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 @gol
178 -ffor-scope -fno-for-scope -fno-gnu-keywords @gol
179 -fno-implicit-templates @gol
180 -fno-implicit-inline-templates @gol
181 -fno-implement-inlines -fms-extensions @gol
182 -fno-nonansi-builtins -fno-operator-names @gol
183 -fno-optional-diags -fpermissive @gol
184 -frepo -fno-rtti -fstats -ftemplate-depth-@var{n} @gol
185 -fuse-cxa-atexit -fno-weak -nostdinc++ @gol
186 -fno-default-inline -Wabi -Wctor-dtor-privacy @gol
187 -Wnon-virtual-dtor -Wreorder @gol
188 -Weffc++ -Wno-deprecated @gol
189 -Wno-non-template-friend -Wold-style-cast @gol
190 -Woverloaded-virtual -Wno-pmf-conversions @gol
191 -Wsign-promo -Wsynth}
193 @item Objective-C Language Options
194 @xref{Objective-C Dialect Options,,Options Controlling Objective-C Dialect}.
196 -fconstant-string-class=@var{class-name} @gol
197 -fgnu-runtime -fnext-runtime @gol
198 -fno-nil-receivers @gol
199 -fobjc-exceptions @gol
200 -freplace-objc-classes @gol
203 -Wno-protocol -Wselector -Wundeclared-selector}
205 @item Language Independent Options
206 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
207 @gccoptlist{-fmessage-length=@var{n} @gol
208 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]}}
210 @item Warning Options
211 @xref{Warning Options,,Options to Request or Suppress Warnings}.
212 @gccoptlist{-fsyntax-only -pedantic -pedantic-errors @gol
213 -w -Wextra -Wall -Waggregate-return @gol
214 -Wcast-align -Wcast-qual -Wchar-subscripts -Wcomment @gol
215 -Wconversion -Wno-deprecated-declarations @gol
216 -Wdisabled-optimization -Wno-div-by-zero -Wendif-labels @gol
217 -Werror -Werror-implicit-function-declaration @gol
218 -Wfloat-equal -Wformat -Wformat=2 @gol
219 -Wno-format-extra-args -Wformat-nonliteral @gol
220 -Wformat-security -Wformat-y2k @gol
221 -Wimplicit -Wimplicit-function-declaration -Wimplicit-int @gol
222 -Wimport -Wno-import -Winit-self -Winline @gol
223 -Wno-invalid-offsetof -Winvalid-pch @gol
224 -Wlarger-than-@var{len} -Wlong-long @gol
225 -Wmain -Wmissing-braces @gol
226 -Wmissing-format-attribute -Wmissing-noreturn @gol
227 -Wno-multichar -Wnonnull -Wpacked -Wpadded @gol
228 -Wparentheses -Wpointer-arith -Wredundant-decls @gol
229 -Wreturn-type -Wsequence-point -Wshadow @gol
230 -Wsign-compare -Wstrict-aliasing @gol
231 -Wswitch -Wswitch-default -Wswitch-enum @gol
232 -Wsystem-headers -Wtrigraphs -Wundef -Wuninitialized @gol
233 -Wunknown-pragmas -Wunreachable-code @gol
234 -Wunused -Wunused-function -Wunused-label -Wunused-parameter @gol
235 -Wunused-value -Wunused-variable -Wwrite-strings}
237 @item C-only Warning Options
238 @gccoptlist{-Wbad-function-cast -Wmissing-declarations @gol
239 -Wmissing-prototypes -Wnested-externs -Wold-style-definition @gol
240 -Wstrict-prototypes -Wtraditional @gol
241 -Wdeclaration-after-statement}
243 @item Debugging Options
244 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
245 @gccoptlist{-d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
246 -fdump-unnumbered -fdump-translation-unit@r{[}-@var{n}@r{]} @gol
247 -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
248 -fdump-tree-original@r{[}-@var{n}@r{]} @gol
249 -fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
250 -fdump-tree-inlined@r{[}-@var{n}@r{]} @gol
251 -feliminate-dwarf2-dups -feliminate-unused-debug-types @gol
252 -feliminate-unused-debug-symbols -fmem-report -fprofile-arcs @gol
253 -frandom-seed=@var{string} -fsched-verbose=@var{n} @gol
254 -ftest-coverage -ftime-report @gol
255 -g -g@var{level} -gcoff -gdwarf-2 @gol
256 -ggdb -gstabs -gstabs+ -gvms -gxcoff -gxcoff+ @gol
257 -p -pg -print-file-name=@var{library} -print-libgcc-file-name @gol
258 -print-multi-directory -print-multi-lib @gol
259 -print-prog-name=@var{program} -print-search-dirs -Q @gol
262 @item Optimization Options
263 @xref{Optimize Options,,Options that Control Optimization}.
264 @gccoptlist{-falign-functions=@var{n} -falign-jumps=@var{n} @gol
265 -falign-labels=@var{n} -falign-loops=@var{n} @gol
266 -fbranch-probabilities -fprofile-values -fvpt -fbranch-target-load-optimize @gol
267 -fbranch-target-load-optimize2 -fcaller-saves -fcprop-registers @gol
268 -fcse-follow-jumps -fcse-skip-blocks -fdata-sections @gol
269 -fdelayed-branch -fdelete-null-pointer-checks @gol
270 -fexpensive-optimizations -ffast-math -ffloat-store @gol
271 -fforce-addr -fforce-mem -ffunction-sections @gol
272 -fgcse -fgcse-lm -fgcse-sm -fgcse-las -floop-optimize @gol
273 -fcrossjumping -fif-conversion -fif-conversion2 @gol
274 -finline-functions -finline-limit=@var{n} -fkeep-inline-functions @gol
275 -fkeep-static-consts -fmerge-constants -fmerge-all-constants @gol
276 -fmove-all-movables -fnew-ra -fno-branch-count-reg @gol
277 -fno-default-inline -fno-defer-pop @gol
278 -fno-function-cse -fno-guess-branch-probability @gol
279 -fno-inline -fno-math-errno -fno-peephole -fno-peephole2 @gol
280 -funsafe-math-optimizations -ffinite-math-only @gol
281 -fno-trapping-math -fno-zero-initialized-in-bss @gol
282 -fomit-frame-pointer -foptimize-register-move @gol
283 -foptimize-sibling-calls -fprefetch-loop-arrays @gol
284 -fprofile-generate -fprofile-use @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 @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 -msep-data -mno-sep-data @gol
345 -mshared-library-id=n -mid-shared-library -mno-id-shared-library}
347 @emph{M68hc1x Options}
348 @gccoptlist{-m6811 -m6812 -m68hc11 -m68hc12 -m68hcs12 @gol
349 -mauto-incdec -minmax -mlong-calls -mshort @gol
350 -msoft-reg-count=@var{count}}
353 @gccoptlist{-mg -mgnu -munix}
356 @gccoptlist{-mcpu=@var{cpu-type} @gol
357 -mtune=@var{cpu-type} @gol
358 -mcmodel=@var{code-model} @gol
360 -mapp-regs -mbroken-saverestore -mcypress @gol
361 -mfaster-structs -mflat @gol
362 -mfpu -mhard-float -mhard-quad-float @gol
363 -mimpure-text -mlittle-endian -mlive-g0 -mno-app-regs @gol
364 -mno-faster-structs -mno-flat -mno-fpu @gol
365 -mno-impure-text -mno-stack-bias -mno-unaligned-doubles @gol
366 -msoft-float -msoft-quad-float -msparclite -mstack-bias @gol
367 -msupersparc -munaligned-doubles -mv8}
370 @gccoptlist{-mapcs-frame -mno-apcs-frame @gol
371 -mapcs-26 -mapcs-32 @gol
372 -mapcs-stack-check -mno-apcs-stack-check @gol
373 -mapcs-float -mno-apcs-float @gol
374 -mapcs-reentrant -mno-apcs-reentrant @gol
375 -msched-prolog -mno-sched-prolog @gol
376 -mlittle-endian -mbig-endian -mwords-little-endian @gol
377 -malignment-traps -mno-alignment-traps @gol
378 -msoft-float -mhard-float -mfpe @gol
379 -mthumb-interwork -mno-thumb-interwork @gol
380 -mcpu=@var{name} -march=@var{name} -mfpe=@var{name} @gol
381 -mstructure-size-boundary=@var{n} @gol
382 -mabort-on-noreturn @gol
383 -mlong-calls -mno-long-calls @gol
384 -msingle-pic-base -mno-single-pic-base @gol
385 -mpic-register=@var{reg} @gol
386 -mnop-fun-dllimport @gol
387 -mcirrus-fix-invalid-insns -mno-cirrus-fix-invalid-insns @gol
388 -mpoke-function-name @gol
390 -mtpcs-frame -mtpcs-leaf-frame @gol
391 -mcaller-super-interworking -mcallee-super-interworking}
393 @emph{MN10200 Options}
396 @emph{MN10300 Options}
397 @gccoptlist{-mmult-bug -mno-mult-bug @gol
398 -mam33 -mno-am33 @gol
399 -mam33-2 -mno-am33-2 @gol
402 @emph{M32R/D Options}
403 @gccoptlist{-m32r2 -m32rx -m32r -mcode-model=@var{model-type} @gol
404 -msdata=@var{sdata-type} -G @var{num}}
407 @gccoptlist{-m88000 -m88100 -m88110 -mbig-pic @gol
408 -mcheck-zero-division -mhandle-large-shift @gol
409 -midentify-revision -mno-check-zero-division @gol
410 -mno-ocs-debug-info -mno-ocs-frame-position @gol
411 -mno-optimize-arg-area -mno-serialize-volatile @gol
412 -mno-underscores -mocs-debug-info @gol
413 -mocs-frame-position -moptimize-arg-area @gol
414 -mserialize-volatile -mshort-data-@var{num} -msvr3 @gol
415 -msvr4 -mtrap-large-shift -muse-div-instruction @gol
416 -mversion-03.00 -mwarn-passed-structs}
418 @emph{RS/6000 and PowerPC Options}
419 @gccoptlist{-mcpu=@var{cpu-type} @gol
420 -mtune=@var{cpu-type} @gol
421 -mpower -mno-power -mpower2 -mno-power2 @gol
422 -mpowerpc -mpowerpc64 -mno-powerpc @gol
423 -maltivec -mno-altivec @gol
424 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
425 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
426 -mnew-mnemonics -mold-mnemonics @gol
427 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
428 -m64 -m32 -mxl-call -mno-xl-call -mpe @gol
429 -malign-power -malign-natural @gol
430 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
431 -mstring -mno-string -mupdate -mno-update @gol
432 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
433 -mstrict-align -mno-strict-align -mrelocatable @gol
434 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
435 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
436 -mdynamic-no-pic @gol
437 -mprioritize-restricted-insns=@var{priority} @gol
438 -msched-costly-dep=@var{dependence_type} @gol
439 -minsert-sched-nops=@var{scheme} @gol
440 -mcall-sysv -mcall-netbsd @gol
441 -maix-struct-return -msvr4-struct-return @gol
442 -mabi=altivec -mabi=no-altivec @gol
443 -mabi=spe -mabi=no-spe @gol
444 -misel=yes -misel=no @gol
445 -mspe=yes -mspe=no @gol
446 -mfloat-gprs=yes -mfloat-gprs=no @gol
447 -mprototype -mno-prototype @gol
448 -msim -mmvme -mads -myellowknife -memb -msdata @gol
449 -msdata=@var{opt} -mvxworks -mwindiss -G @var{num} -pthread}
451 @emph{Darwin Options}
452 @gccoptlist{-all_load -allowable_client -arch -arch_errors_fatal @gol
453 -arch_only -bind_at_load -bundle -bundle_loader @gol
454 -client_name -compatibility_version -current_version @gol
455 -dependency-file -dylib_file -dylinker_install_name @gol
456 -dynamic -dynamiclib -exported_symbols_list @gol
457 -filelist -flat_namespace -force_cpusubtype_ALL @gol
458 -force_flat_namespace -headerpad_max_install_names @gol
459 -image_base -init -install_name -keep_private_externs @gol
460 -multi_module -multiply_defined -multiply_defined_unused @gol
461 -noall_load -nofixprebinding -nomultidefs -noprebind -noseglinkedit @gol
462 -pagezero_size -prebind -prebind_all_twolevel_modules @gol
463 -private_bundle -read_only_relocs -sectalign @gol
464 -sectobjectsymbols -whyload -seg1addr @gol
465 -sectcreate -sectobjectsymbols -sectorder @gol
466 -seg_addr_table -seg_addr_table_filename -seglinkedit @gol
467 -segprot -segs_read_only_addr -segs_read_write_addr @gol
468 -single_module -static -sub_library -sub_umbrella @gol
469 -twolevel_namespace -umbrella -undefined @gol
470 -unexported_symbols_list -weak_reference_mismatches @gol
474 @gccoptlist{-mcall-lib-mul -mfp-arg-in-fpregs -mfp-arg-in-gregs @gol
475 -mfull-fp-blocks -mhc-struct-return -min-line-mul @gol
476 -mminimum-fp-blocks -mnohc-struct-return}
479 @gccoptlist{-mabicalls -march=@var{cpu-type} -mtune=@var{cpu=type} @gol
480 -mcpu=@var{cpu-type} -membedded-data -muninit-const-in-rodata @gol
481 -membedded-pic -mfp32 -mfp64 -mfused-madd -mno-fused-madd @gol
482 -mgas -mgp32 -mgp64 -mhard-float -mint64 -mips1 @gol
483 -mips2 -mips3 -mips4 -mips32 -mips32r2 -mips64 @gol
484 -mlong64 -mlong32 -mlong-calls -mmemcpy @gol
485 -mmips-as -mmips-tfile -mno-abicalls -mxgot @gol
486 -mno-embedded-data -mno-uninit-const-in-rodata @gol
487 -mno-embedded-pic -mno-long-calls @gol
488 -mno-memcpy -mno-mips-tfile -mno-rnames @gol
489 -mrnames -msoft-float @gol
490 -m4650 -msingle-float -mmad @gol
491 -EL -EB -G @var{num} -nocpp @gol
492 -mabi=32 -mabi=n32 -mabi=64 -mabi=eabi -mabi-fake-default @gol
493 -mfix7000 -mfix-sb1 -mno-fix-sb1 @gol
494 -mno-crt0 -mflush-func=@var{func} -mno-flush-func @gol
495 -mbranch-likely -mno-branch-likely}
497 @emph{i386 and x86-64 Options}
498 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
499 -mfpmath=@var{unit} @gol
500 -masm=@var{dialect} -mno-fancy-math-387 @gol
501 -mno-fp-ret-in-387 -msoft-float -msvr3-shlib @gol
502 -mno-wide-multiply -mrtd -malign-double @gol
503 -mpreferred-stack-boundary=@var{num} @gol
504 -mmmx -msse -msse2 -mpni -m3dnow @gol
505 -mthreads -mno-align-stringops -minline-all-stringops @gol
506 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
507 -m96bit-long-double -mregparm=@var{num} -momit-leaf-frame-pointer @gol
508 -mno-red-zone -mno-tls-direct-seg-refs @gol
509 -mcmodel=@var{code-model} @gol
513 @gccoptlist{-march=@var{architecture-type} @gol
514 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
515 -mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
516 -mjump-in-delay -mlinker-opt -mlong-calls @gol
517 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
518 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
519 -mno-jump-in-delay -mno-long-load-store @gol
520 -mno-portable-runtime -mno-soft-float @gol
521 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
522 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
523 -mschedule=@var{cpu-type} -mspace-regs -msio -mwsio @gol
524 -nolibdld -static -threads}
526 @emph{Intel 960 Options}
527 @gccoptlist{-m@var{cpu-type} -masm-compat -mclean-linkage @gol
528 -mcode-align -mcomplex-addr -mleaf-procedures @gol
529 -mic-compat -mic2.0-compat -mic3.0-compat @gol
530 -mintel-asm -mno-clean-linkage -mno-code-align @gol
531 -mno-complex-addr -mno-leaf-procedures @gol
532 -mno-old-align -mno-strict-align -mno-tail-call @gol
533 -mnumerics -mold-align -msoft-float -mstrict-align @gol
536 @emph{DEC Alpha Options}
537 @gccoptlist{-mno-fp-regs -msoft-float -malpha-as -mgas @gol
538 -mieee -mieee-with-inexact -mieee-conformant @gol
539 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
540 -mtrap-precision=@var{mode} -mbuild-constants @gol
541 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} @gol
542 -mbwx -mmax -mfix -mcix @gol
543 -mfloat-vax -mfloat-ieee @gol
544 -mexplicit-relocs -msmall-data -mlarge-data @gol
545 -msmall-text -mlarge-text @gol
546 -mmemory-latency=@var{time}}
548 @emph{DEC Alpha/VMS Options}
549 @gccoptlist{-mvms-return-codes}
551 @emph{H8/300 Options}
552 @gccoptlist{-mrelax -mh -ms -mn -mint32 -malign-300}
555 @gccoptlist{-m1 -m2 -m2e -m3 -m3e @gol
556 -m4-nofpu -m4-single-only -m4-single -m4 @gol
557 -m5-64media -m5-64media-nofpu @gol
558 -m5-32media -m5-32media-nofpu @gol
559 -m5-compact -m5-compact-nofpu @gol
560 -mb -ml -mdalign -mrelax @gol
561 -mbigtable -mfmovd -mhitachi -mnomacsave @gol
562 -mieee -misize -mpadstruct -mspace @gol
563 -mprefergot -musermode}
565 @emph{System V Options}
566 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
569 @gccoptlist{-EB -EL @gol
570 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
571 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
573 @emph{TMS320C3x/C4x Options}
574 @gccoptlist{-mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm @gol
575 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload @gol
576 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned @gol
577 -mparallel-insns -mparallel-mpy -mpreserve-float}
580 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
581 -mprolog-function -mno-prolog-function -mspace @gol
582 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
583 -mapp-regs -mno-app-regs @gol
584 -mdisable-callt -mno-disable-callt @gol
590 @gccoptlist{-m32032 -m32332 -m32532 -m32081 -m32381 @gol
591 -mmult-add -mnomult-add -msoft-float -mrtd -mnortd @gol
592 -mregparam -mnoregparam -msb -mnosb @gol
593 -mbitfield -mnobitfield -mhimem -mnohimem}
596 @gccoptlist{-mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts @gol
597 -mcall-prologues -mno-tablejump -mtiny-stack}
600 @gccoptlist{-mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
601 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
602 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
603 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
604 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
607 @gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
608 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
609 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
610 -mno-base-addresses -msingle-exit -mno-single-exit}
613 @gccoptlist{-mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
614 -mvolatile-asm-stop -mb-step -mregister-names -mno-sdata @gol
615 -mconstant-gp -mauto-pic -minline-float-divide-min-latency @gol
616 -minline-float-divide-max-throughput @gol
617 -minline-int-divide-min-latency @gol
618 -minline-int-divide-max-throughput -mno-dwarf2-asm @gol
619 -mfixed-range=@var{register-range}}
622 @gccoptlist{-mextmem -mextmemory -monchip -mno-asm-optimize @gol
623 -masm-optimize -mbranch-cost=@var{n} -mcond-exec=@var{n}}
625 @emph{S/390 and zSeries Options}
626 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
627 -mhard-float -msoft-float -mbackchain -mno-backchain @gol
628 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
629 -m64 -m31 -mdebug -mno-debug -mesa -mzarch -mfused-madd -mno-fused-madd}
632 @gccoptlist{-mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
633 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
634 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
635 -mstack-align -mdata-align -mconst-align @gol
636 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
637 -melf -maout -melinux -mlinux -sim -sim2}
639 @emph{PDP-11 Options}
640 @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
641 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
642 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
643 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
644 -mbranch-expensive -mbranch-cheap @gol
645 -msplit -mno-split -munix-asm -mdec-asm}
647 @emph{Xstormy16 Options}
650 @emph{Xtensa Options}
651 @gccoptlist{-mconst16 -mno-const16 @gol
652 -mfused-madd -mno-fused-madd @gol
653 -mtext-section-literals -mno-text-section-literals @gol
654 -mtarget-align -mno-target-align @gol
655 -mlongcalls -mno-longcalls}
658 @gccoptlist{-mgpr-32 -mgpr-64 -mfpr-32 -mfpr-64 @gol
659 -mhard-float -msoft-float @gol
660 -malloc-cc -mfixed-cc -mdword -mno-dword @gol
661 -mdouble -mno-double @gol
662 -mmedia -mno-media -mmuladd -mno-muladd @gol
663 -mlibrary-pic -macc-4 -macc-8 @gol
664 -mpack -mno-pack -mno-eflags -mcond-move -mno-cond-move @gol
665 -mscc -mno-scc -mcond-exec -mno-cond-exec @gol
666 -mvliw-branch -mno-vliw-branch @gol
667 -mmulti-cond-exec -mno-multi-cond-exec -mnested-cond-exec @gol
668 -mno-nested-cond-exec -mtomcat-stats @gol
671 @item Code Generation Options
672 @xref{Code Gen Options,,Options for Code Generation Conventions}.
673 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
674 -ffixed-@var{reg} -fexceptions @gol
675 -fnon-call-exceptions -funwind-tables @gol
676 -fasynchronous-unwind-tables @gol
677 -finhibit-size-directive -finstrument-functions @gol
678 -fno-common -fno-ident @gol
679 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
680 -freg-struct-return -fshared-data -fshort-enums @gol
681 -fshort-double -fshort-wchar @gol
682 -fverbose-asm -fpack-struct -fstack-check @gol
683 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
684 -fargument-alias -fargument-noalias @gol
685 -fargument-noalias-global -fleading-underscore @gol
686 -ftls-model=@var{model} @gol
687 -ftrapv -fwrapv -fbounds-check}
691 * Overall Options:: Controlling the kind of output:
692 an executable, object files, assembler files,
693 or preprocessed source.
694 * C Dialect Options:: Controlling the variant of C language compiled.
695 * C++ Dialect Options:: Variations on C++.
696 * Objective-C Dialect Options:: Variations on Objective-C.
697 * Language Independent Options:: Controlling how diagnostics should be
699 * Warning Options:: How picky should the compiler be?
700 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
701 * Optimize Options:: How much optimization?
702 * Preprocessor Options:: Controlling header files and macro definitions.
703 Also, getting dependency information for Make.
704 * Assembler Options:: Passing options to the assembler.
705 * Link Options:: Specifying libraries and so on.
706 * Directory Options:: Where to find header files and libraries.
707 Where to find the compiler executable files.
708 * Spec Files:: How to pass switches to sub-processes.
709 * Target Options:: Running a cross-compiler, or an old version of GCC.
712 @node Overall Options
713 @section Options Controlling the Kind of Output
715 Compilation can involve up to four stages: preprocessing, compilation
716 proper, assembly and linking, always in that order. GCC is capable of
717 preprocessing and compiling several files either into several
718 assembler input files, or into one assembler input file; then each
719 assembler input file produces an object file, and linking combines all
720 the object files (those newly compiled, and those specified as input)
721 into an executable file.
723 @cindex file name suffix
724 For any given input file, the file name suffix determines what kind of
729 C source code which must be preprocessed.
732 C source code which should not be preprocessed.
735 C++ source code which should not be preprocessed.
738 Objective-C source code. Note that you must link with the library
739 @file{libobjc.a} to make an Objective-C program work.
742 Objective-C source code which should not be preprocessed.
745 C or C++ header file to be turned into a precompiled header.
749 @itemx @var{file}.cxx
750 @itemx @var{file}.cpp
751 @itemx @var{file}.CPP
752 @itemx @var{file}.c++
754 C++ source code which must be preprocessed. Note that in @samp{.cxx},
755 the last two letters must both be literally @samp{x}. Likewise,
756 @samp{.C} refers to a literal capital C@.
760 C++ header file to be turned into a precompiled header.
763 @itemx @var{file}.for
764 @itemx @var{file}.FOR
765 Fortran source code which should not be preprocessed.
768 @itemx @var{file}.fpp
769 @itemx @var{file}.FPP
770 Fortran source code which must be preprocessed (with the traditional
774 Fortran source code which must be preprocessed with a RATFOR
775 preprocessor (not included with GCC)@.
777 @xref{Overall Options,,Options Controlling the Kind of Output, g77,
778 Using and Porting GNU Fortran}, for more details of the handling of
781 @c FIXME: Descriptions of Java file types.
788 Ada source code file which contains a library unit declaration (a
789 declaration of a package, subprogram, or generic, or a generic
790 instantiation), or a library unit renaming declaration (a package,
791 generic, or subprogram renaming declaration). Such files are also
794 @itemx @var{file}.adb
795 Ada source code file containing a library unit body (a subprogram or
796 package body). Such files are also called @dfn{bodies}.
798 @c GCC also knows about some suffixes for languages not yet included:
807 Assembler code which must be preprocessed.
810 An object file to be fed straight into linking.
811 Any file name with no recognized suffix is treated this way.
815 You can specify the input language explicitly with the @option{-x} option:
818 @item -x @var{language}
819 Specify explicitly the @var{language} for the following input files
820 (rather than letting the compiler choose a default based on the file
821 name suffix). This option applies to all following input files until
822 the next @option{-x} option. Possible values for @var{language} are:
824 c c-header cpp-output
825 c++ c++-header c++-cpp-output
826 objective-c objective-c-header objc-cpp-output
827 assembler assembler-with-cpp
829 f77 f77-cpp-input ratfor
835 Turn off any specification of a language, so that subsequent files are
836 handled according to their file name suffixes (as they are if @option{-x}
837 has not been used at all).
839 @item -pass-exit-codes
840 @opindex pass-exit-codes
841 Normally the @command{gcc} program will exit with the code of 1 if any
842 phase of the compiler returns a non-success return code. If you specify
843 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
844 numerically highest error produced by any phase that returned an error
848 If you only want some of the stages of compilation, you can use
849 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
850 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
851 @command{gcc} is to stop. Note that some combinations (for example,
852 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
857 Compile or assemble the source files, but do not link. The linking
858 stage simply is not done. The ultimate output is in the form of an
859 object file for each source file.
861 By default, the object file name for a source file is made by replacing
862 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
864 Unrecognized input files, not requiring compilation or assembly, are
869 Stop after the stage of compilation proper; do not assemble. The output
870 is in the form of an assembler code file for each non-assembler input
873 By default, the assembler file name for a source file is made by
874 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
876 Input files that don't require compilation are ignored.
880 Stop after the preprocessing stage; do not run the compiler proper. The
881 output is in the form of preprocessed source code, which is sent to the
884 Input files which don't require preprocessing are ignored.
886 @cindex output file option
889 Place output in file @var{file}. This applies regardless to whatever
890 sort of output is being produced, whether it be an executable file,
891 an object file, an assembler file or preprocessed C code.
893 If you specify @option{-o} when compiling more than one input file, or
894 you are producing an executable file as output, all the source files
895 on the command line will be compiled at once.
897 If @option{-o} is not specified, the default is to put an executable file
898 in @file{a.out}, the object file for @file{@var{source}.@var{suffix}} in
899 @file{@var{source}.o}, its assembler file in @file{@var{source}.s}, and
900 all preprocessed C source on standard output.
904 Print (on standard error output) the commands executed to run the stages
905 of compilation. Also print the version number of the compiler driver
906 program and of the preprocessor and the compiler proper.
910 Like @option{-v} except the commands are not executed and all command
911 arguments are quoted. This is useful for shell scripts to capture the
912 driver-generated command lines.
916 Use pipes rather than temporary files for communication between the
917 various stages of compilation. This fails to work on some systems where
918 the assembler is unable to read from a pipe; but the GNU assembler has
923 Print (on the standard output) a description of the command line options
924 understood by @command{gcc}. If the @option{-v} option is also specified
925 then @option{--help} will also be passed on to the various processes
926 invoked by @command{gcc}, so that they can display the command line options
927 they accept. If the @option{-Wextra} option is also specified then command
928 line options which have no documentation associated with them will also
933 Print (on the standard output) a description of target specific command
934 line options for each tool.
938 Display the version number and copyrights of the invoked GCC.
942 @section Compiling C++ Programs
944 @cindex suffixes for C++ source
945 @cindex C++ source file suffixes
946 C++ source files conventionally use one of the suffixes @samp{.C},
947 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
948 @samp{.cxx}; C++ header files often use @samp{.hh} or @samp{.H}; and
949 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
950 files with these names and compiles them as C++ programs even if you
951 call the compiler the same way as for compiling C programs (usually
952 with the name @command{gcc}).
956 However, C++ programs often require class libraries as well as a
957 compiler that understands the C++ language---and under some
958 circumstances, you might want to compile programs or header files from
959 standard input, or otherwise without a suffix that flags them as C++
960 programs. You might also like to precompile a C header file with a
961 @samp{.h} extension to be used in C++ compilations. @command{g++} is a
962 program that calls GCC with the default language set to C++, and
963 automatically specifies linking against the C++ library. On many
964 systems, @command{g++} is also installed with the name @command{c++}.
966 @cindex invoking @command{g++}
967 When you compile C++ programs, you may specify many of the same
968 command-line options that you use for compiling programs in any
969 language; or command-line options meaningful for C and related
970 languages; or options that are meaningful only for C++ programs.
971 @xref{C Dialect Options,,Options Controlling C Dialect}, for
972 explanations of options for languages related to C@.
973 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
974 explanations of options that are meaningful only for C++ programs.
976 @node C Dialect Options
977 @section Options Controlling C Dialect
978 @cindex dialect options
979 @cindex language dialect options
980 @cindex options, dialect
982 The following options control the dialect of C (or languages derived
983 from C, such as C++ and Objective-C) that the compiler accepts:
990 In C mode, support all ISO C90 programs. In C++ mode,
991 remove GNU extensions that conflict with ISO C++.
993 This turns off certain features of GCC that are incompatible with ISO
994 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
995 such as the @code{asm} and @code{typeof} keywords, and
996 predefined macros such as @code{unix} and @code{vax} that identify the
997 type of system you are using. It also enables the undesirable and
998 rarely used ISO trigraph feature. For the C compiler,
999 it disables recognition of C++ style @samp{//} comments as well as
1000 the @code{inline} keyword.
1002 The alternate keywords @code{__asm__}, @code{__extension__},
1003 @code{__inline__} and @code{__typeof__} continue to work despite
1004 @option{-ansi}. You would not want to use them in an ISO C program, of
1005 course, but it is useful to put them in header files that might be included
1006 in compilations done with @option{-ansi}. Alternate predefined macros
1007 such as @code{__unix__} and @code{__vax__} are also available, with or
1008 without @option{-ansi}.
1010 The @option{-ansi} option does not cause non-ISO programs to be
1011 rejected gratuitously. For that, @option{-pedantic} is required in
1012 addition to @option{-ansi}. @xref{Warning Options}.
1014 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1015 option is used. Some header files may notice this macro and refrain
1016 from declaring certain functions or defining certain macros that the
1017 ISO standard doesn't call for; this is to avoid interfering with any
1018 programs that might use these names for other things.
1020 Functions which would normally be built in but do not have semantics
1021 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1022 functions with @option{-ansi} is used. @xref{Other Builtins,,Other
1023 built-in functions provided by GCC}, for details of the functions
1028 Determine the language standard. This option is currently only
1029 supported when compiling C or C++. A value for this option must be
1030 provided; possible values are
1035 ISO C90 (same as @option{-ansi}).
1037 @item iso9899:199409
1038 ISO C90 as modified in amendment 1.
1044 ISO C99. Note that this standard is not yet fully supported; see
1045 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1046 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1049 Default, ISO C90 plus GNU extensions (including some C99 features).
1053 ISO C99 plus GNU extensions. When ISO C99 is fully implemented in GCC,
1054 this will become the default. The name @samp{gnu9x} is deprecated.
1057 The 1998 ISO C++ standard plus amendments.
1060 The same as @option{-std=c++98} plus GNU extensions. This is the
1061 default for C++ code.
1064 Even when this option is not specified, you can still use some of the
1065 features of newer standards in so far as they do not conflict with
1066 previous C standards. For example, you may use @code{__restrict__} even
1067 when @option{-std=c99} is not specified.
1069 The @option{-std} options specifying some version of ISO C have the same
1070 effects as @option{-ansi}, except that features that were not in ISO C90
1071 but are in the specified version (for example, @samp{//} comments and
1072 the @code{inline} keyword in ISO C99) are not disabled.
1074 @xref{Standards,,Language Standards Supported by GCC}, for details of
1075 these standard versions.
1077 @item -aux-info @var{filename}
1079 Output to the given filename prototyped declarations for all functions
1080 declared and/or defined in a translation unit, including those in header
1081 files. This option is silently ignored in any language other than C@.
1083 Besides declarations, the file indicates, in comments, the origin of
1084 each declaration (source file and line), whether the declaration was
1085 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1086 @samp{O} for old, respectively, in the first character after the line
1087 number and the colon), and whether it came from a declaration or a
1088 definition (@samp{C} or @samp{F}, respectively, in the following
1089 character). In the case of function definitions, a K&R-style list of
1090 arguments followed by their declarations is also provided, inside
1091 comments, after the declaration.
1095 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1096 keyword, so that code can use these words as identifiers. You can use
1097 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1098 instead. @option{-ansi} implies @option{-fno-asm}.
1100 In C++, this switch only affects the @code{typeof} keyword, since
1101 @code{asm} and @code{inline} are standard keywords. You may want to
1102 use the @option{-fno-gnu-keywords} flag instead, which has the same
1103 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1104 switch only affects the @code{asm} and @code{typeof} keywords, since
1105 @code{inline} is a standard keyword in ISO C99.
1108 @itemx -fno-builtin-@var{function}
1109 @opindex fno-builtin
1110 @cindex built-in functions
1111 Don't recognize built-in functions that do not begin with
1112 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1113 functions provided by GCC}, for details of the functions affected,
1114 including those which are not built-in functions when @option{-ansi} or
1115 @option{-std} options for strict ISO C conformance are used because they
1116 do not have an ISO standard meaning.
1118 GCC normally generates special code to handle certain built-in functions
1119 more efficiently; for instance, calls to @code{alloca} may become single
1120 instructions that adjust the stack directly, and calls to @code{memcpy}
1121 may become inline copy loops. The resulting code is often both smaller
1122 and faster, but since the function calls no longer appear as such, you
1123 cannot set a breakpoint on those calls, nor can you change the behavior
1124 of the functions by linking with a different library.
1126 With the @option{-fno-builtin-@var{function}} option
1127 only the built-in function @var{function} is
1128 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1129 function is named this is not built-in in this version of GCC, this
1130 option is ignored. There is no corresponding
1131 @option{-fbuiltin-@var{function}} option; if you wish to enable
1132 built-in functions selectively when using @option{-fno-builtin} or
1133 @option{-ffreestanding}, you may define macros such as:
1136 #define abs(n) __builtin_abs ((n))
1137 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1142 @cindex hosted environment
1144 Assert that compilation takes place in a hosted environment. This implies
1145 @option{-fbuiltin}. A hosted environment is one in which the
1146 entire standard library is available, and in which @code{main} has a return
1147 type of @code{int}. Examples are nearly everything except a kernel.
1148 This is equivalent to @option{-fno-freestanding}.
1150 @item -ffreestanding
1151 @opindex ffreestanding
1152 @cindex hosted environment
1154 Assert that compilation takes place in a freestanding environment. This
1155 implies @option{-fno-builtin}. A freestanding environment
1156 is one in which the standard library may not exist, and program startup may
1157 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1158 This is equivalent to @option{-fno-hosted}.
1160 @xref{Standards,,Language Standards Supported by GCC}, for details of
1161 freestanding and hosted environments.
1163 @item -fms-extensions
1164 @opindex fms-extensions
1165 Accept some non-standard constructs used in Microsoft header files.
1169 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1170 options for strict ISO C conformance) implies @option{-trigraphs}.
1172 @item -no-integrated-cpp
1173 @opindex no-integrated-cpp
1174 Performs a compilation in two passes: preprocessing and compiling. This
1175 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1176 @option{-B} option. The user supplied compilation step can then add in
1177 an additional preprocessing step after normal preprocessing but before
1178 compiling. The default is to use the integrated cpp (internal cpp)
1180 The semantics of this option will change if "cc1", "cc1plus", and
1181 "cc1obj" are merged.
1183 @cindex traditional C language
1184 @cindex C language, traditional
1186 @itemx -traditional-cpp
1187 @opindex traditional-cpp
1188 @opindex traditional
1189 Formerly, these options caused GCC to attempt to emulate a pre-standard
1190 C compiler. They are now only supported with the @option{-E} switch.
1191 The preprocessor continues to support a pre-standard mode. See the GNU
1192 CPP manual for details.
1194 @item -fcond-mismatch
1195 @opindex fcond-mismatch
1196 Allow conditional expressions with mismatched types in the second and
1197 third arguments. The value of such an expression is void. This option
1198 is not supported for C++.
1200 @item -funsigned-char
1201 @opindex funsigned-char
1202 Let the type @code{char} be unsigned, like @code{unsigned char}.
1204 Each kind of machine has a default for what @code{char} should
1205 be. It is either like @code{unsigned char} by default or like
1206 @code{signed char} by default.
1208 Ideally, a portable program should always use @code{signed char} or
1209 @code{unsigned char} when it depends on the signedness of an object.
1210 But many programs have been written to use plain @code{char} and
1211 expect it to be signed, or expect it to be unsigned, depending on the
1212 machines they were written for. This option, and its inverse, let you
1213 make such a program work with the opposite default.
1215 The type @code{char} is always a distinct type from each of
1216 @code{signed char} or @code{unsigned char}, even though its behavior
1217 is always just like one of those two.
1220 @opindex fsigned-char
1221 Let the type @code{char} be signed, like @code{signed char}.
1223 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1224 the negative form of @option{-funsigned-char}. Likewise, the option
1225 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1227 @item -fsigned-bitfields
1228 @itemx -funsigned-bitfields
1229 @itemx -fno-signed-bitfields
1230 @itemx -fno-unsigned-bitfields
1231 @opindex fsigned-bitfields
1232 @opindex funsigned-bitfields
1233 @opindex fno-signed-bitfields
1234 @opindex fno-unsigned-bitfields
1235 These options control whether a bit-field is signed or unsigned, when the
1236 declaration does not use either @code{signed} or @code{unsigned}. By
1237 default, such a bit-field is signed, because this is consistent: the
1238 basic integer types such as @code{int} are signed types.
1240 @item -fwritable-strings
1241 @opindex fwritable-strings
1242 Store string constants in the writable data segment and don't uniquize
1243 them. This is for compatibility with old programs which assume they can
1244 write into string constants.
1246 Writing into string constants is a very bad idea; ``constants'' should
1249 This option is deprecated.
1252 @node C++ Dialect Options
1253 @section Options Controlling C++ Dialect
1255 @cindex compiler options, C++
1256 @cindex C++ options, command line
1257 @cindex options, C++
1258 This section describes the command-line options that are only meaningful
1259 for C++ programs; but you can also use most of the GNU compiler options
1260 regardless of what language your program is in. For example, you
1261 might compile a file @code{firstClass.C} like this:
1264 g++ -g -frepo -O -c firstClass.C
1268 In this example, only @option{-frepo} is an option meant
1269 only for C++ programs; you can use the other options with any
1270 language supported by GCC@.
1272 Here is a list of options that are @emph{only} for compiling C++ programs:
1276 @item -fabi-version=@var{n}
1277 @opindex fabi-version
1278 Use version @var{n} of the C++ ABI. Version 2 is the version of the
1279 C++ ABI that first appeared in G++ 3.4. Version 1 is the version of
1280 the C++ ABI that first appeared in G++ 3.2. Version 0 will always be
1281 the version that conforms most closely to the C++ ABI specification.
1282 Therefore, the ABI obtained using version 0 will change as ABI bugs
1285 The default is version 1.
1287 @item -fno-access-control
1288 @opindex fno-access-control
1289 Turn off all access checking. This switch is mainly useful for working
1290 around bugs in the access control code.
1294 Check that the pointer returned by @code{operator new} is non-null
1295 before attempting to modify the storage allocated. This check is
1296 normally unnecessary because the C++ standard specifies that
1297 @code{operator new} will only return @code{0} if it is declared
1298 @samp{throw()}, in which case the compiler will always check the
1299 return value even without this option. In all other cases, when
1300 @code{operator new} has a non-empty exception specification, memory
1301 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1302 @samp{new (nothrow)}.
1304 @item -fconserve-space
1305 @opindex fconserve-space
1306 Put uninitialized or runtime-initialized global variables into the
1307 common segment, as C does. This saves space in the executable at the
1308 cost of not diagnosing duplicate definitions. If you compile with this
1309 flag and your program mysteriously crashes after @code{main()} has
1310 completed, you may have an object that is being destroyed twice because
1311 two definitions were merged.
1313 This option is no longer useful on most targets, now that support has
1314 been added for putting variables into BSS without making them common.
1316 @item -fno-const-strings
1317 @opindex fno-const-strings
1318 Give string constants type @code{char *} instead of type @code{const
1319 char *}. By default, G++ uses type @code{const char *} as required by
1320 the standard. Even if you use @option{-fno-const-strings}, you cannot
1321 actually modify the value of a string constant, unless you also use
1322 @option{-fwritable-strings}.
1324 This option might be removed in a future release of G++. For maximum
1325 portability, you should structure your code so that it works with
1326 string constants that have type @code{const char *}.
1328 @item -fno-elide-constructors
1329 @opindex fno-elide-constructors
1330 The C++ standard allows an implementation to omit creating a temporary
1331 which is only used to initialize another object of the same type.
1332 Specifying this option disables that optimization, and forces G++ to
1333 call the copy constructor in all cases.
1335 @item -fno-enforce-eh-specs
1336 @opindex fno-enforce-eh-specs
1337 Don't check for violation of exception specifications at runtime. This
1338 option violates the C++ standard, but may be useful for reducing code
1339 size in production builds, much like defining @samp{NDEBUG}. The compiler
1340 will still optimize based on the exception specifications.
1343 @itemx -fno-for-scope
1345 @opindex fno-for-scope
1346 If @option{-ffor-scope} is specified, the scope of variables declared in
1347 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1348 as specified by the C++ standard.
1349 If @option{-fno-for-scope} is specified, the scope of variables declared in
1350 a @i{for-init-statement} extends to the end of the enclosing scope,
1351 as was the case in old versions of G++, and other (traditional)
1352 implementations of C++.
1354 The default if neither flag is given to follow the standard,
1355 but to allow and give a warning for old-style code that would
1356 otherwise be invalid, or have different behavior.
1358 @item -fno-gnu-keywords
1359 @opindex fno-gnu-keywords
1360 Do not recognize @code{typeof} as a keyword, so that code can use this
1361 word as an identifier. You can use the keyword @code{__typeof__} instead.
1362 @option{-ansi} implies @option{-fno-gnu-keywords}.
1364 @item -fno-implicit-templates
1365 @opindex fno-implicit-templates
1366 Never emit code for non-inline templates which are instantiated
1367 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1368 @xref{Template Instantiation}, for more information.
1370 @item -fno-implicit-inline-templates
1371 @opindex fno-implicit-inline-templates
1372 Don't emit code for implicit instantiations of inline templates, either.
1373 The default is to handle inlines differently so that compiles with and
1374 without optimization will need the same set of explicit instantiations.
1376 @item -fno-implement-inlines
1377 @opindex fno-implement-inlines
1378 To save space, do not emit out-of-line copies of inline functions
1379 controlled by @samp{#pragma implementation}. This will cause linker
1380 errors if these functions are not inlined everywhere they are called.
1382 @item -fms-extensions
1383 @opindex fms-extensions
1384 Disable pedantic warnings about constructs used in MFC, such as implicit
1385 int and getting a pointer to member function via non-standard syntax.
1387 @item -fno-nonansi-builtins
1388 @opindex fno-nonansi-builtins
1389 Disable built-in declarations of functions that are not mandated by
1390 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1391 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1393 @item -fno-operator-names
1394 @opindex fno-operator-names
1395 Do not treat the operator name keywords @code{and}, @code{bitand},
1396 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1397 synonyms as keywords.
1399 @item -fno-optional-diags
1400 @opindex fno-optional-diags
1401 Disable diagnostics that the standard says a compiler does not need to
1402 issue. Currently, the only such diagnostic issued by G++ is the one for
1403 a name having multiple meanings within a class.
1406 @opindex fpermissive
1407 Downgrade some diagnostics about nonconformant code from errors to
1408 warnings. Thus, using @option{-fpermissive} will allow some
1409 nonconforming code to compile.
1413 Enable automatic template instantiation at link time. This option also
1414 implies @option{-fno-implicit-templates}. @xref{Template
1415 Instantiation}, for more information.
1419 Disable generation of information about every class with virtual
1420 functions for use by the C++ runtime type identification features
1421 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1422 of the language, you can save some space by using this flag. Note that
1423 exception handling uses the same information, but it will generate it as
1428 Emit statistics about front-end processing at the end of the compilation.
1429 This information is generally only useful to the G++ development team.
1431 @item -ftemplate-depth-@var{n}
1432 @opindex ftemplate-depth
1433 Set the maximum instantiation depth for template classes to @var{n}.
1434 A limit on the template instantiation depth is needed to detect
1435 endless recursions during template class instantiation. ANSI/ISO C++
1436 conforming programs must not rely on a maximum depth greater than 17.
1438 @item -fuse-cxa-atexit
1439 @opindex fuse-cxa-atexit
1440 Register destructors for objects with static storage duration with the
1441 @code{__cxa_atexit} function rather than the @code{atexit} function.
1442 This option is required for fully standards-compliant handling of static
1443 destructors, but will only work if your C library supports
1444 @code{__cxa_atexit}.
1448 Do not use weak symbol support, even if it is provided by the linker.
1449 By default, G++ will use weak symbols if they are available. This
1450 option exists only for testing, and should not be used by end-users;
1451 it will result in inferior code and has no benefits. This option may
1452 be removed in a future release of G++.
1456 Do not search for header files in the standard directories specific to
1457 C++, but do still search the other standard directories. (This option
1458 is used when building the C++ library.)
1461 In addition, these optimization, warning, and code generation options
1462 have meanings only for C++ programs:
1465 @item -fno-default-inline
1466 @opindex fno-default-inline
1467 Do not assume @samp{inline} for functions defined inside a class scope.
1468 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1469 functions will have linkage like inline functions; they just won't be
1472 @item -Wabi @r{(C++ only)}
1474 Warn when G++ generates code that is probably not compatible with the
1475 vendor-neutral C++ ABI. Although an effort has been made to warn about
1476 all such cases, there are probably some cases that are not warned about,
1477 even though G++ is generating incompatible code. There may also be
1478 cases where warnings are emitted even though the code that is generated
1481 You should rewrite your code to avoid these warnings if you are
1482 concerned about the fact that code generated by G++ may not be binary
1483 compatible with code generated by other compilers.
1485 The known incompatibilities at this point include:
1490 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
1491 pack data into the same byte as a base class. For example:
1494 struct A @{ virtual void f(); int f1 : 1; @};
1495 struct B : public A @{ int f2 : 1; @};
1499 In this case, G++ will place @code{B::f2} into the same byte
1500 as@code{A::f1}; other compilers will not. You can avoid this problem
1501 by explicitly padding @code{A} so that its size is a multiple of the
1502 byte size on your platform; that will cause G++ and other compilers to
1503 layout @code{B} identically.
1506 Incorrect handling of tail-padding for virtual bases. G++ does not use
1507 tail padding when laying out virtual bases. For example:
1510 struct A @{ virtual void f(); char c1; @};
1511 struct B @{ B(); char c2; @};
1512 struct C : public A, public virtual B @{@};
1516 In this case, G++ will not place @code{B} into the tail-padding for
1517 @code{A}; other compilers will. You can avoid this problem by
1518 explicitly padding @code{A} so that its size is a multiple of its
1519 alignment (ignoring virtual base classes); that will cause G++ and other
1520 compilers to layout @code{C} identically.
1523 Incorrect handling of bit-fields with declared widths greater than that
1524 of their underlying types, when the bit-fields appear in a union. For
1528 union U @{ int i : 4096; @};
1532 Assuming that an @code{int} does not have 4096 bits, G++ will make the
1533 union too small by the number of bits in an @code{int}.
1536 Empty classes can be placed at incorrect offsets. For example:
1546 struct C : public B, public A @{@};
1550 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
1551 it should be placed at offset zero. G++ mistakenly believes that the
1552 @code{A} data member of @code{B} is already at offset zero.
1555 Names of template functions whose types involve @code{typename} or
1556 template template parameters can be mangled incorrectly.
1559 template <typename Q>
1560 void f(typename Q::X) @{@}
1562 template <template <typename> class Q>
1563 void f(typename Q<int>::X) @{@}
1567 Instantiations of these templates may be mangled incorrectly.
1571 @item -Wctor-dtor-privacy @r{(C++ only)}
1572 @opindex Wctor-dtor-privacy
1573 Warn when a class seems unusable because all the constructors or
1574 destructors in that class are private, and it has neither friends nor
1575 public static member functions.
1577 @item -Wnon-virtual-dtor @r{(C++ only)}
1578 @opindex Wnon-virtual-dtor
1579 Warn when a class appears to be polymorphic, thereby requiring a virtual
1580 destructor, yet it declares a non-virtual one.
1581 This warning is enabled by @option{-Wall}.
1583 @item -Wreorder @r{(C++ only)}
1585 @cindex reordering, warning
1586 @cindex warning for reordering of member initializers
1587 Warn when the order of member initializers given in the code does not
1588 match the order in which they must be executed. For instance:
1594 A(): j (0), i (1) @{ @}
1598 The compiler will rearrange the member initializers for @samp{i}
1599 and @samp{j} to match the declaration order of the members, emitting
1600 a warning to that effect. This warning is enabled by @option{-Wall}.
1603 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
1606 @item -Weffc++ @r{(C++ only)}
1608 Warn about violations of the following style guidelines from Scott Meyers'
1609 @cite{Effective C++} book:
1613 Item 11: Define a copy constructor and an assignment operator for classes
1614 with dynamically allocated memory.
1617 Item 12: Prefer initialization to assignment in constructors.
1620 Item 14: Make destructors virtual in base classes.
1623 Item 15: Have @code{operator=} return a reference to @code{*this}.
1626 Item 23: Don't try to return a reference when you must return an object.
1630 Also warn about violations of the following style guidelines from
1631 Scott Meyers' @cite{More Effective C++} book:
1635 Item 6: Distinguish between prefix and postfix forms of increment and
1636 decrement operators.
1639 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
1643 When selecting this option, be aware that the standard library
1644 headers do not obey all of these guidelines; use @samp{grep -v}
1645 to filter out those warnings.
1647 @item -Wno-deprecated @r{(C++ only)}
1648 @opindex Wno-deprecated
1649 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1651 @item -Wno-non-template-friend @r{(C++ only)}
1652 @opindex Wno-non-template-friend
1653 Disable warnings when non-templatized friend functions are declared
1654 within a template. Since the advent of explicit template specification
1655 support in G++, if the name of the friend is an unqualified-id (i.e.,
1656 @samp{friend foo(int)}), the C++ language specification demands that the
1657 friend declare or define an ordinary, nontemplate function. (Section
1658 14.5.3). Before G++ implemented explicit specification, unqualified-ids
1659 could be interpreted as a particular specialization of a templatized
1660 function. Because this non-conforming behavior is no longer the default
1661 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
1662 check existing code for potential trouble spots and is on by default.
1663 This new compiler behavior can be turned off with
1664 @option{-Wno-non-template-friend} which keeps the conformant compiler code
1665 but disables the helpful warning.
1667 @item -Wold-style-cast @r{(C++ only)}
1668 @opindex Wold-style-cast
1669 Warn if an old-style (C-style) cast to a non-void type is used within
1670 a C++ program. The new-style casts (@samp{static_cast},
1671 @samp{reinterpret_cast}, and @samp{const_cast}) are less vulnerable to
1672 unintended effects and much easier to search for.
1674 @item -Woverloaded-virtual @r{(C++ only)}
1675 @opindex Woverloaded-virtual
1676 @cindex overloaded virtual fn, warning
1677 @cindex warning for overloaded virtual fn
1678 Warn when a function declaration hides virtual functions from a
1679 base class. For example, in:
1686 struct B: public A @{
1691 the @code{A} class version of @code{f} is hidden in @code{B}, and code
1699 will fail to compile.
1701 @item -Wno-pmf-conversions @r{(C++ only)}
1702 @opindex Wno-pmf-conversions
1703 Disable the diagnostic for converting a bound pointer to member function
1706 @item -Wsign-promo @r{(C++ only)}
1707 @opindex Wsign-promo
1708 Warn when overload resolution chooses a promotion from unsigned or
1709 enumeral type to a signed type, over a conversion to an unsigned type of
1710 the same size. Previous versions of G++ would try to preserve
1711 unsignedness, but the standard mandates the current behavior.
1713 @item -Wsynth @r{(C++ only)}
1715 @cindex warning for synthesized methods
1716 @cindex synthesized methods, warning
1717 Warn when G++'s synthesis behavior does not match that of cfront. For
1723 A& operator = (int);
1733 In this example, G++ will synthesize a default @samp{A& operator =
1734 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1737 @node Objective-C Dialect Options
1738 @section Options Controlling Objective-C Dialect
1740 @cindex compiler options, Objective-C
1741 @cindex Objective-C options, command line
1742 @cindex options, Objective-C
1743 (NOTE: This manual does not describe the Objective-C language itself. See
1744 @w{@uref{http://gcc.gnu.org/readings.html}} for references.)
1746 This section describes the command-line options that are only meaningful
1747 for Objective-C programs, but you can also use most of the GNU compiler
1748 options regardless of what language your program is in. For example,
1749 you might compile a file @code{some_class.m} like this:
1752 gcc -g -fgnu-runtime -O -c some_class.m
1756 In this example, @option{-fgnu-runtime} is an option meant only for
1757 Objective-C programs; you can use the other options with any language
1760 Here is a list of options that are @emph{only} for compiling Objective-C
1764 @item -fconstant-string-class=@var{class-name}
1765 @opindex fconstant-string-class
1766 Use @var{class-name} as the name of the class to instantiate for each
1767 literal string specified with the syntax @code{@@"@dots{}"}. The default
1768 class name is @code{NXConstantString} if the GNU runtime is being used, and
1769 @code{NSConstantString} if the NeXT runtime is being used (see below). The
1770 @option{-fconstant-cfstrings} option, if also present, will override the
1771 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
1772 to be laid out as constant CoreFoundation strings.
1775 @opindex fgnu-runtime
1776 Generate object code compatible with the standard GNU Objective-C
1777 runtime. This is the default for most types of systems.
1779 @item -fnext-runtime
1780 @opindex fnext-runtime
1781 Generate output compatible with the NeXT runtime. This is the default
1782 for NeXT-based systems, including Darwin and Mac OS X@. The macro
1783 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
1786 @item -fno-nil-receivers
1787 @opindex -fno-nil-receivers
1788 Assume that all Objective-C message dispatches (e.g.,
1789 @code{[receiver message:arg]}) in this translation unit ensure that the receiver
1790 is not @code{nil}. This allows for more efficient entry points in the runtime to be
1791 used. Currently, this option is only available in conjunction with
1792 the NeXT runtime on Mac OS X 10.3 and later.
1794 @item -fobjc-exceptions
1795 @opindex -fobjc-exceptions
1796 Enable syntactic support for structured exception handling in Objective-C,
1797 similar to what is offered by C++ and Java. Currently, this option is only
1798 available in conjunction with the NeXT runtime on Mac OS X 10.3 and later.
1806 @@catch (AnObjCClass *exc) @{
1813 @@catch (AnotherClass *exc) @{
1816 @@catch (id allOthers) @{
1826 The @code{@@throw} statement may appear anywhere in an Objective-C or
1827 Objective-C++ program; when used inside of a @code{@@catch} block, the
1828 @code{@@throw} may appear without an argument (as shown above), in which case
1829 the object caught by the @code{@@catch} will be rethrown.
1831 Note that only (pointers to) Objective-C objects may be thrown and
1832 caught using this scheme. When an object is thrown, it will be caught
1833 by the nearest @code{@@catch} clause capable of handling objects of that type,
1834 analogously to how @code{catch} blocks work in C++ and Java. A
1835 @code{@@catch(id @dots{})} clause (as shown above) may also be provided to catch
1836 any and all Objective-C exceptions not caught by previous @code{@@catch}
1839 The @code{@@finally} clause, if present, will be executed upon exit from the
1840 immediately preceding @code{@@try @dots{} @@catch} section. This will happen
1841 regardless of whether any exceptions are thrown, caught or rethrown
1842 inside the @code{@@try @dots{} @@catch} section, analogously to the behavior
1843 of the @code{finally} clause in Java.
1845 There are several caveats to using the new exception mechanism:
1849 Although currently designed to be binary compatible with @code{NS_HANDLER}-style
1850 idioms provided by the @code{NSException} class, the new
1851 exceptions can only be used on Mac OS X 10.3 (Panther) and later
1852 systems, due to additional functionality needed in the (NeXT) Objective-C
1856 As mentioned above, the new exceptions do not support handling
1857 types other than Objective-C objects. Furthermore, when used from
1858 Objective-C++, the Objective-C exception model does not interoperate with C++
1859 exceptions at this time. This means you cannot @code{@@throw} an exception
1860 from Objective-C and @code{catch} it in C++, or vice versa
1861 (i.e., @code{throw @dots{} @@catch}).
1864 The @option{-fobjc-exceptions} switch also enables the use of synchronization
1865 blocks for thread-safe execution:
1868 @@synchronized (ObjCClass *guard) @{
1873 Upon entering the @code{@@synchronized} block, a thread of execution shall
1874 first check whether a lock has been placed on the corresponding @code{guard}
1875 object by another thread. If it has, the current thread shall wait until
1876 the other thread relinquishes its lock. Once @code{guard} becomes available,
1877 the current thread will place its own lock on it, execute the code contained in
1878 the @code{@@synchronized} block, and finally relinquish the lock (thereby
1879 making @code{guard} available to other threads).
1881 Unlike Java, Objective-C does not allow for entire methods to be marked
1882 @code{@@synchronized}. Note that throwing exceptions out of
1883 @code{@@synchronized} blocks is allowed, and will cause the guarding object
1884 to be unlocked properly.
1886 @item -freplace-objc-classes
1887 @opindex -freplace-objc-classes
1888 Emit a special marker instructing @command{ld(1)} not to statically link in
1889 the resulting object file, and allow @command{dyld(1)} to load it in at
1890 run time instead. This is used in conjunction with the Fix-and-Continue
1891 debugging mode, where the object file in question may be recompiled and
1892 dynamically reloaded in the course of program execution, without the need
1893 to restart the program itself. Currently, Fix-and-Continue functionality
1894 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
1898 @opindex -fzero-link
1899 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
1900 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
1901 compile time) with static class references that get initialized at load time,
1902 which improves run-time performance. Specifying the @option{-fzero-link} flag
1903 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
1904 to be retained. This is useful in Zero-Link debugging mode, since it allows
1905 for individual class implementations to be modified during program execution.
1909 Dump interface declarations for all classes seen in the source file to a
1910 file named @file{@var{sourcename}.decl}.
1913 @opindex Wno-protocol
1914 If a class is declared to implement a protocol, a warning is issued for
1915 every method in the protocol that is not implemented by the class. The
1916 default behavior is to issue a warning for every method not explicitly
1917 implemented in the class, even if a method implementation is inherited
1918 from the superclass. If you use the @code{-Wno-protocol} option, then
1919 methods inherited from the superclass are considered to be implemented,
1920 and no warning is issued for them.
1924 Warn if multiple methods of different types for the same selector are
1925 found during compilation. The check is performed on the list of methods
1926 in the final stage of compilation. Additionally, a check is performed
1927 for each selector appearing in a @code{@@selector(@dots{})}
1928 expression, and a corresponding method for that selector has been found
1929 during compilation. Because these checks scan the method table only at
1930 the end of compilation, these warnings are not produced if the final
1931 stage of compilation is not reached, for example because an error is
1932 found during compilation, or because the @code{-fsyntax-only} option is
1935 @item -Wundeclared-selector
1936 @opindex Wundeclared-selector
1937 Warn if a @code{@@selector(@dots{})} expression referring to an
1938 undeclared selector is found. A selector is considered undeclared if no
1939 method with that name has been declared before the
1940 @code{@@selector(@dots{})} expression, either explicitly in an
1941 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
1942 an @code{@@implementation} section. This option always performs its
1943 checks as soon as a @code{@@selector(@dots{})} expression is found,
1944 while @code{-Wselector} only performs its checks in the final stage of
1945 compilation. This also enforces the coding style convention
1946 that methods and selectors must be declared before being used.
1948 @item -print-objc-runtime-info
1949 @opindex -print-objc-runtime-info
1950 Generate C header describing the largest structure that is passed by
1955 @node Language Independent Options
1956 @section Options to Control Diagnostic Messages Formatting
1957 @cindex options to control diagnostics formatting
1958 @cindex diagnostic messages
1959 @cindex message formatting
1961 Traditionally, diagnostic messages have been formatted irrespective of
1962 the output device's aspect (e.g.@: its width, @dots{}). The options described
1963 below can be used to control the diagnostic messages formatting
1964 algorithm, e.g.@: how many characters per line, how often source location
1965 information should be reported. Right now, only the C++ front end can
1966 honor these options. However it is expected, in the near future, that
1967 the remaining front ends would be able to digest them correctly.
1970 @item -fmessage-length=@var{n}
1971 @opindex fmessage-length
1972 Try to format error messages so that they fit on lines of about @var{n}
1973 characters. The default is 72 characters for @command{g++} and 0 for the rest of
1974 the front ends supported by GCC@. If @var{n} is zero, then no
1975 line-wrapping will be done; each error message will appear on a single
1978 @opindex fdiagnostics-show-location
1979 @item -fdiagnostics-show-location=once
1980 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
1981 reporter to emit @emph{once} source location information; that is, in
1982 case the message is too long to fit on a single physical line and has to
1983 be wrapped, the source location won't be emitted (as prefix) again,
1984 over and over, in subsequent continuation lines. This is the default
1987 @item -fdiagnostics-show-location=every-line
1988 Only meaningful in line-wrapping mode. Instructs the diagnostic
1989 messages reporter to emit the same source location information (as
1990 prefix) for physical lines that result from the process of breaking
1991 a message which is too long to fit on a single line.
1995 @node Warning Options
1996 @section Options to Request or Suppress Warnings
1997 @cindex options to control warnings
1998 @cindex warning messages
1999 @cindex messages, warning
2000 @cindex suppressing warnings
2002 Warnings are diagnostic messages that report constructions which
2003 are not inherently erroneous but which are risky or suggest there
2004 may have been an error.
2006 You can request many specific warnings with options beginning @samp{-W},
2007 for example @option{-Wimplicit} to request warnings on implicit
2008 declarations. Each of these specific warning options also has a
2009 negative form beginning @samp{-Wno-} to turn off warnings;
2010 for example, @option{-Wno-implicit}. This manual lists only one of the
2011 two forms, whichever is not the default.
2013 The following options control the amount and kinds of warnings produced
2014 by GCC; for further, language-specific options also refer to
2015 @ref{C++ Dialect Options} and @ref{Objective-C Dialect Options}.
2018 @cindex syntax checking
2020 @opindex fsyntax-only
2021 Check the code for syntax errors, but don't do anything beyond that.
2025 Issue all the warnings demanded by strict ISO C and ISO C++;
2026 reject all programs that use forbidden extensions, and some other
2027 programs that do not follow ISO C and ISO C++. For ISO C, follows the
2028 version of the ISO C standard specified by any @option{-std} option used.
2030 Valid ISO C and ISO C++ programs should compile properly with or without
2031 this option (though a rare few will require @option{-ansi} or a
2032 @option{-std} option specifying the required version of ISO C)@. However,
2033 without this option, certain GNU extensions and traditional C and C++
2034 features are supported as well. With this option, they are rejected.
2036 @option{-pedantic} does not cause warning messages for use of the
2037 alternate keywords whose names begin and end with @samp{__}. Pedantic
2038 warnings are also disabled in the expression that follows
2039 @code{__extension__}. However, only system header files should use
2040 these escape routes; application programs should avoid them.
2041 @xref{Alternate Keywords}.
2043 Some users try to use @option{-pedantic} to check programs for strict ISO
2044 C conformance. They soon find that it does not do quite what they want:
2045 it finds some non-ISO practices, but not all---only those for which
2046 ISO C @emph{requires} a diagnostic, and some others for which
2047 diagnostics have been added.
2049 A feature to report any failure to conform to ISO C might be useful in
2050 some instances, but would require considerable additional work and would
2051 be quite different from @option{-pedantic}. We don't have plans to
2052 support such a feature in the near future.
2054 Where the standard specified with @option{-std} represents a GNU
2055 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
2056 corresponding @dfn{base standard}, the version of ISO C on which the GNU
2057 extended dialect is based. Warnings from @option{-pedantic} are given
2058 where they are required by the base standard. (It would not make sense
2059 for such warnings to be given only for features not in the specified GNU
2060 C dialect, since by definition the GNU dialects of C include all
2061 features the compiler supports with the given option, and there would be
2062 nothing to warn about.)
2064 @item -pedantic-errors
2065 @opindex pedantic-errors
2066 Like @option{-pedantic}, except that errors are produced rather than
2071 Inhibit all warning messages.
2075 Inhibit warning messages about the use of @samp{#import}.
2077 @item -Wchar-subscripts
2078 @opindex Wchar-subscripts
2079 Warn if an array subscript has type @code{char}. This is a common cause
2080 of error, as programmers often forget that this type is signed on some
2085 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
2086 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
2090 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
2091 the arguments supplied have types appropriate to the format string
2092 specified, and that the conversions specified in the format string make
2093 sense. This includes standard functions, and others specified by format
2094 attributes (@pxref{Function Attributes}), in the @code{printf},
2095 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
2096 not in the C standard) families.
2098 The formats are checked against the format features supported by GNU
2099 libc version 2.2. These include all ISO C90 and C99 features, as well
2100 as features from the Single Unix Specification and some BSD and GNU
2101 extensions. Other library implementations may not support all these
2102 features; GCC does not support warning about features that go beyond a
2103 particular library's limitations. However, if @option{-pedantic} is used
2104 with @option{-Wformat}, warnings will be given about format features not
2105 in the selected standard version (but not for @code{strfmon} formats,
2106 since those are not in any version of the C standard). @xref{C Dialect
2107 Options,,Options Controlling C Dialect}.
2109 Since @option{-Wformat} also checks for null format arguments for
2110 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
2112 @option{-Wformat} is included in @option{-Wall}. For more control over some
2113 aspects of format checking, the options @option{-Wformat-y2k},
2114 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
2115 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
2116 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
2119 @opindex Wformat-y2k
2120 If @option{-Wformat} is specified, also warn about @code{strftime}
2121 formats which may yield only a two-digit year.
2123 @item -Wno-format-extra-args
2124 @opindex Wno-format-extra-args
2125 If @option{-Wformat} is specified, do not warn about excess arguments to a
2126 @code{printf} or @code{scanf} format function. The C standard specifies
2127 that such arguments are ignored.
2129 Where the unused arguments lie between used arguments that are
2130 specified with @samp{$} operand number specifications, normally
2131 warnings are still given, since the implementation could not know what
2132 type to pass to @code{va_arg} to skip the unused arguments. However,
2133 in the case of @code{scanf} formats, this option will suppress the
2134 warning if the unused arguments are all pointers, since the Single
2135 Unix Specification says that such unused arguments are allowed.
2137 @item -Wno-format-zero-length
2138 @opindex Wno-format-zero-length
2139 If @option{-Wformat} is specified, do not warn about zero-length formats.
2140 The C standard specifies that zero-length formats are allowed.
2142 @item -Wformat-nonliteral
2143 @opindex Wformat-nonliteral
2144 If @option{-Wformat} is specified, also warn if the format string is not a
2145 string literal and so cannot be checked, unless the format function
2146 takes its format arguments as a @code{va_list}.
2148 @item -Wformat-security
2149 @opindex Wformat-security
2150 If @option{-Wformat} is specified, also warn about uses of format
2151 functions that represent possible security problems. At present, this
2152 warns about calls to @code{printf} and @code{scanf} functions where the
2153 format string is not a string literal and there are no format arguments,
2154 as in @code{printf (foo);}. This may be a security hole if the format
2155 string came from untrusted input and contains @samp{%n}. (This is
2156 currently a subset of what @option{-Wformat-nonliteral} warns about, but
2157 in future warnings may be added to @option{-Wformat-security} that are not
2158 included in @option{-Wformat-nonliteral}.)
2162 Enable @option{-Wformat} plus format checks not included in
2163 @option{-Wformat}. Currently equivalent to @samp{-Wformat
2164 -Wformat-nonliteral -Wformat-security -Wformat-y2k}.
2168 Warn about passing a null pointer for arguments marked as
2169 requiring a non-null value by the @code{nonnull} function attribute.
2171 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
2172 can be disabled with the @option{-Wno-nonnull} option.
2174 @item -Winit-self @r{(C, C++, and Objective-C only)}
2176 Warn about uninitialized variables which are initialized with themselves.
2177 Note this option can only be used with the @option{-Wuninitialized} option,
2178 which in turn only works with @option{-O1} and above.
2180 For example, GCC will warn about @code{i} being uninitialized in the
2181 following snippet only when @option{-Winit-self} has been specified:
2192 @item -Wimplicit-int
2193 @opindex Wimplicit-int
2194 Warn when a declaration does not specify a type.
2196 @item -Wimplicit-function-declaration
2197 @itemx -Werror-implicit-function-declaration
2198 @opindex Wimplicit-function-declaration
2199 @opindex Werror-implicit-function-declaration
2200 Give a warning (or error) whenever a function is used before being
2205 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
2209 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
2210 function with external linkage, returning int, taking either zero
2211 arguments, two, or three arguments of appropriate types.
2213 @item -Wmissing-braces
2214 @opindex Wmissing-braces
2215 Warn if an aggregate or union initializer is not fully bracketed. In
2216 the following example, the initializer for @samp{a} is not fully
2217 bracketed, but that for @samp{b} is fully bracketed.
2220 int a[2][2] = @{ 0, 1, 2, 3 @};
2221 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
2225 @opindex Wparentheses
2226 Warn if parentheses are omitted in certain contexts, such
2227 as when there is an assignment in a context where a truth value
2228 is expected, or when operators are nested whose precedence people
2229 often get confused about.
2231 Also warn about constructions where there may be confusion to which
2232 @code{if} statement an @code{else} branch belongs. Here is an example of
2247 In C, every @code{else} branch belongs to the innermost possible @code{if}
2248 statement, which in this example is @code{if (b)}. This is often not
2249 what the programmer expected, as illustrated in the above example by
2250 indentation the programmer chose. When there is the potential for this
2251 confusion, GCC will issue a warning when this flag is specified.
2252 To eliminate the warning, add explicit braces around the innermost
2253 @code{if} statement so there is no way the @code{else} could belong to
2254 the enclosing @code{if}. The resulting code would look like this:
2270 @item -Wsequence-point
2271 @opindex Wsequence-point
2272 Warn about code that may have undefined semantics because of violations
2273 of sequence point rules in the C standard.
2275 The C standard defines the order in which expressions in a C program are
2276 evaluated in terms of @dfn{sequence points}, which represent a partial
2277 ordering between the execution of parts of the program: those executed
2278 before the sequence point, and those executed after it. These occur
2279 after the evaluation of a full expression (one which is not part of a
2280 larger expression), after the evaluation of the first operand of a
2281 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
2282 function is called (but after the evaluation of its arguments and the
2283 expression denoting the called function), and in certain other places.
2284 Other than as expressed by the sequence point rules, the order of
2285 evaluation of subexpressions of an expression is not specified. All
2286 these rules describe only a partial order rather than a total order,
2287 since, for example, if two functions are called within one expression
2288 with no sequence point between them, the order in which the functions
2289 are called is not specified. However, the standards committee have
2290 ruled that function calls do not overlap.
2292 It is not specified when between sequence points modifications to the
2293 values of objects take effect. Programs whose behavior depends on this
2294 have undefined behavior; the C standard specifies that ``Between the
2295 previous and next sequence point an object shall have its stored value
2296 modified at most once by the evaluation of an expression. Furthermore,
2297 the prior value shall be read only to determine the value to be
2298 stored.''. If a program breaks these rules, the results on any
2299 particular implementation are entirely unpredictable.
2301 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
2302 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
2303 diagnosed by this option, and it may give an occasional false positive
2304 result, but in general it has been found fairly effective at detecting
2305 this sort of problem in programs.
2307 The present implementation of this option only works for C programs. A
2308 future implementation may also work for C++ programs.
2310 The C standard is worded confusingly, therefore there is some debate
2311 over the precise meaning of the sequence point rules in subtle cases.
2312 Links to discussions of the problem, including proposed formal
2313 definitions, may be found on our readings page, at
2314 @w{@uref{http://gcc.gnu.org/readings.html}}.
2317 @opindex Wreturn-type
2318 Warn whenever a function is defined with a return-type that defaults to
2319 @code{int}. Also warn about any @code{return} statement with no
2320 return-value in a function whose return-type is not @code{void}.
2322 For C++, a function without return type always produces a diagnostic
2323 message, even when @option{-Wno-return-type} is specified. The only
2324 exceptions are @samp{main} and functions defined in system headers.
2328 Warn whenever a @code{switch} statement has an index of enumeral type
2329 and lacks a @code{case} for one or more of the named codes of that
2330 enumeration. (The presence of a @code{default} label prevents this
2331 warning.) @code{case} labels outside the enumeration range also
2332 provoke warnings when this option is used.
2334 @item -Wswitch-default
2335 @opindex Wswitch-switch
2336 Warn whenever a @code{switch} statement does not have a @code{default}
2340 @opindex Wswitch-enum
2341 Warn whenever a @code{switch} statement has an index of enumeral type
2342 and lacks a @code{case} for one or more of the named codes of that
2343 enumeration. @code{case} labels outside the enumeration range also
2344 provoke warnings when this option is used.
2348 Warn if any trigraphs are encountered that might change the meaning of
2349 the program (trigraphs within comments are not warned about).
2351 @item -Wunused-function
2352 @opindex Wunused-function
2353 Warn whenever a static function is declared but not defined or a
2354 non\-inline static function is unused.
2356 @item -Wunused-label
2357 @opindex Wunused-label
2358 Warn whenever a label is declared but not used.
2360 To suppress this warning use the @samp{unused} attribute
2361 (@pxref{Variable Attributes}).
2363 @item -Wunused-parameter
2364 @opindex Wunused-parameter
2365 Warn whenever a function parameter is unused aside from its declaration.
2367 To suppress this warning use the @samp{unused} attribute
2368 (@pxref{Variable Attributes}).
2370 @item -Wunused-variable
2371 @opindex Wunused-variable
2372 Warn whenever a local variable or non-constant static variable is unused
2373 aside from its declaration
2375 To suppress this warning use the @samp{unused} attribute
2376 (@pxref{Variable Attributes}).
2378 @item -Wunused-value
2379 @opindex Wunused-value
2380 Warn whenever a statement computes a result that is explicitly not used.
2382 To suppress this warning cast the expression to @samp{void}.
2386 All the above @option{-Wunused} options combined.
2388 In order to get a warning about an unused function parameter, you must
2389 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
2390 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
2392 @item -Wuninitialized
2393 @opindex Wuninitialized
2394 Warn if an automatic variable is used without first being initialized or
2395 if a variable may be clobbered by a @code{setjmp} call.
2397 These warnings are possible only in optimizing compilation,
2398 because they require data flow information that is computed only
2399 when optimizing. If you don't specify @option{-O}, you simply won't
2402 If you want to warn about code which uses the uninitialized value of the
2403 variable in its own initializer, use the @option{-Winit-self} option.
2405 These warnings occur only for variables that are candidates for
2406 register allocation. Therefore, they do not occur for a variable that
2407 is declared @code{volatile}, or whose address is taken, or whose size
2408 is other than 1, 2, 4 or 8 bytes. Also, they do not occur for
2409 structures, unions or arrays, even when they are in registers.
2411 Note that there may be no warning about a variable that is used only
2412 to compute a value that itself is never used, because such
2413 computations may be deleted by data flow analysis before the warnings
2416 These warnings are made optional because GCC is not smart
2417 enough to see all the reasons why the code might be correct
2418 despite appearing to have an error. Here is one example of how
2439 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
2440 always initialized, but GCC doesn't know this. Here is
2441 another common case:
2446 if (change_y) save_y = y, y = new_y;
2448 if (change_y) y = save_y;
2453 This has no bug because @code{save_y} is used only if it is set.
2455 @cindex @code{longjmp} warnings
2456 This option also warns when a non-volatile automatic variable might be
2457 changed by a call to @code{longjmp}. These warnings as well are possible
2458 only in optimizing compilation.
2460 The compiler sees only the calls to @code{setjmp}. It cannot know
2461 where @code{longjmp} will be called; in fact, a signal handler could
2462 call it at any point in the code. As a result, you may get a warning
2463 even when there is in fact no problem because @code{longjmp} cannot
2464 in fact be called at the place which would cause a problem.
2466 Some spurious warnings can be avoided if you declare all the functions
2467 you use that never return as @code{noreturn}. @xref{Function
2470 @item -Wunknown-pragmas
2471 @opindex Wunknown-pragmas
2472 @cindex warning for unknown pragmas
2473 @cindex unknown pragmas, warning
2474 @cindex pragmas, warning of unknown
2475 Warn when a #pragma directive is encountered which is not understood by
2476 GCC@. If this command line option is used, warnings will even be issued
2477 for unknown pragmas in system header files. This is not the case if
2478 the warnings were only enabled by the @option{-Wall} command line option.
2480 @item -Wstrict-aliasing
2481 @opindex Wstrict-aliasing
2482 This option is only active when @option{-fstrict-aliasing} is active.
2483 It warns about code which might break the strict aliasing rules that the
2484 compiler is using for optimization. The warning does not catch all
2485 cases, but does attempt to catch the more common pitfalls. It is
2486 included in @option{-Wall}.
2490 All of the above @samp{-W} options combined. This enables all the
2491 warnings about constructions that some users consider questionable, and
2492 that are easy to avoid (or modify to prevent the warning), even in
2493 conjunction with macros. This also enables some language-specific
2494 warnings described in @ref{C++ Dialect Options} and
2495 @ref{Objective-C Dialect Options}.
2498 The following @option{-W@dots{}} options are not implied by @option{-Wall}.
2499 Some of them warn about constructions that users generally do not
2500 consider questionable, but which occasionally you might wish to check
2501 for; others warn about constructions that are necessary or hard to avoid
2502 in some cases, and there is no simple way to modify the code to suppress
2509 (This option used to be called @option{-W}. The older name is still
2510 supported, but the newer name is more descriptive.) Print extra warning
2511 messages for these events:
2515 A function can return either with or without a value. (Falling
2516 off the end of the function body is considered returning without
2517 a value.) For example, this function would evoke such a
2531 An expression-statement or the left-hand side of a comma expression
2532 contains no side effects.
2533 To suppress the warning, cast the unused expression to void.
2534 For example, an expression such as @samp{x[i,j]} will cause a warning,
2535 but @samp{x[(void)i,j]} will not.
2538 An unsigned value is compared against zero with @samp{<} or @samp{>=}.
2541 A comparison like @samp{x<=y<=z} appears; this is equivalent to
2542 @samp{(x<=y ? 1 : 0) <= z}, which is a different interpretation from
2543 that of ordinary mathematical notation.
2546 Storage-class specifiers like @code{static} are not the first things in
2547 a declaration. According to the C Standard, this usage is obsolescent.
2550 The return type of a function has a type qualifier such as @code{const}.
2551 Such a type qualifier has no effect, since the value returned by a
2552 function is not an lvalue. (But don't warn about the GNU extension of
2553 @code{volatile void} return types. That extension will be warned about
2554 if @option{-pedantic} is specified.)
2557 If @option{-Wall} or @option{-Wunused} is also specified, warn about unused
2561 A comparison between signed and unsigned values could produce an
2562 incorrect result when the signed value is converted to unsigned.
2563 (But don't warn if @option{-Wno-sign-compare} is also specified.)
2566 An aggregate has an initializer which does not initialize all members.
2567 For example, the following code would cause such a warning, because
2568 @code{x.h} would be implicitly initialized to zero:
2571 struct s @{ int f, g, h; @};
2572 struct s x = @{ 3, 4 @};
2576 A function parameter is declared without a type specifier in K&R-style
2584 An empty body occurs in an @samp{if} or @samp{else} statement.
2587 A pointer is compared against integer zero with @samp{<}, @samp{<=},
2588 @samp{>}, or @samp{>=}.
2591 A variable might be changed by @samp{longjmp} or @samp{vfork}.
2594 Any of several floating-point events that often indicate errors, such as
2595 overflow, underflow, loss of precision, etc.
2597 @item @r{(C++ only)}
2598 An enumerator and a non-enumerator both appear in a conditional expression.
2600 @item @r{(C++ only)}
2601 A non-static reference or non-static @samp{const} member appears in a
2602 class without constructors.
2604 @item @r{(C++ only)}
2605 Ambiguous virtual bases.
2607 @item @r{(C++ only)}
2608 Subscripting an array which has been declared @samp{register}.
2610 @item @r{(C++ only)}
2611 Taking the address of a variable which has been declared @samp{register}.
2613 @item @r{(C++ only)}
2614 A base class is not initialized in a derived class' copy constructor.
2617 @item -Wno-div-by-zero
2618 @opindex Wno-div-by-zero
2619 @opindex Wdiv-by-zero
2620 Do not warn about compile-time integer division by zero. Floating point
2621 division by zero is not warned about, as it can be a legitimate way of
2622 obtaining infinities and NaNs.
2624 @item -Wsystem-headers
2625 @opindex Wsystem-headers
2626 @cindex warnings from system headers
2627 @cindex system headers, warnings from
2628 Print warning messages for constructs found in system header files.
2629 Warnings from system headers are normally suppressed, on the assumption
2630 that they usually do not indicate real problems and would only make the
2631 compiler output harder to read. Using this command line option tells
2632 GCC to emit warnings from system headers as if they occurred in user
2633 code. However, note that using @option{-Wall} in conjunction with this
2634 option will @emph{not} warn about unknown pragmas in system
2635 headers---for that, @option{-Wunknown-pragmas} must also be used.
2638 @opindex Wfloat-equal
2639 Warn if floating point values are used in equality comparisons.
2641 The idea behind this is that sometimes it is convenient (for the
2642 programmer) to consider floating-point values as approximations to
2643 infinitely precise real numbers. If you are doing this, then you need
2644 to compute (by analyzing the code, or in some other way) the maximum or
2645 likely maximum error that the computation introduces, and allow for it
2646 when performing comparisons (and when producing output, but that's a
2647 different problem). In particular, instead of testing for equality, you
2648 would check to see whether the two values have ranges that overlap; and
2649 this is done with the relational operators, so equality comparisons are
2652 @item -Wtraditional @r{(C only)}
2653 @opindex Wtraditional
2654 Warn about certain constructs that behave differently in traditional and
2655 ISO C@. Also warn about ISO C constructs that have no traditional C
2656 equivalent, and/or problematic constructs which should be avoided.
2660 Macro parameters that appear within string literals in the macro body.
2661 In traditional C macro replacement takes place within string literals,
2662 but does not in ISO C@.
2665 In traditional C, some preprocessor directives did not exist.
2666 Traditional preprocessors would only consider a line to be a directive
2667 if the @samp{#} appeared in column 1 on the line. Therefore
2668 @option{-Wtraditional} warns about directives that traditional C
2669 understands but would ignore because the @samp{#} does not appear as the
2670 first character on the line. It also suggests you hide directives like
2671 @samp{#pragma} not understood by traditional C by indenting them. Some
2672 traditional implementations would not recognize @samp{#elif}, so it
2673 suggests avoiding it altogether.
2676 A function-like macro that appears without arguments.
2679 The unary plus operator.
2682 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
2683 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
2684 constants.) Note, these suffixes appear in macros defined in the system
2685 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
2686 Use of these macros in user code might normally lead to spurious
2687 warnings, however gcc's integrated preprocessor has enough context to
2688 avoid warning in these cases.
2691 A function declared external in one block and then used after the end of
2695 A @code{switch} statement has an operand of type @code{long}.
2698 A non-@code{static} function declaration follows a @code{static} one.
2699 This construct is not accepted by some traditional C compilers.
2702 The ISO type of an integer constant has a different width or
2703 signedness from its traditional type. This warning is only issued if
2704 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
2705 typically represent bit patterns, are not warned about.
2708 Usage of ISO string concatenation is detected.
2711 Initialization of automatic aggregates.
2714 Identifier conflicts with labels. Traditional C lacks a separate
2715 namespace for labels.
2718 Initialization of unions. If the initializer is zero, the warning is
2719 omitted. This is done under the assumption that the zero initializer in
2720 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
2721 initializer warnings and relies on default initialization to zero in the
2725 Conversions by prototypes between fixed/floating point values and vice
2726 versa. The absence of these prototypes when compiling with traditional
2727 C would cause serious problems. This is a subset of the possible
2728 conversion warnings, for the full set use @option{-Wconversion}.
2731 Use of ISO C style function definitions. This warning intentionally is
2732 @emph{not} issued for prototype declarations or variadic functions
2733 because these ISO C features will appear in your code when using
2734 libiberty's traditional C compatibility macros, @code{PARAMS} and
2735 @code{VPARAMS}. This warning is also bypassed for nested functions
2736 because that feature is already a gcc extension and thus not relevant to
2737 traditional C compatibility.
2740 @item -Wdeclaration-after-statement @r{(C only)}
2741 @opindex Wdeclaration-after-statement
2742 Warn when a declaration is found after a statement in a block. This
2743 construct, known from C++, was introduced with ISO C99 and is by default
2744 allowed in GCC@. It is not supported by ISO C90 and was not supported by
2745 GCC versions before GCC 3.0. @xref{Mixed Declarations}.
2749 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
2751 @item -Wendif-labels
2752 @opindex Wendif-labels
2753 Warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
2757 Warn whenever a local variable shadows another local variable, parameter or
2758 global variable or whenever a built-in function is shadowed.
2760 @item -Wlarger-than-@var{len}
2761 @opindex Wlarger-than
2762 Warn whenever an object of larger than @var{len} bytes is defined.
2764 @item -Wpointer-arith
2765 @opindex Wpointer-arith
2766 Warn about anything that depends on the ``size of'' a function type or
2767 of @code{void}. GNU C assigns these types a size of 1, for
2768 convenience in calculations with @code{void *} pointers and pointers
2771 @item -Wbad-function-cast @r{(C only)}
2772 @opindex Wbad-function-cast
2773 Warn whenever a function call is cast to a non-matching type.
2774 For example, warn if @code{int malloc()} is cast to @code{anything *}.
2778 Warn whenever a pointer is cast so as to remove a type qualifier from
2779 the target type. For example, warn if a @code{const char *} is cast
2780 to an ordinary @code{char *}.
2783 @opindex Wcast-align
2784 Warn whenever a pointer is cast such that the required alignment of the
2785 target is increased. For example, warn if a @code{char *} is cast to
2786 an @code{int *} on machines where integers can only be accessed at
2787 two- or four-byte boundaries.
2789 @item -Wwrite-strings
2790 @opindex Wwrite-strings
2791 When compiling C, give string constants the type @code{const
2792 char[@var{length}]} so that
2793 copying the address of one into a non-@code{const} @code{char *}
2794 pointer will get a warning; when compiling C++, warn about the
2795 deprecated conversion from string constants to @code{char *}.
2796 These warnings will help you find at
2797 compile time code that can try to write into a string constant, but
2798 only if you have been very careful about using @code{const} in
2799 declarations and prototypes. Otherwise, it will just be a nuisance;
2800 this is why we did not make @option{-Wall} request these warnings.
2803 @opindex Wconversion
2804 Warn if a prototype causes a type conversion that is different from what
2805 would happen to the same argument in the absence of a prototype. This
2806 includes conversions of fixed point to floating and vice versa, and
2807 conversions changing the width or signedness of a fixed point argument
2808 except when the same as the default promotion.
2810 Also, warn if a negative integer constant expression is implicitly
2811 converted to an unsigned type. For example, warn about the assignment
2812 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
2813 casts like @code{(unsigned) -1}.
2815 @item -Wsign-compare
2816 @opindex Wsign-compare
2817 @cindex warning for comparison of signed and unsigned values
2818 @cindex comparison of signed and unsigned values, warning
2819 @cindex signed and unsigned values, comparison warning
2820 Warn when a comparison between signed and unsigned values could produce
2821 an incorrect result when the signed value is converted to unsigned.
2822 This warning is also enabled by @option{-Wextra}; to get the other warnings
2823 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
2825 @item -Waggregate-return
2826 @opindex Waggregate-return
2827 Warn if any functions that return structures or unions are defined or
2828 called. (In languages where you can return an array, this also elicits
2831 @item -Wstrict-prototypes @r{(C only)}
2832 @opindex Wstrict-prototypes
2833 Warn if a function is declared or defined without specifying the
2834 argument types. (An old-style function definition is permitted without
2835 a warning if preceded by a declaration which specifies the argument
2838 @item -Wold-style-definition @r{(C only)}
2839 @opindex Wold-style-definition
2840 Warn if an old-style function definition is used. A warning is given
2841 even if there is a previous prototype.
2843 @item -Wmissing-prototypes @r{(C only)}
2844 @opindex Wmissing-prototypes
2845 Warn if a global function is defined without a previous prototype
2846 declaration. This warning is issued even if the definition itself
2847 provides a prototype. The aim is to detect global functions that fail
2848 to be declared in header files.
2850 @item -Wmissing-declarations @r{(C only)}
2851 @opindex Wmissing-declarations
2852 Warn if a global function is defined without a previous declaration.
2853 Do so even if the definition itself provides a prototype.
2854 Use this option to detect global functions that are not declared in
2857 @item -Wmissing-noreturn
2858 @opindex Wmissing-noreturn
2859 Warn about functions which might be candidates for attribute @code{noreturn}.
2860 Note these are only possible candidates, not absolute ones. Care should
2861 be taken to manually verify functions actually do not ever return before
2862 adding the @code{noreturn} attribute, otherwise subtle code generation
2863 bugs could be introduced. You will not get a warning for @code{main} in
2864 hosted C environments.
2866 @item -Wmissing-format-attribute
2867 @opindex Wmissing-format-attribute
2869 If @option{-Wformat} is enabled, also warn about functions which might be
2870 candidates for @code{format} attributes. Note these are only possible
2871 candidates, not absolute ones. GCC will guess that @code{format}
2872 attributes might be appropriate for any function that calls a function
2873 like @code{vprintf} or @code{vscanf}, but this might not always be the
2874 case, and some functions for which @code{format} attributes are
2875 appropriate may not be detected. This option has no effect unless
2876 @option{-Wformat} is enabled (possibly by @option{-Wall}).
2878 @item -Wno-multichar
2879 @opindex Wno-multichar
2881 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
2882 Usually they indicate a typo in the user's code, as they have
2883 implementation-defined values, and should not be used in portable code.
2885 @item -Wno-deprecated-declarations
2886 @opindex Wno-deprecated-declarations
2887 Do not warn about uses of functions, variables, and types marked as
2888 deprecated by using the @code{deprecated} attribute.
2889 (@pxref{Function Attributes}, @pxref{Variable Attributes},
2890 @pxref{Type Attributes}.)
2894 Warn if a structure is given the packed attribute, but the packed
2895 attribute has no effect on the layout or size of the structure.
2896 Such structures may be mis-aligned for little benefit. For
2897 instance, in this code, the variable @code{f.x} in @code{struct bar}
2898 will be misaligned even though @code{struct bar} does not itself
2899 have the packed attribute:
2906 @} __attribute__((packed));
2916 Warn if padding is included in a structure, either to align an element
2917 of the structure or to align the whole structure. Sometimes when this
2918 happens it is possible to rearrange the fields of the structure to
2919 reduce the padding and so make the structure smaller.
2921 @item -Wredundant-decls
2922 @opindex Wredundant-decls
2923 Warn if anything is declared more than once in the same scope, even in
2924 cases where multiple declaration is valid and changes nothing.
2926 @item -Wnested-externs @r{(C only)}
2927 @opindex Wnested-externs
2928 Warn if an @code{extern} declaration is encountered within a function.
2930 @item -Wunreachable-code
2931 @opindex Wunreachable-code
2932 Warn if the compiler detects that code will never be executed.
2934 This option is intended to warn when the compiler detects that at
2935 least a whole line of source code will never be executed, because
2936 some condition is never satisfied or because it is after a
2937 procedure that never returns.
2939 It is possible for this option to produce a warning even though there
2940 are circumstances under which part of the affected line can be executed,
2941 so care should be taken when removing apparently-unreachable code.
2943 For instance, when a function is inlined, a warning may mean that the
2944 line is unreachable in only one inlined copy of the function.
2946 This option is not made part of @option{-Wall} because in a debugging
2947 version of a program there is often substantial code which checks
2948 correct functioning of the program and is, hopefully, unreachable
2949 because the program does work. Another common use of unreachable
2950 code is to provide behavior which is selectable at compile-time.
2954 Warn if a function can not be inlined and it was declared as inline.
2955 Even with this option, the compiler will not warn about failures to
2956 inline functions declared in system headers.
2958 The compiler uses a variety of heuristics to determine whether or not
2959 to inline a function. For example, the compiler takes into account
2960 the size of the function being inlined and the the amount of inlining
2961 that has already been done in the current function. Therefore,
2962 seemingly insignificant changes in the source program can cause the
2963 warnings produced by @option{-Winline} to appear or disappear.
2965 @item -Wno-invalid-offsetof @r{(C++ only)}
2966 @opindex Wno-invalid-offsetof
2967 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
2968 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
2969 to a non-POD type is undefined. In existing C++ implementations,
2970 however, @samp{offsetof} typically gives meaningful results even when
2971 applied to certain kinds of non-POD types. (Such as a simple
2972 @samp{struct} that fails to be a POD type only by virtue of having a
2973 constructor.) This flag is for users who are aware that they are
2974 writing nonportable code and who have deliberately chosen to ignore the
2977 The restrictions on @samp{offsetof} may be relaxed in a future version
2978 of the C++ standard.
2981 @opindex Winvalid-pch
2982 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
2983 the search path but can't be used.
2987 @opindex Wno-long-long
2988 Warn if @samp{long long} type is used. This is default. To inhibit
2989 the warning messages, use @option{-Wno-long-long}. Flags
2990 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
2991 only when @option{-pedantic} flag is used.
2993 @item -Wdisabled-optimization
2994 @opindex Wdisabled-optimization
2995 Warn if a requested optimization pass is disabled. This warning does
2996 not generally indicate that there is anything wrong with your code; it
2997 merely indicates that GCC's optimizers were unable to handle the code
2998 effectively. Often, the problem is that your code is too big or too
2999 complex; GCC will refuse to optimize programs when the optimization
3000 itself is likely to take inordinate amounts of time.
3004 Make all warnings into errors.
3007 @node Debugging Options
3008 @section Options for Debugging Your Program or GCC
3009 @cindex options, debugging
3010 @cindex debugging information options
3012 GCC has various special options that are used for debugging
3013 either your program or GCC:
3018 Produce debugging information in the operating system's native format
3019 (stabs, COFF, XCOFF, or DWARF)@. GDB can work with this debugging
3022 On most systems that use stabs format, @option{-g} enables use of extra
3023 debugging information that only GDB can use; this extra information
3024 makes debugging work better in GDB but will probably make other debuggers
3026 refuse to read the program. If you want to control for certain whether
3027 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
3028 @option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
3030 Unlike most other C compilers, GCC allows you to use @option{-g} with
3031 @option{-O}. The shortcuts taken by optimized code may occasionally
3032 produce surprising results: some variables you declared may not exist
3033 at all; flow of control may briefly move where you did not expect it;
3034 some statements may not be executed because they compute constant
3035 results or their values were already at hand; some statements may
3036 execute in different places because they were moved out of loops.
3038 Nevertheless it proves possible to debug optimized output. This makes
3039 it reasonable to use the optimizer for programs that might have bugs.
3041 The following options are useful when GCC is generated with the
3042 capability for more than one debugging format.
3046 Produce debugging information for use by GDB@. This means to use the
3047 most expressive format available (DWARF 2, stabs, or the native format
3048 if neither of those are supported), including GDB extensions if at all
3053 Produce debugging information in stabs format (if that is supported),
3054 without GDB extensions. This is the format used by DBX on most BSD
3055 systems. On MIPS, Alpha and System V Release 4 systems this option
3056 produces stabs debugging output which is not understood by DBX or SDB@.
3057 On System V Release 4 systems this option requires the GNU assembler.
3059 @item -feliminate-unused-debug-symbols
3060 @opindex feliminate-unused-debug-symbols
3061 Produce debugging information in stabs format (if that is supported),
3062 for only symbols that are actually used.
3066 Produce debugging information in stabs format (if that is supported),
3067 using GNU extensions understood only by the GNU debugger (GDB)@. The
3068 use of these extensions is likely to make other debuggers crash or
3069 refuse to read the program.
3073 Produce debugging information in COFF format (if that is supported).
3074 This is the format used by SDB on most System V systems prior to
3079 Produce debugging information in XCOFF format (if that is supported).
3080 This is the format used by the DBX debugger on IBM RS/6000 systems.
3084 Produce debugging information in XCOFF format (if that is supported),
3085 using GNU extensions understood only by the GNU debugger (GDB)@. The
3086 use of these extensions is likely to make other debuggers crash or
3087 refuse to read the program, and may cause assemblers other than the GNU
3088 assembler (GAS) to fail with an error.
3092 Produce debugging information in DWARF version 2 format (if that is
3093 supported). This is the format used by DBX on IRIX 6.
3097 Produce debugging information in VMS debug format (if that is
3098 supported). This is the format used by DEBUG on VMS systems.
3101 @itemx -ggdb@var{level}
3102 @itemx -gstabs@var{level}
3103 @itemx -gcoff@var{level}
3104 @itemx -gxcoff@var{level}
3105 @itemx -gvms@var{level}
3106 Request debugging information and also use @var{level} to specify how
3107 much information. The default level is 2.
3109 Level 1 produces minimal information, enough for making backtraces in
3110 parts of the program that you don't plan to debug. This includes
3111 descriptions of functions and external variables, but no information
3112 about local variables and no line numbers.
3114 Level 3 includes extra information, such as all the macro definitions
3115 present in the program. Some debuggers support macro expansion when
3116 you use @option{-g3}.
3118 Note that in order to avoid confusion between DWARF1 debug level 2,
3119 and DWARF2 @option{-gdwarf-2} does not accept a concatenated debug
3120 level. Instead use an additional @option{-g@var{level}} option to
3121 change the debug level for DWARF2.
3123 @item -feliminate-dwarf2-dups
3124 @opindex feliminate-dwarf2-dups
3125 Compress DWARF2 debugging information by eliminating duplicated
3126 information about each symbol. This option only makes sense when
3127 generating DWARF2 debugging information with @option{-gdwarf-2}.
3129 @cindex @command{prof}
3132 Generate extra code to write profile information suitable for the
3133 analysis program @command{prof}. You must use this option when compiling
3134 the source files you want data about, and you must also use it when
3137 @cindex @command{gprof}
3140 Generate extra code to write profile information suitable for the
3141 analysis program @command{gprof}. You must use this option when compiling
3142 the source files you want data about, and you must also use it when
3147 Makes the compiler print out each function name as it is compiled, and
3148 print some statistics about each pass when it finishes.
3151 @opindex ftime-report
3152 Makes the compiler print some statistics about the time consumed by each
3153 pass when it finishes.
3156 @opindex fmem-report
3157 Makes the compiler print some statistics about permanent memory
3158 allocation when it finishes.
3160 @item -fprofile-arcs
3161 @opindex fprofile-arcs
3162 Add code so that program flow @dfn{arcs} are instrumented. During
3163 execution the program records how many times each branch and call is
3164 executed and how many times it is taken or returns. When the compiled
3165 program exits it saves this data to a file called
3166 @file{@var{auxname}.gcda} for each source file. The data may be used for
3167 profile-directed optimizations (@option{-fbranch-probabilities}), or for
3168 test coverage analysis (@option{-ftest-coverage}). Each object file's
3169 @var{auxname} is generated from the name of the output file, if
3170 explicitly specified and it is not the final executable, otherwise it is
3171 the basename of the source file. In both cases any suffix is removed
3172 (e.g. @file{foo.gcda} for input file @file{dir/foo.c}, or
3173 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
3178 Compile the source files with @option{-fprofile-arcs} plus optimization
3179 and code generation options. For test coverage analysis, use the
3180 additional @option{-ftest-coverage} option. You do not need to profile
3181 every source file in a program.
3184 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
3185 (the latter implies the former).
3188 Run the program on a representative workload to generate the arc profile
3189 information. This may be repeated any number of times. You can run
3190 concurrent instances of your program, and provided that the file system
3191 supports locking, the data files will be correctly updated. Also
3192 @code{fork} calls are detected and correctly handled (double counting
3196 For profile-directed optimizations, compile the source files again with
3197 the same optimization and code generation options plus
3198 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
3199 Control Optimization}).
3202 For test coverage analysis, use @command{gcov} to produce human readable
3203 information from the @file{.gcno} and @file{.gcda} files. Refer to the
3204 @command{gcov} documentation for further information.
3208 With @option{-fprofile-arcs}, for each function of your program GCC
3209 creates a program flow graph, then finds a spanning tree for the graph.
3210 Only arcs that are not on the spanning tree have to be instrumented: the
3211 compiler adds code to count the number of times that these arcs are
3212 executed. When an arc is the only exit or only entrance to a block, the
3213 instrumentation code can be added to the block; otherwise, a new basic
3214 block must be created to hold the instrumentation code.
3217 @item -ftest-coverage
3218 @opindex ftest-coverage
3219 Produce a notes file that the @command{gcov} code-coverage utility
3220 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
3221 show program coverage. Each source file's note file is called
3222 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
3223 above for a description of @var{auxname} and instructions on how to
3224 generate test coverage data. Coverage data will match the source files
3225 more closely, if you do not optimize.
3227 @item -d@var{letters}
3229 Says to make debugging dumps during compilation at times specified by
3230 @var{letters}. This is used for debugging the compiler. The file names
3231 for most of the dumps are made by appending a pass number and a word to
3232 the @var{dumpname}. @var{dumpname} is generated from the name of the
3233 output file, if explicitly specified and it is not an executable,
3234 otherwise it is the basename of the source file. In both cases any
3235 suffix is removed (e.g. @file{foo.01.rtl} or @file{foo.02.sibling}).
3236 Here are the possible letters for use in @var{letters}, and their
3242 Annotate the assembler output with miscellaneous debugging information.
3245 Dump after computing branch probabilities, to @file{@var{file}.12.bp}.
3248 Dump after block reordering, to @file{@var{file}.30.bbro}.
3251 Dump after instruction combination, to the file @file{@var{file}.20.combine}.
3254 Dump after the first if conversion, to the file @file{@var{file}.14.ce1}.
3255 Also dump after the second if conversion, to the file @file{@var{file}.21.ce2}.
3258 Dump after branch target load optimization, to to @file{@var{file}.32.btl}.
3259 Also dump after delayed branch scheduling, to @file{@var{file}.36.dbr}.
3262 Dump all macro definitions, at the end of preprocessing, in addition to
3266 Dump after the second if conversion, to @file{@var{file}.31.ce3}.
3269 Dump after control and data flow analysis, to @file{@var{file}.11.cfg}.
3270 Also dump after life analysis, to @file{@var{file}.19.life}.
3273 Dump after purging @code{ADDRESSOF} codes, to @file{@var{file}.07.addressof}.
3276 Dump after global register allocation, to @file{@var{file}.25.greg}.
3279 Dump after GCSE, to @file{@var{file}.08.gcse}.
3280 Also dump after jump bypassing and control flow optimizations, to
3281 @file{@var{file}.10.bypass}.
3284 Dump after finalization of EH handling code, to @file{@var{file}.03.eh}.
3287 Dump after sibling call optimizations, to @file{@var{file}.02.sibling}.
3290 Dump after the first jump optimization, to @file{@var{file}.04.jump}.
3293 Dump after conversion from registers to stack, to @file{@var{file}.34.stack}.
3296 Dump after local register allocation, to @file{@var{file}.24.lreg}.
3299 Dump after loop optimization passes, to @file{@var{file}.09.loop} and
3300 @file{@var{file}.16.loop2}.
3303 Dump after performing the machine dependent reorganization pass, to
3304 @file{@var{file}.35.mach}.
3307 Dump after register renumbering, to @file{@var{file}.29.rnreg}.
3310 Dump after the register move pass, to @file{@var{file}.22.regmove}.
3313 Dump after post-reload optimizations, to @file{@var{file}.26.postreload}.
3316 Dump after RTL generation, to @file{@var{file}.01.rtl}.
3319 Dump after the second scheduling pass, to @file{@var{file}.33.sched2}.
3322 Dump after CSE (including the jump optimization that sometimes follows
3323 CSE), to @file{@var{file}.06.cse}.
3326 Dump after the first scheduling pass, to @file{@var{file}.23.sched}.
3329 Dump after the second CSE pass (including the jump optimization that
3330 sometimes follows CSE), to @file{@var{file}.18.cse2}.
3333 Dump after running tracer, to @file{@var{file}.15.tracer}.
3336 Dump after null pointer elimination pass to @file{@var{file}.05.null}.
3339 Dump callgraph and unit-at-a-time optimization @file{@var{file}.00.unit}.
3342 Dump after the value profile transformations, to @file{@var{file}.13.vpt}.
3345 Dump after the second flow pass, to @file{@var{file}.27.flow2}.
3348 Dump after the peephole pass, to @file{@var{file}.28.peephole2}.
3351 Dump after constructing the web, to @file{@var{file}.17.web}.
3354 Produce all the dumps listed above.
3357 Produce a core dump whenever an error occurs.
3360 Print statistics on memory usage, at the end of the run, to
3364 Annotate the assembler output with a comment indicating which
3365 pattern and alternative was used. The length of each instruction is
3369 Dump the RTL in the assembler output as a comment before each instruction.
3370 Also turns on @option{-dp} annotation.
3373 For each of the other indicated dump files (except for
3374 @file{@var{file}.01.rtl}), dump a representation of the control flow graph
3375 suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
3378 Just generate RTL for a function instead of compiling it. Usually used
3382 Dump debugging information during parsing, to standard error.
3385 @item -fdump-unnumbered
3386 @opindex fdump-unnumbered
3387 When doing debugging dumps (see @option{-d} option above), suppress instruction
3388 numbers and line number note output. This makes it more feasible to
3389 use diff on debugging dumps for compiler invocations with different
3390 options, in particular with and without @option{-g}.
3392 @item -fdump-translation-unit @r{(C and C++ only)}
3393 @itemx -fdump-translation-unit-@var{options} @r{(C and C++ only)}
3394 @opindex fdump-translation-unit
3395 Dump a representation of the tree structure for the entire translation
3396 unit to a file. The file name is made by appending @file{.tu} to the
3397 source file name. If the @samp{-@var{options}} form is used, @var{options}
3398 controls the details of the dump as described for the
3399 @option{-fdump-tree} options.
3401 @item -fdump-class-hierarchy @r{(C++ only)}
3402 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
3403 @opindex fdump-class-hierarchy
3404 Dump a representation of each class's hierarchy and virtual function
3405 table layout to a file. The file name is made by appending @file{.class}
3406 to the source file name. If the @samp{-@var{options}} form is used,
3407 @var{options} controls the details of the dump as described for the
3408 @option{-fdump-tree} options.
3410 @item -fdump-tree-@var{switch} @r{(C++ only)}
3411 @itemx -fdump-tree-@var{switch}-@var{options} @r{(C++ only)}
3413 Control the dumping at various stages of processing the intermediate
3414 language tree to a file. The file name is generated by appending a switch
3415 specific suffix to the source file name. If the @samp{-@var{options}}
3416 form is used, @var{options} is a list of @samp{-} separated options that
3417 control the details of the dump. Not all options are applicable to all
3418 dumps, those which are not meaningful will be ignored. The following
3419 options are available
3423 Print the address of each node. Usually this is not meaningful as it
3424 changes according to the environment and source file. Its primary use
3425 is for tying up a dump file with a debug environment.
3427 Inhibit dumping of members of a scope or body of a function merely
3428 because that scope has been reached. Only dump such items when they
3429 are directly reachable by some other path.
3431 Turn on all options.
3434 The following tree dumps are possible:
3437 Dump before any tree based optimization, to @file{@var{file}.original}.
3439 Dump after all tree based optimization, to @file{@var{file}.optimized}.
3441 Dump after function inlining, to @file{@var{file}.inlined}.
3444 @item -frandom-seed=@var{string}
3445 @opindex frandom-string
3446 This option provides a seed that GCC uses when it would otherwise use
3447 random numbers. It is used to generate certain symbol names
3448 that have to be different in every compiled file. It is also used to
3449 place unique stamps in coverage data files and the object files that
3450 produce them. You can use the @option{-frandom-seed} option to produce
3451 reproducibly identical object files.
3453 The @var{string} should be different for every file you compile.
3455 @item -fsched-verbose=@var{n}
3456 @opindex fsched-verbose
3457 On targets that use instruction scheduling, this option controls the
3458 amount of debugging output the scheduler prints. This information is
3459 written to standard error, unless @option{-dS} or @option{-dR} is
3460 specified, in which case it is output to the usual dump
3461 listing file, @file{.sched} or @file{.sched2} respectively. However
3462 for @var{n} greater than nine, the output is always printed to standard
3465 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
3466 same information as @option{-dRS}. For @var{n} greater than one, it
3467 also output basic block probabilities, detailed ready list information
3468 and unit/insn info. For @var{n} greater than two, it includes RTL
3469 at abort point, control-flow and regions info. And for @var{n} over
3470 four, @option{-fsched-verbose} also includes dependence info.
3474 Store the usual ``temporary'' intermediate files permanently; place them
3475 in the current directory and name them based on the source file. Thus,
3476 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
3477 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
3478 preprocessed @file{foo.i} output file even though the compiler now
3479 normally uses an integrated preprocessor.
3483 Report the CPU time taken by each subprocess in the compilation
3484 sequence. For C source files, this is the compiler proper and assembler
3485 (plus the linker if linking is done). The output looks like this:
3492 The first number on each line is the ``user time,'' that is time spent
3493 executing the program itself. The second number is ``system time,''
3494 time spent executing operating system routines on behalf of the program.
3495 Both numbers are in seconds.
3497 @item -print-file-name=@var{library}
3498 @opindex print-file-name
3499 Print the full absolute name of the library file @var{library} that
3500 would be used when linking---and don't do anything else. With this
3501 option, GCC does not compile or link anything; it just prints the
3504 @item -print-multi-directory
3505 @opindex print-multi-directory
3506 Print the directory name corresponding to the multilib selected by any
3507 other switches present in the command line. This directory is supposed
3508 to exist in @env{GCC_EXEC_PREFIX}.
3510 @item -print-multi-lib
3511 @opindex print-multi-lib
3512 Print the mapping from multilib directory names to compiler switches
3513 that enable them. The directory name is separated from the switches by
3514 @samp{;}, and each switch starts with an @samp{@@} instead of the
3515 @samp{-}, without spaces between multiple switches. This is supposed to
3516 ease shell-processing.
3518 @item -print-prog-name=@var{program}
3519 @opindex print-prog-name
3520 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
3522 @item -print-libgcc-file-name
3523 @opindex print-libgcc-file-name
3524 Same as @option{-print-file-name=libgcc.a}.
3526 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
3527 but you do want to link with @file{libgcc.a}. You can do
3530 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
3533 @item -print-search-dirs
3534 @opindex print-search-dirs
3535 Print the name of the configured installation directory and a list of
3536 program and library directories gcc will search---and don't do anything else.
3538 This is useful when gcc prints the error message
3539 @samp{installation problem, cannot exec cpp0: No such file or directory}.
3540 To resolve this you either need to put @file{cpp0} and the other compiler
3541 components where gcc expects to find them, or you can set the environment
3542 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
3543 Don't forget the trailing '/'.
3544 @xref{Environment Variables}.
3547 @opindex dumpmachine
3548 Print the compiler's target machine (for example,
3549 @samp{i686-pc-linux-gnu})---and don't do anything else.
3552 @opindex dumpversion
3553 Print the compiler version (for example, @samp{3.0})---and don't do
3558 Print the compiler's built-in specs---and don't do anything else. (This
3559 is used when GCC itself is being built.) @xref{Spec Files}.
3561 @item -feliminate-unused-debug-types
3562 @opindex feliminate-unused-debug-types
3563 Normally, when producing DWARF2 output, GCC will emit debugging
3564 information for all types declared in a compilation
3565 unit, regardless of whether or not they are actually used
3566 in that compilation unit. Sometimes this is useful, such as
3567 if, in the debugger, you want to cast a value to a type that is
3568 not actually used in your program (but is declared). More often,
3569 however, this results in a significant amount of wasted space.
3570 With this option, GCC will avoid producing debug symbol output
3571 for types that are nowhere used in the source file being compiled.
3574 @node Optimize Options
3575 @section Options That Control Optimization
3576 @cindex optimize options
3577 @cindex options, optimization
3579 These options control various sorts of optimizations.
3581 Without any optimization option, the compiler's goal is to reduce the
3582 cost of compilation and to make debugging produce the expected
3583 results. Statements are independent: if you stop the program with a
3584 breakpoint between statements, you can then assign a new value to any
3585 variable or change the program counter to any other statement in the
3586 function and get exactly the results you would expect from the source
3589 Turning on optimization flags makes the compiler attempt to improve
3590 the performance and/or code size at the expense of compilation time
3591 and possibly the ability to debug the program.
3593 The compiler performs optimisation based on the knowledge it has of
3594 the program. Using the @option{-funit-at-a-time} flag will allow the
3595 compiler to consider information gained from later functions in the
3596 file when compiling a function. Compiling multiple files at once to a
3597 single output file (and using @option{-funit-at-a-time}) will allow
3598 the compiler to use information gained from all of the files when
3599 compiling each of them.
3601 Not all optimizations are controlled directly by a flag. Only
3602 optimizations that have a flag are listed.
3609 Optimize. Optimizing compilation takes somewhat more time, and a lot
3610 more memory for a large function.
3612 With @option{-O}, the compiler tries to reduce code size and execution
3613 time, without performing any optimizations that take a great deal of
3616 @option{-O} turns on the following optimization flags:
3617 @gccoptlist{-fdefer-pop @gol
3618 -fmerge-constants @gol
3620 -floop-optimize @gol
3622 -fif-conversion @gol
3623 -fif-conversion2 @gol
3624 -fdelayed-branch @gol
3625 -fguess-branch-probability @gol
3628 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
3629 where doing so does not interfere with debugging.
3633 Optimize even more. GCC performs nearly all supported optimizations
3634 that do not involve a space-speed tradeoff. The compiler does not
3635 perform loop unrolling or function inlining when you specify @option{-O2}.
3636 As compared to @option{-O}, this option increases both compilation time
3637 and the performance of the generated code.
3639 @option{-O2} turns on all optimization flags specified by @option{-O}. It
3640 also turns on the following optimization flags:
3641 @gccoptlist{-fforce-mem @gol
3642 -foptimize-sibling-calls @gol
3643 -fstrength-reduce @gol
3644 -fcse-follow-jumps -fcse-skip-blocks @gol
3645 -frerun-cse-after-loop -frerun-loop-opt @gol
3646 -fgcse -fgcse-lm -fgcse-sm -fgcse-las @gol
3647 -fdelete-null-pointer-checks @gol
3648 -fexpensive-optimizations @gol
3650 -fschedule-insns -fschedule-insns2 @gol
3651 -fsched-interblock -fsched-spec @gol
3654 -freorder-blocks -freorder-functions @gol
3655 -fstrict-aliasing @gol
3656 -funit-at-a-time @gol
3657 -falign-functions -falign-jumps @gol
3658 -falign-loops -falign-labels}
3660 Please note the warning under @option{-fgcse} about
3661 invoking @option{-O2} on programs that use computed gotos.
3665 Optimize yet more. @option{-O3} turns on all optimizations specified by
3666 @option{-O2} and also turns on the @option{-finline-functions},
3667 @option{-fweb} and @option{-frename-registers} options.
3671 Do not optimize. This is the default.
3675 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
3676 do not typically increase code size. It also performs further
3677 optimizations designed to reduce code size.
3679 @option{-Os} disables the following optimization flags:
3680 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
3681 -falign-labels -freorder-blocks -fprefetch-loop-arrays}
3683 If you use multiple @option{-O} options, with or without level numbers,
3684 the last such option is the one that is effective.
3687 Options of the form @option{-f@var{flag}} specify machine-independent
3688 flags. Most flags have both positive and negative forms; the negative
3689 form of @option{-ffoo} would be @option{-fno-foo}. In the table
3690 below, only one of the forms is listed---the one you typically will
3691 use. You can figure out the other form by either removing @samp{no-}
3694 The following options control specific optimizations. They are either
3695 activated by @option{-O} options or are related to ones that are. You
3696 can use the following flags in the rare cases when ``fine-tuning'' of
3697 optimizations to be performed is desired.
3700 @item -fno-default-inline
3701 @opindex fno-default-inline
3702 Do not make member functions inline by default merely because they are
3703 defined inside the class scope (C++ only). Otherwise, when you specify
3704 @w{@option{-O}}, member functions defined inside class scope are compiled
3705 inline by default; i.e., you don't need to add @samp{inline} in front of
3706 the member function name.
3708 @item -fno-defer-pop
3709 @opindex fno-defer-pop
3710 Always pop the arguments to each function call as soon as that function
3711 returns. For machines which must pop arguments after a function call,
3712 the compiler normally lets arguments accumulate on the stack for several
3713 function calls and pops them all at once.
3715 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3719 Force memory operands to be copied into registers before doing
3720 arithmetic on them. This produces better code by making all memory
3721 references potential common subexpressions. When they are not common
3722 subexpressions, instruction combination should eliminate the separate
3725 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3728 @opindex fforce-addr
3729 Force memory address constants to be copied into registers before
3730 doing arithmetic on them. This may produce better code just as
3731 @option{-fforce-mem} may.
3733 @item -fomit-frame-pointer
3734 @opindex fomit-frame-pointer
3735 Don't keep the frame pointer in a register for functions that
3736 don't need one. This avoids the instructions to save, set up and
3737 restore frame pointers; it also makes an extra register available
3738 in many functions. @strong{It also makes debugging impossible on
3741 On some machines, such as the VAX, this flag has no effect, because
3742 the standard calling sequence automatically handles the frame pointer
3743 and nothing is saved by pretending it doesn't exist. The
3744 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
3745 whether a target machine supports this flag. @xref{Registers,,Register
3746 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
3748 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3750 @item -foptimize-sibling-calls
3751 @opindex foptimize-sibling-calls
3752 Optimize sibling and tail recursive calls.
3754 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3758 Don't pay attention to the @code{inline} keyword. Normally this option
3759 is used to keep the compiler from expanding any functions inline.
3760 Note that if you are not optimizing, no functions can be expanded inline.
3762 @item -finline-functions
3763 @opindex finline-functions
3764 Integrate all simple functions into their callers. The compiler
3765 heuristically decides which functions are simple enough to be worth
3766 integrating in this way.
3768 If all calls to a given function are integrated, and the function is
3769 declared @code{static}, then the function is normally not output as
3770 assembler code in its own right.
3772 Enabled at level @option{-O3}.
3774 @item -finline-limit=@var{n}
3775 @opindex finline-limit
3776 By default, gcc limits the size of functions that can be inlined. This flag
3777 allows the control of this limit for functions that are explicitly marked as
3778 inline (i.e., marked with the inline keyword or defined within the class
3779 definition in c++). @var{n} is the size of functions that can be inlined in
3780 number of pseudo instructions (not counting parameter handling). The default
3781 value of @var{n} is 600.
3782 Increasing this value can result in more inlined code at
3783 the cost of compilation time and memory consumption. Decreasing usually makes
3784 the compilation faster and less code will be inlined (which presumably
3785 means slower programs). This option is particularly useful for programs that
3786 use inlining heavily such as those based on recursive templates with C++.
3788 Inlining is actually controlled by a number of parameters, which may be
3789 specified individually by using @option{--param @var{name}=@var{value}}.
3790 The @option{-finline-limit=@var{n}} option sets some of these parameters
3794 @item max-inline-insns-single
3795 is set to @var{n}/2.
3796 @item max-inline-insns-auto
3797 is set to @var{n}/2.
3798 @item min-inline-insns
3799 is set to 130 or @var{n}/4, whichever is smaller.
3800 @item max-inline-insns-rtl
3804 See below for a documentation of the individual
3805 parameters controlling inlining.
3807 @emph{Note:} pseudo instruction represents, in this particular context, an
3808 abstract measurement of function's size. In no way, it represents a count
3809 of assembly instructions and as such its exact meaning might change from one
3810 release to an another.
3812 @item -fkeep-inline-functions
3813 @opindex fkeep-inline-functions
3814 Even if all calls to a given function are integrated, and the function
3815 is declared @code{static}, nevertheless output a separate run-time
3816 callable version of the function. This switch does not affect
3817 @code{extern inline} functions.
3819 @item -fkeep-static-consts
3820 @opindex fkeep-static-consts
3821 Emit variables declared @code{static const} when optimization isn't turned
3822 on, even if the variables aren't referenced.
3824 GCC enables this option by default. If you want to force the compiler to
3825 check if the variable was referenced, regardless of whether or not
3826 optimization is turned on, use the @option{-fno-keep-static-consts} option.
3828 @item -fmerge-constants
3829 Attempt to merge identical constants (string constants and floating point
3830 constants) across compilation units.
3832 This option is the default for optimized compilation if the assembler and
3833 linker support it. Use @option{-fno-merge-constants} to inhibit this
3836 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3838 @item -fmerge-all-constants
3839 Attempt to merge identical constants and identical variables.
3841 This option implies @option{-fmerge-constants}. In addition to
3842 @option{-fmerge-constants} this considers e.g. even constant initialized
3843 arrays or initialized constant variables with integral or floating point
3844 types. Languages like C or C++ require each non-automatic variable to
3845 have distinct location, so using this option will result in non-conforming
3850 Use a graph coloring register allocator. Currently this option is meant
3851 for testing, so we are interested to hear about miscompilations with
3854 @item -fno-branch-count-reg
3855 @opindex fno-branch-count-reg
3856 Do not use ``decrement and branch'' instructions on a count register,
3857 but instead generate a sequence of instructions that decrement a
3858 register, compare it against zero, then branch based upon the result.
3859 This option is only meaningful on architectures that support such
3860 instructions, which include x86, PowerPC, IA-64 and S/390.
3862 The default is @option{-fbranch-count-reg}, enabled when
3863 @option{-fstrength-reduce} is enabled.
3865 @item -fno-function-cse
3866 @opindex fno-function-cse
3867 Do not put function addresses in registers; make each instruction that
3868 calls a constant function contain the function's address explicitly.
3870 This option results in less efficient code, but some strange hacks
3871 that alter the assembler output may be confused by the optimizations
3872 performed when this option is not used.
3874 The default is @option{-ffunction-cse}
3876 @item -fno-zero-initialized-in-bss
3877 @opindex fno-zero-initialized-in-bss
3878 If the target supports a BSS section, GCC by default puts variables that
3879 are initialized to zero into BSS@. This can save space in the resulting
3882 This option turns off this behavior because some programs explicitly
3883 rely on variables going to the data section. E.g., so that the
3884 resulting executable can find the beginning of that section and/or make
3885 assumptions based on that.
3887 The default is @option{-fzero-initialized-in-bss}.
3889 @item -fstrength-reduce
3890 @opindex fstrength-reduce
3891 Perform the optimizations of loop strength reduction and
3892 elimination of iteration variables.
3894 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3896 @item -fthread-jumps
3897 @opindex fthread-jumps
3898 Perform optimizations where we check to see if a jump branches to a
3899 location where another comparison subsumed by the first is found. If
3900 so, the first branch is redirected to either the destination of the
3901 second branch or a point immediately following it, depending on whether
3902 the condition is known to be true or false.
3904 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3906 @item -fcse-follow-jumps
3907 @opindex fcse-follow-jumps
3908 In common subexpression elimination, scan through jump instructions
3909 when the target of the jump is not reached by any other path. For
3910 example, when CSE encounters an @code{if} statement with an
3911 @code{else} clause, CSE will follow the jump when the condition
3914 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3916 @item -fcse-skip-blocks
3917 @opindex fcse-skip-blocks
3918 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
3919 follow jumps which conditionally skip over blocks. When CSE
3920 encounters a simple @code{if} statement with no else clause,
3921 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
3922 body of the @code{if}.
3924 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3926 @item -frerun-cse-after-loop
3927 @opindex frerun-cse-after-loop
3928 Re-run common subexpression elimination after loop optimizations has been
3931 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3933 @item -frerun-loop-opt
3934 @opindex frerun-loop-opt
3935 Run the loop optimizer twice.
3937 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3941 Perform a global common subexpression elimination pass.
3942 This pass also performs global constant and copy propagation.
3944 @emph{Note:} When compiling a program using computed gotos, a GCC
3945 extension, you may get better runtime performance if you disable
3946 the global common subexpression elimination pass by adding
3947 @option{-fno-gcse} to the command line.
3949 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3953 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
3954 attempt to move loads which are only killed by stores into themselves. This
3955 allows a loop containing a load/store sequence to be changed to a load outside
3956 the loop, and a copy/store within the loop.
3958 Enabled by default when gcse is enabled.
3962 When @option{-fgcse-sm} is enabled, a store motion pass is run after
3963 global common subexpression elimination. This pass will attempt to move
3964 stores out of loops. When used in conjunction with @option{-fgcse-lm},
3965 loops containing a load/store sequence can be changed to a load before
3966 the loop and a store after the loop.
3968 Enabled by default when gcse is enabled.
3972 When @option{-fgcse-las} is enabled, the global common subexpression
3973 elimination pass eliminates redundant loads that come after stores to the
3974 same memory location (both partial and full redundancies).
3976 Enabled by default when gcse is enabled.
3978 @item -floop-optimize
3979 @opindex floop-optimize
3980 Perform loop optimizations: move constant expressions out of loops, simplify
3981 exit test conditions and optionally do strength-reduction and loop unrolling as
3984 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3986 @item -fcrossjumping
3987 @opindex crossjumping
3988 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
3989 resulting code may or may not perform better than without cross-jumping.
3991 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3993 @item -fif-conversion
3994 @opindex if-conversion
3995 Attempt to transform conditional jumps into branch-less equivalents. This
3996 include use of conditional moves, min, max, set flags and abs instructions, and
3997 some tricks doable by standard arithmetics. The use of conditional execution
3998 on chips where it is available is controlled by @code{if-conversion2}.
4000 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4002 @item -fif-conversion2
4003 @opindex if-conversion2
4004 Use conditional execution (where available) to transform conditional jumps into
4005 branch-less equivalents.
4007 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4009 @item -fdelete-null-pointer-checks
4010 @opindex fdelete-null-pointer-checks
4011 Use global dataflow analysis to identify and eliminate useless checks
4012 for null pointers. The compiler assumes that dereferencing a null
4013 pointer would have halted the program. If a pointer is checked after
4014 it has already been dereferenced, it cannot be null.
4016 In some environments, this assumption is not true, and programs can
4017 safely dereference null pointers. Use
4018 @option{-fno-delete-null-pointer-checks} to disable this optimization
4019 for programs which depend on that behavior.
4021 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4023 @item -fexpensive-optimizations
4024 @opindex fexpensive-optimizations
4025 Perform a number of minor optimizations that are relatively expensive.
4027 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4029 @item -foptimize-register-move
4031 @opindex foptimize-register-move
4033 Attempt to reassign register numbers in move instructions and as
4034 operands of other simple instructions in order to maximize the amount of
4035 register tying. This is especially helpful on machines with two-operand
4038 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
4041 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4043 @item -fdelayed-branch
4044 @opindex fdelayed-branch
4045 If supported for the target machine, attempt to reorder instructions
4046 to exploit instruction slots available after delayed branch
4049 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4051 @item -fschedule-insns
4052 @opindex fschedule-insns
4053 If supported for the target machine, attempt to reorder instructions to
4054 eliminate execution stalls due to required data being unavailable. This
4055 helps machines that have slow floating point or memory load instructions
4056 by allowing other instructions to be issued until the result of the load
4057 or floating point instruction is required.
4059 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4061 @item -fschedule-insns2
4062 @opindex fschedule-insns2
4063 Similar to @option{-fschedule-insns}, but requests an additional pass of
4064 instruction scheduling after register allocation has been done. This is
4065 especially useful on machines with a relatively small number of
4066 registers and where memory load instructions take more than one cycle.
4068 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4070 @item -fno-sched-interblock
4071 @opindex fno-sched-interblock
4072 Don't schedule instructions across basic blocks. This is normally
4073 enabled by default when scheduling before register allocation, i.e.@:
4074 with @option{-fschedule-insns} or at @option{-O2} or higher.
4076 @item -fno-sched-spec
4077 @opindex fno-sched-spec
4078 Don't allow speculative motion of non-load instructions. This is normally
4079 enabled by default when scheduling before register allocation, i.e.@:
4080 with @option{-fschedule-insns} or at @option{-O2} or higher.
4082 @item -fsched-spec-load
4083 @opindex fsched-spec-load
4084 Allow speculative motion of some load instructions. This only makes
4085 sense when scheduling before register allocation, i.e.@: with
4086 @option{-fschedule-insns} or at @option{-O2} or higher.
4088 @item -fsched-spec-load-dangerous
4089 @opindex fsched-spec-load-dangerous
4090 Allow speculative motion of more load instructions. This only makes
4091 sense when scheduling before register allocation, i.e.@: with
4092 @option{-fschedule-insns} or at @option{-O2} or higher.
4094 @item -fsched-stalled-insns=@var{n}
4095 @opindex fsched-stalled-insns
4096 Define how many insns (if any) can be moved prematurely from the queue
4097 of stalled insns into the ready list, during the second scheduling pass.
4099 @item -fsched-stalled-insns-dep=@var{n}
4100 @opindex fsched-stalled-insns-dep
4101 Define how many insn groups (cycles) will be examined for a dependency
4102 on a stalled insn that is candidate for premature removal from the queue
4103 of stalled insns. Has an effect only during the second scheduling pass,
4104 and only if @option{-fsched-stalled-insns} is used and its value is not zero.
4106 @item -fsched2-use-superblocks
4107 @opindex fsched2-use-superblocks
4108 When scheduling after register allocation, do use superblock scheduling
4109 algorithm. Superblock scheduling allows motion across basic block boundaries
4110 resulting on faster schedules. This option is experimental, as not all machine
4111 descriptions used by GCC model the CPU closely enough to avoid unreliable
4112 results from the algorithm.
4114 This only makes sense when scheduling after register allocation, i.e.@: with
4115 @option{-fschedule-insns2} or at @option{-O2} or higher.
4117 @item -fsched2-use-traces
4118 @opindex fsched2-use-traces
4119 Use @option{-fsched2-use-superblocks} algorithm when scheduling after register
4120 allocation and additionally perform code duplication in order to increase the
4121 size of superblocks using tracer pass. See @option{-ftracer} for details on
4124 This mode should produce faster but significantly longer programs. Also
4125 without @code{-fbranch-probabilities} the traces constructed may not match the
4126 reality and hurt the performance. This only makes
4127 sense when scheduling after register allocation, i.e.@: with
4128 @option{-fschedule-insns2} or at @option{-O2} or higher.
4130 @item -fcaller-saves
4131 @opindex fcaller-saves
4132 Enable values to be allocated in registers that will be clobbered by
4133 function calls, by emitting extra instructions to save and restore the
4134 registers around such calls. Such allocation is done only when it
4135 seems to result in better code than would otherwise be produced.
4137 This option is always enabled by default on certain machines, usually
4138 those which have no call-preserved registers to use instead.
4140 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4142 @item -fmove-all-movables
4143 @opindex fmove-all-movables
4144 Forces all invariant computations in loops to be moved
4147 @item -freduce-all-givs
4148 @opindex freduce-all-givs
4149 Forces all general-induction variables in loops to be
4152 @emph{Note:} When compiling programs written in Fortran,
4153 @option{-fmove-all-movables} and @option{-freduce-all-givs} are enabled
4154 by default when you use the optimizer.
4156 These options may generate better or worse code; results are highly
4157 dependent on the structure of loops within the source code.
4159 These two options are intended to be removed someday, once
4160 they have helped determine the efficacy of various
4161 approaches to improving loop optimizations.
4163 Please let us (@w{@email{gcc@@gcc.gnu.org}} and @w{@email{fortran@@gnu.org}})
4164 know how use of these options affects
4165 the performance of your production code.
4166 We're very interested in code that runs @emph{slower}
4167 when these options are @emph{enabled}.
4170 @itemx -fno-peephole2
4171 @opindex fno-peephole
4172 @opindex fno-peephole2
4173 Disable any machine-specific peephole optimizations. The difference
4174 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
4175 are implemented in the compiler; some targets use one, some use the
4176 other, a few use both.
4178 @option{-fpeephole} is enabled by default.
4179 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4181 @item -fno-guess-branch-probability
4182 @opindex fno-guess-branch-probability
4183 Do not guess branch probabilities using a randomized model.
4185 Sometimes gcc will opt to use a randomized model to guess branch
4186 probabilities, when none are available from either profiling feedback
4187 (@option{-fprofile-arcs}) or @samp{__builtin_expect}. This means that
4188 different runs of the compiler on the same program may produce different
4191 In a hard real-time system, people don't want different runs of the
4192 compiler to produce code that has different behavior; minimizing
4193 non-determinism is of paramount import. This switch allows users to
4194 reduce non-determinism, possibly at the expense of inferior
4197 The default is @option{-fguess-branch-probability} at levels
4198 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4200 @item -freorder-blocks
4201 @opindex freorder-blocks
4202 Reorder basic blocks in the compiled function in order to reduce number of
4203 taken branches and improve code locality.
4205 Enabled at levels @option{-O2}, @option{-O3}.
4207 @item -freorder-functions
4208 @opindex freorder-functions
4209 Reorder basic blocks in the compiled function in order to reduce number of
4210 taken branches and improve code locality. This is implemented by using special
4211 subsections @code{text.hot} for most frequently executed functions and
4212 @code{text.unlikely} for unlikely executed functions. Reordering is done by
4213 the linker so object file format must support named sections and linker must
4214 place them in a reasonable way.
4216 Also profile feedback must be available in to make this option effective. See
4217 @option{-fprofile-arcs} for details.
4219 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4221 @item -fstrict-aliasing
4222 @opindex fstrict-aliasing
4223 Allows the compiler to assume the strictest aliasing rules applicable to
4224 the language being compiled. For C (and C++), this activates
4225 optimizations based on the type of expressions. In particular, an
4226 object of one type is assumed never to reside at the same address as an
4227 object of a different type, unless the types are almost the same. For
4228 example, an @code{unsigned int} can alias an @code{int}, but not a
4229 @code{void*} or a @code{double}. A character type may alias any other
4232 Pay special attention to code like this:
4245 The practice of reading from a different union member than the one most
4246 recently written to (called ``type-punning'') is common. Even with
4247 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
4248 is accessed through the union type. So, the code above will work as
4249 expected. However, this code might not:
4260 Every language that wishes to perform language-specific alias analysis
4261 should define a function that computes, given an @code{tree}
4262 node, an alias set for the node. Nodes in different alias sets are not
4263 allowed to alias. For an example, see the C front-end function
4264 @code{c_get_alias_set}.
4266 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4268 @item -falign-functions
4269 @itemx -falign-functions=@var{n}
4270 @opindex falign-functions
4271 Align the start of functions to the next power-of-two greater than
4272 @var{n}, skipping up to @var{n} bytes. For instance,
4273 @option{-falign-functions=32} aligns functions to the next 32-byte
4274 boundary, but @option{-falign-functions=24} would align to the next
4275 32-byte boundary only if this can be done by skipping 23 bytes or less.
4277 @option{-fno-align-functions} and @option{-falign-functions=1} are
4278 equivalent and mean that functions will not be aligned.
4280 Some assemblers only support this flag when @var{n} is a power of two;
4281 in that case, it is rounded up.
4283 If @var{n} is not specified or is zero, use a machine-dependent default.
4285 Enabled at levels @option{-O2}, @option{-O3}.
4287 @item -falign-labels
4288 @itemx -falign-labels=@var{n}
4289 @opindex falign-labels
4290 Align all branch targets to a power-of-two boundary, skipping up to
4291 @var{n} bytes like @option{-falign-functions}. This option can easily
4292 make code slower, because it must insert dummy operations for when the
4293 branch target is reached in the usual flow of the code.
4295 @option{-fno-align-labels} and @option{-falign-labels=1} are
4296 equivalent and mean that labels will not be aligned.
4298 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
4299 are greater than this value, then their values are used instead.
4301 If @var{n} is not specified or is zero, use a machine-dependent default
4302 which is very likely to be @samp{1}, meaning no alignment.
4304 Enabled at levels @option{-O2}, @option{-O3}.
4307 @itemx -falign-loops=@var{n}
4308 @opindex falign-loops
4309 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
4310 like @option{-falign-functions}. The hope is that the loop will be
4311 executed many times, which will make up for any execution of the dummy
4314 @option{-fno-align-loops} and @option{-falign-loops=1} are
4315 equivalent and mean that loops will not be aligned.
4317 If @var{n} is not specified or is zero, use a machine-dependent default.
4319 Enabled at levels @option{-O2}, @option{-O3}.
4322 @itemx -falign-jumps=@var{n}
4323 @opindex falign-jumps
4324 Align branch targets to a power-of-two boundary, for branch targets
4325 where the targets can only be reached by jumping, skipping up to @var{n}
4326 bytes like @option{-falign-functions}. In this case, no dummy operations
4329 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
4330 equivalent and mean that loops will not be aligned.
4332 If @var{n} is not specified or is zero, use a machine-dependent default.
4334 Enabled at levels @option{-O2}, @option{-O3}.
4336 @item -frename-registers
4337 @opindex frename-registers
4338 Attempt to avoid false dependencies in scheduled code by making use
4339 of registers left over after register allocation. This optimization
4340 will most benefit processors with lots of registers. It can, however,
4341 make debugging impossible, since variables will no longer stay in
4342 a ``home register''.
4346 Constructs webs as commonly used for register allocation purposes and assign
4347 each web individual pseudo register. This allows our register allocation pass
4348 to operate on pseudos directly, but also strengthens several other optimization
4349 passes, such as CSE, loop optimizer and trivial dead code remover. It can,
4350 however, make debugging impossible, since variables will no longer stay in a
4353 Enabled at levels @option{-O3}.
4355 @item -fno-cprop-registers
4356 @opindex fno-cprop-registers
4357 After register allocation and post-register allocation instruction splitting,
4358 we perform a copy-propagation pass to try to reduce scheduling dependencies
4359 and occasionally eliminate the copy.
4361 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4363 @item -fprofile-generate
4364 @opindex fprofile-generate
4365 Enable options usually used for instrumenting application to produce profile usefull
4366 for later recompilation profile feedback based optimization.
4368 The following options are enabled: @code{-fprofile-arcs}, @code{-fprofile-values}, @code{-fvpt}
4371 @opindex fprofile-use
4372 Enable profile feedback directed optimizations, and optimizations
4373 generally profitable only with profile feedback available.
4375 The following options are enabled: @code{-fbranch-probabilities},
4376 @code{-fvpt}, @code{-funroll-loops}, @code{-fpeel-loops}, @code{-ftracer}.
4380 The following options control compiler behavior regarding floating
4381 point arithmetic. These options trade off between speed and
4382 correctness. All must be specifically enabled.
4386 @opindex ffloat-store
4387 Do not store floating point variables in registers, and inhibit other
4388 options that might change whether a floating point value is taken from a
4391 @cindex floating point precision
4392 This option prevents undesirable excess precision on machines such as
4393 the 68000 where the floating registers (of the 68881) keep more
4394 precision than a @code{double} is supposed to have. Similarly for the
4395 x86 architecture. For most programs, the excess precision does only
4396 good, but a few programs rely on the precise definition of IEEE floating
4397 point. Use @option{-ffloat-store} for such programs, after modifying
4398 them to store all pertinent intermediate computations into variables.
4402 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations}, @*
4403 @option{-fno-trapping-math}, @option{-ffinite-math-only},
4404 @option{-fno-rounding-math} and @option{-fno-signaling-nans}.
4406 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
4408 This option should never be turned on by any @option{-O} option since
4409 it can result in incorrect output for programs which depend on
4410 an exact implementation of IEEE or ISO rules/specifications for
4413 @item -fno-math-errno
4414 @opindex fno-math-errno
4415 Do not set ERRNO after calling math functions that are executed
4416 with a single instruction, e.g., sqrt. A program that relies on
4417 IEEE exceptions for math error handling may want to use this flag
4418 for speed while maintaining IEEE arithmetic compatibility.
4420 This option should never be turned on by any @option{-O} option since
4421 it can result in incorrect output for programs which depend on
4422 an exact implementation of IEEE or ISO rules/specifications for
4425 The default is @option{-fmath-errno}.
4427 @item -funsafe-math-optimizations
4428 @opindex funsafe-math-optimizations
4429 Allow optimizations for floating-point arithmetic that (a) assume
4430 that arguments and results are valid and (b) may violate IEEE or
4431 ANSI standards. When used at link-time, it may include libraries
4432 or startup files that change the default FPU control word or other
4433 similar optimizations.
4435 This option should never be turned on by any @option{-O} option since
4436 it can result in incorrect output for programs which depend on
4437 an exact implementation of IEEE or ISO rules/specifications for
4440 The default is @option{-fno-unsafe-math-optimizations}.
4442 @item -ffinite-math-only
4443 @opindex ffinite-math-only
4444 Allow optimizations for floating-point arithmetic that assume
4445 that arguments and results are not NaNs or +-Infs.
4447 This option should never be turned on by any @option{-O} option since
4448 it can result in incorrect output for programs which depend on
4449 an exact implementation of IEEE or ISO rules/specifications.
4451 The default is @option{-fno-finite-math-only}.
4453 @item -fno-trapping-math
4454 @opindex fno-trapping-math
4455 Compile code assuming that floating-point operations cannot generate
4456 user-visible traps. These traps include division by zero, overflow,
4457 underflow, inexact result and invalid operation. This option implies
4458 @option{-fno-signaling-nans}. Setting this option may allow faster
4459 code if one relies on ``non-stop'' IEEE arithmetic, for example.
4461 This option should never be turned on by any @option{-O} option since
4462 it can result in incorrect output for programs which depend on
4463 an exact implementation of IEEE or ISO rules/specifications for
4466 The default is @option{-ftrapping-math}.
4468 @item -frounding-math
4469 @opindex frounding-math
4470 Disable transformations and optimizations that assume default floating
4471 point rounding behavior. This is round-to-zero for all floating point
4472 to integer conversions, and round-to-nearest for all other arithmetic
4473 truncations. This option should be specified for programs that change
4474 the FP rounding mode dynamically, or that may be executed with a
4475 non-default rounding mode. This option disables constant folding of
4476 floating point expressions at compile-time (which may be affected by
4477 rounding mode) and arithmetic transformations that are unsafe in the
4478 presence of sign-dependent rounding modes.
4480 The default is @option{-fno-rounding-math}.
4482 This option is experimental and does not currently guarantee to
4483 disable all GCC optimizations that are affected by rounding mode.
4484 Future versions of gcc may provide finer control of this setting
4485 using C99's @code{FENV_ACCESS} pragma. This command line option
4486 will be used to specify the default state for @code{FENV_ACCESS}.
4488 @item -fsignaling-nans
4489 @opindex fsignaling-nans
4490 Compile code assuming that IEEE signaling NaNs may generate user-visible
4491 traps during floating-point operations. Setting this option disables
4492 optimizations that may change the number of exceptions visible with
4493 signaling NaNs. This option implies @option{-ftrapping-math}.
4495 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
4498 The default is @option{-fno-signaling-nans}.
4500 This option is experimental and does not currently guarantee to
4501 disable all GCC optimizations that affect signaling NaN behavior.
4503 @item -fsingle-precision-constant
4504 @opindex fsingle-precision-constant
4505 Treat floating point constant as single precision constant instead of
4506 implicitly converting it to double precision constant.
4511 The following options control optimizations that may improve
4512 performance, but are not enabled by any @option{-O} options. This
4513 section includes experimental options that may produce broken code.
4516 @item -fbranch-probabilities
4517 @opindex fbranch-probabilities
4518 After running a program compiled with @option{-fprofile-arcs}
4519 (@pxref{Debugging Options,, Options for Debugging Your Program or
4520 @command{gcc}}), you can compile it a second time using
4521 @option{-fbranch-probabilities}, to improve optimizations based on
4522 the number of times each branch was taken. When the program
4523 compiled with @option{-fprofile-arcs} exits it saves arc execution
4524 counts to a file called @file{@var{sourcename}.gcda} for each source
4525 file The information in this data file is very dependent on the
4526 structure of the generated code, so you must use the same source code
4527 and the same optimization options for both compilations.
4529 With @option{-fbranch-probabilities}, GCC puts a
4530 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
4531 These can be used to improve optimization. Currently, they are only
4532 used in one place: in @file{reorg.c}, instead of guessing which path a
4533 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
4534 exactly determine which path is taken more often.
4536 @item -fprofile-values
4537 @opindex fprofile-values
4538 If combined with @option{-fprofile-arcs}, it adds code so that some
4539 data about values of expressions in the program is gathered.
4541 With @option{-fbranch-probabilities}, it reads back the data gathered
4542 from profiling values of expressions and adds @samp{REG_VALUE_PROFILE}
4543 notes to instructions for their later usage in optimizations.
4547 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
4548 a code to gather information about values of expressions.
4550 With @option{-fbranch-probabilities}, it reads back the data gathered
4551 and actually performs the optimizations based on them.
4552 Currently the optimizations include specialization of division operation
4553 using the knowledge about the value of the denominator.
4557 Use a graph coloring register allocator. Currently this option is meant
4558 for testing, so we are interested to hear about miscompilations with
4563 Perform tail duplication to enlarge superblock size. This transformation
4564 simplifies the control flow of the function allowing other optimizations to do
4567 @item -funit-at-a-time
4568 @opindex funit-at-a-time
4569 Parse the whole compilation unit before starting to produce code.
4570 This allows some extra optimizations to take place but consumes more
4573 @item -funroll-loops
4574 @opindex funroll-loops
4575 Unroll loops whose number of iterations can be determined at compile time or
4576 upon entry to the loop. @option{-funroll-loops} implies
4577 @option{-frerun-cse-after-loop}. It also turns on complete loop peeling
4578 (i.e. complete removal of loops with small constant number of iterations).
4579 This option makes code larger, and may or may not make it run faster.
4581 @item -funroll-all-loops
4582 @opindex funroll-all-loops
4583 Unroll all loops, even if their number of iterations is uncertain when
4584 the loop is entered. This usually makes programs run more slowly.
4585 @option{-funroll-all-loops} implies the same options as
4586 @option{-funroll-loops}.
4589 @opindex fpeel-loops
4590 Peels the loops for that there is enough information that they do not
4591 roll much (from profile feedback). It also turns on complete loop peeling
4592 (i.e. complete removal of loops with small constant number of iterations).
4594 @item -funswitch-loops
4595 @opindex funswitch-loops
4596 Move branches with loop invariant conditions out of the loop, with duplicates
4597 of the loop on both branches (modified according to result of the condition).
4599 @item -fold-unroll-loops
4600 @opindex fold-unroll-loops
4601 Unroll loops whose number of iterations can be determined at compile
4602 time or upon entry to the loop, using the old loop unroller whose loop
4603 recognition is based on notes from frontend. @option{-fold-unroll-loops} implies
4604 both @option{-fstrength-reduce} and @option{-frerun-cse-after-loop}. This
4605 option makes code larger, and may or may not make it run faster.
4607 @item -fold-unroll-all-loops
4608 @opindex fold-unroll-all-loops
4609 Unroll all loops, even if their number of iterations is uncertain when
4610 the loop is entered. This is done using the old loop unroller whose loop
4611 recognition is based on notes from frontend. This usually makes programs run more slowly.
4612 @option{-fold-unroll-all-loops} implies the same options as
4613 @option{-fold-unroll-loops}.
4615 @item -funswitch-loops
4616 @opindex funswitch-loops
4617 Move branches with loop invariant conditions out of the loop, with duplicates
4618 of the loop on both branches (modified according to result of the condition).
4620 @item -funswitch-loops
4621 @opindex funswitch-loops
4622 Move branches with loop invariant conditions out of the loop, with duplicates
4623 of the loop on both branches (modified according to result of the condition).
4625 @item -fprefetch-loop-arrays
4626 @opindex fprefetch-loop-arrays
4627 If supported by the target machine, generate instructions to prefetch
4628 memory to improve the performance of loops that access large arrays.
4630 Disabled at level @option{-Os}.
4632 @item -ffunction-sections
4633 @itemx -fdata-sections
4634 @opindex ffunction-sections
4635 @opindex fdata-sections
4636 Place each function or data item into its own section in the output
4637 file if the target supports arbitrary sections. The name of the
4638 function or the name of the data item determines the section's name
4641 Use these options on systems where the linker can perform optimizations
4642 to improve locality of reference in the instruction space. Most systems
4643 using the ELF object format and SPARC processors running Solaris 2 have
4644 linkers with such optimizations. AIX may have these optimizations in
4647 Only use these options when there are significant benefits from doing
4648 so. When you specify these options, the assembler and linker will
4649 create larger object and executable files and will also be slower.
4650 You will not be able to use @code{gprof} on all systems if you
4651 specify this option and you may have problems with debugging if
4652 you specify both this option and @option{-g}.
4654 @item -fbranch-target-load-optimize
4655 @opindex fbranch-target-load-optimize
4656 Perform branch target register load optimization before prologue / epilogue
4658 The use of target registers can typically be exposed only during reload,
4659 thus hoisting loads out of loops and doing inter-block scheduling needs
4660 a separate optimization pass.
4662 @item -fbranch-target-load-optimize2
4663 @opindex fbranch-target-load-optimize2
4664 Perform branch target register load optimization after prologue / epilogue
4667 @item --param @var{name}=@var{value}
4669 In some places, GCC uses various constants to control the amount of
4670 optimization that is done. For example, GCC will not inline functions
4671 that contain more that a certain number of instructions. You can
4672 control some of these constants on the command-line using the
4673 @option{--param} option.
4675 In each case, the @var{value} is an integer. The allowable choices for
4676 @var{name} are given in the following table:
4679 @item max-crossjump-edges
4680 The maximum number of incoming edges to consider for crossjumping.
4681 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
4682 the number of edges incoming to each block. Increasing values mean
4683 more aggressive optimization, making the compile time increase with
4684 probably small improvement in executable size.
4686 @item max-delay-slot-insn-search
4687 The maximum number of instructions to consider when looking for an
4688 instruction to fill a delay slot. If more than this arbitrary number of
4689 instructions is searched, the time savings from filling the delay slot
4690 will be minimal so stop searching. Increasing values mean more
4691 aggressive optimization, making the compile time increase with probably
4692 small improvement in executable run time.
4694 @item max-delay-slot-live-search
4695 When trying to fill delay slots, the maximum number of instructions to
4696 consider when searching for a block with valid live register
4697 information. Increasing this arbitrarily chosen value means more
4698 aggressive optimization, increasing the compile time. This parameter
4699 should be removed when the delay slot code is rewritten to maintain the
4702 @item max-gcse-memory
4703 The approximate maximum amount of memory that will be allocated in
4704 order to perform the global common subexpression elimination
4705 optimization. If more memory than specified is required, the
4706 optimization will not be done.
4708 @item max-gcse-passes
4709 The maximum number of passes of GCSE to run.
4711 @item max-pending-list-length
4712 The maximum number of pending dependencies scheduling will allow
4713 before flushing the current state and starting over. Large functions
4714 with few branches or calls can create excessively large lists which
4715 needlessly consume memory and resources.
4717 @item max-inline-insns-single
4718 Several parameters control the tree inliner used in gcc.
4719 This number sets the maximum number of instructions (counted in gcc's
4720 internal representation) in a single function that the tree inliner
4721 will consider for inlining. This only affects functions declared
4722 inline and methods implemented in a class declaration (C++).
4723 The default value is 500.
4725 @item max-inline-insns-auto
4726 When you use @option{-finline-functions} (included in @option{-O3}),
4727 a lot of functions that would otherwise not be considered for inlining
4728 by the compiler will be investigated. To those functions, a different
4729 (more restrictive) limit compared to functions declared inline can
4731 The default value is 150.
4733 @item large-function-insns
4734 The limit specifying really large functions. For functions greater than this
4735 limit inlining is constrained by @option{--param large-function-growth}.
4736 This parameter is useful primarily to avoid extreme compilation time caused by non-linear
4737 algorithms used by the backend.
4738 This parameter is ignored when @option{-funit-at-a-time} is not used.
4739 The default value is 30000.
4741 @item large-function-growth
4742 Specifies maximal growth of large functtion caused by inlining in percents.
4743 This parameter is ignored when @option{-funit-at-a-time} is not used.
4744 The default value is 200.
4746 @item inline-unit-growth
4747 Specifies maximal overall growth of the compilation unit caused by inlining.
4748 This parameter is ignored when @option{-funit-at-a-time} is not used.
4749 The default value is 150.
4751 @item max-inline-insns-rtl
4752 For languages that use the RTL inliner (this happens at a later stage
4753 than tree inlining), you can set the maximum allowable size (counted
4754 in RTL instructions) for the RTL inliner with this parameter.
4755 The default value is 600.
4757 @item max-unrolled-insns
4758 The maximum number of instructions that a loop should have if that loop
4759 is unrolled, and if the loop is unrolled, it determines how many times
4760 the loop code is unrolled.
4762 @item max-average-unrolled-insns
4763 The maximum number of instructions biased by probabilities of their execution
4764 that a loop should have if that loop is unrolled, and if the loop is unrolled,
4765 it determines how many times the loop code is unrolled.
4767 @item max-unroll-times
4768 The maximum number of unrollings of a single loop.
4770 @item max-peeled-insns
4771 The maximum number of instructions that a loop should have if that loop
4772 is peeled, and if the loop is peeled, it determines how many times
4773 the loop code is peeled.
4775 @item max-peel-times
4776 The maximum number of peelings of a single loop.
4778 @item max-completely-peeled-insns
4779 The maximum number of insns of a completely peeled loop.
4781 @item max-completely-peel-times
4782 The maximum number of iterations of a loop to be suitable for complete peeling.
4784 @item max-unswitch-insns
4785 The maximum number of insns of an unswitched loop.
4787 @item max-unswitch-level
4788 The maximum number of branches unswitched in a single loop.
4790 @item hot-bb-count-fraction
4791 Select fraction of the maximal count of repetitions of basic block in program
4792 given basic block needs to have to be considered hot.
4794 @item hot-bb-frequency-fraction
4795 Select fraction of the maximal frequency of executions of basic block in
4796 function given basic block needs to have to be considered hot
4798 @item tracer-dynamic-coverage
4799 @itemx tracer-dynamic-coverage-feedback
4801 This value is used to limit superblock formation once the given percentage of
4802 executed instructions is covered. This limits unnecessary code size
4805 The @option{tracer-dynamic-coverage-feedback} is used only when profile
4806 feedback is available. The real profiles (as opposed to statically estimated
4807 ones) are much less balanced allowing the threshold to be larger value.
4809 @item tracer-max-code-growth
4810 Stop tail duplication once code growth has reached given percentage. This is
4811 rather hokey argument, as most of the duplicates will be eliminated later in
4812 cross jumping, so it may be set to much higher values than is the desired code
4815 @item tracer-min-branch-ratio
4817 Stop reverse growth when the reverse probability of best edge is less than this
4818 threshold (in percent).
4820 @item tracer-min-branch-ratio
4821 @itemx tracer-min-branch-ratio-feedback
4823 Stop forward growth if the best edge do have probability lower than this
4826 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
4827 compilation for profile feedback and one for compilation without. The value
4828 for compilation with profile feedback needs to be more conservative (higher) in
4829 order to make tracer effective.
4831 @item max-cse-path-length
4833 Maximum number of basic blocks on path that cse considers.
4835 @item ggc-min-expand
4837 GCC uses a garbage collector to manage its own memory allocation. This
4838 parameter specifies the minimum percentage by which the garbage
4839 collector's heap should be allowed to expand between collections.
4840 Tuning this may improve compilation speed; it has no effect on code
4843 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
4844 RAM >= 1GB. If @code{getrlimit} is available, the notion of "RAM" is
4845 the smallest of actual RAM, RLIMIT_RSS, RLIMIT_DATA and RLIMIT_AS. If
4846 GCC is not able to calculate RAM on a particular platform, the lower
4847 bound of 30% is used. Setting this parameter and
4848 @option{ggc-min-heapsize} to zero causes a full collection to occur at
4849 every opportunity. This is extremely slow, but can be useful for
4852 @item ggc-min-heapsize
4854 Minimum size of the garbage collector's heap before it begins bothering
4855 to collect garbage. The first collection occurs after the heap expands
4856 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
4857 tuning this may improve compilation speed, and has no effect on code
4860 The default is RAM/8, with a lower bound of 4096 (four megabytes) and an
4861 upper bound of 131072 (128 megabytes). If @code{getrlimit} is
4862 available, the notion of "RAM" is the smallest of actual RAM,
4863 RLIMIT_RSS, RLIMIT_DATA and RLIMIT_AS. If GCC is not able to calculate
4864 RAM on a particular platform, the lower bound is used. Setting this
4865 parameter very large effectively disables garbage collection. Setting
4866 this parameter and @option{ggc-min-expand} to zero causes a full
4867 collection to occur at every opportunity.
4869 @item reorder-blocks-duplicate
4870 @itemx reorder-blocks-duplicate-feedback
4872 Used by basic block reordering pass to decide whether to use unconditional
4873 branch or duplicate the code on its destination. Code is duplicated when its
4874 estimated size is smaller than this value multiplied by the estimated size of
4875 unconditional jump in the hot spots of the program.
4877 The @option{reorder-block-duplicate-feedback} is used only when profile
4878 feedback is available and may be set to higher values than
4879 @option{reorder-block-duplicate} since information about the hot spots is more
4884 @node Preprocessor Options
4885 @section Options Controlling the Preprocessor
4886 @cindex preprocessor options
4887 @cindex options, preprocessor
4889 These options control the C preprocessor, which is run on each C source
4890 file before actual compilation.
4892 If you use the @option{-E} option, nothing is done except preprocessing.
4893 Some of these options make sense only together with @option{-E} because
4894 they cause the preprocessor output to be unsuitable for actual
4899 You can use @option{-Wp,@var{option}} to bypass the compiler driver
4900 and pass @var{option} directly through to the preprocessor. If
4901 @var{option} contains commas, it is split into multiple options at the
4902 commas. However, many options are modified, translated or interpreted
4903 by the compiler driver before being passed to the preprocessor, and
4904 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
4905 interface is undocumented and subject to change, so whenever possible
4906 you should avoid using @option{-Wp} and let the driver handle the
4909 @item -Xpreprocessor @var{option}
4910 @opindex preprocessor
4911 Pass @var{option} as an option to the preprocessor. You can use this to
4912 supply system-specific preprocessor options which GCC does not know how to
4915 If you want to pass an option that takes an argument, you must use
4916 @option{-Xpreprocessor} twice, once for the option and once for the argument.
4919 @include cppopts.texi
4921 @node Assembler Options
4922 @section Passing Options to the Assembler
4924 @c prevent bad page break with this line
4925 You can pass options to the assembler.
4928 @item -Wa,@var{option}
4930 Pass @var{option} as an option to the assembler. If @var{option}
4931 contains commas, it is split into multiple options at the commas.
4933 @item -Xassembler @var{option}
4935 Pass @var{option} as an option to the assembler. You can use this to
4936 supply system-specific assembler options which GCC does not know how to
4939 If you want to pass an option that takes an argument, you must use
4940 @option{-Xassembler} twice, once for the option and once for the argument.
4945 @section Options for Linking
4946 @cindex link options
4947 @cindex options, linking
4949 These options come into play when the compiler links object files into
4950 an executable output file. They are meaningless if the compiler is
4951 not doing a link step.
4955 @item @var{object-file-name}
4956 A file name that does not end in a special recognized suffix is
4957 considered to name an object file or library. (Object files are
4958 distinguished from libraries by the linker according to the file
4959 contents.) If linking is done, these object files are used as input
4968 If any of these options is used, then the linker is not run, and
4969 object file names should not be used as arguments. @xref{Overall
4973 @item -l@var{library}
4974 @itemx -l @var{library}
4976 Search the library named @var{library} when linking. (The second
4977 alternative with the library as a separate argument is only for
4978 POSIX compliance and is not recommended.)
4980 It makes a difference where in the command you write this option; the
4981 linker searches and processes libraries and object files in the order they
4982 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
4983 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
4984 to functions in @samp{z}, those functions may not be loaded.
4986 The linker searches a standard list of directories for the library,
4987 which is actually a file named @file{lib@var{library}.a}. The linker
4988 then uses this file as if it had been specified precisely by name.
4990 The directories searched include several standard system directories
4991 plus any that you specify with @option{-L}.
4993 Normally the files found this way are library files---archive files
4994 whose members are object files. The linker handles an archive file by
4995 scanning through it for members which define symbols that have so far
4996 been referenced but not defined. But if the file that is found is an
4997 ordinary object file, it is linked in the usual fashion. The only
4998 difference between using an @option{-l} option and specifying a file name
4999 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
5000 and searches several directories.
5004 You need this special case of the @option{-l} option in order to
5005 link an Objective-C program.
5008 @opindex nostartfiles
5009 Do not use the standard system startup files when linking.
5010 The standard system libraries are used normally, unless @option{-nostdlib}
5011 or @option{-nodefaultlibs} is used.
5013 @item -nodefaultlibs
5014 @opindex nodefaultlibs
5015 Do not use the standard system libraries when linking.
5016 Only the libraries you specify will be passed to the linker.
5017 The standard startup files are used normally, unless @option{-nostartfiles}
5018 is used. The compiler may generate calls to memcmp, memset, and memcpy
5019 for System V (and ISO C) environments or to bcopy and bzero for
5020 BSD environments. These entries are usually resolved by entries in
5021 libc. These entry points should be supplied through some other
5022 mechanism when this option is specified.
5026 Do not use the standard system startup files or libraries when linking.
5027 No startup files and only the libraries you specify will be passed to
5028 the linker. The compiler may generate calls to memcmp, memset, and memcpy
5029 for System V (and ISO C) environments or to bcopy and bzero for
5030 BSD environments. These entries are usually resolved by entries in
5031 libc. These entry points should be supplied through some other
5032 mechanism when this option is specified.
5034 @cindex @option{-lgcc}, use with @option{-nostdlib}
5035 @cindex @option{-nostdlib} and unresolved references
5036 @cindex unresolved references and @option{-nostdlib}
5037 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
5038 @cindex @option{-nodefaultlibs} and unresolved references
5039 @cindex unresolved references and @option{-nodefaultlibs}
5040 One of the standard libraries bypassed by @option{-nostdlib} and
5041 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
5042 that GCC uses to overcome shortcomings of particular machines, or special
5043 needs for some languages.
5044 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
5045 Collection (GCC) Internals},
5046 for more discussion of @file{libgcc.a}.)
5047 In most cases, you need @file{libgcc.a} even when you want to avoid
5048 other standard libraries. In other words, when you specify @option{-nostdlib}
5049 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
5050 This ensures that you have no unresolved references to internal GCC
5051 library subroutines. (For example, @samp{__main}, used to ensure C++
5052 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
5053 GNU Compiler Collection (GCC) Internals}.)
5057 Produce a position independent executable on targets which support it.
5058 For predictable results, you must also specify the same set of options
5059 that were used to generate code (@option{-fpie}, @option{-fPIE},
5060 or model suboptions) when you specify this option.
5064 Remove all symbol table and relocation information from the executable.
5068 On systems that support dynamic linking, this prevents linking with the shared
5069 libraries. On other systems, this option has no effect.
5073 Produce a shared object which can then be linked with other objects to
5074 form an executable. Not all systems support this option. For predictable
5075 results, you must also specify the same set of options that were used to
5076 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
5077 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
5078 needs to build supplementary stub code for constructors to work. On
5079 multi-libbed systems, @samp{gcc -shared} must select the correct support
5080 libraries to link against. Failing to supply the correct flags may lead
5081 to subtle defects. Supplying them in cases where they are not necessary
5084 @item -shared-libgcc
5085 @itemx -static-libgcc
5086 @opindex shared-libgcc
5087 @opindex static-libgcc
5088 On systems that provide @file{libgcc} as a shared library, these options
5089 force the use of either the shared or static version respectively.
5090 If no shared version of @file{libgcc} was built when the compiler was
5091 configured, these options have no effect.
5093 There are several situations in which an application should use the
5094 shared @file{libgcc} instead of the static version. The most common
5095 of these is when the application wishes to throw and catch exceptions
5096 across different shared libraries. In that case, each of the libraries
5097 as well as the application itself should use the shared @file{libgcc}.
5099 Therefore, the G++ and GCJ drivers automatically add
5100 @option{-shared-libgcc} whenever you build a shared library or a main
5101 executable, because C++ and Java programs typically use exceptions, so
5102 this is the right thing to do.
5104 If, instead, you use the GCC driver to create shared libraries, you may
5105 find that they will not always be linked with the shared @file{libgcc}.
5106 If GCC finds, at its configuration time, that you have a GNU linker that
5107 does not support option @option{--eh-frame-hdr}, it will link the shared
5108 version of @file{libgcc} into shared libraries by default. Otherwise,
5109 it will take advantage of the linker and optimize away the linking with
5110 the shared version of @file{libgcc}, linking with the static version of
5111 libgcc by default. This allows exceptions to propagate through such
5112 shared libraries, without incurring relocation costs at library load
5115 However, if a library or main executable is supposed to throw or catch
5116 exceptions, you must link it using the G++ or GCJ driver, as appropriate
5117 for the languages used in the program, or using the option
5118 @option{-shared-libgcc}, such that it is linked with the shared
5123 Bind references to global symbols when building a shared object. Warn
5124 about any unresolved references (unless overridden by the link editor
5125 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
5128 @item -Xlinker @var{option}
5130 Pass @var{option} as an option to the linker. You can use this to
5131 supply system-specific linker options which GCC does not know how to
5134 If you want to pass an option that takes an argument, you must use
5135 @option{-Xlinker} twice, once for the option and once for the argument.
5136 For example, to pass @option{-assert definitions}, you must write
5137 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
5138 @option{-Xlinker "-assert definitions"}, because this passes the entire
5139 string as a single argument, which is not what the linker expects.
5141 @item -Wl,@var{option}
5143 Pass @var{option} as an option to the linker. If @var{option} contains
5144 commas, it is split into multiple options at the commas.
5146 @item -u @var{symbol}
5148 Pretend the symbol @var{symbol} is undefined, to force linking of
5149 library modules to define it. You can use @option{-u} multiple times with
5150 different symbols to force loading of additional library modules.
5153 @node Directory Options
5154 @section Options for Directory Search
5155 @cindex directory options
5156 @cindex options, directory search
5159 These options specify directories to search for header files, for
5160 libraries and for parts of the compiler:
5165 Add the directory @var{dir} to the head of the list of directories to be
5166 searched for header files. This can be used to override a system header
5167 file, substituting your own version, since these directories are
5168 searched before the system header file directories. However, you should
5169 not use this option to add directories that contain vendor-supplied
5170 system header files (use @option{-isystem} for that). If you use more than
5171 one @option{-I} option, the directories are scanned in left-to-right
5172 order; the standard system directories come after.
5174 If a standard system include directory, or a directory specified with
5175 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
5176 option will be ignored. The directory will still be searched but as a
5177 system directory at its normal position in the system include chain.
5178 This is to ensure that GCC's procedure to fix buggy system headers and
5179 the ordering for the include_next directive are not inadvertently changed.
5180 If you really need to change the search order for system directories,
5181 use the @option{-nostdinc} and/or @option{-isystem} options.
5185 Any directories you specify with @option{-I} options before the @option{-I-}
5186 option are searched only for the case of @samp{#include "@var{file}"};
5187 they are not searched for @samp{#include <@var{file}>}.
5189 If additional directories are specified with @option{-I} options after
5190 the @option{-I-}, these directories are searched for all @samp{#include}
5191 directives. (Ordinarily @emph{all} @option{-I} directories are used
5194 In addition, the @option{-I-} option inhibits the use of the current
5195 directory (where the current input file came from) as the first search
5196 directory for @samp{#include "@var{file}"}. There is no way to
5197 override this effect of @option{-I-}. With @option{-I.} you can specify
5198 searching the directory which was current when the compiler was
5199 invoked. That is not exactly the same as what the preprocessor does
5200 by default, but it is often satisfactory.
5202 @option{-I-} does not inhibit the use of the standard system directories
5203 for header files. Thus, @option{-I-} and @option{-nostdinc} are
5208 Add directory @var{dir} to the list of directories to be searched
5211 @item -B@var{prefix}
5213 This option specifies where to find the executables, libraries,
5214 include files, and data files of the compiler itself.
5216 The compiler driver program runs one or more of the subprograms
5217 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
5218 @var{prefix} as a prefix for each program it tries to run, both with and
5219 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
5221 For each subprogram to be run, the compiler driver first tries the
5222 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
5223 was not specified, the driver tries two standard prefixes, which are
5224 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
5225 those results in a file name that is found, the unmodified program
5226 name is searched for using the directories specified in your
5227 @env{PATH} environment variable.
5229 The compiler will check to see if the path provided by the @option{-B}
5230 refers to a directory, and if necessary it will add a directory
5231 separator character at the end of the path.
5233 @option{-B} prefixes that effectively specify directory names also apply
5234 to libraries in the linker, because the compiler translates these
5235 options into @option{-L} options for the linker. They also apply to
5236 includes files in the preprocessor, because the compiler translates these
5237 options into @option{-isystem} options for the preprocessor. In this case,
5238 the compiler appends @samp{include} to the prefix.
5240 The run-time support file @file{libgcc.a} can also be searched for using
5241 the @option{-B} prefix, if needed. If it is not found there, the two
5242 standard prefixes above are tried, and that is all. The file is left
5243 out of the link if it is not found by those means.
5245 Another way to specify a prefix much like the @option{-B} prefix is to use
5246 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
5249 As a special kludge, if the path provided by @option{-B} is
5250 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
5251 9, then it will be replaced by @file{[dir/]include}. This is to help
5252 with boot-strapping the compiler.
5254 @item -specs=@var{file}
5256 Process @var{file} after the compiler reads in the standard @file{specs}
5257 file, in order to override the defaults that the @file{gcc} driver
5258 program uses when determining what switches to pass to @file{cc1},
5259 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
5260 @option{-specs=@var{file}} can be specified on the command line, and they
5261 are processed in order, from left to right.
5267 @section Specifying subprocesses and the switches to pass to them
5270 @command{gcc} is a driver program. It performs its job by invoking a
5271 sequence of other programs to do the work of compiling, assembling and
5272 linking. GCC interprets its command-line parameters and uses these to
5273 deduce which programs it should invoke, and which command-line options
5274 it ought to place on their command lines. This behavior is controlled
5275 by @dfn{spec strings}. In most cases there is one spec string for each
5276 program that GCC can invoke, but a few programs have multiple spec
5277 strings to control their behavior. The spec strings built into GCC can
5278 be overridden by using the @option{-specs=} command-line switch to specify
5281 @dfn{Spec files} are plaintext files that are used to construct spec
5282 strings. They consist of a sequence of directives separated by blank
5283 lines. The type of directive is determined by the first non-whitespace
5284 character on the line and it can be one of the following:
5287 @item %@var{command}
5288 Issues a @var{command} to the spec file processor. The commands that can
5292 @item %include <@var{file}>
5294 Search for @var{file} and insert its text at the current point in the
5297 @item %include_noerr <@var{file}>
5298 @cindex %include_noerr
5299 Just like @samp{%include}, but do not generate an error message if the include
5300 file cannot be found.
5302 @item %rename @var{old_name} @var{new_name}
5304 Rename the spec string @var{old_name} to @var{new_name}.
5308 @item *[@var{spec_name}]:
5309 This tells the compiler to create, override or delete the named spec
5310 string. All lines after this directive up to the next directive or
5311 blank line are considered to be the text for the spec string. If this
5312 results in an empty string then the spec will be deleted. (Or, if the
5313 spec did not exist, then nothing will happened.) Otherwise, if the spec
5314 does not currently exist a new spec will be created. If the spec does
5315 exist then its contents will be overridden by the text of this
5316 directive, unless the first character of that text is the @samp{+}
5317 character, in which case the text will be appended to the spec.
5319 @item [@var{suffix}]:
5320 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
5321 and up to the next directive or blank line are considered to make up the
5322 spec string for the indicated suffix. When the compiler encounters an
5323 input file with the named suffix, it will processes the spec string in
5324 order to work out how to compile that file. For example:
5331 This says that any input file whose name ends in @samp{.ZZ} should be
5332 passed to the program @samp{z-compile}, which should be invoked with the
5333 command-line switch @option{-input} and with the result of performing the
5334 @samp{%i} substitution. (See below.)
5336 As an alternative to providing a spec string, the text that follows a
5337 suffix directive can be one of the following:
5340 @item @@@var{language}
5341 This says that the suffix is an alias for a known @var{language}. This is
5342 similar to using the @option{-x} command-line switch to GCC to specify a
5343 language explicitly. For example:
5350 Says that .ZZ files are, in fact, C++ source files.
5353 This causes an error messages saying:
5356 @var{name} compiler not installed on this system.
5360 GCC already has an extensive list of suffixes built into it.
5361 This directive will add an entry to the end of the list of suffixes, but
5362 since the list is searched from the end backwards, it is effectively
5363 possible to override earlier entries using this technique.
5367 GCC has the following spec strings built into it. Spec files can
5368 override these strings or create their own. Note that individual
5369 targets can also add their own spec strings to this list.
5372 asm Options to pass to the assembler
5373 asm_final Options to pass to the assembler post-processor
5374 cpp Options to pass to the C preprocessor
5375 cc1 Options to pass to the C compiler
5376 cc1plus Options to pass to the C++ compiler
5377 endfile Object files to include at the end of the link
5378 link Options to pass to the linker
5379 lib Libraries to include on the command line to the linker
5380 libgcc Decides which GCC support library to pass to the linker
5381 linker Sets the name of the linker
5382 predefines Defines to be passed to the C preprocessor
5383 signed_char Defines to pass to CPP to say whether @code{char} is signed
5385 startfile Object files to include at the start of the link
5388 Here is a small example of a spec file:
5394 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
5397 This example renames the spec called @samp{lib} to @samp{old_lib} and
5398 then overrides the previous definition of @samp{lib} with a new one.
5399 The new definition adds in some extra command-line options before
5400 including the text of the old definition.
5402 @dfn{Spec strings} are a list of command-line options to be passed to their
5403 corresponding program. In addition, the spec strings can contain
5404 @samp{%}-prefixed sequences to substitute variable text or to
5405 conditionally insert text into the command line. Using these constructs
5406 it is possible to generate quite complex command lines.
5408 Here is a table of all defined @samp{%}-sequences for spec
5409 strings. Note that spaces are not generated automatically around the
5410 results of expanding these sequences. Therefore you can concatenate them
5411 together or combine them with constant text in a single argument.
5415 Substitute one @samp{%} into the program name or argument.
5418 Substitute the name of the input file being processed.
5421 Substitute the basename of the input file being processed.
5422 This is the substring up to (and not including) the last period
5423 and not including the directory.
5426 This is the same as @samp{%b}, but include the file suffix (text after
5430 Marks the argument containing or following the @samp{%d} as a
5431 temporary file name, so that that file will be deleted if GCC exits
5432 successfully. Unlike @samp{%g}, this contributes no text to the
5435 @item %g@var{suffix}
5436 Substitute a file name that has suffix @var{suffix} and is chosen
5437 once per compilation, and mark the argument in the same way as
5438 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
5439 name is now chosen in a way that is hard to predict even when previously
5440 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
5441 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
5442 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
5443 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
5444 was simply substituted with a file name chosen once per compilation,
5445 without regard to any appended suffix (which was therefore treated
5446 just like ordinary text), making such attacks more likely to succeed.
5448 @item %u@var{suffix}
5449 Like @samp{%g}, but generates a new temporary file name even if
5450 @samp{%u@var{suffix}} was already seen.
5452 @item %U@var{suffix}
5453 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
5454 new one if there is no such last file name. In the absence of any
5455 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
5456 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
5457 would involve the generation of two distinct file names, one
5458 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
5459 simply substituted with a file name chosen for the previous @samp{%u},
5460 without regard to any appended suffix.
5462 @item %j@var{suffix}
5463 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
5464 writable, and if save-temps is off; otherwise, substitute the name
5465 of a temporary file, just like @samp{%u}. This temporary file is not
5466 meant for communication between processes, but rather as a junk
5469 @item %|@var{suffix}
5470 @itemx %m@var{suffix}
5471 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
5472 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
5473 all. These are the two most common ways to instruct a program that it
5474 should read from standard input or write to standard output. If you
5475 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
5476 construct: see for example @file{f/lang-specs.h}.
5478 @item %.@var{SUFFIX}
5479 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
5480 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
5481 terminated by the next space or %.
5484 Marks the argument containing or following the @samp{%w} as the
5485 designated output file of this compilation. This puts the argument
5486 into the sequence of arguments that @samp{%o} will substitute later.
5489 Substitutes the names of all the output files, with spaces
5490 automatically placed around them. You should write spaces
5491 around the @samp{%o} as well or the results are undefined.
5492 @samp{%o} is for use in the specs for running the linker.
5493 Input files whose names have no recognized suffix are not compiled
5494 at all, but they are included among the output files, so they will
5498 Substitutes the suffix for object files. Note that this is
5499 handled specially when it immediately follows @samp{%g, %u, or %U},
5500 because of the need for those to form complete file names. The
5501 handling is such that @samp{%O} is treated exactly as if it had already
5502 been substituted, except that @samp{%g, %u, and %U} do not currently
5503 support additional @var{suffix} characters following @samp{%O} as they would
5504 following, for example, @samp{.o}.
5507 Substitutes the standard macro predefinitions for the
5508 current target machine. Use this when running @code{cpp}.
5511 Like @samp{%p}, but puts @samp{__} before and after the name of each
5512 predefined macro, except for macros that start with @samp{__} or with
5513 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
5517 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
5518 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}), and
5519 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
5523 Current argument is the name of a library or startup file of some sort.
5524 Search for that file in a standard list of directories and substitute
5525 the full name found.
5528 Print @var{str} as an error message. @var{str} is terminated by a newline.
5529 Use this when inconsistent options are detected.
5532 Substitute the contents of spec string @var{name} at this point.
5535 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
5537 @item %x@{@var{option}@}
5538 Accumulate an option for @samp{%X}.
5541 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
5545 Output the accumulated assembler options specified by @option{-Wa}.
5548 Output the accumulated preprocessor options specified by @option{-Wp}.
5551 Process the @code{asm} spec. This is used to compute the
5552 switches to be passed to the assembler.
5555 Process the @code{asm_final} spec. This is a spec string for
5556 passing switches to an assembler post-processor, if such a program is
5560 Process the @code{link} spec. This is the spec for computing the
5561 command line passed to the linker. Typically it will make use of the
5562 @samp{%L %G %S %D and %E} sequences.
5565 Dump out a @option{-L} option for each directory that GCC believes might
5566 contain startup files. If the target supports multilibs then the
5567 current multilib directory will be prepended to each of these paths.
5570 Output the multilib directory with directory separators replaced with
5571 @samp{_}. If multilib directories are not set, or the multilib directory is
5572 @file{.} then this option emits nothing.
5575 Process the @code{lib} spec. This is a spec string for deciding which
5576 libraries should be included on the command line to the linker.
5579 Process the @code{libgcc} spec. This is a spec string for deciding
5580 which GCC support library should be included on the command line to the linker.
5583 Process the @code{startfile} spec. This is a spec for deciding which
5584 object files should be the first ones passed to the linker. Typically
5585 this might be a file named @file{crt0.o}.
5588 Process the @code{endfile} spec. This is a spec string that specifies
5589 the last object files that will be passed to the linker.
5592 Process the @code{cpp} spec. This is used to construct the arguments
5593 to be passed to the C preprocessor.
5596 Process the @code{signed_char} spec. This is intended to be used
5597 to tell cpp whether a char is signed. It typically has the definition:
5599 %@{funsigned-char:-D__CHAR_UNSIGNED__@}
5603 Process the @code{cc1} spec. This is used to construct the options to be
5604 passed to the actual C compiler (@samp{cc1}).
5607 Process the @code{cc1plus} spec. This is used to construct the options to be
5608 passed to the actual C++ compiler (@samp{cc1plus}).
5611 Substitute the variable part of a matched option. See below.
5612 Note that each comma in the substituted string is replaced by
5616 Remove all occurrences of @code{-S} from the command line. Note---this
5617 command is position dependent. @samp{%} commands in the spec string
5618 before this one will see @code{-S}, @samp{%} commands in the spec string
5619 after this one will not.
5621 @item %:@var{function}(@var{args})
5622 Call the named function @var{function}, passing it @var{args}.
5623 @var{args} is first processed as a nested spec string, then split
5624 into an argument vector in the usual fashion. The function returns
5625 a string which is processed as if it had appeared literally as part
5626 of the current spec.
5628 The following built-in spec functions are provided:
5631 @item @code{if-exists}
5632 The @code{if-exists} spec function takes one argument, an absolute
5633 pathname to a file. If the file exists, @code{if-exists} returns the
5634 pathname. Here is a small example of its usage:
5638 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
5641 @item @code{if-exists-else}
5642 The @code{if-exists-else} spec function is similar to the @code{if-exists}
5643 spec function, except that it takes two arguments. The first argument is
5644 an absolute pathname to a file. If the file exists, @code{if-exists-else}
5645 returns the pathname. If it does not exist, it returns the second argument.
5646 This way, @code{if-exists-else} can be used to select one file or another,
5647 based on the existence of the first. Here is a small example of its usage:
5651 crt0%O%s %:if-exists(crti%O%s) \
5652 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
5657 Substitutes the @code{-S} switch, if that switch was given to GCC@.
5658 If that switch was not specified, this substitutes nothing. Note that
5659 the leading dash is omitted when specifying this option, and it is
5660 automatically inserted if the substitution is performed. Thus the spec
5661 string @samp{%@{foo@}} would match the command-line option @option{-foo}
5662 and would output the command line option @option{-foo}.
5664 @item %W@{@code{S}@}
5665 Like %@{@code{S}@} but mark last argument supplied within as a file to be
5668 @item %@{@code{S}*@}
5669 Substitutes all the switches specified to GCC whose names start
5670 with @code{-S}, but which also take an argument. This is used for
5671 switches like @option{-o}, @option{-D}, @option{-I}, etc.
5672 GCC considers @option{-o foo} as being
5673 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
5674 text, including the space. Thus two arguments would be generated.
5676 @item %@{@code{S}*&@code{T}*@}
5677 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
5678 (the order of @code{S} and @code{T} in the spec is not significant).
5679 There can be any number of ampersand-separated variables; for each the
5680 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
5682 @item %@{@code{S}:@code{X}@}
5683 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
5685 @item %@{!@code{S}:@code{X}@}
5686 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
5688 @item %@{@code{S}*:@code{X}@}
5689 Substitutes @code{X} if one or more switches whose names start with
5690 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
5691 once, no matter how many such switches appeared. However, if @code{%*}
5692 appears somewhere in @code{X}, then @code{X} will be substituted once
5693 for each matching switch, with the @code{%*} replaced by the part of
5694 that switch that matched the @code{*}.
5696 @item %@{.@code{S}:@code{X}@}
5697 Substitutes @code{X}, if processing a file with suffix @code{S}.
5699 @item %@{!.@code{S}:@code{X}@}
5700 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
5702 @item %@{@code{S}|@code{P}:@code{X}@}
5703 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC@.
5704 This may be combined with @samp{!}, @samp{.}, and @code{*} sequences as well,
5705 although they have a stronger binding than the @samp{|}. If @code{%*}
5706 appears in @code{X}, all of the alternatives must be starred, and only
5707 the first matching alternative is substituted.
5709 For example, a spec string like this:
5712 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
5715 will output the following command-line options from the following input
5716 command-line options:
5721 -d fred.c -foo -baz -boggle
5722 -d jim.d -bar -baz -boggle
5725 @item %@{S:X; T:Y; :D@}
5727 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
5728 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
5729 be as many clauses as you need. This may be combined with @code{.},
5730 @code{!}, @code{|}, and @code{*} as needed.
5735 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
5736 construct may contain other nested @samp{%} constructs or spaces, or
5737 even newlines. They are processed as usual, as described above.
5738 Trailing white space in @code{X} is ignored. White space may also
5739 appear anywhere on the left side of the colon in these constructs,
5740 except between @code{.} or @code{*} and the corresponding word.
5742 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
5743 handled specifically in these constructs. If another value of
5744 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
5745 @option{-W} switch is found later in the command line, the earlier
5746 switch value is ignored, except with @{@code{S}*@} where @code{S} is
5747 just one letter, which passes all matching options.
5749 The character @samp{|} at the beginning of the predicate text is used to
5750 indicate that a command should be piped to the following command, but
5751 only if @option{-pipe} is specified.
5753 It is built into GCC which switches take arguments and which do not.
5754 (You might think it would be useful to generalize this to allow each
5755 compiler's spec to say which switches take arguments. But this cannot
5756 be done in a consistent fashion. GCC cannot even decide which input
5757 files have been specified without knowing which switches take arguments,
5758 and it must know which input files to compile in order to tell which
5761 GCC also knows implicitly that arguments starting in @option{-l} are to be
5762 treated as compiler output files, and passed to the linker in their
5763 proper position among the other output files.
5765 @c man begin OPTIONS
5767 @node Target Options
5768 @section Specifying Target Machine and Compiler Version
5769 @cindex target options
5770 @cindex cross compiling
5771 @cindex specifying machine version
5772 @cindex specifying compiler version and target machine
5773 @cindex compiler version, specifying
5774 @cindex target machine, specifying
5776 The usual way to run GCC is to run the executable called @file{gcc}, or
5777 @file{<machine>-gcc} when cross-compiling, or
5778 @file{<machine>-gcc-<version>} to run a version other than the one that
5779 was installed last. Sometimes this is inconvenient, so GCC provides
5780 options that will switch to another cross-compiler or version.
5783 @item -b @var{machine}
5785 The argument @var{machine} specifies the target machine for compilation.
5787 The value to use for @var{machine} is the same as was specified as the
5788 machine type when configuring GCC as a cross-compiler. For
5789 example, if a cross-compiler was configured with @samp{configure
5790 i386v}, meaning to compile for an 80386 running System V, then you
5791 would specify @option{-b i386v} to run that cross compiler.
5793 @item -V @var{version}
5795 The argument @var{version} specifies which version of GCC to run.
5796 This is useful when multiple versions are installed. For example,
5797 @var{version} might be @samp{2.0}, meaning to run GCC version 2.0.
5800 The @option{-V} and @option{-b} options work by running the
5801 @file{<machine>-gcc-<version>} executable, so there's no real reason to
5802 use them if you can just run that directly.
5804 @node Submodel Options
5805 @section Hardware Models and Configurations
5806 @cindex submodel options
5807 @cindex specifying hardware config
5808 @cindex hardware models and configurations, specifying
5809 @cindex machine dependent options
5811 Earlier we discussed the standard option @option{-b} which chooses among
5812 different installed compilers for completely different target
5813 machines, such as VAX vs.@: 68000 vs.@: 80386.
5815 In addition, each of these target machine types can have its own
5816 special options, starting with @samp{-m}, to choose among various
5817 hardware models or configurations---for example, 68010 vs 68020,
5818 floating coprocessor or none. A single installed version of the
5819 compiler can compile for any model or configuration, according to the
5822 Some configurations of the compiler also support additional special
5823 options, usually for compatibility with other compilers on the same
5826 These options are defined by the macro @code{TARGET_SWITCHES} in the
5827 machine description. The default for the options is also defined by
5828 that macro, which enables you to change the defaults.
5840 * RS/6000 and PowerPC Options::
5844 * i386 and x86-64 Options::
5846 * Intel 960 Options::
5847 * DEC Alpha Options::
5848 * DEC Alpha/VMS Options::
5851 * System V Options::
5852 * TMS320C3x/C4x Options::
5860 * S/390 and zSeries Options::
5864 * Xstormy16 Options::
5869 @node M680x0 Options
5870 @subsection M680x0 Options
5871 @cindex M680x0 options
5873 These are the @samp{-m} options defined for the 68000 series. The default
5874 values for these options depends on which style of 68000 was selected when
5875 the compiler was configured; the defaults for the most common choices are
5883 Generate output for a 68000. This is the default
5884 when the compiler is configured for 68000-based systems.
5886 Use this option for microcontrollers with a 68000 or EC000 core,
5887 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
5893 Generate output for a 68020. This is the default
5894 when the compiler is configured for 68020-based systems.
5898 Generate output containing 68881 instructions for floating point.
5899 This is the default for most 68020 systems unless @option{--nfp} was
5900 specified when the compiler was configured.
5904 Generate output for a 68030. This is the default when the compiler is
5905 configured for 68030-based systems.
5909 Generate output for a 68040. This is the default when the compiler is
5910 configured for 68040-based systems.
5912 This option inhibits the use of 68881/68882 instructions that have to be
5913 emulated by software on the 68040. Use this option if your 68040 does not
5914 have code to emulate those instructions.
5918 Generate output for a 68060. This is the default when the compiler is
5919 configured for 68060-based systems.
5921 This option inhibits the use of 68020 and 68881/68882 instructions that
5922 have to be emulated by software on the 68060. Use this option if your 68060
5923 does not have code to emulate those instructions.
5927 Generate output for a CPU32. This is the default
5928 when the compiler is configured for CPU32-based systems.
5930 Use this option for microcontrollers with a
5931 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
5932 68336, 68340, 68341, 68349 and 68360.
5936 Generate output for a 520X ``coldfire'' family cpu. This is the default
5937 when the compiler is configured for 520X-based systems.
5939 Use this option for microcontroller with a 5200 core, including
5940 the MCF5202, MCF5203, MCF5204 and MCF5202.
5945 Generate output for a 68040, without using any of the new instructions.
5946 This results in code which can run relatively efficiently on either a
5947 68020/68881 or a 68030 or a 68040. The generated code does use the
5948 68881 instructions that are emulated on the 68040.
5952 Generate output for a 68060, without using any of the new instructions.
5953 This results in code which can run relatively efficiently on either a
5954 68020/68881 or a 68030 or a 68040. The generated code does use the
5955 68881 instructions that are emulated on the 68060.
5958 @opindex msoft-float
5959 Generate output containing library calls for floating point.
5960 @strong{Warning:} the requisite libraries are not available for all m68k
5961 targets. Normally the facilities of the machine's usual C compiler are
5962 used, but this can't be done directly in cross-compilation. You must
5963 make your own arrangements to provide suitable library functions for
5964 cross-compilation. The embedded targets @samp{m68k-*-aout} and
5965 @samp{m68k-*-coff} do provide software floating point support.
5969 Consider type @code{int} to be 16 bits wide, like @code{short int}.
5972 @opindex mnobitfield
5973 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
5974 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
5978 Do use the bit-field instructions. The @option{-m68020} option implies
5979 @option{-mbitfield}. This is the default if you use a configuration
5980 designed for a 68020.
5984 Use a different function-calling convention, in which functions
5985 that take a fixed number of arguments return with the @code{rtd}
5986 instruction, which pops their arguments while returning. This
5987 saves one instruction in the caller since there is no need to pop
5988 the arguments there.
5990 This calling convention is incompatible with the one normally
5991 used on Unix, so you cannot use it if you need to call libraries
5992 compiled with the Unix compiler.
5994 Also, you must provide function prototypes for all functions that
5995 take variable numbers of arguments (including @code{printf});
5996 otherwise incorrect code will be generated for calls to those
5999 In addition, seriously incorrect code will result if you call a
6000 function with too many arguments. (Normally, extra arguments are
6001 harmlessly ignored.)
6003 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
6004 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
6007 @itemx -mno-align-int
6009 @opindex mno-align-int
6010 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
6011 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
6012 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
6013 Aligning variables on 32-bit boundaries produces code that runs somewhat
6014 faster on processors with 32-bit busses at the expense of more memory.
6016 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
6017 align structures containing the above types differently than
6018 most published application binary interface specifications for the m68k.
6022 Use the pc-relative addressing mode of the 68000 directly, instead of
6023 using a global offset table. At present, this option implies @option{-fpic},
6024 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
6025 not presently supported with @option{-mpcrel}, though this could be supported for
6026 68020 and higher processors.
6028 @item -mno-strict-align
6029 @itemx -mstrict-align
6030 @opindex mno-strict-align
6031 @opindex mstrict-align
6032 Do not (do) assume that unaligned memory references will be handled by
6036 Generate code that allows the data segment to be located in a different
6037 area of memory from the text segment. This allows for execute in place in
6038 an environment without virtual memory management. This option implies -fPIC.
6041 Generate code that assumes that the data segment follows the text segment.
6042 This is the default.
6044 @item -mid-shared-library
6045 Generate code that supports shared libraries via the library ID method.
6046 This allows for execute in place and shared libraries in an environment
6047 without virtual memory management. This option implies -fPIC.
6049 @item -mno-id-shared-library
6050 Generate code that doesn't assume ID based shared libraries are being used.
6051 This is the default.
6053 @item -mshared-library-id=n
6054 Specified the identification number of the ID based shared library being
6055 compiled. Specifying a value of 0 will generate more compact code, specifying
6056 other values will force the allocation of that number to the current
6057 library but is no more space or time efficient than omitting this option.
6061 @node M68hc1x Options
6062 @subsection M68hc1x Options
6063 @cindex M68hc1x options
6065 These are the @samp{-m} options defined for the 68hc11 and 68hc12
6066 microcontrollers. The default values for these options depends on
6067 which style of microcontroller was selected when the compiler was configured;
6068 the defaults for the most common choices are given below.
6075 Generate output for a 68HC11. This is the default
6076 when the compiler is configured for 68HC11-based systems.
6082 Generate output for a 68HC12. This is the default
6083 when the compiler is configured for 68HC12-based systems.
6089 Generate output for a 68HCS12.
6092 @opindex mauto-incdec
6093 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
6100 Enable the use of 68HC12 min and max instructions.
6103 @itemx -mno-long-calls
6104 @opindex mlong-calls
6105 @opindex mno-long-calls
6106 Treat all calls as being far away (near). If calls are assumed to be
6107 far away, the compiler will use the @code{call} instruction to
6108 call a function and the @code{rtc} instruction for returning.
6112 Consider type @code{int} to be 16 bits wide, like @code{short int}.
6114 @item -msoft-reg-count=@var{count}
6115 @opindex msoft-reg-count
6116 Specify the number of pseudo-soft registers which are used for the
6117 code generation. The maximum number is 32. Using more pseudo-soft
6118 register may or may not result in better code depending on the program.
6119 The default is 4 for 68HC11 and 2 for 68HC12.
6124 @subsection VAX Options
6127 These @samp{-m} options are defined for the VAX:
6132 Do not output certain jump instructions (@code{aobleq} and so on)
6133 that the Unix assembler for the VAX cannot handle across long
6138 Do output those jump instructions, on the assumption that you
6139 will assemble with the GNU assembler.
6143 Output code for g-format floating point numbers instead of d-format.
6147 @subsection SPARC Options
6148 @cindex SPARC options
6150 These @samp{-m} switches are supported on the SPARC:
6155 @opindex mno-app-regs
6157 Specify @option{-mapp-regs} to generate output using the global registers
6158 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
6161 To be fully SVR4 ABI compliant at the cost of some performance loss,
6162 specify @option{-mno-app-regs}. You should compile libraries and system
6163 software with this option.
6168 @opindex mhard-float
6169 Generate output containing floating point instructions. This is the
6175 @opindex msoft-float
6176 Generate output containing library calls for floating point.
6177 @strong{Warning:} the requisite libraries are not available for all SPARC
6178 targets. Normally the facilities of the machine's usual C compiler are
6179 used, but this cannot be done directly in cross-compilation. You must make
6180 your own arrangements to provide suitable library functions for
6181 cross-compilation. The embedded targets @samp{sparc-*-aout} and
6182 @samp{sparclite-*-*} do provide software floating point support.
6184 @option{-msoft-float} changes the calling convention in the output file;
6185 therefore, it is only useful if you compile @emph{all} of a program with
6186 this option. In particular, you need to compile @file{libgcc.a}, the
6187 library that comes with GCC, with @option{-msoft-float} in order for
6190 @item -mhard-quad-float
6191 @opindex mhard-quad-float
6192 Generate output containing quad-word (long double) floating point
6195 @item -msoft-quad-float
6196 @opindex msoft-quad-float
6197 Generate output containing library calls for quad-word (long double)
6198 floating point instructions. The functions called are those specified
6199 in the SPARC ABI@. This is the default.
6201 As of this writing, there are no sparc implementations that have hardware
6202 support for the quad-word floating point instructions. They all invoke
6203 a trap handler for one of these instructions, and then the trap handler
6204 emulates the effect of the instruction. Because of the trap handler overhead,
6205 this is much slower than calling the ABI library routines. Thus the
6206 @option{-msoft-quad-float} option is the default.
6212 With @option{-mflat}, the compiler does not generate save/restore instructions
6213 and will use a ``flat'' or single register window calling convention.
6214 This model uses %i7 as the frame pointer and is compatible with the normal
6215 register window model. Code from either may be intermixed.
6216 The local registers and the input registers (0--5) are still treated as
6217 ``call saved'' registers and will be saved on the stack as necessary.
6219 With @option{-mno-flat} (the default), the compiler emits save/restore
6220 instructions (except for leaf functions) and is the normal mode of operation.
6222 @item -mno-unaligned-doubles
6223 @itemx -munaligned-doubles
6224 @opindex mno-unaligned-doubles
6225 @opindex munaligned-doubles
6226 Assume that doubles have 8 byte alignment. This is the default.
6228 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
6229 alignment only if they are contained in another type, or if they have an
6230 absolute address. Otherwise, it assumes they have 4 byte alignment.
6231 Specifying this option avoids some rare compatibility problems with code
6232 generated by other compilers. It is not the default because it results
6233 in a performance loss, especially for floating point code.
6235 @item -mno-faster-structs
6236 @itemx -mfaster-structs
6237 @opindex mno-faster-structs
6238 @opindex mfaster-structs
6239 With @option{-mfaster-structs}, the compiler assumes that structures
6240 should have 8 byte alignment. This enables the use of pairs of
6241 @code{ldd} and @code{std} instructions for copies in structure
6242 assignment, in place of twice as many @code{ld} and @code{st} pairs.
6243 However, the use of this changed alignment directly violates the SPARC
6244 ABI@. Thus, it's intended only for use on targets where the developer
6245 acknowledges that their resulting code will not be directly in line with
6246 the rules of the ABI@.
6249 @opindex mimpure-text
6250 @option{-mimpure-text}, used in addition to @option{-shared}, tells
6251 the compiler to not pass @option{-z text} to the linker when linking a
6252 shared object. Using this option, you can link position-dependent
6253 code into a shared object.
6255 @option{-mimpure-text} suppresses the ``relocations remain against
6256 allocatable but non-writable sections'' linker error message.
6257 However, the necessary relocations will trigger copy-on-write, and the
6258 shared object is not actually shared across processes. Instead of
6259 using @option{-mimpure-text}, you should compile all source code with
6260 @option{-fpic} or @option{-fPIC}.
6262 This option is only available on SunOS and Solaris.
6268 These two options select variations on the SPARC architecture.
6270 By default (unless specifically configured for the Fujitsu SPARClite),
6271 GCC generates code for the v7 variant of the SPARC architecture.
6273 @option{-mv8} will give you SPARC v8 code. The only difference from v7
6274 code is that the compiler emits the integer multiply and integer
6275 divide instructions which exist in SPARC v8 but not in SPARC v7.
6277 @option{-msparclite} will give you SPARClite code. This adds the integer
6278 multiply, integer divide step and scan (@code{ffs}) instructions which
6279 exist in SPARClite but not in SPARC v7.
6281 These options are deprecated and will be deleted in a future GCC release.
6282 They have been replaced with @option{-mcpu=xxx}.
6287 @opindex msupersparc
6288 These two options select the processor for which the code is optimized.
6290 With @option{-mcypress} (the default), the compiler optimizes code for the
6291 Cypress CY7C602 chip, as used in the SPARCStation/SPARCServer 3xx series.
6292 This is also appropriate for the older SPARCStation 1, 2, IPX etc.
6294 With @option{-msupersparc} the compiler optimizes code for the SuperSPARC cpu, as
6295 used in the SPARCStation 10, 1000 and 2000 series. This flag also enables use
6296 of the full SPARC v8 instruction set.
6298 These options are deprecated and will be deleted in a future GCC release.
6299 They have been replaced with @option{-mcpu=xxx}.
6301 @item -mcpu=@var{cpu_type}
6303 Set the instruction set, register set, and instruction scheduling parameters
6304 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
6305 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
6306 @samp{hypersparc}, @samp{sparclite86x}, @samp{f930}, @samp{f934},
6307 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc}, and
6310 Default instruction scheduling parameters are used for values that select
6311 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
6312 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
6314 Here is a list of each supported architecture and their supported
6319 v8: supersparc, hypersparc
6320 sparclite: f930, f934, sparclite86x
6322 v9: ultrasparc, ultrasparc3
6325 @item -mtune=@var{cpu_type}
6327 Set the instruction scheduling parameters for machine type
6328 @var{cpu_type}, but do not set the instruction set or register set that the
6329 option @option{-mcpu=@var{cpu_type}} would.
6331 The same values for @option{-mcpu=@var{cpu_type}} can be used for
6332 @option{-mtune=@var{cpu_type}}, but the only useful values are those
6333 that select a particular cpu implementation. Those are @samp{cypress},
6334 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
6335 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc}, and
6340 These @samp{-m} switches are supported in addition to the above
6341 on the SPARCLET processor.
6344 @item -mlittle-endian
6345 @opindex mlittle-endian
6346 Generate code for a processor running in little-endian mode.
6350 Treat register @code{%g0} as a normal register.
6351 GCC will continue to clobber it as necessary but will not assume
6352 it always reads as 0.
6354 @item -mbroken-saverestore
6355 @opindex mbroken-saverestore
6356 Generate code that does not use non-trivial forms of the @code{save} and
6357 @code{restore} instructions. Early versions of the SPARCLET processor do
6358 not correctly handle @code{save} and @code{restore} instructions used with
6359 arguments. They correctly handle them used without arguments. A @code{save}
6360 instruction used without arguments increments the current window pointer
6361 but does not allocate a new stack frame. It is assumed that the window
6362 overflow trap handler will properly handle this case as will interrupt
6366 These @samp{-m} switches are supported in addition to the above
6367 on SPARC V9 processors in 64-bit environments.
6370 @item -mlittle-endian
6371 @opindex mlittle-endian
6372 Generate code for a processor running in little-endian mode. It is only
6373 available for a few configurations and most notably not on Solaris.
6379 Generate code for a 32-bit or 64-bit environment.
6380 The 32-bit environment sets int, long and pointer to 32 bits.
6381 The 64-bit environment sets int to 32 bits and long and pointer
6384 @item -mcmodel=medlow
6385 @opindex mcmodel=medlow
6386 Generate code for the Medium/Low code model: the program must be linked
6387 in the low 32 bits of the address space. Pointers are 64 bits.
6388 Programs can be statically or dynamically linked.
6390 @item -mcmodel=medmid
6391 @opindex mcmodel=medmid
6392 Generate code for the Medium/Middle code model: the program must be linked
6393 in the low 44 bits of the address space, the text segment must be less than
6394 2G bytes, and data segment must be within 2G of the text segment.
6395 Pointers are 64 bits.
6397 @item -mcmodel=medany
6398 @opindex mcmodel=medany
6399 Generate code for the Medium/Anywhere code model: the program may be linked
6400 anywhere in the address space, the text segment must be less than
6401 2G bytes, and data segment must be within 2G of the text segment.
6402 Pointers are 64 bits.
6404 @item -mcmodel=embmedany
6405 @opindex mcmodel=embmedany
6406 Generate code for the Medium/Anywhere code model for embedded systems:
6407 assume a 32-bit text and a 32-bit data segment, both starting anywhere
6408 (determined at link time). Register %g4 points to the base of the
6409 data segment. Pointers are still 64 bits.
6410 Programs are statically linked, PIC is not supported.
6413 @itemx -mno-stack-bias
6414 @opindex mstack-bias
6415 @opindex mno-stack-bias
6416 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
6417 frame pointer if present, are offset by @minus{}2047 which must be added back
6418 when making stack frame references.
6419 Otherwise, assume no such offset is present.
6423 @subsection ARM Options
6426 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
6431 @opindex mapcs-frame
6432 Generate a stack frame that is compliant with the ARM Procedure Call
6433 Standard for all functions, even if this is not strictly necessary for
6434 correct execution of the code. Specifying @option{-fomit-frame-pointer}
6435 with this option will cause the stack frames not to be generated for
6436 leaf functions. The default is @option{-mno-apcs-frame}.
6440 This is a synonym for @option{-mapcs-frame}.
6444 Generate code for a processor running with a 26-bit program counter,
6445 and conforming to the function calling standards for the APCS 26-bit
6446 option. This option replaces the @option{-m2} and @option{-m3} options
6447 of previous releases of the compiler.
6451 Generate code for a processor running with a 32-bit program counter,
6452 and conforming to the function calling standards for the APCS 32-bit
6453 option. This option replaces the @option{-m6} option of previous releases
6457 @c not currently implemented
6458 @item -mapcs-stack-check
6459 @opindex mapcs-stack-check
6460 Generate code to check the amount of stack space available upon entry to
6461 every function (that actually uses some stack space). If there is
6462 insufficient space available then either the function
6463 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
6464 called, depending upon the amount of stack space required. The run time
6465 system is required to provide these functions. The default is
6466 @option{-mno-apcs-stack-check}, since this produces smaller code.
6468 @c not currently implemented
6470 @opindex mapcs-float
6471 Pass floating point arguments using the float point registers. This is
6472 one of the variants of the APCS@. This option is recommended if the
6473 target hardware has a floating point unit or if a lot of floating point
6474 arithmetic is going to be performed by the code. The default is
6475 @option{-mno-apcs-float}, since integer only code is slightly increased in
6476 size if @option{-mapcs-float} is used.
6478 @c not currently implemented
6479 @item -mapcs-reentrant
6480 @opindex mapcs-reentrant
6481 Generate reentrant, position independent code. The default is
6482 @option{-mno-apcs-reentrant}.
6485 @item -mthumb-interwork
6486 @opindex mthumb-interwork
6487 Generate code which supports calling between the ARM and Thumb
6488 instruction sets. Without this option the two instruction sets cannot
6489 be reliably used inside one program. The default is
6490 @option{-mno-thumb-interwork}, since slightly larger code is generated
6491 when @option{-mthumb-interwork} is specified.
6493 @item -mno-sched-prolog
6494 @opindex mno-sched-prolog
6495 Prevent the reordering of instructions in the function prolog, or the
6496 merging of those instruction with the instructions in the function's
6497 body. This means that all functions will start with a recognizable set
6498 of instructions (or in fact one of a choice from a small set of
6499 different function prologues), and this information can be used to
6500 locate the start if functions inside an executable piece of code. The
6501 default is @option{-msched-prolog}.
6504 @opindex mhard-float
6505 Generate output containing floating point instructions. This is the
6509 @opindex msoft-float
6510 Generate output containing library calls for floating point.
6511 @strong{Warning:} the requisite libraries are not available for all ARM
6512 targets. Normally the facilities of the machine's usual C compiler are
6513 used, but this cannot be done directly in cross-compilation. You must make
6514 your own arrangements to provide suitable library functions for
6517 @option{-msoft-float} changes the calling convention in the output file;
6518 therefore, it is only useful if you compile @emph{all} of a program with
6519 this option. In particular, you need to compile @file{libgcc.a}, the
6520 library that comes with GCC, with @option{-msoft-float} in order for
6523 @item -mlittle-endian
6524 @opindex mlittle-endian
6525 Generate code for a processor running in little-endian mode. This is
6526 the default for all standard configurations.
6529 @opindex mbig-endian
6530 Generate code for a processor running in big-endian mode; the default is
6531 to compile code for a little-endian processor.
6533 @item -mwords-little-endian
6534 @opindex mwords-little-endian
6535 This option only applies when generating code for big-endian processors.
6536 Generate code for a little-endian word order but a big-endian byte
6537 order. That is, a byte order of the form @samp{32107654}. Note: this
6538 option should only be used if you require compatibility with code for
6539 big-endian ARM processors generated by versions of the compiler prior to
6542 @item -malignment-traps
6543 @opindex malignment-traps
6544 Generate code that will not trap if the MMU has alignment traps enabled.
6545 On ARM architectures prior to ARMv4, there were no instructions to
6546 access half-word objects stored in memory. However, when reading from
6547 memory a feature of the ARM architecture allows a word load to be used,
6548 even if the address is unaligned, and the processor core will rotate the
6549 data as it is being loaded. This option tells the compiler that such
6550 misaligned accesses will cause a MMU trap and that it should instead
6551 synthesize the access as a series of byte accesses. The compiler can
6552 still use word accesses to load half-word data if it knows that the
6553 address is aligned to a word boundary.
6555 This option is ignored when compiling for ARM architecture 4 or later,
6556 since these processors have instructions to directly access half-word
6559 @item -mno-alignment-traps
6560 @opindex mno-alignment-traps
6561 Generate code that assumes that the MMU will not trap unaligned
6562 accesses. This produces better code when the target instruction set
6563 does not have half-word memory operations (i.e.@: implementations prior to
6566 Note that you cannot use this option to access unaligned word objects,
6567 since the processor will only fetch one 32-bit aligned object from
6570 The default setting for most targets is @option{-mno-alignment-traps}, since
6571 this produces better code when there are no half-word memory
6572 instructions available.
6574 @item -mshort-load-bytes
6575 @itemx -mno-short-load-words
6576 @opindex mshort-load-bytes
6577 @opindex mno-short-load-words
6578 These are deprecated aliases for @option{-malignment-traps}.
6580 @item -mno-short-load-bytes
6581 @itemx -mshort-load-words
6582 @opindex mno-short-load-bytes
6583 @opindex mshort-load-words
6584 This are deprecated aliases for @option{-mno-alignment-traps}.
6586 @item -mcpu=@var{name}
6588 This specifies the name of the target ARM processor. GCC uses this name
6589 to determine what kind of instructions it can emit when generating
6590 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
6591 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
6592 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
6593 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
6594 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
6595 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm8},
6596 @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
6597 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
6598 @samp{arm920t}, @samp{arm926ejs}, @samp{arm940t}, @samp{arm9tdmi},
6599 @samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ejs},
6600 @samp{arm1136js}, @samp{arm1136jfs} ,@samp{xscale}, @samp{iwmmxt},
6603 @itemx -mtune=@var{name}
6605 This option is very similar to the @option{-mcpu=} option, except that
6606 instead of specifying the actual target processor type, and hence
6607 restricting which instructions can be used, it specifies that GCC should
6608 tune the performance of the code as if the target were of the type
6609 specified in this option, but still choosing the instructions that it
6610 will generate based on the cpu specified by a @option{-mcpu=} option.
6611 For some ARM implementations better performance can be obtained by using
6614 @item -march=@var{name}
6616 This specifies the name of the target ARM architecture. GCC uses this
6617 name to determine what kind of instructions it can emit when generating
6618 assembly code. This option can be used in conjunction with or instead
6619 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
6620 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
6621 @samp{armv5}, @samp{armv5t}, @samp{armv5te}, @samp{armv6j},
6622 @samp{iwmmxt}, @samp{ep9312}.
6624 @item -mfpe=@var{number}
6625 @itemx -mfp=@var{number}
6628 This specifies the version of the floating point emulation available on
6629 the target. Permissible values are 2 and 3. @option{-mfp=} is a synonym
6630 for @option{-mfpe=}, for compatibility with older versions of GCC@.
6632 @item -mstructure-size-boundary=@var{n}
6633 @opindex mstructure-size-boundary
6634 The size of all structures and unions will be rounded up to a multiple
6635 of the number of bits set by this option. Permissible values are 8 and
6636 32. The default value varies for different toolchains. For the COFF
6637 targeted toolchain the default value is 8. Specifying the larger number
6638 can produce faster, more efficient code, but can also increase the size
6639 of the program. The two values are potentially incompatible. Code
6640 compiled with one value cannot necessarily expect to work with code or
6641 libraries compiled with the other value, if they exchange information
6642 using structures or unions.
6644 @item -mabort-on-noreturn
6645 @opindex mabort-on-noreturn
6646 Generate a call to the function @code{abort} at the end of a
6647 @code{noreturn} function. It will be executed if the function tries to
6651 @itemx -mno-long-calls
6652 @opindex mlong-calls
6653 @opindex mno-long-calls
6654 Tells the compiler to perform function calls by first loading the
6655 address of the function into a register and then performing a subroutine
6656 call on this register. This switch is needed if the target function
6657 will lie outside of the 64 megabyte addressing range of the offset based
6658 version of subroutine call instruction.
6660 Even if this switch is enabled, not all function calls will be turned
6661 into long calls. The heuristic is that static functions, functions
6662 which have the @samp{short-call} attribute, functions that are inside
6663 the scope of a @samp{#pragma no_long_calls} directive and functions whose
6664 definitions have already been compiled within the current compilation
6665 unit, will not be turned into long calls. The exception to this rule is
6666 that weak function definitions, functions with the @samp{long-call}
6667 attribute or the @samp{section} attribute, and functions that are within
6668 the scope of a @samp{#pragma long_calls} directive, will always be
6669 turned into long calls.
6671 This feature is not enabled by default. Specifying
6672 @option{-mno-long-calls} will restore the default behavior, as will
6673 placing the function calls within the scope of a @samp{#pragma
6674 long_calls_off} directive. Note these switches have no effect on how
6675 the compiler generates code to handle function calls via function
6678 @item -mnop-fun-dllimport
6679 @opindex mnop-fun-dllimport
6680 Disable support for the @code{dllimport} attribute.
6682 @item -msingle-pic-base
6683 @opindex msingle-pic-base
6684 Treat the register used for PIC addressing as read-only, rather than
6685 loading it in the prologue for each function. The run-time system is
6686 responsible for initializing this register with an appropriate value
6687 before execution begins.
6689 @item -mpic-register=@var{reg}
6690 @opindex mpic-register
6691 Specify the register to be used for PIC addressing. The default is R10
6692 unless stack-checking is enabled, when R9 is used.
6694 @item -mcirrus-fix-invalid-insns
6695 @opindex mcirrus-fix-invalid-insns
6696 @opindex mno-cirrus-fix-invalid-insns
6697 Insert NOPs into the instruction stream to in order to work around
6698 problems with invalid Maverick instruction combinations. This option
6699 is only valid if the @option{-mcpu=ep9312} option has been used to
6700 enable generation of instructions for the Cirrus Maverick floating
6701 point co-processor. This option is not enabled by default, since the
6702 problem is only present in older Maverick implementations. The default
6703 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
6706 @item -mpoke-function-name
6707 @opindex mpoke-function-name
6708 Write the name of each function into the text section, directly
6709 preceding the function prologue. The generated code is similar to this:
6713 .ascii "arm_poke_function_name", 0
6716 .word 0xff000000 + (t1 - t0)
6717 arm_poke_function_name
6719 stmfd sp!, @{fp, ip, lr, pc@}
6723 When performing a stack backtrace, code can inspect the value of
6724 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
6725 location @code{pc - 12} and the top 8 bits are set, then we know that
6726 there is a function name embedded immediately preceding this location
6727 and has length @code{((pc[-3]) & 0xff000000)}.
6731 Generate code for the 16-bit Thumb instruction set. The default is to
6732 use the 32-bit ARM instruction set.
6735 @opindex mtpcs-frame
6736 Generate a stack frame that is compliant with the Thumb Procedure Call
6737 Standard for all non-leaf functions. (A leaf function is one that does
6738 not call any other functions.) The default is @option{-mno-tpcs-frame}.
6740 @item -mtpcs-leaf-frame
6741 @opindex mtpcs-leaf-frame
6742 Generate a stack frame that is compliant with the Thumb Procedure Call
6743 Standard for all leaf functions. (A leaf function is one that does
6744 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
6746 @item -mcallee-super-interworking
6747 @opindex mcallee-super-interworking
6748 Gives all externally visible functions in the file being compiled an ARM
6749 instruction set header which switches to Thumb mode before executing the
6750 rest of the function. This allows these functions to be called from
6751 non-interworking code.
6753 @item -mcaller-super-interworking
6754 @opindex mcaller-super-interworking
6755 Allows calls via function pointers (including virtual functions) to
6756 execute correctly regardless of whether the target code has been
6757 compiled for interworking or not. There is a small overhead in the cost
6758 of executing a function pointer if this option is enabled.
6762 @node MN10200 Options
6763 @subsection MN10200 Options
6764 @cindex MN10200 options
6766 These @option{-m} options are defined for Matsushita MN10200 architectures:
6771 Indicate to the linker that it should perform a relaxation optimization pass
6772 to shorten branches, calls and absolute memory addresses. This option only
6773 has an effect when used on the command line for the final link step.
6775 This option makes symbolic debugging impossible.
6778 @node MN10300 Options
6779 @subsection MN10300 Options
6780 @cindex MN10300 options
6782 These @option{-m} options are defined for Matsushita MN10300 architectures:
6787 Generate code to avoid bugs in the multiply instructions for the MN10300
6788 processors. This is the default.
6791 @opindex mno-mult-bug
6792 Do not generate code to avoid bugs in the multiply instructions for the
6797 Generate code which uses features specific to the AM33 processor.
6801 Do not generate code which uses features specific to the AM33 processor. This
6806 Do not link in the C run-time initialization object file.
6810 Indicate to the linker that it should perform a relaxation optimization pass
6811 to shorten branches, calls and absolute memory addresses. This option only
6812 has an effect when used on the command line for the final link step.
6814 This option makes symbolic debugging impossible.
6818 @node M32R/D Options
6819 @subsection M32R/D Options
6820 @cindex M32R/D options
6822 These @option{-m} options are defined for Renesas M32R/D architectures:
6827 Generate code for the M32R/2@.
6831 Generate code for the M32R/X@.
6835 Generate code for the M32R@. This is the default.
6837 @item -mcode-model=small
6838 @opindex mcode-model=small
6839 Assume all objects live in the lower 16MB of memory (so that their addresses
6840 can be loaded with the @code{ld24} instruction), and assume all subroutines
6841 are reachable with the @code{bl} instruction.
6842 This is the default.
6844 The addressability of a particular object can be set with the
6845 @code{model} attribute.
6847 @item -mcode-model=medium
6848 @opindex mcode-model=medium
6849 Assume objects may be anywhere in the 32-bit address space (the compiler
6850 will generate @code{seth/add3} instructions to load their addresses), and
6851 assume all subroutines are reachable with the @code{bl} instruction.
6853 @item -mcode-model=large
6854 @opindex mcode-model=large
6855 Assume objects may be anywhere in the 32-bit address space (the compiler
6856 will generate @code{seth/add3} instructions to load their addresses), and
6857 assume subroutines may not be reachable with the @code{bl} instruction
6858 (the compiler will generate the much slower @code{seth/add3/jl}
6859 instruction sequence).
6862 @opindex msdata=none
6863 Disable use of the small data area. Variables will be put into
6864 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
6865 @code{section} attribute has been specified).
6866 This is the default.
6868 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
6869 Objects may be explicitly put in the small data area with the
6870 @code{section} attribute using one of these sections.
6873 @opindex msdata=sdata
6874 Put small global and static data in the small data area, but do not
6875 generate special code to reference them.
6879 Put small global and static data in the small data area, and generate
6880 special instructions to reference them.
6884 @cindex smaller data references
6885 Put global and static objects less than or equal to @var{num} bytes
6886 into the small data or bss sections instead of the normal data or bss
6887 sections. The default value of @var{num} is 8.
6888 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
6889 for this option to have any effect.
6891 All modules should be compiled with the same @option{-G @var{num}} value.
6892 Compiling with different values of @var{num} may or may not work; if it
6893 doesn't the linker will give an error message---incorrect code will not be
6899 @subsection M88K Options
6900 @cindex M88k options
6902 These @samp{-m} options are defined for Motorola 88k architectures:
6907 Generate code that works well on both the m88100 and the
6912 Generate code that works best for the m88100, but that also
6917 Generate code that works best for the m88110, and may not run
6922 Obsolete option to be removed from the next revision.
6925 @item -midentify-revision
6926 @opindex midentify-revision
6927 @cindex identifying source, compiler (88k)
6928 Include an @code{ident} directive in the assembler output recording the
6929 source file name, compiler name and version, timestamp, and compilation
6932 @item -mno-underscores
6933 @opindex mno-underscores
6934 @cindex underscores, avoiding (88k)
6935 In assembler output, emit symbol names without adding an underscore
6936 character at the beginning of each name. The default is to use an
6937 underscore as prefix on each name.
6939 @item -mocs-debug-info
6940 @itemx -mno-ocs-debug-info
6941 @opindex mocs-debug-info
6942 @opindex mno-ocs-debug-info
6944 @cindex debugging, 88k OCS
6945 Include (or omit) additional debugging information (about registers used
6946 in each stack frame) as specified in the 88open Object Compatibility
6947 Standard, ``OCS''@. This extra information allows debugging of code that
6948 has had the frame pointer eliminated. The default for SVr4 and Delta 88
6949 SVr3.2 is to include this information; other 88k configurations omit this
6950 information by default.
6952 @item -mocs-frame-position
6953 @opindex mocs-frame-position
6954 @cindex register positions in frame (88k)
6955 When emitting COFF debugging information for automatic variables and
6956 parameters stored on the stack, use the offset from the canonical frame
6957 address, which is the stack pointer (register 31) on entry to the
6958 function. The SVr4 and Delta88 SVr3.2, and BCS configurations use
6959 @option{-mocs-frame-position}; other 88k configurations have the default
6960 @option{-mno-ocs-frame-position}.
6962 @item -mno-ocs-frame-position
6963 @opindex mno-ocs-frame-position
6964 @cindex register positions in frame (88k)
6965 When emitting COFF debugging information for automatic variables and
6966 parameters stored on the stack, use the offset from the frame pointer
6967 register (register 30). When this option is in effect, the frame
6968 pointer is not eliminated when debugging information is selected by the
6971 @item -moptimize-arg-area
6972 @opindex moptimize-arg-area
6973 @cindex arguments in frame (88k)
6974 Save space by reorganizing the stack frame. This option generates code
6975 that does not agree with the 88open specifications, but uses less
6978 @itemx -mno-optimize-arg-area
6979 @opindex mno-optimize-arg-area
6980 Do not reorganize the stack frame to save space. This is the default.
6981 The generated conforms to the specification, but uses more memory.
6983 @item -mshort-data-@var{num}
6984 @opindex mshort-data
6985 @cindex smaller data references (88k)
6986 @cindex r0-relative references (88k)
6987 Generate smaller data references by making them relative to @code{r0},
6988 which allows loading a value using a single instruction (rather than the
6989 usual two). You control which data references are affected by
6990 specifying @var{num} with this option. For example, if you specify
6991 @option{-mshort-data-512}, then the data references affected are those
6992 involving displacements of less than 512 bytes.
6993 @option{-mshort-data-@var{num}} is not effective for @var{num} greater
6996 @item -mserialize-volatile
6997 @opindex mserialize-volatile
6998 @itemx -mno-serialize-volatile
6999 @opindex mno-serialize-volatile
7000 @cindex sequential consistency on 88k
7001 Do, or don't, generate code to guarantee sequential consistency
7002 of volatile memory references. By default, consistency is
7005 The order of memory references made by the MC88110 processor does
7006 not always match the order of the instructions requesting those
7007 references. In particular, a load instruction may execute before
7008 a preceding store instruction. Such reordering violates
7009 sequential consistency of volatile memory references, when there
7010 are multiple processors. When consistency must be guaranteed,
7011 GCC generates special instructions, as needed, to force
7012 execution in the proper order.
7014 The MC88100 processor does not reorder memory references and so
7015 always provides sequential consistency. However, by default, GCC
7016 generates the special instructions to guarantee consistency
7017 even when you use @option{-m88100}, so that the code may be run on an
7018 MC88110 processor. If you intend to run your code only on the
7019 MC88100 processor, you may use @option{-mno-serialize-volatile}.
7021 The extra code generated to guarantee consistency may affect the
7022 performance of your application. If you know that you can safely
7023 forgo this guarantee, you may use @option{-mno-serialize-volatile}.
7029 @cindex assembler syntax, 88k
7031 Turn on (@option{-msvr4}) or off (@option{-msvr3}) compiler extensions
7032 related to System V release 4 (SVr4). This controls the following:
7036 Which variant of the assembler syntax to emit.
7038 @option{-msvr4} makes the C preprocessor recognize @samp{#pragma weak}
7039 that is used on System V release 4.
7041 @option{-msvr4} makes GCC issue additional declaration directives used in
7045 @option{-msvr4} is the default for the m88k-motorola-sysv4 configuration.
7046 @option{-msvr3} is the default for all other m88k configurations.
7048 @item -mversion-03.00
7049 @opindex mversion-03.00
7050 This option is obsolete, and is ignored.
7051 @c ??? which asm syntax better for GAS? option there too?
7053 @item -mno-check-zero-division
7054 @itemx -mcheck-zero-division
7055 @opindex mno-check-zero-division
7056 @opindex mcheck-zero-division
7057 @cindex zero division on 88k
7058 Do, or don't, generate code to guarantee that integer division by
7059 zero will be detected. By default, detection is guaranteed.
7061 Some models of the MC88100 processor fail to trap upon integer
7062 division by zero under certain conditions. By default, when
7063 compiling code that might be run on such a processor, GCC
7064 generates code that explicitly checks for zero-valued divisors
7065 and traps with exception number 503 when one is detected. Use of
7066 @option{-mno-check-zero-division} suppresses such checking for code
7067 generated to run on an MC88100 processor.
7069 GCC assumes that the MC88110 processor correctly detects all instances
7070 of integer division by zero. When @option{-m88110} is specified, no
7071 explicit checks for zero-valued divisors are generated, and both
7072 @option{-mcheck-zero-division} and @option{-mno-check-zero-division} are
7075 @item -muse-div-instruction
7076 @opindex muse-div-instruction
7077 @cindex divide instruction, 88k
7078 Use the div instruction for signed integer division on the
7079 MC88100 processor. By default, the div instruction is not used.
7081 On the MC88100 processor the signed integer division instruction
7082 div) traps to the operating system on a negative operand. The
7083 operating system transparently completes the operation, but at a
7084 large cost in execution time. By default, when compiling code
7085 that might be run on an MC88100 processor, GCC emulates signed
7086 integer division using the unsigned integer division instruction
7087 divu), thereby avoiding the large penalty of a trap to the
7088 operating system. Such emulation has its own, smaller, execution
7089 cost in both time and space. To the extent that your code's
7090 important signed integer division operations are performed on two
7091 nonnegative operands, it may be desirable to use the div
7092 instruction directly.
7094 On the MC88110 processor the div instruction (also known as the
7095 divs instruction) processes negative operands without trapping to
7096 the operating system. When @option{-m88110} is specified,
7097 @option{-muse-div-instruction} is ignored, and the div instruction is used
7098 for signed integer division.
7100 Note that the result of dividing @code{INT_MIN} by @minus{}1 is undefined. In
7101 particular, the behavior of such a division with and without
7102 @option{-muse-div-instruction} may differ.
7104 @item -mtrap-large-shift
7105 @itemx -mhandle-large-shift
7106 @opindex mtrap-large-shift
7107 @opindex mhandle-large-shift
7108 @cindex bit shift overflow (88k)
7109 @cindex large bit shifts (88k)
7110 Include code to detect bit-shifts of more than 31 bits; respectively,
7111 trap such shifts or emit code to handle them properly. By default GCC
7112 makes no special provision for large bit shifts.
7114 @item -mwarn-passed-structs
7115 @opindex mwarn-passed-structs
7116 @cindex structure passing (88k)
7117 Warn when a function passes a struct as an argument or result.
7118 Structure-passing conventions have changed during the evolution of the C
7119 language, and are often the source of portability problems. By default,
7120 GCC issues no such warning.
7123 @c break page here to avoid unsightly interparagraph stretch.
7127 @node RS/6000 and PowerPC Options
7128 @subsection IBM RS/6000 and PowerPC Options
7129 @cindex RS/6000 and PowerPC Options
7130 @cindex IBM RS/6000 and PowerPC Options
7132 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
7140 @itemx -mpowerpc-gpopt
7141 @itemx -mno-powerpc-gpopt
7142 @itemx -mpowerpc-gfxopt
7143 @itemx -mno-powerpc-gfxopt
7145 @itemx -mno-powerpc64
7151 @opindex mno-powerpc
7152 @opindex mpowerpc-gpopt
7153 @opindex mno-powerpc-gpopt
7154 @opindex mpowerpc-gfxopt
7155 @opindex mno-powerpc-gfxopt
7157 @opindex mno-powerpc64
7158 GCC supports two related instruction set architectures for the
7159 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
7160 instructions supported by the @samp{rios} chip set used in the original
7161 RS/6000 systems and the @dfn{PowerPC} instruction set is the
7162 architecture of the Motorola MPC5xx, MPC6xx, MPC8xx microprocessors, and
7163 the IBM 4xx microprocessors.
7165 Neither architecture is a subset of the other. However there is a
7166 large common subset of instructions supported by both. An MQ
7167 register is included in processors supporting the POWER architecture.
7169 You use these options to specify which instructions are available on the
7170 processor you are using. The default value of these options is
7171 determined when configuring GCC@. Specifying the
7172 @option{-mcpu=@var{cpu_type}} overrides the specification of these
7173 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
7174 rather than the options listed above.
7176 The @option{-mpower} option allows GCC to generate instructions that
7177 are found only in the POWER architecture and to use the MQ register.
7178 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
7179 to generate instructions that are present in the POWER2 architecture but
7180 not the original POWER architecture.
7182 The @option{-mpowerpc} option allows GCC to generate instructions that
7183 are found only in the 32-bit subset of the PowerPC architecture.
7184 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
7185 GCC to use the optional PowerPC architecture instructions in the
7186 General Purpose group, including floating-point square root. Specifying
7187 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
7188 use the optional PowerPC architecture instructions in the Graphics
7189 group, including floating-point select.
7191 The @option{-mpowerpc64} option allows GCC to generate the additional
7192 64-bit instructions that are found in the full PowerPC64 architecture
7193 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
7194 @option{-mno-powerpc64}.
7196 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
7197 will use only the instructions in the common subset of both
7198 architectures plus some special AIX common-mode calls, and will not use
7199 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
7200 permits GCC to use any instruction from either architecture and to
7201 allow use of the MQ register; specify this for the Motorola MPC601.
7203 @item -mnew-mnemonics
7204 @itemx -mold-mnemonics
7205 @opindex mnew-mnemonics
7206 @opindex mold-mnemonics
7207 Select which mnemonics to use in the generated assembler code. With
7208 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
7209 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
7210 assembler mnemonics defined for the POWER architecture. Instructions
7211 defined in only one architecture have only one mnemonic; GCC uses that
7212 mnemonic irrespective of which of these options is specified.
7214 GCC defaults to the mnemonics appropriate for the architecture in
7215 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
7216 value of these option. Unless you are building a cross-compiler, you
7217 should normally not specify either @option{-mnew-mnemonics} or
7218 @option{-mold-mnemonics}, but should instead accept the default.
7220 @item -mcpu=@var{cpu_type}
7222 Set architecture type, register usage, choice of mnemonics, and
7223 instruction scheduling parameters for machine type @var{cpu_type}.
7224 Supported values for @var{cpu_type} are @samp{401}, @samp{403},
7225 @samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{505},
7226 @samp{601}, @samp{602}, @samp{603}, @samp{603e}, @samp{604},
7227 @samp{604e}, @samp{620}, @samp{630}, @samp{740}, @samp{7400},
7228 @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
7229 @samp{860}, @samp{970}, @samp{common}, @samp{ec603e}, @samp{G3},
7230 @samp{G4}, @samp{G5}, @samp{power}, @samp{power2}, @samp{power3},
7231 @samp{power4}, @samp{powerpc}, @samp{powerpc64}, @samp{rios},
7232 @samp{rios1}, @samp{rios2}, @samp{rsc}, and @samp{rs64a}.
7234 @option{-mcpu=common} selects a completely generic processor. Code
7235 generated under this option will run on any POWER or PowerPC processor.
7236 GCC will use only the instructions in the common subset of both
7237 architectures, and will not use the MQ register. GCC assumes a generic
7238 processor model for scheduling purposes.
7240 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
7241 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
7242 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
7243 types, with an appropriate, generic processor model assumed for
7244 scheduling purposes.
7246 The other options specify a specific processor. Code generated under
7247 those options will run best on that processor, and may not run at all on
7250 The @option{-mcpu} options automatically enable or disable the
7251 following options: @option{-maltivec}, @option{-mhard-float},
7252 @option{-mmfcrf}, @option{-mmultiple}, @option{-mnew-mnemonics},
7253 @option{-mpower}, @option{-mpower2}, @option{-mpowerpc64},
7254 @option{-mpowerpc-gpopt}, @option{-mpowerpc-gfxopt},
7255 @option{-mstring}. The particular options set for any particular CPU
7256 will vary between compiler versions, depending on what setting seems
7257 to produce optimal code for that CPU; it doesn't necessarily reflect
7258 the actual hardware's capabilities. If you wish to set an individual
7259 option to a particular value, you may specify it after the
7260 @option{-mcpu} option, like @samp{-mcpu=970 -mno-altivec}.
7262 On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
7263 not enabled or disabled by the @option{-mcpu} option at present, since
7264 AIX does not have full support for these options. You may still
7265 enable or disable them individually if you're sure it'll work in your
7268 @item -mtune=@var{cpu_type}
7270 Set the instruction scheduling parameters for machine type
7271 @var{cpu_type}, but do not set the architecture type, register usage, or
7272 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
7273 values for @var{cpu_type} are used for @option{-mtune} as for
7274 @option{-mcpu}. If both are specified, the code generated will use the
7275 architecture, registers, and mnemonics set by @option{-mcpu}, but the
7276 scheduling parameters set by @option{-mtune}.
7281 @opindex mno-altivec
7282 These switches enable or disable the use of built-in functions that
7283 allow access to the AltiVec instruction set. You may also need to set
7284 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
7289 Extend the current ABI with SPE ABI extensions. This does not change
7290 the default ABI, instead it adds the SPE ABI extensions to the current
7294 @opindex mabi=no-spe
7295 Disable Booke SPE ABI extensions for the current ABI.
7297 @item -misel=@var{yes/no}
7300 This switch enables or disables the generation of ISEL instructions.
7302 @item -mspe=@var{yes/no}
7305 This switch enables or disables the generation of SPE simd
7308 @item -mfloat-gprs=@var{yes/no}
7310 @opindex mfloat-gprs
7311 This switch enables or disables the generation of floating point
7312 operations on the general purpose registers for architectures that
7313 support it. This option is currently only available on the MPC8540.
7316 @itemx -mno-fp-in-toc
7317 @itemx -mno-sum-in-toc
7318 @itemx -mminimal-toc
7320 @opindex mno-fp-in-toc
7321 @opindex mno-sum-in-toc
7322 @opindex mminimal-toc
7323 Modify generation of the TOC (Table Of Contents), which is created for
7324 every executable file. The @option{-mfull-toc} option is selected by
7325 default. In that case, GCC will allocate at least one TOC entry for
7326 each unique non-automatic variable reference in your program. GCC
7327 will also place floating-point constants in the TOC@. However, only
7328 16,384 entries are available in the TOC@.
7330 If you receive a linker error message that saying you have overflowed
7331 the available TOC space, you can reduce the amount of TOC space used
7332 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
7333 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
7334 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
7335 generate code to calculate the sum of an address and a constant at
7336 run-time instead of putting that sum into the TOC@. You may specify one
7337 or both of these options. Each causes GCC to produce very slightly
7338 slower and larger code at the expense of conserving TOC space.
7340 If you still run out of space in the TOC even when you specify both of
7341 these options, specify @option{-mminimal-toc} instead. This option causes
7342 GCC to make only one TOC entry for every file. When you specify this
7343 option, GCC will produce code that is slower and larger but which
7344 uses extremely little TOC space. You may wish to use this option
7345 only on files that contain less frequently executed code.
7351 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
7352 @code{long} type, and the infrastructure needed to support them.
7353 Specifying @option{-maix64} implies @option{-mpowerpc64} and
7354 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
7355 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
7360 @opindex mno-xl-call
7361 On AIX, pass floating-point arguments to prototyped functions beyond the
7362 register save area (RSA) on the stack in addition to argument FPRs. The
7363 AIX calling convention was extended but not initially documented to
7364 handle an obscure K&R C case of calling a function that takes the
7365 address of its arguments with fewer arguments than declared. AIX XL
7366 compilers access floating point arguments which do not fit in the
7367 RSA from the stack when a subroutine is compiled without
7368 optimization. Because always storing floating-point arguments on the
7369 stack is inefficient and rarely needed, this option is not enabled by
7370 default and only is necessary when calling subroutines compiled by AIX
7371 XL compilers without optimization.
7375 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
7376 application written to use message passing with special startup code to
7377 enable the application to run. The system must have PE installed in the
7378 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
7379 must be overridden with the @option{-specs=} option to specify the
7380 appropriate directory location. The Parallel Environment does not
7381 support threads, so the @option{-mpe} option and the @option{-pthread}
7382 option are incompatible.
7384 @item -malign-natural
7385 @itemx -malign-power
7386 @opindex malign-natural
7387 @opindex malign-power
7388 On AIX, Darwin, and 64-bit PowerPC Linux, the option
7389 @option{-malign-natural} overrides the ABI-defined alignment of larger
7390 types, such as floating-point doubles, on their natural size-based boundary.
7391 The option @option{-malign-power} instructs GCC to follow the ABI-specified
7392 alignment rules. GCC defaults to the standard alignment defined in the ABI.
7396 @opindex msoft-float
7397 @opindex mhard-float
7398 Generate code that does not use (uses) the floating-point register set.
7399 Software floating point emulation is provided if you use the
7400 @option{-msoft-float} option, and pass the option to GCC when linking.
7403 @itemx -mno-multiple
7405 @opindex mno-multiple
7406 Generate code that uses (does not use) the load multiple word
7407 instructions and the store multiple word instructions. These
7408 instructions are generated by default on POWER systems, and not
7409 generated on PowerPC systems. Do not use @option{-mmultiple} on little
7410 endian PowerPC systems, since those instructions do not work when the
7411 processor is in little endian mode. The exceptions are PPC740 and
7412 PPC750 which permit the instructions usage in little endian mode.
7418 Generate code that uses (does not use) the load string instructions
7419 and the store string word instructions to save multiple registers and
7420 do small block moves. These instructions are generated by default on
7421 POWER systems, and not generated on PowerPC systems. Do not use
7422 @option{-mstring} on little endian PowerPC systems, since those
7423 instructions do not work when the processor is in little endian mode.
7424 The exceptions are PPC740 and PPC750 which permit the instructions
7425 usage in little endian mode.
7431 Generate code that uses (does not use) the load or store instructions
7432 that update the base register to the address of the calculated memory
7433 location. These instructions are generated by default. If you use
7434 @option{-mno-update}, there is a small window between the time that the
7435 stack pointer is updated and the address of the previous frame is
7436 stored, which means code that walks the stack frame across interrupts or
7437 signals may get corrupted data.
7440 @itemx -mno-fused-madd
7441 @opindex mfused-madd
7442 @opindex mno-fused-madd
7443 Generate code that uses (does not use) the floating point multiply and
7444 accumulate instructions. These instructions are generated by default if
7445 hardware floating is used.
7447 @item -mno-bit-align
7449 @opindex mno-bit-align
7451 On System V.4 and embedded PowerPC systems do not (do) force structures
7452 and unions that contain bit-fields to be aligned to the base type of the
7455 For example, by default a structure containing nothing but 8
7456 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
7457 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
7458 the structure would be aligned to a 1 byte boundary and be one byte in
7461 @item -mno-strict-align
7462 @itemx -mstrict-align
7463 @opindex mno-strict-align
7464 @opindex mstrict-align
7465 On System V.4 and embedded PowerPC systems do not (do) assume that
7466 unaligned memory references will be handled by the system.
7469 @itemx -mno-relocatable
7470 @opindex mrelocatable
7471 @opindex mno-relocatable
7472 On embedded PowerPC systems generate code that allows (does not allow)
7473 the program to be relocated to a different address at runtime. If you
7474 use @option{-mrelocatable} on any module, all objects linked together must
7475 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
7477 @item -mrelocatable-lib
7478 @itemx -mno-relocatable-lib
7479 @opindex mrelocatable-lib
7480 @opindex mno-relocatable-lib
7481 On embedded PowerPC systems generate code that allows (does not allow)
7482 the program to be relocated to a different address at runtime. Modules
7483 compiled with @option{-mrelocatable-lib} can be linked with either modules
7484 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
7485 with modules compiled with the @option{-mrelocatable} options.
7491 On System V.4 and embedded PowerPC systems do not (do) assume that
7492 register 2 contains a pointer to a global area pointing to the addresses
7493 used in the program.
7496 @itemx -mlittle-endian
7498 @opindex mlittle-endian
7499 On System V.4 and embedded PowerPC systems compile code for the
7500 processor in little endian mode. The @option{-mlittle-endian} option is
7501 the same as @option{-mlittle}.
7506 @opindex mbig-endian
7507 On System V.4 and embedded PowerPC systems compile code for the
7508 processor in big endian mode. The @option{-mbig-endian} option is
7509 the same as @option{-mbig}.
7511 @item -mdynamic-no-pic
7512 @opindex mdynamic-no-pic
7513 On Darwin and Mac OS X systems, compile code so that it is not
7514 relocatable, but that its external references are relocatable. The
7515 resulting code is suitable for applications, but not shared
7518 @item -mprioritize-restricted-insns=@var{priority}
7519 @opindex mprioritize-restricted-insns
7520 This option controls the priority that is assigned to
7521 dispatch-slot restricted instructions during the second scheduling
7522 pass. The argument @var{priority} takes the value @var{0/1/2} to assign
7523 @var{no/highest/second-highest} priority to dispatch slot restricted
7526 @item -msched-costly-dep=@var{dependence_type}
7527 @opindex msched-costly-dep
7528 This option controls which dependences are considered costly
7529 by the target during instruction scheduling. The argument
7530 @var{dependence_type} takes one of the following values:
7531 @var{no}: no dependence is costly,
7532 @var{all}: all dependences are costly,
7533 @var{true_store_to_load}: a true dependence from store to load is costly,
7534 @var{store_to_load}: any dependence from store to load is costly,
7535 @var{number}: any dependence which latency >= @var{number} is costly.
7537 @item -minsert-sched-nops=@var{scheme}
7538 @opindex minsert-sched-nops
7539 This option controls which nop insertion scheme will be used during
7540 the second scheduling pass. The argument @var{scheme} takes one of the
7542 @var{no}: Don't insert nops.
7543 @var{pad}: Pad with nops any dispatch group which has vacant issue slots,
7544 according to the scheduler's grouping.
7545 @var{regroup_exact}: Insert nops to force costly dependent insns into
7546 separate groups. Insert exactly as many nops as needed to force an insn
7547 to a new group, according to the estimatied processor grouping.
7548 @var{number}: Insert nops to force costly dependent insns into
7549 separate groups. Insert @var{number} nops to force an insn to a new group.
7553 On System V.4 and embedded PowerPC systems compile code using calling
7554 conventions that adheres to the March 1995 draft of the System V
7555 Application Binary Interface, PowerPC processor supplement. This is the
7556 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
7558 @item -mcall-sysv-eabi
7559 @opindex mcall-sysv-eabi
7560 Specify both @option{-mcall-sysv} and @option{-meabi} options.
7562 @item -mcall-sysv-noeabi
7563 @opindex mcall-sysv-noeabi
7564 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
7566 @item -mcall-solaris
7567 @opindex mcall-solaris
7568 On System V.4 and embedded PowerPC systems compile code for the Solaris
7572 @opindex mcall-linux
7573 On System V.4 and embedded PowerPC systems compile code for the
7574 Linux-based GNU system.
7578 On System V.4 and embedded PowerPC systems compile code for the
7579 Hurd-based GNU system.
7582 @opindex mcall-netbsd
7583 On System V.4 and embedded PowerPC systems compile code for the
7584 NetBSD operating system.
7586 @item -maix-struct-return
7587 @opindex maix-struct-return
7588 Return all structures in memory (as specified by the AIX ABI)@.
7590 @item -msvr4-struct-return
7591 @opindex msvr4-struct-return
7592 Return structures smaller than 8 bytes in registers (as specified by the
7596 @opindex mabi=altivec
7597 Extend the current ABI with AltiVec ABI extensions. This does not
7598 change the default ABI, instead it adds the AltiVec ABI extensions to
7601 @item -mabi=no-altivec
7602 @opindex mabi=no-altivec
7603 Disable AltiVec ABI extensions for the current ABI.
7606 @itemx -mno-prototype
7608 @opindex mno-prototype
7609 On System V.4 and embedded PowerPC systems assume that all calls to
7610 variable argument functions are properly prototyped. Otherwise, the
7611 compiler must insert an instruction before every non prototyped call to
7612 set or clear bit 6 of the condition code register (@var{CR}) to
7613 indicate whether floating point values were passed in the floating point
7614 registers in case the function takes a variable arguments. With
7615 @option{-mprototype}, only calls to prototyped variable argument functions
7616 will set or clear the bit.
7620 On embedded PowerPC systems, assume that the startup module is called
7621 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
7622 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
7627 On embedded PowerPC systems, assume that the startup module is called
7628 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
7633 On embedded PowerPC systems, assume that the startup module is called
7634 @file{crt0.o} and the standard C libraries are @file{libads.a} and
7638 @opindex myellowknife
7639 On embedded PowerPC systems, assume that the startup module is called
7640 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
7645 On System V.4 and embedded PowerPC systems, specify that you are
7646 compiling for a VxWorks system.
7650 Specify that you are compiling for the WindISS simulation environment.
7654 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
7655 header to indicate that @samp{eabi} extended relocations are used.
7661 On System V.4 and embedded PowerPC systems do (do not) adhere to the
7662 Embedded Applications Binary Interface (eabi) which is a set of
7663 modifications to the System V.4 specifications. Selecting @option{-meabi}
7664 means that the stack is aligned to an 8 byte boundary, a function
7665 @code{__eabi} is called to from @code{main} to set up the eabi
7666 environment, and the @option{-msdata} option can use both @code{r2} and
7667 @code{r13} to point to two separate small data areas. Selecting
7668 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
7669 do not call an initialization function from @code{main}, and the
7670 @option{-msdata} option will only use @code{r13} to point to a single
7671 small data area. The @option{-meabi} option is on by default if you
7672 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
7675 @opindex msdata=eabi
7676 On System V.4 and embedded PowerPC systems, put small initialized
7677 @code{const} global and static data in the @samp{.sdata2} section, which
7678 is pointed to by register @code{r2}. Put small initialized
7679 non-@code{const} global and static data in the @samp{.sdata} section,
7680 which is pointed to by register @code{r13}. Put small uninitialized
7681 global and static data in the @samp{.sbss} section, which is adjacent to
7682 the @samp{.sdata} section. The @option{-msdata=eabi} option is
7683 incompatible with the @option{-mrelocatable} option. The
7684 @option{-msdata=eabi} option also sets the @option{-memb} option.
7687 @opindex msdata=sysv
7688 On System V.4 and embedded PowerPC systems, put small global and static
7689 data in the @samp{.sdata} section, which is pointed to by register
7690 @code{r13}. Put small uninitialized global and static data in the
7691 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
7692 The @option{-msdata=sysv} option is incompatible with the
7693 @option{-mrelocatable} option.
7695 @item -msdata=default
7697 @opindex msdata=default
7699 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
7700 compile code the same as @option{-msdata=eabi}, otherwise compile code the
7701 same as @option{-msdata=sysv}.
7704 @opindex msdata-data
7705 On System V.4 and embedded PowerPC systems, put small global and static
7706 data in the @samp{.sdata} section. Put small uninitialized global and
7707 static data in the @samp{.sbss} section. Do not use register @code{r13}
7708 to address small data however. This is the default behavior unless
7709 other @option{-msdata} options are used.
7713 @opindex msdata=none
7715 On embedded PowerPC systems, put all initialized global and static data
7716 in the @samp{.data} section, and all uninitialized data in the
7717 @samp{.bss} section.
7721 @cindex smaller data references (PowerPC)
7722 @cindex .sdata/.sdata2 references (PowerPC)
7723 On embedded PowerPC systems, put global and static items less than or
7724 equal to @var{num} bytes into the small data or bss sections instead of
7725 the normal data or bss section. By default, @var{num} is 8. The
7726 @option{-G @var{num}} switch is also passed to the linker.
7727 All modules should be compiled with the same @option{-G @var{num}} value.
7730 @itemx -mno-regnames
7732 @opindex mno-regnames
7733 On System V.4 and embedded PowerPC systems do (do not) emit register
7734 names in the assembly language output using symbolic forms.
7737 @itemx -mno-longcall
7739 @opindex mno-longcall
7740 Default to making all function calls via pointers, so that functions
7741 which reside further than 64 megabytes (67,108,864 bytes) from the
7742 current location can be called. This setting can be overridden by the
7743 @code{shortcall} function attribute, or by @code{#pragma longcall(0)}.
7745 Some linkers are capable of detecting out-of-range calls and generating
7746 glue code on the fly. On these systems, long calls are unnecessary and
7747 generate slower code. As of this writing, the AIX linker can do this,
7748 as can the GNU linker for PowerPC/64. It is planned to add this feature
7749 to the GNU linker for 32-bit PowerPC systems as well.
7751 On Mach-O (Darwin) systems, this option directs the compiler emit to
7752 the glue for every direct call, and the Darwin linker decides whether
7753 to use or discard it.
7755 In the future, we may cause GCC to ignore all longcall specifications
7756 when the linker is known to generate glue.
7760 Adds support for multithreading with the @dfn{pthreads} library.
7761 This option sets flags for both the preprocessor and linker.
7765 @node Darwin Options
7766 @subsection Darwin Options
7767 @cindex Darwin options
7769 These options are defined for all architectures running the Darwin operating
7770 system. They are useful for compatibility with other Mac OS compilers.
7775 Loads all members of static archive libraries.
7776 See man ld(1) for more information.
7778 @item -arch_errors_fatal
7779 @opindex arch_errors_fatal
7780 Cause the errors having to do with files that have the wrong architecture
7784 @opindex bind_at_load
7785 Causes the output file to be marked such that the dynamic linker will
7786 bind all undefined references when the file is loaded or launched.
7790 Produce a Mach-o bundle format file.
7791 See man ld(1) for more information.
7793 @item -bundle_loader @var{executable}
7794 @opindex bundle_loader
7795 This specifies the @var{executable} that will be loading the build
7796 output file being linked. See man ld(1) for more information.
7798 @item -allowable_client @var{client_name}
7802 @item -compatibility_version
7803 @item -current_version
7804 @item -dependency-file
7806 @item -dylinker_install_name
7809 @item -exported_symbols_list
7811 @item -flat_namespace
7812 @item -force_cpusubtype_ALL
7813 @item -force_flat_namespace
7814 @item -headerpad_max_install_names
7818 @item -keep_private_externs
7820 @item -multiply_defined
7821 @item -multiply_defined_unused
7823 @item -nofixprebinding
7826 @item -noseglinkedit
7827 @item -pagezero_size
7829 @item -prebind_all_twolevel_modules
7830 @item -private_bundle
7831 @item -read_only_relocs
7833 @item -sectobjectsymbols
7837 @item -sectobjectsymbols
7839 @item -seg_addr_table
7840 @item -seg_addr_table_filename
7843 @item -segs_read_only_addr
7844 @item -segs_read_write_addr
7845 @item -single_module
7849 @item -twolevel_namespace
7852 @item -unexported_symbols_list
7853 @item -weak_reference_mismatches
7856 @opindex allowable_client
7858 @opindex client_name
7859 @opindex compatibility_version
7860 @opindex current_version
7861 @opindex dependency-file
7863 @opindex dylinker_install_name
7866 @opindex exported_symbols_list
7868 @opindex flat_namespace
7869 @opindex force_cpusubtype_ALL
7870 @opindex force_flat_namespace
7871 @opindex headerpad_max_install_names
7874 @opindex install_name
7875 @opindex keep_private_externs
7876 @opindex multi_module
7877 @opindex multiply_defined
7878 @opindex multiply_defined_unused
7880 @opindex nofixprebinding
7881 @opindex nomultidefs
7883 @opindex noseglinkedit
7884 @opindex pagezero_size
7886 @opindex prebind_all_twolevel_modules
7887 @opindex private_bundle
7888 @opindex read_only_relocs
7890 @opindex sectobjectsymbols
7894 @opindex sectobjectsymbols
7896 @opindex seg_addr_table
7897 @opindex seg_addr_table_filename
7898 @opindex seglinkedit
7900 @opindex segs_read_only_addr
7901 @opindex segs_read_write_addr
7902 @opindex single_module
7904 @opindex sub_library
7905 @opindex sub_umbrella
7906 @opindex twolevel_namespace
7909 @opindex unexported_symbols_list
7910 @opindex weak_reference_mismatches
7911 @opindex whatsloaded
7913 This options are available for Darwin linker. Darwin linker man page
7914 describes them in detail.
7919 @subsection IBM RT Options
7921 @cindex IBM RT options
7923 These @samp{-m} options are defined for the IBM RT PC:
7927 @opindex min-line-mul
7928 Use an in-line code sequence for integer multiplies. This is the
7931 @item -mcall-lib-mul
7932 @opindex mcall-lib-mul
7933 Call @code{lmul$$} for integer multiples.
7935 @item -mfull-fp-blocks
7936 @opindex mfull-fp-blocks
7937 Generate full-size floating point data blocks, including the minimum
7938 amount of scratch space recommended by IBM@. This is the default.
7940 @item -mminimum-fp-blocks
7941 @opindex mminimum-fp-blocks
7942 Do not include extra scratch space in floating point data blocks. This
7943 results in smaller code, but slower execution, since scratch space must
7944 be allocated dynamically.
7946 @cindex @file{stdarg.h} and RT PC
7947 @item -mfp-arg-in-fpregs
7948 @opindex mfp-arg-in-fpregs
7949 Use a calling sequence incompatible with the IBM calling convention in
7950 which floating point arguments are passed in floating point registers.
7951 Note that @code{stdarg.h} will not work with floating point operands
7952 if this option is specified.
7954 @item -mfp-arg-in-gregs
7955 @opindex mfp-arg-in-gregs
7956 Use the normal calling convention for floating point arguments. This is
7959 @item -mhc-struct-return
7960 @opindex mhc-struct-return
7961 Return structures of more than one word in memory, rather than in a
7962 register. This provides compatibility with the MetaWare HighC (hc)
7963 compiler. Use the option @option{-fpcc-struct-return} for compatibility
7964 with the Portable C Compiler (pcc).
7966 @item -mnohc-struct-return
7967 @opindex mnohc-struct-return
7968 Return some structures of more than one word in registers, when
7969 convenient. This is the default. For compatibility with the
7970 IBM-supplied compilers, use the option @option{-fpcc-struct-return} or the
7971 option @option{-mhc-struct-return}.
7975 @subsection MIPS Options
7976 @cindex MIPS options
7978 These @samp{-m} options are defined for the MIPS family of computers:
7982 @item -march=@var{arch}
7984 Generate code that will run on @var{arch}, which can be the name of a
7985 generic MIPS ISA, or the name of a particular processor.
7987 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
7988 @samp{mips32}, @samp{mips32r2}, and @samp{mips64}.
7989 The processor names are:
7990 @samp{4kc}, @samp{4kp}, @samp{5kc}, @samp{20kc},
7992 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
7993 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000}, @samp{rm7000},
7997 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4300},
7998 @samp{vr5000}, @samp{vr5400} and @samp{vr5500}.
7999 The special value @samp{from-abi} selects the
8000 most compatible architecture for the selected ABI (that is,
8001 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
8003 In processor names, a final @samp{000} can be abbreviated as @samp{k}
8004 (for example, @samp{-march=r2k}). Prefixes are optional, and
8005 @samp{vr} may be written @samp{r}.
8007 GCC defines two macros based on the value of this option. The first
8008 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
8009 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
8010 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
8011 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
8012 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
8014 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
8015 above. In other words, it will have the full prefix and will not
8016 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
8017 the macro names the resolved architecture (either @samp{"mips1"} or
8018 @samp{"mips3"}). It names the default architecture when no
8019 @option{-march} option is given.
8021 @item -mtune=@var{arch}
8023 Optimize for @var{arch}. Among other things, this option controls
8024 the way instructions are scheduled, and the perceived cost of arithmetic
8025 operations. The list of @var{arch} values is the same as for
8028 When this option is not used, GCC will optimize for the processor
8029 specified by @option{-march}. By using @option{-march} and
8030 @option{-mtune} together, it is possible to generate code that will
8031 run on a family of processors, but optimize the code for one
8032 particular member of that family.
8034 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
8035 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
8036 @samp{-march} ones described above.
8040 Equivalent to @samp{-march=mips1}.
8044 Equivalent to @samp{-march=mips2}.
8048 Equivalent to @samp{-march=mips3}.
8052 Equivalent to @samp{-march=mips4}.
8056 Equivalent to @samp{-march=mips32}.
8060 Equivalent to @samp{-march=mips32r2}.
8064 Equivalent to @samp{-march=mips64}.
8067 @itemx -mno-fused-madd
8068 @opindex mfused-madd
8069 @opindex mno-fused-madd
8070 Generate code that uses (does not use) the floating point multiply and
8071 accumulate instructions, when they are available. These instructions
8072 are generated by default if they are available, but this may be
8073 undesirable if the extra precision causes problems or on certain chips
8074 in the mode where denormals are rounded to zero where denormals
8075 generated by multiply and accumulate instructions cause exceptions
8080 Assume that floating point registers are 32 bits wide.
8084 Assume that floating point registers are 64 bits wide.
8088 Assume that general purpose registers are 32 bits wide.
8092 Assume that general purpose registers are 64 bits wide.
8096 Force int and long types to be 64 bits wide. See @option{-mlong32} for an
8097 explanation of the default, and the width of pointers.
8101 Force long types to be 64 bits wide. See @option{-mlong32} for an
8102 explanation of the default, and the width of pointers.
8106 Force long, int, and pointer types to be 32 bits wide.
8108 The default size of ints, longs and pointers depends on the ABI@. All
8109 the supported ABIs use 32-bit ints. The n64 ABI uses 64-bit longs, as
8110 does the 64-bit Cygnus EABI; the others use 32-bit longs. Pointers
8111 are the same size as longs, or the same size as integer registers,
8112 whichever is smaller.
8126 Generate code for the given ABI@.
8128 Note that there are two embedded ABIs: @option{-mabi=eabi}
8129 selects the one defined by Cygnus while @option{-meabi=meabi}
8130 selects the one defined by MIPS@. Both these ABIs have
8131 32-bit and 64-bit variants. Normally, GCC will generate
8132 64-bit code when you select a 64-bit architecture, but you
8133 can use @option{-mgp32} to get 32-bit code instead.
8135 @item -mabi-fake-default
8136 @opindex mabi-fake-default
8137 You don't want to know what this option does. No, really. I mean
8138 it. Move on to the next option.
8140 What? You're still here? Oh, well@enddots{} Ok, here's the deal. GCC
8141 wants the default set of options to get the root of the multilib tree,
8142 and the shared library SONAMEs without any multilib-indicating
8143 suffixes. This is not convenience for @samp{mips64-linux-gnu}, since
8144 we want to default to the N32 ABI, while still being binary-compatible
8145 with @samp{mips-linux-gnu} if you stick to the O32 ABI@. Being
8146 binary-compatible means shared libraries should have the same SONAMEs,
8147 and libraries should live in the same location. Having O32 libraries
8148 in a sub-directory named say @file{o32} is not acceptable.
8150 So we trick GCC into believing that O32 is the default ABI, except
8151 that we override the default with some internal command-line
8152 processing magic. Problem is, if we stopped at that, and you then
8153 created a multilib-aware package that used the output of @command{gcc
8154 -print-multi-lib} to decide which multilibs to build, and how, and
8155 you'd find yourself in an awkward situation when you found out that
8156 some of the options listed ended up mapping to the same multilib, and
8157 none of your libraries was actually built for the multilib that
8158 @option{-print-multi-lib} claims to be the default. So we added this
8159 option that disables the default switcher, falling back to GCC's
8160 original notion of the default library. Confused yet?
8162 For short: don't ever use this option, unless you find it in the list
8163 of additional options to be used when building for multilibs, in the
8164 output of @option{gcc -print-multi-lib}.
8168 Generate code for the MIPS assembler, and invoke @file{mips-tfile} to
8169 add normal debug information. This is the default for all
8170 platforms except for the OSF/1 reference platform, using the OSF/rose
8171 object format. If the either of the @option{-gstabs} or @option{-gstabs+}
8172 switches are used, the @file{mips-tfile} program will encapsulate the
8173 stabs within MIPS ECOFF@.
8177 Generate code for the GNU assembler. This is the default on the OSF/1
8178 reference platform, using the OSF/rose object format. Also, this is
8179 the default if the configure option @option{--with-gnu-as} is used.
8181 @item -msplit-addresses
8182 @itemx -mno-split-addresses
8183 @opindex msplit-addresses
8184 @opindex mno-split-addresses
8185 Generate code to load the high and low parts of address constants separately.
8186 This allows GCC to optimize away redundant loads of the high order
8187 bits of addresses. This optimization requires GNU as and GNU ld.
8188 This optimization is enabled by default for some embedded targets where
8189 GNU as and GNU ld are standard.
8195 The @option{-mrnames} switch says to output code using the MIPS software
8196 names for the registers, instead of the hardware names (ie, @var{a0}
8197 instead of @var{$4}). The only known assembler that supports this option
8198 is the Algorithmics assembler.
8204 The @option{-mmemcpy} switch makes all block moves call the appropriate
8205 string function (@samp{memcpy} or @samp{bcopy}) instead of possibly
8206 generating inline code.
8209 @itemx -mno-mips-tfile
8210 @opindex mmips-tfile
8211 @opindex mno-mips-tfile
8212 The @option{-mno-mips-tfile} switch causes the compiler not
8213 postprocess the object file with the @file{mips-tfile} program,
8214 after the MIPS assembler has generated it to add debug support. If
8215 @file{mips-tfile} is not run, then no local variables will be
8216 available to the debugger. In addition, @file{stage2} and
8217 @file{stage3} objects will have the temporary file names passed to the
8218 assembler embedded in the object file, which means the objects will
8219 not compare the same. The @option{-mno-mips-tfile} switch should only
8220 be used when there are bugs in the @file{mips-tfile} program that
8221 prevents compilation.
8224 @opindex msoft-float
8225 Generate output containing library calls for floating point.
8226 @strong{Warning:} the requisite libraries are not part of GCC@.
8227 Normally the facilities of the machine's usual C compiler are used, but
8228 this can't be done directly in cross-compilation. You must make your
8229 own arrangements to provide suitable library functions for
8233 @opindex mhard-float
8234 Generate output containing floating point instructions. This is the
8235 default if you use the unmodified sources.
8238 @itemx -mno-abicalls
8240 @opindex mno-abicalls
8241 Emit (or do not emit) the pseudo operations @samp{.abicalls},
8242 @samp{.cpload}, and @samp{.cprestore} that some System V.4 ports use for
8243 position independent code.
8249 Lift (or do not lift) the usual restrictions on the size of the global
8252 GCC normally uses a single instruction to load values from the GOT.
8253 While this is relatively efficient, it will only work if the GOT
8254 is smaller than about 64k. Anything larger will cause the linker
8255 to report an error such as:
8257 @cindex relocation truncated to fit (MIPS)
8259 relocation truncated to fit: R_MIPS_GOT16 foobar
8262 If this happens, you should recompile your code with @option{-mxgot}.
8263 It should then work with very large GOTs, although it will also be
8264 less efficient, since it will take three instructions to fetch the
8265 value of a global symbol.
8267 Note that some linkers can create multiple GOTs. If you have such a
8268 linker, you should only need to use @option{-mxgot} when a single object
8269 file accesses more than 64k's worth of GOT entries. Very few do.
8271 These options have no effect unless GCC is generating position
8275 @itemx -mno-long-calls
8276 @opindex mlong-calls
8277 @opindex mno-long-calls
8278 Do all calls with the @samp{JALR} instruction, which requires
8279 loading up a function's address into a register before the call.
8280 You need to use this switch, if you call outside of the current
8281 512 megabyte segment to functions that are not through pointers.
8283 @item -membedded-pic
8284 @itemx -mno-embedded-pic
8285 @opindex membedded-pic
8286 @opindex mno-embedded-pic
8287 Generate PIC code suitable for some embedded systems. All calls are
8288 made using PC relative address, and all data is addressed using the $gp
8289 register. No more than 65536 bytes of global data may be used. This
8290 requires GNU as and GNU ld which do most of the work. This currently
8291 only works on targets which use ECOFF; it does not work with ELF@.
8293 @item -membedded-data
8294 @itemx -mno-embedded-data
8295 @opindex membedded-data
8296 @opindex mno-embedded-data
8297 Allocate variables to the read-only data section first if possible, then
8298 next in the small data section if possible, otherwise in data. This gives
8299 slightly slower code than the default, but reduces the amount of RAM required
8300 when executing, and thus may be preferred for some embedded systems.
8302 @item -muninit-const-in-rodata
8303 @itemx -mno-uninit-const-in-rodata
8304 @opindex muninit-const-in-rodata
8305 @opindex mno-uninit-const-in-rodata
8306 When used together with @option{-membedded-data}, it will always store uninitialized
8307 const variables in the read-only data section.
8309 @item -msingle-float
8310 @itemx -mdouble-float
8311 @opindex msingle-float
8312 @opindex mdouble-float
8313 The @option{-msingle-float} switch tells gcc to assume that the floating
8314 point coprocessor only supports single precision operations, as on the
8315 @samp{r4650} chip. The @option{-mdouble-float} switch permits gcc to use
8316 double precision operations. This is the default.
8322 Permit use of the @samp{mad}, @samp{madu} and @samp{mul} instructions,
8323 as on the @samp{r4650} chip.
8327 Turns on @option{-msingle-float}, @option{-mmad}, and, at least for now,
8328 @option{-mcpu=r4650}.
8334 Enable 16-bit instructions.
8338 Compile code for the processor in little endian mode.
8339 The requisite libraries are assumed to exist.
8343 Compile code for the processor in big endian mode.
8344 The requisite libraries are assumed to exist.
8348 @cindex smaller data references (MIPS)
8349 @cindex gp-relative references (MIPS)
8350 Put global and static items less than or equal to @var{num} bytes into
8351 the small data or bss sections instead of the normal data or bss
8352 section. This allows the assembler to emit one word memory reference
8353 instructions based on the global pointer (@var{gp} or @var{$28}),
8354 instead of the normal two words used. By default, @var{num} is 8 when
8355 the MIPS assembler is used, and 0 when the GNU assembler is used. The
8356 @option{-G @var{num}} switch is also passed to the assembler and linker.
8357 All modules should be compiled with the same @option{-G @var{num}}
8362 Tell the MIPS assembler to not run its preprocessor over user
8363 assembler files (with a @samp{.s} suffix) when assembling them.
8367 Pass an option to gas which will cause nops to be inserted if
8368 the read of the destination register of an mfhi or mflo instruction
8369 occurs in the following two instructions.
8374 Work around certain SB-1 CPU core errata.
8375 (This flag currently works around the SB-1 revision 2
8376 ``F1'' and ``F2'' floating point errata.)
8380 Do not include the default crt0.
8382 @item -mflush-func=@var{func}
8383 @itemx -mno-flush-func
8384 @opindex mflush-func
8385 Specifies the function to call to flush the I and D caches, or to not
8386 call any such function. If called, the function must take the same
8387 arguments as the common @code{_flush_func()}, that is, the address of the
8388 memory range for which the cache is being flushed, the size of the
8389 memory range, and the number 3 (to flush both caches). The default
8390 depends on the target gcc was configured for, but commonly is either
8391 @samp{_flush_func} or @samp{__cpu_flush}.
8393 @item -mbranch-likely
8394 @itemx -mno-branch-likely
8395 @opindex mbranch-likely
8396 @opindex mno-branch-likely
8397 Enable or disable use of Branch Likely instructions, regardless of the
8398 default for the selected architecture. By default, Branch Likely
8399 instructions may be generated if they are supported by the selected
8400 architecture. An exception is for the MIPS32 and MIPS64 architectures
8401 and processors which implement those architectures; for those, Branch
8402 Likely instructions will not be generated by default because the MIPS32
8403 and MIPS64 architectures specifically deprecate their use.
8406 @node i386 and x86-64 Options
8407 @subsection Intel 386 and AMD x86-64 Options
8408 @cindex i386 Options
8409 @cindex x86-64 Options
8410 @cindex Intel 386 Options
8411 @cindex AMD x86-64 Options
8413 These @samp{-m} options are defined for the i386 and x86-64 family of
8417 @item -mtune=@var{cpu-type}
8419 Tune to @var{cpu-type} everything applicable about the generated code, except
8420 for the ABI and the set of available instructions. The choices for
8421 @var{cpu-type} are @samp{i386}, @samp{i486}, @samp{i586}, @samp{i686},
8422 @samp{pentium}, @samp{pentium-mmx}, @samp{pentiumpro}, @samp{pentium2},
8423 @samp{pentium3}, @samp{pentium4}, @samp{k6}, @samp{k6-2}, @samp{k6-3},
8424 @samp{athlon}, @samp{athlon-tbird}, @samp{athlon-4}, @samp{athlon-xp},
8425 @samp{athlon-mp}, @samp{winchip-c6}, @samp{winchip2}, @samp{k8}, @samp{c3}
8428 While picking a specific @var{cpu-type} will schedule things appropriately
8429 for that particular chip, the compiler will not generate any code that
8430 does not run on the i386 without the @option{-march=@var{cpu-type}} option
8431 being used. @samp{i586} is equivalent to @samp{pentium} and @samp{i686}
8432 is equivalent to @samp{pentiumpro}. @samp{k6} and @samp{athlon} are the
8433 AMD chips as opposed to the Intel ones.
8435 @item -march=@var{cpu-type}
8437 Generate instructions for the machine type @var{cpu-type}. The choices
8438 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
8439 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
8441 @item -mcpu=@var{cpu-type}
8443 A deprecated synonym for @option{-mtune}.
8452 @opindex mpentiumpro
8453 These options are synonyms for @option{-mtune=i386}, @option{-mtune=i486},
8454 @option{-mtune=pentium}, and @option{-mtune=pentiumpro} respectively.
8455 These synonyms are deprecated.
8457 @item -mfpmath=@var{unit}
8459 generate floating point arithmetics for selected unit @var{unit}. the choices
8464 Use the standard 387 floating point coprocessor present majority of chips and
8465 emulated otherwise. Code compiled with this option will run almost everywhere.
8466 The temporary results are computed in 80bit precision instead of precision
8467 specified by the type resulting in slightly different results compared to most
8468 of other chips. See @option{-ffloat-store} for more detailed description.
8470 This is the default choice for i386 compiler.
8473 Use scalar floating point instructions present in the SSE instruction set.
8474 This instruction set is supported by Pentium3 and newer chips, in the AMD line
8475 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
8476 instruction set supports only single precision arithmetics, thus the double and
8477 extended precision arithmetics is still done using 387. Later version, present
8478 only in Pentium4 and the future AMD x86-64 chips supports double precision
8481 For i387 you need to use @option{-march=@var{cpu-type}}, @option{-msse} or
8482 @option{-msse2} switches to enable SSE extensions and make this option
8483 effective. For x86-64 compiler, these extensions are enabled by default.
8485 The resulting code should be considerably faster in majority of cases and avoid
8486 the numerical instability problems of 387 code, but may break some existing
8487 code that expects temporaries to be 80bit.
8489 This is the default choice for x86-64 compiler.
8492 Use all SSE extensions enabled by @option{-msse2} as well as the new
8493 SSE extensions in Prescott New Instructions. @option{-mpni} also
8494 enables 2 builtin functions, @code{__builtin_ia32_monitor} and
8495 @code{__builtin_ia32_mwait}, for new instructions @code{monitor} and
8499 Attempt to utilize both instruction sets at once. This effectively double the
8500 amount of available registers and on chips with separate execution units for
8501 387 and SSE the execution resources too. Use this option with care, as it is
8502 still experimental, because gcc register allocator does not model separate
8503 functional units well resulting in instable performance.
8506 @item -masm=@var{dialect}
8507 @opindex masm=@var{dialect}
8508 Output asm instructions using selected @var{dialect}. Supported choices are
8509 @samp{intel} or @samp{att} (the default one).
8514 @opindex mno-ieee-fp
8515 Control whether or not the compiler uses IEEE floating point
8516 comparisons. These handle correctly the case where the result of a
8517 comparison is unordered.
8520 @opindex msoft-float
8521 Generate output containing library calls for floating point.
8522 @strong{Warning:} the requisite libraries are not part of GCC@.
8523 Normally the facilities of the machine's usual C compiler are used, but
8524 this can't be done directly in cross-compilation. You must make your
8525 own arrangements to provide suitable library functions for
8528 On machines where a function returns floating point results in the 80387
8529 register stack, some floating point opcodes may be emitted even if
8530 @option{-msoft-float} is used.
8532 @item -mno-fp-ret-in-387
8533 @opindex mno-fp-ret-in-387
8534 Do not use the FPU registers for return values of functions.
8536 The usual calling convention has functions return values of types
8537 @code{float} and @code{double} in an FPU register, even if there
8538 is no FPU@. The idea is that the operating system should emulate
8541 The option @option{-mno-fp-ret-in-387} causes such values to be returned
8542 in ordinary CPU registers instead.
8544 @item -mno-fancy-math-387
8545 @opindex mno-fancy-math-387
8546 Some 387 emulators do not support the @code{sin}, @code{cos} and
8547 @code{sqrt} instructions for the 387. Specify this option to avoid
8548 generating those instructions. This option is the default on FreeBSD,
8549 OpenBSD and NetBSD@. This option is overridden when @option{-march}
8550 indicates that the target cpu will always have an FPU and so the
8551 instruction will not need emulation. As of revision 2.6.1, these
8552 instructions are not generated unless you also use the
8553 @option{-funsafe-math-optimizations} switch.
8555 @item -malign-double
8556 @itemx -mno-align-double
8557 @opindex malign-double
8558 @opindex mno-align-double
8559 Control whether GCC aligns @code{double}, @code{long double}, and
8560 @code{long long} variables on a two word boundary or a one word
8561 boundary. Aligning @code{double} variables on a two word boundary will
8562 produce code that runs somewhat faster on a @samp{Pentium} at the
8563 expense of more memory.
8565 @strong{Warning:} if you use the @option{-malign-double} switch,
8566 structures containing the above types will be aligned differently than
8567 the published application binary interface specifications for the 386
8568 and will not be binary compatible with structures in code compiled
8569 without that switch.
8571 @item -m96bit-long-double
8572 @item -m128bit-long-double
8573 @opindex m96bit-long-double
8574 @opindex m128bit-long-double
8575 These switches control the size of @code{long double} type. The i386
8576 application binary interface specifies the size to be 96 bits,
8577 so @option{-m96bit-long-double} is the default in 32 bit mode.
8579 Modern architectures (Pentium and newer) would prefer @code{long double}
8580 to be aligned to an 8 or 16 byte boundary. In arrays or structures
8581 conforming to the ABI, this would not be possible. So specifying a
8582 @option{-m128bit-long-double} will align @code{long double}
8583 to a 16 byte boundary by padding the @code{long double} with an additional
8586 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
8587 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
8589 Notice that neither of these options enable any extra precision over the x87
8590 standard of 80 bits for a @code{long double}.
8592 @strong{Warning:} if you override the default value for your target ABI, the
8593 structures and arrays containing @code{long double} will change their size as
8594 well as function calling convention for function taking @code{long double}
8595 will be modified. Hence they will not be binary compatible with arrays or
8596 structures in code compiled without that switch.
8600 @itemx -mno-svr3-shlib
8601 @opindex msvr3-shlib
8602 @opindex mno-svr3-shlib
8603 Control whether GCC places uninitialized local variables into the
8604 @code{bss} or @code{data} segments. @option{-msvr3-shlib} places them
8605 into @code{bss}. These options are meaningful only on System V Release 3.
8609 Use a different function-calling convention, in which functions that
8610 take a fixed number of arguments return with the @code{ret} @var{num}
8611 instruction, which pops their arguments while returning. This saves one
8612 instruction in the caller since there is no need to pop the arguments
8615 You can specify that an individual function is called with this calling
8616 sequence with the function attribute @samp{stdcall}. You can also
8617 override the @option{-mrtd} option by using the function attribute
8618 @samp{cdecl}. @xref{Function Attributes}.
8620 @strong{Warning:} this calling convention is incompatible with the one
8621 normally used on Unix, so you cannot use it if you need to call
8622 libraries compiled with the Unix compiler.
8624 Also, you must provide function prototypes for all functions that
8625 take variable numbers of arguments (including @code{printf});
8626 otherwise incorrect code will be generated for calls to those
8629 In addition, seriously incorrect code will result if you call a
8630 function with too many arguments. (Normally, extra arguments are
8631 harmlessly ignored.)
8633 @item -mregparm=@var{num}
8635 Control how many registers are used to pass integer arguments. By
8636 default, no registers are used to pass arguments, and at most 3
8637 registers can be used. You can control this behavior for a specific
8638 function by using the function attribute @samp{regparm}.
8639 @xref{Function Attributes}.
8641 @strong{Warning:} if you use this switch, and
8642 @var{num} is nonzero, then you must build all modules with the same
8643 value, including any libraries. This includes the system libraries and
8646 @item -mpreferred-stack-boundary=@var{num}
8647 @opindex mpreferred-stack-boundary
8648 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
8649 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
8650 the default is 4 (16 bytes or 128 bits), except when optimizing for code
8651 size (@option{-Os}), in which case the default is the minimum correct
8652 alignment (4 bytes for x86, and 8 bytes for x86-64).
8654 On Pentium and PentiumPro, @code{double} and @code{long double} values
8655 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
8656 suffer significant run time performance penalties. On Pentium III, the
8657 Streaming SIMD Extension (SSE) data type @code{__m128} suffers similar
8658 penalties if it is not 16 byte aligned.
8660 To ensure proper alignment of this values on the stack, the stack boundary
8661 must be as aligned as that required by any value stored on the stack.
8662 Further, every function must be generated such that it keeps the stack
8663 aligned. Thus calling a function compiled with a higher preferred
8664 stack boundary from a function compiled with a lower preferred stack
8665 boundary will most likely misalign the stack. It is recommended that
8666 libraries that use callbacks always use the default setting.
8668 This extra alignment does consume extra stack space, and generally
8669 increases code size. Code that is sensitive to stack space usage, such
8670 as embedded systems and operating system kernels, may want to reduce the
8671 preferred alignment to @option{-mpreferred-stack-boundary=2}.
8689 These switches enable or disable the use of built-in functions that allow
8690 direct access to the MMX, SSE and 3Dnow extensions of the instruction set.
8692 @xref{X86 Built-in Functions}, for details of the functions enabled
8693 and disabled by these switches.
8695 To have SSE/SSE2 instructions generated automatically from floating-point
8696 code, see @option{-mfpmath=sse}.
8699 @itemx -mno-push-args
8701 @opindex mno-push-args
8702 Use PUSH operations to store outgoing parameters. This method is shorter
8703 and usually equally fast as method using SUB/MOV operations and is enabled
8704 by default. In some cases disabling it may improve performance because of
8705 improved scheduling and reduced dependencies.
8707 @item -maccumulate-outgoing-args
8708 @opindex maccumulate-outgoing-args
8709 If enabled, the maximum amount of space required for outgoing arguments will be
8710 computed in the function prologue. This is faster on most modern CPUs
8711 because of reduced dependencies, improved scheduling and reduced stack usage
8712 when preferred stack boundary is not equal to 2. The drawback is a notable
8713 increase in code size. This switch implies @option{-mno-push-args}.
8717 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
8718 on thread-safe exception handling must compile and link all code with the
8719 @option{-mthreads} option. When compiling, @option{-mthreads} defines
8720 @option{-D_MT}; when linking, it links in a special thread helper library
8721 @option{-lmingwthrd} which cleans up per thread exception handling data.
8723 @item -mno-align-stringops
8724 @opindex mno-align-stringops
8725 Do not align destination of inlined string operations. This switch reduces
8726 code size and improves performance in case the destination is already aligned,
8727 but gcc don't know about it.
8729 @item -minline-all-stringops
8730 @opindex minline-all-stringops
8731 By default GCC inlines string operations only when destination is known to be
8732 aligned at least to 4 byte boundary. This enables more inlining, increase code
8733 size, but may improve performance of code that depends on fast memcpy, strlen
8734 and memset for short lengths.
8736 @item -momit-leaf-frame-pointer
8737 @opindex momit-leaf-frame-pointer
8738 Don't keep the frame pointer in a register for leaf functions. This
8739 avoids the instructions to save, set up and restore frame pointers and
8740 makes an extra register available in leaf functions. The option
8741 @option{-fomit-frame-pointer} removes the frame pointer for all functions
8742 which might make debugging harder.
8744 @item -mtls-direct-seg-refs
8745 @itemx -mno-tls-direct-seg-refs
8746 @opindex mtls-direct-seg-refs
8747 Controls whether TLS variables may be accessed with offsets from the
8748 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
8749 or whether the thread base pointer must be added. Whether or not this
8750 is legal depends on the operating system, and whether it maps the
8751 segment to cover the entire TLS area.
8753 For systems that use GNU libc, the default is on.
8756 These @samp{-m} switches are supported in addition to the above
8757 on AMD x86-64 processors in 64-bit environments.
8764 Generate code for a 32-bit or 64-bit environment.
8765 The 32-bit environment sets int, long and pointer to 32 bits and
8766 generates code that runs on any i386 system.
8767 The 64-bit environment sets int to 32 bits and long and pointer
8768 to 64 bits and generates code for AMD's x86-64 architecture.
8771 @opindex no-red-zone
8772 Do not use a so called red zone for x86-64 code. The red zone is mandated
8773 by the x86-64 ABI, it is a 128-byte area beyond the location of the
8774 stack pointer that will not be modified by signal or interrupt handlers
8775 and therefore can be used for temporary data without adjusting the stack
8776 pointer. The flag @option{-mno-red-zone} disables this red zone.
8778 @item -mcmodel=small
8779 @opindex mcmodel=small
8780 Generate code for the small code model: the program and its symbols must
8781 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
8782 Programs can be statically or dynamically linked. This is the default
8785 @item -mcmodel=kernel
8786 @opindex mcmodel=kernel
8787 Generate code for the kernel code model. The kernel runs in the
8788 negative 2 GB of the address space.
8789 This model has to be used for Linux kernel code.
8791 @item -mcmodel=medium
8792 @opindex mcmodel=medium
8793 Generate code for the medium model: The program is linked in the lower 2
8794 GB of the address space but symbols can be located anywhere in the
8795 address space. Programs can be statically or dynamically linked, but
8796 building of shared libraries are not supported with the medium model.
8798 @item -mcmodel=large
8799 @opindex mcmodel=large
8800 Generate code for the large model: This model makes no assumptions
8801 about addresses and sizes of sections. Currently GCC does not implement
8806 @subsection HPPA Options
8807 @cindex HPPA Options
8809 These @samp{-m} options are defined for the HPPA family of computers:
8812 @item -march=@var{architecture-type}
8814 Generate code for the specified architecture. The choices for
8815 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
8816 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
8817 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
8818 architecture option for your machine. Code compiled for lower numbered
8819 architectures will run on higher numbered architectures, but not the
8822 PA 2.0 support currently requires gas snapshot 19990413 or later. The
8823 next release of binutils (current is 2.9.1) will probably contain PA 2.0
8827 @itemx -mpa-risc-1-1
8828 @itemx -mpa-risc-2-0
8829 @opindex mpa-risc-1-0
8830 @opindex mpa-risc-1-1
8831 @opindex mpa-risc-2-0
8832 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
8835 @opindex mbig-switch
8836 Generate code suitable for big switch tables. Use this option only if
8837 the assembler/linker complain about out of range branches within a switch
8840 @item -mjump-in-delay
8841 @opindex mjump-in-delay
8842 Fill delay slots of function calls with unconditional jump instructions
8843 by modifying the return pointer for the function call to be the target
8844 of the conditional jump.
8846 @item -mdisable-fpregs
8847 @opindex mdisable-fpregs
8848 Prevent floating point registers from being used in any manner. This is
8849 necessary for compiling kernels which perform lazy context switching of
8850 floating point registers. If you use this option and attempt to perform
8851 floating point operations, the compiler will abort.
8853 @item -mdisable-indexing
8854 @opindex mdisable-indexing
8855 Prevent the compiler from using indexing address modes. This avoids some
8856 rather obscure problems when compiling MIG generated code under MACH@.
8858 @item -mno-space-regs
8859 @opindex mno-space-regs
8860 Generate code that assumes the target has no space registers. This allows
8861 GCC to generate faster indirect calls and use unscaled index address modes.
8863 Such code is suitable for level 0 PA systems and kernels.
8865 @item -mfast-indirect-calls
8866 @opindex mfast-indirect-calls
8867 Generate code that assumes calls never cross space boundaries. This
8868 allows GCC to emit code which performs faster indirect calls.
8870 This option will not work in the presence of shared libraries or nested
8873 @item -mlong-load-store
8874 @opindex mlong-load-store
8875 Generate 3-instruction load and store sequences as sometimes required by
8876 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
8879 @item -mportable-runtime
8880 @opindex mportable-runtime
8881 Use the portable calling conventions proposed by HP for ELF systems.
8885 Enable the use of assembler directives only GAS understands.
8887 @item -mschedule=@var{cpu-type}
8889 Schedule code according to the constraints for the machine type
8890 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
8891 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
8892 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
8893 proper scheduling option for your machine. The default scheduling is
8897 @opindex mlinker-opt
8898 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
8899 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
8900 linkers in which they give bogus error messages when linking some programs.
8903 @opindex msoft-float
8904 Generate output containing library calls for floating point.
8905 @strong{Warning:} the requisite libraries are not available for all HPPA
8906 targets. Normally the facilities of the machine's usual C compiler are
8907 used, but this cannot be done directly in cross-compilation. You must make
8908 your own arrangements to provide suitable library functions for
8909 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
8910 does provide software floating point support.
8912 @option{-msoft-float} changes the calling convention in the output file;
8913 therefore, it is only useful if you compile @emph{all} of a program with
8914 this option. In particular, you need to compile @file{libgcc.a}, the
8915 library that comes with GCC, with @option{-msoft-float} in order for
8920 Generate the predefine, @code{_SIO}, for server IO. The default is
8921 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
8922 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO. These
8923 options are available under HP-UX and HI-UX.
8927 Use GNU ld specific options. This passes @option{-shared} to ld when
8928 building a shared library. It is the default when GCC is configured,
8929 explicitly or implicitly, with the GNU linker. This option does not
8930 have any affect on which ld is called, it only changes what parameters
8931 are passed to that ld. The ld that is called is determined by the
8932 @option{--with-ld} configure option, gcc's program search path, and
8933 finally by the user's @env{PATH}. The linker used by GCC can be printed
8934 using @samp{which `gcc -print-prog-name=ld`}.
8938 Use HP ld specific options. This passes @option{-b} to ld when building
8939 a shared library and passes @option{+Accept TypeMismatch} to ld on all
8940 links. It is the default when GCC is configured, explicitly or
8941 implicitly, with the HP linker. This option does not have any affect on
8942 which ld is called, it only changes what parameters are passed to that
8943 ld. The ld that is called is determined by the @option{--with-ld}
8944 configure option, gcc's program search path, and finally by the user's
8945 @env{PATH}. The linker used by GCC can be printed using @samp{which
8946 `gcc -print-prog-name=ld`}.
8949 @opindex mno-long-calls
8950 Generate code that uses long call sequences. This ensures that a call
8951 is always able to reach linker generated stubs. The default is to generate
8952 long calls only when the distance from the call site to the beginning
8953 of the function or translation unit, as the case may be, exceeds a
8954 predefined limit set by the branch type being used. The limits for
8955 normal calls are 7,600,000 and 240,000 bytes, respectively for the
8956 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
8959 Distances are measured from the beginning of functions when using the
8960 @option{-ffunction-sections} option, or when using the @option{-mgas}
8961 and @option{-mno-portable-runtime} options together under HP-UX with
8964 It is normally not desirable to use this option as it will degrade
8965 performance. However, it may be useful in large applications,
8966 particularly when partial linking is used to build the application.
8968 The types of long calls used depends on the capabilities of the
8969 assembler and linker, and the type of code being generated. The
8970 impact on systems that support long absolute calls, and long pic
8971 symbol-difference or pc-relative calls should be relatively small.
8972 However, an indirect call is used on 32-bit ELF systems in pic code
8973 and it is quite long.
8977 Suppress the generation of link options to search libdld.sl when the
8978 @option{-static} option is specified on HP-UX 10 and later.
8982 The HP-UX implementation of setlocale in libc has a dependency on
8983 libdld.sl. There isn't an archive version of libdld.sl. Thus,
8984 when the @option{-static} option is specified, special link options
8985 are needed to resolve this dependency.
8987 On HP-UX 10 and later, the GCC driver adds the necessary options to
8988 link with libdld.sl when the @option{-static} option is specified.
8989 This causes the resulting binary to be dynamic. On the 64-bit port,
8990 the linkers generate dynamic binaries by default in any case. The
8991 @option{-nolibdld} option can be used to prevent the GCC driver from
8992 adding these link options.
8996 Add support for multithreading with the @dfn{dce thread} library
8997 under HP-UX. This option sets flags for both the preprocessor and
9001 @node Intel 960 Options
9002 @subsection Intel 960 Options
9004 These @samp{-m} options are defined for the Intel 960 implementations:
9007 @item -m@var{cpu-type}
9015 Assume the defaults for the machine type @var{cpu-type} for some of
9016 the other options, including instruction scheduling, floating point
9017 support, and addressing modes. The choices for @var{cpu-type} are
9018 @samp{ka}, @samp{kb}, @samp{mc}, @samp{ca}, @samp{cf},
9019 @samp{sa}, and @samp{sb}.
9026 @opindex msoft-float
9027 The @option{-mnumerics} option indicates that the processor does support
9028 floating-point instructions. The @option{-msoft-float} option indicates
9029 that floating-point support should not be assumed.
9031 @item -mleaf-procedures
9032 @itemx -mno-leaf-procedures
9033 @opindex mleaf-procedures
9034 @opindex mno-leaf-procedures
9035 Do (or do not) attempt to alter leaf procedures to be callable with the
9036 @code{bal} instruction as well as @code{call}. This will result in more
9037 efficient code for explicit calls when the @code{bal} instruction can be
9038 substituted by the assembler or linker, but less efficient code in other
9039 cases, such as calls via function pointers, or using a linker that doesn't
9040 support this optimization.
9043 @itemx -mno-tail-call
9045 @opindex mno-tail-call
9046 Do (or do not) make additional attempts (beyond those of the
9047 machine-independent portions of the compiler) to optimize tail-recursive
9048 calls into branches. You may not want to do this because the detection of
9049 cases where this is not valid is not totally complete. The default is
9050 @option{-mno-tail-call}.
9052 @item -mcomplex-addr
9053 @itemx -mno-complex-addr
9054 @opindex mcomplex-addr
9055 @opindex mno-complex-addr
9056 Assume (or do not assume) that the use of a complex addressing mode is a
9057 win on this implementation of the i960. Complex addressing modes may not
9058 be worthwhile on the K-series, but they definitely are on the C-series.
9059 The default is currently @option{-mcomplex-addr} for all processors except
9063 @itemx -mno-code-align
9064 @opindex mcode-align
9065 @opindex mno-code-align
9066 Align code to 8-byte boundaries for faster fetching (or don't bother).
9067 Currently turned on by default for C-series implementations only.
9070 @item -mclean-linkage
9071 @itemx -mno-clean-linkage
9072 @opindex mclean-linkage
9073 @opindex mno-clean-linkage
9074 These options are not fully implemented.
9078 @itemx -mic2.0-compat
9079 @itemx -mic3.0-compat
9081 @opindex mic2.0-compat
9082 @opindex mic3.0-compat
9083 Enable compatibility with iC960 v2.0 or v3.0.
9087 @opindex masm-compat
9089 Enable compatibility with the iC960 assembler.
9091 @item -mstrict-align
9092 @itemx -mno-strict-align
9093 @opindex mstrict-align
9094 @opindex mno-strict-align
9095 Do not permit (do permit) unaligned accesses.
9099 Enable structure-alignment compatibility with Intel's gcc release version
9100 1.3 (based on gcc 1.37). This option implies @option{-mstrict-align}.
9102 @item -mlong-double-64
9103 @opindex mlong-double-64
9104 Implement type @samp{long double} as 64-bit floating point numbers.
9105 Without the option @samp{long double} is implemented by 80-bit
9106 floating point numbers. The only reason we have it because there is
9107 no 128-bit @samp{long double} support in @samp{fp-bit.c} yet. So it
9108 is only useful for people using soft-float targets. Otherwise, we
9109 should recommend against use of it.
9113 @node DEC Alpha Options
9114 @subsection DEC Alpha Options
9116 These @samp{-m} options are defined for the DEC Alpha implementations:
9119 @item -mno-soft-float
9121 @opindex mno-soft-float
9122 @opindex msoft-float
9123 Use (do not use) the hardware floating-point instructions for
9124 floating-point operations. When @option{-msoft-float} is specified,
9125 functions in @file{libgcc.a} will be used to perform floating-point
9126 operations. Unless they are replaced by routines that emulate the
9127 floating-point operations, or compiled in such a way as to call such
9128 emulations routines, these routines will issue floating-point
9129 operations. If you are compiling for an Alpha without floating-point
9130 operations, you must ensure that the library is built so as not to call
9133 Note that Alpha implementations without floating-point operations are
9134 required to have floating-point registers.
9139 @opindex mno-fp-regs
9140 Generate code that uses (does not use) the floating-point register set.
9141 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
9142 register set is not used, floating point operands are passed in integer
9143 registers as if they were integers and floating-point results are passed
9144 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
9145 so any function with a floating-point argument or return value called by code
9146 compiled with @option{-mno-fp-regs} must also be compiled with that
9149 A typical use of this option is building a kernel that does not use,
9150 and hence need not save and restore, any floating-point registers.
9154 The Alpha architecture implements floating-point hardware optimized for
9155 maximum performance. It is mostly compliant with the IEEE floating
9156 point standard. However, for full compliance, software assistance is
9157 required. This option generates code fully IEEE compliant code
9158 @emph{except} that the @var{inexact-flag} is not maintained (see below).
9159 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
9160 defined during compilation. The resulting code is less efficient but is
9161 able to correctly support denormalized numbers and exceptional IEEE
9162 values such as not-a-number and plus/minus infinity. Other Alpha
9163 compilers call this option @option{-ieee_with_no_inexact}.
9165 @item -mieee-with-inexact
9166 @opindex mieee-with-inexact
9167 This is like @option{-mieee} except the generated code also maintains
9168 the IEEE @var{inexact-flag}. Turning on this option causes the
9169 generated code to implement fully-compliant IEEE math. In addition to
9170 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
9171 macro. On some Alpha implementations the resulting code may execute
9172 significantly slower than the code generated by default. Since there is
9173 very little code that depends on the @var{inexact-flag}, you should
9174 normally not specify this option. Other Alpha compilers call this
9175 option @option{-ieee_with_inexact}.
9177 @item -mfp-trap-mode=@var{trap-mode}
9178 @opindex mfp-trap-mode
9179 This option controls what floating-point related traps are enabled.
9180 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
9181 The trap mode can be set to one of four values:
9185 This is the default (normal) setting. The only traps that are enabled
9186 are the ones that cannot be disabled in software (e.g., division by zero
9190 In addition to the traps enabled by @samp{n}, underflow traps are enabled
9194 Like @samp{su}, but the instructions are marked to be safe for software
9195 completion (see Alpha architecture manual for details).
9198 Like @samp{su}, but inexact traps are enabled as well.
9201 @item -mfp-rounding-mode=@var{rounding-mode}
9202 @opindex mfp-rounding-mode
9203 Selects the IEEE rounding mode. Other Alpha compilers call this option
9204 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
9209 Normal IEEE rounding mode. Floating point numbers are rounded towards
9210 the nearest machine number or towards the even machine number in case
9214 Round towards minus infinity.
9217 Chopped rounding mode. Floating point numbers are rounded towards zero.
9220 Dynamic rounding mode. A field in the floating point control register
9221 (@var{fpcr}, see Alpha architecture reference manual) controls the
9222 rounding mode in effect. The C library initializes this register for
9223 rounding towards plus infinity. Thus, unless your program modifies the
9224 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
9227 @item -mtrap-precision=@var{trap-precision}
9228 @opindex mtrap-precision
9229 In the Alpha architecture, floating point traps are imprecise. This
9230 means without software assistance it is impossible to recover from a
9231 floating trap and program execution normally needs to be terminated.
9232 GCC can generate code that can assist operating system trap handlers
9233 in determining the exact location that caused a floating point trap.
9234 Depending on the requirements of an application, different levels of
9235 precisions can be selected:
9239 Program precision. This option is the default and means a trap handler
9240 can only identify which program caused a floating point exception.
9243 Function precision. The trap handler can determine the function that
9244 caused a floating point exception.
9247 Instruction precision. The trap handler can determine the exact
9248 instruction that caused a floating point exception.
9251 Other Alpha compilers provide the equivalent options called
9252 @option{-scope_safe} and @option{-resumption_safe}.
9254 @item -mieee-conformant
9255 @opindex mieee-conformant
9256 This option marks the generated code as IEEE conformant. You must not
9257 use this option unless you also specify @option{-mtrap-precision=i} and either
9258 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
9259 is to emit the line @samp{.eflag 48} in the function prologue of the
9260 generated assembly file. Under DEC Unix, this has the effect that
9261 IEEE-conformant math library routines will be linked in.
9263 @item -mbuild-constants
9264 @opindex mbuild-constants
9265 Normally GCC examines a 32- or 64-bit integer constant to
9266 see if it can construct it from smaller constants in two or three
9267 instructions. If it cannot, it will output the constant as a literal and
9268 generate code to load it from the data segment at runtime.
9270 Use this option to require GCC to construct @emph{all} integer constants
9271 using code, even if it takes more instructions (the maximum is six).
9273 You would typically use this option to build a shared library dynamic
9274 loader. Itself a shared library, it must relocate itself in memory
9275 before it can find the variables and constants in its own data segment.
9281 Select whether to generate code to be assembled by the vendor-supplied
9282 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
9300 Indicate whether GCC should generate code to use the optional BWX,
9301 CIX, FIX and MAX instruction sets. The default is to use the instruction
9302 sets supported by the CPU type specified via @option{-mcpu=} option or that
9303 of the CPU on which GCC was built if none was specified.
9308 @opindex mfloat-ieee
9309 Generate code that uses (does not use) VAX F and G floating point
9310 arithmetic instead of IEEE single and double precision.
9312 @item -mexplicit-relocs
9313 @itemx -mno-explicit-relocs
9314 @opindex mexplicit-relocs
9315 @opindex mno-explicit-relocs
9316 Older Alpha assemblers provided no way to generate symbol relocations
9317 except via assembler macros. Use of these macros does not allow
9318 optimal instruction scheduling. GNU binutils as of version 2.12
9319 supports a new syntax that allows the compiler to explicitly mark
9320 which relocations should apply to which instructions. This option
9321 is mostly useful for debugging, as GCC detects the capabilities of
9322 the assembler when it is built and sets the default accordingly.
9326 @opindex msmall-data
9327 @opindex mlarge-data
9328 When @option{-mexplicit-relocs} is in effect, static data is
9329 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
9330 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
9331 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
9332 16-bit relocations off of the @code{$gp} register. This limits the
9333 size of the small data area to 64KB, but allows the variables to be
9334 directly accessed via a single instruction.
9336 The default is @option{-mlarge-data}. With this option the data area
9337 is limited to just below 2GB. Programs that require more than 2GB of
9338 data must use @code{malloc} or @code{mmap} to allocate the data in the
9339 heap instead of in the program's data segment.
9341 When generating code for shared libraries, @option{-fpic} implies
9342 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
9346 @opindex msmall-text
9347 @opindex mlarge-text
9348 When @option{-msmall-text} is used, the compiler assumes that the
9349 code of the entire program (or shared library) fits in 4MB, and is
9350 thus reachable with a branch instruction. When @option{-msmall-data}
9351 is used, the compiler can assume that all local symbols share the
9352 same @code{$gp} value, and thus reduce the number of instructions
9353 required for a function call from 4 to 1.
9355 The default is @option{-mlarge-text}.
9357 @item -mcpu=@var{cpu_type}
9359 Set the instruction set and instruction scheduling parameters for
9360 machine type @var{cpu_type}. You can specify either the @samp{EV}
9361 style name or the corresponding chip number. GCC supports scheduling
9362 parameters for the EV4, EV5 and EV6 family of processors and will
9363 choose the default values for the instruction set from the processor
9364 you specify. If you do not specify a processor type, GCC will default
9365 to the processor on which the compiler was built.
9367 Supported values for @var{cpu_type} are
9373 Schedules as an EV4 and has no instruction set extensions.
9377 Schedules as an EV5 and has no instruction set extensions.
9381 Schedules as an EV5 and supports the BWX extension.
9386 Schedules as an EV5 and supports the BWX and MAX extensions.
9390 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
9394 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
9397 @item -mtune=@var{cpu_type}
9399 Set only the instruction scheduling parameters for machine type
9400 @var{cpu_type}. The instruction set is not changed.
9402 @item -mmemory-latency=@var{time}
9403 @opindex mmemory-latency
9404 Sets the latency the scheduler should assume for typical memory
9405 references as seen by the application. This number is highly
9406 dependent on the memory access patterns used by the application
9407 and the size of the external cache on the machine.
9409 Valid options for @var{time} are
9413 A decimal number representing clock cycles.
9419 The compiler contains estimates of the number of clock cycles for
9420 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
9421 (also called Dcache, Scache, and Bcache), as well as to main memory.
9422 Note that L3 is only valid for EV5.
9427 @node DEC Alpha/VMS Options
9428 @subsection DEC Alpha/VMS Options
9430 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
9433 @item -mvms-return-codes
9434 @opindex mvms-return-codes
9435 Return VMS condition codes from main. The default is to return POSIX
9436 style condition (e.g.@ error) codes.
9439 @node H8/300 Options
9440 @subsection H8/300 Options
9442 These @samp{-m} options are defined for the H8/300 implementations:
9447 Shorten some address references at link time, when possible; uses the
9448 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
9449 ld, Using ld}, for a fuller description.
9453 Generate code for the H8/300H@.
9457 Generate code for the H8S@.
9461 Generate code for the H8S and H8/300H in the normal mode. This switch
9462 must be used either with -mh or -ms.
9466 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
9470 Make @code{int} data 32 bits by default.
9474 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
9475 The default for the H8/300H and H8S is to align longs and floats on 4
9477 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
9478 This option has no effect on the H8/300.
9482 @subsection SH Options
9484 These @samp{-m} options are defined for the SH implementations:
9489 Generate code for the SH1.
9493 Generate code for the SH2.
9496 Generate code for the SH2e.
9500 Generate code for the SH3.
9504 Generate code for the SH3e.
9508 Generate code for the SH4 without a floating-point unit.
9510 @item -m4-single-only
9511 @opindex m4-single-only
9512 Generate code for the SH4 with a floating-point unit that only
9513 supports single-precision arithmetic.
9517 Generate code for the SH4 assuming the floating-point unit is in
9518 single-precision mode by default.
9522 Generate code for the SH4.
9526 Compile code for the processor in big endian mode.
9530 Compile code for the processor in little endian mode.
9534 Align doubles at 64-bit boundaries. Note that this changes the calling
9535 conventions, and thus some functions from the standard C library will
9536 not work unless you recompile it first with @option{-mdalign}.
9540 Shorten some address references at link time, when possible; uses the
9541 linker option @option{-relax}.
9545 Use 32-bit offsets in @code{switch} tables. The default is to use
9550 Enable the use of the instruction @code{fmovd}.
9554 Comply with the calling conventions defined by Renesas.
9558 Mark the @code{MAC} register as call-clobbered, even if
9559 @option{-mhitachi} is given.
9563 Increase IEEE-compliance of floating-point code.
9567 Dump instruction size and location in the assembly code.
9571 This option is deprecated. It pads structures to multiple of 4 bytes,
9572 which is incompatible with the SH ABI@.
9576 Optimize for space instead of speed. Implied by @option{-Os}.
9580 When generating position-independent code, emit function calls using
9581 the Global Offset Table instead of the Procedure Linkage Table.
9585 Generate a library function call to invalidate instruction cache
9586 entries, after fixing up a trampoline. This library function call
9587 doesn't assume it can write to the whole memory address space. This
9588 is the default when the target is @code{sh-*-linux*}.
9591 @node System V Options
9592 @subsection Options for System V
9594 These additional options are available on System V Release 4 for
9595 compatibility with other compilers on those systems:
9600 Create a shared object.
9601 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
9605 Identify the versions of each tool used by the compiler, in a
9606 @code{.ident} assembler directive in the output.
9610 Refrain from adding @code{.ident} directives to the output file (this is
9613 @item -YP,@var{dirs}
9615 Search the directories @var{dirs}, and no others, for libraries
9616 specified with @option{-l}.
9620 Look in the directory @var{dir} to find the M4 preprocessor.
9621 The assembler uses this option.
9622 @c This is supposed to go with a -Yd for predefined M4 macro files, but
9623 @c the generic assembler that comes with Solaris takes just -Ym.
9626 @node TMS320C3x/C4x Options
9627 @subsection TMS320C3x/C4x Options
9628 @cindex TMS320C3x/C4x Options
9630 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
9634 @item -mcpu=@var{cpu_type}
9636 Set the instruction set, register set, and instruction scheduling
9637 parameters for machine type @var{cpu_type}. Supported values for
9638 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
9639 @samp{c44}. The default is @samp{c40} to generate code for the
9644 @itemx -msmall-memory
9646 @opindex mbig-memory
9648 @opindex msmall-memory
9650 Generates code for the big or small memory model. The small memory
9651 model assumed that all data fits into one 64K word page. At run-time
9652 the data page (DP) register must be set to point to the 64K page
9653 containing the .bss and .data program sections. The big memory model is
9654 the default and requires reloading of the DP register for every direct
9661 Allow (disallow) allocation of general integer operands into the block
9668 Enable (disable) generation of code using decrement and branch,
9669 DBcond(D), instructions. This is enabled by default for the C4x. To be
9670 on the safe side, this is disabled for the C3x, since the maximum
9671 iteration count on the C3x is @math{2^{23} + 1} (but who iterates loops more than
9672 @math{2^{23}} times on the C3x?). Note that GCC will try to reverse a loop so
9673 that it can utilize the decrement and branch instruction, but will give
9674 up if there is more than one memory reference in the loop. Thus a loop
9675 where the loop counter is decremented can generate slightly more
9676 efficient code, in cases where the RPTB instruction cannot be utilized.
9678 @item -mdp-isr-reload
9680 @opindex mdp-isr-reload
9682 Force the DP register to be saved on entry to an interrupt service
9683 routine (ISR), reloaded to point to the data section, and restored on
9684 exit from the ISR@. This should not be required unless someone has
9685 violated the small memory model by modifying the DP register, say within
9692 For the C3x use the 24-bit MPYI instruction for integer multiplies
9693 instead of a library call to guarantee 32-bit results. Note that if one
9694 of the operands is a constant, then the multiplication will be performed
9695 using shifts and adds. If the @option{-mmpyi} option is not specified for the C3x,
9696 then squaring operations are performed inline instead of a library call.
9699 @itemx -mno-fast-fix
9701 @opindex mno-fast-fix
9702 The C3x/C4x FIX instruction to convert a floating point value to an
9703 integer value chooses the nearest integer less than or equal to the
9704 floating point value rather than to the nearest integer. Thus if the
9705 floating point number is negative, the result will be incorrectly
9706 truncated an additional code is necessary to detect and correct this
9707 case. This option can be used to disable generation of the additional
9708 code required to correct the result.
9714 Enable (disable) generation of repeat block sequences using the RPTB
9715 instruction for zero overhead looping. The RPTB construct is only used
9716 for innermost loops that do not call functions or jump across the loop
9717 boundaries. There is no advantage having nested RPTB loops due to the
9718 overhead required to save and restore the RC, RS, and RE registers.
9719 This is enabled by default with @option{-O2}.
9721 @item -mrpts=@var{count}
9725 Enable (disable) the use of the single instruction repeat instruction
9726 RPTS@. If a repeat block contains a single instruction, and the loop
9727 count can be guaranteed to be less than the value @var{count}, GCC will
9728 emit a RPTS instruction instead of a RPTB@. If no value is specified,
9729 then a RPTS will be emitted even if the loop count cannot be determined
9730 at compile time. Note that the repeated instruction following RPTS does
9731 not have to be reloaded from memory each iteration, thus freeing up the
9732 CPU buses for operands. However, since interrupts are blocked by this
9733 instruction, it is disabled by default.
9735 @item -mloop-unsigned
9736 @itemx -mno-loop-unsigned
9737 @opindex mloop-unsigned
9738 @opindex mno-loop-unsigned
9739 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
9740 is @math{2^{31} + 1} since these instructions test if the iteration count is
9741 negative to terminate the loop. If the iteration count is unsigned
9742 there is a possibility than the @math{2^{31} + 1} maximum iteration count may be
9743 exceeded. This switch allows an unsigned iteration count.
9747 Try to emit an assembler syntax that the TI assembler (asm30) is happy
9748 with. This also enforces compatibility with the API employed by the TI
9749 C3x C compiler. For example, long doubles are passed as structures
9750 rather than in floating point registers.
9756 Generate code that uses registers (stack) for passing arguments to functions.
9757 By default, arguments are passed in registers where possible rather
9758 than by pushing arguments on to the stack.
9760 @item -mparallel-insns
9761 @itemx -mno-parallel-insns
9762 @opindex mparallel-insns
9763 @opindex mno-parallel-insns
9764 Allow the generation of parallel instructions. This is enabled by
9765 default with @option{-O2}.
9767 @item -mparallel-mpy
9768 @itemx -mno-parallel-mpy
9769 @opindex mparallel-mpy
9770 @opindex mno-parallel-mpy
9771 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
9772 provided @option{-mparallel-insns} is also specified. These instructions have
9773 tight register constraints which can pessimize the code generation
9779 @subsection V850 Options
9780 @cindex V850 Options
9782 These @samp{-m} options are defined for V850 implementations:
9786 @itemx -mno-long-calls
9787 @opindex mlong-calls
9788 @opindex mno-long-calls
9789 Treat all calls as being far away (near). If calls are assumed to be
9790 far away, the compiler will always load the functions address up into a
9791 register, and call indirect through the pointer.
9797 Do not optimize (do optimize) basic blocks that use the same index
9798 pointer 4 or more times to copy pointer into the @code{ep} register, and
9799 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
9800 option is on by default if you optimize.
9802 @item -mno-prolog-function
9803 @itemx -mprolog-function
9804 @opindex mno-prolog-function
9805 @opindex mprolog-function
9806 Do not use (do use) external functions to save and restore registers
9807 at the prologue and epilogue of a function. The external functions
9808 are slower, but use less code space if more than one function saves
9809 the same number of registers. The @option{-mprolog-function} option
9810 is on by default if you optimize.
9814 Try to make the code as small as possible. At present, this just turns
9815 on the @option{-mep} and @option{-mprolog-function} options.
9819 Put static or global variables whose size is @var{n} bytes or less into
9820 the tiny data area that register @code{ep} points to. The tiny data
9821 area can hold up to 256 bytes in total (128 bytes for byte references).
9825 Put static or global variables whose size is @var{n} bytes or less into
9826 the small data area that register @code{gp} points to. The small data
9827 area can hold up to 64 kilobytes.
9831 Put static or global variables whose size is @var{n} bytes or less into
9832 the first 32 kilobytes of memory.
9836 Specify that the target processor is the V850.
9839 @opindex mbig-switch
9840 Generate code suitable for big switch tables. Use this option only if
9841 the assembler/linker complain about out of range branches within a switch
9846 This option will cause r2 and r5 to be used in the code generated by
9847 the compiler. This setting is the default.
9850 @opindex mno-app-regs
9851 This option will cause r2 and r5 to be treated as fixed registers.
9855 Specify that the target processor is the V850E1. The preprocessor
9856 constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
9857 this option is used.
9861 Specify that the target processor is the V850E. The preprocessor
9862 constant @samp{__v850e__} will be defined if this option is used.
9864 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
9865 are defined then a default target processor will be chosen and the
9866 relevant @samp{__v850*__} preprocessor constant will be defined.
9868 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
9869 defined, regardless of which processor variant is the target.
9871 @item -mdisable-callt
9872 @opindex mdisable-callt
9873 This option will suppress generation of the CALLT instruction for the
9874 v850e and v850e1 flavors of the v850 architecture. The default is
9875 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
9880 @subsection ARC Options
9883 These options are defined for ARC implementations:
9888 Compile code for little endian mode. This is the default.
9892 Compile code for big endian mode.
9895 @opindex mmangle-cpu
9896 Prepend the name of the cpu to all public symbol names.
9897 In multiple-processor systems, there are many ARC variants with different
9898 instruction and register set characteristics. This flag prevents code
9899 compiled for one cpu to be linked with code compiled for another.
9900 No facility exists for handling variants that are ``almost identical''.
9901 This is an all or nothing option.
9903 @item -mcpu=@var{cpu}
9905 Compile code for ARC variant @var{cpu}.
9906 Which variants are supported depend on the configuration.
9907 All variants support @option{-mcpu=base}, this is the default.
9909 @item -mtext=@var{text-section}
9910 @itemx -mdata=@var{data-section}
9911 @itemx -mrodata=@var{readonly-data-section}
9915 Put functions, data, and readonly data in @var{text-section},
9916 @var{data-section}, and @var{readonly-data-section} respectively
9917 by default. This can be overridden with the @code{section} attribute.
9918 @xref{Variable Attributes}.
9923 @subsection NS32K Options
9924 @cindex NS32K options
9926 These are the @samp{-m} options defined for the 32000 series. The default
9927 values for these options depends on which style of 32000 was selected when
9928 the compiler was configured; the defaults for the most common choices are
9936 Generate output for a 32032. This is the default
9937 when the compiler is configured for 32032 and 32016 based systems.
9943 Generate output for a 32332. This is the default
9944 when the compiler is configured for 32332-based systems.
9950 Generate output for a 32532. This is the default
9951 when the compiler is configured for 32532-based systems.
9955 Generate output containing 32081 instructions for floating point.
9956 This is the default for all systems.
9960 Generate output containing 32381 instructions for floating point. This
9961 also implies @option{-m32081}. The 32381 is only compatible with the 32332
9962 and 32532 cpus. This is the default for the pc532-netbsd configuration.
9966 Try and generate multiply-add floating point instructions @code{polyF}
9967 and @code{dotF}. This option is only available if the @option{-m32381}
9968 option is in effect. Using these instructions requires changes to
9969 register allocation which generally has a negative impact on
9970 performance. This option should only be enabled when compiling code
9971 particularly likely to make heavy use of multiply-add instructions.
9974 @opindex mnomulti-add
9975 Do not try and generate multiply-add floating point instructions
9976 @code{polyF} and @code{dotF}. This is the default on all platforms.
9979 @opindex msoft-float
9980 Generate output containing library calls for floating point.
9981 @strong{Warning:} the requisite libraries may not be available.
9983 @item -mieee-compare
9984 @itemx -mno-ieee-compare
9985 @opindex mieee-compare
9986 @opindex mno-ieee-compare
9987 Control whether or not the compiler uses IEEE floating point
9988 comparisons. These handle correctly the case where the result of a
9989 comparison is unordered.
9990 @strong{Warning:} the requisite kernel support may not be available.
9993 @opindex mnobitfield
9994 Do not use the bit-field instructions. On some machines it is faster to
9995 use shifting and masking operations. This is the default for the pc532.
9999 Do use the bit-field instructions. This is the default for all platforms
10004 Use a different function-calling convention, in which functions
10005 that take a fixed number of arguments return pop their
10006 arguments on return with the @code{ret} instruction.
10008 This calling convention is incompatible with the one normally
10009 used on Unix, so you cannot use it if you need to call libraries
10010 compiled with the Unix compiler.
10012 Also, you must provide function prototypes for all functions that
10013 take variable numbers of arguments (including @code{printf});
10014 otherwise incorrect code will be generated for calls to those
10017 In addition, seriously incorrect code will result if you call a
10018 function with too many arguments. (Normally, extra arguments are
10019 harmlessly ignored.)
10021 This option takes its name from the 680x0 @code{rtd} instruction.
10026 Use a different function-calling convention where the first two arguments
10027 are passed in registers.
10029 This calling convention is incompatible with the one normally
10030 used on Unix, so you cannot use it if you need to call libraries
10031 compiled with the Unix compiler.
10034 @opindex mnoregparam
10035 Do not pass any arguments in registers. This is the default for all
10040 It is OK to use the sb as an index register which is always loaded with
10041 zero. This is the default for the pc532-netbsd target.
10045 The sb register is not available for use or has not been initialized to
10046 zero by the run time system. This is the default for all targets except
10047 the pc532-netbsd. It is also implied whenever @option{-mhimem} or
10048 @option{-fpic} is set.
10052 Many ns32000 series addressing modes use displacements of up to 512MB@.
10053 If an address is above 512MB then displacements from zero can not be used.
10054 This option causes code to be generated which can be loaded above 512MB@.
10055 This may be useful for operating systems or ROM code.
10059 Assume code will be loaded in the first 512MB of virtual address space.
10060 This is the default for all platforms.
10066 @subsection AVR Options
10067 @cindex AVR Options
10069 These options are defined for AVR implementations:
10072 @item -mmcu=@var{mcu}
10074 Specify ATMEL AVR instruction set or MCU type.
10076 Instruction set avr1 is for the minimal AVR core, not supported by the C
10077 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
10078 attiny11, attiny12, attiny15, attiny28).
10080 Instruction set avr2 (default) is for the classic AVR core with up to
10081 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
10082 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
10083 at90c8534, at90s8535).
10085 Instruction set avr3 is for the classic AVR core with up to 128K program
10086 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
10088 Instruction set avr4 is for the enhanced AVR core with up to 8K program
10089 memory space (MCU types: atmega8, atmega83, atmega85).
10091 Instruction set avr5 is for the enhanced AVR core with up to 128K program
10092 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
10093 atmega64, atmega128, at43usb355, at94k).
10097 Output instruction sizes to the asm file.
10099 @item -minit-stack=@var{N}
10100 @opindex minit-stack
10101 Specify the initial stack address, which may be a symbol or numeric value,
10102 @samp{__stack} is the default.
10104 @item -mno-interrupts
10105 @opindex mno-interrupts
10106 Generated code is not compatible with hardware interrupts.
10107 Code size will be smaller.
10109 @item -mcall-prologues
10110 @opindex mcall-prologues
10111 Functions prologues/epilogues expanded as call to appropriate
10112 subroutines. Code size will be smaller.
10114 @item -mno-tablejump
10115 @opindex mno-tablejump
10116 Do not generate tablejump insns which sometimes increase code size.
10119 @opindex mtiny-stack
10120 Change only the low 8 bits of the stack pointer.
10123 @node MCore Options
10124 @subsection MCore Options
10125 @cindex MCore options
10127 These are the @samp{-m} options defined for the Motorola M*Core
10133 @itemx -mno-hardlit
10135 @opindex mno-hardlit
10136 Inline constants into the code stream if it can be done in two
10137 instructions or less.
10143 Use the divide instruction. (Enabled by default).
10145 @item -mrelax-immediate
10146 @itemx -mno-relax-immediate
10147 @opindex mrelax-immediate
10148 @opindex mno-relax-immediate
10149 Allow arbitrary sized immediates in bit operations.
10151 @item -mwide-bitfields
10152 @itemx -mno-wide-bitfields
10153 @opindex mwide-bitfields
10154 @opindex mno-wide-bitfields
10155 Always treat bit-fields as int-sized.
10157 @item -m4byte-functions
10158 @itemx -mno-4byte-functions
10159 @opindex m4byte-functions
10160 @opindex mno-4byte-functions
10161 Force all functions to be aligned to a four byte boundary.
10163 @item -mcallgraph-data
10164 @itemx -mno-callgraph-data
10165 @opindex mcallgraph-data
10166 @opindex mno-callgraph-data
10167 Emit callgraph information.
10170 @itemx -mno-slow-bytes
10171 @opindex mslow-bytes
10172 @opindex mno-slow-bytes
10173 Prefer word access when reading byte quantities.
10175 @item -mlittle-endian
10176 @itemx -mbig-endian
10177 @opindex mlittle-endian
10178 @opindex mbig-endian
10179 Generate code for a little endian target.
10185 Generate code for the 210 processor.
10188 @node IA-64 Options
10189 @subsection IA-64 Options
10190 @cindex IA-64 Options
10192 These are the @samp{-m} options defined for the Intel IA-64 architecture.
10196 @opindex mbig-endian
10197 Generate code for a big endian target. This is the default for HP-UX@.
10199 @item -mlittle-endian
10200 @opindex mlittle-endian
10201 Generate code for a little endian target. This is the default for AIX5
10207 @opindex mno-gnu-as
10208 Generate (or don't) code for the GNU assembler. This is the default.
10209 @c Also, this is the default if the configure option @option{--with-gnu-as}
10215 @opindex mno-gnu-ld
10216 Generate (or don't) code for the GNU linker. This is the default.
10217 @c Also, this is the default if the configure option @option{--with-gnu-ld}
10222 Generate code that does not use a global pointer register. The result
10223 is not position independent code, and violates the IA-64 ABI@.
10225 @item -mvolatile-asm-stop
10226 @itemx -mno-volatile-asm-stop
10227 @opindex mvolatile-asm-stop
10228 @opindex mno-volatile-asm-stop
10229 Generate (or don't) a stop bit immediately before and after volatile asm
10234 Generate code that works around Itanium B step errata.
10236 @item -mregister-names
10237 @itemx -mno-register-names
10238 @opindex mregister-names
10239 @opindex mno-register-names
10240 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
10241 the stacked registers. This may make assembler output more readable.
10247 Disable (or enable) optimizations that use the small data section. This may
10248 be useful for working around optimizer bugs.
10250 @item -mconstant-gp
10251 @opindex mconstant-gp
10252 Generate code that uses a single constant global pointer value. This is
10253 useful when compiling kernel code.
10257 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
10258 This is useful when compiling firmware code.
10260 @item -minline-float-divide-min-latency
10261 @opindex minline-float-divide-min-latency
10262 Generate code for inline divides of floating point values
10263 using the minimum latency algorithm.
10265 @item -minline-float-divide-max-throughput
10266 @opindex minline-float-divide-max-throughput
10267 Generate code for inline divides of floating point values
10268 using the maximum throughput algorithm.
10270 @item -minline-int-divide-min-latency
10271 @opindex minline-int-divide-min-latency
10272 Generate code for inline divides of integer values
10273 using the minimum latency algorithm.
10275 @item -minline-int-divide-max-throughput
10276 @opindex minline-int-divide-max-throughput
10277 Generate code for inline divides of integer values
10278 using the maximum throughput algorithm.
10280 @item -mno-dwarf2-asm
10281 @itemx -mdwarf2-asm
10282 @opindex mno-dwarf2-asm
10283 @opindex mdwarf2-asm
10284 Don't (or do) generate assembler code for the DWARF2 line number debugging
10285 info. This may be useful when not using the GNU assembler.
10287 @item -mfixed-range=@var{register-range}
10288 @opindex mfixed-range
10289 Generate code treating the given register range as fixed registers.
10290 A fixed register is one that the register allocator can not use. This is
10291 useful when compiling kernel code. A register range is specified as
10292 two registers separated by a dash. Multiple register ranges can be
10293 specified separated by a comma.
10295 @item -mearly-stop-bits
10296 @itemx -mno-early-stop-bits
10297 @opindex mearly-stop-bits
10298 @opindex mno-early-stop-bits
10299 Allow stop bits to be placed earlier than immediately preceding the
10300 instruction that triggered the stop bit. This can improve instruction
10301 scheduling, but does not always do so.
10305 @subsection D30V Options
10306 @cindex D30V Options
10308 These @samp{-m} options are defined for D30V implementations:
10313 Link the @samp{.text}, @samp{.data}, @samp{.bss}, @samp{.strings},
10314 @samp{.rodata}, @samp{.rodata1}, @samp{.data1} sections into external
10315 memory, which starts at location @code{0x80000000}.
10318 @opindex mextmemory
10319 Same as the @option{-mextmem} switch.
10323 Link the @samp{.text} section into onchip text memory, which starts at
10324 location @code{0x0}. Also link @samp{.data}, @samp{.bss},
10325 @samp{.strings}, @samp{.rodata}, @samp{.rodata1}, @samp{.data1} sections
10326 into onchip data memory, which starts at location @code{0x20000000}.
10328 @item -mno-asm-optimize
10329 @itemx -masm-optimize
10330 @opindex mno-asm-optimize
10331 @opindex masm-optimize
10332 Disable (enable) passing @option{-O} to the assembler when optimizing.
10333 The assembler uses the @option{-O} option to automatically parallelize
10334 adjacent short instructions where possible.
10336 @item -mbranch-cost=@var{n}
10337 @opindex mbranch-cost
10338 Increase the internal costs of branches to @var{n}. Higher costs means
10339 that the compiler will issue more instructions to avoid doing a branch.
10342 @item -mcond-exec=@var{n}
10343 @opindex mcond-exec
10344 Specify the maximum number of conditionally executed instructions that
10345 replace a branch. The default is 4.
10348 @node S/390 and zSeries Options
10349 @subsection S/390 and zSeries Options
10350 @cindex S/390 and zSeries Options
10352 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
10356 @itemx -msoft-float
10357 @opindex mhard-float
10358 @opindex msoft-float
10359 Use (do not use) the hardware floating-point instructions and registers
10360 for floating-point operations. When @option{-msoft-float} is specified,
10361 functions in @file{libgcc.a} will be used to perform floating-point
10362 operations. When @option{-mhard-float} is specified, the compiler
10363 generates IEEE floating-point instructions. This is the default.
10366 @itemx -mno-backchain
10367 @opindex mbackchain
10368 @opindex mno-backchain
10369 Generate (or do not generate) code which maintains an explicit
10370 backchain within the stack frame that points to the caller's frame.
10371 This is currently needed to allow debugging. The default is to
10372 generate the backchain.
10375 @itemx -mno-small-exec
10376 @opindex msmall-exec
10377 @opindex mno-small-exec
10378 Generate (or do not generate) code using the @code{bras} instruction
10379 to do subroutine calls.
10380 This only works reliably if the total executable size does not
10381 exceed 64k. The default is to use the @code{basr} instruction instead,
10382 which does not have this limitation.
10388 When @option{-m31} is specified, generate code compliant to the
10389 Linux for S/390 ABI@. When @option{-m64} is specified, generate
10390 code compliant to the Linux for zSeries ABI@. This allows GCC in
10391 particular to generate 64-bit instructions. For the @samp{s390}
10392 targets, the default is @option{-m31}, while the @samp{s390x}
10393 targets default to @option{-m64}.
10399 When @option{-mzarch} is specified, generate code using the
10400 instructions available on z/Architecture.
10401 When @option{-mesa} is specified, generate code using the
10402 instructions available on ESA/390. Note that @option{-mesa} is
10403 not possible with @option{-m64}.
10404 When generating code compliant to the Linux for S/390 ABI,
10405 the default is @option{-mesa}. When generating code compliant
10406 to the Linux for zSeries ABI, the default is @option{-mzarch}.
10412 Generate (or do not generate) code using the @code{mvcle} instruction
10413 to perform block moves. When @option{-mno-mvcle} is specified,
10414 use a @code{mvc} loop instead. This is the default.
10420 Print (or do not print) additional debug information when compiling.
10421 The default is to not print debug information.
10423 @item -march=@var{cpu-type}
10425 Generate code that will run on @var{cpu-type}, which is the name of a system
10426 representing a certain processor type. Possible values for
10427 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, and @samp{z990}.
10428 When generating code using the instructions available on z/Architecture,
10429 the default is @option{-march=z900}. Otherwise, the default is
10430 @option{-march=g5}.
10432 @item -mtune=@var{cpu-type}
10434 Tune to @var{cpu-type} everything applicable about the generated code,
10435 except for the ABI and the set of available instructions.
10436 The list of @var{cpu-type} values is the same as for @option{-march}.
10437 The default is the value used for @option{-march}.
10440 @itemx -mno-fused-madd
10441 @opindex mfused-madd
10442 @opindex mno-fused-madd
10443 Generate code that uses (does not use) the floating point multiply and
10444 accumulate instructions. These instructions are generated by default if
10445 hardware floating point is used.
10449 @subsection CRIS Options
10450 @cindex CRIS Options
10452 These options are defined specifically for the CRIS ports.
10455 @item -march=@var{architecture-type}
10456 @itemx -mcpu=@var{architecture-type}
10459 Generate code for the specified architecture. The choices for
10460 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
10461 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX.
10462 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
10465 @item -mtune=@var{architecture-type}
10467 Tune to @var{architecture-type} everything applicable about the generated
10468 code, except for the ABI and the set of available instructions. The
10469 choices for @var{architecture-type} are the same as for
10470 @option{-march=@var{architecture-type}}.
10472 @item -mmax-stack-frame=@var{n}
10473 @opindex mmax-stack-frame
10474 Warn when the stack frame of a function exceeds @var{n} bytes.
10476 @item -melinux-stacksize=@var{n}
10477 @opindex melinux-stacksize
10478 Only available with the @samp{cris-axis-aout} target. Arranges for
10479 indications in the program to the kernel loader that the stack of the
10480 program should be set to @var{n} bytes.
10486 The options @option{-metrax4} and @option{-metrax100} are synonyms for
10487 @option{-march=v3} and @option{-march=v8} respectively.
10491 Enable CRIS-specific verbose debug-related information in the assembly
10492 code. This option also has the effect to turn off the @samp{#NO_APP}
10493 formatted-code indicator to the assembler at the beginning of the
10498 Do not use condition-code results from previous instruction; always emit
10499 compare and test instructions before use of condition codes.
10501 @item -mno-side-effects
10502 @opindex mno-side-effects
10503 Do not emit instructions with side-effects in addressing modes other than
10506 @item -mstack-align
10507 @itemx -mno-stack-align
10508 @itemx -mdata-align
10509 @itemx -mno-data-align
10510 @itemx -mconst-align
10511 @itemx -mno-const-align
10512 @opindex mstack-align
10513 @opindex mno-stack-align
10514 @opindex mdata-align
10515 @opindex mno-data-align
10516 @opindex mconst-align
10517 @opindex mno-const-align
10518 These options (no-options) arranges (eliminate arrangements) for the
10519 stack-frame, individual data and constants to be aligned for the maximum
10520 single data access size for the chosen CPU model. The default is to
10521 arrange for 32-bit alignment. ABI details such as structure layout are
10522 not affected by these options.
10530 Similar to the stack- data- and const-align options above, these options
10531 arrange for stack-frame, writable data and constants to all be 32-bit,
10532 16-bit or 8-bit aligned. The default is 32-bit alignment.
10534 @item -mno-prologue-epilogue
10535 @itemx -mprologue-epilogue
10536 @opindex mno-prologue-epilogue
10537 @opindex mprologue-epilogue
10538 With @option{-mno-prologue-epilogue}, the normal function prologue and
10539 epilogue that sets up the stack-frame are omitted and no return
10540 instructions or return sequences are generated in the code. Use this
10541 option only together with visual inspection of the compiled code: no
10542 warnings or errors are generated when call-saved registers must be saved,
10543 or storage for local variable needs to be allocated.
10547 @opindex mno-gotplt
10549 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
10550 instruction sequences that load addresses for functions from the PLT part
10551 of the GOT rather than (traditional on other architectures) calls to the
10552 PLT. The default is @option{-mgotplt}.
10556 Legacy no-op option only recognized with the cris-axis-aout target.
10560 Legacy no-op option only recognized with the cris-axis-elf and
10561 cris-axis-linux-gnu targets.
10565 Only recognized with the cris-axis-aout target, where it selects a
10566 GNU/linux-like multilib, include files and instruction set for
10567 @option{-march=v8}.
10571 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
10575 This option, recognized for the cris-axis-aout and cris-axis-elf arranges
10576 to link with input-output functions from a simulator library. Code,
10577 initialized data and zero-initialized data are allocated consecutively.
10581 Like @option{-sim}, but pass linker options to locate initialized data at
10582 0x40000000 and zero-initialized data at 0x80000000.
10586 @subsection MMIX Options
10587 @cindex MMIX Options
10589 These options are defined for the MMIX:
10593 @itemx -mno-libfuncs
10595 @opindex mno-libfuncs
10596 Specify that intrinsic library functions are being compiled, passing all
10597 values in registers, no matter the size.
10600 @itemx -mno-epsilon
10602 @opindex mno-epsilon
10603 Generate floating-point comparison instructions that compare with respect
10604 to the @code{rE} epsilon register.
10606 @item -mabi=mmixware
10608 @opindex mabi-mmixware
10610 Generate code that passes function parameters and return values that (in
10611 the called function) are seen as registers @code{$0} and up, as opposed to
10612 the GNU ABI which uses global registers @code{$231} and up.
10614 @item -mzero-extend
10615 @itemx -mno-zero-extend
10616 @opindex mzero-extend
10617 @opindex mno-zero-extend
10618 When reading data from memory in sizes shorter than 64 bits, use (do not
10619 use) zero-extending load instructions by default, rather than
10620 sign-extending ones.
10623 @itemx -mno-knuthdiv
10625 @opindex mno-knuthdiv
10626 Make the result of a division yielding a remainder have the same sign as
10627 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
10628 remainder follows the sign of the dividend. Both methods are
10629 arithmetically valid, the latter being almost exclusively used.
10631 @item -mtoplevel-symbols
10632 @itemx -mno-toplevel-symbols
10633 @opindex mtoplevel-symbols
10634 @opindex mno-toplevel-symbols
10635 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
10636 code can be used with the @code{PREFIX} assembly directive.
10640 Generate an executable in the ELF format, rather than the default
10641 @samp{mmo} format used by the @command{mmix} simulator.
10643 @item -mbranch-predict
10644 @itemx -mno-branch-predict
10645 @opindex mbranch-predict
10646 @opindex mno-branch-predict
10647 Use (do not use) the probable-branch instructions, when static branch
10648 prediction indicates a probable branch.
10650 @item -mbase-addresses
10651 @itemx -mno-base-addresses
10652 @opindex mbase-addresses
10653 @opindex mno-base-addresses
10654 Generate (do not generate) code that uses @emph{base addresses}. Using a
10655 base address automatically generates a request (handled by the assembler
10656 and the linker) for a constant to be set up in a global register. The
10657 register is used for one or more base address requests within the range 0
10658 to 255 from the value held in the register. The generally leads to short
10659 and fast code, but the number of different data items that can be
10660 addressed is limited. This means that a program that uses lots of static
10661 data may require @option{-mno-base-addresses}.
10663 @item -msingle-exit
10664 @itemx -mno-single-exit
10665 @opindex msingle-exit
10666 @opindex mno-single-exit
10667 Force (do not force) generated code to have a single exit point in each
10671 @node PDP-11 Options
10672 @subsection PDP-11 Options
10673 @cindex PDP-11 Options
10675 These options are defined for the PDP-11:
10680 Use hardware FPP floating point. This is the default. (FIS floating
10681 point on the PDP-11/40 is not supported.)
10684 @opindex msoft-float
10685 Do not use hardware floating point.
10689 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
10693 Return floating-point results in memory. This is the default.
10697 Generate code for a PDP-11/40.
10701 Generate code for a PDP-11/45. This is the default.
10705 Generate code for a PDP-11/10.
10707 @item -mbcopy-builtin
10708 @opindex bcopy-builtin
10709 Use inline @code{movstrhi} patterns for copying memory. This is the
10714 Do not use inline @code{movstrhi} patterns for copying memory.
10720 Use 16-bit @code{int}. This is the default.
10726 Use 32-bit @code{int}.
10729 @itemx -mno-float32
10731 @opindex mno-float32
10732 Use 64-bit @code{float}. This is the default.
10737 @opindex mno-float64
10738 Use 32-bit @code{float}.
10742 Use @code{abshi2} pattern. This is the default.
10746 Do not use @code{abshi2} pattern.
10748 @item -mbranch-expensive
10749 @opindex mbranch-expensive
10750 Pretend that branches are expensive. This is for experimenting with
10751 code generation only.
10753 @item -mbranch-cheap
10754 @opindex mbranch-cheap
10755 Do not pretend that branches are expensive. This is the default.
10759 Generate code for a system with split I&D.
10763 Generate code for a system without split I&D. This is the default.
10767 Use Unix assembler syntax. This is the default when configured for
10768 @samp{pdp11-*-bsd}.
10772 Use DEC assembler syntax. This is the default when configured for any
10773 PDP-11 target other than @samp{pdp11-*-bsd}.
10776 @node Xstormy16 Options
10777 @subsection Xstormy16 Options
10778 @cindex Xstormy16 Options
10780 These options are defined for Xstormy16:
10785 Choose startup files and linker script suitable for the simulator.
10789 @subsection FRV Options
10790 @cindex FRV Options
10796 Only use the first 32 general purpose registers.
10801 Use all 64 general purpose registers.
10806 Use only the first 32 floating point registers.
10811 Use all 64 floating point registers
10814 @opindex mhard-float
10816 Use hardware instructions for floating point operations.
10819 @opindex msoft-float
10821 Use library routines for floating point operations.
10826 Dynamically allocate condition code registers.
10831 Do not try to dynamically allocate condition code registers, only
10832 use @code{icc0} and @code{fcc0}.
10837 Change ABI to use double word insns.
10842 Do not use double word instructions.
10847 Use floating point double instructions.
10850 @opindex mno-double
10852 Do not use floating point double instructions.
10857 Use media instructions.
10862 Do not use media instructions.
10867 Use multiply and add/subtract instructions.
10870 @opindex mno-muladd
10872 Do not use multiply and add/subtract instructions.
10874 @item -mlibrary-pic
10875 @opindex mlibrary-pic
10877 Enable PIC support for building libraries
10882 Use only the first four media accumulator registers.
10887 Use all eight media accumulator registers.
10892 Pack VLIW instructions.
10897 Do not pack VLIW instructions.
10900 @opindex mno-eflags
10902 Do not mark ABI switches in e_flags.
10905 @opindex mcond-move
10907 Enable the use of conditional-move instructions (default).
10909 This switch is mainly for debugging the compiler and will likely be removed
10910 in a future version.
10912 @item -mno-cond-move
10913 @opindex mno-cond-move
10915 Disable the use of conditional-move instructions.
10917 This switch is mainly for debugging the compiler and will likely be removed
10918 in a future version.
10923 Enable the use of conditional set instructions (default).
10925 This switch is mainly for debugging the compiler and will likely be removed
10926 in a future version.
10931 Disable the use of conditional set instructions.
10933 This switch is mainly for debugging the compiler and will likely be removed
10934 in a future version.
10937 @opindex mcond-exec
10939 Enable the use of conditional execution (default).
10941 This switch is mainly for debugging the compiler and will likely be removed
10942 in a future version.
10944 @item -mno-cond-exec
10945 @opindex mno-cond-exec
10947 Disable the use of conditional execution.
10949 This switch is mainly for debugging the compiler and will likely be removed
10950 in a future version.
10952 @item -mvliw-branch
10953 @opindex mvliw-branch
10955 Run a pass to pack branches into VLIW instructions (default).
10957 This switch is mainly for debugging the compiler and will likely be removed
10958 in a future version.
10960 @item -mno-vliw-branch
10961 @opindex mno-vliw-branch
10963 Do not run a pass to pack branches into VLIW instructions.
10965 This switch is mainly for debugging the compiler and will likely be removed
10966 in a future version.
10968 @item -mmulti-cond-exec
10969 @opindex mmulti-cond-exec
10971 Enable optimization of @code{&&} and @code{||} in conditional execution
10974 This switch is mainly for debugging the compiler and will likely be removed
10975 in a future version.
10977 @item -mno-multi-cond-exec
10978 @opindex mno-multi-cond-exec
10980 Disable optimization of @code{&&} and @code{||} in conditional execution.
10982 This switch is mainly for debugging the compiler and will likely be removed
10983 in a future version.
10985 @item -mnested-cond-exec
10986 @opindex mnested-cond-exec
10988 Enable nested conditional execution optimizations (default).
10990 This switch is mainly for debugging the compiler and will likely be removed
10991 in a future version.
10993 @item -mno-nested-cond-exec
10994 @opindex mno-nested-cond-exec
10996 Disable nested conditional execution optimizations.
10998 This switch is mainly for debugging the compiler and will likely be removed
10999 in a future version.
11001 @item -mtomcat-stats
11002 @opindex mtomcat-stats
11004 Cause gas to print out tomcat statistics.
11006 @item -mcpu=@var{cpu}
11009 Select the processor type for which to generate code. Possible values are
11010 @samp{simple}, @samp{tomcat}, @samp{fr500}, @samp{fr400}, @samp{fr300},
11015 @node Xtensa Options
11016 @subsection Xtensa Options
11017 @cindex Xtensa Options
11019 These options are supported for Xtensa targets:
11023 @itemx -mno-const16
11025 @opindex mno-const16
11026 Enable or disable use of @code{CONST16} instructions for loading
11027 constant values. The @code{CONST16} instruction is currently not a
11028 standard option from Tensilica. When enabled, @code{CONST16}
11029 instructions are always used in place of the standard @code{L32R}
11030 instructions. The use of @code{CONST16} is enabled by default only if
11031 the @code{L32R} instruction is not available.
11034 @itemx -mno-fused-madd
11035 @opindex mfused-madd
11036 @opindex mno-fused-madd
11037 Enable or disable use of fused multiply/add and multiply/subtract
11038 instructions in the floating-point option. This has no effect if the
11039 floating-point option is not also enabled. Disabling fused multiply/add
11040 and multiply/subtract instructions forces the compiler to use separate
11041 instructions for the multiply and add/subtract operations. This may be
11042 desirable in some cases where strict IEEE 754-compliant results are
11043 required: the fused multiply add/subtract instructions do not round the
11044 intermediate result, thereby producing results with @emph{more} bits of
11045 precision than specified by the IEEE standard. Disabling fused multiply
11046 add/subtract instructions also ensures that the program output is not
11047 sensitive to the compiler's ability to combine multiply and add/subtract
11050 @item -mtext-section-literals
11051 @itemx -mno-text-section-literals
11052 @opindex mtext-section-literals
11053 @opindex mno-text-section-literals
11054 Control the treatment of literal pools. The default is
11055 @option{-mno-text-section-literals}, which places literals in a separate
11056 section in the output file. This allows the literal pool to be placed
11057 in a data RAM/ROM, and it also allows the linker to combine literal
11058 pools from separate object files to remove redundant literals and
11059 improve code size. With @option{-mtext-section-literals}, the literals
11060 are interspersed in the text section in order to keep them as close as
11061 possible to their references. This may be necessary for large assembly
11064 @item -mtarget-align
11065 @itemx -mno-target-align
11066 @opindex mtarget-align
11067 @opindex mno-target-align
11068 When this option is enabled, GCC instructs the assembler to
11069 automatically align instructions to reduce branch penalties at the
11070 expense of some code density. The assembler attempts to widen density
11071 instructions to align branch targets and the instructions following call
11072 instructions. If there are not enough preceding safe density
11073 instructions to align a target, no widening will be performed. The
11074 default is @option{-mtarget-align}. These options do not affect the
11075 treatment of auto-aligned instructions like @code{LOOP}, which the
11076 assembler will always align, either by widening density instructions or
11077 by inserting no-op instructions.
11080 @itemx -mno-longcalls
11081 @opindex mlongcalls
11082 @opindex mno-longcalls
11083 When this option is enabled, GCC instructs the assembler to translate
11084 direct calls to indirect calls unless it can determine that the target
11085 of a direct call is in the range allowed by the call instruction. This
11086 translation typically occurs for calls to functions in other source
11087 files. Specifically, the assembler translates a direct @code{CALL}
11088 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
11089 The default is @option{-mno-longcalls}. This option should be used in
11090 programs where the call target can potentially be out of range. This
11091 option is implemented in the assembler, not the compiler, so the
11092 assembly code generated by GCC will still show direct call
11093 instructions---look at the disassembled object code to see the actual
11094 instructions. Note that the assembler will use an indirect call for
11095 every cross-file call, not just those that really will be out of range.
11098 @node Code Gen Options
11099 @section Options for Code Generation Conventions
11100 @cindex code generation conventions
11101 @cindex options, code generation
11102 @cindex run-time options
11104 These machine-independent options control the interface conventions
11105 used in code generation.
11107 Most of them have both positive and negative forms; the negative form
11108 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
11109 one of the forms is listed---the one which is not the default. You
11110 can figure out the other form by either removing @samp{no-} or adding
11114 @item -fbounds-check
11115 @opindex fbounds-check
11116 For front-ends that support it, generate additional code to check that
11117 indices used to access arrays are within the declared range. This is
11118 currently only supported by the Java and Fortran 77 front-ends, where
11119 this option defaults to true and false respectively.
11123 This option generates traps for signed overflow on addition, subtraction,
11124 multiplication operations.
11128 This option instructs the compiler to assume that signed arithmetic
11129 overflow of addition, subtraction and multiplication wraps around
11130 using twos-complement representation. This flag enables some optimizations
11131 and disables other. This option is enabled by default for the Java
11132 front-end, as required by the Java language specification.
11135 @opindex fexceptions
11136 Enable exception handling. Generates extra code needed to propagate
11137 exceptions. For some targets, this implies GCC will generate frame
11138 unwind information for all functions, which can produce significant data
11139 size overhead, although it does not affect execution. If you do not
11140 specify this option, GCC will enable it by default for languages like
11141 C++ which normally require exception handling, and disable it for
11142 languages like C that do not normally require it. However, you may need
11143 to enable this option when compiling C code that needs to interoperate
11144 properly with exception handlers written in C++. You may also wish to
11145 disable this option if you are compiling older C++ programs that don't
11146 use exception handling.
11148 @item -fnon-call-exceptions
11149 @opindex fnon-call-exceptions
11150 Generate code that allows trapping instructions to throw exceptions.
11151 Note that this requires platform-specific runtime support that does
11152 not exist everywhere. Moreover, it only allows @emph{trapping}
11153 instructions to throw exceptions, i.e.@: memory references or floating
11154 point instructions. It does not allow exceptions to be thrown from
11155 arbitrary signal handlers such as @code{SIGALRM}.
11157 @item -funwind-tables
11158 @opindex funwind-tables
11159 Similar to @option{-fexceptions}, except that it will just generate any needed
11160 static data, but will not affect the generated code in any other way.
11161 You will normally not enable this option; instead, a language processor
11162 that needs this handling would enable it on your behalf.
11164 @item -fasynchronous-unwind-tables
11165 @opindex funwind-tables
11166 Generate unwind table in dwarf2 format, if supported by target machine. The
11167 table is exact at each instruction boundary, so it can be used for stack
11168 unwinding from asynchronous events (such as debugger or garbage collector).
11170 @item -fpcc-struct-return
11171 @opindex fpcc-struct-return
11172 Return ``short'' @code{struct} and @code{union} values in memory like
11173 longer ones, rather than in registers. This convention is less
11174 efficient, but it has the advantage of allowing intercallability between
11175 GCC-compiled files and files compiled with other compilers, particularly
11176 the Portable C Compiler (pcc).
11178 The precise convention for returning structures in memory depends
11179 on the target configuration macros.
11181 Short structures and unions are those whose size and alignment match
11182 that of some integer type.
11184 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
11185 switch is not binary compatible with code compiled with the
11186 @option{-freg-struct-return} switch.
11187 Use it to conform to a non-default application binary interface.
11189 @item -freg-struct-return
11190 @opindex freg-struct-return
11191 Return @code{struct} and @code{union} values in registers when possible.
11192 This is more efficient for small structures than
11193 @option{-fpcc-struct-return}.
11195 If you specify neither @option{-fpcc-struct-return} nor
11196 @option{-freg-struct-return}, GCC defaults to whichever convention is
11197 standard for the target. If there is no standard convention, GCC
11198 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
11199 the principal compiler. In those cases, we can choose the standard, and
11200 we chose the more efficient register return alternative.
11202 @strong{Warning:} code compiled with the @option{-freg-struct-return}
11203 switch is not binary compatible with code compiled with the
11204 @option{-fpcc-struct-return} switch.
11205 Use it to conform to a non-default application binary interface.
11207 @item -fshort-enums
11208 @opindex fshort-enums
11209 Allocate to an @code{enum} type only as many bytes as it needs for the
11210 declared range of possible values. Specifically, the @code{enum} type
11211 will be equivalent to the smallest integer type which has enough room.
11213 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
11214 code that is not binary compatible with code generated without that switch.
11215 Use it to conform to a non-default application binary interface.
11217 @item -fshort-double
11218 @opindex fshort-double
11219 Use the same size for @code{double} as for @code{float}.
11221 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
11222 code that is not binary compatible with code generated without that switch.
11223 Use it to conform to a non-default application binary interface.
11225 @item -fshort-wchar
11226 @opindex fshort-wchar
11227 Override the underlying type for @samp{wchar_t} to be @samp{short
11228 unsigned int} instead of the default for the target. This option is
11229 useful for building programs to run under WINE@.
11231 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
11232 code that is not binary compatible with code generated without that switch.
11233 Use it to conform to a non-default application binary interface.
11235 @item -fshared-data
11236 @opindex fshared-data
11237 Requests that the data and non-@code{const} variables of this
11238 compilation be shared data rather than private data. The distinction
11239 makes sense only on certain operating systems, where shared data is
11240 shared between processes running the same program, while private data
11241 exists in one copy per process.
11244 @opindex fno-common
11245 In C, allocate even uninitialized global variables in the data section of the
11246 object file, rather than generating them as common blocks. This has the
11247 effect that if the same variable is declared (without @code{extern}) in
11248 two different compilations, you will get an error when you link them.
11249 The only reason this might be useful is if you wish to verify that the
11250 program will work on other systems which always work this way.
11254 Ignore the @samp{#ident} directive.
11256 @item -finhibit-size-directive
11257 @opindex finhibit-size-directive
11258 Don't output a @code{.size} assembler directive, or anything else that
11259 would cause trouble if the function is split in the middle, and the
11260 two halves are placed at locations far apart in memory. This option is
11261 used when compiling @file{crtstuff.c}; you should not need to use it
11264 @item -fverbose-asm
11265 @opindex fverbose-asm
11266 Put extra commentary information in the generated assembly code to
11267 make it more readable. This option is generally only of use to those
11268 who actually need to read the generated assembly code (perhaps while
11269 debugging the compiler itself).
11271 @option{-fno-verbose-asm}, the default, causes the
11272 extra information to be omitted and is useful when comparing two assembler
11277 @cindex global offset table
11279 Generate position-independent code (PIC) suitable for use in a shared
11280 library, if supported for the target machine. Such code accesses all
11281 constant addresses through a global offset table (GOT)@. The dynamic
11282 loader resolves the GOT entries when the program starts (the dynamic
11283 loader is not part of GCC; it is part of the operating system). If
11284 the GOT size for the linked executable exceeds a machine-specific
11285 maximum size, you get an error message from the linker indicating that
11286 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
11287 instead. (These maximums are 16k on the m88k, 8k on the SPARC, and 32k
11288 on the m68k and RS/6000. The 386 has no such limit.)
11290 Position-independent code requires special support, and therefore works
11291 only on certain machines. For the 386, GCC supports PIC for System V
11292 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
11293 position-independent.
11297 If supported for the target machine, emit position-independent code,
11298 suitable for dynamic linking and avoiding any limit on the size of the
11299 global offset table. This option makes a difference on the m68k, m88k,
11302 Position-independent code requires special support, and therefore works
11303 only on certain machines.
11309 These options are similar to @option{-fpic} and @option{-fPIC}, but
11310 generated position independent code can be only linked into executables.
11311 Usually these options are used when @option{-pie} GCC option will be
11312 used during linking.
11314 @item -ffixed-@var{reg}
11316 Treat the register named @var{reg} as a fixed register; generated code
11317 should never refer to it (except perhaps as a stack pointer, frame
11318 pointer or in some other fixed role).
11320 @var{reg} must be the name of a register. The register names accepted
11321 are machine-specific and are defined in the @code{REGISTER_NAMES}
11322 macro in the machine description macro file.
11324 This flag does not have a negative form, because it specifies a
11327 @item -fcall-used-@var{reg}
11328 @opindex fcall-used
11329 Treat the register named @var{reg} as an allocable register that is
11330 clobbered by function calls. It may be allocated for temporaries or
11331 variables that do not live across a call. Functions compiled this way
11332 will not save and restore the register @var{reg}.
11334 It is an error to used this flag with the frame pointer or stack pointer.
11335 Use of this flag for other registers that have fixed pervasive roles in
11336 the machine's execution model will produce disastrous results.
11338 This flag does not have a negative form, because it specifies a
11341 @item -fcall-saved-@var{reg}
11342 @opindex fcall-saved
11343 Treat the register named @var{reg} as an allocable register saved by
11344 functions. It may be allocated even for temporaries or variables that
11345 live across a call. Functions compiled this way will save and restore
11346 the register @var{reg} if they use it.
11348 It is an error to used this flag with the frame pointer or stack pointer.
11349 Use of this flag for other registers that have fixed pervasive roles in
11350 the machine's execution model will produce disastrous results.
11352 A different sort of disaster will result from the use of this flag for
11353 a register in which function values may be returned.
11355 This flag does not have a negative form, because it specifies a
11358 @item -fpack-struct
11359 @opindex fpack-struct
11360 Pack all structure members together without holes.
11362 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
11363 code that is not binary compatible with code generated without that switch.
11364 Additionally, it makes the code suboptimal.
11365 Use it to conform to a non-default application binary interface.
11367 @item -finstrument-functions
11368 @opindex finstrument-functions
11369 Generate instrumentation calls for entry and exit to functions. Just
11370 after function entry and just before function exit, the following
11371 profiling functions will be called with the address of the current
11372 function and its call site. (On some platforms,
11373 @code{__builtin_return_address} does not work beyond the current
11374 function, so the call site information may not be available to the
11375 profiling functions otherwise.)
11378 void __cyg_profile_func_enter (void *this_fn,
11380 void __cyg_profile_func_exit (void *this_fn,
11384 The first argument is the address of the start of the current function,
11385 which may be looked up exactly in the symbol table.
11387 This instrumentation is also done for functions expanded inline in other
11388 functions. The profiling calls will indicate where, conceptually, the
11389 inline function is entered and exited. This means that addressable
11390 versions of such functions must be available. If all your uses of a
11391 function are expanded inline, this may mean an additional expansion of
11392 code size. If you use @samp{extern inline} in your C code, an
11393 addressable version of such functions must be provided. (This is
11394 normally the case anyways, but if you get lucky and the optimizer always
11395 expands the functions inline, you might have gotten away without
11396 providing static copies.)
11398 A function may be given the attribute @code{no_instrument_function}, in
11399 which case this instrumentation will not be done. This can be used, for
11400 example, for the profiling functions listed above, high-priority
11401 interrupt routines, and any functions from which the profiling functions
11402 cannot safely be called (perhaps signal handlers, if the profiling
11403 routines generate output or allocate memory).
11405 @item -fstack-check
11406 @opindex fstack-check
11407 Generate code to verify that you do not go beyond the boundary of the
11408 stack. You should specify this flag if you are running in an
11409 environment with multiple threads, but only rarely need to specify it in
11410 a single-threaded environment since stack overflow is automatically
11411 detected on nearly all systems if there is only one stack.
11413 Note that this switch does not actually cause checking to be done; the
11414 operating system must do that. The switch causes generation of code
11415 to ensure that the operating system sees the stack being extended.
11417 @item -fstack-limit-register=@var{reg}
11418 @itemx -fstack-limit-symbol=@var{sym}
11419 @itemx -fno-stack-limit
11420 @opindex fstack-limit-register
11421 @opindex fstack-limit-symbol
11422 @opindex fno-stack-limit
11423 Generate code to ensure that the stack does not grow beyond a certain value,
11424 either the value of a register or the address of a symbol. If the stack
11425 would grow beyond the value, a signal is raised. For most targets,
11426 the signal is raised before the stack overruns the boundary, so
11427 it is possible to catch the signal without taking special precautions.
11429 For instance, if the stack starts at absolute address @samp{0x80000000}
11430 and grows downwards, you can use the flags
11431 @option{-fstack-limit-symbol=__stack_limit} and
11432 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
11433 of 128KB@. Note that this may only work with the GNU linker.
11435 @cindex aliasing of parameters
11436 @cindex parameters, aliased
11437 @item -fargument-alias
11438 @itemx -fargument-noalias
11439 @itemx -fargument-noalias-global
11440 @opindex fargument-alias
11441 @opindex fargument-noalias
11442 @opindex fargument-noalias-global
11443 Specify the possible relationships among parameters and between
11444 parameters and global data.
11446 @option{-fargument-alias} specifies that arguments (parameters) may
11447 alias each other and may alias global storage.@*
11448 @option{-fargument-noalias} specifies that arguments do not alias
11449 each other, but may alias global storage.@*
11450 @option{-fargument-noalias-global} specifies that arguments do not
11451 alias each other and do not alias global storage.
11453 Each language will automatically use whatever option is required by
11454 the language standard. You should not need to use these options yourself.
11456 @item -fleading-underscore
11457 @opindex fleading-underscore
11458 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
11459 change the way C symbols are represented in the object file. One use
11460 is to help link with legacy assembly code.
11462 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
11463 generate code that is not binary compatible with code generated without that
11464 switch. Use it to conform to a non-default application binary interface.
11465 Not all targets provide complete support for this switch.
11467 @item -ftls-model=@var{model}
11468 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
11469 The @var{model} argument should be one of @code{global-dynamic},
11470 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
11472 The default without @option{-fpic} is @code{initial-exec}; with
11473 @option{-fpic} the default is @code{global-dynamic}.
11478 @node Environment Variables
11479 @section Environment Variables Affecting GCC
11480 @cindex environment variables
11482 @c man begin ENVIRONMENT
11483 This section describes several environment variables that affect how GCC
11484 operates. Some of them work by specifying directories or prefixes to use
11485 when searching for various kinds of files. Some are used to specify other
11486 aspects of the compilation environment.
11488 Note that you can also specify places to search using options such as
11489 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
11490 take precedence over places specified using environment variables, which
11491 in turn take precedence over those specified by the configuration of GCC@.
11492 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
11493 GNU Compiler Collection (GCC) Internals}.
11498 @c @itemx LC_COLLATE
11500 @c @itemx LC_MONETARY
11501 @c @itemx LC_NUMERIC
11506 @c @findex LC_COLLATE
11507 @findex LC_MESSAGES
11508 @c @findex LC_MONETARY
11509 @c @findex LC_NUMERIC
11513 These environment variables control the way that GCC uses
11514 localization information that allow GCC to work with different
11515 national conventions. GCC inspects the locale categories
11516 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
11517 so. These locale categories can be set to any value supported by your
11518 installation. A typical value is @samp{en_UK} for English in the United
11521 The @env{LC_CTYPE} environment variable specifies character
11522 classification. GCC uses it to determine the character boundaries in
11523 a string; this is needed for some multibyte encodings that contain quote
11524 and escape characters that would otherwise be interpreted as a string
11527 The @env{LC_MESSAGES} environment variable specifies the language to
11528 use in diagnostic messages.
11530 If the @env{LC_ALL} environment variable is set, it overrides the value
11531 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
11532 and @env{LC_MESSAGES} default to the value of the @env{LANG}
11533 environment variable. If none of these variables are set, GCC
11534 defaults to traditional C English behavior.
11538 If @env{TMPDIR} is set, it specifies the directory to use for temporary
11539 files. GCC uses temporary files to hold the output of one stage of
11540 compilation which is to be used as input to the next stage: for example,
11541 the output of the preprocessor, which is the input to the compiler
11544 @item GCC_EXEC_PREFIX
11545 @findex GCC_EXEC_PREFIX
11546 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
11547 names of the subprograms executed by the compiler. No slash is added
11548 when this prefix is combined with the name of a subprogram, but you can
11549 specify a prefix that ends with a slash if you wish.
11551 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
11552 an appropriate prefix to use based on the pathname it was invoked with.
11554 If GCC cannot find the subprogram using the specified prefix, it
11555 tries looking in the usual places for the subprogram.
11557 The default value of @env{GCC_EXEC_PREFIX} is
11558 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the value
11559 of @code{prefix} when you ran the @file{configure} script.
11561 Other prefixes specified with @option{-B} take precedence over this prefix.
11563 This prefix is also used for finding files such as @file{crt0.o} that are
11566 In addition, the prefix is used in an unusual way in finding the
11567 directories to search for header files. For each of the standard
11568 directories whose name normally begins with @samp{/usr/local/lib/gcc}
11569 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
11570 replacing that beginning with the specified prefix to produce an
11571 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
11572 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
11573 These alternate directories are searched first; the standard directories
11576 @item COMPILER_PATH
11577 @findex COMPILER_PATH
11578 The value of @env{COMPILER_PATH} is a colon-separated list of
11579 directories, much like @env{PATH}. GCC tries the directories thus
11580 specified when searching for subprograms, if it can't find the
11581 subprograms using @env{GCC_EXEC_PREFIX}.
11584 @findex LIBRARY_PATH
11585 The value of @env{LIBRARY_PATH} is a colon-separated list of
11586 directories, much like @env{PATH}. When configured as a native compiler,
11587 GCC tries the directories thus specified when searching for special
11588 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
11589 using GCC also uses these directories when searching for ordinary
11590 libraries for the @option{-l} option (but directories specified with
11591 @option{-L} come first).
11595 @cindex locale definition
11596 This variable is used to pass locale information to the compiler. One way in
11597 which this information is used is to determine the character set to be used
11598 when character literals, string literals and comments are parsed in C and C++.
11599 When the compiler is configured to allow multibyte characters,
11600 the following values for @env{LANG} are recognized:
11604 Recognize JIS characters.
11606 Recognize SJIS characters.
11608 Recognize EUCJP characters.
11611 If @env{LANG} is not defined, or if it has some other value, then the
11612 compiler will use mblen and mbtowc as defined by the default locale to
11613 recognize and translate multibyte characters.
11617 Some additional environments variables affect the behavior of the
11620 @include cppenv.texi
11624 @node Precompiled Headers
11625 @section Using Precompiled Headers
11626 @cindex precompiled headers
11627 @cindex speed of compilation
11629 Often large projects have many header files that are included in every
11630 source file. The time the compiler takes to process these header files
11631 over and over again can account for nearly all of the time required to
11632 build the project. To make builds faster, GCC allows users to
11633 `precompile' a header file; then, if builds can use the precompiled
11634 header file they will be much faster.
11636 To create a precompiled header file, simply compile it as you would any
11637 other file, if necessary using the @option{-x} option to make the driver
11638 treat it as a C or C++ header file. You will probably want to use a
11639 tool like @command{make} to keep the precompiled header up-to-date when
11640 the headers it contains change.
11642 A precompiled header file will be searched for when @code{#include} is
11643 seen in the compilation. As it searches for the included file
11644 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
11645 compiler looks for a precompiled header in each directory just before it
11646 looks for the include file in that directory. The name searched for is
11647 the name specified in the @code{#include} with @samp{.gch} appended. If
11648 the precompiled header file can't be used, it is ignored.
11650 For instance, if you have @code{#include "all.h"}, and you have
11651 @file{all.h.gch} in the same directory as @file{all.h}, then the
11652 precompiled header file will be used if possible, and the original
11653 header will be used otherwise.
11655 Alternatively, you might decide to put the precompiled header file in a
11656 directory and use @option{-I} to ensure that directory is searched
11657 before (or instead of) the directory containing the original header.
11658 Then, if you want to check that the precompiled header file is always
11659 used, you can put a file of the same name as the original header in this
11660 directory containing an @code{#error} command.
11662 This also works with @option{-include}. So yet another way to use
11663 precompiled headers, good for projects not designed with precompiled
11664 header files in mind, is to simply take most of the header files used by
11665 a project, include them from another header file, precompile that header
11666 file, and @option{-include} the precompiled header. If the header files
11667 have guards against multiple inclusion, they will be skipped because
11668 they've already been included (in the precompiled header).
11670 If you need to precompile the same header file for different
11671 languages, targets, or compiler options, you can instead make a
11672 @emph{directory} named like @file{all.h.gch}, and put each precompiled
11673 header in the directory. (It doesn't matter what you call the files
11674 in the directory, every precompiled header in the directory will be
11675 considered.) The first precompiled header encountered in the
11676 directory that is valid for this compilation will be used; they're
11677 searched in no particular order.
11679 There are many other possibilities, limited only by your imagination,
11680 good sense, and the constraints of your build system.
11682 A precompiled header file can be used only when these conditions apply:
11686 Only one precompiled header can be used in a particular compilation.
11688 A precompiled header can't be used once the first C token is seen. You
11689 can have preprocessor directives before a precompiled header; you can
11690 even include a precompiled header from inside another header, so long as
11691 there are no C tokens before the @code{#include}.
11693 The precompiled header file must be produced for the same language as
11694 the current compilation. You can't use a C precompiled header for a C++
11697 The precompiled header file must be produced by the same compiler
11698 version and configuration as the current compilation is using.
11699 The easiest way to guarantee this is to use the same compiler binary
11700 for creating and using precompiled headers.
11702 Any macros defined before the precompiled header (including with
11703 @option{-D}) must either be defined in the same way as when the
11704 precompiled header was generated, or must not affect the precompiled
11705 header, which usually means that the they don't appear in the
11706 precompiled header at all.
11708 Certain command-line options must be defined in the same way as when the
11709 precompiled header was generated. At present, it's not clear which
11710 options are safe to change and which are not; the safest choice is to
11711 use exactly the same options when generating and using the precompiled
11715 For all of these but the last, the compiler will automatically ignore
11716 the precompiled header if the conditions aren't met. For the last item,
11717 some option changes will cause the precompiled header to be rejected,
11718 but not all incompatible option combinations have yet been found. If
11719 you find a new incompatible combination, please consider filing a bug
11720 report, see @ref{Bugs}.
11722 @node Running Protoize
11723 @section Running Protoize
11725 The program @code{protoize} is an optional part of GCC@. You can use
11726 it to add prototypes to a program, thus converting the program to ISO
11727 C in one respect. The companion program @code{unprotoize} does the
11728 reverse: it removes argument types from any prototypes that are found.
11730 When you run these programs, you must specify a set of source files as
11731 command line arguments. The conversion programs start out by compiling
11732 these files to see what functions they define. The information gathered
11733 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
11735 After scanning comes actual conversion. The specified files are all
11736 eligible to be converted; any files they include (whether sources or
11737 just headers) are eligible as well.
11739 But not all the eligible files are converted. By default,
11740 @code{protoize} and @code{unprotoize} convert only source and header
11741 files in the current directory. You can specify additional directories
11742 whose files should be converted with the @option{-d @var{directory}}
11743 option. You can also specify particular files to exclude with the
11744 @option{-x @var{file}} option. A file is converted if it is eligible, its
11745 directory name matches one of the specified directory names, and its
11746 name within the directory has not been excluded.
11748 Basic conversion with @code{protoize} consists of rewriting most
11749 function definitions and function declarations to specify the types of
11750 the arguments. The only ones not rewritten are those for varargs
11753 @code{protoize} optionally inserts prototype declarations at the
11754 beginning of the source file, to make them available for any calls that
11755 precede the function's definition. Or it can insert prototype
11756 declarations with block scope in the blocks where undeclared functions
11759 Basic conversion with @code{unprotoize} consists of rewriting most
11760 function declarations to remove any argument types, and rewriting
11761 function definitions to the old-style pre-ISO form.
11763 Both conversion programs print a warning for any function declaration or
11764 definition that they can't convert. You can suppress these warnings
11767 The output from @code{protoize} or @code{unprotoize} replaces the
11768 original source file. The original file is renamed to a name ending
11769 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
11770 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
11771 for DOS) file already exists, then the source file is simply discarded.
11773 @code{protoize} and @code{unprotoize} both depend on GCC itself to
11774 scan the program and collect information about the functions it uses.
11775 So neither of these programs will work until GCC is installed.
11777 Here is a table of the options you can use with @code{protoize} and
11778 @code{unprotoize}. Each option works with both programs unless
11782 @item -B @var{directory}
11783 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
11784 usual directory (normally @file{/usr/local/lib}). This file contains
11785 prototype information about standard system functions. This option
11786 applies only to @code{protoize}.
11788 @item -c @var{compilation-options}
11789 Use @var{compilation-options} as the options when running @command{gcc} to
11790 produce the @samp{.X} files. The special option @option{-aux-info} is
11791 always passed in addition, to tell @command{gcc} to write a @samp{.X} file.
11793 Note that the compilation options must be given as a single argument to
11794 @code{protoize} or @code{unprotoize}. If you want to specify several
11795 @command{gcc} options, you must quote the entire set of compilation options
11796 to make them a single word in the shell.
11798 There are certain @command{gcc} arguments that you cannot use, because they
11799 would produce the wrong kind of output. These include @option{-g},
11800 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
11801 the @var{compilation-options}, they are ignored.
11804 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
11805 systems) instead of @samp{.c}. This is convenient if you are converting
11806 a C program to C++. This option applies only to @code{protoize}.
11809 Add explicit global declarations. This means inserting explicit
11810 declarations at the beginning of each source file for each function
11811 that is called in the file and was not declared. These declarations
11812 precede the first function definition that contains a call to an
11813 undeclared function. This option applies only to @code{protoize}.
11815 @item -i @var{string}
11816 Indent old-style parameter declarations with the string @var{string}.
11817 This option applies only to @code{protoize}.
11819 @code{unprotoize} converts prototyped function definitions to old-style
11820 function definitions, where the arguments are declared between the
11821 argument list and the initial @samp{@{}. By default, @code{unprotoize}
11822 uses five spaces as the indentation. If you want to indent with just
11823 one space instead, use @option{-i " "}.
11826 Keep the @samp{.X} files. Normally, they are deleted after conversion
11830 Add explicit local declarations. @code{protoize} with @option{-l} inserts
11831 a prototype declaration for each function in each block which calls the
11832 function without any declaration. This option applies only to
11836 Make no real changes. This mode just prints information about the conversions
11837 that would have been done without @option{-n}.
11840 Make no @samp{.save} files. The original files are simply deleted.
11841 Use this option with caution.
11843 @item -p @var{program}
11844 Use the program @var{program} as the compiler. Normally, the name
11845 @file{gcc} is used.
11848 Work quietly. Most warnings are suppressed.
11851 Print the version number, just like @option{-v} for @command{gcc}.
11854 If you need special compiler options to compile one of your program's
11855 source files, then you should generate that file's @samp{.X} file
11856 specially, by running @command{gcc} on that source file with the
11857 appropriate options and the option @option{-aux-info}. Then run
11858 @code{protoize} on the entire set of files. @code{protoize} will use
11859 the existing @samp{.X} file because it is newer than the source file.
11863 gcc -Dfoo=bar file1.c -aux-info file1.X
11868 You need to include the special files along with the rest in the
11869 @code{protoize} command, even though their @samp{.X} files already
11870 exist, because otherwise they won't get converted.
11872 @xref{Protoize Caveats}, for more information on how to use
11873 @code{protoize} successfully.