8595 application binary interface specifies the size to be 96 bits,
8596 so @option{-m96bit-long-double} is the default in 32 bit mode.
8598 Modern architectures (Pentium and newer) would prefer @code{long double}
8599 to be aligned to an 8 or 16 byte boundary. In arrays or structures
8600 conforming to the ABI, this would not be possible. So specifying a
8601 @option{-m128bit-long-double} will align @code{long double}
8602 to a 16 byte boundary by padding the @code{long double} with an additional
8605 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
8606 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
8608 Notice that neither of these options enable any extra precision over the x87
8609 standard of 80 bits for a @code{long double}.
8611 @strong{Warning:} if you override the default value for your target ABI, the
8612 structures and arrays containing @code{long double} will change their size as
8613 well as function calling convention for function taking @code{long double}
8614 will be modified. Hence they will not be binary compatible with arrays or
8615 structures in code compiled without that switch.
8619 @itemx -mno-svr3-shlib
8620 @opindex msvr3-shlib
8621 @opindex mno-svr3-shlib
8622 Control whether GCC places uninitialized local variables into the
8623 @code{bss} or @code{data} segments. @option{-msvr3-shlib} places them
8624 into @code{bss}. These options are meaningful only on System V Release 3.
8628 Use a different function-calling convention, in which functions that
8629 take a fixed number of arguments return with the @code{ret} @var{num}
8630 instruction, which pops their arguments while returning. This saves one
8631 instruction in the caller since there is no need to pop the arguments
8634 You can specify that an individual function is called with this calling
8635 sequence with the function attribute @samp{stdcall}. You can also
8636 override the @option{-mrtd} option by using the function attribute
8637 @samp{cdecl}. @xref{Function Attributes}.
8639 @strong{Warning:} this calling convention is incompatible with the one
8640 normally used on Unix, so you cannot use it if you need to call
8641 libraries compiled with the Unix compiler.
8643 Also, you must provide function prototypes for all functions that
8644 take variable numbers of arguments (including @code{printf});
8645 otherwise incorrect code will be generated for calls to those
8648 In addition, seriously incorrect code will result if you call a
8649 function with too many arguments. (Normally, extra arguments are
8650 harmlessly ignored.)
8652 @item -mregparm=@var{num}
8654 Control how many registers are used to pass integer arguments. By
8655 default, no registers are used to pass arguments, and at most 3
8656 registers can be used. You can control this behavior for a specific
8657 function by using the function attribute @samp{regparm}.
8658 @xref{Function Attributes}.
8660 @strong{Warning:} if you use this switch, and
8661 @var{num} is nonzero, then you must build all modules with the same
8662 value, including any libraries. This includes the system libraries and
8665 @item -mpreferred-stack-boundary=@var{num}
8666 @opindex mpreferred-stack-boundary
8667 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
8668 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
8669 the default is 4 (16 bytes or 128 bits), except when optimizing for code
8670 size (@option{-Os}), in which case the default is the minimum correct
8671 alignment (4 bytes for x86, and 8 bytes for x86-64).
8673 On Pentium and PentiumPro, @code{double} and @code{long double} values
8674 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
8675 suffer significant run time performance penalties. On Pentium III, the
8676 Streaming SIMD Extension (SSE) data type @code{__m128} suffers similar
8677 penalties if it is not 16 byte aligned.
8679 To ensure proper alignment of this values on the stack, the stack boundary
8680 must be as aligned as that required by any value stored on the stack.
8681 Further, every function must be generated such that it keeps the stack
8682 aligned. Thus calling a function compiled with a higher preferred
8683 stack boundary from a function compiled with a lower preferred stack
8684 boundary will most likely misalign the stack. It is recommended that
8685 libraries that use callbacks always use the default setting.
8687 This extra alignment does consume extra stack space, and generally
8688 increases code size. Code that is sensitive to stack space usage, such
8689 as embedded systems and operating system kernels, may want to reduce the
8690 preferred alignment to @option{-mpreferred-stack-boundary=2}.
8708 These switches enable or disable the use of built-in functions that allow
8709 direct access to the MMX, SSE and 3Dnow extensions of the instruction set.
8711 @xref{X86 Built-in Functions}, for details of the functions enabled
8712 and disabled by these switches.
8714 To have SSE/SSE2 instructions generated automatically from floating-point
8715 code, see @option{-mfpmath=sse}.
8718 @itemx -mno-push-args
8720 @opindex mno-push-args
8721 Use PUSH operations to store outgoing parameters. This method is shorter
8722 and usually equally fast as method using SUB/MOV operations and is enabled
8723 by default. In some cases disabling it may improve performance because of
8724 improved scheduling and reduced dependencies.
8726 @item -maccumulate-outgoing-args
8727 @opindex maccumulate-outgoing-args
8728 If enabled, the maximum amount of space required for outgoing arguments will be
8729 computed in the function prologue. This is faster on most modern CPUs
8730 because of reduced dependencies, improved scheduling and reduced stack usage
8731 when preferred stack boundary is not equal to 2. The drawback is a notable
8732 increase in code size. This switch implies @option{-mno-push-args}.
8736 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
8737 on thread-safe exception handling must compile and link all code with the
8738 @option{-mthreads} option. When compiling, @option{-mthreads} defines
8739 @option{-D_MT}; when linking, it links in a special thread helper library
8740 @option{-lmingwthrd} which cleans up per thread exception handling data.
8742 @item -mno-align-stringops
8743 @opindex mno-align-stringops
8744 Do not align destination of inlined string operations. This switch reduces
8745 code size and improves performance in case the destination is already aligned,
8746 but gcc don't know about it.
8748 @item -minline-all-stringops
8749 @opindex minline-all-stringops
8750 By default GCC inlines string operations only when destination is known to be
8751 aligned at least to 4 byte boundary. This enables more inlining, increase code
8752 size, but may improve performance of code that depends on fast memcpy, strlen
8753 and memset for short lengths.
8755 @item -momit-leaf-frame-pointer
8756 @opindex momit-leaf-frame-pointer
8757 Don't keep the frame pointer in a register for leaf functions. This
8758 avoids the instructions to save, set up and restore frame pointers and
8759 makes an extra register available in leaf functions. The option
8760 @option{-fomit-frame-pointer} removes the frame pointer for all functions
8761 which might make debugging harder.
8763 @item -mtls-direct-seg-refs
8764 @itemx -mno-tls-direct-seg-refs
8765 @opindex mtls-direct-seg-refs
8766 Controls whether TLS variables may be accessed with offsets from the
8767 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
8768 or whether the thread base pointer must be added. Whether or not this
8769 is legal depends on the operating system, and whether it maps the
8770 segment to cover the entire TLS area.
8772 For systems that use GNU libc, the default is on.
8775 These @samp{-m} switches are supported in addition to the above
8776 on AMD x86-64 processors in 64-bit environments.
8783 Generate code for a 32-bit or 64-bit environment.
8784 The 32-bit environment sets int, long and pointer to 32 bits and
8785 generates code that runs on any i386 system.
8786 The 64-bit environment sets int to 32 bits and long and pointer
8787 to 64 bits and generates code for AMD's x86-64 architecture.
8790 @opindex no-red-zone
8791 Do not use a so called red zone for x86-64 code. The red zone is mandated
8792 by the x86-64 ABI, it is a 128-byte area beyond the location of the
8793 stack pointer that will not be modified by signal or interrupt handlers
8794 and therefore can be used for temporary data without adjusting the stack
8795 pointer. The flag @option{-mno-red-zone} disables this red zone.
8797 @item -mcmodel=small
8798 @opindex mcmodel=small
8799 Generate code for the small code model: the program and its symbols must
8800 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
8801 Programs can be statically or dynamically linked. This is the default
8804 @item -mcmodel=kernel
8805 @opindex mcmodel=kernel
8806 Generate code for the kernel code model. The kernel runs in the
8807 negative 2 GB of the address space.
8808 This model has to be used for Linux kernel code.
8810 @item -mcmodel=medium
8811 @opindex mcmodel=medium
8812 Generate code for the medium model: The program is linked in the lower 2
8813 GB of the address space but symbols can be located anywhere in the
8814 address space. Programs can be statically or dynamically linked, but
8815 building of shared libraries are not supported with the medium model.
8817 @item -mcmodel=large
8818 @opindex mcmodel=large
8819 Generate code for the large model: This model makes no assumptions
8820 about addresses and sizes of sections. Currently GCC does not implement
8825 @subsection HPPA Options
8826 @cindex HPPA Options
8828 These @samp{-m} options are defined for the HPPA family of computers:
8831 @item -march=@var{architecture-type}
8833 Generate code for the specified architecture. The choices for
8834 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
8835 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
8836 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
8837 architecture option for your machine. Code compiled for lower numbered
8838 architectures will run on higher numbered architectures, but not the
8841 PA 2.0 support currently requires gas snapshot 19990413 or later. The
8842 next release of binutils (current is 2.9.1) will probably contain PA 2.0
8846 @itemx -mpa-risc-1-1
8847 @itemx -mpa-risc-2-0
8848 @opindex mpa-risc-1-0
8849 @opindex mpa-risc-1-1
8850 @opindex mpa-risc-2-0
8851 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
8854 @opindex mbig-switch
8855 Generate code suitable for big switch tables. Use this option only if
8856 the assembler/linker complain about out of range branches within a switch
8859 @item -mjump-in-delay
8860 @opindex mjump-in-delay
8861 Fill delay slots of function calls with unconditional jump instructions
8862 by modifying the return pointer for the function call to be the target
8863 of the conditional jump.
8865 @item -mdisable-fpregs
8866 @opindex mdisable-fpregs
8867 Prevent floating point registers from being used in any manner. This is
8868 necessary for compiling kernels which perform lazy context switching of
8869 floating point registers. If you use this option and attempt to perform
8870 floating point operations, the compiler will abort.
8872 @item -mdisable-indexing
8873 @opindex mdisable-indexing
8874 Prevent the compiler from using indexing address modes. This avoids some
8875 rather obscure problems when compiling MIG generated code under MACH@.
8877 @item -mno-space-regs
8878 @opindex mno-space-regs
8879 Generate code that assumes the target has no space registers. This allows
8880 GCC to generate faster indirect calls and use unscaled index address modes.
8882 Such code is suitable for level 0 PA systems and kernels.
8884 @item -mfast-indirect-calls
8885 @opindex mfast-indirect-calls
8886 Generate code that assumes calls never cross space boundaries. This
8887 allows GCC to emit code which performs faster indirect calls.
8889 This option will not work in the presence of shared libraries or nested
8892 @item -mlong-load-store
8893 @opindex mlong-load-store
8894 Generate 3-instruction load and store sequences as sometimes required by
8895 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
8898 @item -mportable-runtime
8899 @opindex mportable-runtime
8900 Use the portable calling conventions proposed by HP for ELF systems.
8904 Enable the use of assembler directives only GAS understands.
8906 @item -mschedule=@var{cpu-type}
8908 Schedule code according to the constraints for the machine type
8909 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
8910 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
8911 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
8912 proper scheduling option for your machine. The default scheduling is
8916 @opindex mlinker-opt
8917 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
8918 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
8919 linkers in which they give bogus error messages when linking some programs.
8922 @opindex msoft-float
8923 Generate output containing library calls for floating point.
8924 @strong{Warning:} the requisite libraries are not available for all HPPA
8925 targets. Normally the facilities of the machine's usual C compiler are
8926 used, but this cannot be done directly in cross-compilation. You must make
8927 your own arrangements to provide suitable library functions for
8928 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
8929 does provide software floating point support.
8931 @option{-msoft-float} changes the calling convention in the output file;
8932 therefore, it is only useful if you compile @emph{all} of a program with
8933 this option. In particular, you need to compile @file{libgcc.a}, the
8934 library that comes with GCC, with @option{-msoft-float} in order for
8939 Generate the predefine, @code{_SIO}, for server IO. The default is
8940 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
8941 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO. These
8942 options are available under HP-UX and HI-UX.
8946 Use GNU ld specific options. This passes @option{-shared} to ld when
8947 building a shared library. It is the default when GCC is configured,
8948 explicitly or implicitly, with the GNU linker. This option does not
8949 have any affect on which ld is called, it only changes what parameters
8950 are passed to that ld. The ld that is called is determined by the
8951 @option{--with-ld} configure option, gcc's program search path, and
8952 finally by the user's @env{PATH}. The linker used by GCC can be printed
8953 using @samp{which `gcc -print-prog-name=ld`}.
8957 Use HP ld specific options. This passes @option{-b} to ld when building
8958 a shared library and passes @option{+Accept TypeMismatch} to ld on all
8959 links. It is the default when GCC is configured, explicitly or
8960 implicitly, with the HP linker. This option does not have any affect on
8961 which ld is called, it only changes what parameters are passed to that
8962 ld. The ld that is called is determined by the @option{--with-ld}
8963 configure option, gcc's program search path, and finally by the user's
8964 @env{PATH}. The linker used by GCC can be printed using @samp{which
8965 `gcc -print-prog-name=ld`}.
8968 @opindex mno-long-calls
8969 Generate code that uses long call sequences. This ensures that a call
8970 is always able to reach linker generated stubs. The default is to generate
8971 long calls only when the distance from the call site to the beginning
8972 of the function or translation unit, as the case may be, exceeds a
8973 predefined limit set by the branch type being used. The limits for
8974 normal calls are 7,600,000 and 240,000 bytes, respectively for the
8975 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
8978 Distances are measured from the beginning of functions when using the
8979 @option{-ffunction-sections} option, or when using the @option{-mgas}
8980 and @option{-mno-portable-runtime} options together under HP-UX with
8983 It is normally not desirable to use this option as it will degrade
8984 performance. However, it may be useful in large applications,
8985 particularly when partial linking is used to build the application.
8987 The types of long calls used depends on the capabilities of the
8988 assembler and linker, and the type of code being generated. The
8989 impact on systems that support long absolute calls, and long pic
8990 symbol-difference or pc-relative calls should be relatively small.
8991 However, an indirect call is used on 32-bit ELF systems in pic code
8992 and it is quite long.
8996 Suppress the generation of link options to search libdld.sl when the
8997 @option{-static} option is specified on HP-UX 10 and later.
9001 The HP-UX implementation of setlocale in libc has a dependency on
9002 libdld.sl. There isn't an archive version of libdld.sl. Thus,
9003 when the @option{-static} option is specified, special link options
9004 are needed to resolve this dependency.
9006 On HP-UX 10 and later, the GCC driver adds the necessary options to
9007 link with libdld.sl when the @option{-static} option is specified.
9008 This causes the resulting binary to be dynamic. On the 64-bit port,
9009 the linkers generate dynamic binaries by default in any case. The
9010 @option{-nolibdld} option can be used to prevent the GCC driver from
9011 adding these link options.
9015 Add support for multithreading with the @dfn{dce thread} library
9016 under HP-UX. This option sets flags for both the preprocessor and
9020 @node Intel 960 Options
9021 @subsection Intel 960 Options
9023 These @samp{-m} options are defined for the Intel 960 implementations:
9026 @item -m@var{cpu-type}
9034 Assume the defaults for the machine type @var{cpu-type} for some of
9035 the other options, including instruction scheduling, floating point
9036 support, and addressing modes. The choices for @var{cpu-type} are
9037 @samp{ka}, @samp{kb}, @samp{mc}, @samp{ca}, @samp{cf},
9038 @samp{sa}, and @samp{sb}.
9045 @opindex msoft-float
9046 The @option{-mnumerics} option indicates that the processor does support
9047 floating-point instructions. The @option{-msoft-float} option indicates
9048 that floating-point support should not be assumed.
9050 @item -mleaf-procedures
9051 @itemx -mno-leaf-procedures
9052 @opindex mleaf-procedures
9053 @opindex mno-leaf-procedures
9054 Do (or do not) attempt to alter leaf procedures to be callable with the
9055 @code{bal} instruction as well as @code{call}. This will result in more
9056 efficient code for explicit calls when the @code{bal} instruction can be
9057 substituted by the assembler or linker, but less efficient code in other
9058 cases, such as calls via function pointers, or using a linker that doesn't
9059 support this optimization.
9062 @itemx -mno-tail-call
9064 @opindex mno-tail-call
9065 Do (or do not) make additional attempts (beyond those of the
9066 machine-independent portions of the compiler) to optimize tail-recursive
9067 calls into branches. You may not want to do this because the detection of
9068 cases where this is not valid is not totally complete. The default is
9069 @option{-mno-tail-call}.
9071 @item -mcomplex-addr
9072 @itemx -mno-complex-addr
9073 @opindex mcomplex-addr
9074 @opindex mno-complex-addr
9075 Assume (or do not assume) that the use of a complex addressing mode is a
9076 win on this implementation of the i960. Complex addressing modes may not
9077 be worthwhile on the K-series, but they definitely are on the C-series.
9078 The default is currently @option{-mcomplex-addr} for all processors except
9082 @itemx -mno-code-align
9083 @opindex mcode-align
9084 @opindex mno-code-align
9085 Align code to 8-byte boundaries for faster fetching (or don't bother).
9086 Currently turned on by default for C-series implementations only.
9089 @item -mclean-linkage
9090 @itemx -mno-clean-linkage
9091 @opindex mclean-linkage
9092 @opindex mno-clean-linkage
9093 These options are not fully implemented.
9097 @itemx -mic2.0-compat
9098 @itemx -mic3.0-compat
9100 @opindex mic2.0-compat
9101 @opindex mic3.0-compat
9102 Enable compatibility with iC960 v2.0 or v3.0.
9106 @opindex masm-compat
9108 Enable compatibility with the iC960 assembler.
9110 @item -mstrict-align
9111 @itemx -mno-strict-align
9112 @opindex mstrict-align
9113 @opindex mno-strict-align
9114 Do not permit (do permit) unaligned accesses.
9118 Enable structure-alignment compatibility with Intel's gcc release version
9119 1.3 (based on gcc 1.37). This option implies @option{-mstrict-align}.
9121 @item -mlong-double-64
9122 @opindex mlong-double-64
9123 Implement type @samp{long double} as 64-bit floating point numbers.
9124 Without the option @samp{long double} is implemented by 80-bit
9125 floating point numbers. The only reason we have it because there is
9126 no 128-bit @samp{long double} support in @samp{fp-bit.c} yet. So it
9127 is only useful for people using soft-float targets. Otherwise, we
9128 should recommend against use of it.
9132 @node DEC Alpha Options
9133 @subsection DEC Alpha Options
9135 These @samp{-m} options are defined for the DEC Alpha implementations:
9138 @item -mno-soft-float
9140 @opindex mno-soft-float
9141 @opindex msoft-float
9142 Use (do not use) the hardware floating-point instructions for
9143 floating-point operations. When @option{-msoft-float} is specified,
9144 functions in @file{libgcc.a} will be used to perform floating-point
9145 operations. Unless they are replaced by routines that emulate the
9146 floating-point operations, or compiled in such a way as to call such
9147 emulations routines, these routines will issue floating-point
9148 operations. If you are compiling for an Alpha without floating-point
9149 operations, you must ensure that the library is built so as not to call
9152 Note that Alpha implementations without floating-point operations are
9153 required to have floating-point registers.
9158 @opindex mno-fp-regs
9159 Generate code that uses (does not use) the floating-point register set.
9160 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
9161 register set is not used, floating point operands are passed in integer
9162 registers as if they were integers and floating-point results are passed
9163 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
9164 so any function with a floating-point argument or return value called by code
9165 compiled with @option{-mno-fp-regs} must also be compiled with that
9168 A typical use of this option is building a kernel that does not use,
9169 and hence need not save and restore, any floating-point registers.
9173 The Alpha architecture implements floating-point hardware optimized for
9174 maximum performance. It is mostly compliant with the IEEE floating
9175 point standard. However, for full compliance, software assistance is
9176 required. This option generates code fully IEEE compliant code
9177 @emph{except} that the @var{inexact-flag} is not maintained (see below).
9178 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
9179 defined during compilation. The resulting code is less efficient but is
9180 able to correctly support denormalized numbers and exceptional IEEE
9181 values such as not-a-number and plus/minus infinity. Other Alpha
9182 compilers call this option @option{-ieee_with_no_inexact}.
9184 @item -mieee-with-inexact
9185 @opindex mieee-with-inexact
9186 This is like @option{-mieee} except the generated code also maintains
9187 the IEEE @var{inexact-flag}. Turning on this option causes the
9188 generated code to implement fully-compliant IEEE math. In addition to
9189 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
9190 macro. On some Alpha implementations the resulting code may execute
9191 significantly slower than the code generated by default. Since there is
9192 very little code that depends on the @var{inexact-flag}, you should
9193 normally not specify this option. Other Alpha compilers call this
9194 option @option{-ieee_with_inexact}.
9196 @item -mfp-trap-mode=@var{trap-mode}
9197 @opindex mfp-trap-mode
9198 This option controls what floating-point related traps are enabled.
9199 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
9200 The trap mode can be set to one of four values:
9204 This is the default (normal) setting. The only traps that are enabled
9205 are the ones that cannot be disabled in software (e.g., division by zero
9209 In addition to the traps enabled by @samp{n}, underflow traps are enabled
9213 Like @samp{su}, but the instructions are marked to be safe for software
9214 completion (see Alpha architecture manual for details).
9217 Like @samp{su}, but inexact traps are enabled as well.
9220 @item -mfp-rounding-mode=@var{rounding-mode}
9221 @opindex mfp-rounding-mode
9222 Selects the IEEE rounding mode. Other Alpha compilers call this option
9223 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
9228 Normal IEEE rounding mode. Floating point numbers are rounded towards
9229 the nearest machine number or towards the even machine number in case
9233 Round towards minus infinity.
9236 Chopped rounding mode. Floating point numbers are rounded towards zero.
9239 Dynamic rounding mode. A field in the floating point control register
9240 (@var{fpcr}, see Alpha architecture reference manual) controls the
9241 rounding mode in effect. The C library initializes this register for
9242 rounding towards plus infinity. Thus, unless your program modifies the
9243 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
9246 @item -mtrap-precision=@var{trap-precision}
9247 @opindex mtrap-precision
9248 In the Alpha architecture, floating point traps are imprecise. This
9249 means without software assistance it is impossible to recover from a
9250 floating trap and program execution normally needs to be terminated.
9251 GCC can generate code that can assist operating system trap handlers
9252 in determining the exact location that caused a floating point trap.
9253 Depending on the requirements of an application, different levels of
9254 precisions can be selected:
9258 Program precision. This option is the default and means a trap handler
9259 can only identify which program caused a floating point exception.
9262 Function precision. The trap handler can determine the function that
9263 caused a floating point exception.
9266 Instruction precision. The trap handler can determine the exact
9267 instruction that caused a floating point exception.
9270 Other Alpha compilers provide the equivalent options called
9271 @option{-scope_safe} and @option{-resumption_safe}.
9273 @item -mieee-conformant
9274 @opindex mieee-conformant
9275 This option marks the generated code as IEEE conformant. You must not
9276 use this option unless you also specify @option{-mtrap-precision=i} and either
9277 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
9278 is to emit the line @samp{.eflag 48} in the function prologue of the
9279 generated assembly file. Under DEC Unix, this has the effect that
9280 IEEE-conformant math library routines will be linked in.
9282 @item -mbuild-constants
9283 @opindex mbuild-constants
9284 Normally GCC examines a 32- or 64-bit integer constant to
9285 see if it can construct it from smaller constants in two or three
9286 instructions. If it cannot, it will output the constant as a literal and
9287 generate code to load it from the data segment at runtime.
9289 Use this option to require GCC to construct @emph{all} integer constants
9290 using code, even if it takes more instructions (the maximum is six).
9292 You would typically use this option to build a shared library dynamic
9293 loader. Itself a shared library, it must relocate itself in memory
9294 before it can find the variables and constants in its own data segment.
9300 Select whether to generate code to be assembled by the vendor-supplied
9301 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
9319 Indicate whether GCC should generate code to use the optional BWX,
9320 CIX, FIX and MAX instruction sets. The default is to use the instruction
9321 sets supported by the CPU type specified via @option{-mcpu=} option or that
9322 of the CPU on which GCC was built if none was specified.
9327 @opindex mfloat-ieee
9328 Generate code that uses (does not use) VAX F and G floating point
9329 arithmetic instead of IEEE single and double precision.
9331 @item -mexplicit-relocs
9332 @itemx -mno-explicit-relocs
9333 @opindex mexplicit-relocs
9334 @opindex mno-explicit-relocs
9335 Older Alpha assemblers provided no way to generate symbol relocations
9336 except via assembler macros. Use of these macros does not allow
9337 optimal instruction scheduling. GNU binutils as of version 2.12
9338 supports a new syntax that allows the compiler to explicitly mark
9339 which relocations should apply to which instructions. This option
9340 is mostly useful for debugging, as GCC detects the capabilities of
9341 the assembler when it is built and sets the default accordingly.
9345 @opindex msmall-data
9346 @opindex mlarge-data
9347 When @option{-mexplicit-relocs} is in effect, static data is
9348 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
9349 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
9350 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
9351 16-bit relocations off of the @code{$gp} register. This limits the
9352 size of the small data area to 64KB, but allows the variables to be
9353 directly accessed via a single instruction.
9355 The default is @option{-mlarge-data}. With this option the data area
9356 is limited to just below 2GB. Programs that require more than 2GB of
9357 data must use @code{malloc} or @code{mmap} to allocate the data in the
9358 heap instead of in the program's data segment.
9360 When generating code for shared libraries, @option{-fpic} implies
9361 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
9365 @opindex msmall-text
9366 @opindex mlarge-text
9367 When @option{-msmall-text} is used, the compiler assumes that the
9368 code of the entire program (or shared library) fits in 4MB, and is
9369 thus reachable with a branch instruction. When @option{-msmall-data}
9370 is used, the compiler can assume that all local symbols share the
9371 same @code{$gp} value, and thus reduce the number of instructions
9372 required for a function call from 4 to 1.
9374 The default is @option{-mlarge-text}.
9376 @item -mcpu=@var{cpu_type}
9378 Set the instruction set and instruction scheduling parameters for
9379 machine type @var{cpu_type}. You can specify either the @samp{EV}
9380 style name or the corresponding chip number. GCC supports scheduling
9381 parameters for the EV4, EV5 and EV6 family of processors and will
9382 choose the default values for the instruction set from the processor
9383 you specify. If you do not specify a processor type, GCC will default
9384 to the processor on which the compiler was built.
9386 Supported values for @var{cpu_type} are
9392 Schedules as an EV4 and has no instruction set extensions.
9396 Schedules as an EV5 and has no instruction set extensions.
9400 Schedules as an EV5 and supports the BWX extension.
9405 Schedules as an EV5 and supports the BWX and MAX extensions.
9409 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
9413 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
9416 @item -mtune=@var{cpu_type}
9418 Set only the instruction scheduling parameters for machine type
9419 @var{cpu_type}. The instruction set is not changed.
9421 @item -mmemory-latency=@var{time}
9422 @opindex mmemory-latency
9423 Sets the latency the scheduler should assume for typical memory
9424 references as seen by the application. This number is highly
9425 dependent on the memory access patterns used by the application
9426 and the size of the external cache on the machine.
9428 Valid options for @var{time} are
9432 A decimal number representing clock cycles.
9438 The compiler contains estimates of the number of clock cycles for
9439 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
9440 (also called Dcache, Scache, and Bcache), as well as to main memory.
9441 Note that L3 is only valid for EV5.
9446 @node DEC Alpha/VMS Options
9447 @subsection DEC Alpha/VMS Options
9449 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
9452 @item -mvms-return-codes
9453 @opindex mvms-return-codes
9454 Return VMS condition codes from main. The default is to return POSIX
9455 style condition (e.g.@ error) codes.
9458 @node H8/300 Options
9459 @subsection H8/300 Options
9461 These @samp{-m} options are defined for the H8/300 implementations:
9466 Shorten some address references at link time, when possible; uses the
9467 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
9468 ld.info, Using ld}, for a fuller description.
9472 Generate code for the H8/300H@.
9476 Generate code for the H8S@.
9480 Generate code for the H8S and H8/300H in the normal mode. This switch
9481 must be used either with -mh or -ms.
9485 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
9489 Make @code{int} data 32 bits by default.
9493 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
9494 The default for the H8/300H and H8S is to align longs and floats on 4
9496 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
9497 This option has no effect on the H8/300.
9501 @subsection SH Options
9503 These @samp{-m} options are defined for the SH implementations:
9508 Generate code for the SH1.
9512 Generate code for the SH2.
9515 Generate code for the SH2e.
9519 Generate code for the SH3.
9523 Generate code for the SH3e.
9527 Generate code for the SH4 without a floating-point unit.
9529 @item -m4-single-only
9530 @opindex m4-single-only
9531 Generate code for the SH4 with a floating-point unit that only
9532 supports single-precision arithmetic.
9536 Generate code for the SH4 assuming the floating-point unit is in
9537 single-precision mode by default.
9541 Generate code for the SH4.
9545 Compile code for the processor in big endian mode.
9549 Compile code for the processor in little endian mode.
9553 Align doubles at 64-bit boundaries. Note that this changes the calling
9554 conventions, and thus some functions from the standard C library will
9555 not work unless you recompile it first with @option{-mdalign}.
9559 Shorten some address references at link time, when possible; uses the
9560 linker option @option{-relax}.
9564 Use 32-bit offsets in @code{switch} tables. The default is to use
9569 Enable the use of the instruction @code{fmovd}.
9573 Comply with the calling conventions defined by Renesas.
9577 Mark the @code{MAC} register as call-clobbered, even if
9578 @option{-mhitachi} is given.
9582 Increase IEEE-compliance of floating-point code.
9586 Dump instruction size and location in the assembly code.
9590 This option is deprecated. It pads structures to multiple of 4 bytes,
9591 which is incompatible with the SH ABI@.
9595 Optimize for space instead of speed. Implied by @option{-Os}.
9599 When generating position-independent code, emit function calls using
9600 the Global Offset Table instead of the Procedure Linkage Table.
9604 Generate a library function call to invalidate instruction cache
9605 entries, after fixing up a trampoline. This library function call
9606 doesn't assume it can write to the whole memory address space. This
9607 is the default when the target is @code{sh-*-linux*}.
9610 @node System V Options
9611 @subsection Options for System V
9613 These additional options are available on System V Release 4 for
9614 compatibility with other compilers on those systems:
9619 Create a shared object.
9620 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
9624 Identify the versions of each tool used by the compiler, in a
9625 @code{.ident} assembler directive in the output.
9629 Refrain from adding @code{.ident} directives to the output file (this is
9632 @item -YP,@var{dirs}
9634 Search the directories @var{dirs}, and no others, for libraries
9635 specified with @option{-l}.
9639 Look in the directory @var{dir} to find the M4 preprocessor.
9640 The assembler uses this option.
9641 @c This is supposed to go with a -Yd for predefined M4 macro files, but
9642 @c the generic assembler that comes with Solaris takes just -Ym.
9645 @node TMS320C3x/C4x Options
9646 @subsection TMS320C3x/C4x Options
9647 @cindex TMS320C3x/C4x Options
9649 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
9653 @item -mcpu=@var{cpu_type}
9655 Set the instruction set, register set, and instruction scheduling
9656 parameters for machine type @var{cpu_type}. Supported values for
9657 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
9658 @samp{c44}. The default is @samp{c40} to generate code for the
9663 @itemx -msmall-memory
9665 @opindex mbig-memory
9667 @opindex msmall-memory
9669 Generates code for the big or small memory model. The small memory
9670 model assumed that all data fits into one 64K word page. At run-time
9671 the data page (DP) register must be set to point to the 64K page
9672 containing the .bss and .data program sections. The big memory model is
9673 the default and requires reloading of the DP register for every direct
9680 Allow (disallow) allocation of general integer operands into the block
9687 Enable (disable) generation of code using decrement and branch,
9688 DBcond(D), instructions. This is enabled by default for the C4x. To be
9689 on the safe side, this is disabled for the C3x, since the maximum
9690 iteration count on the C3x is @math{2^{23} + 1} (but who iterates loops more than
9691 @math{2^{23}} times on the C3x?). Note that GCC will try to reverse a loop so
9692 that it can utilize the decrement and branch instruction, but will give
9693 up if there is more than one memory reference in the loop. Thus a loop
9694 where the loop counter is decremented can generate slightly more
9695 efficient code, in cases where the RPTB instruction cannot be utilized.
9697 @item -mdp-isr-reload
9699 @opindex mdp-isr-reload
9701 Force the DP register to be saved on entry to an interrupt service
9702 routine (ISR), reloaded to point to the data section, and restored on
9703 exit from the ISR@. This should not be required unless someone has
9704 violated the small memory model by modifying the DP register, say within
9711 For the C3x use the 24-bit MPYI instruction for integer multiplies
9712 instead of a library call to guarantee 32-bit results. Note that if one
9713 of the operands is a constant, then the multiplication will be performed
9714 using shifts and adds. If the @option{-mmpyi} option is not specified for the C3x,
9715 then squaring operations are performed inline instead of a library call.
9718 @itemx -mno-fast-fix
9720 @opindex mno-fast-fix
9721 The C3x/C4x FIX instruction to convert a floating point value to an
9722 integer value chooses the nearest integer less than or equal to the
9723 floating point value rather than to the nearest integer. Thus if the
9724 floating point number is negative, the result will be incorrectly
9725 truncated an additional code is necessary to detect and correct this
9726 case. This option can be used to disable generation of the additional
9727 code required to correct the result.
9733 Enable (disable) generation of repeat block sequences using the RPTB
9734 instruction for zero overhead looping. The RPTB construct is only used
9735 for innermost loops that do not call functions or jump across the loop
9736 boundaries. There is no advantage having nested RPTB loops due to the
9737 overhead required to save and restore the RC, RS, and RE registers.
9738 This is enabled by default with @option{-O2}.
9740 @item -mrpts=@var{count}
9744 Enable (disable) the use of the single instruction repeat instruction
9745 RPTS@. If a repeat block contains a single instruction, and the loop
9746 count can be guaranteed to be less than the value @var{count}, GCC will
9747 emit a RPTS instruction instead of a RPTB@. If no value is specified,
9748 then a RPTS will be emitted even if the loop count cannot be determined
9749 at compile time. Note that the repeated instruction following RPTS does
9750 not have to be reloaded from memory each iteration, thus freeing up the
9751 CPU buses for operands. However, since interrupts are blocked by this
9752 instruction, it is disabled by default.
9754 @item -mloop-unsigned
9755 @itemx -mno-loop-unsigned
9756 @opindex mloop-unsigned
9757 @opindex mno-loop-unsigned
9758 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
9759 is @math{2^{31} + 1} since these instructions test if the iteration count is
9760 negative to terminate the loop. If the iteration count is unsigned
9761 there is a possibility than the @math{2^{31} + 1} maximum iteration count may be
9762 exceeded. This switch allows an unsigned iteration count.
9766 Try to emit an assembler syntax that the TI assembler (asm30) is happy
9767 with. This also enforces compatibility with the API employed by the TI
9768 C3x C compiler. For example, long doubles are passed as structures
9769 rather than in floating point registers.
9775 Generate code that uses registers (stack) for passing arguments to functions.
9776 By default, arguments are passed in registers where possible rather
9777 than by pushing arguments on to the stack.
9779 @item -mparallel-insns
9780 @itemx -mno-parallel-insns
9781 @opindex mparallel-insns
9782 @opindex mno-parallel-insns
9783 Allow the generation of parallel instructions. This is enabled by
9784 default with @option{-O2}.
9786 @item -mparallel-mpy
9787 @itemx -mno-parallel-mpy
9788 @opindex mparallel-mpy
9789 @opindex mno-parallel-mpy
9790 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
9791 provided @option{-mparallel-insns} is also specified. These instructions have
9792 tight register constraints which can pessimize the code generation
9798 @subsection V850 Options
9799 @cindex V850 Options
9801 These @samp{-m} options are defined for V850 implementations:
9805 @itemx -mno-long-calls
9806 @opindex mlong-calls
9807 @opindex mno-long-calls
9808 Treat all calls as being far away (near). If calls are assumed to be
9809 far away, the compiler will always load the functions address up into a
9810 register, and call indirect through the pointer.
9816 Do not optimize (do optimize) basic blocks that use the same index
9817 pointer 4 or more times to copy pointer into the @code{ep} register, and
9818 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
9819 option is on by default if you optimize.
9821 @item -mno-prolog-function
9822 @itemx -mprolog-function
9823 @opindex mno-prolog-function
9824 @opindex mprolog-function
9825 Do not use (do use) external functions to save and restore registers
9826 at the prologue and epilogue of a function. The external functions
9827 are slower, but use less code space if more than one function saves
9828 the same number of registers. The @option{-mprolog-function} option
9829 is on by default if you optimize.
9833 Try to make the code as small as possible. At present, this just turns
9834 on the @option{-mep} and @option{-mprolog-function} options.
9838 Put static or global variables whose size is @var{n} bytes or less into
9839 the tiny data area that register @code{ep} points to. The tiny data
9840 area can hold up to 256 bytes in total (128 bytes for byte references).
9844 Put static or global variables whose size is @var{n} bytes or less into
9845 the small data area that register @code{gp} points to. The small data
9846 area can hold up to 64 kilobytes.
9850 Put static or global variables whose size is @var{n} bytes or less into
9851 the first 32 kilobytes of memory.
9855 Specify that the target processor is the V850.
9858 @opindex mbig-switch
9859 Generate code suitable for big switch tables. Use this option only if
9860 the assembler/linker complain about out of range branches within a switch
9865 This option will cause r2 and r5 to be used in the code generated by
9866 the compiler. This setting is the default.
9869 @opindex mno-app-regs
9870 This option will cause r2 and r5 to be treated as fixed registers.
9874 Specify that the target processor is the V850E1. The preprocessor
9875 constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
9876 this option is used.
9880 Specify that the target processor is the V850E. The preprocessor
9881 constant @samp{__v850e__} will be defined if this option is used.
9883 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
9884 are defined then a default target processor will be chosen and the
9885 relevant @samp{__v850*__} preprocessor constant will be defined.
9887 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
9888 defined, regardless of which processor variant is the target.
9890 @item -mdisable-callt
9891 @opindex mdisable-callt
9892 This option will suppress generation of the CALLT instruction for the
9893 v850e and v850e1 flavors of the v850 architecture. The default is
9894 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
9899 @subsection ARC Options
9902 These options are defined for ARC implementations:
9907 Compile code for little endian mode. This is the default.
9911 Compile code for big endian mode.
9914 @opindex mmangle-cpu
9915 Prepend the name of the cpu to all public symbol names.
9916 In multiple-processor systems, there are many ARC variants with different
9917 instruction and register set characteristics. This flag prevents code
9918 compiled for one cpu to be linked with code compiled for another.
9919 No facility exists for handling variants that are ``almost identical''.
9920 This is an all or nothing option.
9922 @item -mcpu=@var{cpu}
9924 Compile code for ARC variant @var{cpu}.
9925 Which variants are supported depend on the configuration.
9926 All variants support @option{-mcpu=base}, this is the default.
9928 @item -mtext=@var{text-section}
9929 @itemx -mdata=@var{data-section}
9930 @itemx -mrodata=@var{readonly-data-section}
9934 Put functions, data, and readonly data in @var{text-section},
9935 @var{data-section}, and @var{readonly-data-section} respectively
9936 by default. This can be overridden with the @code{section} attribute.
9937 @xref{Variable Attributes}.
9942 @subsection NS32K Options
9943 @cindex NS32K options
9945 These are the @samp{-m} options defined for the 32000 series. The default
9946 values for these options depends on which style of 32000 was selected when
9947 the compiler was configured; the defaults for the most common choices are
9955 Generate output for a 32032. This is the default
9956 when the compiler is configured for 32032 and 32016 based systems.
9962 Generate output for a 32332. This is the default
9963 when the compiler is configured for 32332-based systems.
9969 Generate output for a 32532. This is the default
9970 when the compiler is configured for 32532-based systems.
9974 Generate output containing 32081 instructions for floating point.
9975 This is the default for all systems.
9979 Generate output containing 32381 instructions for floating point. This
9980 also implies @option{-m32081}. The 32381 is only compatible with the 32332
9981 and 32532 cpus. This is the default for the pc532-netbsd configuration.
9985 Try and generate multiply-add floating point instructions @code{polyF}
9986 and @code{dotF}. This option is only available if the @option{-m32381}
9987 option is in effect. Using these instructions requires changes to
9988 register allocation which generally has a negative impact on
9989 performance. This option should only be enabled when compiling code
9990 particularly likely to make heavy use of multiply-add instructions.
9993 @opindex mnomulti-add
9994 Do not try and generate multiply-add floating point instructions
9995 @code{polyF} and @code{dotF}. This is the default on all platforms.
9998 @opindex msoft-float
9999 Generate output containing library calls for floating point.
10000 @strong{Warning:} the requisite libraries may not be available.
10002 @item -mieee-compare
10003 @itemx -mno-ieee-compare
10004 @opindex mieee-compare
10005 @opindex mno-ieee-compare
10006 Control whether or not the compiler uses IEEE floating point
10007 comparisons. These handle correctly the case where the result of a
10008 comparison is unordered.
10009 @strong{Warning:} the requisite kernel support may not be available.
10012 @opindex mnobitfield
10013 Do not use the bit-field instructions. On some machines it is faster to
10014 use shifting and masking operations. This is the default for the pc532.
10018 Do use the bit-field instructions. This is the default for all platforms
10023 Use a different function-calling convention, in which functions
10024 that take a fixed number of arguments return pop their
10025 arguments on return with the @code{ret} instruction.
10027 This calling convention is incompatible with the one normally
10028 used on Unix, so you cannot use it if you need to call libraries
10029 compiled with the Unix compiler.
10031 Also, you must provide function prototypes for all functions that
10032 take variable numbers of arguments (including @code{printf});
10033 otherwise incorrect code will be generated for calls to those
10036 In addition, seriously incorrect code will result if you call a
10037 function with too many arguments. (Normally, extra arguments are
10038 harmlessly ignored.)
10040 This option takes its name from the 680x0 @code{rtd} instruction.
10045 Use a different function-calling convention where the first two arguments
10046 are passed in registers.
10048 This calling convention is incompatible with the one normally
10049 used on Unix, so you cannot use it if you need to call libraries
10050 compiled with the Unix compiler.
10053 @opindex mnoregparam
10054 Do not pass any arguments in registers. This is the default for all
10059 It is OK to use the sb as an index register which is always loaded with
10060 zero. This is the default for the pc532-netbsd target.
10064 The sb register is not available for use or has not been initialized to
10065 zero by the run time system. This is the default for all targets except
10066 the pc532-netbsd. It is also implied whenever @option{-mhimem} or
10067 @option{-fpic} is set.
10071 Many ns32000 series addressing modes use displacements of up to 512MB@.
10072 If an address is above 512MB then displacements from zero can not be used.
10073 This option causes code to be generated which can be loaded above 512MB@.
10074 This may be useful for operating systems or ROM code.
10078 Assume code will be loaded in the first 512MB of virtual address space.
10079 This is the default for all platforms.
10085 @subsection AVR Options
10086 @cindex AVR Options
10088 These options are defined for AVR implementations:
10091 @item -mmcu=@var{mcu}
10093 Specify ATMEL AVR instruction set or MCU type.
10095 Instruction set avr1 is for the minimal AVR core, not supported by the C
10096 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
10097 attiny11, attiny12, attiny15, attiny28).
10099 Instruction set avr2 (default) is for the classic AVR core with up to
10100 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
10101 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
10102 at90c8534, at90s8535).
10104 Instruction set avr3 is for the classic AVR core with up to 128K program
10105 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
10107 Instruction set avr4 is for the enhanced AVR core with up to 8K program
10108 memory space (MCU types: atmega8, atmega83, atmega85).
10110 Instruction set avr5 is for the enhanced AVR core with up to 128K program
10111 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
10112 atmega64, atmega128, at43usb355, at94k).
10116 Output instruction sizes to the asm file.
10118 @item -minit-stack=@var{N}
10119 @opindex minit-stack
10120 Specify the initial stack address, which may be a symbol or numeric value,
10121 @samp{__stack} is the default.
10123 @item -mno-interrupts
10124 @opindex mno-interrupts
10125 Generated code is not compatible with hardware interrupts.
10126 Code size will be smaller.
10128 @item -mcall-prologues
10129 @opindex mcall-prologues
10130 Functions prologues/epilogues expanded as call to appropriate
10131 subroutines. Code size will be smaller.
10133 @item -mno-tablejump
10134 @opindex mno-tablejump
10135 Do not generate tablejump insns which sometimes increase code size.
10138 @opindex mtiny-stack
10139 Change only the low 8 bits of the stack pointer.
10142 @node MCore Options
10143 @subsection MCore Options
10144 @cindex MCore options
10146 These are the @samp{-m} options defined for the Motorola M*Core
10152 @itemx -mno-hardlit
10154 @opindex mno-hardlit
10155 Inline constants into the code stream if it can be done in two
10156 instructions or less.
10162 Use the divide instruction. (Enabled by default).
10164 @item -mrelax-immediate
10165 @itemx -mno-relax-immediate
10166 @opindex mrelax-immediate
10167 @opindex mno-relax-immediate
10168 Allow arbitrary sized immediates in bit operations.
10170 @item -mwide-bitfields
10171 @itemx -mno-wide-bitfields
10172 @opindex mwide-bitfields
10173 @opindex mno-wide-bitfields
10174 Always treat bit-fields as int-sized.
10176 @item -m4byte-functions
10177 @itemx -mno-4byte-functions
10178 @opindex m4byte-functions
10179 @opindex mno-4byte-functions
10180 Force all functions to be aligned to a four byte boundary.
10182 @item -mcallgraph-data
10183 @itemx -mno-callgraph-data
10184 @opindex mcallgraph-data
10185 @opindex mno-callgraph-data
10186 Emit callgraph information.
10189 @itemx -mno-slow-bytes
10190 @opindex mslow-bytes
10191 @opindex mno-slow-bytes
10192 Prefer word access when reading byte quantities.
10194 @item -mlittle-endian
10195 @itemx -mbig-endian
10196 @opindex mlittle-endian
10197 @opindex mbig-endian
10198 Generate code for a little endian target.
10204 Generate code for the 210 processor.
10207 @node IA-64 Options
10208 @subsection IA-64 Options
10209 @cindex IA-64 Options
10211 These are the @samp{-m} options defined for the Intel IA-64 architecture.
10215 @opindex mbig-endian
10216 Generate code for a big endian target. This is the default for HP-UX@.
10218 @item -mlittle-endian
10219 @opindex mlittle-endian
10220 Generate code for a little endian target. This is the default for AIX5
10226 @opindex mno-gnu-as
10227 Generate (or don't) code for the GNU assembler. This is the default.
10228 @c Also, this is the default if the configure option @option{--with-gnu-as}
10234 @opindex mno-gnu-ld
10235 Generate (or don't) code for the GNU linker. This is the default.
10236 @c Also, this is the default if the configure option @option{--with-gnu-ld}
10241 Generate code that does not use a global pointer register. The result
10242 is not position independent code, and violates the IA-64 ABI@.
10244 @item -mvolatile-asm-stop
10245 @itemx -mno-volatile-asm-stop
10246 @opindex mvolatile-asm-stop
10247 @opindex mno-volatile-asm-stop
10248 Generate (or don't) a stop bit immediately before and after volatile asm
10253 Generate code that works around Itanium B step errata.
10255 @item -mregister-names
10256 @itemx -mno-register-names
10257 @opindex mregister-names
10258 @opindex mno-register-names
10259 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
10260 the stacked registers. This may make assembler output more readable.
10266 Disable (or enable) optimizations that use the small data section. This may
10267 be useful for working around optimizer bugs.
10269 @item -mconstant-gp
10270 @opindex mconstant-gp
10271 Generate code that uses a single constant global pointer value. This is
10272 useful when compiling kernel code.
10276 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
10277 This is useful when compiling firmware code.
10279 @item -minline-float-divide-min-latency
10280 @opindex minline-float-divide-min-latency
10281 Generate code for inline divides of floating point values
10282 using the minimum latency algorithm.
10284 @item -minline-float-divide-max-throughput
10285 @opindex minline-float-divide-max-throughput
10286 Generate code for inline divides of floating point values
10287 using the maximum throughput algorithm.
10289 @item -minline-int-divide-min-latency
10290 @opindex minline-int-divide-min-latency
10291 Generate code for inline divides of integer values
10292 using the minimum latency algorithm.
10294 @item -minline-int-divide-max-throughput
10295 @opindex minline-int-divide-max-throughput
10296 Generate code for inline divides of integer values
10297 using the maximum throughput algorithm.
10299 @item -mno-dwarf2-asm
10300 @itemx -mdwarf2-asm
10301 @opindex mno-dwarf2-asm
10302 @opindex mdwarf2-asm
10303 Don't (or do) generate assembler code for the DWARF2 line number debugging
10304 info. This may be useful when not using the GNU assembler.
10306 @item -mfixed-range=@var{register-range}
10307 @opindex mfixed-range
10308 Generate code treating the given register range as fixed registers.
10309 A fixed register is one that the register allocator can not use. This is
10310 useful when compiling kernel code. A register range is specified as
10311 two registers separated by a dash. Multiple register ranges can be
10312 specified separated by a comma.
10314 @item -mearly-stop-bits
10315 @itemx -mno-early-stop-bits
10316 @opindex mearly-stop-bits
10317 @opindex mno-early-stop-bits
10318 Allow stop bits to be placed earlier than immediately preceding the
10319 instruction that triggered the stop bit. This can improve instruction
10320 scheduling, but does not always do so.
10324 @subsection D30V Options
10325 @cindex D30V Options
10327 These @samp{-m} options are defined for D30V implementations:
10332 Link the @samp{.text}, @samp{.data}, @samp{.bss}, @samp{.strings},
10333 @samp{.rodata}, @samp{.rodata1}, @samp{.data1} sections into external
10334 memory, which starts at location @code{0x80000000}.
10337 @opindex mextmemory
10338 Same as the @option{-mextmem} switch.
10342 Link the @samp{.text} section into onchip text memory, which starts at
10343 location @code{0x0}. Also link @samp{.data}, @samp{.bss},
10344 @samp{.strings}, @samp{.rodata}, @samp{.rodata1}, @samp{.data1} sections
10345 into onchip data memory, which starts at location @code{0x20000000}.
10347 @item -mno-asm-optimize
10348 @itemx -masm-optimize
10349 @opindex mno-asm-optimize
10350 @opindex masm-optimize
10351 Disable (enable) passing @option{-O} to the assembler when optimizing.
10352 The assembler uses the @option{-O} option to automatically parallelize
10353 adjacent short instructions where possible.
10355 @item -mbranch-cost=@var{n}
10356 @opindex mbranch-cost
10357 Increase the internal costs of branches to @var{n}. Higher costs means
10358 that the compiler will issue more instructions to avoid doing a branch.
10361 @item -mcond-exec=@var{n}
10362 @opindex mcond-exec
10363 Specify the maximum number of conditionally executed instructions that
10364 replace a branch. The default is 4.
10367 @node S/390 and zSeries Options
10368 @subsection S/390 and zSeries Options
10369 @cindex S/390 and zSeries Options
10371 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
10375 @itemx -msoft-float
10376 @opindex mhard-float
10377 @opindex msoft-float
10378 Use (do not use) the hardware floating-point instructions and registers
10379 for floating-point operations. When @option{-msoft-float} is specified,
10380 functions in @file{libgcc.a} will be used to perform floating-point
10381 operations. When @option{-mhard-float} is specified, the compiler
10382 generates IEEE floating-point instructions. This is the default.
10385 @itemx -mno-backchain
10386 @opindex mbackchain
10387 @opindex mno-backchain
10388 Generate (or do not generate) code which maintains an explicit
10389 backchain within the stack frame that points to the caller's frame.
10390 This is currently needed to allow debugging. The default is to
10391 generate the backchain.
10394 @itemx -mno-small-exec
10395 @opindex msmall-exec
10396 @opindex mno-small-exec
10397 Generate (or do not generate) code using the @code{bras} instruction
10398 to do subroutine calls.
10399 This only works reliably if the total executable size does not
10400 exceed 64k. The default is to use the @code{basr} instruction instead,
10401 which does not have this limitation.
10407 When @option{-m31} is specified, generate code compliant to the
10408 Linux for S/390 ABI@. When @option{-m64} is specified, generate
10409 code compliant to the Linux for zSeries ABI@. This allows GCC in
10410 particular to generate 64-bit instructions. For the @samp{s390}
10411 targets, the default is @option{-m31}, while the @samp{s390x}
10412 targets default to @option{-m64}.
10418 When @option{-mzarch} is specified, generate code using the
10419 instructions available on z/Architecture.
10420 When @option{-mesa} is specified, generate code using the
10421 instructions available on ESA/390. Note that @option{-mesa} is
10422 not possible with @option{-m64}.
10423 When generating code compliant to the Linux for S/390 ABI,
10424 the default is @option{-mesa}. When generating code compliant
10425 to the Linux for zSeries ABI, the default is @option{-mzarch}.
10431 Generate (or do not generate) code using the @code{mvcle} instruction
10432 to perform block moves. When @option{-mno-mvcle} is specified,
10433 use a @code{mvc} loop instead. This is the default.
10439 Print (or do not print) additional debug information when compiling.
10440 The default is to not print debug information.
10442 @item -march=@var{cpu-type}
10444 Generate code that will run on @var{cpu-type}, which is the name of a system
10445 representing a certain processor type. Possible values for
10446 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, and @samp{z990}.
10447 When generating code using the instructions available on z/Architecture,
10448 the default is @option{-march=z900}. Otherwise, the default is
10449 @option{-march=g5}.
10451 @item -mtune=@var{cpu-type}
10453 Tune to @var{cpu-type} everything applicable about the generated code,
10454 except for the ABI and the set of available instructions.
10455 The list of @var{cpu-type} values is the same as for @option{-march}.
10456 The default is the value used for @option{-march}.
10459 @itemx -mno-fused-madd
10460 @opindex mfused-madd
10461 @opindex mno-fused-madd
10462 Generate code that uses (does not use) the floating point multiply and
10463 accumulate instructions. These instructions are generated by default if
10464 hardware floating point is used.
10468 @subsection CRIS Options
10469 @cindex CRIS Options
10471 These options are defined specifically for the CRIS ports.
10474 @item -march=@var{architecture-type}
10475 @itemx -mcpu=@var{architecture-type}
10478 Generate code for the specified architecture. The choices for
10479 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
10480 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX.
10481 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
10484 @item -mtune=@var{architecture-type}
10486 Tune to @var{architecture-type} everything applicable about the generated
10487 code, except for the ABI and the set of available instructions. The
10488 choices for @var{architecture-type} are the same as for
10489 @option{-march=@var{architecture-type}}.
10491 @item -mmax-stack-frame=@var{n}
10492 @opindex mmax-stack-frame
10493 Warn when the stack frame of a function exceeds @var{n} bytes.
10495 @item -melinux-stacksize=@var{n}
10496 @opindex melinux-stacksize
10497 Only available with the @samp{cris-axis-aout} target. Arranges for
10498 indications in the program to the kernel loader that the stack of the
10499 program should be set to @var{n} bytes.
10505 The options @option{-metrax4} and @option{-metrax100} are synonyms for
10506 @option{-march=v3} and @option{-march=v8} respectively.
10510 Enable CRIS-specific verbose debug-related information in the assembly
10511 code. This option also has the effect to turn off the @samp{#NO_APP}
10512 formatted-code indicator to the assembler at the beginning of the
10517 Do not use condition-code results from previous instruction; always emit
10518 compare and test instructions before use of condition codes.
10520 @item -mno-side-effects
10521 @opindex mno-side-effects
10522 Do not emit instructions with side-effects in addressing modes other than
10525 @item -mstack-align
10526 @itemx -mno-stack-align
10527 @itemx -mdata-align
10528 @itemx -mno-data-align
10529 @itemx -mconst-align
10530 @itemx -mno-const-align
10531 @opindex mstack-align
10532 @opindex mno-stack-align
10533 @opindex mdata-align
10534 @opindex mno-data-align
10535 @opindex mconst-align
10536 @opindex mno-const-align
10537 These options (no-options) arranges (eliminate arrangements) for the
10538 stack-frame, individual data and constants to be aligned for the maximum
10539 single data access size for the chosen CPU model. The default is to
10540 arrange for 32-bit alignment. ABI details such as structure layout are
10541 not affected by these options.
10549 Similar to the stack- data- and const-align options above, these options
10550 arrange for stack-frame, writable data and constants to all be 32-bit,
10551 16-bit or 8-bit aligned. The default is 32-bit alignment.
10553 @item -mno-prologue-epilogue
10554 @itemx -mprologue-epilogue
10555 @opindex mno-prologue-epilogue
10556 @opindex mprologue-epilogue
10557 With @option{-mno-prologue-epilogue}, the normal function prologue and
10558 epilogue that sets up the stack-frame are omitted and no return
10559 instructions or return sequences are generated in the code. Use this
10560 option only together with visual inspection of the compiled code: no
10561 warnings or errors are generated when call-saved registers must be saved,
10562 or storage for local variable needs to be allocated.
10566 @opindex mno-gotplt
10568 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
10569 instruction sequences that load addresses for functions from the PLT part
10570 of the GOT rather than (traditional on other architectures) calls to the
10571 PLT. The default is @option{-mgotplt}.
10575 Legacy no-op option only recognized with the cris-axis-aout target.
10579 Legacy no-op option only recognized with the cris-axis-elf and
10580 cris-axis-linux-gnu targets.
10584 Only recognized with the cris-axis-aout target, where it selects a
10585 GNU/linux-like multilib, include files and instruction set for
10586 @option{-march=v8}.
10590 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
10594 This option, recognized for the cris-axis-aout and cris-axis-elf arranges
10595 to link with input-output functions from a simulator library. Code,
10596 initialized data and zero-initialized data are allocated consecutively.
10600 Like @option{-sim}, but pass linker options to locate initialized data at
10601 0x40000000 and zero-initialized data at 0x80000000.
10605 @subsection MMIX Options
10606 @cindex MMIX Options
10608 These options are defined for the MMIX:
10612 @itemx -mno-libfuncs
10614 @opindex mno-libfuncs
10615 Specify that intrinsic library functions are being compiled, passing all
10616 values in registers, no matter the size.
10619 @itemx -mno-epsilon
10621 @opindex mno-epsilon
10622 Generate floating-point comparison instructions that compare with respect
10623 to the @code{rE} epsilon register.
10625 @item -mabi=mmixware
10627 @opindex mabi-mmixware
10629 Generate code that passes function parameters and return values that (in
10630 the called function) are seen as registers @code{$0} and up, as opposed to
10631 the GNU ABI which uses global registers @code{$231} and up.
10633 @item -mzero-extend
10634 @itemx -mno-zero-extend
10635 @opindex mzero-extend
10636 @opindex mno-zero-extend
10637 When reading data from memory in sizes shorter than 64 bits, use (do not
10638 use) zero-extending load instructions by default, rather than
10639 sign-extending ones.
10642 @itemx -mno-knuthdiv
10644 @opindex mno-knuthdiv
10645 Make the result of a division yielding a remainder have the same sign as
10646 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
10647 remainder follows the sign of the dividend. Both methods are
10648 arithmetically valid, the latter being almost exclusively used.
10650 @item -mtoplevel-symbols
10651 @itemx -mno-toplevel-symbols
10652 @opindex mtoplevel-symbols
10653 @opindex mno-toplevel-symbols
10654 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
10655 code can be used with the @code{PREFIX} assembly directive.
10659 Generate an executable in the ELF format, rather than the default
10660 @samp{mmo} format used by the @command{mmix} simulator.
10662 @item -mbranch-predict
10663 @itemx -mno-branch-predict
10664 @opindex mbranch-predict
10665 @opindex mno-branch-predict
10666 Use (do not use) the probable-branch instructions, when static branch
10667 prediction indicates a probable branch.
10669 @item -mbase-addresses
10670 @itemx -mno-base-addresses
10671 @opindex mbase-addresses
10672 @opindex mno-base-addresses
10673 Generate (do not generate) code that uses @emph{base addresses}. Using a
10674 base address automatically generates a request (handled by the assembler
10675 and the linker) for a constant to be set up in a global register. The
10676 register is used for one or more base address requests within the range 0
10677 to 255 from the value held in the register. The generally leads to short
10678 and fast code, but the number of different data items that can be
10679 addressed is limited. This means that a program that uses lots of static
10680 data may require @option{-mno-base-addresses}.
10682 @item -msingle-exit
10683 @itemx -mno-single-exit
10684 @opindex msingle-exit
10685 @opindex mno-single-exit
10686 Force (do not force) generated code to have a single exit point in each
10690 @node PDP-11 Options
10691 @subsection PDP-11 Options
10692 @cindex PDP-11 Options
10694 These options are defined for the PDP-11:
10699 Use hardware FPP floating point. This is the default. (FIS floating
10700 point on the PDP-11/40 is not supported.)
10703 @opindex msoft-float
10704 Do not use hardware floating point.
10708 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
10712 Return floating-point results in memory. This is the default.
10716 Generate code for a PDP-11/40.
10720 Generate code for a PDP-11/45. This is the default.
10724 Generate code for a PDP-11/10.
10726 @item -mbcopy-builtin
10727 @opindex bcopy-builtin
10728 Use inline @code{movstrhi} patterns for copying memory. This is the
10733 Do not use inline @code{movstrhi} patterns for copying memory.
10739 Use 16-bit @code{int}. This is the default.
10745 Use 32-bit @code{int}.
10748 @itemx -mno-float32
10750 @opindex mno-float32
10751 Use 64-bit @code{float}. This is the default.
10756 @opindex mno-float64
10757 Use 32-bit @code{float}.
10761 Use @code{abshi2} pattern. This is the default.
10765 Do not use @code{abshi2} pattern.
10767 @item -mbranch-expensive
10768 @opindex mbranch-expensive
10769 Pretend that branches are expensive. This is for experimenting with
10770 code generation only.
10772 @item -mbranch-cheap
10773 @opindex mbranch-cheap
10774 Do not pretend that branches are expensive. This is the default.
10778 Generate code for a system with split I&D.
10782 Generate code for a system without split I&D. This is the default.
10786 Use Unix assembler syntax. This is the default when configured for
10787 @samp{pdp11-*-bsd}.
10791 Use DEC assembler syntax. This is the default when configured for any
10792 PDP-11 target other than @samp{pdp11-*-bsd}.
10795 @node Xstormy16 Options
10796 @subsection Xstormy16 Options
10797 @cindex Xstormy16 Options
10799 These options are defined for Xstormy16:
10804 Choose startup files and linker script suitable for the simulator.
10808 @subsection FRV Options
10809 @cindex FRV Options
10815 Only use the first 32 general purpose registers.
10820 Use all 64 general purpose registers.
10825 Use only the first 32 floating point registers.
10830 Use all 64 floating point registers
10833 @opindex mhard-float
10835 Use hardware instructions for floating point operations.
10838 @opindex msoft-float
10840 Use library routines for floating point operations.
10845 Dynamically allocate condition code registers.
10850 Do not try to dynamically allocate condition code registers, only
10851 use @code{icc0} and @code{fcc0}.
10856 Change ABI to use double word insns.
10861 Do not use double word instructions.
10866 Use floating point double instructions.
10869 @opindex mno-double
10871 Do not use floating point double instructions.
10876 Use media instructions.
10881 Do not use media instructions.
10886 Use multiply and add/subtract instructions.
10889 @opindex mno-muladd
10891 Do not use multiply and add/subtract instructions.
10893 @item -mlibrary-pic
10894 @opindex mlibrary-pic
10896 Enable PIC support for building libraries
10901 Use only the first four media accumulator registers.
10906 Use all eight media accumulator registers.
10911 Pack VLIW instructions.
10916 Do not pack VLIW instructions.
10919 @opindex mno-eflags
10921 Do not mark ABI switches in e_flags.
10924 @opindex mcond-move
10926 Enable the use of conditional-move instructions (default).
10928 This switch is mainly for debugging the compiler and will likely be removed
10929 in a future version.
10931 @item -mno-cond-move
10932 @opindex mno-cond-move
10934 Disable the use of conditional-move instructions.
10936 This switch is mainly for debugging the compiler and will likely be removed
10937 in a future version.
10942 Enable the use of conditional set instructions (default).
10944 This switch is mainly for debugging the compiler and will likely be removed
10945 in a future version.
10950 Disable the use of conditional set instructions.
10952 This switch is mainly for debugging the compiler and will likely be removed
10953 in a future version.
10956 @opindex mcond-exec
10958 Enable the use of conditional execution (default).
10960 This switch is mainly for debugging the compiler and will likely be removed
10961 in a future version.
10963 @item -mno-cond-exec
10964 @opindex mno-cond-exec
10966 Disable the use of conditional execution.
10968 This switch is mainly for debugging the compiler and will likely be removed
10969 in a future version.
10971 @item -mvliw-branch
10972 @opindex mvliw-branch
10974 Run a pass to pack branches into VLIW instructions (default).
10976 This switch is mainly for debugging the compiler and will likely be removed
10977 in a future version.
10979 @item -mno-vliw-branch
10980 @opindex mno-vliw-branch
10982 Do not run a pass to pack branches into VLIW instructions.
10984 This switch is mainly for debugging the compiler and will likely be removed
10985 in a future version.
10987 @item -mmulti-cond-exec
10988 @opindex mmulti-cond-exec
10990 Enable optimization of @code{&&} and @code{||} in conditional execution
10993 This switch is mainly for debugging the compiler and will likely be removed
10994 in a future version.
10996 @item -mno-multi-cond-exec
10997 @opindex mno-multi-cond-exec
10999 Disable optimization of @code{&&} and @code{||} in conditional execution.
11001 This switch is mainly for debugging the compiler and will likely be removed
11002 in a future version.
11004 @item -mnested-cond-exec
11005 @opindex mnested-cond-exec
11007 Enable nested conditional execution optimizations (default).
11009 This switch is mainly for debugging the compiler and will likely be removed
11010 in a future version.
11012 @item -mno-nested-cond-exec
11013 @opindex mno-nested-cond-exec
11015 Disable nested conditional execution optimizations.
11017 This switch is mainly for debugging the compiler and will likely be removed
11018 in a future version.
11020 @item -mtomcat-stats
11021 @opindex mtomcat-stats
11023 Cause gas to print out tomcat statistics.
11025 @item -mcpu=@var{cpu}
11028 Select the processor type for which to generate code. Possible values are
11029 @samp{simple}, @samp{tomcat}, @samp{fr500}, @samp{fr400}, @samp{fr300},
11034 @node Xtensa Options
11035 @subsection Xtensa Options
11036 @cindex Xtensa Options
11038 These options are supported for Xtensa targets:
11042 @itemx -mno-const16
11044 @opindex mno-const16
11045 Enable or disable use of @code{CONST16} instructions for loading
11046 constant values. The @code{CONST16} instruction is currently not a
11047 standard option from Tensilica. When enabled, @code{CONST16}
11048 instructions are always used in place of the standard @code{L32R}
11049 instructions. The use of @code{CONST16} is enabled by default only if
11050 the @code{L32R} instruction is not available.
11053 @itemx -mno-fused-madd
11054 @opindex mfused-madd
11055 @opindex mno-fused-madd
11056 Enable or disable use of fused multiply/add and multiply/subtract
11057 instructions in the floating-point option. This has no effect if the
11058 floating-point option is not also enabled. Disabling fused multiply/add
11059 and multiply/subtract instructions forces the compiler to use separate
11060 instructions for the multiply and add/subtract operations. This may be
11061 desirable in some cases where strict IEEE 754-compliant results are
11062 required: the fused multiply add/subtract instructions do not round the
11063 intermediate result, thereby producing results with @emph{more} bits of
11064 precision than specified by the IEEE standard. Disabling fused multiply
11065 add/subtract instructions also ensures that the program output is not
11066 sensitive to the compiler's ability to combine multiply and add/subtract
11069 @item -mtext-section-literals
11070 @itemx -mno-text-section-literals
11071 @opindex mtext-section-literals
11072 @opindex mno-text-section-literals
11073 Control the treatment of literal pools. The default is
11074 @option{-mno-text-section-literals}, which places literals in a separate
11075 section in the output file. This allows the literal pool to be placed
11076 in a data RAM/ROM, and it also allows the linker to combine literal
11077 pools from separate object files to remove redundant literals and
11078 improve code size. With @option{-mtext-section-literals}, the literals
11079 are interspersed in the text section in order to keep them as close as
11080 possible to their references. This may be necessary for large assembly
11083 @item -mtarget-align
11084 @itemx -mno-target-align
11085 @opindex mtarget-align
11086 @opindex mno-target-align
11087 When this option is enabled, GCC instructs the assembler to
11088 automatically align instructions to reduce branch penalties at the
11089 expense of some code density. The assembler attempts to widen density
11090 instructions to align branch targets and the instructions following call
11091 instructions. If there are not enough preceding safe density
11092 instructions to align a target, no widening will be performed. The
11093 default is @option{-mtarget-align}. These options do not affect the
11094 treatment of auto-aligned instructions like @code{LOOP}, which the
11095 assembler will always align, either by widening density instructions or
11096 by inserting no-op instructions.
11099 @itemx -mno-longcalls
11100 @opindex mlongcalls
11101 @opindex mno-longcalls
11102 When this option is enabled, GCC instructs the assembler to translate
11103 direct calls to indirect calls unless it can determine that the target
11104 of a direct call is in the range allowed by the call instruction. This
11105 translation typically occurs for calls to functions in other source
11106 files. Specifically, the assembler translates a direct @code{CALL}
11107 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
11108 The default is @option{-mno-longcalls}. This option should be used in
11109 programs where the call target can potentially be out of range. This
11110 option is implemented in the assembler, not the compiler, so the
11111 assembly code generated by GCC will still show direct call
11112 instructions---look at the disassembled object code to see the actual
11113 instructions. Note that the assembler will use an indirect call for
11114 every cross-file call, not just those that really will be out of range.
11117 @node Code Gen Options
11118 @section Options for Code Generation Conventions
11119 @cindex code generation conventions
11120 @cindex options, code generation
11121 @cindex run-time options
11123 These machine-independent options control the interface conventions
11124 used in code generation.
11126 Most of them have both positive and negative forms; the negative form
11127 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
11128 one of the forms is listed---the one which is not the default. You
11129 can figure out the other form by either removing @samp{no-} or adding
11133 @item -fbounds-check
11134 @opindex fbounds-check
11135 For front-ends that support it, generate additional code to check that
11136 indices used to access arrays are within the declared range. This is
11137 currently only supported by the Java and Fortran 77 front-ends, where
11138 this option defaults to true and false respectively.
11142 This option generates traps for signed overflow on addition, subtraction,
11143 multiplication operations.
11147 This option instructs the compiler to assume that signed arithmetic
11148 overflow of addition, subtraction and multiplication wraps around
11149 using twos-complement representation. This flag enables some optimizations
11150 and disables other. This option is enabled by default for the Java
11151 front-end, as required by the Java language specification.
11154 @opindex fexceptions
11155 Enable exception handling. Generates extra code needed to propagate
11156 exceptions. For some targets, this implies GCC will generate frame
11157 unwind information for all functions, which can produce significant data
11158 size overhead, although it does not affect execution. If you do not
11159 specify this option, GCC will enable it by default for languages like
11160 C++ which normally require exception handling, and disable it for
11161 languages like C that do not normally require it. However, you may need
11162 to enable this option when compiling C code that needs to interoperate
11163 properly with exception handlers written in C++. You may also wish to
11164 disable this option if you are compiling older C++ programs that don't
11165 use exception handling.
11167 @item -fnon-call-exceptions
11168 @opindex fnon-call-exceptions
11169 Generate code that allows trapping instructions to throw exceptions.
11170 Note that this requires platform-specific runtime support that does
11171 not exist everywhere. Moreover, it only allows @emph{trapping}
11172 instructions to throw exceptions, i.e.@: memory references or floating
11173 point instructions. It does not allow exceptions to be thrown from
11174 arbitrary signal handlers such as @code{SIGALRM}.
11176 @item -funwind-tables
11177 @opindex funwind-tables
11178 Similar to @option{-fexceptions}, except that it will just generate any needed
11179 static data, but will not affect the generated code in any other way.
11180 You will normally not enable this option; instead, a language processor
11181 that needs this handling would enable it on your behalf.
11183 @item -fasynchronous-unwind-tables
11184 @opindex funwind-tables
11185 Generate unwind table in dwarf2 format, if supported by target machine. The
11186 table is exact at each instruction boundary, so it can be used for stack
11187 unwinding from asynchronous events (such as debugger or garbage collector).
11189 @item -fpcc-struct-return
11190 @opindex fpcc-struct-return
11191 Return ``short'' @code{struct} and @code{union} values in memory like
11192 longer ones, rather than in registers. This convention is less
11193 efficient, but it has the advantage of allowing intercallability between
11194 GCC-compiled files and files compiled with other compilers, particularly
11195 the Portable C Compiler (pcc).
11197 The precise convention for returning structures in memory depends
11198 on the target configuration macros.
11200 Short structures and unions are those whose size and alignment match
11201 that of some integer type.
11203 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
11204 switch is not binary compatible with code compiled with the
11205 @option{-freg-struct-return} switch.
11206 Use it to conform to a non-default application binary interface.
11208 @item -freg-struct-return
11209 @opindex freg-struct-return
11210 Return @code{struct} and @code{union} values in registers when possible.
11211 This is more efficient for small structures than
11212 @option{-fpcc-struct-return}.
11214 If you specify neither @option{-fpcc-struct-return} nor
11215 @option{-freg-struct-return}, GCC defaults to whichever convention is
11216 standard for the target. If there is no standard convention, GCC
11217 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
11218 the principal compiler. In those cases, we can choose the standard, and
11219 we chose the more efficient register return alternative.
11221 @strong{Warning:} code compiled with the @option{-freg-struct-return}
11222 switch is not binary compatible with code compiled with the
11223 @option{-fpcc-struct-return} switch.
11224 Use it to conform to a non-default application binary interface.
11226 @item -fshort-enums
11227 @opindex fshort-enums
11228 Allocate to an @code{enum} type only as many bytes as it needs for the
11229 declared range of possible values. Specifically, the @code{enum} type
11230 will be equivalent to the smallest integer type which has enough room.
11232 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
11233 code that is not binary compatible with code generated without that switch.
11234 Use it to conform to a non-default application binary interface.
11236 @item -fshort-double
11237 @opindex fshort-double
11238 Use the same size for @code{double} as for @code{float}.
11240 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
11241 code that is not binary compatible with code generated without that switch.
11242 Use it to conform to a non-default application binary interface.
11244 @item -fshort-wchar
11245 @opindex fshort-wchar
11246 Override the underlying type for @samp{wchar_t} to be @samp{short
11247 unsigned int} instead of the default for the target. This option is
11248 useful for building programs to run under WINE@.
11250 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
11251 code that is not binary compatible with code generated without that switch.
11252 Use it to conform to a non-default application binary interface.
11254 @item -fshared-data
11255 @opindex fshared-data
11256 Requests that the data and non-@code{const} variables of this
11257 compilation be shared data rather than private data. The distinction
11258 makes sense only on certain operating systems, where shared data is
11259 shared between processes running the same program, while private data
11260 exists in one copy per process.
11263 @opindex fno-common
11264 In C, allocate even uninitialized global variables in the data section of the
11265 object file, rather than generating them as common blocks. This has the
11266 effect that if the same variable is declared (without @code{extern}) in
11267 two different compilations, you will get an error when you link them.
11268 The only reason this might be useful is if you wish to verify that the
11269 program will work on other systems which always work this way.
11273 Ignore the @samp{#ident} directive.
11275 @item -fno-gnu-linker
11276 @opindex fno-gnu-linker
11277 Do not output global initializations (such as C++ constructors and
11278 destructors) in the form used by the GNU linker (on systems where the GNU
11279 linker is the standard method of handling them). Use this option when
11280 you want to use a non-GNU linker, which also requires using the
11281 @command{collect2} program to make sure the system linker includes
11282 constructors and destructors. (@command{collect2} is included in the GCC
11283 distribution.) For systems which @emph{must} use @command{collect2}, the
11284 compiler driver @command{gcc} is configured to do this automatically.
11286 @item -finhibit-size-directive
11287 @opindex finhibit-size-directive
11288 Don't output a @code{.size} assembler directive, or anything else that
11289 would cause trouble if the function is split in the middle, and the
11290 two halves are placed at locations far apart in memory. This option is
11291 used when compiling @file{crtstuff.c}; you should not need to use it
11294 @item -fverbose-asm
11295 @opindex fverbose-asm
11296 Put extra commentary information in the generated assembly code to
11297 make it more readable. This option is generally only of use to those
11298 who actually need to read the generated assembly code (perhaps while
11299 debugging the compiler itself).
11301 @option{-fno-verbose-asm}, the default, causes the
11302 extra information to be omitted and is useful when comparing two assembler
11307 @cindex global offset table
11309 Generate position-independent code (PIC) suitable for use in a shared
11310 library, if supported for the target machine. Such code accesses all
11311 constant addresses through a global offset table (GOT)@. The dynamic
11312 loader resolves the GOT entries when the program starts (the dynamic
11313 loader is not part of GCC; it is part of the operating system). If
11314 the GOT size for the linked executable exceeds a machine-specific
11315 maximum size, you get an error message from the linker indicating that
11316 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
11317 instead. (These maximums are 16k on the m88k, 8k on the SPARC, and 32k
11318 on the m68k and RS/6000. The 386 has no such limit.)
11320 Position-independent code requires special support, and therefore works
11321 only on certain machines. For the 386, GCC supports PIC for System V
11322 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
11323 position-independent.
11327 If supported for the target machine, emit position-independent code,
11328 suitable for dynamic linking and avoiding any limit on the size of the
11329 global offset table. This option makes a difference on the m68k, m88k,
11332 Position-independent code requires special support, and therefore works
11333 only on certain machines.
11339 These options are similar to @option{-fpic} and @option{-fPIC}, but
11340 generated position independent code can be only linked into executables.
11341 Usually these options are used when @option{-pie} GCC option will be
11342 used during linking.
11344 @item -ffixed-@var{reg}
11346 Treat the register named @var{reg} as a fixed register; generated code
11347 should never refer to it (except perhaps as a stack pointer, frame
11348 pointer or in some other fixed role).
11350 @var{reg} must be the name of a register. The register names accepted
11351 are machine-specific and are defined in the @code{REGISTER_NAMES}
11352 macro in the machine description macro file.
11354 This flag does not have a negative form, because it specifies a
11357 @item -fcall-used-@var{reg}
11358 @opindex fcall-used
11359 Treat the register named @var{reg} as an allocable register that is
11360 clobbered by function calls. It may be allocated for temporaries or
11361 variables that do not live across a call. Functions compiled this way
11362 will not save and restore the register @var{reg}.
11364 It is an error to used this flag with the frame pointer or stack pointer.
11365 Use of this flag for other registers that have fixed pervasive roles in
11366 the machine's execution model will produce disastrous results.
11368 This flag does not have a negative form, because it specifies a
11371 @item -fcall-saved-@var{reg}
11372 @opindex fcall-saved
11373 Treat the register named @var{reg} as an allocable register saved by
11374 functions. It may be allocated even for temporaries or variables that
11375 live across a call. Functions compiled this way will save and restore
11376 the register @var{reg} if they use it.
11378 It is an error to used this flag with the frame pointer or stack pointer.
11379 Use of this flag for other registers that have fixed pervasive roles in
11380 the machine's execution model will produce disastrous results.
11382 A different sort of disaster will result from the use of this flag for
11383 a register in which function values may be returned.
11385 This flag does not have a negative form, because it specifies a
11388 @item -fpack-struct
11389 @opindex fpack-struct
11390 Pack all structure members together without holes.
11392 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
11393 code that is not binary compatible with code generated without that switch.
11394 Additionally, it makes the code suboptimal.
11395 Use it to conform to a non-default application binary interface.
11397 @item -finstrument-functions
11398 @opindex finstrument-functions
11399 Generate instrumentation calls for entry and exit to functions. Just
11400 after function entry and just before function exit, the following
11401 profiling functions will be called with the address of the current
11402 function and its call site. (On some platforms,
11403 @code{__builtin_return_address} does not work beyond the current
11404 function, so the call site information may not be available to the
11405 profiling functions otherwise.)
11408 void __cyg_profile_func_enter (void *this_fn,
11410 void __cyg_profile_func_exit (void *this_fn,
11414 The first argument is the address of the start of the current function,
11415 which may be looked up exactly in the symbol table.
11417 This instrumentation is also done for functions expanded inline in other
11418 functions. The profiling calls will indicate where, conceptually, the
11419 inline function is entered and exited. This means that addressable
11420 versions of such functions must be available. If all your uses of a
11421 function are expanded inline, this may mean an additional expansion of
11422 code size. If you use @samp{extern inline} in your C code, an
11423 addressable version of such functions must be provided. (This is
11424 normally the case anyways, but if you get lucky and the optimizer always
11425 expands the functions inline, you might have gotten away without
11426 providing static copies.)
11428 A function may be given the attribute @code{no_instrument_function}, in
11429 which case this instrumentation will not be done. This can be used, for
11430 example, for the profiling functions listed above, high-priority
11431 interrupt routines, and any functions from which the profiling functions
11432 cannot safely be called (perhaps signal handlers, if the profiling
11433 routines generate output or allocate memory).
11435 @item -fstack-check
11436 @opindex fstack-check
11437 Generate code to verify that you do not go beyond the boundary of the
11438 stack. You should specify this flag if you are running in an
11439 environment with multiple threads, but only rarely need to specify it in
11440 a single-threaded environment since stack overflow is automatically
11441 detected on nearly all systems if there is only one stack.
11443 Note that this switch does not actually cause checking to be done; the
11444 operating system must do that. The switch causes generation of code
11445 to ensure that the operating system sees the stack being extended.
11447 @item -fstack-limit-register=@var{reg}
11448 @itemx -fstack-limit-symbol=@var{sym}
11449 @itemx -fno-stack-limit
11450 @opindex fstack-limit-register
11451 @opindex fstack-limit-symbol
11452 @opindex fno-stack-limit
11453 Generate code to ensure that the stack does not grow beyond a certain value,
11454 either the value of a register or the address of a symbol. If the stack
11455 would grow beyond the value, a signal is raised. For most targets,
11456 the signal is raised before the stack overruns the boundary, so
11457 it is possible to catch the signal without taking special precautions.
11459 For instance, if the stack starts at absolute address @samp{0x80000000}
11460 and grows downwards, you can use the flags
11461 @option{-fstack-limit-symbol=__stack_limit} and
11462 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
11463 of 128KB@. Note that this may only work with the GNU linker.
11465 @cindex aliasing of parameters
11466 @cindex parameters, aliased
11467 @item -fargument-alias
11468 @itemx -fargument-noalias
11469 @itemx -fargument-noalias-global
11470 @opindex fargument-alias
11471 @opindex fargument-noalias
11472 @opindex fargument-noalias-global
11473 Specify the possible relationships among parameters and between
11474 parameters and global data.
11476 @option{-fargument-alias} specifies that arguments (parameters) may
11477 alias each other and may alias global storage.@*
11478 @option{-fargument-noalias} specifies that arguments do not alias
11479 each other, but may alias global storage.@*
11480 @option{-fargument-noalias-global} specifies that arguments do not
11481 alias each other and do not alias global storage.
11483 Each language will automatically use whatever option is required by
11484 the language standard. You should not need to use these options yourself.
11486 @item -fleading-underscore
11487 @opindex fleading-underscore
11488 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
11489 change the way C symbols are represented in the object file. One use
11490 is to help link with legacy assembly code.
11492 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
11493 generate code that is not binary compatible with code generated without that
11494 switch. Use it to conform to a non-default application binary interface.
11495 Not all targets provide complete support for this switch.
11497 @item -ftls-model=@var{model}
11498 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
11499 The @var{model} argument should be one of @code{global-dynamic},
11500 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
11502 The default without @option{-fpic} is @code{initial-exec}; with
11503 @option{-fpic} the default is @code{global-dynamic}.
11508 @node Environment Variables
11509 @section Environment Variables Affecting GCC
11510 @cindex environment variables
11512 @c man begin ENVIRONMENT
11513 This section describes several environment variables that affect how GCC
11514 operates. Some of them work by specifying directories or prefixes to use
11515 when searching for various kinds of files. Some are used to specify other
11516 aspects of the compilation environment.
11518 Note that you can also specify places to search using options such as
11519 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
11520 take precedence over places specified using environment variables, which
11521 in turn take precedence over those specified by the configuration of GCC@.
11522 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
11523 GNU Compiler Collection (GCC) Internals}.
11528 @c @itemx LC_COLLATE
11530 @c @itemx LC_MONETARY
11531 @c @itemx LC_NUMERIC
11536 @c @findex LC_COLLATE
11537 @findex LC_MESSAGES
11538 @c @findex LC_MONETARY
11539 @c @findex LC_NUMERIC
11543 These environment variables control the way that GCC uses
11544 localization information that allow GCC to work with different
11545 national conventions. GCC inspects the locale categories
11546 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
11547 so. These locale categories can be set to any value supported by your
11548 installation. A typical value is @samp{en_UK} for English in the United
11551 The @env{LC_CTYPE} environment variable specifies character
11552 classification. GCC uses it to determine the character boundaries in
11553 a string; this is needed for some multibyte encodings that contain quote
11554 and escape characters that would otherwise be interpreted as a string
11557 The @env{LC_MESSAGES} environment variable specifies the language to
11558 use in diagnostic messages.
11560 If the @env{LC_ALL} environment variable is set, it overrides the value
11561 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
11562 and @env{LC_MESSAGES} default to the value of the @env{LANG}
11563 environment variable. If none of these variables are set, GCC
11564 defaults to traditional C English behavior.
11568 If @env{TMPDIR} is set, it specifies the directory to use for temporary
11569 files. GCC uses temporary files to hold the output of one stage of
11570 compilation which is to be used as input to the next stage: for example,
11571 the output of the preprocessor, which is the input to the compiler
11574 @item GCC_EXEC_PREFIX
11575 @findex GCC_EXEC_PREFIX
11576 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
11577 names of the subprograms executed by the compiler. No slash is added
11578 when this prefix is combined with the name of a subprogram, but you can
11579 specify a prefix that ends with a slash if you wish.
11581 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
11582 an appropriate prefix to use based on the pathname it was invoked with.
11584 If GCC cannot find the subprogram using the specified prefix, it
11585 tries looking in the usual places for the subprogram.
11587 The default value of @env{GCC_EXEC_PREFIX} is
11588 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the value
11589 of @code{prefix} when you ran the @file{configure} script.
11591 Other prefixes specified with @option{-B} take precedence over this prefix.
11593 This prefix is also used for finding files such as @file{crt0.o} that are
11596 In addition, the prefix is used in an unusual way in finding the
11597 directories to search for header files. For each of the standard
11598 directories whose name normally begins with @samp{/usr/local/lib/gcc}
11599 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
11600 replacing that beginning with the specified prefix to produce an
11601 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
11602 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
11603 These alternate directories are searched first; the standard directories
11606 @item COMPILER_PATH
11607 @findex COMPILER_PATH
11608 The value of @env{COMPILER_PATH} is a colon-separated list of
11609 directories, much like @env{PATH}. GCC tries the directories thus
11610 specified when searching for subprograms, if it can't find the
11611 subprograms using @env{GCC_EXEC_PREFIX}.
11614 @findex LIBRARY_PATH
11615 The value of @env{LIBRARY_PATH} is a colon-separated list of
11616 directories, much like @env{PATH}. When configured as a native compiler,
11617 GCC tries the directories thus specified when searching for special
11618 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
11619 using GCC also uses these directories when searching for ordinary
11620 libraries for the @option{-l} option (but directories specified with
11621 @option{-L} come first).
11625 @cindex locale definition
11626 This variable is used to pass locale information to the compiler. One way in
11627 which this information is used is to determine the character set to be used
11628 when character literals, string literals and comments are parsed in C and C++.
11629 When the compiler is configured to allow multibyte characters,
11630 the following values for @env{LANG} are recognized:
11634 Recognize JIS characters.
11636 Recognize SJIS characters.
11638 Recognize EUCJP characters.
11641 If @env{LANG} is not defined, or if it has some other value, then the
11642 compiler will use mblen and mbtowc as defined by the default locale to
11643 recognize and translate multibyte characters.
11647 Some additional environments variables affect the behavior of the
11650 @include cppenv.texi
11654 @node Precompiled Headers
11655 @section Using Precompiled Headers
11656 @cindex precompiled headers
11657 @cindex speed of compilation
11659 Often large projects have many header files that are included in every
11660 source file. The time the compiler takes to process these header files
11661 over and over again can account for nearly all of the time required to
11662 build the project. To make builds faster, GCC allows users to
11663 `precompile' a header file; then, if builds can use the precompiled
11664 header file they will be much faster.
11666 To create a precompiled header file, simply compile it as you would any
11667 other file, if necessary using the @option{-x} option to make the driver
11668 treat it as a C or C++ header file. You will probably want to use a
11669 tool like @command{make} to keep the precompiled header up-to-date when
11670 the headers it contains change.
11672 A precompiled header file will be searched for when @code{#include} is
11673 seen in the compilation. As it searches for the included file
11674 (@pxref{Search Path,,Search Path,cpp.info,The C Preprocessor}) the
11675 compiler looks for a precompiled header in each directory just before it
11676 looks for the include file in that directory. The name searched for is
11677 the name specified in the @code{#include} with @samp{.gch} appended. If
11678 the precompiled header file can't be used, it is ignored.
11680 For instance, if you have @code{#include "all.h"}, and you have
11681 @file{all.h.gch} in the same directory as @file{all.h}, then the
11682 precompiled header file will be used if possible, and the original
11683 header will be used otherwise.
11685 Alternatively, you might decide to put the precompiled header file in a
11686 directory and use @option{-I} to ensure that directory is searched
11687 before (or instead of) the directory containing the original header.
11688 Then, if you want to check that the precompiled header file is always
11689 used, you can put a file of the same name as the original header in this
11690 directory containing an @code{#error} command.
11692 This also works with @option{-include}. So yet another way to use
11693 precompiled headers, good for projects not designed with precompiled
11694 header files in mind, is to simply take most of the header files used by
11695 a project, include them from another header file, precompile that header
11696 file, and @option{-include} the precompiled header. If the header files
11697 have guards against multiple inclusion, they will be skipped because
11698 they've already been included (in the precompiled header).
11700 If you need to precompile the same header file for different
11701 languages, targets, or compiler options, you can instead make a
11702 @emph{directory} named like @file{all.h.gch}, and put each precompiled
11703 header in the directory. (It doesn't matter what you call the files
11704 in the directory, every precompiled header in the directory will be
11705 considered.) The first precompiled header encountered in the
11706 directory that is valid for this compilation will be used; they're
11707 searched in no particular order.
11709 There are many other possibilities, limited only by your imagination,
11710 good sense, and the constraints of your build system.
11712 A precompiled header file can be used only when these conditions apply:
11716 Only one precompiled header can be used in a particular compilation.
11718 A precompiled header can't be used once the first C token is seen. You
11719 can have preprocessor directives before a precompiled header; you can
11720 even include a precompiled header from inside another header, so long as
11721 there are no C tokens before the @code{#include}.
11723 The precompiled header file must be produced for the same language as
11724 the current compilation. You can't use a C precompiled header for a C++
11727 The precompiled header file must be produced by the same compiler
11728 version and configuration as the current compilation is using.
11729 The easiest way to guarantee this is to use the same compiler binary
11730 for creating and using precompiled headers.
11732 Any macros defined before the precompiled header (including with
11733 @option{-D}) must either be defined in the same way as when the
11734 precompiled header was generated, or must not affect the precompiled
11735 header, which usually means that the they don't appear in the
11736 precompiled header at all.
11738 Certain command-line options must be defined in the same way as when the
11739 precompiled header was generated. At present, it's not clear which
11740 options are safe to change and which are not; the safest choice is to
11741 use exactly the same options when generating and using the precompiled
11745 For all of these but the last, the compiler will automatically ignore
11746 the precompiled header if the conditions aren't met. For the last item,
11747 some option changes will cause the precompiled header to be rejected,
11748 but not all incompatible option combinations have yet been found. If
11749 you find a new incompatible combination, please consider filing a bug
11750 report, see @ref{Bugs}.
11752 @node Running Protoize
11753 @section Running Protoize
11755 The program @code{protoize} is an optional part of GCC@. You can use
11756 it to add prototypes to a program, thus converting the program to ISO
11757 C in one respect. The companion program @code{unprotoize} does the
11758 reverse: it removes argument types from any prototypes that are found.
11760 When you run these programs, you must specify a set of source files as
11761 command line arguments. The conversion programs start out by compiling
11762 these files to see what functions they define. The information gathered
11763 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
11765 After scanning comes actual conversion. The specified files are all
11766 eligible to be converted; any files they include (whether sources or
11767 just headers) are eligible as well.
11769 But not all the eligible files are converted. By default,
11770 @code{protoize} and @code{unprotoize} convert only source and header
11771 files in the current directory. You can specify additional directories
11772 whose files should be converted with the @option{-d @var{directory}}
11773 option. You can also specify particular files to exclude with the
11774 @option{-x @var{file}} option. A file is converted if it is eligible, its
11775 directory name matches one of the specified directory names, and its
11776 name within the directory has not been excluded.
11778 Basic conversion with @code{protoize} consists of rewriting most
11779 function definitions and function declarations to specify the types of
11780 the arguments. The only ones not rewritten are those for varargs
11783 @code{protoize} optionally inserts prototype declarations at the
11784 beginning of the source file, to make them available for any calls that
11785 precede the function's definition. Or it can insert prototype
11786 declarations with block scope in the blocks where undeclared functions
11789 Basic conversion with @code{unprotoize} consists of rewriting most
11790 function declarations to remove any argument types, and rewriting
11791 function definitions to the old-style pre-ISO form.
11793 Both conversion programs print a warning for any function declaration or
11794 definition that they can't convert. You can suppress these warnings
11797 The output from @code{protoize} or @code{unprotoize} replaces the
11798 original source file. The original file is renamed to a name ending
11799 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
11800 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
11801 for DOS) file already exists, then the source file is simply discarded.
11803 @code{protoize} and @code{unprotoize} both depend on GCC itself to
11804 scan the program and collect information about the functions it uses.
11805 So neither of these programs will work until GCC is installed.
11807 Here is a table of the options you can use with @code{protoize} and
11808 @code{unprotoize}. Each option works with both programs unless
11812 @item -B @var{directory}
11813 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
11814 usual directory (normally @file{/usr/local/lib}). This file contains
11815 prototype information about standard system functions. This option
11816 applies only to @code{protoize}.
11818 @item -c @var{compilation-options}
11819 Use @var{compilation-options} as the options when running @command{gcc} to
11820 produce the @samp{.X} files. The special option @option{-aux-info} is
11821 always passed in addition, to tell @command{gcc} to write a @samp{.X} file.
11823 Note that the compilation options must be given as a single argument to
11824 @code{protoize} or @code{unprotoize}. If you want to specify several
11825 @command{gcc} options, you must quote the entire set of compilation options
11826 to make them a single word in the shell.
11828 There are certain @command{gcc} arguments that you cannot use, because they
11829 would produce the wrong kind of output. These include @option{-g},
11830 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
11831 the @var{compilation-options}, they are ignored.
11834 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
11835 systems) instead of @samp{.c}. This is convenient if you are converting
11836 a C program to C++. This option applies only to @code{protoize}.
11839 Add explicit global declarations. This means inserting explicit
11840 declarations at the beginning of each source file for each function
11841 that is called in the file and was not declared. These declarations
11842 precede the first function definition that contains a call to an
11843 undeclared function. This option applies only to @code{protoize}.
11845 @item -i @var{string}
11846 Indent old-style parameter declarations with the string @var{string}.
11847 This option applies only to @code{protoize}.
11849 @code{unprotoize} converts prototyped function definitions to old-style
11850 function definitions, where the arguments are declared between the
11851 argument list and the initial @samp{@{}. By default, @code{unprotoize}
11852 uses five spaces as the indentation. If you want to indent with just
11853 one space instead, use @option{-i " "}.
11856 Keep the @samp{.X} files. Normally, they are deleted after conversion
11860 Add explicit local declarations. @code{protoize} with @option{-l} inserts
11861 a prototype declaration for each function in each block which calls the
11862 function without any declaration. This option applies only to
11866 Make no real changes. This mode just prints information about the conversions
11867 that would have been done without @option{-n}.
11870 Make no @samp{.save} files. The original files are simply deleted.
11871 Use this option with caution.
11873 @item -p @var{program}
11874 Use the program @var{program} as the compiler. Normally, the name
11875 @file{gcc} is used.
11878 Work quietly. Most warnings are suppressed.
11881 Print the version number, just like @option{-v} for @command{gcc}.
11884 If you need special compiler options to compile one of your program's
11885 source files, then you should generate that file's @samp{.X} file
11886 specially, by running @command{gcc} on that source file with the
11887 appropriate options and the option @option{-aux-info}. Then run
11888 @code{protoize} on the entire set of files. @code{protoize} will use
11889 the existing @samp{.X} file because it is newer than the source file.
11893 gcc -Dfoo=bar file1.c -aux-info file1.X
11898 You need to include the special files along with the rest in the
11899 @code{protoize} command, even though their @samp{.X} files already
11900 exist, because otherwise they won't get converted.
11902 @xref{Protoize Caveats}, for more information on how to use
11903 @code{protoize} successfully.