1 @c Copyright (C) 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
2 @c 2000, 2001, 2002, 2003, 2004 Free Software Foundation, Inc.
3 @c This is part of the GCC manual.
4 @c For copying conditions, see the file gcc.texi.
8 Copyright @copyright{} 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997,
9 1998, 1999, 2000, 2001, 2002, 2003, 2004 Free Software Foundation, Inc.
11 Permission is granted to copy, distribute and/or modify this document
12 under the terms of the GNU Free Documentation License, Version 1.2 or
13 any later version published by the Free Software Foundation; with the
14 Invariant Sections being ``GNU General Public License'' and ``Funding
15 Free Software'', the Front-Cover texts being (a) (see below), and with
16 the Back-Cover Texts being (b) (see below). A copy of the license is
17 included in the gfdl(7) man page.
19 (a) The FSF's Front-Cover Text is:
23 (b) The FSF's Back-Cover Text is:
25 You have freedom to copy and modify this GNU Manual, like GNU
26 software. Copies published by the Free Software Foundation raise
27 funds for GNU development.
29 @c Set file name and title for the man page.
31 @settitle GNU project C and C++ compiler
33 gcc [@option{-c}|@option{-S}|@option{-E}] [@option{-std=}@var{standard}]
34 [@option{-g}] [@option{-pg}] [@option{-O}@var{level}]
35 [@option{-W}@var{warn}@dots{}] [@option{-pedantic}]
36 [@option{-I}@var{dir}@dots{}] [@option{-L}@var{dir}@dots{}]
37 [@option{-D}@var{macro}[=@var{defn}]@dots{}] [@option{-U}@var{macro}]
38 [@option{-f}@var{option}@dots{}] [@option{-m}@var{machine-option}@dots{}]
39 [@option{-o} @var{outfile}] @var{infile}@dots{}
41 Only the most useful options are listed here; see below for the
42 remainder. @samp{g++} accepts mostly the same options as @samp{gcc}.
45 gpl(7), gfdl(7), fsf-funding(7),
46 cpp(1), gcov(1), as(1), ld(1), gdb(1), adb(1), dbx(1), sdb(1)
47 and the Info entries for @file{gcc}, @file{cpp}, @file{as},
48 @file{ld}, @file{binutils} and @file{gdb}.
51 For instructions on reporting bugs, see
52 @w{@uref{http://gcc.gnu.org/bugs.html}}. Use of the @command{gccbug}
53 script to report bugs is recommended.
56 See the Info entry for @command{gcc}, or
57 @w{@uref{http://gcc.gnu.org/onlinedocs/gcc/Contributors.html}},
58 for contributors to GCC@.
63 @chapter GCC Command Options
64 @cindex GCC command options
65 @cindex command options
66 @cindex options, GCC command
68 @c man begin DESCRIPTION
69 When you invoke GCC, it normally does preprocessing, compilation,
70 assembly and linking. The ``overall options'' allow you to stop this
71 process at an intermediate stage. For example, the @option{-c} option
72 says not to run the linker. Then the output consists of object files
73 output by the assembler.
75 Other options are passed on to one stage of processing. Some options
76 control the preprocessor and others the compiler itself. Yet other
77 options control the assembler and linker; most of these are not
78 documented here, since you rarely need to use any of them.
80 @cindex C compilation options
81 Most of the command line options that you can use with GCC are useful
82 for C programs; when an option is only useful with another language
83 (usually C++), the explanation says so explicitly. If the description
84 for a particular option does not mention a source language, you can use
85 that option with all supported languages.
87 @cindex C++ compilation options
88 @xref{Invoking G++,,Compiling C++ Programs}, for a summary of special
89 options for compiling C++ programs.
91 @cindex grouping options
92 @cindex options, grouping
93 The @command{gcc} program accepts options and file names as operands. Many
94 options have multi-letter names; therefore multiple single-letter options
95 may @emph{not} be grouped: @option{-dr} is very different from @w{@samp{-d
98 @cindex order of options
99 @cindex options, order
100 You can mix options and other arguments. For the most part, the order
101 you use doesn't matter. Order does matter when you use several options
102 of the same kind; for example, if you specify @option{-L} more than once,
103 the directories are searched in the order specified.
105 Many options have long names starting with @samp{-f} or with
106 @samp{-W}---for example, @option{-fforce-mem},
107 @option{-fstrength-reduce}, @option{-Wformat} and so on. Most of
108 these have both positive and negative forms; the negative form of
109 @option{-ffoo} would be @option{-fno-foo}. This manual documents
110 only one of these two forms, whichever one is not the default.
114 @xref{Option Index}, for an index to GCC's options.
117 * Option Summary:: Brief list of all options, without explanations.
118 * Overall Options:: Controlling the kind of output:
119 an executable, object files, assembler files,
120 or preprocessed source.
121 * Invoking G++:: Compiling C++ programs.
122 * C Dialect Options:: Controlling the variant of C language compiled.
123 * C++ Dialect Options:: Variations on C++.
124 * Objective-C Dialect Options:: Variations on Objective-C.
125 * Language Independent Options:: Controlling how diagnostics should be
127 * Warning Options:: How picky should the compiler be?
128 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
129 * Optimize Options:: How much optimization?
130 * Preprocessor Options:: Controlling header files and macro definitions.
131 Also, getting dependency information for Make.
132 * Assembler Options:: Passing options to the assembler.
133 * Link Options:: Specifying libraries and so on.
134 * Directory Options:: Where to find header files and libraries.
135 Where to find the compiler executable files.
136 * Spec Files:: How to pass switches to sub-processes.
137 * Target Options:: Running a cross-compiler, or an old version of GCC.
138 * Submodel Options:: Specifying minor hardware or convention variations,
139 such as 68010 vs 68020.
140 * Code Gen Options:: Specifying conventions for function calls, data layout
142 * Environment Variables:: Env vars that affect GCC.
143 * Precompiled Headers:: Compiling a header once, and using it many times.
144 * Running Protoize:: Automatically adding or removing function prototypes.
150 @section Option Summary
152 Here is a summary of all the options, grouped by type. Explanations are
153 in the following sections.
156 @item Overall Options
157 @xref{Overall Options,,Options Controlling the Kind of Output}.
158 @gccoptlist{-c -S -E -o @var{file} -combine -pipe -pass-exit-codes @gol
159 -x @var{language} -v -### --help --target-help --version}
161 @item C Language Options
162 @xref{C Dialect Options,,Options Controlling C Dialect}.
163 @gccoptlist{-ansi -std=@var{standard} -aux-info @var{filename} @gol
164 -fno-asm -fno-builtin -fno-builtin-@var{function} @gol
165 -fhosted -ffreestanding -fms-extensions @gol
166 -trigraphs -no-integrated-cpp -traditional -traditional-cpp @gol
167 -fallow-single-precision -fcond-mismatch @gol
168 -fsigned-bitfields -fsigned-char @gol
169 -funsigned-bitfields -funsigned-char}
171 @item C++ Language Options
172 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
173 @gccoptlist{-fabi-version=@var{n} -fno-access-control -fcheck-new @gol
174 -fconserve-space -fno-const-strings @gol
175 -fno-elide-constructors @gol
176 -fno-enforce-eh-specs @gol
177 -ffor-scope -fno-for-scope -fno-gnu-keywords @gol
178 -fno-implicit-templates @gol
179 -fno-implicit-inline-templates @gol
180 -fno-implement-inlines -fms-extensions @gol
181 -fno-nonansi-builtins -fno-operator-names @gol
182 -fno-optional-diags -fpermissive @gol
183 -frepo -fno-rtti -fstats -ftemplate-depth-@var{n} @gol
184 -fuse-cxa-atexit -fno-weak -nostdinc++ @gol
185 -fno-default-inline -Wabi -Wctor-dtor-privacy @gol
186 -Wnon-virtual-dtor -Wreorder @gol
187 -Weffc++ -Wno-deprecated @gol
188 -Wno-non-template-friend -Wold-style-cast @gol
189 -Woverloaded-virtual -Wno-pmf-conversions @gol
190 -Wsign-promo -Wsynth}
192 @item Objective-C Language Options
193 @xref{Objective-C Dialect Options,,Options Controlling Objective-C Dialect}.
195 -fconstant-string-class=@var{class-name} @gol
196 -fgnu-runtime -fnext-runtime @gol
197 -fno-nil-receivers @gol
198 -fobjc-exceptions @gol
199 -freplace-objc-classes @gol
202 -Wno-protocol -Wselector -Wundeclared-selector}
204 @item Language Independent Options
205 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
206 @gccoptlist{-fmessage-length=@var{n} @gol
207 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]}}
209 @item Warning Options
210 @xref{Warning Options,,Options to Request or Suppress Warnings}.
211 @gccoptlist{-fsyntax-only -pedantic -pedantic-errors @gol
212 -w -Wextra -Wall -Waggregate-return @gol
213 -Wcast-align -Wcast-qual -Wchar-subscripts -Wcomment @gol
214 -Wconversion -Wno-deprecated-declarations @gol
215 -Wdisabled-optimization -Wno-div-by-zero -Wendif-labels @gol
216 -Werror -Werror-implicit-function-declaration @gol
217 -Wfatal-errors -Wfloat-equal -Wformat -Wformat=2 @gol
218 -Wno-format-extra-args -Wformat-nonliteral @gol
219 -Wformat-security -Wformat-y2k @gol
220 -Wimplicit -Wimplicit-function-declaration -Wimplicit-int @gol
221 -Wimport -Wno-import -Winit-self -Winline @gol
222 -Wno-invalid-offsetof -Winvalid-pch @gol
223 -Wlarger-than-@var{len} -Wlong-long @gol
224 -Wmain -Wmissing-braces @gol
225 -Wmissing-format-attribute -Wmissing-include-dirs @gol
226 -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 -Wstrict-aliasing=2 @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 @gol
238 @item C-only Warning Options
239 @gccoptlist{-Wbad-function-cast -Wmissing-declarations @gol
240 -Wmissing-prototypes -Wnested-externs -Wold-style-definition @gol
241 -Wstrict-prototypes -Wtraditional @gol
242 -Wdeclaration-after-statement}
244 @item Debugging Options
245 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
246 @gccoptlist{-d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
247 -fdump-unnumbered -fdump-translation-unit@r{[}-@var{n}@r{]} @gol
248 -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
250 -fdump-tree-original@r{[}-@var{n}@r{]} @gol
251 -fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
252 -fdump-tree-inlined@r{[}-@var{n}@r{]} @gol
253 -fdump-tree-cfg -fdump-tree-vcg -fdump-tree-alias @gol
255 -fdump-tree-ssa@r{[}-@var{n}@r{]} -fdump-tree-pre@r{[}-@var{n}@r{]} @gol
256 -fdump-tree-ccp@r{[}-@var{n}@r{]} -fdump-tree-dce@r{[}-@var{n}@r{]} @gol
257 -fdump-tree-gimple@r{[}-raw@r{]} -fdump-tree-mudflap@r{[}-@var{n}@r{]} @gol
258 -fdump-tree-dom@r{[}-@var{n}@r{]} @gol
259 -fdump-tree-dse@r{[}-@var{n}@r{]} @gol
260 -fdump-tree-phiopt@r{[}-@var{n}@r{]} @gol
261 -fdump-tree-forwprop@r{[}-@var{n}@r{]} @gol
262 -fdump-tree-copyrename@r{[}-@var{n}@r{]} @gol
264 -fdump-tree-sra@r{[}-@var{n}@r{]} @gol
265 -fdump-tree-fre@r{[}-@var{n}@r{]} @gol
266 -feliminate-dwarf2-dups -feliminate-unused-debug-types @gol
267 -feliminate-unused-debug-symbols -fmem-report -fprofile-arcs -ftree-based-profiling @gol
268 -frandom-seed=@var{string} -fsched-verbose=@var{n} @gol
269 -ftest-coverage -ftime-report -fvar-tracking @gol
270 -g -g@var{level} -gcoff -gdwarf-2 @gol
271 -ggdb -gstabs -gstabs+ -gvms -gxcoff -gxcoff+ @gol
272 -p -pg -print-file-name=@var{library} -print-libgcc-file-name @gol
273 -print-multi-directory -print-multi-lib @gol
274 -print-prog-name=@var{program} -print-search-dirs -Q @gol
277 @item Optimization Options
278 @xref{Optimize Options,,Options that Control Optimization}.
279 @gccoptlist{-falign-functions=@var{n} -falign-jumps=@var{n} @gol
280 -falign-labels=@var{n} -falign-loops=@var{n} @gol
281 -fbounds-check -fmudflap -fmudflapth -fmudflapir @gol
282 -fbranch-probabilities -fprofile-values -fvpt -fbranch-target-load-optimize @gol
283 -fbranch-target-load-optimize2 -fbtr-bb-exclusive @gol
284 -fcaller-saves -fcprop-registers @gol
285 -fcse-follow-jumps -fcse-skip-blocks -fdata-sections @gol
286 -fdelayed-branch -fdelete-null-pointer-checks @gol
287 -fexpensive-optimizations -ffast-math -ffloat-store @gol
288 -fforce-addr -fforce-mem -ffunction-sections @gol
289 -fgcse -fgcse-lm -fgcse-sm -fgcse-las -fgcse-after-reload @gol
290 -floop-optimize -fcrossjumping -fif-conversion -fif-conversion2 @gol
291 -finline-functions -finline-limit=@var{n} -fkeep-inline-functions @gol
292 -fkeep-static-consts -fmerge-constants -fmerge-all-constants @gol
293 -fmodulo-sched -fmove-all-movables -fnew-ra -fno-branch-count-reg @gol
294 -fno-default-inline -fno-defer-pop -floop-optimize2 -fmove-loop-invariants @gol
295 -fno-function-cse -fno-guess-branch-probability @gol
296 -fno-inline -fno-math-errno -fno-peephole -fno-peephole2 @gol
297 -funsafe-math-optimizations -ffinite-math-only @gol
298 -fno-trapping-math -fno-zero-initialized-in-bss @gol
299 -fomit-frame-pointer -foptimize-register-move @gol
300 -foptimize-sibling-calls -fprefetch-loop-arrays @gol
301 -fprofile-generate -fprofile-use @gol
302 -freduce-all-givs -fregmove -frename-registers @gol
303 -freorder-blocks -freorder-blocks-and-partition -freorder-functions @gol
304 -frerun-cse-after-loop -frerun-loop-opt @gol
305 -frounding-math -fschedule-insns -fschedule-insns2 @gol
306 -fno-sched-interblock -fno-sched-spec -fsched-spec-load @gol
307 -fsched-spec-load-dangerous @gol
308 -fsched-stalled-insns=@var{n} -sched-stalled-insns-dep=@var{n} @gol
309 -fsched2-use-superblocks @gol
310 -fsched2-use-traces -fsignaling-nans @gol
311 -fsingle-precision-constant @gol
312 -fstrength-reduce -fstrict-aliasing -ftracer -fthread-jumps @gol
313 -funroll-all-loops -funroll-loops -fpeel-loops @gol
314 -funswitch-loops -fold-unroll-loops -fold-unroll-all-loops @gol
315 -ftree-pre -ftree-ccp -ftree-dce -ftree-loop-optimize @gol
316 -ftree-dominator-opts -ftree-dse -ftree-copyrename @gol
317 -ftree-ch -ftree-sra -ftree-ter -ftree-lrs -ftree-fre @gol
318 --param @var{name}=@var{value}
319 -O -O0 -O1 -O2 -O3 -Os}
321 @item Preprocessor Options
322 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
323 @gccoptlist{-A@var{question}=@var{answer} @gol
324 -A-@var{question}@r{[}=@var{answer}@r{]} @gol
325 -C -dD -dI -dM -dN @gol
326 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
327 -idirafter @var{dir} @gol
328 -include @var{file} -imacros @var{file} @gol
329 -iprefix @var{file} -iwithprefix @var{dir} @gol
330 -iwithprefixbefore @var{dir} -isystem @var{dir} @gol
331 -M -MM -MF -MG -MP -MQ -MT -nostdinc @gol
332 -P -fworking-directory -remap @gol
333 -trigraphs -undef -U@var{macro} -Wp,@var{option} @gol
334 -Xpreprocessor @var{option}}
336 @item Assembler Option
337 @xref{Assembler Options,,Passing Options to the Assembler}.
338 @gccoptlist{-Wa,@var{option} -Xassembler @var{option}}
341 @xref{Link Options,,Options for Linking}.
342 @gccoptlist{@var{object-file-name} -l@var{library} @gol
343 -nostartfiles -nodefaultlibs -nostdlib -pie @gol
344 -s -static -static-libgcc -shared -shared-libgcc -symbolic @gol
345 -Wl,@var{option} -Xlinker @var{option} @gol
348 @item Directory Options
349 @xref{Directory Options,,Options for Directory Search}.
350 @gccoptlist{-B@var{prefix} -I@var{dir} -iquote@var{dir} -L@var{dir} -specs=@var{file} -I-}
353 @c I wrote this xref this way to avoid overfull hbox. -- rms
354 @xref{Target Options}.
355 @gccoptlist{-V @var{version} -b @var{machine}}
357 @item Machine Dependent Options
358 @xref{Submodel Options,,Hardware Models and Configurations}.
359 @c This list is ordered alphanumerically by subsection name.
360 @c Try and put the significant identifier (CPU or system) first,
361 @c so users have a clue at guessing where the ones they want will be.
364 @gccoptlist{-EB -EL @gol
365 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
366 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
369 @gccoptlist{-mapcs-frame -mno-apcs-frame @gol
370 -mabi=@var{name} @gol
371 -mapcs-stack-check -mno-apcs-stack-check @gol
372 -mapcs-float -mno-apcs-float @gol
373 -mapcs-reentrant -mno-apcs-reentrant @gol
374 -msched-prolog -mno-sched-prolog @gol
375 -mlittle-endian -mbig-endian -mwords-little-endian @gol
376 -mfloat-abi=@var{name} -msoft-float -mhard-float -mfpe @gol
377 -mthumb-interwork -mno-thumb-interwork @gol
378 -mcpu=@var{name} -march=@var{name} -mfpu=@var{name} @gol
379 -mstructure-size-boundary=@var{n} @gol
380 -mabort-on-noreturn @gol
381 -mlong-calls -mno-long-calls @gol
382 -msingle-pic-base -mno-single-pic-base @gol
383 -mpic-register=@var{reg} @gol
384 -mnop-fun-dllimport @gol
385 -mcirrus-fix-invalid-insns -mno-cirrus-fix-invalid-insns @gol
386 -mpoke-function-name @gol
388 -mtpcs-frame -mtpcs-leaf-frame @gol
389 -mcaller-super-interworking -mcallee-super-interworking}
392 @gccoptlist{-mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts @gol
393 -mcall-prologues -mno-tablejump -mtiny-stack -mint8}
396 @gccoptlist{-mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
397 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
398 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
399 -mstack-align -mdata-align -mconst-align @gol
400 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
401 -melf -maout -melinux -mlinux -sim -sim2 @gol
402 -mmul-bug-workaround -mno-mul-bug-workaround}
404 @emph{Darwin Options}
405 @gccoptlist{-all_load -allowable_client -arch -arch_errors_fatal @gol
406 -arch_only -bind_at_load -bundle -bundle_loader @gol
407 -client_name -compatibility_version -current_version @gol
408 -dependency-file -dylib_file -dylinker_install_name @gol
409 -dynamic -dynamiclib -exported_symbols_list @gol
410 -filelist -flat_namespace -force_cpusubtype_ALL @gol
411 -force_flat_namespace -headerpad_max_install_names @gol
412 -image_base -init -install_name -keep_private_externs @gol
413 -multi_module -multiply_defined -multiply_defined_unused @gol
414 -noall_load -nofixprebinding -nomultidefs -noprebind -noseglinkedit @gol
415 -pagezero_size -prebind -prebind_all_twolevel_modules @gol
416 -private_bundle -read_only_relocs -sectalign @gol
417 -sectobjectsymbols -whyload -seg1addr @gol
418 -sectcreate -sectobjectsymbols -sectorder @gol
419 -seg_addr_table -seg_addr_table_filename -seglinkedit @gol
420 -segprot -segs_read_only_addr -segs_read_write_addr @gol
421 -single_module -static -sub_library -sub_umbrella @gol
422 -twolevel_namespace -umbrella -undefined @gol
423 -unexported_symbols_list -weak_reference_mismatches @gol
426 @emph{DEC Alpha Options}
427 @gccoptlist{-mno-fp-regs -msoft-float -malpha-as -mgas @gol
428 -mieee -mieee-with-inexact -mieee-conformant @gol
429 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
430 -mtrap-precision=@var{mode} -mbuild-constants @gol
431 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} @gol
432 -mbwx -mmax -mfix -mcix @gol
433 -mfloat-vax -mfloat-ieee @gol
434 -mexplicit-relocs -msmall-data -mlarge-data @gol
435 -msmall-text -mlarge-text @gol
436 -mmemory-latency=@var{time}}
438 @emph{DEC Alpha/VMS Options}
439 @gccoptlist{-mvms-return-codes}
442 @gccoptlist{-mgpr-32 -mgpr-64 -mfpr-32 -mfpr-64 @gol
443 -mhard-float -msoft-float @gol
444 -malloc-cc -mfixed-cc -mdword -mno-dword @gol
445 -mdouble -mno-double @gol
446 -mmedia -mno-media -mmuladd -mno-muladd @gol
447 -mfdpic -minline-plt -mgprel-ro -multilib-library-pic -mlinked-fp @gol
448 -mlibrary-pic -macc-4 -macc-8 @gol
449 -mpack -mno-pack -mno-eflags -mcond-move -mno-cond-move @gol
450 -mscc -mno-scc -mcond-exec -mno-cond-exec @gol
451 -mvliw-branch -mno-vliw-branch @gol
452 -mmulti-cond-exec -mno-multi-cond-exec -mnested-cond-exec @gol
453 -mno-nested-cond-exec -mtomcat-stats @gol
456 @emph{H8/300 Options}
457 @gccoptlist{-mrelax -mh -ms -mn -mint32 -malign-300}
460 @gccoptlist{-march=@var{architecture-type} @gol
461 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
462 -mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
463 -mjump-in-delay -mlinker-opt -mlong-calls @gol
464 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
465 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
466 -mno-jump-in-delay -mno-long-load-store @gol
467 -mno-portable-runtime -mno-soft-float @gol
468 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
469 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
470 -mschedule=@var{cpu-type} -mspace-regs -msio -mwsio @gol
471 -nolibdld -static -threads}
473 @emph{i386 and x86-64 Options}
474 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
475 -mfpmath=@var{unit} @gol
476 -masm=@var{dialect} -mno-fancy-math-387 @gol
477 -mno-fp-ret-in-387 -msoft-float -msvr3-shlib @gol
478 -mno-wide-multiply -mrtd -malign-double @gol
479 -mpreferred-stack-boundary=@var{num} @gol
480 -mmmx -msse -msse2 -msse3 -m3dnow @gol
481 -mthreads -mno-align-stringops -minline-all-stringops @gol
482 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
483 -m96bit-long-double -mregparm=@var{num} -momit-leaf-frame-pointer @gol
484 -mno-red-zone -mno-tls-direct-seg-refs @gol
485 -mcmodel=@var{code-model} @gol
489 @gccoptlist{-mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
490 -mvolatile-asm-stop -mb-step -mregister-names -mno-sdata @gol
491 -mconstant-gp -mauto-pic -minline-float-divide-min-latency @gol
492 -minline-float-divide-max-throughput @gol
493 -minline-int-divide-min-latency @gol
494 -minline-int-divide-max-throughput -mno-dwarf2-asm @gol
495 -mfixed-range=@var{register-range}}
497 @emph{M32R/D Options}
498 @gccoptlist{-m32r2 -m32rx -m32r @gol
500 -malign-loops -mno-align-loops @gol
501 -missue-rate=@var{number} @gol
502 -mbranch-cost=@var{number} @gol
503 -mmodel=@var{code-size-model-type} @gol
504 -msdata=@var{sdata-type} @gol
505 -mno-flush-func -mflush-func=@var{name} @gol
506 -mno-flush-trap -mflush-trap=@var{number} @gol
509 @emph{M680x0 Options}
510 @gccoptlist{-m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
511 -m68060 -mcpu32 -m5200 -m68881 -mbitfield -mc68000 -mc68020 @gol
512 -mnobitfield -mrtd -mshort -msoft-float -mpcrel @gol
513 -malign-int -mstrict-align -msep-data -mno-sep-data @gol
514 -mshared-library-id=n -mid-shared-library -mno-id-shared-library}
516 @emph{M68hc1x Options}
517 @gccoptlist{-m6811 -m6812 -m68hc11 -m68hc12 -m68hcs12 @gol
518 -mauto-incdec -minmax -mlong-calls -mshort @gol
519 -msoft-reg-count=@var{count}}
522 @gccoptlist{-mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
523 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
524 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
525 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
526 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
529 @gccoptlist{-EL -EB -march=@var{arch} -mtune=@var{arch} @gol
530 -mips1 -mips2 -mips3 -mips4 -mips32 -mips32r2 -mips64 @gol
531 -mips16 -mno-mips16 -mabi=@var{abi} -mabicalls -mno-abicalls @gol
532 -mxgot -mno-xgot -mgp32 -mgp64 -mfp32 -mfp64 @gol
533 -mhard-float -msoft-float -msingle-float -mdouble-float @gol
534 -mint64 -mlong64 -mlong32 @gol
535 -G@var{num} -membedded-data -mno-embedded-data @gol
536 -muninit-const-in-rodata -mno-uninit-const-in-rodata @gol
537 -msplit-addresses -mno-split-addresses @gol
538 -mexplicit-relocs -mno-explicit-relocs @gol
539 -mrnames -mno-rnames @gol
540 -mcheck-zero-division -mno-check-zero-division @gol
541 -mmemcpy -mno-memcpy -mlong-calls -mno-long-calls @gol
542 -mmad -mno-mad -mfused-madd -mno-fused-madd -nocpp @gol
543 -mfix-r4000 -mno-fix-r4000 -mfix-r4400 -mno-fix-r4400 @gol
544 -mfix-vr4120 -mno-fix-vr4120 -mfix-sb1 -mno-fix-sb1 @gol
545 -mflush-func=@var{func} -mno-flush-func @gol
546 -mbranch-likely -mno-branch-likely @gol
547 -mfp-exceptions -mno-fp-exceptions @gol
548 -mvr4130-align -mno-vr4130-align}
551 @gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
552 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
553 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
554 -mno-base-addresses -msingle-exit -mno-single-exit}
556 @emph{MN10300 Options}
557 @gccoptlist{-mmult-bug -mno-mult-bug @gol
558 -mam33 -mno-am33 @gol
559 -mam33-2 -mno-am33-2 @gol
563 @gccoptlist{-m32032 -m32332 -m32532 -m32081 -m32381 @gol
564 -mmult-add -mnomult-add -msoft-float -mrtd -mnortd @gol
565 -mregparam -mnoregparam -msb -mnosb @gol
566 -mbitfield -mnobitfield -mhimem -mnohimem}
568 @emph{PDP-11 Options}
569 @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
570 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
571 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
572 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
573 -mbranch-expensive -mbranch-cheap @gol
574 -msplit -mno-split -munix-asm -mdec-asm}
576 @emph{PowerPC Options}
577 See RS/6000 and PowerPC Options.
579 @emph{RS/6000 and PowerPC Options}
580 @gccoptlist{-mcpu=@var{cpu-type} @gol
581 -mtune=@var{cpu-type} @gol
582 -mpower -mno-power -mpower2 -mno-power2 @gol
583 -mpowerpc -mpowerpc64 -mno-powerpc @gol
584 -maltivec -mno-altivec @gol
585 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
586 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
587 -mnew-mnemonics -mold-mnemonics @gol
588 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
589 -m64 -m32 -mxl-call -mno-xl-call -mpe @gol
590 -malign-power -malign-natural @gol
591 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
592 -mstring -mno-string -mupdate -mno-update @gol
593 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
594 -mstrict-align -mno-strict-align -mrelocatable @gol
595 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
596 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
597 -mdynamic-no-pic @gol
598 -mprioritize-restricted-insns=@var{priority} @gol
599 -msched-costly-dep=@var{dependence_type} @gol
600 -minsert-sched-nops=@var{scheme} @gol
601 -mcall-sysv -mcall-netbsd @gol
602 -maix-struct-return -msvr4-struct-return @gol
603 -mabi=altivec -mabi=no-altivec @gol
604 -mabi=spe -mabi=no-spe @gol
605 -misel=yes -misel=no @gol
606 -mspe=yes -mspe=no @gol
607 -mfloat-gprs=yes -mfloat-gprs=no @gol
608 -mprototype -mno-prototype @gol
609 -msim -mmvme -mads -myellowknife -memb -msdata @gol
610 -msdata=@var{opt} -mvxworks -mwindiss -G @var{num} -pthread}
612 @emph{S/390 and zSeries Options}
613 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
614 -mhard-float -msoft-float -mbackchain -mno-backchain @gol
615 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
616 -m64 -m31 -mdebug -mno-debug -mesa -mzarch @gol
617 -mtpf-trace -mno-tpf-trace -mfused-madd -mno-fused-madd}
620 @gccoptlist{-m1 -m2 -m2e -m3 -m3e @gol
621 -m4-nofpu -m4-single-only -m4-single -m4 @gol
622 -m5-64media -m5-64media-nofpu @gol
623 -m5-32media -m5-32media-nofpu @gol
624 -m5-compact -m5-compact-nofpu @gol
625 -mb -ml -mdalign -mrelax @gol
626 -mbigtable -mfmovd -mhitachi -mnomacsave @gol
627 -mieee -misize -mpadstruct -mspace @gol
628 -mprefergot -musermode}
631 @gccoptlist{-mcpu=@var{cpu-type} @gol
632 -mtune=@var{cpu-type} @gol
633 -mcmodel=@var{code-model} @gol
634 -m32 -m64 -mapp-regs -mno-app-regs @gol
635 -mfaster-structs -mno-faster-structs @gol
636 -mfpu -mno-fpu -mhard-float -msoft-float @gol
637 -mhard-quad-float -msoft-quad-float @gol
638 -mimpure-text -mno-impure-text -mlittle-endian @gol
639 -mstack-bias -mno-stack-bias @gol
640 -munaligned-doubles -mno-unaligned-doubles @gol
641 -mv8plus -mno-v8plus -mvis -mno-vis}
643 @emph{System V Options}
644 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
646 @emph{TMS320C3x/C4x Options}
647 @gccoptlist{-mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm @gol
648 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload @gol
649 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned @gol
650 -mparallel-insns -mparallel-mpy -mpreserve-float}
653 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
654 -mprolog-function -mno-prolog-function -mspace @gol
655 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
656 -mapp-regs -mno-app-regs @gol
657 -mdisable-callt -mno-disable-callt @gol
663 @gccoptlist{-mg -mgnu -munix}
665 @emph{x86-64 Options}
666 See i386 and x86-64 Options.
668 @emph{Xstormy16 Options}
671 @emph{Xtensa Options}
672 @gccoptlist{-mconst16 -mno-const16 @gol
673 -mfused-madd -mno-fused-madd @gol
674 -mtext-section-literals -mno-text-section-literals @gol
675 -mtarget-align -mno-target-align @gol
676 -mlongcalls -mno-longcalls}
678 @emph{zSeries Options}
679 See S/390 and zSeries Options.
681 @item Code Generation Options
682 @xref{Code Gen Options,,Options for Code Generation Conventions}.
683 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
684 -ffixed-@var{reg} -fexceptions @gol
685 -fnon-call-exceptions -funwind-tables @gol
686 -fasynchronous-unwind-tables @gol
687 -finhibit-size-directive -finstrument-functions @gol
688 -fno-common -fno-ident @gol
689 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
690 -freg-struct-return -fshared-data -fshort-enums @gol
691 -fshort-double -fshort-wchar @gol
692 -fverbose-asm -fpack-struct -fstack-check @gol
693 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
694 -fargument-alias -fargument-noalias @gol
695 -fargument-noalias-global -fleading-underscore @gol
696 -ftls-model=@var{model} @gol
697 -ftrapv -fwrapv -fbounds-check}
701 * Overall Options:: Controlling the kind of output:
702 an executable, object files, assembler files,
703 or preprocessed source.
704 * C Dialect Options:: Controlling the variant of C language compiled.
705 * C++ Dialect Options:: Variations on C++.
706 * Objective-C Dialect Options:: Variations on Objective-C.
707 * Language Independent Options:: Controlling how diagnostics should be
709 * Warning Options:: How picky should the compiler be?
710 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
711 * Optimize Options:: How much optimization?
712 * Preprocessor Options:: Controlling header files and macro definitions.
713 Also, getting dependency information for Make.
714 * Assembler Options:: Passing options to the assembler.
715 * Link Options:: Specifying libraries and so on.
716 * Directory Options:: Where to find header files and libraries.
717 Where to find the compiler executable files.
718 * Spec Files:: How to pass switches to sub-processes.
719 * Target Options:: Running a cross-compiler, or an old version of GCC.
722 @node Overall Options
723 @section Options Controlling the Kind of Output
725 Compilation can involve up to four stages: preprocessing, compilation
726 proper, assembly and linking, always in that order. GCC is capable of
727 preprocessing and compiling several files either into several
728 assembler input files, or into one assembler input file; then each
729 assembler input file produces an object file, and linking combines all
730 the object files (those newly compiled, and those specified as input)
731 into an executable file.
733 @cindex file name suffix
734 For any given input file, the file name suffix determines what kind of
739 C source code which must be preprocessed.
742 C source code which should not be preprocessed.
745 C++ source code which should not be preprocessed.
748 Objective-C source code. Note that you must link with the library
749 @file{libobjc.a} to make an Objective-C program work.
752 Objective-C source code which should not be preprocessed.
755 C or C++ header file to be turned into a precompiled header.
759 @itemx @var{file}.cxx
760 @itemx @var{file}.cpp
761 @itemx @var{file}.CPP
762 @itemx @var{file}.c++
764 C++ source code which must be preprocessed. Note that in @samp{.cxx},
765 the last two letters must both be literally @samp{x}. Likewise,
766 @samp{.C} refers to a literal capital C@.
770 C++ header file to be turned into a precompiled header.
773 @itemx @var{file}.for
774 @itemx @var{file}.FOR
775 Fortran source code which should not be preprocessed.
778 @itemx @var{file}.fpp
779 @itemx @var{file}.FPP
780 Fortran source code which must be preprocessed (with the traditional
784 Fortran source code which must be preprocessed with a RATFOR
785 preprocessor (not included with GCC)@.
788 @itemx @var{file}.f95
789 Fortran 90/95 source code which should not be preprocessed.
791 @c FIXME: Descriptions of Java file types.
798 Ada source code file which contains a library unit declaration (a
799 declaration of a package, subprogram, or generic, or a generic
800 instantiation), or a library unit renaming declaration (a package,
801 generic, or subprogram renaming declaration). Such files are also
804 @itemx @var{file}.adb
805 Ada source code file containing a library unit body (a subprogram or
806 package body). Such files are also called @dfn{bodies}.
808 @c GCC also knows about some suffixes for languages not yet included:
817 Assembler code which must be preprocessed.
820 An object file to be fed straight into linking.
821 Any file name with no recognized suffix is treated this way.
825 You can specify the input language explicitly with the @option{-x} option:
828 @item -x @var{language}
829 Specify explicitly the @var{language} for the following input files
830 (rather than letting the compiler choose a default based on the file
831 name suffix). This option applies to all following input files until
832 the next @option{-x} option. Possible values for @var{language} are:
834 c c-header cpp-output
835 c++ c++-header c++-cpp-output
836 objective-c objective-c-header objc-cpp-output
837 assembler assembler-with-cpp
839 f77 f77-cpp-input ratfor
846 Turn off any specification of a language, so that subsequent files are
847 handled according to their file name suffixes (as they are if @option{-x}
848 has not been used at all).
850 @item -pass-exit-codes
851 @opindex pass-exit-codes
852 Normally the @command{gcc} program will exit with the code of 1 if any
853 phase of the compiler returns a non-success return code. If you specify
854 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
855 numerically highest error produced by any phase that returned an error
859 If you only want some of the stages of compilation, you can use
860 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
861 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
862 @command{gcc} is to stop. Note that some combinations (for example,
863 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
868 Compile or assemble the source files, but do not link. The linking
869 stage simply is not done. The ultimate output is in the form of an
870 object file for each source file.
872 By default, the object file name for a source file is made by replacing
873 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
875 Unrecognized input files, not requiring compilation or assembly, are
880 Stop after the stage of compilation proper; do not assemble. The output
881 is in the form of an assembler code file for each non-assembler input
884 By default, the assembler file name for a source file is made by
885 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
887 Input files that don't require compilation are ignored.
891 Stop after the preprocessing stage; do not run the compiler proper. The
892 output is in the form of preprocessed source code, which is sent to the
895 Input files which don't require preprocessing are ignored.
897 @cindex output file option
900 Place output in file @var{file}. This applies regardless to whatever
901 sort of output is being produced, whether it be an executable file,
902 an object file, an assembler file or preprocessed C code.
904 If @option{-o} is not specified, the default is to put an executable
905 file in @file{a.out}, the object file for
906 @file{@var{source}.@var{suffix}} in @file{@var{source}.o}, its
907 assembler file in @file{@var{source}.s}, a precompiled header file in
908 @file{@var{source}.@var{suffix}.gch}, and all preprocessed C source on
913 Print (on standard error output) the commands executed to run the stages
914 of compilation. Also print the version number of the compiler driver
915 program and of the preprocessor and the compiler proper.
919 Like @option{-v} except the commands are not executed and all command
920 arguments are quoted. This is useful for shell scripts to capture the
921 driver-generated command lines.
925 Use pipes rather than temporary files for communication between the
926 various stages of compilation. This fails to work on some systems where
927 the assembler is unable to read from a pipe; but the GNU assembler has
932 If you are compiling multiple source files, this option tells the driver
933 to pass all the source files to the compiler at once (for those
934 languages for which the compiler can handle this). This will allow
935 intermodule analysis (IMA) to be performed by the compiler. Currently the only
936 language for which this is supported is C. If you pass source files for
937 multiple languages to the driver, using this option, the driver will invoke
938 the compiler(s) that support IMA once each, passing each compiler all the
939 source files appropriate for it. For those languages that do not support
940 IMA this option will be ignored, and the compiler will be invoked once for
941 each source file in that language. If you use this option in conjunction
942 with -save-temps, the compiler will generate multiple pre-processed files
943 (one for each source file), but only one (combined) .o or .s file.
947 Print (on the standard output) a description of the command line options
948 understood by @command{gcc}. If the @option{-v} option is also specified
949 then @option{--help} will also be passed on to the various processes
950 invoked by @command{gcc}, so that they can display the command line options
951 they accept. If the @option{-Wextra} option is also specified then command
952 line options which have no documentation associated with them will also
957 Print (on the standard output) a description of target specific command
958 line options for each tool.
962 Display the version number and copyrights of the invoked GCC.
966 @section Compiling C++ Programs
968 @cindex suffixes for C++ source
969 @cindex C++ source file suffixes
970 C++ source files conventionally use one of the suffixes @samp{.C},
971 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
972 @samp{.cxx}; C++ header files often use @samp{.hh} or @samp{.H}; and
973 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
974 files with these names and compiles them as C++ programs even if you
975 call the compiler the same way as for compiling C programs (usually
976 with the name @command{gcc}).
980 However, C++ programs often require class libraries as well as a
981 compiler that understands the C++ language---and under some
982 circumstances, you might want to compile programs or header files from
983 standard input, or otherwise without a suffix that flags them as C++
984 programs. You might also like to precompile a C header file with a
985 @samp{.h} extension to be used in C++ compilations. @command{g++} is a
986 program that calls GCC with the default language set to C++, and
987 automatically specifies linking against the C++ library. On many
988 systems, @command{g++} is also installed with the name @command{c++}.
990 @cindex invoking @command{g++}
991 When you compile C++ programs, you may specify many of the same
992 command-line options that you use for compiling programs in any
993 language; or command-line options meaningful for C and related
994 languages; or options that are meaningful only for C++ programs.
995 @xref{C Dialect Options,,Options Controlling C Dialect}, for
996 explanations of options for languages related to C@.
997 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
998 explanations of options that are meaningful only for C++ programs.
1000 @node C Dialect Options
1001 @section Options Controlling C Dialect
1002 @cindex dialect options
1003 @cindex language dialect options
1004 @cindex options, dialect
1006 The following options control the dialect of C (or languages derived
1007 from C, such as C++ and Objective-C) that the compiler accepts:
1010 @cindex ANSI support
1014 In C mode, support all ISO C90 programs. In C++ mode,
1015 remove GNU extensions that conflict with ISO C++.
1017 This turns off certain features of GCC that are incompatible with ISO
1018 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
1019 such as the @code{asm} and @code{typeof} keywords, and
1020 predefined macros such as @code{unix} and @code{vax} that identify the
1021 type of system you are using. It also enables the undesirable and
1022 rarely used ISO trigraph feature. For the C compiler,
1023 it disables recognition of C++ style @samp{//} comments as well as
1024 the @code{inline} keyword.
1026 The alternate keywords @code{__asm__}, @code{__extension__},
1027 @code{__inline__} and @code{__typeof__} continue to work despite
1028 @option{-ansi}. You would not want to use them in an ISO C program, of
1029 course, but it is useful to put them in header files that might be included
1030 in compilations done with @option{-ansi}. Alternate predefined macros
1031 such as @code{__unix__} and @code{__vax__} are also available, with or
1032 without @option{-ansi}.
1034 The @option{-ansi} option does not cause non-ISO programs to be
1035 rejected gratuitously. For that, @option{-pedantic} is required in
1036 addition to @option{-ansi}. @xref{Warning Options}.
1038 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1039 option is used. Some header files may notice this macro and refrain
1040 from declaring certain functions or defining certain macros that the
1041 ISO standard doesn't call for; this is to avoid interfering with any
1042 programs that might use these names for other things.
1044 Functions which would normally be built in but do not have semantics
1045 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1046 functions with @option{-ansi} is used. @xref{Other Builtins,,Other
1047 built-in functions provided by GCC}, for details of the functions
1052 Determine the language standard. This option is currently only
1053 supported when compiling C or C++. A value for this option must be
1054 provided; possible values are
1059 ISO C90 (same as @option{-ansi}).
1061 @item iso9899:199409
1062 ISO C90 as modified in amendment 1.
1068 ISO C99. Note that this standard is not yet fully supported; see
1069 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1070 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1073 Default, ISO C90 plus GNU extensions (including some C99 features).
1077 ISO C99 plus GNU extensions. When ISO C99 is fully implemented in GCC,
1078 this will become the default. The name @samp{gnu9x} is deprecated.
1081 The 1998 ISO C++ standard plus amendments.
1084 The same as @option{-std=c++98} plus GNU extensions. This is the
1085 default for C++ code.
1088 Even when this option is not specified, you can still use some of the
1089 features of newer standards in so far as they do not conflict with
1090 previous C standards. For example, you may use @code{__restrict__} even
1091 when @option{-std=c99} is not specified.
1093 The @option{-std} options specifying some version of ISO C have the same
1094 effects as @option{-ansi}, except that features that were not in ISO C90
1095 but are in the specified version (for example, @samp{//} comments and
1096 the @code{inline} keyword in ISO C99) are not disabled.
1098 @xref{Standards,,Language Standards Supported by GCC}, for details of
1099 these standard versions.
1101 @item -aux-info @var{filename}
1103 Output to the given filename prototyped declarations for all functions
1104 declared and/or defined in a translation unit, including those in header
1105 files. This option is silently ignored in any language other than C@.
1107 Besides declarations, the file indicates, in comments, the origin of
1108 each declaration (source file and line), whether the declaration was
1109 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1110 @samp{O} for old, respectively, in the first character after the line
1111 number and the colon), and whether it came from a declaration or a
1112 definition (@samp{C} or @samp{F}, respectively, in the following
1113 character). In the case of function definitions, a K&R-style list of
1114 arguments followed by their declarations is also provided, inside
1115 comments, after the declaration.
1119 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1120 keyword, so that code can use these words as identifiers. You can use
1121 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1122 instead. @option{-ansi} implies @option{-fno-asm}.
1124 In C++, this switch only affects the @code{typeof} keyword, since
1125 @code{asm} and @code{inline} are standard keywords. You may want to
1126 use the @option{-fno-gnu-keywords} flag instead, which has the same
1127 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1128 switch only affects the @code{asm} and @code{typeof} keywords, since
1129 @code{inline} is a standard keyword in ISO C99.
1132 @itemx -fno-builtin-@var{function}
1133 @opindex fno-builtin
1134 @cindex built-in functions
1135 Don't recognize built-in functions that do not begin with
1136 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1137 functions provided by GCC}, for details of the functions affected,
1138 including those which are not built-in functions when @option{-ansi} or
1139 @option{-std} options for strict ISO C conformance are used because they
1140 do not have an ISO standard meaning.
1142 GCC normally generates special code to handle certain built-in functions
1143 more efficiently; for instance, calls to @code{alloca} may become single
1144 instructions that adjust the stack directly, and calls to @code{memcpy}
1145 may become inline copy loops. The resulting code is often both smaller
1146 and faster, but since the function calls no longer appear as such, you
1147 cannot set a breakpoint on those calls, nor can you change the behavior
1148 of the functions by linking with a different library.
1150 With the @option{-fno-builtin-@var{function}} option
1151 only the built-in function @var{function} is
1152 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1153 function is named this is not built-in in this version of GCC, this
1154 option is ignored. There is no corresponding
1155 @option{-fbuiltin-@var{function}} option; if you wish to enable
1156 built-in functions selectively when using @option{-fno-builtin} or
1157 @option{-ffreestanding}, you may define macros such as:
1160 #define abs(n) __builtin_abs ((n))
1161 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1166 @cindex hosted environment
1168 Assert that compilation takes place in a hosted environment. This implies
1169 @option{-fbuiltin}. A hosted environment is one in which the
1170 entire standard library is available, and in which @code{main} has a return
1171 type of @code{int}. Examples are nearly everything except a kernel.
1172 This is equivalent to @option{-fno-freestanding}.
1174 @item -ffreestanding
1175 @opindex ffreestanding
1176 @cindex hosted environment
1178 Assert that compilation takes place in a freestanding environment. This
1179 implies @option{-fno-builtin}. A freestanding environment
1180 is one in which the standard library may not exist, and program startup may
1181 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1182 This is equivalent to @option{-fno-hosted}.
1184 @xref{Standards,,Language Standards Supported by GCC}, for details of
1185 freestanding and hosted environments.
1187 @item -fms-extensions
1188 @opindex fms-extensions
1189 Accept some non-standard constructs used in Microsoft header files.
1193 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1194 options for strict ISO C conformance) implies @option{-trigraphs}.
1196 @item -no-integrated-cpp
1197 @opindex no-integrated-cpp
1198 Performs a compilation in two passes: preprocessing and compiling. This
1199 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1200 @option{-B} option. The user supplied compilation step can then add in
1201 an additional preprocessing step after normal preprocessing but before
1202 compiling. The default is to use the integrated cpp (internal cpp)
1204 The semantics of this option will change if "cc1", "cc1plus", and
1205 "cc1obj" are merged.
1207 @cindex traditional C language
1208 @cindex C language, traditional
1210 @itemx -traditional-cpp
1211 @opindex traditional-cpp
1212 @opindex traditional
1213 Formerly, these options caused GCC to attempt to emulate a pre-standard
1214 C compiler. They are now only supported with the @option{-E} switch.
1215 The preprocessor continues to support a pre-standard mode. See the GNU
1216 CPP manual for details.
1218 @item -fcond-mismatch
1219 @opindex fcond-mismatch
1220 Allow conditional expressions with mismatched types in the second and
1221 third arguments. The value of such an expression is void. This option
1222 is not supported for C++.
1224 @item -funsigned-char
1225 @opindex funsigned-char
1226 Let the type @code{char} be unsigned, like @code{unsigned char}.
1228 Each kind of machine has a default for what @code{char} should
1229 be. It is either like @code{unsigned char} by default or like
1230 @code{signed char} by default.
1232 Ideally, a portable program should always use @code{signed char} or
1233 @code{unsigned char} when it depends on the signedness of an object.
1234 But many programs have been written to use plain @code{char} and
1235 expect it to be signed, or expect it to be unsigned, depending on the
1236 machines they were written for. This option, and its inverse, let you
1237 make such a program work with the opposite default.
1239 The type @code{char} is always a distinct type from each of
1240 @code{signed char} or @code{unsigned char}, even though its behavior
1241 is always just like one of those two.
1244 @opindex fsigned-char
1245 Let the type @code{char} be signed, like @code{signed char}.
1247 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1248 the negative form of @option{-funsigned-char}. Likewise, the option
1249 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1251 @item -fsigned-bitfields
1252 @itemx -funsigned-bitfields
1253 @itemx -fno-signed-bitfields
1254 @itemx -fno-unsigned-bitfields
1255 @opindex fsigned-bitfields
1256 @opindex funsigned-bitfields
1257 @opindex fno-signed-bitfields
1258 @opindex fno-unsigned-bitfields
1259 These options control whether a bit-field is signed or unsigned, when the
1260 declaration does not use either @code{signed} or @code{unsigned}. By
1261 default, such a bit-field is signed, because this is consistent: the
1262 basic integer types such as @code{int} are signed types.
1265 @node C++ Dialect Options
1266 @section Options Controlling C++ Dialect
1268 @cindex compiler options, C++
1269 @cindex C++ options, command line
1270 @cindex options, C++
1271 This section describes the command-line options that are only meaningful
1272 for C++ programs; but you can also use most of the GNU compiler options
1273 regardless of what language your program is in. For example, you
1274 might compile a file @code{firstClass.C} like this:
1277 g++ -g -frepo -O -c firstClass.C
1281 In this example, only @option{-frepo} is an option meant
1282 only for C++ programs; you can use the other options with any
1283 language supported by GCC@.
1285 Here is a list of options that are @emph{only} for compiling C++ programs:
1289 @item -fabi-version=@var{n}
1290 @opindex fabi-version
1291 Use version @var{n} of the C++ ABI. Version 2 is the version of the
1292 C++ ABI that first appeared in G++ 3.4. Version 1 is the version of
1293 the C++ ABI that first appeared in G++ 3.2. Version 0 will always be
1294 the version that conforms most closely to the C++ ABI specification.
1295 Therefore, the ABI obtained using version 0 will change as ABI bugs
1298 The default is version 2.
1300 @item -fno-access-control
1301 @opindex fno-access-control
1302 Turn off all access checking. This switch is mainly useful for working
1303 around bugs in the access control code.
1307 Check that the pointer returned by @code{operator new} is non-null
1308 before attempting to modify the storage allocated. This check is
1309 normally unnecessary because the C++ standard specifies that
1310 @code{operator new} will only return @code{0} if it is declared
1311 @samp{throw()}, in which case the compiler will always check the
1312 return value even without this option. In all other cases, when
1313 @code{operator new} has a non-empty exception specification, memory
1314 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1315 @samp{new (nothrow)}.
1317 @item -fconserve-space
1318 @opindex fconserve-space
1319 Put uninitialized or runtime-initialized global variables into the
1320 common segment, as C does. This saves space in the executable at the
1321 cost of not diagnosing duplicate definitions. If you compile with this
1322 flag and your program mysteriously crashes after @code{main()} has
1323 completed, you may have an object that is being destroyed twice because
1324 two definitions were merged.
1326 This option is no longer useful on most targets, now that support has
1327 been added for putting variables into BSS without making them common.
1329 @item -fno-const-strings
1330 @opindex fno-const-strings
1331 Give string constants type @code{char *} instead of type @code{const
1332 char *}. By default, G++ uses type @code{const char *} as required by
1333 the standard. Even if you use @option{-fno-const-strings}, you cannot
1334 actually modify the value of a string constant.
1336 This option might be removed in a future release of G++. For maximum
1337 portability, you should structure your code so that it works with
1338 string constants that have type @code{const char *}.
1340 @item -fno-elide-constructors
1341 @opindex fno-elide-constructors
1342 The C++ standard allows an implementation to omit creating a temporary
1343 which is only used to initialize another object of the same type.
1344 Specifying this option disables that optimization, and forces G++ to
1345 call the copy constructor in all cases.
1347 @item -fno-enforce-eh-specs
1348 @opindex fno-enforce-eh-specs
1349 Don't check for violation of exception specifications at runtime. This
1350 option violates the C++ standard, but may be useful for reducing code
1351 size in production builds, much like defining @samp{NDEBUG}. The compiler
1352 will still optimize based on the exception specifications.
1355 @itemx -fno-for-scope
1357 @opindex fno-for-scope
1358 If @option{-ffor-scope} is specified, the scope of variables declared in
1359 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1360 as specified by the C++ standard.
1361 If @option{-fno-for-scope} is specified, the scope of variables declared in
1362 a @i{for-init-statement} extends to the end of the enclosing scope,
1363 as was the case in old versions of G++, and other (traditional)
1364 implementations of C++.
1366 The default if neither flag is given to follow the standard,
1367 but to allow and give a warning for old-style code that would
1368 otherwise be invalid, or have different behavior.
1370 @item -fno-gnu-keywords
1371 @opindex fno-gnu-keywords
1372 Do not recognize @code{typeof} as a keyword, so that code can use this
1373 word as an identifier. You can use the keyword @code{__typeof__} instead.
1374 @option{-ansi} implies @option{-fno-gnu-keywords}.
1376 @item -fno-implicit-templates
1377 @opindex fno-implicit-templates
1378 Never emit code for non-inline templates which are instantiated
1379 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1380 @xref{Template Instantiation}, for more information.
1382 @item -fno-implicit-inline-templates
1383 @opindex fno-implicit-inline-templates
1384 Don't emit code for implicit instantiations of inline templates, either.
1385 The default is to handle inlines differently so that compiles with and
1386 without optimization will need the same set of explicit instantiations.
1388 @item -fno-implement-inlines
1389 @opindex fno-implement-inlines
1390 To save space, do not emit out-of-line copies of inline functions
1391 controlled by @samp{#pragma implementation}. This will cause linker
1392 errors if these functions are not inlined everywhere they are called.
1394 @item -fms-extensions
1395 @opindex fms-extensions
1396 Disable pedantic warnings about constructs used in MFC, such as implicit
1397 int and getting a pointer to member function via non-standard syntax.
1399 @item -fno-nonansi-builtins
1400 @opindex fno-nonansi-builtins
1401 Disable built-in declarations of functions that are not mandated by
1402 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1403 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1405 @item -fno-operator-names
1406 @opindex fno-operator-names
1407 Do not treat the operator name keywords @code{and}, @code{bitand},
1408 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1409 synonyms as keywords.
1411 @item -fno-optional-diags
1412 @opindex fno-optional-diags
1413 Disable diagnostics that the standard says a compiler does not need to
1414 issue. Currently, the only such diagnostic issued by G++ is the one for
1415 a name having multiple meanings within a class.
1418 @opindex fpermissive
1419 Downgrade some diagnostics about nonconformant code from errors to
1420 warnings. Thus, using @option{-fpermissive} will allow some
1421 nonconforming code to compile.
1425 Enable automatic template instantiation at link time. This option also
1426 implies @option{-fno-implicit-templates}. @xref{Template
1427 Instantiation}, for more information.
1431 Disable generation of information about every class with virtual
1432 functions for use by the C++ runtime type identification features
1433 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1434 of the language, you can save some space by using this flag. Note that
1435 exception handling uses the same information, but it will generate it as
1440 Emit statistics about front-end processing at the end of the compilation.
1441 This information is generally only useful to the G++ development team.
1443 @item -ftemplate-depth-@var{n}
1444 @opindex ftemplate-depth
1445 Set the maximum instantiation depth for template classes to @var{n}.
1446 A limit on the template instantiation depth is needed to detect
1447 endless recursions during template class instantiation. ANSI/ISO C++
1448 conforming programs must not rely on a maximum depth greater than 17.
1450 @item -fuse-cxa-atexit
1451 @opindex fuse-cxa-atexit
1452 Register destructors for objects with static storage duration with the
1453 @code{__cxa_atexit} function rather than the @code{atexit} function.
1454 This option is required for fully standards-compliant handling of static
1455 destructors, but will only work if your C library supports
1456 @code{__cxa_atexit}.
1460 Do not use weak symbol support, even if it is provided by the linker.
1461 By default, G++ will use weak symbols if they are available. This
1462 option exists only for testing, and should not be used by end-users;
1463 it will result in inferior code and has no benefits. This option may
1464 be removed in a future release of G++.
1468 Do not search for header files in the standard directories specific to
1469 C++, but do still search the other standard directories. (This option
1470 is used when building the C++ library.)
1473 In addition, these optimization, warning, and code generation options
1474 have meanings only for C++ programs:
1477 @item -fno-default-inline
1478 @opindex fno-default-inline
1479 Do not assume @samp{inline} for functions defined inside a class scope.
1480 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1481 functions will have linkage like inline functions; they just won't be
1484 @item -Wabi @r{(C++ only)}
1486 Warn when G++ generates code that is probably not compatible with the
1487 vendor-neutral C++ ABI. Although an effort has been made to warn about
1488 all such cases, there are probably some cases that are not warned about,
1489 even though G++ is generating incompatible code. There may also be
1490 cases where warnings are emitted even though the code that is generated
1493 You should rewrite your code to avoid these warnings if you are
1494 concerned about the fact that code generated by G++ may not be binary
1495 compatible with code generated by other compilers.
1497 The known incompatibilities at this point include:
1502 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
1503 pack data into the same byte as a base class. For example:
1506 struct A @{ virtual void f(); int f1 : 1; @};
1507 struct B : public A @{ int f2 : 1; @};
1511 In this case, G++ will place @code{B::f2} into the same byte
1512 as@code{A::f1}; other compilers will not. You can avoid this problem
1513 by explicitly padding @code{A} so that its size is a multiple of the
1514 byte size on your platform; that will cause G++ and other compilers to
1515 layout @code{B} identically.
1518 Incorrect handling of tail-padding for virtual bases. G++ does not use
1519 tail padding when laying out virtual bases. For example:
1522 struct A @{ virtual void f(); char c1; @};
1523 struct B @{ B(); char c2; @};
1524 struct C : public A, public virtual B @{@};
1528 In this case, G++ will not place @code{B} into the tail-padding for
1529 @code{A}; other compilers will. You can avoid this problem by
1530 explicitly padding @code{A} so that its size is a multiple of its
1531 alignment (ignoring virtual base classes); that will cause G++ and other
1532 compilers to layout @code{C} identically.
1535 Incorrect handling of bit-fields with declared widths greater than that
1536 of their underlying types, when the bit-fields appear in a union. For
1540 union U @{ int i : 4096; @};
1544 Assuming that an @code{int} does not have 4096 bits, G++ will make the
1545 union too small by the number of bits in an @code{int}.
1548 Empty classes can be placed at incorrect offsets. For example:
1558 struct C : public B, public A @{@};
1562 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
1563 it should be placed at offset zero. G++ mistakenly believes that the
1564 @code{A} data member of @code{B} is already at offset zero.
1567 Names of template functions whose types involve @code{typename} or
1568 template template parameters can be mangled incorrectly.
1571 template <typename Q>
1572 void f(typename Q::X) @{@}
1574 template <template <typename> class Q>
1575 void f(typename Q<int>::X) @{@}
1579 Instantiations of these templates may be mangled incorrectly.
1583 @item -Wctor-dtor-privacy @r{(C++ only)}
1584 @opindex Wctor-dtor-privacy
1585 Warn when a class seems unusable because all the constructors or
1586 destructors in that class are private, and it has neither friends nor
1587 public static member functions.
1589 @item -Wnon-virtual-dtor @r{(C++ only)}
1590 @opindex Wnon-virtual-dtor
1591 Warn when a class appears to be polymorphic, thereby requiring a virtual
1592 destructor, yet it declares a non-virtual one.
1593 This warning is enabled by @option{-Wall}.
1595 @item -Wreorder @r{(C++ only)}
1597 @cindex reordering, warning
1598 @cindex warning for reordering of member initializers
1599 Warn when the order of member initializers given in the code does not
1600 match the order in which they must be executed. For instance:
1606 A(): j (0), i (1) @{ @}
1610 The compiler will rearrange the member initializers for @samp{i}
1611 and @samp{j} to match the declaration order of the members, emitting
1612 a warning to that effect. This warning is enabled by @option{-Wall}.
1615 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
1618 @item -Weffc++ @r{(C++ only)}
1620 Warn about violations of the following style guidelines from Scott Meyers'
1621 @cite{Effective C++} book:
1625 Item 11: Define a copy constructor and an assignment operator for classes
1626 with dynamically allocated memory.
1629 Item 12: Prefer initialization to assignment in constructors.
1632 Item 14: Make destructors virtual in base classes.
1635 Item 15: Have @code{operator=} return a reference to @code{*this}.
1638 Item 23: Don't try to return a reference when you must return an object.
1642 Also warn about violations of the following style guidelines from
1643 Scott Meyers' @cite{More Effective C++} book:
1647 Item 6: Distinguish between prefix and postfix forms of increment and
1648 decrement operators.
1651 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
1655 When selecting this option, be aware that the standard library
1656 headers do not obey all of these guidelines; use @samp{grep -v}
1657 to filter out those warnings.
1659 @item -Wno-deprecated @r{(C++ only)}
1660 @opindex Wno-deprecated
1661 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1663 @item -Wno-non-template-friend @r{(C++ only)}
1664 @opindex Wno-non-template-friend
1665 Disable warnings when non-templatized friend functions are declared
1666 within a template. Since the advent of explicit template specification
1667 support in G++, if the name of the friend is an unqualified-id (i.e.,
1668 @samp{friend foo(int)}), the C++ language specification demands that the
1669 friend declare or define an ordinary, nontemplate function. (Section
1670 14.5.3). Before G++ implemented explicit specification, unqualified-ids
1671 could be interpreted as a particular specialization of a templatized
1672 function. Because this non-conforming behavior is no longer the default
1673 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
1674 check existing code for potential trouble spots and is on by default.
1675 This new compiler behavior can be turned off with
1676 @option{-Wno-non-template-friend} which keeps the conformant compiler code
1677 but disables the helpful warning.
1679 @item -Wold-style-cast @r{(C++ only)}
1680 @opindex Wold-style-cast
1681 Warn if an old-style (C-style) cast to a non-void type is used within
1682 a C++ program. The new-style casts (@samp{static_cast},
1683 @samp{reinterpret_cast}, and @samp{const_cast}) are less vulnerable to
1684 unintended effects and much easier to search for.
1686 @item -Woverloaded-virtual @r{(C++ only)}
1687 @opindex Woverloaded-virtual
1688 @cindex overloaded virtual fn, warning
1689 @cindex warning for overloaded virtual fn
1690 Warn when a function declaration hides virtual functions from a
1691 base class. For example, in:
1698 struct B: public A @{
1703 the @code{A} class version of @code{f} is hidden in @code{B}, and code
1711 will fail to compile.
1713 @item -Wno-pmf-conversions @r{(C++ only)}
1714 @opindex Wno-pmf-conversions
1715 Disable the diagnostic for converting a bound pointer to member function
1718 @item -Wsign-promo @r{(C++ only)}
1719 @opindex Wsign-promo
1720 Warn when overload resolution chooses a promotion from unsigned or
1721 enumerated type to a signed type, over a conversion to an unsigned type of
1722 the same size. Previous versions of G++ would try to preserve
1723 unsignedness, but the standard mandates the current behavior.
1725 @item -Wsynth @r{(C++ only)}
1727 @cindex warning for synthesized methods
1728 @cindex synthesized methods, warning
1729 Warn when G++'s synthesis behavior does not match that of cfront. For
1735 A& operator = (int);
1745 In this example, G++ will synthesize a default @samp{A& operator =
1746 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1749 @node Objective-C Dialect Options
1750 @section Options Controlling Objective-C Dialect
1752 @cindex compiler options, Objective-C
1753 @cindex Objective-C options, command line
1754 @cindex options, Objective-C
1755 (NOTE: This manual does not describe the Objective-C language itself. See
1756 @w{@uref{http://gcc.gnu.org/readings.html}} for references.)
1758 This section describes the command-line options that are only meaningful
1759 for Objective-C programs, but you can also use most of the GNU compiler
1760 options regardless of what language your program is in. For example,
1761 you might compile a file @code{some_class.m} like this:
1764 gcc -g -fgnu-runtime -O -c some_class.m
1768 In this example, @option{-fgnu-runtime} is an option meant only for
1769 Objective-C programs; you can use the other options with any language
1772 Here is a list of options that are @emph{only} for compiling Objective-C
1776 @item -fconstant-string-class=@var{class-name}
1777 @opindex fconstant-string-class
1778 Use @var{class-name} as the name of the class to instantiate for each
1779 literal string specified with the syntax @code{@@"@dots{}"}. The default
1780 class name is @code{NXConstantString} if the GNU runtime is being used, and
1781 @code{NSConstantString} if the NeXT runtime is being used (see below). The
1782 @option{-fconstant-cfstrings} option, if also present, will override the
1783 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
1784 to be laid out as constant CoreFoundation strings.
1787 @opindex fgnu-runtime
1788 Generate object code compatible with the standard GNU Objective-C
1789 runtime. This is the default for most types of systems.
1791 @item -fnext-runtime
1792 @opindex fnext-runtime
1793 Generate output compatible with the NeXT runtime. This is the default
1794 for NeXT-based systems, including Darwin and Mac OS X@. The macro
1795 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
1798 @item -fno-nil-receivers
1799 @opindex fno-nil-receivers
1800 Assume that all Objective-C message dispatches (e.g.,
1801 @code{[receiver message:arg]}) in this translation unit ensure that the receiver
1802 is not @code{nil}. This allows for more efficient entry points in the runtime to be
1803 used. Currently, this option is only available in conjunction with
1804 the NeXT runtime on Mac OS X 10.3 and later.
1806 @item -fobjc-exceptions
1807 @opindex fobjc-exceptions
1808 Enable syntactic support for structured exception handling in Objective-C,
1809 similar to what is offered by C++ and Java. Currently, this option is only
1810 available in conjunction with the NeXT runtime on Mac OS X 10.3 and later.
1818 @@catch (AnObjCClass *exc) @{
1825 @@catch (AnotherClass *exc) @{
1828 @@catch (id allOthers) @{
1838 The @code{@@throw} statement may appear anywhere in an Objective-C or
1839 Objective-C++ program; when used inside of a @code{@@catch} block, the
1840 @code{@@throw} may appear without an argument (as shown above), in which case
1841 the object caught by the @code{@@catch} will be rethrown.
1843 Note that only (pointers to) Objective-C objects may be thrown and
1844 caught using this scheme. When an object is thrown, it will be caught
1845 by the nearest @code{@@catch} clause capable of handling objects of that type,
1846 analogously to how @code{catch} blocks work in C++ and Java. A
1847 @code{@@catch(id @dots{})} clause (as shown above) may also be provided to catch
1848 any and all Objective-C exceptions not caught by previous @code{@@catch}
1851 The @code{@@finally} clause, if present, will be executed upon exit from the
1852 immediately preceding @code{@@try @dots{} @@catch} section. This will happen
1853 regardless of whether any exceptions are thrown, caught or rethrown
1854 inside the @code{@@try @dots{} @@catch} section, analogously to the behavior
1855 of the @code{finally} clause in Java.
1857 There are several caveats to using the new exception mechanism:
1861 Although currently designed to be binary compatible with @code{NS_HANDLER}-style
1862 idioms provided by the @code{NSException} class, the new
1863 exceptions can only be used on Mac OS X 10.3 (Panther) and later
1864 systems, due to additional functionality needed in the (NeXT) Objective-C
1868 As mentioned above, the new exceptions do not support handling
1869 types other than Objective-C objects. Furthermore, when used from
1870 Objective-C++, the Objective-C exception model does not interoperate with C++
1871 exceptions at this time. This means you cannot @code{@@throw} an exception
1872 from Objective-C and @code{catch} it in C++, or vice versa
1873 (i.e., @code{throw @dots{} @@catch}).
1876 The @option{-fobjc-exceptions} switch also enables the use of synchronization
1877 blocks for thread-safe execution:
1880 @@synchronized (ObjCClass *guard) @{
1885 Upon entering the @code{@@synchronized} block, a thread of execution shall
1886 first check whether a lock has been placed on the corresponding @code{guard}
1887 object by another thread. If it has, the current thread shall wait until
1888 the other thread relinquishes its lock. Once @code{guard} becomes available,
1889 the current thread will place its own lock on it, execute the code contained in
1890 the @code{@@synchronized} block, and finally relinquish the lock (thereby
1891 making @code{guard} available to other threads).
1893 Unlike Java, Objective-C does not allow for entire methods to be marked
1894 @code{@@synchronized}. Note that throwing exceptions out of
1895 @code{@@synchronized} blocks is allowed, and will cause the guarding object
1896 to be unlocked properly.
1898 @item -freplace-objc-classes
1899 @opindex freplace-objc-classes
1900 Emit a special marker instructing @command{ld(1)} not to statically link in
1901 the resulting object file, and allow @command{dyld(1)} to load it in at
1902 run time instead. This is used in conjunction with the Fix-and-Continue
1903 debugging mode, where the object file in question may be recompiled and
1904 dynamically reloaded in the course of program execution, without the need
1905 to restart the program itself. Currently, Fix-and-Continue functionality
1906 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
1911 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
1912 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
1913 compile time) with static class references that get initialized at load time,
1914 which improves run-time performance. Specifying the @option{-fzero-link} flag
1915 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
1916 to be retained. This is useful in Zero-Link debugging mode, since it allows
1917 for individual class implementations to be modified during program execution.
1921 Dump interface declarations for all classes seen in the source file to a
1922 file named @file{@var{sourcename}.decl}.
1925 @opindex Wno-protocol
1926 If a class is declared to implement a protocol, a warning is issued for
1927 every method in the protocol that is not implemented by the class. The
1928 default behavior is to issue a warning for every method not explicitly
1929 implemented in the class, even if a method implementation is inherited
1930 from the superclass. If you use the @code{-Wno-protocol} option, then
1931 methods inherited from the superclass are considered to be implemented,
1932 and no warning is issued for them.
1936 Warn if multiple methods of different types for the same selector are
1937 found during compilation. The check is performed on the list of methods
1938 in the final stage of compilation. Additionally, a check is performed
1939 for each selector appearing in a @code{@@selector(@dots{})}
1940 expression, and a corresponding method for that selector has been found
1941 during compilation. Because these checks scan the method table only at
1942 the end of compilation, these warnings are not produced if the final
1943 stage of compilation is not reached, for example because an error is
1944 found during compilation, or because the @code{-fsyntax-only} option is
1947 @item -Wundeclared-selector
1948 @opindex Wundeclared-selector
1949 Warn if a @code{@@selector(@dots{})} expression referring to an
1950 undeclared selector is found. A selector is considered undeclared if no
1951 method with that name has been declared before the
1952 @code{@@selector(@dots{})} expression, either explicitly in an
1953 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
1954 an @code{@@implementation} section. This option always performs its
1955 checks as soon as a @code{@@selector(@dots{})} expression is found,
1956 while @code{-Wselector} only performs its checks in the final stage of
1957 compilation. This also enforces the coding style convention
1958 that methods and selectors must be declared before being used.
1960 @item -print-objc-runtime-info
1961 @opindex print-objc-runtime-info
1962 Generate C header describing the largest structure that is passed by
1967 @node Language Independent Options
1968 @section Options to Control Diagnostic Messages Formatting
1969 @cindex options to control diagnostics formatting
1970 @cindex diagnostic messages
1971 @cindex message formatting
1973 Traditionally, diagnostic messages have been formatted irrespective of
1974 the output device's aspect (e.g.@: its width, @dots{}). The options described
1975 below can be used to control the diagnostic messages formatting
1976 algorithm, e.g.@: how many characters per line, how often source location
1977 information should be reported. Right now, only the C++ front end can
1978 honor these options. However it is expected, in the near future, that
1979 the remaining front ends would be able to digest them correctly.
1982 @item -fmessage-length=@var{n}
1983 @opindex fmessage-length
1984 Try to format error messages so that they fit on lines of about @var{n}
1985 characters. The default is 72 characters for @command{g++} and 0 for the rest of
1986 the front ends supported by GCC@. If @var{n} is zero, then no
1987 line-wrapping will be done; each error message will appear on a single
1990 @opindex fdiagnostics-show-location
1991 @item -fdiagnostics-show-location=once
1992 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
1993 reporter to emit @emph{once} source location information; that is, in
1994 case the message is too long to fit on a single physical line and has to
1995 be wrapped, the source location won't be emitted (as prefix) again,
1996 over and over, in subsequent continuation lines. This is the default
1999 @item -fdiagnostics-show-location=every-line
2000 Only meaningful in line-wrapping mode. Instructs the diagnostic
2001 messages reporter to emit the same source location information (as
2002 prefix) for physical lines that result from the process of breaking
2003 a message which is too long to fit on a single line.
2007 @node Warning Options
2008 @section Options to Request or Suppress Warnings
2009 @cindex options to control warnings
2010 @cindex warning messages
2011 @cindex messages, warning
2012 @cindex suppressing warnings
2014 Warnings are diagnostic messages that report constructions which
2015 are not inherently erroneous but which are risky or suggest there
2016 may have been an error.
2018 You can request many specific warnings with options beginning @samp{-W},
2019 for example @option{-Wimplicit} to request warnings on implicit
2020 declarations. Each of these specific warning options also has a
2021 negative form beginning @samp{-Wno-} to turn off warnings;
2022 for example, @option{-Wno-implicit}. This manual lists only one of the
2023 two forms, whichever is not the default.
2025 The following options control the amount and kinds of warnings produced
2026 by GCC; for further, language-specific options also refer to
2027 @ref{C++ Dialect Options} and @ref{Objective-C Dialect Options}.
2030 @cindex syntax checking
2032 @opindex fsyntax-only
2033 Check the code for syntax errors, but don't do anything beyond that.
2037 Issue all the warnings demanded by strict ISO C and ISO C++;
2038 reject all programs that use forbidden extensions, and some other
2039 programs that do not follow ISO C and ISO C++. For ISO C, follows the
2040 version of the ISO C standard specified by any @option{-std} option used.
2042 Valid ISO C and ISO C++ programs should compile properly with or without
2043 this option (though a rare few will require @option{-ansi} or a
2044 @option{-std} option specifying the required version of ISO C)@. However,
2045 without this option, certain GNU extensions and traditional C and C++
2046 features are supported as well. With this option, they are rejected.
2048 @option{-pedantic} does not cause warning messages for use of the
2049 alternate keywords whose names begin and end with @samp{__}. Pedantic
2050 warnings are also disabled in the expression that follows
2051 @code{__extension__}. However, only system header files should use
2052 these escape routes; application programs should avoid them.
2053 @xref{Alternate Keywords}.
2055 Some users try to use @option{-pedantic} to check programs for strict ISO
2056 C conformance. They soon find that it does not do quite what they want:
2057 it finds some non-ISO practices, but not all---only those for which
2058 ISO C @emph{requires} a diagnostic, and some others for which
2059 diagnostics have been added.
2061 A feature to report any failure to conform to ISO C might be useful in
2062 some instances, but would require considerable additional work and would
2063 be quite different from @option{-pedantic}. We don't have plans to
2064 support such a feature in the near future.
2066 Where the standard specified with @option{-std} represents a GNU
2067 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
2068 corresponding @dfn{base standard}, the version of ISO C on which the GNU
2069 extended dialect is based. Warnings from @option{-pedantic} are given
2070 where they are required by the base standard. (It would not make sense
2071 for such warnings to be given only for features not in the specified GNU
2072 C dialect, since by definition the GNU dialects of C include all
2073 features the compiler supports with the given option, and there would be
2074 nothing to warn about.)
2076 @item -pedantic-errors
2077 @opindex pedantic-errors
2078 Like @option{-pedantic}, except that errors are produced rather than
2083 Inhibit all warning messages.
2087 Inhibit warning messages about the use of @samp{#import}.
2089 @item -Wchar-subscripts
2090 @opindex Wchar-subscripts
2091 Warn if an array subscript has type @code{char}. This is a common cause
2092 of error, as programmers often forget that this type is signed on some
2097 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
2098 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
2100 @item -Wfatal-errors
2101 @opindex Wfatal-errors
2102 This option causes the compiler to abort compilation on the first error
2103 occurred rather than trying to keep going and printing further error
2108 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
2109 the arguments supplied have types appropriate to the format string
2110 specified, and that the conversions specified in the format string make
2111 sense. This includes standard functions, and others specified by format
2112 attributes (@pxref{Function Attributes}), in the @code{printf},
2113 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
2114 not in the C standard) families.
2116 The formats are checked against the format features supported by GNU
2117 libc version 2.2. These include all ISO C90 and C99 features, as well
2118 as features from the Single Unix Specification and some BSD and GNU
2119 extensions. Other library implementations may not support all these
2120 features; GCC does not support warning about features that go beyond a
2121 particular library's limitations. However, if @option{-pedantic} is used
2122 with @option{-Wformat}, warnings will be given about format features not
2123 in the selected standard version (but not for @code{strfmon} formats,
2124 since those are not in any version of the C standard). @xref{C Dialect
2125 Options,,Options Controlling C Dialect}.
2127 Since @option{-Wformat} also checks for null format arguments for
2128 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
2130 @option{-Wformat} is included in @option{-Wall}. For more control over some
2131 aspects of format checking, the options @option{-Wformat-y2k},
2132 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
2133 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
2134 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
2137 @opindex Wformat-y2k
2138 If @option{-Wformat} is specified, also warn about @code{strftime}
2139 formats which may yield only a two-digit year.
2141 @item -Wno-format-extra-args
2142 @opindex Wno-format-extra-args
2143 If @option{-Wformat} is specified, do not warn about excess arguments to a
2144 @code{printf} or @code{scanf} format function. The C standard specifies
2145 that such arguments are ignored.
2147 Where the unused arguments lie between used arguments that are
2148 specified with @samp{$} operand number specifications, normally
2149 warnings are still given, since the implementation could not know what
2150 type to pass to @code{va_arg} to skip the unused arguments. However,
2151 in the case of @code{scanf} formats, this option will suppress the
2152 warning if the unused arguments are all pointers, since the Single
2153 Unix Specification says that such unused arguments are allowed.
2155 @item -Wno-format-zero-length
2156 @opindex Wno-format-zero-length
2157 If @option{-Wformat} is specified, do not warn about zero-length formats.
2158 The C standard specifies that zero-length formats are allowed.
2160 @item -Wformat-nonliteral
2161 @opindex Wformat-nonliteral
2162 If @option{-Wformat} is specified, also warn if the format string is not a
2163 string literal and so cannot be checked, unless the format function
2164 takes its format arguments as a @code{va_list}.
2166 @item -Wformat-security
2167 @opindex Wformat-security
2168 If @option{-Wformat} is specified, also warn about uses of format
2169 functions that represent possible security problems. At present, this
2170 warns about calls to @code{printf} and @code{scanf} functions where the
2171 format string is not a string literal and there are no format arguments,
2172 as in @code{printf (foo);}. This may be a security hole if the format
2173 string came from untrusted input and contains @samp{%n}. (This is
2174 currently a subset of what @option{-Wformat-nonliteral} warns about, but
2175 in future warnings may be added to @option{-Wformat-security} that are not
2176 included in @option{-Wformat-nonliteral}.)
2180 Enable @option{-Wformat} plus format checks not included in
2181 @option{-Wformat}. Currently equivalent to @samp{-Wformat
2182 -Wformat-nonliteral -Wformat-security -Wformat-y2k}.
2186 Warn about passing a null pointer for arguments marked as
2187 requiring a non-null value by the @code{nonnull} function attribute.
2189 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
2190 can be disabled with the @option{-Wno-nonnull} option.
2192 @item -Winit-self @r{(C, C++, and Objective-C only)}
2194 Warn about uninitialized variables which are initialized with themselves.
2195 Note this option can only be used with the @option{-Wuninitialized} option,
2196 which in turn only works with @option{-O1} and above.
2198 For example, GCC will warn about @code{i} being uninitialized in the
2199 following snippet only when @option{-Winit-self} has been specified:
2210 @item -Wimplicit-int
2211 @opindex Wimplicit-int
2212 Warn when a declaration does not specify a type.
2214 @item -Wimplicit-function-declaration
2215 @itemx -Werror-implicit-function-declaration
2216 @opindex Wimplicit-function-declaration
2217 @opindex Werror-implicit-function-declaration
2218 Give a warning (or error) whenever a function is used before being
2223 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
2227 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
2228 function with external linkage, returning int, taking either zero
2229 arguments, two, or three arguments of appropriate types.
2231 @item -Wmissing-braces
2232 @opindex Wmissing-braces
2233 Warn if an aggregate or union initializer is not fully bracketed. In
2234 the following example, the initializer for @samp{a} is not fully
2235 bracketed, but that for @samp{b} is fully bracketed.
2238 int a[2][2] = @{ 0, 1, 2, 3 @};
2239 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
2242 @item -Wmissing-include-dirs @r{(C, C++, and Objective-C only)}
2243 @opindex Wmissing-include-dirs
2244 Warn if a user-supplied include directory does not exist.
2247 @opindex Wparentheses
2248 Warn if parentheses are omitted in certain contexts, such
2249 as when there is an assignment in a context where a truth value
2250 is expected, or when operators are nested whose precedence people
2251 often get confused about.
2253 Also warn about constructions where there may be confusion to which
2254 @code{if} statement an @code{else} branch belongs. Here is an example of
2269 In C, every @code{else} branch belongs to the innermost possible @code{if}
2270 statement, which in this example is @code{if (b)}. This is often not
2271 what the programmer expected, as illustrated in the above example by
2272 indentation the programmer chose. When there is the potential for this
2273 confusion, GCC will issue a warning when this flag is specified.
2274 To eliminate the warning, add explicit braces around the innermost
2275 @code{if} statement so there is no way the @code{else} could belong to
2276 the enclosing @code{if}. The resulting code would look like this:
2292 @item -Wsequence-point
2293 @opindex Wsequence-point
2294 Warn about code that may have undefined semantics because of violations
2295 of sequence point rules in the C standard.
2297 The C standard defines the order in which expressions in a C program are
2298 evaluated in terms of @dfn{sequence points}, which represent a partial
2299 ordering between the execution of parts of the program: those executed
2300 before the sequence point, and those executed after it. These occur
2301 after the evaluation of a full expression (one which is not part of a
2302 larger expression), after the evaluation of the first operand of a
2303 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
2304 function is called (but after the evaluation of its arguments and the
2305 expression denoting the called function), and in certain other places.
2306 Other than as expressed by the sequence point rules, the order of
2307 evaluation of subexpressions of an expression is not specified. All
2308 these rules describe only a partial order rather than a total order,
2309 since, for example, if two functions are called within one expression
2310 with no sequence point between them, the order in which the functions
2311 are called is not specified. However, the standards committee have
2312 ruled that function calls do not overlap.
2314 It is not specified when between sequence points modifications to the
2315 values of objects take effect. Programs whose behavior depends on this
2316 have undefined behavior; the C standard specifies that ``Between the
2317 previous and next sequence point an object shall have its stored value
2318 modified at most once by the evaluation of an expression. Furthermore,
2319 the prior value shall be read only to determine the value to be
2320 stored.''. If a program breaks these rules, the results on any
2321 particular implementation are entirely unpredictable.
2323 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
2324 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
2325 diagnosed by this option, and it may give an occasional false positive
2326 result, but in general it has been found fairly effective at detecting
2327 this sort of problem in programs.
2329 The present implementation of this option only works for C programs. A
2330 future implementation may also work for C++ programs.
2332 The C standard is worded confusingly, therefore there is some debate
2333 over the precise meaning of the sequence point rules in subtle cases.
2334 Links to discussions of the problem, including proposed formal
2335 definitions, may be found on our readings page, at
2336 @w{@uref{http://gcc.gnu.org/readings.html}}.
2339 @opindex Wreturn-type
2340 Warn whenever a function is defined with a return-type that defaults to
2341 @code{int}. Also warn about any @code{return} statement with no
2342 return-value in a function whose return-type is not @code{void}.
2344 For C++, a function without return type always produces a diagnostic
2345 message, even when @option{-Wno-return-type} is specified. The only
2346 exceptions are @samp{main} and functions defined in system headers.
2350 Warn whenever a @code{switch} statement has an index of enumerated type
2351 and lacks a @code{case} for one or more of the named codes of that
2352 enumeration. (The presence of a @code{default} label prevents this
2353 warning.) @code{case} labels outside the enumeration range also
2354 provoke warnings when this option is used.
2356 @item -Wswitch-default
2357 @opindex Wswitch-switch
2358 Warn whenever a @code{switch} statement does not have a @code{default}
2362 @opindex Wswitch-enum
2363 Warn whenever a @code{switch} statement has an index of enumerated type
2364 and lacks a @code{case} for one or more of the named codes of that
2365 enumeration. @code{case} labels outside the enumeration range also
2366 provoke warnings when this option is used.
2370 Warn if any trigraphs are encountered that might change the meaning of
2371 the program (trigraphs within comments are not warned about).
2373 @item -Wunused-function
2374 @opindex Wunused-function
2375 Warn whenever a static function is declared but not defined or a
2376 non\-inline static function is unused.
2378 @item -Wunused-label
2379 @opindex Wunused-label
2380 Warn whenever a label is declared but not used.
2382 To suppress this warning use the @samp{unused} attribute
2383 (@pxref{Variable Attributes}).
2385 @item -Wunused-parameter
2386 @opindex Wunused-parameter
2387 Warn whenever a function parameter is unused aside from its declaration.
2389 To suppress this warning use the @samp{unused} attribute
2390 (@pxref{Variable Attributes}).
2392 @item -Wunused-variable
2393 @opindex Wunused-variable
2394 Warn whenever a local variable or non-constant static variable is unused
2395 aside from its declaration
2397 To suppress this warning use the @samp{unused} attribute
2398 (@pxref{Variable Attributes}).
2400 @item -Wunused-value
2401 @opindex Wunused-value
2402 Warn whenever a statement computes a result that is explicitly not used.
2404 To suppress this warning cast the expression to @samp{void}.
2408 All the above @option{-Wunused} options combined.
2410 In order to get a warning about an unused function parameter, you must
2411 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
2412 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
2414 @item -Wuninitialized
2415 @opindex Wuninitialized
2416 Warn if an automatic variable is used without first being initialized or
2417 if a variable may be clobbered by a @code{setjmp} call.
2419 These warnings are possible only in optimizing compilation,
2420 because they require data flow information that is computed only
2421 when optimizing. If you don't specify @option{-O}, you simply won't
2424 If you want to warn about code which uses the uninitialized value of the
2425 variable in its own initializer, use the @option{-Winit-self} option.
2427 These warnings occur only for variables that are candidates for
2428 register allocation. Therefore, they do not occur for a variable that
2429 is declared @code{volatile}, or whose address is taken, or whose size
2430 is other than 1, 2, 4 or 8 bytes. Also, they do not occur for
2431 structures, unions or arrays, even when they are in registers.
2433 Note that there may be no warning about a variable that is used only
2434 to compute a value that itself is never used, because such
2435 computations may be deleted by data flow analysis before the warnings
2438 These warnings are made optional because GCC is not smart
2439 enough to see all the reasons why the code might be correct
2440 despite appearing to have an error. Here is one example of how
2461 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
2462 always initialized, but GCC doesn't know this. Here is
2463 another common case:
2468 if (change_y) save_y = y, y = new_y;
2470 if (change_y) y = save_y;
2475 This has no bug because @code{save_y} is used only if it is set.
2477 @cindex @code{longjmp} warnings
2478 This option also warns when a non-volatile automatic variable might be
2479 changed by a call to @code{longjmp}. These warnings as well are possible
2480 only in optimizing compilation.
2482 The compiler sees only the calls to @code{setjmp}. It cannot know
2483 where @code{longjmp} will be called; in fact, a signal handler could
2484 call it at any point in the code. As a result, you may get a warning
2485 even when there is in fact no problem because @code{longjmp} cannot
2486 in fact be called at the place which would cause a problem.
2488 Some spurious warnings can be avoided if you declare all the functions
2489 you use that never return as @code{noreturn}. @xref{Function
2492 @item -Wunknown-pragmas
2493 @opindex Wunknown-pragmas
2494 @cindex warning for unknown pragmas
2495 @cindex unknown pragmas, warning
2496 @cindex pragmas, warning of unknown
2497 Warn when a #pragma directive is encountered which is not understood by
2498 GCC@. If this command line option is used, warnings will even be issued
2499 for unknown pragmas in system header files. This is not the case if
2500 the warnings were only enabled by the @option{-Wall} command line option.
2502 @item -Wstrict-aliasing
2503 @opindex Wstrict-aliasing
2504 This option is only active when @option{-fstrict-aliasing} is active.
2505 It warns about code which might break the strict aliasing rules that the
2506 compiler is using for optimization. The warning does not catch all
2507 cases, but does attempt to catch the more common pitfalls. It is
2508 included in @option{-Wall}.
2510 @item -Wstrict-aliasing=2
2511 @opindex Wstrict-aliasing=2
2512 This option is only active when @option{-fstrict-aliasing} is active.
2513 It warns about all code which might break the strict aliasing rules that the
2514 compiler is using for optimization. This warning catches all cases, but
2515 it will also give a warning for some ambiguous cases that are safe.
2519 All of the above @samp{-W} options combined. This enables all the
2520 warnings about constructions that some users consider questionable, and
2521 that are easy to avoid (or modify to prevent the warning), even in
2522 conjunction with macros. This also enables some language-specific
2523 warnings described in @ref{C++ Dialect Options} and
2524 @ref{Objective-C Dialect Options}.
2527 The following @option{-W@dots{}} options are not implied by @option{-Wall}.
2528 Some of them warn about constructions that users generally do not
2529 consider questionable, but which occasionally you might wish to check
2530 for; others warn about constructions that are necessary or hard to avoid
2531 in some cases, and there is no simple way to modify the code to suppress
2538 (This option used to be called @option{-W}. The older name is still
2539 supported, but the newer name is more descriptive.) Print extra warning
2540 messages for these events:
2544 A function can return either with or without a value. (Falling
2545 off the end of the function body is considered returning without
2546 a value.) For example, this function would evoke such a
2560 An expression-statement or the left-hand side of a comma expression
2561 contains no side effects.
2562 To suppress the warning, cast the unused expression to void.
2563 For example, an expression such as @samp{x[i,j]} will cause a warning,
2564 but @samp{x[(void)i,j]} will not.
2567 An unsigned value is compared against zero with @samp{<} or @samp{>=}.
2570 A comparison like @samp{x<=y<=z} appears; this is equivalent to
2571 @samp{(x<=y ? 1 : 0) <= z}, which is a different interpretation from
2572 that of ordinary mathematical notation.
2575 Storage-class specifiers like @code{static} are not the first things in
2576 a declaration. According to the C Standard, this usage is obsolescent.
2579 The return type of a function has a type qualifier such as @code{const}.
2580 Such a type qualifier has no effect, since the value returned by a
2581 function is not an lvalue. (But don't warn about the GNU extension of
2582 @code{volatile void} return types. That extension will be warned about
2583 if @option{-pedantic} is specified.)
2586 If @option{-Wall} or @option{-Wunused} is also specified, warn about unused
2590 A comparison between signed and unsigned values could produce an
2591 incorrect result when the signed value is converted to unsigned.
2592 (But don't warn if @option{-Wno-sign-compare} is also specified.)
2595 An aggregate has an initializer which does not initialize all members.
2596 For example, the following code would cause such a warning, because
2597 @code{x.h} would be implicitly initialized to zero:
2600 struct s @{ int f, g, h; @};
2601 struct s x = @{ 3, 4 @};
2605 A function parameter is declared without a type specifier in K&R-style
2613 An empty body occurs in an @samp{if} or @samp{else} statement.
2616 A pointer is compared against integer zero with @samp{<}, @samp{<=},
2617 @samp{>}, or @samp{>=}.
2620 A variable might be changed by @samp{longjmp} or @samp{vfork}.
2623 Any of several floating-point events that often indicate errors, such as
2624 overflow, underflow, loss of precision, etc.
2626 @item @r{(C++ only)}
2627 An enumerator and a non-enumerator both appear in a conditional expression.
2629 @item @r{(C++ only)}
2630 A non-static reference or non-static @samp{const} member appears in a
2631 class without constructors.
2633 @item @r{(C++ only)}
2634 Ambiguous virtual bases.
2636 @item @r{(C++ only)}
2637 Subscripting an array which has been declared @samp{register}.
2639 @item @r{(C++ only)}
2640 Taking the address of a variable which has been declared @samp{register}.
2642 @item @r{(C++ only)}
2643 A base class is not initialized in a derived class' copy constructor.
2646 @item -Wno-div-by-zero
2647 @opindex Wno-div-by-zero
2648 @opindex Wdiv-by-zero
2649 Do not warn about compile-time integer division by zero. Floating point
2650 division by zero is not warned about, as it can be a legitimate way of
2651 obtaining infinities and NaNs.
2653 @item -Wsystem-headers
2654 @opindex Wsystem-headers
2655 @cindex warnings from system headers
2656 @cindex system headers, warnings from
2657 Print warning messages for constructs found in system header files.
2658 Warnings from system headers are normally suppressed, on the assumption
2659 that they usually do not indicate real problems and would only make the
2660 compiler output harder to read. Using this command line option tells
2661 GCC to emit warnings from system headers as if they occurred in user
2662 code. However, note that using @option{-Wall} in conjunction with this
2663 option will @emph{not} warn about unknown pragmas in system
2664 headers---for that, @option{-Wunknown-pragmas} must also be used.
2667 @opindex Wfloat-equal
2668 Warn if floating point values are used in equality comparisons.
2670 The idea behind this is that sometimes it is convenient (for the
2671 programmer) to consider floating-point values as approximations to
2672 infinitely precise real numbers. If you are doing this, then you need
2673 to compute (by analyzing the code, or in some other way) the maximum or
2674 likely maximum error that the computation introduces, and allow for it
2675 when performing comparisons (and when producing output, but that's a
2676 different problem). In particular, instead of testing for equality, you
2677 would check to see whether the two values have ranges that overlap; and
2678 this is done with the relational operators, so equality comparisons are
2681 @item -Wtraditional @r{(C only)}
2682 @opindex Wtraditional
2683 Warn about certain constructs that behave differently in traditional and
2684 ISO C@. Also warn about ISO C constructs that have no traditional C
2685 equivalent, and/or problematic constructs which should be avoided.
2689 Macro parameters that appear within string literals in the macro body.
2690 In traditional C macro replacement takes place within string literals,
2691 but does not in ISO C@.
2694 In traditional C, some preprocessor directives did not exist.
2695 Traditional preprocessors would only consider a line to be a directive
2696 if the @samp{#} appeared in column 1 on the line. Therefore
2697 @option{-Wtraditional} warns about directives that traditional C
2698 understands but would ignore because the @samp{#} does not appear as the
2699 first character on the line. It also suggests you hide directives like
2700 @samp{#pragma} not understood by traditional C by indenting them. Some
2701 traditional implementations would not recognize @samp{#elif}, so it
2702 suggests avoiding it altogether.
2705 A function-like macro that appears without arguments.
2708 The unary plus operator.
2711 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
2712 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
2713 constants.) Note, these suffixes appear in macros defined in the system
2714 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
2715 Use of these macros in user code might normally lead to spurious
2716 warnings, however GCC's integrated preprocessor has enough context to
2717 avoid warning in these cases.
2720 A function declared external in one block and then used after the end of
2724 A @code{switch} statement has an operand of type @code{long}.
2727 A non-@code{static} function declaration follows a @code{static} one.
2728 This construct is not accepted by some traditional C compilers.
2731 The ISO type of an integer constant has a different width or
2732 signedness from its traditional type. This warning is only issued if
2733 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
2734 typically represent bit patterns, are not warned about.
2737 Usage of ISO string concatenation is detected.
2740 Initialization of automatic aggregates.
2743 Identifier conflicts with labels. Traditional C lacks a separate
2744 namespace for labels.
2747 Initialization of unions. If the initializer is zero, the warning is
2748 omitted. This is done under the assumption that the zero initializer in
2749 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
2750 initializer warnings and relies on default initialization to zero in the
2754 Conversions by prototypes between fixed/floating point values and vice
2755 versa. The absence of these prototypes when compiling with traditional
2756 C would cause serious problems. This is a subset of the possible
2757 conversion warnings, for the full set use @option{-Wconversion}.
2760 Use of ISO C style function definitions. This warning intentionally is
2761 @emph{not} issued for prototype declarations or variadic functions
2762 because these ISO C features will appear in your code when using
2763 libiberty's traditional C compatibility macros, @code{PARAMS} and
2764 @code{VPARAMS}. This warning is also bypassed for nested functions
2765 because that feature is already a GCC extension and thus not relevant to
2766 traditional C compatibility.
2769 @item -Wdeclaration-after-statement @r{(C only)}
2770 @opindex Wdeclaration-after-statement
2771 Warn when a declaration is found after a statement in a block. This
2772 construct, known from C++, was introduced with ISO C99 and is by default
2773 allowed in GCC@. It is not supported by ISO C90 and was not supported by
2774 GCC versions before GCC 3.0. @xref{Mixed Declarations}.
2778 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
2780 @item -Wendif-labels
2781 @opindex Wendif-labels
2782 Warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
2786 Warn whenever a local variable shadows another local variable, parameter or
2787 global variable or whenever a built-in function is shadowed.
2789 @item -Wlarger-than-@var{len}
2790 @opindex Wlarger-than
2791 Warn whenever an object of larger than @var{len} bytes is defined.
2793 @item -Wpointer-arith
2794 @opindex Wpointer-arith
2795 Warn about anything that depends on the ``size of'' a function type or
2796 of @code{void}. GNU C assigns these types a size of 1, for
2797 convenience in calculations with @code{void *} pointers and pointers
2800 @item -Wbad-function-cast @r{(C only)}
2801 @opindex Wbad-function-cast
2802 Warn whenever a function call is cast to a non-matching type.
2803 For example, warn if @code{int malloc()} is cast to @code{anything *}.
2807 Warn whenever a pointer is cast so as to remove a type qualifier from
2808 the target type. For example, warn if a @code{const char *} is cast
2809 to an ordinary @code{char *}.
2812 @opindex Wcast-align
2813 Warn whenever a pointer is cast such that the required alignment of the
2814 target is increased. For example, warn if a @code{char *} is cast to
2815 an @code{int *} on machines where integers can only be accessed at
2816 two- or four-byte boundaries.
2818 @item -Wwrite-strings
2819 @opindex Wwrite-strings
2820 When compiling C, give string constants the type @code{const
2821 char[@var{length}]} so that
2822 copying the address of one into a non-@code{const} @code{char *}
2823 pointer will get a warning; when compiling C++, warn about the
2824 deprecated conversion from string constants to @code{char *}.
2825 These warnings will help you find at
2826 compile time code that can try to write into a string constant, but
2827 only if you have been very careful about using @code{const} in
2828 declarations and prototypes. Otherwise, it will just be a nuisance;
2829 this is why we did not make @option{-Wall} request these warnings.
2832 @opindex Wconversion
2833 Warn if a prototype causes a type conversion that is different from what
2834 would happen to the same argument in the absence of a prototype. This
2835 includes conversions of fixed point to floating and vice versa, and
2836 conversions changing the width or signedness of a fixed point argument
2837 except when the same as the default promotion.
2839 Also, warn if a negative integer constant expression is implicitly
2840 converted to an unsigned type. For example, warn about the assignment
2841 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
2842 casts like @code{(unsigned) -1}.
2844 @item -Wsign-compare
2845 @opindex Wsign-compare
2846 @cindex warning for comparison of signed and unsigned values
2847 @cindex comparison of signed and unsigned values, warning
2848 @cindex signed and unsigned values, comparison warning
2849 Warn when a comparison between signed and unsigned values could produce
2850 an incorrect result when the signed value is converted to unsigned.
2851 This warning is also enabled by @option{-Wextra}; to get the other warnings
2852 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
2854 @item -Waggregate-return
2855 @opindex Waggregate-return
2856 Warn if any functions that return structures or unions are defined or
2857 called. (In languages where you can return an array, this also elicits
2860 @item -Wstrict-prototypes @r{(C only)}
2861 @opindex Wstrict-prototypes
2862 Warn if a function is declared or defined without specifying the
2863 argument types. (An old-style function definition is permitted without
2864 a warning if preceded by a declaration which specifies the argument
2867 @item -Wold-style-definition @r{(C only)}
2868 @opindex Wold-style-definition
2869 Warn if an old-style function definition is used. A warning is given
2870 even if there is a previous prototype.
2872 @item -Wmissing-prototypes @r{(C only)}
2873 @opindex Wmissing-prototypes
2874 Warn if a global function is defined without a previous prototype
2875 declaration. This warning is issued even if the definition itself
2876 provides a prototype. The aim is to detect global functions that fail
2877 to be declared in header files.
2879 @item -Wmissing-declarations @r{(C only)}
2880 @opindex Wmissing-declarations
2881 Warn if a global function is defined without a previous declaration.
2882 Do so even if the definition itself provides a prototype.
2883 Use this option to detect global functions that are not declared in
2886 @item -Wmissing-noreturn
2887 @opindex Wmissing-noreturn
2888 Warn about functions which might be candidates for attribute @code{noreturn}.
2889 Note these are only possible candidates, not absolute ones. Care should
2890 be taken to manually verify functions actually do not ever return before
2891 adding the @code{noreturn} attribute, otherwise subtle code generation
2892 bugs could be introduced. You will not get a warning for @code{main} in
2893 hosted C environments.
2895 @item -Wmissing-format-attribute
2896 @opindex Wmissing-format-attribute
2898 If @option{-Wformat} is enabled, also warn about functions which might be
2899 candidates for @code{format} attributes. Note these are only possible
2900 candidates, not absolute ones. GCC will guess that @code{format}
2901 attributes might be appropriate for any function that calls a function
2902 like @code{vprintf} or @code{vscanf}, but this might not always be the
2903 case, and some functions for which @code{format} attributes are
2904 appropriate may not be detected. This option has no effect unless
2905 @option{-Wformat} is enabled (possibly by @option{-Wall}).
2907 @item -Wno-multichar
2908 @opindex Wno-multichar
2910 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
2911 Usually they indicate a typo in the user's code, as they have
2912 implementation-defined values, and should not be used in portable code.
2914 @item -Wno-deprecated-declarations
2915 @opindex Wno-deprecated-declarations
2916 Do not warn about uses of functions, variables, and types marked as
2917 deprecated by using the @code{deprecated} attribute.
2918 (@pxref{Function Attributes}, @pxref{Variable Attributes},
2919 @pxref{Type Attributes}.)
2923 Warn if a structure is given the packed attribute, but the packed
2924 attribute has no effect on the layout or size of the structure.
2925 Such structures may be mis-aligned for little benefit. For
2926 instance, in this code, the variable @code{f.x} in @code{struct bar}
2927 will be misaligned even though @code{struct bar} does not itself
2928 have the packed attribute:
2935 @} __attribute__((packed));
2945 Warn if padding is included in a structure, either to align an element
2946 of the structure or to align the whole structure. Sometimes when this
2947 happens it is possible to rearrange the fields of the structure to
2948 reduce the padding and so make the structure smaller.
2950 @item -Wredundant-decls
2951 @opindex Wredundant-decls
2952 Warn if anything is declared more than once in the same scope, even in
2953 cases where multiple declaration is valid and changes nothing.
2955 @item -Wnested-externs @r{(C only)}
2956 @opindex Wnested-externs
2957 Warn if an @code{extern} declaration is encountered within a function.
2959 @item -Wunreachable-code
2960 @opindex Wunreachable-code
2961 Warn if the compiler detects that code will never be executed.
2963 This option is intended to warn when the compiler detects that at
2964 least a whole line of source code will never be executed, because
2965 some condition is never satisfied or because it is after a
2966 procedure that never returns.
2968 It is possible for this option to produce a warning even though there
2969 are circumstances under which part of the affected line can be executed,
2970 so care should be taken when removing apparently-unreachable code.
2972 For instance, when a function is inlined, a warning may mean that the
2973 line is unreachable in only one inlined copy of the function.
2975 This option is not made part of @option{-Wall} because in a debugging
2976 version of a program there is often substantial code which checks
2977 correct functioning of the program and is, hopefully, unreachable
2978 because the program does work. Another common use of unreachable
2979 code is to provide behavior which is selectable at compile-time.
2983 Warn if a function can not be inlined and it was declared as inline.
2984 Even with this option, the compiler will not warn about failures to
2985 inline functions declared in system headers.
2987 The compiler uses a variety of heuristics to determine whether or not
2988 to inline a function. For example, the compiler takes into account
2989 the size of the function being inlined and the the amount of inlining
2990 that has already been done in the current function. Therefore,
2991 seemingly insignificant changes in the source program can cause the
2992 warnings produced by @option{-Winline} to appear or disappear.
2994 @item -Wno-invalid-offsetof @r{(C++ only)}
2995 @opindex Wno-invalid-offsetof
2996 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
2997 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
2998 to a non-POD type is undefined. In existing C++ implementations,
2999 however, @samp{offsetof} typically gives meaningful results even when
3000 applied to certain kinds of non-POD types. (Such as a simple
3001 @samp{struct} that fails to be a POD type only by virtue of having a
3002 constructor.) This flag is for users who are aware that they are
3003 writing nonportable code and who have deliberately chosen to ignore the
3006 The restrictions on @samp{offsetof} may be relaxed in a future version
3007 of the C++ standard.
3010 @opindex Winvalid-pch
3011 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
3012 the search path but can't be used.
3016 @opindex Wno-long-long
3017 Warn if @samp{long long} type is used. This is default. To inhibit
3018 the warning messages, use @option{-Wno-long-long}. Flags
3019 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
3020 only when @option{-pedantic} flag is used.
3022 @item -Wvariadic-macros
3023 @opindex Wvariadic-macros
3024 @opindex Wno-variadic-macros
3025 Warn if variadic macros are used in pedantic ISO C90 mode, or the GNU
3026 alternate syntax when in pedantic ISO C99 mode. This is default.
3027 To inhibit the warning messages, use @option{-Wno-variadic-macros}.
3029 @item -Wdisabled-optimization
3030 @opindex Wdisabled-optimization
3031 Warn if a requested optimization pass is disabled. This warning does
3032 not generally indicate that there is anything wrong with your code; it
3033 merely indicates that GCC's optimizers were unable to handle the code
3034 effectively. Often, the problem is that your code is too big or too
3035 complex; GCC will refuse to optimize programs when the optimization
3036 itself is likely to take inordinate amounts of time.
3040 Make all warnings into errors.
3043 @node Debugging Options
3044 @section Options for Debugging Your Program or GCC
3045 @cindex options, debugging
3046 @cindex debugging information options
3048 GCC has various special options that are used for debugging
3049 either your program or GCC:
3054 Produce debugging information in the operating system's native format
3055 (stabs, COFF, XCOFF, or DWARF)@. GDB can work with this debugging
3058 On most systems that use stabs format, @option{-g} enables use of extra
3059 debugging information that only GDB can use; this extra information
3060 makes debugging work better in GDB but will probably make other debuggers
3062 refuse to read the program. If you want to control for certain whether
3063 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
3064 @option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
3066 Unlike most other C compilers, GCC allows you to use @option{-g} with
3067 @option{-O}. The shortcuts taken by optimized code may occasionally
3068 produce surprising results: some variables you declared may not exist
3069 at all; flow of control may briefly move where you did not expect it;
3070 some statements may not be executed because they compute constant
3071 results or their values were already at hand; some statements may
3072 execute in different places because they were moved out of loops.
3074 Nevertheless it proves possible to debug optimized output. This makes
3075 it reasonable to use the optimizer for programs that might have bugs.
3077 The following options are useful when GCC is generated with the
3078 capability for more than one debugging format.
3082 Produce debugging information for use by GDB@. This means to use the
3083 most expressive format available (DWARF 2, stabs, or the native format
3084 if neither of those are supported), including GDB extensions if at all
3089 Produce debugging information in stabs format (if that is supported),
3090 without GDB extensions. This is the format used by DBX on most BSD
3091 systems. On MIPS, Alpha and System V Release 4 systems this option
3092 produces stabs debugging output which is not understood by DBX or SDB@.
3093 On System V Release 4 systems this option requires the GNU assembler.
3095 @item -feliminate-unused-debug-symbols
3096 @opindex feliminate-unused-debug-symbols
3097 Produce debugging information in stabs format (if that is supported),
3098 for only symbols that are actually used.
3102 Produce debugging information in stabs format (if that is supported),
3103 using GNU extensions understood only by the GNU debugger (GDB)@. The
3104 use of these extensions is likely to make other debuggers crash or
3105 refuse to read the program.
3109 Produce debugging information in COFF format (if that is supported).
3110 This is the format used by SDB on most System V systems prior to
3115 Produce debugging information in XCOFF format (if that is supported).
3116 This is the format used by the DBX debugger on IBM RS/6000 systems.
3120 Produce debugging information in XCOFF format (if that is supported),
3121 using GNU extensions understood only by the GNU debugger (GDB)@. The
3122 use of these extensions is likely to make other debuggers crash or
3123 refuse to read the program, and may cause assemblers other than the GNU
3124 assembler (GAS) to fail with an error.
3128 Produce debugging information in DWARF version 2 format (if that is
3129 supported). This is the format used by DBX on IRIX 6.
3133 Produce debugging information in VMS debug format (if that is
3134 supported). This is the format used by DEBUG on VMS systems.
3137 @itemx -ggdb@var{level}
3138 @itemx -gstabs@var{level}
3139 @itemx -gcoff@var{level}
3140 @itemx -gxcoff@var{level}
3141 @itemx -gvms@var{level}
3142 Request debugging information and also use @var{level} to specify how
3143 much information. The default level is 2.
3145 Level 1 produces minimal information, enough for making backtraces in
3146 parts of the program that you don't plan to debug. This includes
3147 descriptions of functions and external variables, but no information
3148 about local variables and no line numbers.
3150 Level 3 includes extra information, such as all the macro definitions
3151 present in the program. Some debuggers support macro expansion when
3152 you use @option{-g3}.
3154 Note that in order to avoid confusion between DWARF1 debug level 2,
3155 and DWARF2 @option{-gdwarf-2} does not accept a concatenated debug
3156 level. Instead use an additional @option{-g@var{level}} option to
3157 change the debug level for DWARF2.
3159 @item -feliminate-dwarf2-dups
3160 @opindex feliminate-dwarf2-dups
3161 Compress DWARF2 debugging information by eliminating duplicated
3162 information about each symbol. This option only makes sense when
3163 generating DWARF2 debugging information with @option{-gdwarf-2}.
3165 @cindex @command{prof}
3168 Generate extra code to write profile information suitable for the
3169 analysis program @command{prof}. You must use this option when compiling
3170 the source files you want data about, and you must also use it when
3173 @cindex @command{gprof}
3176 Generate extra code to write profile information suitable for the
3177 analysis program @command{gprof}. You must use this option when compiling
3178 the source files you want data about, and you must also use it when
3183 Makes the compiler print out each function name as it is compiled, and
3184 print some statistics about each pass when it finishes.
3187 @opindex ftime-report
3188 Makes the compiler print some statistics about the time consumed by each
3189 pass when it finishes.
3192 @opindex fmem-report
3193 Makes the compiler print some statistics about permanent memory
3194 allocation when it finishes.
3196 @item -fprofile-arcs
3197 @opindex fprofile-arcs
3198 Add code so that program flow @dfn{arcs} are instrumented. During
3199 execution the program records how many times each branch and call is
3200 executed and how many times it is taken or returns. When the compiled
3201 program exits it saves this data to a file called
3202 @file{@var{auxname}.gcda} for each source file. The data may be used for
3203 profile-directed optimizations (@option{-fbranch-probabilities}), or for
3204 test coverage analysis (@option{-ftest-coverage}). Each object file's
3205 @var{auxname} is generated from the name of the output file, if
3206 explicitly specified and it is not the final executable, otherwise it is
3207 the basename of the source file. In both cases any suffix is removed
3208 (e.g. @file{foo.gcda} for input file @file{dir/foo.c}, or
3209 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
3214 Compile the source files with @option{-fprofile-arcs} plus optimization
3215 and code generation options. For test coverage analysis, use the
3216 additional @option{-ftest-coverage} option. You do not need to profile
3217 every source file in a program.
3220 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
3221 (the latter implies the former).
3224 Run the program on a representative workload to generate the arc profile
3225 information. This may be repeated any number of times. You can run
3226 concurrent instances of your program, and provided that the file system
3227 supports locking, the data files will be correctly updated. Also
3228 @code{fork} calls are detected and correctly handled (double counting
3232 For profile-directed optimizations, compile the source files again with
3233 the same optimization and code generation options plus
3234 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
3235 Control Optimization}).
3238 For test coverage analysis, use @command{gcov} to produce human readable
3239 information from the @file{.gcno} and @file{.gcda} files. Refer to the
3240 @command{gcov} documentation for further information.
3244 With @option{-fprofile-arcs}, for each function of your program GCC
3245 creates a program flow graph, then finds a spanning tree for the graph.
3246 Only arcs that are not on the spanning tree have to be instrumented: the
3247 compiler adds code to count the number of times that these arcs are
3248 executed. When an arc is the only exit or only entrance to a block, the
3249 instrumentation code can be added to the block; otherwise, a new basic
3250 block must be created to hold the instrumentation code.
3252 @item -ftree-based-profiling
3253 @opindex ftree-based-profiling
3254 This option is used in addition to @option{-fprofile-arcs} or
3255 @option{-fbranch-probabilities} to control whether those optimizations
3256 are performed on a tree-based or rtl-based internal representation.
3257 If you use this option when compiling with @option{-fprofile-arcs},
3258 you must also use it when compiling later with @option{-fbranch-probabilities}.
3259 Currently the tree-based optimization is in an early stage of
3260 development, and this option is recommended only for those people
3261 working on improving it.
3264 @item -ftest-coverage
3265 @opindex ftest-coverage
3266 Produce a notes file that the @command{gcov} code-coverage utility
3267 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
3268 show program coverage. Each source file's note file is called
3269 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
3270 above for a description of @var{auxname} and instructions on how to
3271 generate test coverage data. Coverage data will match the source files
3272 more closely, if you do not optimize.
3274 @item -d@var{letters}
3276 Says to make debugging dumps during compilation at times specified by
3277 @var{letters}. This is used for debugging the compiler. The file names
3278 for most of the dumps are made by appending a pass number and a word to
3279 the @var{dumpname}. @var{dumpname} is generated from the name of the
3280 output file, if explicitly specified and it is not an executable,
3281 otherwise it is the basename of the source file. In both cases any
3282 suffix is removed (e.g. @file{foo.01.rtl} or @file{foo.02.sibling}).
3283 Here are the possible letters for use in @var{letters}, and their
3289 Annotate the assembler output with miscellaneous debugging information.
3292 Dump after computing branch probabilities, to @file{@var{file}.12.bp}.
3295 Dump after block reordering, to @file{@var{file}.32.bbro}.
3298 Dump after instruction combination, to the file @file{@var{file}.20.combine}.
3301 Dump after the first if conversion, to the file @file{@var{file}.14.ce1}.
3302 Also dump after the second if conversion, to the file @file{@var{file}.21.ce2}.
3305 Dump after branch target load optimization, to to @file{@var{file}.33.btl}.
3306 Also dump after delayed branch scheduling, to @file{@var{file}.37.dbr}.
3309 Dump all macro definitions, at the end of preprocessing, in addition to
3313 Dump after the third if conversion, to @file{@var{file}.31.ce3}.
3316 Dump after control and data flow analysis, to @file{@var{file}.11.cfg}.
3317 Also dump after life analysis, to @file{@var{file}.19.life}.
3320 Dump after purging @code{ADDRESSOF} codes, to @file{@var{file}.07.addressof}.
3323 Dump after global register allocation, to @file{@var{file}.26.greg}.
3326 Dump after GCSE, to @file{@var{file}.08.gcse}.
3327 Also dump after jump bypassing and control flow optimizations, to
3328 @file{@var{file}.10.bypass}.
3331 Dump after finalization of EH handling code, to @file{@var{file}.03.eh}.
3334 Dump after sibling call optimizations, to @file{@var{file}.02.sibling}.
3337 Dump after the first jump optimization, to @file{@var{file}.04.jump}.
3340 Dump after conversion from registers to stack, to @file{@var{file}.35.stack}.
3343 Dump after local register allocation, to @file{@var{file}.25.lreg}.
3346 Dump after loop optimization passes, to @file{@var{file}.09.loop} and
3347 @file{@var{file}.16.loop2}.
3350 Dump after modulo scheduling, to @file{@var{file}.23.sms}.
3353 Dump after performing the machine dependent reorganization pass, to
3354 @file{@var{file}.36.mach}.
3357 Dump after register renumbering, to @file{@var{file}.30.rnreg}.
3360 Dump after the register move pass, to @file{@var{file}.22.regmove}.
3363 Dump after post-reload optimizations, to @file{@var{file}.27.postreload}.
3366 Dump after RTL generation, to @file{@var{file}.01.rtl}.
3369 Dump after the second scheduling pass, to @file{@var{file}.34.sched2}.
3372 Dump after CSE (including the jump optimization that sometimes follows
3373 CSE), to @file{@var{file}.06.cse}.
3376 Dump after the first scheduling pass, to @file{@var{file}.24.sched}.
3379 Dump after the second CSE pass (including the jump optimization that
3380 sometimes follows CSE), to @file{@var{file}.18.cse2}.
3383 Dump after running tracer, to @file{@var{file}.15.tracer}.
3386 Dump after null pointer elimination pass to @file{@var{file}.05.null}.
3389 Dump callgraph and unit-at-a-time optimization @file{@var{file}.00.unit}.
3392 Dump after the value profile transformations, to @file{@var{file}.13.vpt}.
3393 Also dump after variable tracking, to @file{@var{file}.35.vartrack}.
3396 Dump after the second flow pass, to @file{@var{file}.28.flow2}.
3399 Dump after the peephole pass, to @file{@var{file}.29.peephole2}.
3402 Dump after constructing the web, to @file{@var{file}.17.web}.
3405 Produce all the dumps listed above.
3408 Produce a core dump whenever an error occurs.
3411 Print statistics on memory usage, at the end of the run, to
3415 Annotate the assembler output with a comment indicating which
3416 pattern and alternative was used. The length of each instruction is
3420 Dump the RTL in the assembler output as a comment before each instruction.
3421 Also turns on @option{-dp} annotation.
3424 For each of the other indicated dump files (except for
3425 @file{@var{file}.01.rtl}), dump a representation of the control flow graph
3426 suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
3429 Just generate RTL for a function instead of compiling it. Usually used
3433 Dump debugging information during parsing, to standard error.
3436 @item -fdump-unnumbered
3437 @opindex fdump-unnumbered
3438 When doing debugging dumps (see @option{-d} option above), suppress instruction
3439 numbers and line number note output. This makes it more feasible to
3440 use diff on debugging dumps for compiler invocations with different
3441 options, in particular with and without @option{-g}.
3443 @item -fdump-translation-unit @r{(C and C++ only)}
3444 @itemx -fdump-translation-unit-@var{options} @r{(C and C++ only)}
3445 @opindex fdump-translation-unit
3446 Dump a representation of the tree structure for the entire translation
3447 unit to a file. The file name is made by appending @file{.tu} to the
3448 source file name. If the @samp{-@var{options}} form is used, @var{options}
3449 controls the details of the dump as described for the
3450 @option{-fdump-tree} options.
3452 @item -fdump-class-hierarchy @r{(C++ only)}
3453 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
3454 @opindex fdump-class-hierarchy
3455 Dump a representation of each class's hierarchy and virtual function
3456 table layout to a file. The file name is made by appending @file{.class}
3457 to the source file name. If the @samp{-@var{options}} form is used,
3458 @var{options} controls the details of the dump as described for the
3459 @option{-fdump-tree} options.
3461 @item -fdump-tree-@var{switch} @r{(C and C++ only)}
3462 @itemx -fdump-tree-@var{switch}-@var{options} @r{(C and C++ only)}
3464 Control the dumping at various stages of processing the intermediate
3465 language tree to a file. The file name is generated by appending a switch
3466 specific suffix to the source file name. If the @samp{-@var{options}}
3467 form is used, @var{options} is a list of @samp{-} separated options that
3468 control the details of the dump. Not all options are applicable to all
3469 dumps, those which are not meaningful will be ignored. The following
3470 options are available
3474 Print the address of each node. Usually this is not meaningful as it
3475 changes according to the environment and source file. Its primary use
3476 is for tying up a dump file with a debug environment.
3478 Inhibit dumping of members of a scope or body of a function merely
3479 because that scope has been reached. Only dump such items when they
3480 are directly reachable by some other path. When dumping pretty-printed
3481 trees, this option inhibits dumping the bodies of control structures.
3483 Print a raw representation of the tree. By default, trees are
3484 pretty-printed into a C-like representation.
3486 Enable more detailed dumps (not honored by every dump option).
3488 Enable dumping various statistics about the pass (not honored by every dump
3491 Enable showing basic block boundaries (disabled in raw dumps).
3493 Enable showing virtual operands for every statement.
3495 Enable showing line numbers for statements.
3497 Enable showing the unique ID (@code{DECL_UID}) for each variable.
3499 Turn on all options, except @option{raw}, @option{slim} and @option{lineno}.
3502 The following tree dumps are possible:
3506 Dump before any tree based optimization, to @file{@var{file}.original}.
3509 Dump after all tree based optimization, to @file{@var{file}.optimized}.
3512 Dump after function inlining, to @file{@var{file}.inlined}.
3515 @opindex fdump-tree-gimple
3516 Dump each function before and after the gimplification pass to a file. The
3517 file name is made by appending @file{.gimple} to the source file name.
3520 @opindex fdump-tree-cfg
3521 Dump the control flow graph of each function to a file. The file name is
3522 made by appending @file{.cfg} to the source file name.
3525 @opindex fdump-tree-vcg
3526 Dump the control flow graph of each function to a file in VCG format. The
3527 file name is made by appending @file{.vcg} to the source file name. Note
3528 that if the file contains more than one function, the generated file cannot
3529 be used directly by VCG. You will need to cut and paste each function's
3530 graph into its own separate file first.
3533 @opindex fdump-tree-ch
3534 Dump each function after copying loop headers. The file name is made by
3535 appending @file{.ch} to the source file name.
3538 @opindex fdump-tree-ssa
3539 Dump SSA related information to a file. The file name is made by appending
3540 @file{.ssa} to the source file name.
3543 @opindex fdump-tree-alias
3544 Dump aliasing information for each function. The file name is made by
3545 appending @file{.alias} to the source file name.
3548 @opindex fdump-tree-ccp
3549 Dump each function after CCP. The file name is made by appending
3550 @file{.ccp} to the source file name.
3553 @opindex fdump-tree-pre
3554 Dump trees after partial redundancy elimination. The file name is made
3555 by appending @file{.pre} to the source file name.
3558 @opindex fdump-tree-fre
3559 Dump trees after full redundancy elimination. The file name is made
3560 by appending @file{.fre} to the source file name.
3563 @opindex fdump-tree-dce
3564 Dump each function after dead code elimination. The file name is made by
3565 appending @file{.dce} to the source file name.
3568 @opindex fdump-tree-mudflap
3569 Dump each function after adding mudflap instrumentation. The file name is
3570 made by appending @file{.mudflap} to the source file name.
3573 @opindex fdump-tree-sra
3574 Dump each function after performing scalar replacement of aggregates. The
3575 file name is made by appending @file{.sra} to the source file name.
3578 @opindex fdump-tree-dom
3579 Dump each function after applying dominator tree optimizations. The file
3580 name is made by appending @file{.dom} to the source file name.
3583 @opindex fdump-tree-dse
3584 Dump each function after applying dead store elimination. The file
3585 name is made by appending @file{.dse} to the source file name.
3588 @opindex fdump-tree-phiopt
3589 Dump each function after optimizing PHI nodes into straightline code. The file
3590 name is made by appending @file{.phiopt} to the source file name.
3593 @opindex fdump-tree-forwprop
3594 Dump each function after forward propagating single use variables. The file
3595 name is made by appending @file{.forwprop} to the source file name.
3598 @opindex fdump-tree-copyrename
3599 Dump each function after applying the copy rename optimization. The file
3600 name is made by appending @file{.copyrename} to the source file name.
3603 @opindex fdump-tree-nrv
3604 Dump each function after applying the named return value optimization on
3605 generic trees. The file name is made by appending @file{.nrv} to the source
3609 @opindex fdump-tree-all
3610 Enable all the available tree dumps with the flags provided in this option.
3613 @item -frandom-seed=@var{string}
3614 @opindex frandom-string
3615 This option provides a seed that GCC uses when it would otherwise use
3616 random numbers. It is used to generate certain symbol names
3617 that have to be different in every compiled file. It is also used to
3618 place unique stamps in coverage data files and the object files that
3619 produce them. You can use the @option{-frandom-seed} option to produce
3620 reproducibly identical object files.
3622 The @var{string} should be different for every file you compile.
3624 @item -fsched-verbose=@var{n}
3625 @opindex fsched-verbose
3626 On targets that use instruction scheduling, this option controls the
3627 amount of debugging output the scheduler prints. This information is
3628 written to standard error, unless @option{-dS} or @option{-dR} is
3629 specified, in which case it is output to the usual dump
3630 listing file, @file{.sched} or @file{.sched2} respectively. However
3631 for @var{n} greater than nine, the output is always printed to standard
3634 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
3635 same information as @option{-dRS}. For @var{n} greater than one, it
3636 also output basic block probabilities, detailed ready list information
3637 and unit/insn info. For @var{n} greater than two, it includes RTL
3638 at abort point, control-flow and regions info. And for @var{n} over
3639 four, @option{-fsched-verbose} also includes dependence info.
3643 Store the usual ``temporary'' intermediate files permanently; place them
3644 in the current directory and name them based on the source file. Thus,
3645 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
3646 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
3647 preprocessed @file{foo.i} output file even though the compiler now
3648 normally uses an integrated preprocessor.
3652 Report the CPU time taken by each subprocess in the compilation
3653 sequence. For C source files, this is the compiler proper and assembler
3654 (plus the linker if linking is done). The output looks like this:
3661 The first number on each line is the ``user time,'' that is time spent
3662 executing the program itself. The second number is ``system time,''
3663 time spent executing operating system routines on behalf of the program.
3664 Both numbers are in seconds.
3666 @item -fvar-tracking
3667 @opindex fvar-tracking
3668 Run variable tracking pass. It computes where variables are stored at each
3669 position in code. Better debugging information is then generated
3670 (if the debugging information format supports this information).
3672 It is enabled by default when compiling with optimization (@option{-Os},
3673 @option{-O}, @option{-O2}, ...), debugging information (@option{-g}) and
3674 the debug info format supports it.
3676 @item -print-file-name=@var{library}
3677 @opindex print-file-name
3678 Print the full absolute name of the library file @var{library} that
3679 would be used when linking---and don't do anything else. With this
3680 option, GCC does not compile or link anything; it just prints the
3683 @item -print-multi-directory
3684 @opindex print-multi-directory
3685 Print the directory name corresponding to the multilib selected by any
3686 other switches present in the command line. This directory is supposed
3687 to exist in @env{GCC_EXEC_PREFIX}.
3689 @item -print-multi-lib
3690 @opindex print-multi-lib
3691 Print the mapping from multilib directory names to compiler switches
3692 that enable them. The directory name is separated from the switches by
3693 @samp{;}, and each switch starts with an @samp{@@} instead of the
3694 @samp{-}, without spaces between multiple switches. This is supposed to
3695 ease shell-processing.
3697 @item -print-prog-name=@var{program}
3698 @opindex print-prog-name
3699 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
3701 @item -print-libgcc-file-name
3702 @opindex print-libgcc-file-name
3703 Same as @option{-print-file-name=libgcc.a}.
3705 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
3706 but you do want to link with @file{libgcc.a}. You can do
3709 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
3712 @item -print-search-dirs
3713 @opindex print-search-dirs
3714 Print the name of the configured installation directory and a list of
3715 program and library directories @command{gcc} will search---and don't do anything else.
3717 This is useful when @command{gcc} prints the error message
3718 @samp{installation problem, cannot exec cpp0: No such file or directory}.
3719 To resolve this you either need to put @file{cpp0} and the other compiler
3720 components where @command{gcc} expects to find them, or you can set the environment
3721 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
3722 Don't forget the trailing '/'.
3723 @xref{Environment Variables}.
3726 @opindex dumpmachine
3727 Print the compiler's target machine (for example,
3728 @samp{i686-pc-linux-gnu})---and don't do anything else.
3731 @opindex dumpversion
3732 Print the compiler version (for example, @samp{3.0})---and don't do
3737 Print the compiler's built-in specs---and don't do anything else. (This
3738 is used when GCC itself is being built.) @xref{Spec Files}.
3740 @item -feliminate-unused-debug-types
3741 @opindex feliminate-unused-debug-types
3742 Normally, when producing DWARF2 output, GCC will emit debugging
3743 information for all types declared in a compilation
3744 unit, regardless of whether or not they are actually used
3745 in that compilation unit. Sometimes this is useful, such as
3746 if, in the debugger, you want to cast a value to a type that is
3747 not actually used in your program (but is declared). More often,
3748 however, this results in a significant amount of wasted space.
3749 With this option, GCC will avoid producing debug symbol output
3750 for types that are nowhere used in the source file being compiled.
3753 @node Optimize Options
3754 @section Options That Control Optimization
3755 @cindex optimize options
3756 @cindex options, optimization
3758 These options control various sorts of optimizations.
3760 Without any optimization option, the compiler's goal is to reduce the
3761 cost of compilation and to make debugging produce the expected
3762 results. Statements are independent: if you stop the program with a
3763 breakpoint between statements, you can then assign a new value to any
3764 variable or change the program counter to any other statement in the
3765 function and get exactly the results you would expect from the source
3768 Turning on optimization flags makes the compiler attempt to improve
3769 the performance and/or code size at the expense of compilation time
3770 and possibly the ability to debug the program.
3772 The compiler performs optimization based on the knowledge it has of
3773 the program. Optimization levels @option{-O2} and above, in
3774 particular, enable @emph{unit-at-a-time} mode, which allows the
3775 compiler to consider information gained from later functions in
3776 the file when compiling a function. Compiling multiple files at
3777 once to a single output file in @emph{unit-at-a-time} mode allows
3778 the compiler to use information gained from all of the files when
3779 compiling each of them.
3781 Not all optimizations are controlled directly by a flag. Only
3782 optimizations that have a flag are listed.
3789 Optimize. Optimizing compilation takes somewhat more time, and a lot
3790 more memory for a large function.
3792 With @option{-O}, the compiler tries to reduce code size and execution
3793 time, without performing any optimizations that take a great deal of
3796 @option{-O} turns on the following optimization flags:
3797 @gccoptlist{-fdefer-pop @gol
3798 -fmerge-constants @gol
3800 -floop-optimize @gol
3801 -fif-conversion @gol
3802 -fif-conversion2 @gol
3803 -fdelayed-branch @gol
3804 -fguess-branch-probability @gol
3807 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
3808 where doing so does not interfere with debugging.
3812 Optimize even more. GCC performs nearly all supported optimizations
3813 that do not involve a space-speed tradeoff. The compiler does not
3814 perform loop unrolling or function inlining when you specify @option{-O2}.
3815 As compared to @option{-O}, this option increases both compilation time
3816 and the performance of the generated code.
3818 @option{-O2} turns on all optimization flags specified by @option{-O}. It
3819 also turns on the following optimization flags:
3820 @gccoptlist{-fforce-mem @gol
3821 -foptimize-sibling-calls @gol
3822 -fstrength-reduce @gol
3823 -fcse-follow-jumps -fcse-skip-blocks @gol
3824 -frerun-cse-after-loop -frerun-loop-opt @gol
3825 -fgcse -fgcse-lm -fgcse-sm -fgcse-las @gol
3826 -fdelete-null-pointer-checks @gol
3827 -fexpensive-optimizations @gol
3829 -fschedule-insns -fschedule-insns2 @gol
3830 -fsched-interblock -fsched-spec @gol
3833 -freorder-blocks -freorder-functions @gol
3834 -fstrict-aliasing @gol
3835 -funit-at-a-time @gol
3836 -falign-functions -falign-jumps @gol
3837 -falign-loops -falign-labels @gol
3840 Please note the warning under @option{-fgcse} about
3841 invoking @option{-O2} on programs that use computed gotos.
3845 Optimize yet more. @option{-O3} turns on all optimizations specified by
3846 @option{-O2} and also turns on the @option{-finline-functions},
3847 @option{-fweb}, @option{-frename-registers}
3848 and @option{-fgcse-after-reload} options.
3852 Do not optimize. This is the default.
3856 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
3857 do not typically increase code size. It also performs further
3858 optimizations designed to reduce code size.
3860 @option{-Os} disables the following optimization flags:
3861 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
3862 -falign-labels -freorder-blocks -freorder-blocks-and-partition -fprefetch-loop-arrays}
3864 If you use multiple @option{-O} options, with or without level numbers,
3865 the last such option is the one that is effective.
3868 Options of the form @option{-f@var{flag}} specify machine-independent
3869 flags. Most flags have both positive and negative forms; the negative
3870 form of @option{-ffoo} would be @option{-fno-foo}. In the table
3871 below, only one of the forms is listed---the one you typically will
3872 use. You can figure out the other form by either removing @samp{no-}
3875 The following options control specific optimizations. They are either
3876 activated by @option{-O} options or are related to ones that are. You
3877 can use the following flags in the rare cases when ``fine-tuning'' of
3878 optimizations to be performed is desired.
3881 @item -fno-default-inline
3882 @opindex fno-default-inline
3883 Do not make member functions inline by default merely because they are
3884 defined inside the class scope (C++ only). Otherwise, when you specify
3885 @w{@option{-O}}, member functions defined inside class scope are compiled
3886 inline by default; i.e., you don't need to add @samp{inline} in front of
3887 the member function name.
3889 @item -fno-defer-pop
3890 @opindex fno-defer-pop
3891 Always pop the arguments to each function call as soon as that function
3892 returns. For machines which must pop arguments after a function call,
3893 the compiler normally lets arguments accumulate on the stack for several
3894 function calls and pops them all at once.
3896 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3900 Force memory operands to be copied into registers before doing
3901 arithmetic on them. This produces better code by making all memory
3902 references potential common subexpressions. When they are not common
3903 subexpressions, instruction combination should eliminate the separate
3906 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3909 @opindex fforce-addr
3910 Force memory address constants to be copied into registers before
3911 doing arithmetic on them. This may produce better code just as
3912 @option{-fforce-mem} may.
3914 @item -fomit-frame-pointer
3915 @opindex fomit-frame-pointer
3916 Don't keep the frame pointer in a register for functions that
3917 don't need one. This avoids the instructions to save, set up and
3918 restore frame pointers; it also makes an extra register available
3919 in many functions. @strong{It also makes debugging impossible on
3922 On some machines, such as the VAX, this flag has no effect, because
3923 the standard calling sequence automatically handles the frame pointer
3924 and nothing is saved by pretending it doesn't exist. The
3925 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
3926 whether a target machine supports this flag. @xref{Registers,,Register
3927 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
3929 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
3931 @item -foptimize-sibling-calls
3932 @opindex foptimize-sibling-calls
3933 Optimize sibling and tail recursive calls.
3935 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
3939 Don't pay attention to the @code{inline} keyword. Normally this option
3940 is used to keep the compiler from expanding any functions inline.
3941 Note that if you are not optimizing, no functions can be expanded inline.
3943 @item -finline-functions
3944 @opindex finline-functions
3945 Integrate all simple functions into their callers. The compiler
3946 heuristically decides which functions are simple enough to be worth
3947 integrating in this way.
3949 If all calls to a given function are integrated, and the function is
3950 declared @code{static}, then the function is normally not output as
3951 assembler code in its own right.
3953 Enabled at level @option{-O3}.
3955 @item -finline-limit=@var{n}
3956 @opindex finline-limit
3957 By default, GCC limits the size of functions that can be inlined. This flag
3958 allows the control of this limit for functions that are explicitly marked as
3959 inline (i.e., marked with the inline keyword or defined within the class
3960 definition in c++). @var{n} is the size of functions that can be inlined in
3961 number of pseudo instructions (not counting parameter handling). The default
3962 value of @var{n} is 600.
3963 Increasing this value can result in more inlined code at
3964 the cost of compilation time and memory consumption. Decreasing usually makes
3965 the compilation faster and less code will be inlined (which presumably
3966 means slower programs). This option is particularly useful for programs that
3967 use inlining heavily such as those based on recursive templates with C++.
3969 Inlining is actually controlled by a number of parameters, which may be
3970 specified individually by using @option{--param @var{name}=@var{value}}.
3971 The @option{-finline-limit=@var{n}} option sets some of these parameters
3975 @item max-inline-insns-single
3976 is set to @var{n}/2.
3977 @item max-inline-insns-auto
3978 is set to @var{n}/2.
3979 @item min-inline-insns
3980 is set to 130 or @var{n}/4, whichever is smaller.
3981 @item max-inline-insns-rtl
3985 See below for a documentation of the individual
3986 parameters controlling inlining.
3988 @emph{Note:} pseudo instruction represents, in this particular context, an
3989 abstract measurement of function's size. In no way, it represents a count
3990 of assembly instructions and as such its exact meaning might change from one
3991 release to an another.
3993 @item -fkeep-inline-functions
3994 @opindex fkeep-inline-functions
3995 Even if all calls to a given function are integrated, and the function
3996 is declared @code{static}, nevertheless output a separate run-time
3997 callable version of the function. This switch does not affect
3998 @code{extern inline} functions.
4000 @item -fkeep-static-consts
4001 @opindex fkeep-static-consts
4002 Emit variables declared @code{static const} when optimization isn't turned
4003 on, even if the variables aren't referenced.
4005 GCC enables this option by default. If you want to force the compiler to
4006 check if the variable was referenced, regardless of whether or not
4007 optimization is turned on, use the @option{-fno-keep-static-consts} option.
4009 @item -fmerge-constants
4010 Attempt to merge identical constants (string constants and floating point
4011 constants) across compilation units.
4013 This option is the default for optimized compilation if the assembler and
4014 linker support it. Use @option{-fno-merge-constants} to inhibit this
4017 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4019 @item -fmerge-all-constants
4020 Attempt to merge identical constants and identical variables.
4022 This option implies @option{-fmerge-constants}. In addition to
4023 @option{-fmerge-constants} this considers e.g. even constant initialized
4024 arrays or initialized constant variables with integral or floating point
4025 types. Languages like C or C++ require each non-automatic variable to
4026 have distinct location, so using this option will result in non-conforming
4029 @item -fmodulo-sched
4030 @opindex fmodulo-sched
4031 Perform swing modulo scheduling immediately before the first scheduling
4032 pass. This pass looks at innermost loops and reorders their
4033 instructions by overlapping different iterations.
4037 Use a graph coloring register allocator. Currently this option is meant
4038 only for testing. Users should not specify this option, since it is not
4039 yet ready for production use.
4041 @item -fno-branch-count-reg
4042 @opindex fno-branch-count-reg
4043 Do not use ``decrement and branch'' instructions on a count register,
4044 but instead generate a sequence of instructions that decrement a
4045 register, compare it against zero, then branch based upon the result.
4046 This option is only meaningful on architectures that support such
4047 instructions, which include x86, PowerPC, IA-64 and S/390.
4049 The default is @option{-fbranch-count-reg}, enabled when
4050 @option{-fstrength-reduce} is enabled.
4052 @item -fno-function-cse
4053 @opindex fno-function-cse
4054 Do not put function addresses in registers; make each instruction that
4055 calls a constant function contain the function's address explicitly.
4057 This option results in less efficient code, but some strange hacks
4058 that alter the assembler output may be confused by the optimizations
4059 performed when this option is not used.
4061 The default is @option{-ffunction-cse}
4063 @item -fno-zero-initialized-in-bss
4064 @opindex fno-zero-initialized-in-bss
4065 If the target supports a BSS section, GCC by default puts variables that
4066 are initialized to zero into BSS@. This can save space in the resulting
4069 This option turns off this behavior because some programs explicitly
4070 rely on variables going to the data section. E.g., so that the
4071 resulting executable can find the beginning of that section and/or make
4072 assumptions based on that.
4074 The default is @option{-fzero-initialized-in-bss}.
4076 @item -fbounds-check
4077 @opindex fbounds-check
4078 For front-ends that support it, generate additional code to check that
4079 indices used to access arrays are within the declared range. This is
4080 currently only supported by the Java and Fortran front-ends, where
4081 this option defaults to true and false respectively.
4083 @item -fmudflap -fmudflapth -fmudflapir
4087 @cindex bounds checking
4089 For front-ends that support it (C and C++), instrument all risky
4090 pointer/array dereferencing operations, some standard library
4091 string/heap functions, and some other associated constructs with
4092 range/validity tests. Modules so instrumented should be immune to
4093 buffer overflows, invalid heap use, and some other classes of C/C++
4094 programming errors. The instrumentation relies on a separate runtime
4095 library (@file{libmudflap}), which will be linked into a program if
4096 @option{-fmudflap} is given at link time. Run-time behavior of the
4097 instrumented program is controlled by the @env{MUDFLAP_OPTIONS}
4098 environment variable. See @code{env MUDFLAP_OPTIONS=-help a.out}
4101 Use @option{-fmudflapth} instead of @option{-fmudflap} to compile and to
4102 link if your program is multi-threaded. Use @option{-fmudflapir}, in
4103 addition to @option{-fmudflap} or @option{-fmudflapth}, if
4104 instrumentation should ignore pointer reads. This produces less
4105 instrumentation (and therefore faster execution) and still provides
4106 some protection against outright memory corrupting writes, but allows
4107 erroneously read data to propagate within a program.
4109 @item -fstrength-reduce
4110 @opindex fstrength-reduce
4111 Perform the optimizations of loop strength reduction and
4112 elimination of iteration variables.
4114 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4116 @item -fthread-jumps
4117 @opindex fthread-jumps
4118 Perform optimizations where we check to see if a jump branches to a
4119 location where another comparison subsumed by the first is found. If
4120 so, the first branch is redirected to either the destination of the
4121 second branch or a point immediately following it, depending on whether
4122 the condition is known to be true or false.
4124 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4126 @item -fcse-follow-jumps
4127 @opindex fcse-follow-jumps
4128 In common subexpression elimination, scan through jump instructions
4129 when the target of the jump is not reached by any other path. For
4130 example, when CSE encounters an @code{if} statement with an
4131 @code{else} clause, CSE will follow the jump when the condition
4134 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4136 @item -fcse-skip-blocks
4137 @opindex fcse-skip-blocks
4138 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
4139 follow jumps which conditionally skip over blocks. When CSE
4140 encounters a simple @code{if} statement with no else clause,
4141 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
4142 body of the @code{if}.
4144 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4146 @item -frerun-cse-after-loop
4147 @opindex frerun-cse-after-loop
4148 Re-run common subexpression elimination after loop optimizations has been
4151 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4153 @item -frerun-loop-opt
4154 @opindex frerun-loop-opt
4155 Run the loop optimizer twice.
4157 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4161 Perform a global common subexpression elimination pass.
4162 This pass also performs global constant and copy propagation.
4164 @emph{Note:} When compiling a program using computed gotos, a GCC
4165 extension, you may get better runtime performance if you disable
4166 the global common subexpression elimination pass by adding
4167 @option{-fno-gcse} to the command line.
4169 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4173 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
4174 attempt to move loads which are only killed by stores into themselves. This
4175 allows a loop containing a load/store sequence to be changed to a load outside
4176 the loop, and a copy/store within the loop.
4178 Enabled by default when gcse is enabled.
4182 When @option{-fgcse-sm} is enabled, a store motion pass is run after
4183 global common subexpression elimination. This pass will attempt to move
4184 stores out of loops. When used in conjunction with @option{-fgcse-lm},
4185 loops containing a load/store sequence can be changed to a load before
4186 the loop and a store after the loop.
4188 Enabled by default when gcse is enabled.
4192 When @option{-fgcse-las} is enabled, the global common subexpression
4193 elimination pass eliminates redundant loads that come after stores to the
4194 same memory location (both partial and full redundancies).
4196 Enabled by default when gcse is enabled.
4198 @item -fgcse-after-reload
4199 @opindex fgcse-after-reload
4200 When @option{-fgcse-after-reload} is enabled, a redundant load elimination
4201 pass is performed after reload. The purpose of this pass is to cleanup
4204 @item -floop-optimize
4205 @opindex floop-optimize
4206 Perform loop optimizations: move constant expressions out of loops, simplify
4207 exit test conditions and optionally do strength-reduction and loop unrolling as
4210 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4212 @item -floop-optimize2
4213 @opindex floop-optimize2
4214 Perform loop optimizations using the new loop optimizer. The optimizations
4215 (loop unrolling, peeling and unswitching, loop invariant motion) are enabled
4218 @item -fcrossjumping
4219 @opindex crossjumping
4220 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
4221 resulting code may or may not perform better than without cross-jumping.
4223 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4225 @item -fif-conversion
4226 @opindex if-conversion
4227 Attempt to transform conditional jumps into branch-less equivalents. This
4228 include use of conditional moves, min, max, set flags and abs instructions, and
4229 some tricks doable by standard arithmetics. The use of conditional execution
4230 on chips where it is available is controlled by @code{if-conversion2}.
4232 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4234 @item -fif-conversion2
4235 @opindex if-conversion2
4236 Use conditional execution (where available) to transform conditional jumps into
4237 branch-less equivalents.
4239 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4241 @item -fdelete-null-pointer-checks
4242 @opindex fdelete-null-pointer-checks
4243 Use global dataflow analysis to identify and eliminate useless checks
4244 for null pointers. The compiler assumes that dereferencing a null
4245 pointer would have halted the program. If a pointer is checked after
4246 it has already been dereferenced, it cannot be null.
4248 In some environments, this assumption is not true, and programs can
4249 safely dereference null pointers. Use
4250 @option{-fno-delete-null-pointer-checks} to disable this optimization
4251 for programs which depend on that behavior.
4253 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4255 @item -fexpensive-optimizations
4256 @opindex fexpensive-optimizations
4257 Perform a number of minor optimizations that are relatively expensive.
4259 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4261 @item -foptimize-register-move
4263 @opindex foptimize-register-move
4265 Attempt to reassign register numbers in move instructions and as
4266 operands of other simple instructions in order to maximize the amount of
4267 register tying. This is especially helpful on machines with two-operand
4270 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
4273 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4275 @item -fdelayed-branch
4276 @opindex fdelayed-branch
4277 If supported for the target machine, attempt to reorder instructions
4278 to exploit instruction slots available after delayed branch
4281 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4283 @item -fschedule-insns
4284 @opindex fschedule-insns
4285 If supported for the target machine, attempt to reorder instructions to
4286 eliminate execution stalls due to required data being unavailable. This
4287 helps machines that have slow floating point or memory load instructions
4288 by allowing other instructions to be issued until the result of the load
4289 or floating point instruction is required.
4291 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4293 @item -fschedule-insns2
4294 @opindex fschedule-insns2
4295 Similar to @option{-fschedule-insns}, but requests an additional pass of
4296 instruction scheduling after register allocation has been done. This is
4297 especially useful on machines with a relatively small number of
4298 registers and where memory load instructions take more than one cycle.
4300 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4302 @item -fno-sched-interblock
4303 @opindex fno-sched-interblock
4304 Don't schedule instructions across basic blocks. This is normally
4305 enabled by default when scheduling before register allocation, i.e.@:
4306 with @option{-fschedule-insns} or at @option{-O2} or higher.
4308 @item -fno-sched-spec
4309 @opindex fno-sched-spec
4310 Don't allow speculative motion of non-load instructions. This is normally
4311 enabled by default when scheduling before register allocation, i.e.@:
4312 with @option{-fschedule-insns} or at @option{-O2} or higher.
4314 @item -fsched-spec-load
4315 @opindex fsched-spec-load
4316 Allow speculative motion of some load instructions. This only makes
4317 sense when scheduling before register allocation, i.e.@: with
4318 @option{-fschedule-insns} or at @option{-O2} or higher.
4320 @item -fsched-spec-load-dangerous
4321 @opindex fsched-spec-load-dangerous
4322 Allow speculative motion of more load instructions. This only makes
4323 sense when scheduling before register allocation, i.e.@: with
4324 @option{-fschedule-insns} or at @option{-O2} or higher.
4326 @item -fsched-stalled-insns=@var{n}
4327 @opindex fsched-stalled-insns
4328 Define how many insns (if any) can be moved prematurely from the queue
4329 of stalled insns into the ready list, during the second scheduling pass.
4331 @item -fsched-stalled-insns-dep=@var{n}
4332 @opindex fsched-stalled-insns-dep
4333 Define how many insn groups (cycles) will be examined for a dependency
4334 on a stalled insn that is candidate for premature removal from the queue
4335 of stalled insns. Has an effect only during the second scheduling pass,
4336 and only if @option{-fsched-stalled-insns} is used and its value is not zero.
4338 @item -fsched2-use-superblocks
4339 @opindex fsched2-use-superblocks
4340 When scheduling after register allocation, do use superblock scheduling
4341 algorithm. Superblock scheduling allows motion across basic block boundaries
4342 resulting on faster schedules. This option is experimental, as not all machine
4343 descriptions used by GCC model the CPU closely enough to avoid unreliable
4344 results from the algorithm.
4346 This only makes sense when scheduling after register allocation, i.e.@: with
4347 @option{-fschedule-insns2} or at @option{-O2} or higher.
4349 @item -fsched2-use-traces
4350 @opindex fsched2-use-traces
4351 Use @option{-fsched2-use-superblocks} algorithm when scheduling after register
4352 allocation and additionally perform code duplication in order to increase the
4353 size of superblocks using tracer pass. See @option{-ftracer} for details on
4356 This mode should produce faster but significantly longer programs. Also
4357 without @code{-fbranch-probabilities} the traces constructed may not match the
4358 reality and hurt the performance. This only makes
4359 sense when scheduling after register allocation, i.e.@: with
4360 @option{-fschedule-insns2} or at @option{-O2} or higher.
4362 @item -fcaller-saves
4363 @opindex fcaller-saves
4364 Enable values to be allocated in registers that will be clobbered by
4365 function calls, by emitting extra instructions to save and restore the
4366 registers around such calls. Such allocation is done only when it
4367 seems to result in better code than would otherwise be produced.
4369 This option is always enabled by default on certain machines, usually
4370 those which have no call-preserved registers to use instead.
4372 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4375 Perform Partial Redundancy Elimination (PRE) on trees. This flag is
4376 enabled by default at -O and higher.
4379 Perform Full Redundancy Elimination (FRE) on trees. The difference
4380 between FRE and PRE is that FRE only considers expressions
4381 that are computed on all paths leading to the redundant computation.
4382 This analysis faster than PRE, though it exposes fewer redundancies.
4383 This flag is enabled by default at -O and higher.
4386 Perform sparse conditional constant propagation (CCP) on trees. This flag
4387 is enabled by default at -O and higher.
4390 Perform dead code elimination (DCE) on trees. This flag is enabled by
4391 default at -O and higher.
4393 @item -ftree-dominator-opts
4394 Perform dead code elimination (DCE) on trees. This flag is enabled by
4395 default at -O and higher.
4398 Perform loop header copying on trees. This is beneficial since it increases
4399 effectivity of code motion optimizations. It also saves one jump. This flag
4400 is enabled by default at -O and higher. It is not enabled for -Os, since it
4401 usually increases code size.
4403 @item -ftree-loop-optimize
4404 Perform loop optimizations on trees. This flag is enabled by default at -O
4408 Perform scalar replacement of aggregates. This pass replaces structure
4409 references with scalars to prevent committing structures to memory too
4410 early. This flag is enabled by default at -O and higher.
4412 @item -ftree-copyrename
4413 Perform copy renaming on trees. This pass attempts to rename compiler
4414 temporaries to other variables at copy locations, usually resulting in
4415 variable names which more closely resemble the original variables. This flag
4416 is enabled by default at -O and higher.
4419 Perform temporary expression replacement during the SSA->normal phase. Single
4420 use/single def temporaries are replaced at their use location with their
4421 defining expression. This results in non-GIMPLE code, but gives the expanders
4422 much more complex trees to work on resulting in better RTL generation. This is
4423 enabled by default at -O and higher.
4426 Perform live range splitting during the SSA->normal phase. Distinct live
4427 ranges of a variable are split into unique variables, allowing for better
4428 optimization later. This is enabled by default at -O and higher.
4432 Perform tail duplication to enlarge superblock size. This transformation
4433 simplifies the control flow of the function allowing other optimizations to do
4436 @item -funroll-loops
4437 @opindex funroll-loops
4438 Unroll loops whose number of iterations can be determined at compile
4439 time or upon entry to the loop. @option{-funroll-loops} implies both
4440 @option{-fstrength-reduce} and @option{-frerun-cse-after-loop}. This
4441 option makes code larger, and may or may not make it run faster.
4443 @item -funroll-all-loops
4444 @opindex funroll-all-loops
4445 Unroll all loops, even if their number of iterations is uncertain when
4446 the loop is entered. This usually makes programs run more slowly.
4447 @option{-funroll-all-loops} implies the same options as
4448 @option{-funroll-loops},
4450 @item -fprefetch-loop-arrays
4451 @opindex fprefetch-loop-arrays
4452 If supported by the target machine, generate instructions to prefetch
4453 memory to improve the performance of loops that access large arrays.
4455 @item -fmove-all-movables
4456 @opindex fmove-all-movables
4457 Forces all invariant computations in loops to be moved
4460 @item -freduce-all-givs
4461 @opindex freduce-all-givs
4462 Forces all general-induction variables in loops to be
4465 @emph{Note:} When compiling programs written in Fortran,
4466 @option{-fmove-all-movables} and @option{-freduce-all-givs} are enabled
4467 by default when you use the optimizer.
4469 These options may generate better or worse code; results are highly
4470 dependent on the structure of loops within the source code.
4472 These two options are intended to be removed someday, once
4473 they have helped determine the efficacy of various
4474 approaches to improving loop optimizations.
4476 Please let us (@w{@email{gcc@@gcc.gnu.org}} and @w{@email{fortran@@gnu.org}})
4477 know how use of these options affects
4478 the performance of your production code.
4479 We're very interested in code that runs @emph{slower}
4480 when these options are @emph{enabled}.
4483 @itemx -fno-peephole2
4484 @opindex fno-peephole
4485 @opindex fno-peephole2
4486 Disable any machine-specific peephole optimizations. The difference
4487 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
4488 are implemented in the compiler; some targets use one, some use the
4489 other, a few use both.
4491 @option{-fpeephole} is enabled by default.
4492 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4494 @item -fno-guess-branch-probability
4495 @opindex fno-guess-branch-probability
4496 Do not guess branch probabilities using a randomized model.
4498 Sometimes GCC will opt to use a randomized model to guess branch
4499 probabilities, when none are available from either profiling feedback
4500 (@option{-fprofile-arcs}) or @samp{__builtin_expect}. This means that
4501 different runs of the compiler on the same program may produce different
4504 In a hard real-time system, people don't want different runs of the
4505 compiler to produce code that has different behavior; minimizing
4506 non-determinism is of paramount import. This switch allows users to
4507 reduce non-determinism, possibly at the expense of inferior
4510 The default is @option{-fguess-branch-probability} at levels
4511 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4513 @item -freorder-blocks
4514 @opindex freorder-blocks
4515 Reorder basic blocks in the compiled function in order to reduce number of
4516 taken branches and improve code locality.
4518 Enabled at levels @option{-O2}, @option{-O3}.
4520 @item -freorder-blocks-and-partition
4521 @opindex freorder-blocks-and-partition
4522 In addition to reordering basic blocks in the compiled function, in order
4523 to reduce number of taken branches, partitions hot and cold basic blocks
4524 into separate sections of the assembly and .o files, to improve
4525 paging and cache locality performance.
4527 @item -freorder-functions
4528 @opindex freorder-functions
4529 Reorder basic blocks in the compiled function in order to reduce number of
4530 taken branches and improve code locality. This is implemented by using special
4531 subsections @code{.text.hot} for most frequently executed functions and
4532 @code{.text.unlikely} for unlikely executed functions. Reordering is done by
4533 the linker so object file format must support named sections and linker must
4534 place them in a reasonable way.
4536 Also profile feedback must be available in to make this option effective. See
4537 @option{-fprofile-arcs} for details.
4539 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4541 @item -fstrict-aliasing
4542 @opindex fstrict-aliasing
4543 Allows the compiler to assume the strictest aliasing rules applicable to
4544 the language being compiled. For C (and C++), this activates
4545 optimizations based on the type of expressions. In particular, an
4546 object of one type is assumed never to reside at the same address as an
4547 object of a different type, unless the types are almost the same. For
4548 example, an @code{unsigned int} can alias an @code{int}, but not a
4549 @code{void*} or a @code{double}. A character type may alias any other
4552 Pay special attention to code like this:
4565 The practice of reading from a different union member than the one most
4566 recently written to (called ``type-punning'') is common. Even with
4567 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
4568 is accessed through the union type. So, the code above will work as
4569 expected. However, this code might not:
4580 Every language that wishes to perform language-specific alias analysis
4581 should define a function that computes, given an @code{tree}
4582 node, an alias set for the node. Nodes in different alias sets are not
4583 allowed to alias. For an example, see the C front-end function
4584 @code{c_get_alias_set}.
4586 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4588 @item -falign-functions
4589 @itemx -falign-functions=@var{n}
4590 @opindex falign-functions
4591 Align the start of functions to the next power-of-two greater than
4592 @var{n}, skipping up to @var{n} bytes. For instance,
4593 @option{-falign-functions=32} aligns functions to the next 32-byte
4594 boundary, but @option{-falign-functions=24} would align to the next
4595 32-byte boundary only if this can be done by skipping 23 bytes or less.
4597 @option{-fno-align-functions} and @option{-falign-functions=1} are
4598 equivalent and mean that functions will not be aligned.
4600 Some assemblers only support this flag when @var{n} is a power of two;
4601 in that case, it is rounded up.
4603 If @var{n} is not specified or is zero, use a machine-dependent default.
4605 Enabled at levels @option{-O2}, @option{-O3}.
4607 @item -falign-labels
4608 @itemx -falign-labels=@var{n}
4609 @opindex falign-labels
4610 Align all branch targets to a power-of-two boundary, skipping up to
4611 @var{n} bytes like @option{-falign-functions}. This option can easily
4612 make code slower, because it must insert dummy operations for when the
4613 branch target is reached in the usual flow of the code.
4615 @option{-fno-align-labels} and @option{-falign-labels=1} are
4616 equivalent and mean that labels will not be aligned.
4618 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
4619 are greater than this value, then their values are used instead.
4621 If @var{n} is not specified or is zero, use a machine-dependent default
4622 which is very likely to be @samp{1}, meaning no alignment.
4624 Enabled at levels @option{-O2}, @option{-O3}.
4627 @itemx -falign-loops=@var{n}
4628 @opindex falign-loops
4629 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
4630 like @option{-falign-functions}. The hope is that the loop will be
4631 executed many times, which will make up for any execution of the dummy
4634 @option{-fno-align-loops} and @option{-falign-loops=1} are
4635 equivalent and mean that loops will not be aligned.
4637 If @var{n} is not specified or is zero, use a machine-dependent default.
4639 Enabled at levels @option{-O2}, @option{-O3}.
4642 @itemx -falign-jumps=@var{n}
4643 @opindex falign-jumps
4644 Align branch targets to a power-of-two boundary, for branch targets
4645 where the targets can only be reached by jumping, skipping up to @var{n}
4646 bytes like @option{-falign-functions}. In this case, no dummy operations
4649 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
4650 equivalent and mean that loops will not be aligned.
4652 If @var{n} is not specified or is zero, use a machine-dependent default.
4654 Enabled at levels @option{-O2}, @option{-O3}.
4656 @item -funit-at-a-time
4657 @opindex funit-at-a-time
4658 Parse the whole compilation unit before starting to produce code.
4659 This allows some extra optimizations to take place but consumes
4660 more memory (in general). There are some compatibility issues
4661 with @emph{unit-at-at-time} mode:
4664 enabling @emph{unit-at-a-time} mode may change the order
4665 in which functions, variables, and top-level @code{asm} statements
4666 are emitted, and will likely break code relying on some particular
4667 ordering. The majority of such top-level @code{asm} statements,
4668 though, can be replaced by @code{section} attributes.
4671 @emph{unit-at-a-time} mode removes unreferenced static variables
4672 and functions are removed. This may result in undefined references
4673 when an @code{asm} statement refers directly to variables or functions
4674 that are otherwise unused. In that case either the variable/function
4675 shall be listed as an operand of the @code{asm} statement operand or,
4676 in the case of top-level @code{asm} statements the attribute @code{used}
4677 shall be used on the declaration.
4680 Static functions now can use non-standard passing conventions that
4681 may break @code{asm} statements calling functions directly. Again,
4682 attribute @code{used} will prevent this behavior.
4685 As a temporary workaround, @option{-fno-unit-at-a-time} can be used,
4686 but this scheme may not be supported by future releases of GCC.
4688 Enabled at levels @option{-O2}, @option{-O3}.
4692 Constructs webs as commonly used for register allocation purposes and assign
4693 each web individual pseudo register. This allows our register allocation pass
4694 to operate on pseudos directly, but also strengthens several other optimization
4695 passes, such as CSE, loop optimizer and trivial dead code remover. It can,
4696 however, make debugging impossible, since variables will no longer stay in a
4699 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os},
4700 on targets where the default format for debugging information supports
4703 @item -fno-cprop-registers
4704 @opindex fno-cprop-registers
4705 After register allocation and post-register allocation instruction splitting,
4706 we perform a copy-propagation pass to try to reduce scheduling dependencies
4707 and occasionally eliminate the copy.
4709 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4711 @item -fprofile-generate
4712 @opindex fprofile-generate
4714 Enable options usually used for instrumenting application to produce
4715 profile useful for later recompilation with profile feedback based
4716 optimization. You must use @code{-fprofile-generate} both when
4717 compiling and when linking your program.
4719 The following options are enabled: @code{-fprofile-arcs}, @code{-fprofile-values}, @code{-fvpt}.
4722 @opindex fprofile-use
4723 Enable profile feedback directed optimizations, and optimizations
4724 generally profitable only with profile feedback available.
4726 The following options are enabled: @code{-fbranch-probabilities},
4727 @code{-fvpt}, @code{-funroll-loops}, @code{-fpeel-loops}, @code{-ftracer}.
4731 The following options control compiler behavior regarding floating
4732 point arithmetic. These options trade off between speed and
4733 correctness. All must be specifically enabled.
4737 @opindex ffloat-store
4738 Do not store floating point variables in registers, and inhibit other
4739 options that might change whether a floating point value is taken from a
4742 @cindex floating point precision
4743 This option prevents undesirable excess precision on machines such as
4744 the 68000 where the floating registers (of the 68881) keep more
4745 precision than a @code{double} is supposed to have. Similarly for the
4746 x86 architecture. For most programs, the excess precision does only
4747 good, but a few programs rely on the precise definition of IEEE floating
4748 point. Use @option{-ffloat-store} for such programs, after modifying
4749 them to store all pertinent intermediate computations into variables.
4753 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations}, @*
4754 @option{-fno-trapping-math}, @option{-ffinite-math-only},
4755 @option{-fno-rounding-math} and @option{-fno-signaling-nans}.
4757 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
4759 This option should never be turned on by any @option{-O} option since
4760 it can result in incorrect output for programs which depend on
4761 an exact implementation of IEEE or ISO rules/specifications for
4764 @item -fno-math-errno
4765 @opindex fno-math-errno
4766 Do not set ERRNO after calling math functions that are executed
4767 with a single instruction, e.g., sqrt. A program that relies on
4768 IEEE exceptions for math error handling may want to use this flag
4769 for speed while maintaining IEEE arithmetic compatibility.
4771 This option should never be turned on by any @option{-O} option since
4772 it can result in incorrect output for programs which depend on
4773 an exact implementation of IEEE or ISO rules/specifications for
4776 The default is @option{-fmath-errno}.
4778 @item -funsafe-math-optimizations
4779 @opindex funsafe-math-optimizations
4780 Allow optimizations for floating-point arithmetic that (a) assume
4781 that arguments and results are valid and (b) may violate IEEE or
4782 ANSI standards. When used at link-time, it may include libraries
4783 or startup files that change the default FPU control word or other
4784 similar optimizations.
4786 This option should never be turned on by any @option{-O} option since
4787 it can result in incorrect output for programs which depend on
4788 an exact implementation of IEEE or ISO rules/specifications for
4791 The default is @option{-fno-unsafe-math-optimizations}.
4793 @item -ffinite-math-only
4794 @opindex ffinite-math-only
4795 Allow optimizations for floating-point arithmetic that assume
4796 that arguments and results are not NaNs or +-Infs.
4798 This option should never be turned on by any @option{-O} option since
4799 it can result in incorrect output for programs which depend on
4800 an exact implementation of IEEE or ISO rules/specifications.
4802 The default is @option{-fno-finite-math-only}.
4804 @item -fno-trapping-math
4805 @opindex fno-trapping-math
4806 Compile code assuming that floating-point operations cannot generate
4807 user-visible traps. These traps include division by zero, overflow,
4808 underflow, inexact result and invalid operation. This option implies
4809 @option{-fno-signaling-nans}. Setting this option may allow faster
4810 code if one relies on ``non-stop'' IEEE arithmetic, for example.
4812 This option should never be turned on by any @option{-O} option since
4813 it can result in incorrect output for programs which depend on
4814 an exact implementation of IEEE or ISO rules/specifications for
4817 The default is @option{-ftrapping-math}.
4819 @item -frounding-math
4820 @opindex frounding-math
4821 Disable transformations and optimizations that assume default floating
4822 point rounding behavior. This is round-to-zero for all floating point
4823 to integer conversions, and round-to-nearest for all other arithmetic
4824 truncations. This option should be specified for programs that change
4825 the FP rounding mode dynamically, or that may be executed with a
4826 non-default rounding mode. This option disables constant folding of
4827 floating point expressions at compile-time (which may be affected by
4828 rounding mode) and arithmetic transformations that are unsafe in the
4829 presence of sign-dependent rounding modes.
4831 The default is @option{-fno-rounding-math}.
4833 This option is experimental and does not currently guarantee to
4834 disable all GCC optimizations that are affected by rounding mode.
4835 Future versions of GCC may provide finer control of this setting
4836 using C99's @code{FENV_ACCESS} pragma. This command line option
4837 will be used to specify the default state for @code{FENV_ACCESS}.
4839 @item -fsignaling-nans
4840 @opindex fsignaling-nans
4841 Compile code assuming that IEEE signaling NaNs may generate user-visible
4842 traps during floating-point operations. Setting this option disables
4843 optimizations that may change the number of exceptions visible with
4844 signaling NaNs. This option implies @option{-ftrapping-math}.
4846 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
4849 The default is @option{-fno-signaling-nans}.
4851 This option is experimental and does not currently guarantee to
4852 disable all GCC optimizations that affect signaling NaN behavior.
4854 @item -fsingle-precision-constant
4855 @opindex fsingle-precision-constant
4856 Treat floating point constant as single precision constant instead of
4857 implicitly converting it to double precision constant.
4862 The following options control optimizations that may improve
4863 performance, but are not enabled by any @option{-O} options. This
4864 section includes experimental options that may produce broken code.
4867 @item -fbranch-probabilities
4868 @opindex fbranch-probabilities
4869 After running a program compiled with @option{-fprofile-arcs}
4870 (@pxref{Debugging Options,, Options for Debugging Your Program or
4871 @command{gcc}}), you can compile it a second time using
4872 @option{-fbranch-probabilities}, to improve optimizations based on
4873 the number of times each branch was taken. When the program
4874 compiled with @option{-fprofile-arcs} exits it saves arc execution
4875 counts to a file called @file{@var{sourcename}.gcda} for each source
4876 file The information in this data file is very dependent on the
4877 structure of the generated code, so you must use the same source code
4878 and the same optimization options for both compilations.
4880 With @option{-fbranch-probabilities}, GCC puts a
4881 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
4882 These can be used to improve optimization. Currently, they are only
4883 used in one place: in @file{reorg.c}, instead of guessing which path a
4884 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
4885 exactly determine which path is taken more often.
4887 @item -fprofile-values
4888 @opindex fprofile-values
4889 If combined with @option{-fprofile-arcs}, it adds code so that some
4890 data about values of expressions in the program is gathered.
4892 With @option{-fbranch-probabilities}, it reads back the data gathered
4893 from profiling values of expressions and adds @samp{REG_VALUE_PROFILE}
4894 notes to instructions for their later usage in optimizations.
4896 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
4900 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
4901 a code to gather information about values of expressions.
4903 With @option{-fbranch-probabilities}, it reads back the data gathered
4904 and actually performs the optimizations based on them.
4905 Currently the optimizations include specialization of division operation
4906 using the knowledge about the value of the denominator.
4908 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
4910 @item -frename-registers
4911 @opindex frename-registers
4912 Attempt to avoid false dependencies in scheduled code by making use
4913 of registers left over after register allocation. This optimization
4914 will most benefit processors with lots of registers. Depending on the
4915 debug information format adopted by the target, however, it can
4916 make debugging impossible, since variables will no longer stay in
4917 a ``home register''.
4919 Not enabled by default at any level because it has known bugs.
4923 Use a graph coloring register allocator. Currently this option is meant
4924 for testing, so we are interested to hear about miscompilations with
4929 Perform tail duplication to enlarge superblock size. This transformation
4930 simplifies the control flow of the function allowing other optimizations to do
4933 Enabled with @option{-fprofile-use}.
4935 @item -funroll-loops
4936 @opindex funroll-loops
4937 Unroll loops whose number of iterations can be determined at compile time or
4938 upon entry to the loop. @option{-funroll-loops} implies
4939 @option{-frerun-cse-after-loop}. It also turns on complete loop peeling
4940 (i.e. complete removal of loops with small constant number of iterations).
4941 This option makes code larger, and may or may not make it run faster.
4943 Enabled with @option{-fprofile-use}.
4945 @item -funroll-all-loops
4946 @opindex funroll-all-loops
4947 Unroll all loops, even if their number of iterations is uncertain when
4948 the loop is entered. This usually makes programs run more slowly.
4949 @option{-funroll-all-loops} implies the same options as
4950 @option{-funroll-loops}.
4953 @opindex fpeel-loops
4954 Peels the loops for that there is enough information that they do not
4955 roll much (from profile feedback). It also turns on complete loop peeling
4956 (i.e. complete removal of loops with small constant number of iterations).
4958 Enabled with @option{-fprofile-use}.
4960 @item -fmove-loop-invariants
4961 @opindex fmove-loop-invariants
4962 Enables the loop invariant motion pass in the new loop optimizer. Enabled
4963 at level @option{-O1}
4965 @item -funswitch-loops
4966 @opindex funswitch-loops
4967 Move branches with loop invariant conditions out of the loop, with duplicates
4968 of the loop on both branches (modified according to result of the condition).
4970 @item -fold-unroll-loops
4971 @opindex fold-unroll-loops
4972 Unroll loops whose number of iterations can be determined at compile
4973 time or upon entry to the loop, using the old loop unroller whose loop
4974 recognition is based on notes from frontend. @option{-fold-unroll-loops} implies
4975 both @option{-fstrength-reduce} and @option{-frerun-cse-after-loop}. This
4976 option makes code larger, and may or may not make it run faster.
4978 @item -fold-unroll-all-loops
4979 @opindex fold-unroll-all-loops
4980 Unroll all loops, even if their number of iterations is uncertain when
4981 the loop is entered. This is done using the old loop unroller whose loop
4982 recognition is based on notes from frontend. This usually makes programs run more slowly.
4983 @option{-fold-unroll-all-loops} implies the same options as
4984 @option{-fold-unroll-loops}.
4986 @item -fprefetch-loop-arrays
4987 @opindex fprefetch-loop-arrays
4988 If supported by the target machine, generate instructions to prefetch
4989 memory to improve the performance of loops that access large arrays.
4991 Disabled at level @option{-Os}.
4993 @item -ffunction-sections
4994 @itemx -fdata-sections
4995 @opindex ffunction-sections
4996 @opindex fdata-sections
4997 Place each function or data item into its own section in the output
4998 file if the target supports arbitrary sections. The name of the
4999 function or the name of the data item determines the section's name
5002 Use these options on systems where the linker can perform optimizations
5003 to improve locality of reference in the instruction space. Most systems
5004 using the ELF object format and SPARC processors running Solaris 2 have
5005 linkers with such optimizations. AIX may have these optimizations in
5008 Only use these options when there are significant benefits from doing
5009 so. When you specify these options, the assembler and linker will
5010 create larger object and executable files and will also be slower.
5011 You will not be able to use @code{gprof} on all systems if you
5012 specify this option and you may have problems with debugging if
5013 you specify both this option and @option{-g}.
5015 @item -fbranch-target-load-optimize
5016 @opindex fbranch-target-load-optimize
5017 Perform branch target register load optimization before prologue / epilogue
5019 The use of target registers can typically be exposed only during reload,
5020 thus hoisting loads out of loops and doing inter-block scheduling needs
5021 a separate optimization pass.
5023 @item -fbranch-target-load-optimize2
5024 @opindex fbranch-target-load-optimize2
5025 Perform branch target register load optimization after prologue / epilogue
5028 @item -fbtr-bb-exclusive
5029 @opindex fbtr-bb-exclusive
5030 When performing branch target register load optimization, don't reuse
5031 branch target registers in within any basic block.
5033 @item --param @var{name}=@var{value}
5035 In some places, GCC uses various constants to control the amount of
5036 optimization that is done. For example, GCC will not inline functions
5037 that contain more that a certain number of instructions. You can
5038 control some of these constants on the command-line using the
5039 @option{--param} option.
5041 The names of specific parameters, and the meaning of the values, are
5042 tied to the internals of the compiler, and are subject to change
5043 without notice in future releases.
5045 In each case, the @var{value} is an integer. The allowable choices for
5046 @var{name} are given in the following table:
5049 @item max-crossjump-edges
5050 The maximum number of incoming edges to consider for crossjumping.
5051 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
5052 the number of edges incoming to each block. Increasing values mean
5053 more aggressive optimization, making the compile time increase with
5054 probably small improvement in executable size.
5056 @item max-delay-slot-insn-search
5057 The maximum number of instructions to consider when looking for an
5058 instruction to fill a delay slot. If more than this arbitrary number of
5059 instructions is searched, the time savings from filling the delay slot
5060 will be minimal so stop searching. Increasing values mean more
5061 aggressive optimization, making the compile time increase with probably
5062 small improvement in executable run time.
5064 @item max-delay-slot-live-search
5065 When trying to fill delay slots, the maximum number of instructions to
5066 consider when searching for a block with valid live register
5067 information. Increasing this arbitrarily chosen value means more
5068 aggressive optimization, increasing the compile time. This parameter
5069 should be removed when the delay slot code is rewritten to maintain the
5072 @item max-gcse-memory
5073 The approximate maximum amount of memory that will be allocated in
5074 order to perform the global common subexpression elimination
5075 optimization. If more memory than specified is required, the
5076 optimization will not be done.
5078 @item max-gcse-passes
5079 The maximum number of passes of GCSE to run. The default is 1.
5081 @item max-pending-list-length
5082 The maximum number of pending dependencies scheduling will allow
5083 before flushing the current state and starting over. Large functions
5084 with few branches or calls can create excessively large lists which
5085 needlessly consume memory and resources.
5087 @item max-inline-insns-single
5088 Several parameters control the tree inliner used in gcc.
5089 This number sets the maximum number of instructions (counted in GCC's
5090 internal representation) in a single function that the tree inliner
5091 will consider for inlining. This only affects functions declared
5092 inline and methods implemented in a class declaration (C++).
5093 The default value is 500.
5095 @item max-inline-insns-auto
5096 When you use @option{-finline-functions} (included in @option{-O3}),
5097 a lot of functions that would otherwise not be considered for inlining
5098 by the compiler will be investigated. To those functions, a different
5099 (more restrictive) limit compared to functions declared inline can
5101 The default value is 120.
5103 @item large-function-insns
5104 The limit specifying really large functions. For functions greater than this
5105 limit inlining is constrained by @option{--param large-function-growth}.
5106 This parameter is useful primarily to avoid extreme compilation time caused by non-linear
5107 algorithms used by the backend.
5108 This parameter is ignored when @option{-funit-at-a-time} is not used.
5109 The default value is 3000.
5111 @item large-function-growth
5112 Specifies maximal growth of large function caused by inlining in percents.
5113 This parameter is ignored when @option{-funit-at-a-time} is not used.
5114 The default value is 200.
5116 @item inline-unit-growth
5117 Specifies maximal overall growth of the compilation unit caused by inlining.
5118 This parameter is ignored when @option{-funit-at-a-time} is not used.
5119 The default value is 150.
5121 @item max-inline-insns-recursive
5122 @itemx max-inline-insns-recursive-auto
5123 Specifies maximum number of instructions out-of-line copy of self recursive inline
5124 function can grow into by performing recursive inlining.
5126 For functions declared inline @option{--param max-inline-insns-recursive} is
5127 taken into acount. For function not declared inline, recursive inlining
5128 happens only when @option{-finline-functions} (included in @option{-O3}) is
5129 enabled and @option{--param max-inline-insns-recursive-auto} is used. The
5130 default value is 500.
5132 @item max-inline-recursive-depth
5133 @itemx max-inline-recursive-depth-auto
5134 Specifies maximum recursion depth used by the recursive inlining.
5136 For functions declared inline @option{--param max-inline-recursive-depth} is
5137 taken into acount. For function not declared inline, recursive inlining
5138 happens only when @option{-finline-functions} (included in @option{-O3}) is
5139 enabled and @option{--param max-inline-recursive-depth-auto} is used. The
5140 default value is 500.
5142 @item max-inline-insns-rtl
5143 For languages that use the RTL inliner (this happens at a later stage
5144 than tree inlining), you can set the maximum allowable size (counted
5145 in RTL instructions) for the RTL inliner with this parameter.
5146 The default value is 600.
5148 @item max-unrolled-insns
5149 The maximum number of instructions that a loop should have if that loop
5150 is unrolled, and if the loop is unrolled, it determines how many times
5151 the loop code is unrolled.
5153 @item max-average-unrolled-insns
5154 The maximum number of instructions biased by probabilities of their execution
5155 that a loop should have if that loop is unrolled, and if the loop is unrolled,
5156 it determines how many times the loop code is unrolled.
5158 @item max-unroll-times
5159 The maximum number of unrollings of a single loop.
5161 @item max-peeled-insns
5162 The maximum number of instructions that a loop should have if that loop
5163 is peeled, and if the loop is peeled, it determines how many times
5164 the loop code is peeled.
5166 @item max-peel-times
5167 The maximum number of peelings of a single loop.
5169 @item max-completely-peeled-insns
5170 The maximum number of insns of a completely peeled loop.
5172 @item max-completely-peel-times
5173 The maximum number of iterations of a loop to be suitable for complete peeling.
5175 @item max-unswitch-insns
5176 The maximum number of insns of an unswitched loop.
5178 @item max-unswitch-level
5179 The maximum number of branches unswitched in a single loop.
5181 @item hot-bb-count-fraction
5182 Select fraction of the maximal count of repetitions of basic block in program
5183 given basic block needs to have to be considered hot.
5185 @item hot-bb-frequency-fraction
5186 Select fraction of the maximal frequency of executions of basic block in
5187 function given basic block needs to have to be considered hot
5189 @item tracer-dynamic-coverage
5190 @itemx tracer-dynamic-coverage-feedback
5192 This value is used to limit superblock formation once the given percentage of
5193 executed instructions is covered. This limits unnecessary code size
5196 The @option{tracer-dynamic-coverage-feedback} is used only when profile
5197 feedback is available. The real profiles (as opposed to statically estimated
5198 ones) are much less balanced allowing the threshold to be larger value.
5200 @item tracer-max-code-growth
5201 Stop tail duplication once code growth has reached given percentage. This is
5202 rather hokey argument, as most of the duplicates will be eliminated later in
5203 cross jumping, so it may be set to much higher values than is the desired code
5206 @item tracer-min-branch-ratio
5208 Stop reverse growth when the reverse probability of best edge is less than this
5209 threshold (in percent).
5211 @item tracer-min-branch-ratio
5212 @itemx tracer-min-branch-ratio-feedback
5214 Stop forward growth if the best edge do have probability lower than this
5217 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
5218 compilation for profile feedback and one for compilation without. The value
5219 for compilation with profile feedback needs to be more conservative (higher) in
5220 order to make tracer effective.
5222 @item max-cse-path-length
5224 Maximum number of basic blocks on path that cse considers. The default is 10.
5226 @item global-var-threshold
5228 Counts the number of function calls (N) and the number of
5229 call-clobbered variables (V). If NxV is larger than this limit, a
5230 single artificial variable will be created to represent all the
5231 call-clobbered variables at function call sites. This artificial
5232 variable will then be made to alias every call-clobbered variable.
5233 (done as int * size_t on the host machine; beware overflow).
5235 @item max-aliased-vops
5237 Maxiumum number of virtual operands allowed to represent aliases
5238 before triggering the alias grouping heuristic. Alias grouping
5239 reduces compile times and memory consumption needed for aliasing at
5240 the expense of precision loss in alias information.
5242 @item ggc-min-expand
5244 GCC uses a garbage collector to manage its own memory allocation. This
5245 parameter specifies the minimum percentage by which the garbage
5246 collector's heap should be allowed to expand between collections.
5247 Tuning this may improve compilation speed; it has no effect on code
5250 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
5251 RAM >= 1GB. If @code{getrlimit} is available, the notion of "RAM" is
5252 the smallest of actual RAM, RLIMIT_RSS, RLIMIT_DATA and RLIMIT_AS. If
5253 GCC is not able to calculate RAM on a particular platform, the lower
5254 bound of 30% is used. Setting this parameter and
5255 @option{ggc-min-heapsize} to zero causes a full collection to occur at
5256 every opportunity. This is extremely slow, but can be useful for
5259 @item ggc-min-heapsize
5261 Minimum size of the garbage collector's heap before it begins bothering
5262 to collect garbage. The first collection occurs after the heap expands
5263 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
5264 tuning this may improve compilation speed, and has no effect on code
5267 The default is RAM/8, with a lower bound of 4096 (four megabytes) and an
5268 upper bound of 131072 (128 megabytes). If @code{getrlimit} is
5269 available, the notion of "RAM" is the smallest of actual RAM,
5270 RLIMIT_RSS, RLIMIT_DATA and RLIMIT_AS. If GCC is not able to calculate
5271 RAM on a particular platform, the lower bound is used. Setting this
5272 parameter very large effectively disables garbage collection. Setting
5273 this parameter and @option{ggc-min-expand} to zero causes a full
5274 collection to occur at every opportunity.
5276 @item max-reload-search-insns
5277 The maximum number of instruction reload should look backward for equivalent
5278 register. Increasing values mean more aggressive optimization, making the
5279 compile time increase with probably slightly better performance. The default
5282 @item max-cselib-memory-location
5283 The maximum number of memory locations cselib should take into acount.
5284 Increasing values mean more aggressive optimization, making the compile time
5285 increase with probably slightly better performance. The default value is 500.
5287 @item reorder-blocks-duplicate
5288 @itemx reorder-blocks-duplicate-feedback
5290 Used by basic block reordering pass to decide whether to use unconditional
5291 branch or duplicate the code on its destination. Code is duplicated when its
5292 estimated size is smaller than this value multiplied by the estimated size of
5293 unconditional jump in the hot spots of the program.
5295 The @option{reorder-block-duplicate-feedback} is used only when profile
5296 feedback is available and may be set to higher values than
5297 @option{reorder-block-duplicate} since information about the hot spots is more
5300 @item max-sched-region-blocks
5301 The maximum number of blocks in a region to be considered for
5302 interblock scheduling. The default value is 10.
5304 @item max-sched-region-insns
5305 The maximum number of insns in a region to be considered for
5306 interblock scheduling. The default value is 100.
5310 @node Preprocessor Options
5311 @section Options Controlling the Preprocessor
5312 @cindex preprocessor options
5313 @cindex options, preprocessor
5315 These options control the C preprocessor, which is run on each C source
5316 file before actual compilation.
5318 If you use the @option{-E} option, nothing is done except preprocessing.
5319 Some of these options make sense only together with @option{-E} because
5320 they cause the preprocessor output to be unsuitable for actual
5325 You can use @option{-Wp,@var{option}} to bypass the compiler driver
5326 and pass @var{option} directly through to the preprocessor. If
5327 @var{option} contains commas, it is split into multiple options at the
5328 commas. However, many options are modified, translated or interpreted
5329 by the compiler driver before being passed to the preprocessor, and
5330 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
5331 interface is undocumented and subject to change, so whenever possible
5332 you should avoid using @option{-Wp} and let the driver handle the
5335 @item -Xpreprocessor @var{option}
5336 @opindex preprocessor
5337 Pass @var{option} as an option to the preprocessor. You can use this to
5338 supply system-specific preprocessor options which GCC does not know how to
5341 If you want to pass an option that takes an argument, you must use
5342 @option{-Xpreprocessor} twice, once for the option and once for the argument.
5345 @include cppopts.texi
5347 @node Assembler Options
5348 @section Passing Options to the Assembler
5350 @c prevent bad page break with this line
5351 You can pass options to the assembler.
5354 @item -Wa,@var{option}
5356 Pass @var{option} as an option to the assembler. If @var{option}
5357 contains commas, it is split into multiple options at the commas.
5359 @item -Xassembler @var{option}
5361 Pass @var{option} as an option to the assembler. You can use this to
5362 supply system-specific assembler options which GCC does not know how to
5365 If you want to pass an option that takes an argument, you must use
5366 @option{-Xassembler} twice, once for the option and once for the argument.
5371 @section Options for Linking
5372 @cindex link options
5373 @cindex options, linking
5375 These options come into play when the compiler links object files into
5376 an executable output file. They are meaningless if the compiler is
5377 not doing a link step.
5381 @item @var{object-file-name}
5382 A file name that does not end in a special recognized suffix is
5383 considered to name an object file or library. (Object files are
5384 distinguished from libraries by the linker according to the file
5385 contents.) If linking is done, these object files are used as input
5394 If any of these options is used, then the linker is not run, and
5395 object file names should not be used as arguments. @xref{Overall
5399 @item -l@var{library}
5400 @itemx -l @var{library}
5402 Search the library named @var{library} when linking. (The second
5403 alternative with the library as a separate argument is only for
5404 POSIX compliance and is not recommended.)
5406 It makes a difference where in the command you write this option; the
5407 linker searches and processes libraries and object files in the order they
5408 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
5409 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
5410 to functions in @samp{z}, those functions may not be loaded.
5412 The linker searches a standard list of directories for the library,
5413 which is actually a file named @file{lib@var{library}.a}. The linker
5414 then uses this file as if it had been specified precisely by name.
5416 The directories searched include several standard system directories
5417 plus any that you specify with @option{-L}.
5419 Normally the files found this way are library files---archive files
5420 whose members are object files. The linker handles an archive file by
5421 scanning through it for members which define symbols that have so far
5422 been referenced but not defined. But if the file that is found is an
5423 ordinary object file, it is linked in the usual fashion. The only
5424 difference between using an @option{-l} option and specifying a file name
5425 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
5426 and searches several directories.
5430 You need this special case of the @option{-l} option in order to
5431 link an Objective-C program.
5434 @opindex nostartfiles
5435 Do not use the standard system startup files when linking.
5436 The standard system libraries are used normally, unless @option{-nostdlib}
5437 or @option{-nodefaultlibs} is used.
5439 @item -nodefaultlibs
5440 @opindex nodefaultlibs
5441 Do not use the standard system libraries when linking.
5442 Only the libraries you specify will be passed to the linker.
5443 The standard startup files are used normally, unless @option{-nostartfiles}
5444 is used. The compiler may generate calls to memcmp, memset, and memcpy
5445 for System V (and ISO C) environments or to bcopy and bzero for
5446 BSD environments. These entries are usually resolved by entries in
5447 libc. These entry points should be supplied through some other
5448 mechanism when this option is specified.
5452 Do not use the standard system startup files or libraries when linking.
5453 No startup files and only the libraries you specify will be passed to
5454 the linker. The compiler may generate calls to memcmp, memset, and memcpy
5455 for System V (and ISO C) environments or to bcopy and bzero for
5456 BSD environments. These entries are usually resolved by entries in
5457 libc. These entry points should be supplied through some other
5458 mechanism when this option is specified.
5460 @cindex @option{-lgcc}, use with @option{-nostdlib}
5461 @cindex @option{-nostdlib} and unresolved references
5462 @cindex unresolved references and @option{-nostdlib}
5463 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
5464 @cindex @option{-nodefaultlibs} and unresolved references
5465 @cindex unresolved references and @option{-nodefaultlibs}
5466 One of the standard libraries bypassed by @option{-nostdlib} and
5467 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
5468 that GCC uses to overcome shortcomings of particular machines, or special
5469 needs for some languages.
5470 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
5471 Collection (GCC) Internals},
5472 for more discussion of @file{libgcc.a}.)
5473 In most cases, you need @file{libgcc.a} even when you want to avoid
5474 other standard libraries. In other words, when you specify @option{-nostdlib}
5475 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
5476 This ensures that you have no unresolved references to internal GCC
5477 library subroutines. (For example, @samp{__main}, used to ensure C++
5478 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
5479 GNU Compiler Collection (GCC) Internals}.)
5483 Produce a position independent executable on targets which support it.
5484 For predictable results, you must also specify the same set of options
5485 that were used to generate code (@option{-fpie}, @option{-fPIE},
5486 or model suboptions) when you specify this option.
5490 Remove all symbol table and relocation information from the executable.
5494 On systems that support dynamic linking, this prevents linking with the shared
5495 libraries. On other systems, this option has no effect.
5499 Produce a shared object which can then be linked with other objects to
5500 form an executable. Not all systems support this option. For predictable
5501 results, you must also specify the same set of options that were used to
5502 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
5503 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
5504 needs to build supplementary stub code for constructors to work. On
5505 multi-libbed systems, @samp{gcc -shared} must select the correct support
5506 libraries to link against. Failing to supply the correct flags may lead
5507 to subtle defects. Supplying them in cases where they are not necessary
5510 @item -shared-libgcc
5511 @itemx -static-libgcc
5512 @opindex shared-libgcc
5513 @opindex static-libgcc
5514 On systems that provide @file{libgcc} as a shared library, these options
5515 force the use of either the shared or static version respectively.
5516 If no shared version of @file{libgcc} was built when the compiler was
5517 configured, these options have no effect.
5519 There are several situations in which an application should use the
5520 shared @file{libgcc} instead of the static version. The most common
5521 of these is when the application wishes to throw and catch exceptions
5522 across different shared libraries. In that case, each of the libraries
5523 as well as the application itself should use the shared @file{libgcc}.
5525 Therefore, the G++ and GCJ drivers automatically add
5526 @option{-shared-libgcc} whenever you build a shared library or a main
5527 executable, because C++ and Java programs typically use exceptions, so
5528 this is the right thing to do.
5530 If, instead, you use the GCC driver to create shared libraries, you may
5531 find that they will not always be linked with the shared @file{libgcc}.
5532 If GCC finds, at its configuration time, that you have a non-GNU linker
5533 or a GNU linker that does not support option @option{--eh-frame-hdr},
5534 it will link the shared version of @file{libgcc} into shared libraries
5535 by default. Otherwise, it will take advantage of the linker and optimize
5536 away the linking with the shared version of @file{libgcc}, linking with
5537 the static version of libgcc by default. This allows exceptions to
5538 propagate through such shared libraries, without incurring relocation
5539 costs at library load time.
5541 However, if a library or main executable is supposed to throw or catch
5542 exceptions, you must link it using the G++ or GCJ driver, as appropriate
5543 for the languages used in the program, or using the option
5544 @option{-shared-libgcc}, such that it is linked with the shared
5549 Bind references to global symbols when building a shared object. Warn
5550 about any unresolved references (unless overridden by the link editor
5551 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
5554 @item -Xlinker @var{option}
5556 Pass @var{option} as an option to the linker. You can use this to
5557 supply system-specific linker options which GCC does not know how to
5560 If you want to pass an option that takes an argument, you must use
5561 @option{-Xlinker} twice, once for the option and once for the argument.
5562 For example, to pass @option{-assert definitions}, you must write
5563 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
5564 @option{-Xlinker "-assert definitions"}, because this passes the entire
5565 string as a single argument, which is not what the linker expects.
5567 @item -Wl,@var{option}
5569 Pass @var{option} as an option to the linker. If @var{option} contains
5570 commas, it is split into multiple options at the commas.
5572 @item -u @var{symbol}
5574 Pretend the symbol @var{symbol} is undefined, to force linking of
5575 library modules to define it. You can use @option{-u} multiple times with
5576 different symbols to force loading of additional library modules.
5579 @node Directory Options
5580 @section Options for Directory Search
5581 @cindex directory options
5582 @cindex options, directory search
5585 These options specify directories to search for header files, for
5586 libraries and for parts of the compiler:
5591 Add the directory @var{dir} to the head of the list of directories to be
5592 searched for header files. This can be used to override a system header
5593 file, substituting your own version, since these directories are
5594 searched before the system header file directories. However, you should
5595 not use this option to add directories that contain vendor-supplied
5596 system header files (use @option{-isystem} for that). If you use more than
5597 one @option{-I} option, the directories are scanned in left-to-right
5598 order; the standard system directories come after.
5600 If a standard system include directory, or a directory specified with
5601 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
5602 option will be ignored. The directory will still be searched but as a
5603 system directory at its normal position in the system include chain.
5604 This is to ensure that GCC's procedure to fix buggy system headers and
5605 the ordering for the include_next directive are not inadvertently changed.
5606 If you really need to change the search order for system directories,
5607 use the @option{-nostdinc} and/or @option{-isystem} options.
5609 @item -iquote@var{dir}
5611 Add the directory @var{dir} to the head of the list of directories to
5612 be searched for header files only for the case of @samp{#include
5613 "@var{file}"}; they are not searched for @samp{#include <@var{file}>},
5614 otherwise just like @option{-I}.
5618 Add directory @var{dir} to the list of directories to be searched
5621 @item -B@var{prefix}
5623 This option specifies where to find the executables, libraries,
5624 include files, and data files of the compiler itself.
5626 The compiler driver program runs one or more of the subprograms
5627 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
5628 @var{prefix} as a prefix for each program it tries to run, both with and
5629 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
5631 For each subprogram to be run, the compiler driver first tries the
5632 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
5633 was not specified, the driver tries two standard prefixes, which are
5634 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
5635 those results in a file name that is found, the unmodified program
5636 name is searched for using the directories specified in your
5637 @env{PATH} environment variable.
5639 The compiler will check to see if the path provided by the @option{-B}
5640 refers to a directory, and if necessary it will add a directory
5641 separator character at the end of the path.
5643 @option{-B} prefixes that effectively specify directory names also apply
5644 to libraries in the linker, because the compiler translates these
5645 options into @option{-L} options for the linker. They also apply to
5646 includes files in the preprocessor, because the compiler translates these
5647 options into @option{-isystem} options for the preprocessor. In this case,
5648 the compiler appends @samp{include} to the prefix.
5650 The run-time support file @file{libgcc.a} can also be searched for using
5651 the @option{-B} prefix, if needed. If it is not found there, the two
5652 standard prefixes above are tried, and that is all. The file is left
5653 out of the link if it is not found by those means.
5655 Another way to specify a prefix much like the @option{-B} prefix is to use
5656 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
5659 As a special kludge, if the path provided by @option{-B} is
5660 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
5661 9, then it will be replaced by @file{[dir/]include}. This is to help
5662 with boot-strapping the compiler.
5664 @item -specs=@var{file}
5666 Process @var{file} after the compiler reads in the standard @file{specs}
5667 file, in order to override the defaults that the @file{gcc} driver
5668 program uses when determining what switches to pass to @file{cc1},
5669 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
5670 @option{-specs=@var{file}} can be specified on the command line, and they
5671 are processed in order, from left to right.
5675 This option has been deprecated. Please use @option{-iquote} instead for
5676 @option{-I} directories before the @option{-I-} and remove the @option{-I-}.
5677 Any directories you specify with @option{-I} options before the @option{-I-}
5678 option are searched only for the case of @samp{#include "@var{file}"};
5679 they are not searched for @samp{#include <@var{file}>}.
5681 If additional directories are specified with @option{-I} options after
5682 the @option{-I-}, these directories are searched for all @samp{#include}
5683 directives. (Ordinarily @emph{all} @option{-I} directories are used
5686 In addition, the @option{-I-} option inhibits the use of the current
5687 directory (where the current input file came from) as the first search
5688 directory for @samp{#include "@var{file}"}. There is no way to
5689 override this effect of @option{-I-}. With @option{-I.} you can specify
5690 searching the directory which was current when the compiler was
5691 invoked. That is not exactly the same as what the preprocessor does
5692 by default, but it is often satisfactory.
5694 @option{-I-} does not inhibit the use of the standard system directories
5695 for header files. Thus, @option{-I-} and @option{-nostdinc} are
5702 @section Specifying subprocesses and the switches to pass to them
5705 @command{gcc} is a driver program. It performs its job by invoking a
5706 sequence of other programs to do the work of compiling, assembling and
5707 linking. GCC interprets its command-line parameters and uses these to
5708 deduce which programs it should invoke, and which command-line options
5709 it ought to place on their command lines. This behavior is controlled
5710 by @dfn{spec strings}. In most cases there is one spec string for each
5711 program that GCC can invoke, but a few programs have multiple spec
5712 strings to control their behavior. The spec strings built into GCC can
5713 be overridden by using the @option{-specs=} command-line switch to specify
5716 @dfn{Spec files} are plaintext files that are used to construct spec
5717 strings. They consist of a sequence of directives separated by blank
5718 lines. The type of directive is determined by the first non-whitespace
5719 character on the line and it can be one of the following:
5722 @item %@var{command}
5723 Issues a @var{command} to the spec file processor. The commands that can
5727 @item %include <@var{file}>
5729 Search for @var{file} and insert its text at the current point in the
5732 @item %include_noerr <@var{file}>
5733 @cindex %include_noerr
5734 Just like @samp{%include}, but do not generate an error message if the include
5735 file cannot be found.
5737 @item %rename @var{old_name} @var{new_name}
5739 Rename the spec string @var{old_name} to @var{new_name}.
5743 @item *[@var{spec_name}]:
5744 This tells the compiler to create, override or delete the named spec
5745 string. All lines after this directive up to the next directive or
5746 blank line are considered to be the text for the spec string. If this
5747 results in an empty string then the spec will be deleted. (Or, if the
5748 spec did not exist, then nothing will happened.) Otherwise, if the spec
5749 does not currently exist a new spec will be created. If the spec does
5750 exist then its contents will be overridden by the text of this
5751 directive, unless the first character of that text is the @samp{+}
5752 character, in which case the text will be appended to the spec.
5754 @item [@var{suffix}]:
5755 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
5756 and up to the next directive or blank line are considered to make up the
5757 spec string for the indicated suffix. When the compiler encounters an
5758 input file with the named suffix, it will processes the spec string in
5759 order to work out how to compile that file. For example:
5766 This says that any input file whose name ends in @samp{.ZZ} should be
5767 passed to the program @samp{z-compile}, which should be invoked with the
5768 command-line switch @option{-input} and with the result of performing the
5769 @samp{%i} substitution. (See below.)
5771 As an alternative to providing a spec string, the text that follows a
5772 suffix directive can be one of the following:
5775 @item @@@var{language}
5776 This says that the suffix is an alias for a known @var{language}. This is
5777 similar to using the @option{-x} command-line switch to GCC to specify a
5778 language explicitly. For example:
5785 Says that .ZZ files are, in fact, C++ source files.
5788 This causes an error messages saying:
5791 @var{name} compiler not installed on this system.
5795 GCC already has an extensive list of suffixes built into it.
5796 This directive will add an entry to the end of the list of suffixes, but
5797 since the list is searched from the end backwards, it is effectively
5798 possible to override earlier entries using this technique.
5802 GCC has the following spec strings built into it. Spec files can
5803 override these strings or create their own. Note that individual
5804 targets can also add their own spec strings to this list.
5807 asm Options to pass to the assembler
5808 asm_final Options to pass to the assembler post-processor
5809 cpp Options to pass to the C preprocessor
5810 cc1 Options to pass to the C compiler
5811 cc1plus Options to pass to the C++ compiler
5812 endfile Object files to include at the end of the link
5813 link Options to pass to the linker
5814 lib Libraries to include on the command line to the linker
5815 libgcc Decides which GCC support library to pass to the linker
5816 linker Sets the name of the linker
5817 predefines Defines to be passed to the C preprocessor
5818 signed_char Defines to pass to CPP to say whether @code{char} is signed
5820 startfile Object files to include at the start of the link
5823 Here is a small example of a spec file:
5829 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
5832 This example renames the spec called @samp{lib} to @samp{old_lib} and
5833 then overrides the previous definition of @samp{lib} with a new one.
5834 The new definition adds in some extra command-line options before
5835 including the text of the old definition.
5837 @dfn{Spec strings} are a list of command-line options to be passed to their
5838 corresponding program. In addition, the spec strings can contain
5839 @samp{%}-prefixed sequences to substitute variable text or to
5840 conditionally insert text into the command line. Using these constructs
5841 it is possible to generate quite complex command lines.
5843 Here is a table of all defined @samp{%}-sequences for spec
5844 strings. Note that spaces are not generated automatically around the
5845 results of expanding these sequences. Therefore you can concatenate them
5846 together or combine them with constant text in a single argument.
5850 Substitute one @samp{%} into the program name or argument.
5853 Substitute the name of the input file being processed.
5856 Substitute the basename of the input file being processed.
5857 This is the substring up to (and not including) the last period
5858 and not including the directory.
5861 This is the same as @samp{%b}, but include the file suffix (text after
5865 Marks the argument containing or following the @samp{%d} as a
5866 temporary file name, so that that file will be deleted if GCC exits
5867 successfully. Unlike @samp{%g}, this contributes no text to the
5870 @item %g@var{suffix}
5871 Substitute a file name that has suffix @var{suffix} and is chosen
5872 once per compilation, and mark the argument in the same way as
5873 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
5874 name is now chosen in a way that is hard to predict even when previously
5875 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
5876 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
5877 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
5878 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
5879 was simply substituted with a file name chosen once per compilation,
5880 without regard to any appended suffix (which was therefore treated
5881 just like ordinary text), making such attacks more likely to succeed.
5883 @item %u@var{suffix}
5884 Like @samp{%g}, but generates a new temporary file name even if
5885 @samp{%u@var{suffix}} was already seen.
5887 @item %U@var{suffix}
5888 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
5889 new one if there is no such last file name. In the absence of any
5890 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
5891 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
5892 would involve the generation of two distinct file names, one
5893 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
5894 simply substituted with a file name chosen for the previous @samp{%u},
5895 without regard to any appended suffix.
5897 @item %j@var{suffix}
5898 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
5899 writable, and if save-temps is off; otherwise, substitute the name
5900 of a temporary file, just like @samp{%u}. This temporary file is not
5901 meant for communication between processes, but rather as a junk
5904 @item %|@var{suffix}
5905 @itemx %m@var{suffix}
5906 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
5907 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
5908 all. These are the two most common ways to instruct a program that it
5909 should read from standard input or write to standard output. If you
5910 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
5911 construct: see for example @file{f/lang-specs.h}.
5913 @item %.@var{SUFFIX}
5914 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
5915 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
5916 terminated by the next space or %.
5919 Marks the argument containing or following the @samp{%w} as the
5920 designated output file of this compilation. This puts the argument
5921 into the sequence of arguments that @samp{%o} will substitute later.
5924 Substitutes the names of all the output files, with spaces
5925 automatically placed around them. You should write spaces
5926 around the @samp{%o} as well or the results are undefined.
5927 @samp{%o} is for use in the specs for running the linker.
5928 Input files whose names have no recognized suffix are not compiled
5929 at all, but they are included among the output files, so they will
5933 Substitutes the suffix for object files. Note that this is
5934 handled specially when it immediately follows @samp{%g, %u, or %U},
5935 because of the need for those to form complete file names. The
5936 handling is such that @samp{%O} is treated exactly as if it had already
5937 been substituted, except that @samp{%g, %u, and %U} do not currently
5938 support additional @var{suffix} characters following @samp{%O} as they would
5939 following, for example, @samp{.o}.
5942 Substitutes the standard macro predefinitions for the
5943 current target machine. Use this when running @code{cpp}.
5946 Like @samp{%p}, but puts @samp{__} before and after the name of each
5947 predefined macro, except for macros that start with @samp{__} or with
5948 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
5952 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
5953 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}), and
5954 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
5958 Current argument is the name of a library or startup file of some sort.
5959 Search for that file in a standard list of directories and substitute
5960 the full name found.
5963 Print @var{str} as an error message. @var{str} is terminated by a newline.
5964 Use this when inconsistent options are detected.
5967 Substitute the contents of spec string @var{name} at this point.
5970 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
5972 @item %x@{@var{option}@}
5973 Accumulate an option for @samp{%X}.
5976 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
5980 Output the accumulated assembler options specified by @option{-Wa}.
5983 Output the accumulated preprocessor options specified by @option{-Wp}.
5986 Process the @code{asm} spec. This is used to compute the
5987 switches to be passed to the assembler.
5990 Process the @code{asm_final} spec. This is a spec string for
5991 passing switches to an assembler post-processor, if such a program is
5995 Process the @code{link} spec. This is the spec for computing the
5996 command line passed to the linker. Typically it will make use of the
5997 @samp{%L %G %S %D and %E} sequences.
6000 Dump out a @option{-L} option for each directory that GCC believes might
6001 contain startup files. If the target supports multilibs then the
6002 current multilib directory will be prepended to each of these paths.
6005 Output the multilib directory with directory separators replaced with
6006 @samp{_}. If multilib directories are not set, or the multilib directory is
6007 @file{.} then this option emits nothing.
6010 Process the @code{lib} spec. This is a spec string for deciding which
6011 libraries should be included on the command line to the linker.
6014 Process the @code{libgcc} spec. This is a spec string for deciding
6015 which GCC support library should be included on the command line to the linker.
6018 Process the @code{startfile} spec. This is a spec for deciding which
6019 object files should be the first ones passed to the linker. Typically
6020 this might be a file named @file{crt0.o}.
6023 Process the @code{endfile} spec. This is a spec string that specifies
6024 the last object files that will be passed to the linker.
6027 Process the @code{cpp} spec. This is used to construct the arguments
6028 to be passed to the C preprocessor.
6031 Process the @code{cc1} spec. This is used to construct the options to be
6032 passed to the actual C compiler (@samp{cc1}).
6035 Process the @code{cc1plus} spec. This is used to construct the options to be
6036 passed to the actual C++ compiler (@samp{cc1plus}).
6039 Substitute the variable part of a matched option. See below.
6040 Note that each comma in the substituted string is replaced by
6044 Remove all occurrences of @code{-S} from the command line. Note---this
6045 command is position dependent. @samp{%} commands in the spec string
6046 before this one will see @code{-S}, @samp{%} commands in the spec string
6047 after this one will not.
6049 @item %:@var{function}(@var{args})
6050 Call the named function @var{function}, passing it @var{args}.
6051 @var{args} is first processed as a nested spec string, then split
6052 into an argument vector in the usual fashion. The function returns
6053 a string which is processed as if it had appeared literally as part
6054 of the current spec.
6056 The following built-in spec functions are provided:
6059 @item @code{if-exists}
6060 The @code{if-exists} spec function takes one argument, an absolute
6061 pathname to a file. If the file exists, @code{if-exists} returns the
6062 pathname. Here is a small example of its usage:
6066 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
6069 @item @code{if-exists-else}
6070 The @code{if-exists-else} spec function is similar to the @code{if-exists}
6071 spec function, except that it takes two arguments. The first argument is
6072 an absolute pathname to a file. If the file exists, @code{if-exists-else}
6073 returns the pathname. If it does not exist, it returns the second argument.
6074 This way, @code{if-exists-else} can be used to select one file or another,
6075 based on the existence of the first. Here is a small example of its usage:
6079 crt0%O%s %:if-exists(crti%O%s) \
6080 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
6085 Substitutes the @code{-S} switch, if that switch was given to GCC@.
6086 If that switch was not specified, this substitutes nothing. Note that
6087 the leading dash is omitted when specifying this option, and it is
6088 automatically inserted if the substitution is performed. Thus the spec
6089 string @samp{%@{foo@}} would match the command-line option @option{-foo}
6090 and would output the command line option @option{-foo}.
6092 @item %W@{@code{S}@}
6093 Like %@{@code{S}@} but mark last argument supplied within as a file to be
6096 @item %@{@code{S}*@}
6097 Substitutes all the switches specified to GCC whose names start
6098 with @code{-S}, but which also take an argument. This is used for
6099 switches like @option{-o}, @option{-D}, @option{-I}, etc.
6100 GCC considers @option{-o foo} as being
6101 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
6102 text, including the space. Thus two arguments would be generated.
6104 @item %@{@code{S}*&@code{T}*@}
6105 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
6106 (the order of @code{S} and @code{T} in the spec is not significant).
6107 There can be any number of ampersand-separated variables; for each the
6108 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
6110 @item %@{@code{S}:@code{X}@}
6111 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
6113 @item %@{!@code{S}:@code{X}@}
6114 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
6116 @item %@{@code{S}*:@code{X}@}
6117 Substitutes @code{X} if one or more switches whose names start with
6118 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
6119 once, no matter how many such switches appeared. However, if @code{%*}
6120 appears somewhere in @code{X}, then @code{X} will be substituted once
6121 for each matching switch, with the @code{%*} replaced by the part of
6122 that switch that matched the @code{*}.
6124 @item %@{.@code{S}:@code{X}@}
6125 Substitutes @code{X}, if processing a file with suffix @code{S}.
6127 @item %@{!.@code{S}:@code{X}@}
6128 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
6130 @item %@{@code{S}|@code{P}:@code{X}@}
6131 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC@.
6132 This may be combined with @samp{!}, @samp{.}, and @code{*} sequences as well,
6133 although they have a stronger binding than the @samp{|}. If @code{%*}
6134 appears in @code{X}, all of the alternatives must be starred, and only
6135 the first matching alternative is substituted.
6137 For example, a spec string like this:
6140 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
6143 will output the following command-line options from the following input
6144 command-line options:
6149 -d fred.c -foo -baz -boggle
6150 -d jim.d -bar -baz -boggle
6153 @item %@{S:X; T:Y; :D@}
6155 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
6156 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
6157 be as many clauses as you need. This may be combined with @code{.},
6158 @code{!}, @code{|}, and @code{*} as needed.
6163 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
6164 construct may contain other nested @samp{%} constructs or spaces, or
6165 even newlines. They are processed as usual, as described above.
6166 Trailing white space in @code{X} is ignored. White space may also
6167 appear anywhere on the left side of the colon in these constructs,
6168 except between @code{.} or @code{*} and the corresponding word.
6170 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
6171 handled specifically in these constructs. If another value of
6172 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
6173 @option{-W} switch is found later in the command line, the earlier
6174 switch value is ignored, except with @{@code{S}*@} where @code{S} is
6175 just one letter, which passes all matching options.
6177 The character @samp{|} at the beginning of the predicate text is used to
6178 indicate that a command should be piped to the following command, but
6179 only if @option{-pipe} is specified.
6181 It is built into GCC which switches take arguments and which do not.
6182 (You might think it would be useful to generalize this to allow each
6183 compiler's spec to say which switches take arguments. But this cannot
6184 be done in a consistent fashion. GCC cannot even decide which input
6185 files have been specified without knowing which switches take arguments,
6186 and it must know which input files to compile in order to tell which
6189 GCC also knows implicitly that arguments starting in @option{-l} are to be
6190 treated as compiler output files, and passed to the linker in their
6191 proper position among the other output files.
6193 @c man begin OPTIONS
6195 @node Target Options
6196 @section Specifying Target Machine and Compiler Version
6197 @cindex target options
6198 @cindex cross compiling
6199 @cindex specifying machine version
6200 @cindex specifying compiler version and target machine
6201 @cindex compiler version, specifying
6202 @cindex target machine, specifying
6204 The usual way to run GCC is to run the executable called @file{gcc}, or
6205 @file{<machine>-gcc} when cross-compiling, or
6206 @file{<machine>-gcc-<version>} to run a version other than the one that
6207 was installed last. Sometimes this is inconvenient, so GCC provides
6208 options that will switch to another cross-compiler or version.
6211 @item -b @var{machine}
6213 The argument @var{machine} specifies the target machine for compilation.
6215 The value to use for @var{machine} is the same as was specified as the
6216 machine type when configuring GCC as a cross-compiler. For
6217 example, if a cross-compiler was configured with @samp{configure
6218 i386v}, meaning to compile for an 80386 running System V, then you
6219 would specify @option{-b i386v} to run that cross compiler.
6221 @item -V @var{version}
6223 The argument @var{version} specifies which version of GCC to run.
6224 This is useful when multiple versions are installed. For example,
6225 @var{version} might be @samp{2.0}, meaning to run GCC version 2.0.
6228 The @option{-V} and @option{-b} options work by running the
6229 @file{<machine>-gcc-<version>} executable, so there's no real reason to
6230 use them if you can just run that directly.
6232 @node Submodel Options
6233 @section Hardware Models and Configurations
6234 @cindex submodel options
6235 @cindex specifying hardware config
6236 @cindex hardware models and configurations, specifying
6237 @cindex machine dependent options
6239 Earlier we discussed the standard option @option{-b} which chooses among
6240 different installed compilers for completely different target
6241 machines, such as VAX vs.@: 68000 vs.@: 80386.
6243 In addition, each of these target machine types can have its own
6244 special options, starting with @samp{-m}, to choose among various
6245 hardware models or configurations---for example, 68010 vs 68020,
6246 floating coprocessor or none. A single installed version of the
6247 compiler can compile for any model or configuration, according to the
6250 Some configurations of the compiler also support additional special
6251 options, usually for compatibility with other compilers on the same
6254 These options are defined by the macro @code{TARGET_SWITCHES} in the
6255 machine description. The default for the options is also defined by
6256 that macro, which enables you to change the defaults.
6258 @c This list is ordered alphanumerically by subsection name.
6259 @c It should be the same order and spelling as these options are listed
6260 @c in Machine Dependent Options
6268 * DEC Alpha Options::
6269 * DEC Alpha/VMS Options::
6273 * i386 and x86-64 Options::
6285 * RS/6000 and PowerPC Options::
6286 * S/390 and zSeries Options::
6289 * System V Options::
6290 * TMS320C3x/C4x Options::
6294 * Xstormy16 Options::
6300 @subsection ARC Options
6303 These options are defined for ARC implementations:
6308 Compile code for little endian mode. This is the default.
6312 Compile code for big endian mode.
6315 @opindex mmangle-cpu
6316 Prepend the name of the cpu to all public symbol names.
6317 In multiple-processor systems, there are many ARC variants with different
6318 instruction and register set characteristics. This flag prevents code
6319 compiled for one cpu to be linked with code compiled for another.
6320 No facility exists for handling variants that are ``almost identical''.
6321 This is an all or nothing option.
6323 @item -mcpu=@var{cpu}
6325 Compile code for ARC variant @var{cpu}.
6326 Which variants are supported depend on the configuration.
6327 All variants support @option{-mcpu=base}, this is the default.
6329 @item -mtext=@var{text-section}
6330 @itemx -mdata=@var{data-section}
6331 @itemx -mrodata=@var{readonly-data-section}
6335 Put functions, data, and readonly data in @var{text-section},
6336 @var{data-section}, and @var{readonly-data-section} respectively
6337 by default. This can be overridden with the @code{section} attribute.
6338 @xref{Variable Attributes}.
6343 @subsection ARM Options
6346 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
6350 @item -mabi=@var{name}
6352 Generate code for the specified ABI. Permissible values are: @samp{apcs-gnu},
6353 @samp{atpcs}, @samp{aapcs} and @samp{iwmmxt}.
6356 @opindex mapcs-frame
6357 Generate a stack frame that is compliant with the ARM Procedure Call
6358 Standard for all functions, even if this is not strictly necessary for
6359 correct execution of the code. Specifying @option{-fomit-frame-pointer}
6360 with this option will cause the stack frames not to be generated for
6361 leaf functions. The default is @option{-mno-apcs-frame}.
6365 This is a synonym for @option{-mapcs-frame}.
6368 @c not currently implemented
6369 @item -mapcs-stack-check
6370 @opindex mapcs-stack-check
6371 Generate code to check the amount of stack space available upon entry to
6372 every function (that actually uses some stack space). If there is
6373 insufficient space available then either the function
6374 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
6375 called, depending upon the amount of stack space required. The run time
6376 system is required to provide these functions. The default is
6377 @option{-mno-apcs-stack-check}, since this produces smaller code.
6379 @c not currently implemented
6381 @opindex mapcs-float
6382 Pass floating point arguments using the float point registers. This is
6383 one of the variants of the APCS@. This option is recommended if the
6384 target hardware has a floating point unit or if a lot of floating point
6385 arithmetic is going to be performed by the code. The default is
6386 @option{-mno-apcs-float}, since integer only code is slightly increased in
6387 size if @option{-mapcs-float} is used.
6389 @c not currently implemented
6390 @item -mapcs-reentrant
6391 @opindex mapcs-reentrant
6392 Generate reentrant, position independent code. The default is
6393 @option{-mno-apcs-reentrant}.
6396 @item -mthumb-interwork
6397 @opindex mthumb-interwork
6398 Generate code which supports calling between the ARM and Thumb
6399 instruction sets. Without this option the two instruction sets cannot
6400 be reliably used inside one program. The default is
6401 @option{-mno-thumb-interwork}, since slightly larger code is generated
6402 when @option{-mthumb-interwork} is specified.
6404 @item -mno-sched-prolog
6405 @opindex mno-sched-prolog
6406 Prevent the reordering of instructions in the function prolog, or the
6407 merging of those instruction with the instructions in the function's
6408 body. This means that all functions will start with a recognizable set
6409 of instructions (or in fact one of a choice from a small set of
6410 different function prologues), and this information can be used to
6411 locate the start if functions inside an executable piece of code. The
6412 default is @option{-msched-prolog}.
6415 @opindex mhard-float
6416 Generate output containing floating point instructions. This is the
6420 @opindex msoft-float
6421 Generate output containing library calls for floating point.
6422 @strong{Warning:} the requisite libraries are not available for all ARM
6423 targets. Normally the facilities of the machine's usual C compiler are
6424 used, but this cannot be done directly in cross-compilation. You must make
6425 your own arrangements to provide suitable library functions for
6428 @option{-msoft-float} changes the calling convention in the output file;
6429 therefore, it is only useful if you compile @emph{all} of a program with
6430 this option. In particular, you need to compile @file{libgcc.a}, the
6431 library that comes with GCC, with @option{-msoft-float} in order for
6434 @item -mfloat-abi=@var{name}
6436 Specifies which ABI to use for floating point values. Permissible values
6437 are: @samp{soft}, @samp{softfp} and @samp{hard}.
6439 @samp{soft} and @samp{hard} are equivalent to @option{-msoft-float}
6440 and @option{-mhard-float} respectively. @samp{softfp} allows the generation
6441 of floating point instructions, but still uses the soft-float calling
6444 @item -mlittle-endian
6445 @opindex mlittle-endian
6446 Generate code for a processor running in little-endian mode. This is
6447 the default for all standard configurations.
6450 @opindex mbig-endian
6451 Generate code for a processor running in big-endian mode; the default is
6452 to compile code for a little-endian processor.
6454 @item -mwords-little-endian
6455 @opindex mwords-little-endian
6456 This option only applies when generating code for big-endian processors.
6457 Generate code for a little-endian word order but a big-endian byte
6458 order. That is, a byte order of the form @samp{32107654}. Note: this
6459 option should only be used if you require compatibility with code for
6460 big-endian ARM processors generated by versions of the compiler prior to
6463 @item -mcpu=@var{name}
6465 This specifies the name of the target ARM processor. GCC uses this name
6466 to determine what kind of instructions it can emit when generating
6467 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
6468 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
6469 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
6470 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
6471 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
6472 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm8},
6473 @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
6474 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
6475 @samp{arm920t}, @samp{arm926ejs}, @samp{arm940t}, @samp{arm9tdmi},
6476 @samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ejs},
6477 @samp{arm1136js}, @samp{arm1136jfs} ,@samp{xscale}, @samp{iwmmxt},
6480 @itemx -mtune=@var{name}
6482 This option is very similar to the @option{-mcpu=} option, except that
6483 instead of specifying the actual target processor type, and hence
6484 restricting which instructions can be used, it specifies that GCC should
6485 tune the performance of the code as if the target were of the type
6486 specified in this option, but still choosing the instructions that it
6487 will generate based on the cpu specified by a @option{-mcpu=} option.
6488 For some ARM implementations better performance can be obtained by using
6491 @item -march=@var{name}
6493 This specifies the name of the target ARM architecture. GCC uses this
6494 name to determine what kind of instructions it can emit when generating
6495 assembly code. This option can be used in conjunction with or instead
6496 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
6497 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
6498 @samp{armv5}, @samp{armv5t}, @samp{armv5te}, @samp{armv6}, @samp{armv6j},
6499 @samp{iwmmxt}, @samp{ep9312}.
6501 @item -mfpu=@var{name}
6502 @itemx -mfpe=@var{number}
6503 @itemx -mfp=@var{number}
6507 This specifies what floating point hardware (or hardware emulation) is
6508 available on the target. Permissible names are: @samp{fpa}, @samp{fpe2},
6509 @samp{fpe3}, @samp{maverick}, @samp{vfp}. @option{-mfp} and @option{-mfpe}
6510 are synonyms for @option{-mfpu}=@samp{fpe}@var{number}, for compatibility
6511 with older versions of GCC@.
6513 If @option{-msoft-float} is specified this specifies the format of
6514 floating point values.
6516 @item -mstructure-size-boundary=@var{n}
6517 @opindex mstructure-size-boundary
6518 The size of all structures and unions will be rounded up to a multiple
6519 of the number of bits set by this option. Permissible values are 8, 32
6520 and 64. The default value varies for different toolchains. For the COFF
6521 targeted toolchain the default value is 8. A value of 64 is only allowed
6522 if the underlying ABI supports it.
6524 Specifying the larger number can produce faster, more efficient code, but
6525 can also increase the size of the program. Different values are potentially
6526 incompatible. Code compiled with one value cannot necessarily expect to
6527 work with code or libraries compiled with another value, if they exchange
6528 information using structures or unions.
6530 @item -mabort-on-noreturn
6531 @opindex mabort-on-noreturn
6532 Generate a call to the function @code{abort} at the end of a
6533 @code{noreturn} function. It will be executed if the function tries to
6537 @itemx -mno-long-calls
6538 @opindex mlong-calls
6539 @opindex mno-long-calls
6540 Tells the compiler to perform function calls by first loading the
6541 address of the function into a register and then performing a subroutine
6542 call on this register. This switch is needed if the target function
6543 will lie outside of the 64 megabyte addressing range of the offset based
6544 version of subroutine call instruction.
6546 Even if this switch is enabled, not all function calls will be turned
6547 into long calls. The heuristic is that static functions, functions
6548 which have the @samp{short-call} attribute, functions that are inside
6549 the scope of a @samp{#pragma no_long_calls} directive and functions whose
6550 definitions have already been compiled within the current compilation
6551 unit, will not be turned into long calls. The exception to this rule is
6552 that weak function definitions, functions with the @samp{long-call}
6553 attribute or the @samp{section} attribute, and functions that are within
6554 the scope of a @samp{#pragma long_calls} directive, will always be
6555 turned into long calls.
6557 This feature is not enabled by default. Specifying
6558 @option{-mno-long-calls} will restore the default behavior, as will
6559 placing the function calls within the scope of a @samp{#pragma
6560 long_calls_off} directive. Note these switches have no effect on how
6561 the compiler generates code to handle function calls via function
6564 @item -mnop-fun-dllimport
6565 @opindex mnop-fun-dllimport
6566 Disable support for the @code{dllimport} attribute.
6568 @item -msingle-pic-base
6569 @opindex msingle-pic-base
6570 Treat the register used for PIC addressing as read-only, rather than
6571 loading it in the prologue for each function. The run-time system is
6572 responsible for initializing this register with an appropriate value
6573 before execution begins.
6575 @item -mpic-register=@var{reg}
6576 @opindex mpic-register
6577 Specify the register to be used for PIC addressing. The default is R10
6578 unless stack-checking is enabled, when R9 is used.
6580 @item -mcirrus-fix-invalid-insns
6581 @opindex mcirrus-fix-invalid-insns
6582 @opindex mno-cirrus-fix-invalid-insns
6583 Insert NOPs into the instruction stream to in order to work around
6584 problems with invalid Maverick instruction combinations. This option
6585 is only valid if the @option{-mcpu=ep9312} option has been used to
6586 enable generation of instructions for the Cirrus Maverick floating
6587 point co-processor. This option is not enabled by default, since the
6588 problem is only present in older Maverick implementations. The default
6589 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
6592 @item -mpoke-function-name
6593 @opindex mpoke-function-name
6594 Write the name of each function into the text section, directly
6595 preceding the function prologue. The generated code is similar to this:
6599 .ascii "arm_poke_function_name", 0
6602 .word 0xff000000 + (t1 - t0)
6603 arm_poke_function_name
6605 stmfd sp!, @{fp, ip, lr, pc@}
6609 When performing a stack backtrace, code can inspect the value of
6610 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
6611 location @code{pc - 12} and the top 8 bits are set, then we know that
6612 there is a function name embedded immediately preceding this location
6613 and has length @code{((pc[-3]) & 0xff000000)}.
6617 Generate code for the 16-bit Thumb instruction set. The default is to
6618 use the 32-bit ARM instruction set.
6621 @opindex mtpcs-frame
6622 Generate a stack frame that is compliant with the Thumb Procedure Call
6623 Standard for all non-leaf functions. (A leaf function is one that does
6624 not call any other functions.) The default is @option{-mno-tpcs-frame}.
6626 @item -mtpcs-leaf-frame
6627 @opindex mtpcs-leaf-frame
6628 Generate a stack frame that is compliant with the Thumb Procedure Call
6629 Standard for all leaf functions. (A leaf function is one that does
6630 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
6632 @item -mcallee-super-interworking
6633 @opindex mcallee-super-interworking
6634 Gives all externally visible functions in the file being compiled an ARM
6635 instruction set header which switches to Thumb mode before executing the
6636 rest of the function. This allows these functions to be called from
6637 non-interworking code.
6639 @item -mcaller-super-interworking
6640 @opindex mcaller-super-interworking
6641 Allows calls via function pointers (including virtual functions) to
6642 execute correctly regardless of whether the target code has been
6643 compiled for interworking or not. There is a small overhead in the cost
6644 of executing a function pointer if this option is enabled.
6649 @subsection AVR Options
6652 These options are defined for AVR implementations:
6655 @item -mmcu=@var{mcu}
6657 Specify ATMEL AVR instruction set or MCU type.
6659 Instruction set avr1 is for the minimal AVR core, not supported by the C
6660 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
6661 attiny11, attiny12, attiny15, attiny28).
6663 Instruction set avr2 (default) is for the classic AVR core with up to
6664 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
6665 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
6666 at90c8534, at90s8535).
6668 Instruction set avr3 is for the classic AVR core with up to 128K program
6669 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
6671 Instruction set avr4 is for the enhanced AVR core with up to 8K program
6672 memory space (MCU types: atmega8, atmega83, atmega85).
6674 Instruction set avr5 is for the enhanced AVR core with up to 128K program
6675 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
6676 atmega64, atmega128, at43usb355, at94k).
6680 Output instruction sizes to the asm file.
6682 @item -minit-stack=@var{N}
6683 @opindex minit-stack
6684 Specify the initial stack address, which may be a symbol or numeric value,
6685 @samp{__stack} is the default.
6687 @item -mno-interrupts
6688 @opindex mno-interrupts
6689 Generated code is not compatible with hardware interrupts.
6690 Code size will be smaller.
6692 @item -mcall-prologues
6693 @opindex mcall-prologues
6694 Functions prologues/epilogues expanded as call to appropriate
6695 subroutines. Code size will be smaller.
6697 @item -mno-tablejump
6698 @opindex mno-tablejump
6699 Do not generate tablejump insns which sometimes increase code size.
6702 @opindex mtiny-stack
6703 Change only the low 8 bits of the stack pointer.
6707 Assume int to be 8 bit integer. This affects the sizes of all types: A
6708 char will be 1 byte, an int will be 1 byte, an long will be 2 bytes
6709 and long long will be 4 bytes. Please note that this option does not
6710 comply to the C standards, but it will provide you with smaller code
6715 @subsection CRIS Options
6716 @cindex CRIS Options
6718 These options are defined specifically for the CRIS ports.
6721 @item -march=@var{architecture-type}
6722 @itemx -mcpu=@var{architecture-type}
6725 Generate code for the specified architecture. The choices for
6726 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
6727 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX.
6728 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
6731 @item -mtune=@var{architecture-type}
6733 Tune to @var{architecture-type} everything applicable about the generated
6734 code, except for the ABI and the set of available instructions. The
6735 choices for @var{architecture-type} are the same as for
6736 @option{-march=@var{architecture-type}}.
6738 @item -mmax-stack-frame=@var{n}
6739 @opindex mmax-stack-frame
6740 Warn when the stack frame of a function exceeds @var{n} bytes.
6742 @item -melinux-stacksize=@var{n}
6743 @opindex melinux-stacksize
6744 Only available with the @samp{cris-axis-aout} target. Arranges for
6745 indications in the program to the kernel loader that the stack of the
6746 program should be set to @var{n} bytes.
6752 The options @option{-metrax4} and @option{-metrax100} are synonyms for
6753 @option{-march=v3} and @option{-march=v8} respectively.
6755 @item -mmul-bug-workaround
6756 @itemx -mno-mul-bug-workaround
6757 @opindex mmul-bug-workaround
6758 @opindex mno-mul-bug-workaround
6759 Work around a bug in the @code{muls} and @code{mulu} instructions for CPU
6760 models where it applies. This option is active by default.
6764 Enable CRIS-specific verbose debug-related information in the assembly
6765 code. This option also has the effect to turn off the @samp{#NO_APP}
6766 formatted-code indicator to the assembler at the beginning of the
6771 Do not use condition-code results from previous instruction; always emit
6772 compare and test instructions before use of condition codes.
6774 @item -mno-side-effects
6775 @opindex mno-side-effects
6776 Do not emit instructions with side-effects in addressing modes other than
6780 @itemx -mno-stack-align
6782 @itemx -mno-data-align
6783 @itemx -mconst-align
6784 @itemx -mno-const-align
6785 @opindex mstack-align
6786 @opindex mno-stack-align
6787 @opindex mdata-align
6788 @opindex mno-data-align
6789 @opindex mconst-align
6790 @opindex mno-const-align
6791 These options (no-options) arranges (eliminate arrangements) for the
6792 stack-frame, individual data and constants to be aligned for the maximum
6793 single data access size for the chosen CPU model. The default is to
6794 arrange for 32-bit alignment. ABI details such as structure layout are
6795 not affected by these options.
6803 Similar to the stack- data- and const-align options above, these options
6804 arrange for stack-frame, writable data and constants to all be 32-bit,
6805 16-bit or 8-bit aligned. The default is 32-bit alignment.
6807 @item -mno-prologue-epilogue
6808 @itemx -mprologue-epilogue
6809 @opindex mno-prologue-epilogue
6810 @opindex mprologue-epilogue
6811 With @option{-mno-prologue-epilogue}, the normal function prologue and
6812 epilogue that sets up the stack-frame are omitted and no return
6813 instructions or return sequences are generated in the code. Use this
6814 option only together with visual inspection of the compiled code: no
6815 warnings or errors are generated when call-saved registers must be saved,
6816 or storage for local variable needs to be allocated.
6822 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
6823 instruction sequences that load addresses for functions from the PLT part
6824 of the GOT rather than (traditional on other architectures) calls to the
6825 PLT. The default is @option{-mgotplt}.
6829 Legacy no-op option only recognized with the cris-axis-aout target.
6833 Legacy no-op option only recognized with the cris-axis-elf and
6834 cris-axis-linux-gnu targets.
6838 Only recognized with the cris-axis-aout target, where it selects a
6839 GNU/linux-like multilib, include files and instruction set for
6844 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
6848 This option, recognized for the cris-axis-aout and cris-axis-elf arranges
6849 to link with input-output functions from a simulator library. Code,
6850 initialized data and zero-initialized data are allocated consecutively.
6854 Like @option{-sim}, but pass linker options to locate initialized data at
6855 0x40000000 and zero-initialized data at 0x80000000.
6858 @node Darwin Options
6859 @subsection Darwin Options
6860 @cindex Darwin options
6862 These options are defined for all architectures running the Darwin operating
6863 system. They are useful for compatibility with other Mac OS compilers.
6868 Add the framework directory @var{dir} to the head of the list of
6869 directories to be searched for header files. These directories are
6870 interleaved with those specified by @option{-I} options and are
6871 scanned in a left-to-right order.
6873 A framework directory is a directory with frameworks in it. A
6874 framework is a directory with a @samp{"Headers"} and/or
6875 @samp{"PrivateHeaders"} directory contained directly in it that ends
6876 in @samp{".framework"}. The name of a framework is the name of this
6877 directory excluding the @samp{".framework"}. Headers associated with
6878 the framework are found in one of those two directories, with
6879 @samp{"Headers"} being searched first. A subframework is a framework
6880 directory that is in a framework's @samp{"Frameworks"} directory.
6881 Includes of subframework headers can only appear in a header of a
6882 framework that contains the subframework, or in a sibling subframework
6883 header. Two subframeworks are siblings if they occur in the same
6884 framework. A subframework should not have the same name as a
6885 framework, a warning will be issued if this is violated. Currently a
6886 subframework cannot have subframeworks, in the future, the mechanism
6887 may be extended to support this. The standard frameworks can be found
6888 in @samp{"/System/Library/Frameworks"}, @samp{"/Library/Frameworks"}
6889 and @samp{"/Local/Library/Frameworks"}. An example include looks like
6890 @code{#include <Framework/header.h>}, where @samp{Framework} denotes
6891 the name of the framework and header.h is found in the
6892 @samp{"PrivateHeaders"} or @samp{"Headers"} directory.
6896 Loads all members of static archive libraries.
6897 See man ld(1) for more information.
6899 @item -arch_errors_fatal
6900 @opindex arch_errors_fatal
6901 Cause the errors having to do with files that have the wrong architecture
6905 @opindex bind_at_load
6906 Causes the output file to be marked such that the dynamic linker will
6907 bind all undefined references when the file is loaded or launched.
6911 Produce a Mach-o bundle format file.
6912 See man ld(1) for more information.
6914 @item -bundle_loader @var{executable}
6915 @opindex bundle_loader
6916 This specifies the @var{executable} that will be loading the build
6917 output file being linked. See man ld(1) for more information.
6919 @item -allowable_client @var{client_name}
6923 @itemx -compatibility_version
6924 @itemx -current_version
6925 @itemx -dependency-file
6927 @itemx -dylinker_install_name
6930 @itemx -exported_symbols_list
6932 @itemx -flat_namespace
6933 @itemx -force_cpusubtype_ALL
6934 @itemx -force_flat_namespace
6935 @itemx -headerpad_max_install_names
6938 @itemx -install_name
6939 @itemx -keep_private_externs
6940 @itemx -multi_module
6941 @itemx -multiply_defined
6942 @itemx -multiply_defined_unused
6944 @itemx -nofixprebinding
6947 @itemx -noseglinkedit
6948 @itemx -pagezero_size
6950 @itemx -prebind_all_twolevel_modules
6951 @itemx -private_bundle
6952 @itemx -read_only_relocs
6954 @itemx -sectobjectsymbols
6958 @itemx -sectobjectsymbols
6960 @itemx -seg_addr_table
6961 @itemx -seg_addr_table_filename
6964 @itemx -segs_read_only_addr
6965 @itemx -segs_read_write_addr
6966 @itemx -single_module
6969 @itemx -sub_umbrella
6970 @itemx -twolevel_namespace
6973 @itemx -unexported_symbols_list
6974 @itemx -weak_reference_mismatches
6977 @opindex allowable_client
6979 @opindex client_name
6980 @opindex compatibility_version
6981 @opindex current_version
6982 @opindex dependency-file
6984 @opindex dylinker_install_name
6987 @opindex exported_symbols_list
6989 @opindex flat_namespace
6990 @opindex force_cpusubtype_ALL
6991 @opindex force_flat_namespace
6992 @opindex headerpad_max_install_names
6995 @opindex install_name
6996 @opindex keep_private_externs
6997 @opindex multi_module
6998 @opindex multiply_defined
6999 @opindex multiply_defined_unused
7001 @opindex nofixprebinding
7002 @opindex nomultidefs
7004 @opindex noseglinkedit
7005 @opindex pagezero_size
7007 @opindex prebind_all_twolevel_modules
7008 @opindex private_bundle
7009 @opindex read_only_relocs
7011 @opindex sectobjectsymbols
7015 @opindex sectobjectsymbols
7017 @opindex seg_addr_table
7018 @opindex seg_addr_table_filename
7019 @opindex seglinkedit
7021 @opindex segs_read_only_addr
7022 @opindex segs_read_write_addr
7023 @opindex single_module
7025 @opindex sub_library
7026 @opindex sub_umbrella
7027 @opindex twolevel_namespace
7030 @opindex unexported_symbols_list
7031 @opindex weak_reference_mismatches
7032 @opindex whatsloaded
7034 These options are available for Darwin linker. Darwin linker man page
7035 describes them in detail.
7038 @node DEC Alpha Options
7039 @subsection DEC Alpha Options
7041 These @samp{-m} options are defined for the DEC Alpha implementations:
7044 @item -mno-soft-float
7046 @opindex mno-soft-float
7047 @opindex msoft-float
7048 Use (do not use) the hardware floating-point instructions for
7049 floating-point operations. When @option{-msoft-float} is specified,
7050 functions in @file{libgcc.a} will be used to perform floating-point
7051 operations. Unless they are replaced by routines that emulate the
7052 floating-point operations, or compiled in such a way as to call such
7053 emulations routines, these routines will issue floating-point
7054 operations. If you are compiling for an Alpha without floating-point
7055 operations, you must ensure that the library is built so as not to call
7058 Note that Alpha implementations without floating-point operations are
7059 required to have floating-point registers.
7064 @opindex mno-fp-regs
7065 Generate code that uses (does not use) the floating-point register set.
7066 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
7067 register set is not used, floating point operands are passed in integer
7068 registers as if they were integers and floating-point results are passed
7069 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
7070 so any function with a floating-point argument or return value called by code
7071 compiled with @option{-mno-fp-regs} must also be compiled with that
7074 A typical use of this option is building a kernel that does not use,
7075 and hence need not save and restore, any floating-point registers.
7079 The Alpha architecture implements floating-point hardware optimized for
7080 maximum performance. It is mostly compliant with the IEEE floating
7081 point standard. However, for full compliance, software assistance is
7082 required. This option generates code fully IEEE compliant code
7083 @emph{except} that the @var{inexact-flag} is not maintained (see below).
7084 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
7085 defined during compilation. The resulting code is less efficient but is
7086 able to correctly support denormalized numbers and exceptional IEEE
7087 values such as not-a-number and plus/minus infinity. Other Alpha
7088 compilers call this option @option{-ieee_with_no_inexact}.
7090 @item -mieee-with-inexact
7091 @opindex mieee-with-inexact
7092 This is like @option{-mieee} except the generated code also maintains
7093 the IEEE @var{inexact-flag}. Turning on this option causes the
7094 generated code to implement fully-compliant IEEE math. In addition to
7095 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
7096 macro. On some Alpha implementations the resulting code may execute
7097 significantly slower than the code generated by default. Since there is
7098 very little code that depends on the @var{inexact-flag}, you should
7099 normally not specify this option. Other Alpha compilers call this
7100 option @option{-ieee_with_inexact}.
7102 @item -mfp-trap-mode=@var{trap-mode}
7103 @opindex mfp-trap-mode
7104 This option controls what floating-point related traps are enabled.
7105 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
7106 The trap mode can be set to one of four values:
7110 This is the default (normal) setting. The only traps that are enabled
7111 are the ones that cannot be disabled in software (e.g., division by zero
7115 In addition to the traps enabled by @samp{n}, underflow traps are enabled
7119 Like @samp{su}, but the instructions are marked to be safe for software
7120 completion (see Alpha architecture manual for details).
7123 Like @samp{su}, but inexact traps are enabled as well.
7126 @item -mfp-rounding-mode=@var{rounding-mode}
7127 @opindex mfp-rounding-mode
7128 Selects the IEEE rounding mode. Other Alpha compilers call this option
7129 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
7134 Normal IEEE rounding mode. Floating point numbers are rounded towards
7135 the nearest machine number or towards the even machine number in case
7139 Round towards minus infinity.
7142 Chopped rounding mode. Floating point numbers are rounded towards zero.
7145 Dynamic rounding mode. A field in the floating point control register
7146 (@var{fpcr}, see Alpha architecture reference manual) controls the
7147 rounding mode in effect. The C library initializes this register for
7148 rounding towards plus infinity. Thus, unless your program modifies the
7149 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
7152 @item -mtrap-precision=@var{trap-precision}
7153 @opindex mtrap-precision
7154 In the Alpha architecture, floating point traps are imprecise. This
7155 means without software assistance it is impossible to recover from a
7156 floating trap and program execution normally needs to be terminated.
7157 GCC can generate code that can assist operating system trap handlers
7158 in determining the exact location that caused a floating point trap.
7159 Depending on the requirements of an application, different levels of
7160 precisions can be selected:
7164 Program precision. This option is the default and means a trap handler
7165 can only identify which program caused a floating point exception.
7168 Function precision. The trap handler can determine the function that
7169 caused a floating point exception.
7172 Instruction precision. The trap handler can determine the exact
7173 instruction that caused a floating point exception.
7176 Other Alpha compilers provide the equivalent options called
7177 @option{-scope_safe} and @option{-resumption_safe}.
7179 @item -mieee-conformant
7180 @opindex mieee-conformant
7181 This option marks the generated code as IEEE conformant. You must not
7182 use this option unless you also specify @option{-mtrap-precision=i} and either
7183 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
7184 is to emit the line @samp{.eflag 48} in the function prologue of the
7185 generated assembly file. Under DEC Unix, this has the effect that
7186 IEEE-conformant math library routines will be linked in.
7188 @item -mbuild-constants
7189 @opindex mbuild-constants
7190 Normally GCC examines a 32- or 64-bit integer constant to
7191 see if it can construct it from smaller constants in two or three
7192 instructions. If it cannot, it will output the constant as a literal and
7193 generate code to load it from the data segment at runtime.
7195 Use this option to require GCC to construct @emph{all} integer constants
7196 using code, even if it takes more instructions (the maximum is six).
7198 You would typically use this option to build a shared library dynamic
7199 loader. Itself a shared library, it must relocate itself in memory
7200 before it can find the variables and constants in its own data segment.
7206 Select whether to generate code to be assembled by the vendor-supplied
7207 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
7225 Indicate whether GCC should generate code to use the optional BWX,
7226 CIX, FIX and MAX instruction sets. The default is to use the instruction
7227 sets supported by the CPU type specified via @option{-mcpu=} option or that
7228 of the CPU on which GCC was built if none was specified.
7233 @opindex mfloat-ieee
7234 Generate code that uses (does not use) VAX F and G floating point
7235 arithmetic instead of IEEE single and double precision.
7237 @item -mexplicit-relocs
7238 @itemx -mno-explicit-relocs
7239 @opindex mexplicit-relocs
7240 @opindex mno-explicit-relocs
7241 Older Alpha assemblers provided no way to generate symbol relocations
7242 except via assembler macros. Use of these macros does not allow
7243 optimal instruction scheduling. GNU binutils as of version 2.12
7244 supports a new syntax that allows the compiler to explicitly mark
7245 which relocations should apply to which instructions. This option
7246 is mostly useful for debugging, as GCC detects the capabilities of
7247 the assembler when it is built and sets the default accordingly.
7251 @opindex msmall-data
7252 @opindex mlarge-data
7253 When @option{-mexplicit-relocs} is in effect, static data is
7254 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
7255 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
7256 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
7257 16-bit relocations off of the @code{$gp} register. This limits the
7258 size of the small data area to 64KB, but allows the variables to be
7259 directly accessed via a single instruction.
7261 The default is @option{-mlarge-data}. With this option the data area
7262 is limited to just below 2GB. Programs that require more than 2GB of
7263 data must use @code{malloc} or @code{mmap} to allocate the data in the
7264 heap instead of in the program's data segment.
7266 When generating code for shared libraries, @option{-fpic} implies
7267 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
7271 @opindex msmall-text
7272 @opindex mlarge-text
7273 When @option{-msmall-text} is used, the compiler assumes that the
7274 code of the entire program (or shared library) fits in 4MB, and is
7275 thus reachable with a branch instruction. When @option{-msmall-data}
7276 is used, the compiler can assume that all local symbols share the
7277 same @code{$gp} value, and thus reduce the number of instructions
7278 required for a function call from 4 to 1.
7280 The default is @option{-mlarge-text}.
7282 @item -mcpu=@var{cpu_type}
7284 Set the instruction set and instruction scheduling parameters for
7285 machine type @var{cpu_type}. You can specify either the @samp{EV}
7286 style name or the corresponding chip number. GCC supports scheduling
7287 parameters for the EV4, EV5 and EV6 family of processors and will
7288 choose the default values for the instruction set from the processor
7289 you specify. If you do not specify a processor type, GCC will default
7290 to the processor on which the compiler was built.
7292 Supported values for @var{cpu_type} are
7298 Schedules as an EV4 and has no instruction set extensions.
7302 Schedules as an EV5 and has no instruction set extensions.
7306 Schedules as an EV5 and supports the BWX extension.
7311 Schedules as an EV5 and supports the BWX and MAX extensions.
7315 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
7319 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
7322 @item -mtune=@var{cpu_type}
7324 Set only the instruction scheduling parameters for machine type
7325 @var{cpu_type}. The instruction set is not changed.
7327 @item -mmemory-latency=@var{time}
7328 @opindex mmemory-latency
7329 Sets the latency the scheduler should assume for typical memory
7330 references as seen by the application. This number is highly
7331 dependent on the memory access patterns used by the application
7332 and the size of the external cache on the machine.
7334 Valid options for @var{time} are
7338 A decimal number representing clock cycles.
7344 The compiler contains estimates of the number of clock cycles for
7345 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
7346 (also called Dcache, Scache, and Bcache), as well as to main memory.
7347 Note that L3 is only valid for EV5.
7352 @node DEC Alpha/VMS Options
7353 @subsection DEC Alpha/VMS Options
7355 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
7358 @item -mvms-return-codes
7359 @opindex mvms-return-codes
7360 Return VMS condition codes from main. The default is to return POSIX
7361 style condition (e.g.@ error) codes.
7365 @subsection FRV Options
7372 Only use the first 32 general purpose registers.
7377 Use all 64 general purpose registers.
7382 Use only the first 32 floating point registers.
7387 Use all 64 floating point registers
7390 @opindex mhard-float
7392 Use hardware instructions for floating point operations.
7395 @opindex msoft-float
7397 Use library routines for floating point operations.
7402 Dynamically allocate condition code registers.
7407 Do not try to dynamically allocate condition code registers, only
7408 use @code{icc0} and @code{fcc0}.
7413 Change ABI to use double word insns.
7418 Do not use double word instructions.
7423 Use floating point double instructions.
7428 Do not use floating point double instructions.
7433 Use media instructions.
7438 Do not use media instructions.
7443 Use multiply and add/subtract instructions.
7448 Do not use multiply and add/subtract instructions.
7451 @opindex mlibrary-pic
7453 Generate position-independent EABI code.
7458 Use only the first four media accumulator registers.
7463 Use all eight media accumulator registers.
7468 Pack VLIW instructions.
7473 Do not pack VLIW instructions.
7478 Do not mark ABI switches in e_flags.
7483 Enable the use of conditional-move instructions (default).
7485 This switch is mainly for debugging the compiler and will likely be removed
7486 in a future version.
7488 @item -mno-cond-move
7489 @opindex mno-cond-move
7491 Disable the use of conditional-move instructions.
7493 This switch is mainly for debugging the compiler and will likely be removed
7494 in a future version.
7499 Enable the use of conditional set instructions (default).
7501 This switch is mainly for debugging the compiler and will likely be removed
7502 in a future version.
7507 Disable the use of conditional set instructions.
7509 This switch is mainly for debugging the compiler and will likely be removed
7510 in a future version.
7515 Enable the use of conditional execution (default).
7517 This switch is mainly for debugging the compiler and will likely be removed
7518 in a future version.
7520 @item -mno-cond-exec
7521 @opindex mno-cond-exec
7523 Disable the use of conditional execution.
7525 This switch is mainly for debugging the compiler and will likely be removed
7526 in a future version.
7529 @opindex mvliw-branch
7531 Run a pass to pack branches into VLIW instructions (default).
7533 This switch is mainly for debugging the compiler and will likely be removed
7534 in a future version.
7536 @item -mno-vliw-branch
7537 @opindex mno-vliw-branch
7539 Do not run a pass to pack branches into VLIW instructions.
7541 This switch is mainly for debugging the compiler and will likely be removed
7542 in a future version.
7544 @item -mmulti-cond-exec
7545 @opindex mmulti-cond-exec
7547 Enable optimization of @code{&&} and @code{||} in conditional execution
7550 This switch is mainly for debugging the compiler and will likely be removed
7551 in a future version.
7553 @item -mno-multi-cond-exec
7554 @opindex mno-multi-cond-exec
7556 Disable optimization of @code{&&} and @code{||} in conditional execution.
7558 This switch is mainly for debugging the compiler and will likely be removed
7559 in a future version.
7561 @item -mnested-cond-exec
7562 @opindex mnested-cond-exec
7564 Enable nested conditional execution optimizations (default).
7566 This switch is mainly for debugging the compiler and will likely be removed
7567 in a future version.
7569 @item -mno-nested-cond-exec
7570 @opindex mno-nested-cond-exec
7572 Disable nested conditional execution optimizations.
7574 This switch is mainly for debugging the compiler and will likely be removed
7575 in a future version.
7577 @item -mtomcat-stats
7578 @opindex mtomcat-stats
7580 Cause gas to print out tomcat statistics.
7582 @item -mcpu=@var{cpu}
7585 Select the processor type for which to generate code. Possible values are
7586 @samp{simple}, @samp{tomcat}, @samp{fr500}, @samp{fr400}, @samp{fr300},
7591 @node H8/300 Options
7592 @subsection H8/300 Options
7594 These @samp{-m} options are defined for the H8/300 implementations:
7599 Shorten some address references at link time, when possible; uses the
7600 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
7601 ld, Using ld}, for a fuller description.
7605 Generate code for the H8/300H@.
7609 Generate code for the H8S@.
7613 Generate code for the H8S and H8/300H in the normal mode. This switch
7614 must be used either with -mh or -ms.
7618 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
7622 Make @code{int} data 32 bits by default.
7626 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
7627 The default for the H8/300H and H8S is to align longs and floats on 4
7629 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
7630 This option has no effect on the H8/300.
7634 @subsection HPPA Options
7635 @cindex HPPA Options
7637 These @samp{-m} options are defined for the HPPA family of computers:
7640 @item -march=@var{architecture-type}
7642 Generate code for the specified architecture. The choices for
7643 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
7644 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
7645 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
7646 architecture option for your machine. Code compiled for lower numbered
7647 architectures will run on higher numbered architectures, but not the
7650 PA 2.0 support currently requires gas snapshot 19990413 or later. The
7651 next release of binutils (current is 2.9.1) will probably contain PA 2.0
7655 @itemx -mpa-risc-1-1
7656 @itemx -mpa-risc-2-0
7657 @opindex mpa-risc-1-0
7658 @opindex mpa-risc-1-1
7659 @opindex mpa-risc-2-0
7660 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
7663 @opindex mbig-switch
7664 Generate code suitable for big switch tables. Use this option only if
7665 the assembler/linker complain about out of range branches within a switch
7668 @item -mjump-in-delay
7669 @opindex mjump-in-delay
7670 Fill delay slots of function calls with unconditional jump instructions
7671 by modifying the return pointer for the function call to be the target
7672 of the conditional jump.
7674 @item -mdisable-fpregs
7675 @opindex mdisable-fpregs
7676 Prevent floating point registers from being used in any manner. This is
7677 necessary for compiling kernels which perform lazy context switching of
7678 floating point registers. If you use this option and attempt to perform
7679 floating point operations, the compiler will abort.
7681 @item -mdisable-indexing
7682 @opindex mdisable-indexing
7683 Prevent the compiler from using indexing address modes. This avoids some
7684 rather obscure problems when compiling MIG generated code under MACH@.
7686 @item -mno-space-regs
7687 @opindex mno-space-regs
7688 Generate code that assumes the target has no space registers. This allows
7689 GCC to generate faster indirect calls and use unscaled index address modes.
7691 Such code is suitable for level 0 PA systems and kernels.
7693 @item -mfast-indirect-calls
7694 @opindex mfast-indirect-calls
7695 Generate code that assumes calls never cross space boundaries. This
7696 allows GCC to emit code which performs faster indirect calls.
7698 This option will not work in the presence of shared libraries or nested
7701 @item -mlong-load-store
7702 @opindex mlong-load-store
7703 Generate 3-instruction load and store sequences as sometimes required by
7704 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
7707 @item -mportable-runtime
7708 @opindex mportable-runtime
7709 Use the portable calling conventions proposed by HP for ELF systems.
7713 Enable the use of assembler directives only GAS understands.
7715 @item -mschedule=@var{cpu-type}
7717 Schedule code according to the constraints for the machine type
7718 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
7719 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
7720 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
7721 proper scheduling option for your machine. The default scheduling is
7725 @opindex mlinker-opt
7726 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
7727 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
7728 linkers in which they give bogus error messages when linking some programs.
7731 @opindex msoft-float
7732 Generate output containing library calls for floating point.
7733 @strong{Warning:} the requisite libraries are not available for all HPPA
7734 targets. Normally the facilities of the machine's usual C compiler are
7735 used, but this cannot be done directly in cross-compilation. You must make
7736 your own arrangements to provide suitable library functions for
7737 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
7738 does provide software floating point support.
7740 @option{-msoft-float} changes the calling convention in the output file;
7741 therefore, it is only useful if you compile @emph{all} of a program with
7742 this option. In particular, you need to compile @file{libgcc.a}, the
7743 library that comes with GCC, with @option{-msoft-float} in order for
7748 Generate the predefine, @code{_SIO}, for server IO. The default is
7749 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
7750 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO. These
7751 options are available under HP-UX and HI-UX.
7755 Use GNU ld specific options. This passes @option{-shared} to ld when
7756 building a shared library. It is the default when GCC is configured,
7757 explicitly or implicitly, with the GNU linker. This option does not
7758 have any affect on which ld is called, it only changes what parameters
7759 are passed to that ld. The ld that is called is determined by the
7760 @option{--with-ld} configure option, GCC's program search path, and
7761 finally by the user's @env{PATH}. The linker used by GCC can be printed
7762 using @samp{which `gcc -print-prog-name=ld`}.
7766 Use HP ld specific options. This passes @option{-b} to ld when building
7767 a shared library and passes @option{+Accept TypeMismatch} to ld on all
7768 links. It is the default when GCC is configured, explicitly or
7769 implicitly, with the HP linker. This option does not have any affect on
7770 which ld is called, it only changes what parameters are passed to that
7771 ld. The ld that is called is determined by the @option{--with-ld}
7772 configure option, GCC's program search path, and finally by the user's
7773 @env{PATH}. The linker used by GCC can be printed using @samp{which
7774 `gcc -print-prog-name=ld`}.
7779 Select the FDPIC ABI, that uses function descriptors to represent
7780 pointers to functions. Without any PIC/PIE-related options, it
7781 implies @option{-fPIE}. With @option{-fpic} or @option{-fpie}, it
7782 assumes GOT entries and small data are within a 12-bit range from the
7783 GOT base address; with @option{-fPIC} or @option{-fPIE}, GOT offsets
7784 are computed with 32 bits.
7787 @opindex minline-plt
7789 Enable inlining of PLT entries in function calls to functions that are
7790 not known to bind locally. It has no effect without @option{-mfdpic}.
7791 It's enabled by default if optimizing for speed and compiling for
7792 shared libraries (i.e., @option{-fPIC} or @option{-fpic}), or when an
7793 optimization option such as @option{-O3} or above is present in the
7799 Enable the use of @code{GPREL} relocations in the FDPIC ABI for data
7800 that is known to be in read-only sections. It's enabled by default,
7801 except for @option{-fpic} or @option{-fpie}: even though it may help
7802 make the global offset table smaller, it trades 1 instruction for 4.
7803 With @option{-fPIC} or @option{-fPIE}, it trades 3 instructions for 4,
7804 one of which may be shared by multiple symbols, and it avoids the need
7805 for a GOT entry for the referenced symbol, so it's more likely to be a
7806 win. If it is not, @option{-mno-gprel-ro} can be used to disable it.
7808 @item -multilib-library-pic
7809 @opindex multilib-library-pic
7811 Link with the (library, not FD) pic libraries. It's implied by
7812 @option{-mlibrary-pic}, as well as by @option{-fPIC} and
7813 @option{-fpic} without @option{-mfdpic}. You should never have to use
7819 Follow the EABI requirement of always creating a frame pointer whenever
7820 a stack frame is allocated. This option is enabled by default and can
7821 be disabled with @option{-mno-linked-fp}.
7824 @opindex mno-long-calls
7825 Generate code that uses long call sequences. This ensures that a call
7826 is always able to reach linker generated stubs. The default is to generate
7827 long calls only when the distance from the call site to the beginning
7828 of the function or translation unit, as the case may be, exceeds a
7829 predefined limit set by the branch type being used. The limits for
7830 normal calls are 7,600,000 and 240,000 bytes, respectively for the
7831 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
7834 Distances are measured from the beginning of functions when using the
7835 @option{-ffunction-sections} option, or when using the @option{-mgas}
7836 and @option{-mno-portable-runtime} options together under HP-UX with
7839 It is normally not desirable to use this option as it will degrade
7840 performance. However, it may be useful in large applications,
7841 particularly when partial linking is used to build the application.
7843 The types of long calls used depends on the capabilities of the
7844 assembler and linker, and the type of code being generated. The
7845 impact on systems that support long absolute calls, and long pic
7846 symbol-difference or pc-relative calls should be relatively small.
7847 However, an indirect call is used on 32-bit ELF systems in pic code
7848 and it is quite long.
7852 Suppress the generation of link options to search libdld.sl when the
7853 @option{-static} option is specified on HP-UX 10 and later.
7857 The HP-UX implementation of setlocale in libc has a dependency on
7858 libdld.sl. There isn't an archive version of libdld.sl. Thus,
7859 when the @option{-static} option is specified, special link options
7860 are needed to resolve this dependency.
7862 On HP-UX 10 and later, the GCC driver adds the necessary options to
7863 link with libdld.sl when the @option{-static} option is specified.
7864 This causes the resulting binary to be dynamic. On the 64-bit port,
7865 the linkers generate dynamic binaries by default in any case. The
7866 @option{-nolibdld} option can be used to prevent the GCC driver from
7867 adding these link options.
7871 Add support for multithreading with the @dfn{dce thread} library
7872 under HP-UX. This option sets flags for both the preprocessor and
7876 @node i386 and x86-64 Options
7877 @subsection Intel 386 and AMD x86-64 Options
7878 @cindex i386 Options
7879 @cindex x86-64 Options
7880 @cindex Intel 386 Options
7881 @cindex AMD x86-64 Options
7883 These @samp{-m} options are defined for the i386 and x86-64 family of
7887 @item -mtune=@var{cpu-type}
7889 Tune to @var{cpu-type} everything applicable about the generated code, except
7890 for the ABI and the set of available instructions. The choices for
7894 Original Intel's i386 CPU.
7896 Intel's i486 CPU. (No scheduling is implemented for this chip.)
7898 Intel Pentium CPU with no MMX support.
7900 Intel PentiumMMX CPU based on Pentium core with MMX instruction set support.
7901 @item i686, pentiumpro
7902 Intel PentiumPro CPU.
7904 Intel Pentium2 CPU based on PentiumPro core with MMX instruction set support.
7905 @item pentium3, pentium3m
7906 Intel Pentium3 CPU based on PentiumPro core with MMX and SSE instruction set
7909 Low power version of Intel Pentium3 CPU with MMX, SSE and SSE2 instruction set
7910 support. Used by Centrino notebooks.
7911 @item pentium4, pentium4m
7912 Intel Pentium4 CPU with MMX, SSE and SSE2 instruction set support.
7914 Improved version of Intel Pentium4 CPU with MMX, SSE, SSE2 and SSE3 instruction
7917 Improved version of Intel Pentium4 CPU with 64-bit extensions, MMX, SSE,
7918 SSE2 and SSE3 instruction set support.
7920 AMD K6 CPU with MMX instruction set support.
7922 Improved versions of AMD K6 CPU with MMX and 3dNOW! instruction set support.
7923 @item athlon, athlon-tbird
7924 AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and SSE prefetch instructions
7926 @item athlon-4, athlon-xp, athlon-mp
7927 Improved AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and full SSE
7928 instruction set support.
7929 @item k8, opteron, athlon64, athlon-fx
7930 AMD K8 core based CPUs with x86-64 instruction set support. (This supersets
7931 MMX, SSE, SSE2, 3dNOW!, enhanced 3dNOW! and 64-bit instruction set extensions.)
7933 IDT Winchip C6 CPU, dealt in same way as i486 with additional MMX instruction
7936 IDT Winchip2 CPU, dealt in same way as i486 with additional MMX and 3dNOW!
7937 instruction set support.
7939 Via C3 CPU with MMX and 3dNOW! instruction set support. (No scheduling is
7940 implemented for this chip.)
7942 Via C3-2 CPU with MMX and SSE instruction set support. (No scheduling is
7943 implemented for this chip.)
7946 While picking a specific @var{cpu-type} will schedule things appropriately
7947 for that particular chip, the compiler will not generate any code that
7948 does not run on the i386 without the @option{-march=@var{cpu-type}} option
7951 @item -march=@var{cpu-type}
7953 Generate instructions for the machine type @var{cpu-type}. The choices
7954 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
7955 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
7957 @item -mcpu=@var{cpu-type}
7959 A deprecated synonym for @option{-mtune}.
7968 @opindex mpentiumpro
7969 These options are synonyms for @option{-mtune=i386}, @option{-mtune=i486},
7970 @option{-mtune=pentium}, and @option{-mtune=pentiumpro} respectively.
7971 These synonyms are deprecated.
7973 @item -mfpmath=@var{unit}
7975 Generate floating point arithmetics for selected unit @var{unit}. The choices
7980 Use the standard 387 floating point coprocessor present majority of chips and
7981 emulated otherwise. Code compiled with this option will run almost everywhere.
7982 The temporary results are computed in 80bit precision instead of precision
7983 specified by the type resulting in slightly different results compared to most
7984 of other chips. See @option{-ffloat-store} for more detailed description.
7986 This is the default choice for i386 compiler.
7989 Use scalar floating point instructions present in the SSE instruction set.
7990 This instruction set is supported by Pentium3 and newer chips, in the AMD line
7991 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
7992 instruction set supports only single precision arithmetics, thus the double and
7993 extended precision arithmetics is still done using 387. Later version, present
7994 only in Pentium4 and the future AMD x86-64 chips supports double precision
7997 For i387 you need to use @option{-march=@var{cpu-type}}, @option{-msse} or
7998 @option{-msse2} switches to enable SSE extensions and make this option
7999 effective. For x86-64 compiler, these extensions are enabled by default.
8001 The resulting code should be considerably faster in the majority of cases and avoid
8002 the numerical instability problems of 387 code, but may break some existing
8003 code that expects temporaries to be 80bit.
8005 This is the default choice for the x86-64 compiler.
8008 Attempt to utilize both instruction sets at once. This effectively double the
8009 amount of available registers and on chips with separate execution units for
8010 387 and SSE the execution resources too. Use this option with care, as it is
8011 still experimental, because the GCC register allocator does not model separate
8012 functional units well resulting in instable performance.
8015 @item -masm=@var{dialect}
8016 @opindex masm=@var{dialect}
8017 Output asm instructions using selected @var{dialect}. Supported choices are
8018 @samp{intel} or @samp{att} (the default one).
8023 @opindex mno-ieee-fp
8024 Control whether or not the compiler uses IEEE floating point
8025 comparisons. These handle correctly the case where the result of a
8026 comparison is unordered.
8029 @opindex msoft-float
8030 Generate output containing library calls for floating point.
8031 @strong{Warning:} the requisite libraries are not part of GCC@.
8032 Normally the facilities of the machine's usual C compiler are used, but
8033 this can't be done directly in cross-compilation. You must make your
8034 own arrangements to provide suitable library functions for
8037 On machines where a function returns floating point results in the 80387
8038 register stack, some floating point opcodes may be emitted even if
8039 @option{-msoft-float} is used.
8041 @item -mno-fp-ret-in-387
8042 @opindex mno-fp-ret-in-387
8043 Do not use the FPU registers for return values of functions.
8045 The usual calling convention has functions return values of types
8046 @code{float} and @code{double} in an FPU register, even if there
8047 is no FPU@. The idea is that the operating system should emulate
8050 The option @option{-mno-fp-ret-in-387} causes such values to be returned
8051 in ordinary CPU registers instead.
8053 @item -mno-fancy-math-387
8054 @opindex mno-fancy-math-387
8055 Some 387 emulators do not support the @code{sin}, @code{cos} and
8056 @code{sqrt} instructions for the 387. Specify this option to avoid
8057 generating those instructions. This option is the default on FreeBSD,
8058 OpenBSD and NetBSD@. This option is overridden when @option{-march}
8059 indicates that the target cpu will always have an FPU and so the
8060 instruction will not need emulation. As of revision 2.6.1, these
8061 instructions are not generated unless you also use the
8062 @option{-funsafe-math-optimizations} switch.
8064 @item -malign-double
8065 @itemx -mno-align-double
8066 @opindex malign-double
8067 @opindex mno-align-double
8068 Control whether GCC aligns @code{double}, @code{long double}, and
8069 @code{long long} variables on a two word boundary or a one word
8070 boundary. Aligning @code{double} variables on a two word boundary will
8071 produce code that runs somewhat faster on a @samp{Pentium} at the
8072 expense of more memory.
8074 @strong{Warning:} if you use the @option{-malign-double} switch,
8075 structures containing the above types will be aligned differently than
8076 the published application binary interface specifications for the 386
8077 and will not be binary compatible with structures in code compiled
8078 without that switch.
8080 @item -m96bit-long-double
8081 @itemx -m128bit-long-double
8082 @opindex m96bit-long-double
8083 @opindex m128bit-long-double
8084 These switches control the size of @code{long double} type. The i386
8085 application binary interface specifies the size to be 96 bits,
8086 so @option{-m96bit-long-double} is the default in 32 bit mode.
8088 Modern architectures (Pentium and newer) would prefer @code{long double}
8089 to be aligned to an 8 or 16 byte boundary. In arrays or structures
8090 conforming to the ABI, this would not be possible. So specifying a
8091 @option{-m128bit-long-double} will align @code{long double}
8092 to a 16 byte boundary by padding the @code{long double} with an additional
8095 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
8096 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
8098 Notice that neither of these options enable any extra precision over the x87
8099 standard of 80 bits for a @code{long double}.
8101 @strong{Warning:} if you override the default value for your target ABI, the
8102 structures and arrays containing @code{long double} variables will change
8103 their size as well as function calling convention for function taking
8104 @code{long double} will be modified. Hence they will not be binary
8105 compatible with arrays or structures in code compiled without that switch.
8109 @itemx -mno-svr3-shlib
8110 @opindex msvr3-shlib
8111 @opindex mno-svr3-shlib
8112 Control whether GCC places uninitialized local variables into the
8113 @code{bss} or @code{data} segments. @option{-msvr3-shlib} places them
8114 into @code{bss}. These options are meaningful only on System V Release 3.
8118 Use a different function-calling convention, in which functions that
8119 take a fixed number of arguments return with the @code{ret} @var{num}
8120 instruction, which pops their arguments while returning. This saves one
8121 instruction in the caller since there is no need to pop the arguments
8124 You can specify that an individual function is called with this calling
8125 sequence with the function attribute @samp{stdcall}. You can also
8126 override the @option{-mrtd} option by using the function attribute
8127 @samp{cdecl}. @xref{Function Attributes}.
8129 @strong{Warning:} this calling convention is incompatible with the one
8130 normally used on Unix, so you cannot use it if you need to call
8131 libraries compiled with the Unix compiler.
8133 Also, you must provide function prototypes for all functions that
8134 take variable numbers of arguments (including @code{printf});
8135 otherwise incorrect code will be generated for calls to those
8138 In addition, seriously incorrect code will result if you call a
8139 function with too many arguments. (Normally, extra arguments are
8140 harmlessly ignored.)
8142 @item -mregparm=@var{num}
8144 Control how many registers are used to pass integer arguments. By
8145 default, no registers are used to pass arguments, and at most 3
8146 registers can be used. You can control this behavior for a specific
8147 function by using the function attribute @samp{regparm}.
8148 @xref{Function Attributes}.
8150 @strong{Warning:} if you use this switch, and
8151 @var{num} is nonzero, then you must build all modules with the same
8152 value, including any libraries. This includes the system libraries and
8155 @item -mpreferred-stack-boundary=@var{num}
8156 @opindex mpreferred-stack-boundary
8157 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
8158 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
8159 the default is 4 (16 bytes or 128 bits), except when optimizing for code
8160 size (@option{-Os}), in which case the default is the minimum correct
8161 alignment (4 bytes for x86, and 8 bytes for x86-64).
8163 On Pentium and PentiumPro, @code{double} and @code{long double} values
8164 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
8165 suffer significant run time performance penalties. On Pentium III, the
8166 Streaming SIMD Extension (SSE) data type @code{__m128} suffers similar
8167 penalties if it is not 16 byte aligned.
8169 To ensure proper alignment of this values on the stack, the stack boundary
8170 must be as aligned as that required by any value stored on the stack.
8171 Further, every function must be generated such that it keeps the stack
8172 aligned. Thus calling a function compiled with a higher preferred
8173 stack boundary from a function compiled with a lower preferred stack
8174 boundary will most likely misalign the stack. It is recommended that
8175 libraries that use callbacks always use the default setting.
8177 This extra alignment does consume extra stack space, and generally
8178 increases code size. Code that is sensitive to stack space usage, such
8179 as embedded systems and operating system kernels, may want to reduce the
8180 preferred alignment to @option{-mpreferred-stack-boundary=2}.
8198 These switches enable or disable the use of built-in functions that allow
8199 direct access to the MMX, SSE, SSE2, SSE3 and 3Dnow extensions of the
8202 @xref{X86 Built-in Functions}, for details of the functions enabled
8203 and disabled by these switches.
8205 To have SSE/SSE2 instructions generated automatically from floating-point
8206 code, see @option{-mfpmath=sse}.
8209 @itemx -mno-push-args
8211 @opindex mno-push-args
8212 Use PUSH operations to store outgoing parameters. This method is shorter
8213 and usually equally fast as method using SUB/MOV operations and is enabled
8214 by default. In some cases disabling it may improve performance because of
8215 improved scheduling and reduced dependencies.
8217 @item -maccumulate-outgoing-args
8218 @opindex maccumulate-outgoing-args
8219 If enabled, the maximum amount of space required for outgoing arguments will be
8220 computed in the function prologue. This is faster on most modern CPUs
8221 because of reduced dependencies, improved scheduling and reduced stack usage
8222 when preferred stack boundary is not equal to 2. The drawback is a notable
8223 increase in code size. This switch implies @option{-mno-push-args}.
8227 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
8228 on thread-safe exception handling must compile and link all code with the
8229 @option{-mthreads} option. When compiling, @option{-mthreads} defines
8230 @option{-D_MT}; when linking, it links in a special thread helper library
8231 @option{-lmingwthrd} which cleans up per thread exception handling data.
8233 @item -mno-align-stringops
8234 @opindex mno-align-stringops
8235 Do not align destination of inlined string operations. This switch reduces
8236 code size and improves performance in case the destination is already aligned,
8237 but GCC doesn't know about it.
8239 @item -minline-all-stringops
8240 @opindex minline-all-stringops
8241 By default GCC inlines string operations only when destination is known to be
8242 aligned at least to 4 byte boundary. This enables more inlining, increase code
8243 size, but may improve performance of code that depends on fast memcpy, strlen
8244 and memset for short lengths.
8246 @item -momit-leaf-frame-pointer
8247 @opindex momit-leaf-frame-pointer
8248 Don't keep the frame pointer in a register for leaf functions. This
8249 avoids the instructions to save, set up and restore frame pointers and
8250 makes an extra register available in leaf functions. The option
8251 @option{-fomit-frame-pointer} removes the frame pointer for all functions
8252 which might make debugging harder.
8254 @item -mtls-direct-seg-refs
8255 @itemx -mno-tls-direct-seg-refs
8256 @opindex mtls-direct-seg-refs
8257 Controls whether TLS variables may be accessed with offsets from the
8258 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
8259 or whether the thread base pointer must be added. Whether or not this
8260 is legal depends on the operating system, and whether it maps the
8261 segment to cover the entire TLS area.
8263 For systems that use GNU libc, the default is on.
8266 These @samp{-m} switches are supported in addition to the above
8267 on AMD x86-64 processors in 64-bit environments.
8274 Generate code for a 32-bit or 64-bit environment.
8275 The 32-bit environment sets int, long and pointer to 32 bits and
8276 generates code that runs on any i386 system.
8277 The 64-bit environment sets int to 32 bits and long and pointer
8278 to 64 bits and generates code for AMD's x86-64 architecture.
8281 @opindex no-red-zone
8282 Do not use a so called red zone for x86-64 code. The red zone is mandated
8283 by the x86-64 ABI, it is a 128-byte area beyond the location of the
8284 stack pointer that will not be modified by signal or interrupt handlers
8285 and therefore can be used for temporary data without adjusting the stack
8286 pointer. The flag @option{-mno-red-zone} disables this red zone.
8288 @item -mcmodel=small
8289 @opindex mcmodel=small
8290 Generate code for the small code model: the program and its symbols must
8291 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
8292 Programs can be statically or dynamically linked. This is the default
8295 @item -mcmodel=kernel
8296 @opindex mcmodel=kernel
8297 Generate code for the kernel code model. The kernel runs in the
8298 negative 2 GB of the address space.
8299 This model has to be used for Linux kernel code.
8301 @item -mcmodel=medium
8302 @opindex mcmodel=medium
8303 Generate code for the medium model: The program is linked in the lower 2
8304 GB of the address space but symbols can be located anywhere in the
8305 address space. Programs can be statically or dynamically linked, but
8306 building of shared libraries are not supported with the medium model.
8308 @item -mcmodel=large
8309 @opindex mcmodel=large
8310 Generate code for the large model: This model makes no assumptions
8311 about addresses and sizes of sections. Currently GCC does not implement
8316 @subsection IA-64 Options
8317 @cindex IA-64 Options
8319 These are the @samp{-m} options defined for the Intel IA-64 architecture.
8323 @opindex mbig-endian
8324 Generate code for a big endian target. This is the default for HP-UX@.
8326 @item -mlittle-endian
8327 @opindex mlittle-endian
8328 Generate code for a little endian target. This is the default for AIX5
8335 Generate (or don't) code for the GNU assembler. This is the default.
8336 @c Also, this is the default if the configure option @option{--with-gnu-as}
8343 Generate (or don't) code for the GNU linker. This is the default.
8344 @c Also, this is the default if the configure option @option{--with-gnu-ld}
8349 Generate code that does not use a global pointer register. The result
8350 is not position independent code, and violates the IA-64 ABI@.
8352 @item -mvolatile-asm-stop
8353 @itemx -mno-volatile-asm-stop
8354 @opindex mvolatile-asm-stop
8355 @opindex mno-volatile-asm-stop
8356 Generate (or don't) a stop bit immediately before and after volatile asm
8361 Generate code that works around Itanium B step errata.
8363 @item -mregister-names
8364 @itemx -mno-register-names
8365 @opindex mregister-names
8366 @opindex mno-register-names
8367 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
8368 the stacked registers. This may make assembler output more readable.
8374 Disable (or enable) optimizations that use the small data section. This may
8375 be useful for working around optimizer bugs.
8378 @opindex mconstant-gp
8379 Generate code that uses a single constant global pointer value. This is
8380 useful when compiling kernel code.
8384 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
8385 This is useful when compiling firmware code.
8387 @item -minline-float-divide-min-latency
8388 @opindex minline-float-divide-min-latency
8389 Generate code for inline divides of floating point values
8390 using the minimum latency algorithm.
8392 @item -minline-float-divide-max-throughput
8393 @opindex minline-float-divide-max-throughput
8394 Generate code for inline divides of floating point values
8395 using the maximum throughput algorithm.
8397 @item -minline-int-divide-min-latency
8398 @opindex minline-int-divide-min-latency
8399 Generate code for inline divides of integer values
8400 using the minimum latency algorithm.
8402 @item -minline-int-divide-max-throughput
8403 @opindex minline-int-divide-max-throughput
8404 Generate code for inline divides of integer values
8405 using the maximum throughput algorithm.
8407 @item -mno-dwarf2-asm
8409 @opindex mno-dwarf2-asm
8410 @opindex mdwarf2-asm
8411 Don't (or do) generate assembler code for the DWARF2 line number debugging
8412 info. This may be useful when not using the GNU assembler.
8414 @item -mfixed-range=@var{register-range}
8415 @opindex mfixed-range
8416 Generate code treating the given register range as fixed registers.
8417 A fixed register is one that the register allocator can not use. This is
8418 useful when compiling kernel code. A register range is specified as
8419 two registers separated by a dash. Multiple register ranges can be
8420 specified separated by a comma.
8422 @item -mearly-stop-bits
8423 @itemx -mno-early-stop-bits
8424 @opindex mearly-stop-bits
8425 @opindex mno-early-stop-bits
8426 Allow stop bits to be placed earlier than immediately preceding the
8427 instruction that triggered the stop bit. This can improve instruction
8428 scheduling, but does not always do so.
8431 @node M32R/D Options
8432 @subsection M32R/D Options
8433 @cindex M32R/D options
8435 These @option{-m} options are defined for Renesas M32R/D architectures:
8440 Generate code for the M32R/2@.
8444 Generate code for the M32R/X@.
8448 Generate code for the M32R@. This is the default.
8451 @opindex mmodel=small
8452 Assume all objects live in the lower 16MB of memory (so that their addresses
8453 can be loaded with the @code{ld24} instruction), and assume all subroutines
8454 are reachable with the @code{bl} instruction.
8455 This is the default.
8457 The addressability of a particular object can be set with the
8458 @code{model} attribute.
8460 @item -mmodel=medium
8461 @opindex mmodel=medium
8462 Assume objects may be anywhere in the 32-bit address space (the compiler
8463 will generate @code{seth/add3} instructions to load their addresses), and
8464 assume all subroutines are reachable with the @code{bl} instruction.
8467 @opindex mmodel=large
8468 Assume objects may be anywhere in the 32-bit address space (the compiler
8469 will generate @code{seth/add3} instructions to load their addresses), and
8470 assume subroutines may not be reachable with the @code{bl} instruction
8471 (the compiler will generate the much slower @code{seth/add3/jl}
8472 instruction sequence).
8475 @opindex msdata=none
8476 Disable use of the small data area. Variables will be put into
8477 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
8478 @code{section} attribute has been specified).
8479 This is the default.
8481 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
8482 Objects may be explicitly put in the small data area with the
8483 @code{section} attribute using one of these sections.
8486 @opindex msdata=sdata
8487 Put small global and static data in the small data area, but do not
8488 generate special code to reference them.
8492 Put small global and static data in the small data area, and generate
8493 special instructions to reference them.
8497 @cindex smaller data references
8498 Put global and static objects less than or equal to @var{num} bytes
8499 into the small data or bss sections instead of the normal data or bss
8500 sections. The default value of @var{num} is 8.
8501 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
8502 for this option to have any effect.
8504 All modules should be compiled with the same @option{-G @var{num}} value.
8505 Compiling with different values of @var{num} may or may not work; if it
8506 doesn't the linker will give an error message---incorrect code will not be
8511 Makes the M32R specific code in the compiler display some statistics
8512 that might help in debugging programs.
8515 @opindex malign-loops
8516 Align all loops to a 32-byte boundary.
8518 @item -mno-align-loops
8519 @opindex mno-align-loops
8520 Do not enforce a 32-byte alignment for loops. This is the default.
8522 @item -missue-rate=@var{number}
8523 @opindex missue-rate=@var{number}
8524 Issue @var{number} instructions per cycle. @var{number} can only be 1
8527 @item -mbranch-cost=@var{number}
8528 @opindex mbranch-cost=@var{number}
8529 @var{number} can only be 1 or 2. If it is 1 then branches will be
8530 preferred over conditional code, if it is 2, then the opposite will
8533 @item -mflush-trap=@var{number}
8534 @opindex mflush-trap=@var{number}
8535 Specifies the trap number to use to flush the cache. The default is
8536 12. Valid numbers are between 0 and 15 inclusive.
8538 @item -mno-flush-trap
8539 @opindex mno-flush-trap
8540 Specifies that the cache cannot be flushed by using a trap.
8542 @item -mflush-func=@var{name}
8543 @opindex mflush-func=@var{name}
8544 Specifies the name of the operating system function to call to flush
8545 the cache. The default is @emph{_flush_cache}, but a function call
8546 will only be used if a trap is not available.
8548 @item -mno-flush-func
8549 @opindex mno-flush-func
8550 Indicates that there is no OS function for flushing the cache.
8554 @node M680x0 Options
8555 @subsection M680x0 Options
8556 @cindex M680x0 options
8558 These are the @samp{-m} options defined for the 68000 series. The default
8559 values for these options depends on which style of 68000 was selected when
8560 the compiler was configured; the defaults for the most common choices are
8568 Generate output for a 68000. This is the default
8569 when the compiler is configured for 68000-based systems.
8571 Use this option for microcontrollers with a 68000 or EC000 core,
8572 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
8578 Generate output for a 68020. This is the default
8579 when the compiler is configured for 68020-based systems.
8583 Generate output containing 68881 instructions for floating point.
8584 This is the default for most 68020 systems unless @option{--nfp} was
8585 specified when the compiler was configured.
8589 Generate output for a 68030. This is the default when the compiler is
8590 configured for 68030-based systems.
8594 Generate output for a 68040. This is the default when the compiler is
8595 configured for 68040-based systems.
8597 This option inhibits the use of 68881/68882 instructions that have to be
8598 emulated by software on the 68040. Use this option if your 68040 does not
8599 have code to emulate those instructions.
8603 Generate output for a 68060. This is the default when the compiler is
8604 configured for 68060-based systems.
8606 This option inhibits the use of 68020 and 68881/68882 instructions that
8607 have to be emulated by software on the 68060. Use this option if your 68060
8608 does not have code to emulate those instructions.
8612 Generate output for a CPU32. This is the default
8613 when the compiler is configured for CPU32-based systems.
8615 Use this option for microcontrollers with a
8616 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
8617 68336, 68340, 68341, 68349 and 68360.
8621 Generate output for a 520X ``coldfire'' family cpu. This is the default
8622 when the compiler is configured for 520X-based systems.
8624 Use this option for microcontroller with a 5200 core, including
8625 the MCF5202, MCF5203, MCF5204 and MCF5202.
8630 Generate output for a 68040, without using any of the new instructions.
8631 This results in code which can run relatively efficiently on either a
8632 68020/68881 or a 68030 or a 68040. The generated code does use the
8633 68881 instructions that are emulated on the 68040.
8637 Generate output for a 68060, without using any of the new instructions.
8638 This results in code which can run relatively efficiently on either a
8639 68020/68881 or a 68030 or a 68040. The generated code does use the
8640 68881 instructions that are emulated on the 68060.
8643 @opindex msoft-float
8644 Generate output containing library calls for floating point.
8645 @strong{Warning:} the requisite libraries are not available for all m68k
8646 targets. Normally the facilities of the machine's usual C compiler are
8647 used, but this can't be done directly in cross-compilation. You must
8648 make your own arrangements to provide suitable library functions for
8649 cross-compilation. The embedded targets @samp{m68k-*-aout} and
8650 @samp{m68k-*-coff} do provide software floating point support.
8654 Consider type @code{int} to be 16 bits wide, like @code{short int}.
8655 Additionally, parameters passed on the stack are also aligned to a
8656 16-bit boundary even on targets whose API mandates promotion to 32-bit.
8659 @opindex mnobitfield
8660 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
8661 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
8665 Do use the bit-field instructions. The @option{-m68020} option implies
8666 @option{-mbitfield}. This is the default if you use a configuration
8667 designed for a 68020.
8671 Use a different function-calling convention, in which functions
8672 that take a fixed number of arguments return with the @code{rtd}
8673 instruction, which pops their arguments while returning. This
8674 saves one instruction in the caller since there is no need to pop
8675 the arguments there.
8677 This calling convention is incompatible with the one normally
8678 used on Unix, so you cannot use it if you need to call libraries
8679 compiled with the Unix compiler.
8681 Also, you must provide function prototypes for all functions that
8682 take variable numbers of arguments (including @code{printf});
8683 otherwise incorrect code will be generated for calls to those
8686 In addition, seriously incorrect code will result if you call a
8687 function with too many arguments. (Normally, extra arguments are
8688 harmlessly ignored.)
8690 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
8691 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
8694 @itemx -mno-align-int
8696 @opindex mno-align-int
8697 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
8698 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
8699 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
8700 Aligning variables on 32-bit boundaries produces code that runs somewhat
8701 faster on processors with 32-bit busses at the expense of more memory.
8703 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
8704 align structures containing the above types differently than
8705 most published application binary interface specifications for the m68k.
8709 Use the pc-relative addressing mode of the 68000 directly, instead of
8710 using a global offset table. At present, this option implies @option{-fpic},
8711 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
8712 not presently supported with @option{-mpcrel}, though this could be supported for
8713 68020 and higher processors.
8715 @item -mno-strict-align
8716 @itemx -mstrict-align
8717 @opindex mno-strict-align
8718 @opindex mstrict-align
8719 Do not (do) assume that unaligned memory references will be handled by
8723 Generate code that allows the data segment to be located in a different
8724 area of memory from the text segment. This allows for execute in place in
8725 an environment without virtual memory management. This option implies -fPIC.
8728 Generate code that assumes that the data segment follows the text segment.
8729 This is the default.
8731 @item -mid-shared-library
8732 Generate code that supports shared libraries via the library ID method.
8733 This allows for execute in place and shared libraries in an environment
8734 without virtual memory management. This option implies -fPIC.
8736 @item -mno-id-shared-library
8737 Generate code that doesn't assume ID based shared libraries are being used.
8738 This is the default.
8740 @item -mshared-library-id=n
8741 Specified the identification number of the ID based shared library being
8742 compiled. Specifying a value of 0 will generate more compact code, specifying
8743 other values will force the allocation of that number to the current
8744 library but is no more space or time efficient than omitting this option.
8748 @node M68hc1x Options
8749 @subsection M68hc1x Options
8750 @cindex M68hc1x options
8752 These are the @samp{-m} options defined for the 68hc11 and 68hc12
8753 microcontrollers. The default values for these options depends on
8754 which style of microcontroller was selected when the compiler was configured;
8755 the defaults for the most common choices are given below.
8762 Generate output for a 68HC11. This is the default
8763 when the compiler is configured for 68HC11-based systems.
8769 Generate output for a 68HC12. This is the default
8770 when the compiler is configured for 68HC12-based systems.
8776 Generate output for a 68HCS12.
8779 @opindex mauto-incdec
8780 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
8787 Enable the use of 68HC12 min and max instructions.
8790 @itemx -mno-long-calls
8791 @opindex mlong-calls
8792 @opindex mno-long-calls
8793 Treat all calls as being far away (near). If calls are assumed to be
8794 far away, the compiler will use the @code{call} instruction to
8795 call a function and the @code{rtc} instruction for returning.
8799 Consider type @code{int} to be 16 bits wide, like @code{short int}.
8801 @item -msoft-reg-count=@var{count}
8802 @opindex msoft-reg-count
8803 Specify the number of pseudo-soft registers which are used for the
8804 code generation. The maximum number is 32. Using more pseudo-soft
8805 register may or may not result in better code depending on the program.
8806 The default is 4 for 68HC11 and 2 for 68HC12.
8811 @subsection MCore Options
8812 @cindex MCore options
8814 These are the @samp{-m} options defined for the Motorola M*Core
8822 @opindex mno-hardlit
8823 Inline constants into the code stream if it can be done in two
8824 instructions or less.
8830 Use the divide instruction. (Enabled by default).
8832 @item -mrelax-immediate
8833 @itemx -mno-relax-immediate
8834 @opindex mrelax-immediate
8835 @opindex mno-relax-immediate
8836 Allow arbitrary sized immediates in bit operations.
8838 @item -mwide-bitfields
8839 @itemx -mno-wide-bitfields
8840 @opindex mwide-bitfields
8841 @opindex mno-wide-bitfields
8842 Always treat bit-fields as int-sized.
8844 @item -m4byte-functions
8845 @itemx -mno-4byte-functions
8846 @opindex m4byte-functions
8847 @opindex mno-4byte-functions
8848 Force all functions to be aligned to a four byte boundary.
8850 @item -mcallgraph-data
8851 @itemx -mno-callgraph-data
8852 @opindex mcallgraph-data
8853 @opindex mno-callgraph-data
8854 Emit callgraph information.
8857 @itemx -mno-slow-bytes
8858 @opindex mslow-bytes
8859 @opindex mno-slow-bytes
8860 Prefer word access when reading byte quantities.
8862 @item -mlittle-endian
8864 @opindex mlittle-endian
8865 @opindex mbig-endian
8866 Generate code for a little endian target.
8872 Generate code for the 210 processor.
8876 @subsection MIPS Options
8877 @cindex MIPS options
8883 Generate big-endian code.
8887 Generate little-endian code. This is the default for @samp{mips*el-*-*}
8890 @item -march=@var{arch}
8892 Generate code that will run on @var{arch}, which can be the name of a
8893 generic MIPS ISA, or the name of a particular processor.
8895 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
8896 @samp{mips32}, @samp{mips32r2}, and @samp{mips64}.
8897 The processor names are:
8898 @samp{4kc}, @samp{4kp}, @samp{5kc}, @samp{20kc},
8900 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
8901 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000}, @samp{rm7000},
8905 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4130}, @samp{vr4300},
8906 @samp{vr5000}, @samp{vr5400} and @samp{vr5500}.
8907 The special value @samp{from-abi} selects the
8908 most compatible architecture for the selected ABI (that is,
8909 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
8911 In processor names, a final @samp{000} can be abbreviated as @samp{k}
8912 (for example, @samp{-march=r2k}). Prefixes are optional, and
8913 @samp{vr} may be written @samp{r}.
8915 GCC defines two macros based on the value of this option. The first
8916 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
8917 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
8918 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
8919 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
8920 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
8922 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
8923 above. In other words, it will have the full prefix and will not
8924 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
8925 the macro names the resolved architecture (either @samp{"mips1"} or
8926 @samp{"mips3"}). It names the default architecture when no
8927 @option{-march} option is given.
8929 @item -mtune=@var{arch}
8931 Optimize for @var{arch}. Among other things, this option controls
8932 the way instructions are scheduled, and the perceived cost of arithmetic
8933 operations. The list of @var{arch} values is the same as for
8936 When this option is not used, GCC will optimize for the processor
8937 specified by @option{-march}. By using @option{-march} and
8938 @option{-mtune} together, it is possible to generate code that will
8939 run on a family of processors, but optimize the code for one
8940 particular member of that family.
8942 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
8943 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
8944 @samp{-march} ones described above.
8948 Equivalent to @samp{-march=mips1}.
8952 Equivalent to @samp{-march=mips2}.
8956 Equivalent to @samp{-march=mips3}.
8960 Equivalent to @samp{-march=mips4}.
8964 Equivalent to @samp{-march=mips32}.
8968 Equivalent to @samp{-march=mips32r2}.
8972 Equivalent to @samp{-march=mips64}.
8978 Use (do not use) the MIPS16 ISA.
8990 Generate code for the given ABI@.
8992 Note that the EABI has a 32-bit and a 64-bit variant. GCC normally
8993 generates 64-bit code when you select a 64-bit architecture, but you
8994 can use @option{-mgp32} to get 32-bit code instead.
8996 For information about the O64 ABI, see
8997 @w{@uref{http://gcc.gnu.org/projects/mipso64-abi.html}}.
9000 @itemx -mno-abicalls
9002 @opindex mno-abicalls
9003 Generate (do not generate) SVR4-style position-independent code.
9004 @option{-mabicalls} is the default for SVR4-based systems.
9010 Lift (do not lift) the usual restrictions on the size of the global
9013 GCC normally uses a single instruction to load values from the GOT.
9014 While this is relatively efficient, it will only work if the GOT
9015 is smaller than about 64k. Anything larger will cause the linker
9016 to report an error such as:
9018 @cindex relocation truncated to fit (MIPS)
9020 relocation truncated to fit: R_MIPS_GOT16 foobar
9023 If this happens, you should recompile your code with @option{-mxgot}.
9024 It should then work with very large GOTs, although it will also be
9025 less efficient, since it will take three instructions to fetch the
9026 value of a global symbol.
9028 Note that some linkers can create multiple GOTs. If you have such a
9029 linker, you should only need to use @option{-mxgot} when a single object
9030 file accesses more than 64k's worth of GOT entries. Very few do.
9032 These options have no effect unless GCC is generating position
9037 Assume that general-purpose registers are 32 bits wide.
9041 Assume that general-purpose registers are 64 bits wide.
9045 Assume that floating-point registers are 32 bits wide.
9049 Assume that floating-point registers are 64 bits wide.
9052 @opindex mhard-float
9053 Use floating-point coprocessor instructions.
9056 @opindex msoft-float
9057 Do not use floating-point coprocessor instructions. Implement
9058 floating-point calculations using library calls instead.
9060 @item -msingle-float
9061 @opindex msingle-float
9062 Assume that the floating-point coprocessor only supports single-precision
9065 @itemx -mdouble-float
9066 @opindex mdouble-float
9067 Assume that the floating-point coprocessor supports double-precision
9068 operations. This is the default.
9072 Force @code{int} and @code{long} types to be 64 bits wide. See
9073 @option{-mlong32} for an explanation of the default and the way
9074 that the pointer size is determined.
9078 Force @code{long} types to be 64 bits wide. See @option{-mlong32} for
9079 an explanation of the default and the way that the pointer size is
9084 Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
9086 The default size of @code{int}s, @code{long}s and pointers depends on
9087 the ABI@. All the supported ABIs use 32-bit @code{int}s. The n64 ABI
9088 uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
9089 32-bit @code{long}s. Pointers are the same size as @code{long}s,
9090 or the same size as integer registers, whichever is smaller.
9094 @cindex smaller data references (MIPS)
9095 @cindex gp-relative references (MIPS)
9096 Put global and static items less than or equal to @var{num} bytes into
9097 the small data or bss section instead of the normal data or bss section.
9098 This allows the data to be accessed using a single instruction.
9100 All modules should be compiled with the same @option{-G @var{num}}
9103 @item -membedded-data
9104 @itemx -mno-embedded-data
9105 @opindex membedded-data
9106 @opindex mno-embedded-data
9107 Allocate variables to the read-only data section first if possible, then
9108 next in the small data section if possible, otherwise in data. This gives
9109 slightly slower code than the default, but reduces the amount of RAM required
9110 when executing, and thus may be preferred for some embedded systems.
9112 @item -muninit-const-in-rodata
9113 @itemx -mno-uninit-const-in-rodata
9114 @opindex muninit-const-in-rodata
9115 @opindex mno-uninit-const-in-rodata
9116 Put uninitialized @code{const} variables in the read-only data section.
9117 This option is only meaningful in conjunction with @option{-membedded-data}.
9119 @item -msplit-addresses
9120 @itemx -mno-split-addresses
9121 @opindex msplit-addresses
9122 @opindex mno-split-addresses
9123 Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
9124 relocation operators. This option has been superceded by
9125 @option{-mexplicit-relocs} but is retained for backwards compatibility.
9127 @item -mexplicit-relocs
9128 @itemx -mno-explicit-relocs
9129 @opindex mexplicit-relocs
9130 @opindex mno-explicit-relocs
9131 Use (do not use) assembler relocation operators when dealing with symbolic
9132 addresses. The alternative, selected by @option{-mno-explicit-relocs},
9133 is to use assembler macros instead.
9135 @option{-mexplicit-relocs} is usually the default if GCC was configured
9136 to use an assembler that supports relocation operators. However, the
9137 combination of @option{-mabicalls} and @option{-fno-unit-at-a-time}
9138 implies @option{-mno-explicit-relocs} unless explicitly overridden.
9139 This is because, when generating abicalls, the choice of relocation
9140 depends on whether a symbol is local or global. In some rare cases,
9141 GCC will not be able to decide this until the whole compilation unit
9148 Generate (do not generate) code that refers to registers using their
9149 software names. The default is @option{-mno-rnames}, which tells GCC
9150 to use hardware names like @samp{$4} instead of software names like
9151 @samp{a0}. The only assembler known to support @option{-rnames} is
9152 the Algorithmics assembler.
9154 @item -mcheck-zero-division
9155 @itemx -mno-check-zero-division
9156 @opindex mcheck-zero-division
9157 @opindex mno-check-zero-division
9158 Trap (do not trap) on integer division by zero. The default is
9159 @option{-mcheck-zero-division}.
9165 Force (do not force) the use of @code{memcpy()} for non-trivial block
9166 moves. The default is @option{-mno-memcpy}, which allows GCC to inline
9167 most constant-sized copies.
9170 @itemx -mno-long-calls
9171 @opindex mlong-calls
9172 @opindex mno-long-calls
9173 Disable (do not disable) use of the @code{jal} instruction. Calling
9174 functions using @code{jal} is more efficient but requires the caller
9175 and callee to be in the same 256 megabyte segment.
9177 This option has no effect on abicalls code. The default is
9178 @option{-mno-long-calls}.
9184 Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
9185 instructions, as provided by the R4650 ISA.
9188 @itemx -mno-fused-madd
9189 @opindex mfused-madd
9190 @opindex mno-fused-madd
9191 Enable (disable) use of the floating point multiply-accumulate
9192 instructions, when they are available. The default is
9193 @option{-mfused-madd}.
9195 When multiply-accumulate instructions are used, the intermediate
9196 product is calculated to infinite precision and is not subject to
9197 the FCSR Flush to Zero bit. This may be undesirable in some
9202 Tell the MIPS assembler to not run its preprocessor over user
9203 assembler files (with a @samp{.s} suffix) when assembling them.
9206 @itemx -mno-fix-r4000
9208 @opindex mno-fix-r4000
9209 Work around certain R4000 CPU errata:
9212 A double-word or a variable shift may give an incorrect result if executed
9213 immediately after starting an integer division.
9215 A double-word or a variable shift may give an incorrect result if executed
9216 while an integer multiplication is in progress.
9218 An integer division may give an incorrect result if started in a delay slot
9219 of a taken branch or a jump.
9223 @itemx -mno-fix-r4400
9225 @opindex mno-fix-r4400
9226 Work around certain R4400 CPU errata:
9229 A double-word or a variable shift may give an incorrect result if executed
9230 immediately after starting an integer division.
9234 @itemx -mno-fix-vr4120
9235 @opindex mfix-vr4120
9236 Work around certain VR4120 errata:
9239 @code{dmultu} does not always produce the correct result.
9241 @code{div} and @code{ddiv} do not always produce the correct result if one
9242 of the operands is negative.
9244 The workarounds for the division errata rely on special functions in
9245 @file{libgcc.a}. At present, these functions are only provided by
9246 the @code{mips64vr*-elf} configurations.
9248 Other VR4120 errata require a nop to be inserted between certain pairs of
9249 instructions. These errata are handled by the assembler, not by GCC itself.
9254 Work around certain SB-1 CPU core errata.
9255 (This flag currently works around the SB-1 revision 2
9256 ``F1'' and ``F2'' floating point errata.)
9258 @item -mflush-func=@var{func}
9259 @itemx -mno-flush-func
9260 @opindex mflush-func
9261 Specifies the function to call to flush the I and D caches, or to not
9262 call any such function. If called, the function must take the same
9263 arguments as the common @code{_flush_func()}, that is, the address of the
9264 memory range for which the cache is being flushed, the size of the
9265 memory range, and the number 3 (to flush both caches). The default
9266 depends on the target GCC was configured for, but commonly is either
9267 @samp{_flush_func} or @samp{__cpu_flush}.
9269 @item -mbranch-likely
9270 @itemx -mno-branch-likely
9271 @opindex mbranch-likely
9272 @opindex mno-branch-likely
9273 Enable or disable use of Branch Likely instructions, regardless of the
9274 default for the selected architecture. By default, Branch Likely
9275 instructions may be generated if they are supported by the selected
9276 architecture. An exception is for the MIPS32 and MIPS64 architectures
9277 and processors which implement those architectures; for those, Branch
9278 Likely instructions will not be generated by default because the MIPS32
9279 and MIPS64 architectures specifically deprecate their use.
9281 @item -mfp-exceptions
9282 @itemx -mno-fp-exceptions
9283 @opindex mfp-exceptions
9284 Specifies whether FP exceptions are enabled. This affects how we schedule
9285 FP instructions for some processors. The default is that FP exceptions are
9288 For instance, on the SB-1, if FP exceptions are disabled, and we are emitting
9289 64-bit code, then we can use both FP pipes. Otherwise, we can only use one
9292 @item -mvr4130-align
9293 @itemx -mno-vr4130-align
9294 @opindex mvr4130-align
9295 The VR4130 pipeline is two-way superscalar, but can only issue two
9296 instructions together if the first one is 8-byte aligned. When this
9297 option is enabled, GCC will align pairs of instructions that it
9298 thinks should execute in parallel.
9300 This option only has an effect when optimizing for the VR4130.
9301 It normally makes code faster, but at the expense of making it bigger.
9302 It is enabled by default at optimization level @option{-O3}.
9306 @subsection MMIX Options
9307 @cindex MMIX Options
9309 These options are defined for the MMIX:
9313 @itemx -mno-libfuncs
9315 @opindex mno-libfuncs
9316 Specify that intrinsic library functions are being compiled, passing all
9317 values in registers, no matter the size.
9322 @opindex mno-epsilon
9323 Generate floating-point comparison instructions that compare with respect
9324 to the @code{rE} epsilon register.
9326 @item -mabi=mmixware
9328 @opindex mabi-mmixware
9330 Generate code that passes function parameters and return values that (in
9331 the called function) are seen as registers @code{$0} and up, as opposed to
9332 the GNU ABI which uses global registers @code{$231} and up.
9335 @itemx -mno-zero-extend
9336 @opindex mzero-extend
9337 @opindex mno-zero-extend
9338 When reading data from memory in sizes shorter than 64 bits, use (do not
9339 use) zero-extending load instructions by default, rather than
9340 sign-extending ones.
9343 @itemx -mno-knuthdiv
9345 @opindex mno-knuthdiv
9346 Make the result of a division yielding a remainder have the same sign as
9347 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
9348 remainder follows the sign of the dividend. Both methods are
9349 arithmetically valid, the latter being almost exclusively used.
9351 @item -mtoplevel-symbols
9352 @itemx -mno-toplevel-symbols
9353 @opindex mtoplevel-symbols
9354 @opindex mno-toplevel-symbols
9355 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
9356 code can be used with the @code{PREFIX} assembly directive.
9360 Generate an executable in the ELF format, rather than the default
9361 @samp{mmo} format used by the @command{mmix} simulator.
9363 @item -mbranch-predict
9364 @itemx -mno-branch-predict
9365 @opindex mbranch-predict
9366 @opindex mno-branch-predict
9367 Use (do not use) the probable-branch instructions, when static branch
9368 prediction indicates a probable branch.
9370 @item -mbase-addresses
9371 @itemx -mno-base-addresses
9372 @opindex mbase-addresses
9373 @opindex mno-base-addresses
9374 Generate (do not generate) code that uses @emph{base addresses}. Using a
9375 base address automatically generates a request (handled by the assembler
9376 and the linker) for a constant to be set up in a global register. The
9377 register is used for one or more base address requests within the range 0
9378 to 255 from the value held in the register. The generally leads to short
9379 and fast code, but the number of different data items that can be
9380 addressed is limited. This means that a program that uses lots of static
9381 data may require @option{-mno-base-addresses}.
9384 @itemx -mno-single-exit
9385 @opindex msingle-exit
9386 @opindex mno-single-exit
9387 Force (do not force) generated code to have a single exit point in each
9391 @node MN10300 Options
9392 @subsection MN10300 Options
9393 @cindex MN10300 options
9395 These @option{-m} options are defined for Matsushita MN10300 architectures:
9400 Generate code to avoid bugs in the multiply instructions for the MN10300
9401 processors. This is the default.
9404 @opindex mno-mult-bug
9405 Do not generate code to avoid bugs in the multiply instructions for the
9410 Generate code which uses features specific to the AM33 processor.
9414 Do not generate code which uses features specific to the AM33 processor. This
9419 Do not link in the C run-time initialization object file.
9423 Indicate to the linker that it should perform a relaxation optimization pass
9424 to shorten branches, calls and absolute memory addresses. This option only
9425 has an effect when used on the command line for the final link step.
9427 This option makes symbolic debugging impossible.
9431 @subsection NS32K Options
9432 @cindex NS32K options
9434 These are the @samp{-m} options defined for the 32000 series. The default
9435 values for these options depends on which style of 32000 was selected when
9436 the compiler was configured; the defaults for the most common choices are
9444 Generate output for a 32032. This is the default
9445 when the compiler is configured for 32032 and 32016 based systems.
9451 Generate output for a 32332. This is the default
9452 when the compiler is configured for 32332-based systems.
9458 Generate output for a 32532. This is the default
9459 when the compiler is configured for 32532-based systems.
9463 Generate output containing 32081 instructions for floating point.
9464 This is the default for all systems.
9468 Generate output containing 32381 instructions for floating point. This
9469 also implies @option{-m32081}. The 32381 is only compatible with the 32332
9470 and 32532 cpus. This is the default for the pc532-netbsd configuration.
9474 Try and generate multiply-add floating point instructions @code{polyF}
9475 and @code{dotF}. This option is only available if the @option{-m32381}
9476 option is in effect. Using these instructions requires changes to
9477 register allocation which generally has a negative impact on
9478 performance. This option should only be enabled when compiling code
9479 particularly likely to make heavy use of multiply-add instructions.
9482 @opindex mnomulti-add
9483 Do not try and generate multiply-add floating point instructions
9484 @code{polyF} and @code{dotF}. This is the default on all platforms.
9487 @opindex msoft-float
9488 Generate output containing library calls for floating point.
9489 @strong{Warning:} the requisite libraries may not be available.
9491 @item -mieee-compare
9492 @itemx -mno-ieee-compare
9493 @opindex mieee-compare
9494 @opindex mno-ieee-compare
9495 Control whether or not the compiler uses IEEE floating point
9496 comparisons. These handle correctly the case where the result of a
9497 comparison is unordered.
9498 @strong{Warning:} the requisite kernel support may not be available.
9501 @opindex mnobitfield
9502 Do not use the bit-field instructions. On some machines it is faster to
9503 use shifting and masking operations. This is the default for the pc532.
9507 Do use the bit-field instructions. This is the default for all platforms
9512 Use a different function-calling convention, in which functions
9513 that take a fixed number of arguments return pop their
9514 arguments on return with the @code{ret} instruction.
9516 This calling convention is incompatible with the one normally
9517 used on Unix, so you cannot use it if you need to call libraries
9518 compiled with the Unix compiler.
9520 Also, you must provide function prototypes for all functions that
9521 take variable numbers of arguments (including @code{printf});
9522 otherwise incorrect code will be generated for calls to those
9525 In addition, seriously incorrect code will result if you call a
9526 function with too many arguments. (Normally, extra arguments are
9527 harmlessly ignored.)
9529 This option takes its name from the 680x0 @code{rtd} instruction.
9534 Use a different function-calling convention where the first two arguments
9535 are passed in registers.
9537 This calling convention is incompatible with the one normally
9538 used on Unix, so you cannot use it if you need to call libraries
9539 compiled with the Unix compiler.
9542 @opindex mnoregparam
9543 Do not pass any arguments in registers. This is the default for all
9548 It is OK to use the sb as an index register which is always loaded with
9549 zero. This is the default for the pc532-netbsd target.
9553 The sb register is not available for use or has not been initialized to
9554 zero by the run time system. This is the default for all targets except
9555 the pc532-netbsd. It is also implied whenever @option{-mhimem} or
9556 @option{-fpic} is set.
9560 Many ns32000 series addressing modes use displacements of up to 512MB@.
9561 If an address is above 512MB then displacements from zero can not be used.
9562 This option causes code to be generated which can be loaded above 512MB@.
9563 This may be useful for operating systems or ROM code.
9567 Assume code will be loaded in the first 512MB of virtual address space.
9568 This is the default for all platforms.
9572 @node PDP-11 Options
9573 @subsection PDP-11 Options
9574 @cindex PDP-11 Options
9576 These options are defined for the PDP-11:
9581 Use hardware FPP floating point. This is the default. (FIS floating
9582 point on the PDP-11/40 is not supported.)
9585 @opindex msoft-float
9586 Do not use hardware floating point.
9590 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
9594 Return floating-point results in memory. This is the default.
9598 Generate code for a PDP-11/40.
9602 Generate code for a PDP-11/45. This is the default.
9606 Generate code for a PDP-11/10.
9608 @item -mbcopy-builtin
9609 @opindex bcopy-builtin
9610 Use inline @code{movstrhi} patterns for copying memory. This is the
9615 Do not use inline @code{movstrhi} patterns for copying memory.
9621 Use 16-bit @code{int}. This is the default.
9627 Use 32-bit @code{int}.
9632 @opindex mno-float32
9633 Use 64-bit @code{float}. This is the default.
9638 @opindex mno-float64
9639 Use 32-bit @code{float}.
9643 Use @code{abshi2} pattern. This is the default.
9647 Do not use @code{abshi2} pattern.
9649 @item -mbranch-expensive
9650 @opindex mbranch-expensive
9651 Pretend that branches are expensive. This is for experimenting with
9652 code generation only.
9654 @item -mbranch-cheap
9655 @opindex mbranch-cheap
9656 Do not pretend that branches are expensive. This is the default.
9660 Generate code for a system with split I&D.
9664 Generate code for a system without split I&D. This is the default.
9668 Use Unix assembler syntax. This is the default when configured for
9673 Use DEC assembler syntax. This is the default when configured for any
9674 PDP-11 target other than @samp{pdp11-*-bsd}.
9677 @node PowerPC Options
9678 @subsection PowerPC Options
9679 @cindex PowerPC options
9681 These are listed under @xref{RS/6000 and PowerPC Options}.
9683 @node RS/6000 and PowerPC Options
9684 @subsection IBM RS/6000 and PowerPC Options
9685 @cindex RS/6000 and PowerPC Options
9686 @cindex IBM RS/6000 and PowerPC Options
9688 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
9696 @itemx -mpowerpc-gpopt
9697 @itemx -mno-powerpc-gpopt
9698 @itemx -mpowerpc-gfxopt
9699 @itemx -mno-powerpc-gfxopt
9701 @itemx -mno-powerpc64
9707 @opindex mno-powerpc
9708 @opindex mpowerpc-gpopt
9709 @opindex mno-powerpc-gpopt
9710 @opindex mpowerpc-gfxopt
9711 @opindex mno-powerpc-gfxopt
9713 @opindex mno-powerpc64
9714 GCC supports two related instruction set architectures for the
9715 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
9716 instructions supported by the @samp{rios} chip set used in the original
9717 RS/6000 systems and the @dfn{PowerPC} instruction set is the
9718 architecture of the Motorola MPC5xx, MPC6xx, MPC8xx microprocessors, and
9719 the IBM 4xx microprocessors.
9721 Neither architecture is a subset of the other. However there is a
9722 large common subset of instructions supported by both. An MQ
9723 register is included in processors supporting the POWER architecture.
9725 You use these options to specify which instructions are available on the
9726 processor you are using. The default value of these options is
9727 determined when configuring GCC@. Specifying the
9728 @option{-mcpu=@var{cpu_type}} overrides the specification of these
9729 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
9730 rather than the options listed above.
9732 The @option{-mpower} option allows GCC to generate instructions that
9733 are found only in the POWER architecture and to use the MQ register.
9734 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
9735 to generate instructions that are present in the POWER2 architecture but
9736 not the original POWER architecture.
9738 The @option{-mpowerpc} option allows GCC to generate instructions that
9739 are found only in the 32-bit subset of the PowerPC architecture.
9740 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
9741 GCC to use the optional PowerPC architecture instructions in the
9742 General Purpose group, including floating-point square root. Specifying
9743 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
9744 use the optional PowerPC architecture instructions in the Graphics
9745 group, including floating-point select.
9747 The @option{-mpowerpc64} option allows GCC to generate the additional
9748 64-bit instructions that are found in the full PowerPC64 architecture
9749 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
9750 @option{-mno-powerpc64}.
9752 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
9753 will use only the instructions in the common subset of both
9754 architectures plus some special AIX common-mode calls, and will not use
9755 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
9756 permits GCC to use any instruction from either architecture and to
9757 allow use of the MQ register; specify this for the Motorola MPC601.
9759 @item -mnew-mnemonics
9760 @itemx -mold-mnemonics
9761 @opindex mnew-mnemonics
9762 @opindex mold-mnemonics
9763 Select which mnemonics to use in the generated assembler code. With
9764 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
9765 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
9766 assembler mnemonics defined for the POWER architecture. Instructions
9767 defined in only one architecture have only one mnemonic; GCC uses that
9768 mnemonic irrespective of which of these options is specified.
9770 GCC defaults to the mnemonics appropriate for the architecture in
9771 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
9772 value of these option. Unless you are building a cross-compiler, you
9773 should normally not specify either @option{-mnew-mnemonics} or
9774 @option{-mold-mnemonics}, but should instead accept the default.
9776 @item -mcpu=@var{cpu_type}
9778 Set architecture type, register usage, choice of mnemonics, and
9779 instruction scheduling parameters for machine type @var{cpu_type}.
9780 Supported values for @var{cpu_type} are @samp{401}, @samp{403},
9781 @samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{505},
9782 @samp{601}, @samp{602}, @samp{603}, @samp{603e}, @samp{604},
9783 @samp{604e}, @samp{620}, @samp{630}, @samp{740}, @samp{7400},
9784 @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
9785 @samp{860}, @samp{970}, @samp{common}, @samp{ec603e}, @samp{G3},
9786 @samp{G4}, @samp{G5}, @samp{power}, @samp{power2}, @samp{power3},
9787 @samp{power4}, @samp{power5}, @samp{powerpc}, @samp{powerpc64},
9788 @samp{rios}, @samp{rios1}, @samp{rios2}, @samp{rsc}, and @samp{rs64a}.
9790 @option{-mcpu=common} selects a completely generic processor. Code
9791 generated under this option will run on any POWER or PowerPC processor.
9792 GCC will use only the instructions in the common subset of both
9793 architectures, and will not use the MQ register. GCC assumes a generic
9794 processor model for scheduling purposes.
9796 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
9797 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
9798 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
9799 types, with an appropriate, generic processor model assumed for
9800 scheduling purposes.
9802 The other options specify a specific processor. Code generated under
9803 those options will run best on that processor, and may not run at all on
9806 The @option{-mcpu} options automatically enable or disable the
9807 following options: @option{-maltivec}, @option{-mhard-float},
9808 @option{-mmfcrf}, @option{-mmultiple}, @option{-mnew-mnemonics},
9809 @option{-mpower}, @option{-mpower2}, @option{-mpowerpc64},
9810 @option{-mpowerpc-gpopt}, @option{-mpowerpc-gfxopt},
9811 @option{-mstring}. The particular options set for any particular CPU
9812 will vary between compiler versions, depending on what setting seems
9813 to produce optimal code for that CPU; it doesn't necessarily reflect
9814 the actual hardware's capabilities. If you wish to set an individual
9815 option to a particular value, you may specify it after the
9816 @option{-mcpu} option, like @samp{-mcpu=970 -mno-altivec}.
9818 On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
9819 not enabled or disabled by the @option{-mcpu} option at present, since
9820 AIX does not have full support for these options. You may still
9821 enable or disable them individually if you're sure it'll work in your
9824 @item -mtune=@var{cpu_type}
9826 Set the instruction scheduling parameters for machine type
9827 @var{cpu_type}, but do not set the architecture type, register usage, or
9828 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
9829 values for @var{cpu_type} are used for @option{-mtune} as for
9830 @option{-mcpu}. If both are specified, the code generated will use the
9831 architecture, registers, and mnemonics set by @option{-mcpu}, but the
9832 scheduling parameters set by @option{-mtune}.
9837 @opindex mno-altivec
9838 These switches enable or disable the use of built-in functions that
9839 allow access to the AltiVec instruction set. You may also need to set
9840 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
9845 Extend the current ABI with SPE ABI extensions. This does not change
9846 the default ABI, instead it adds the SPE ABI extensions to the current
9850 @opindex mabi=no-spe
9851 Disable Booke SPE ABI extensions for the current ABI.
9853 @item -misel=@var{yes/no}
9856 This switch enables or disables the generation of ISEL instructions.
9858 @item -mspe=@var{yes/no}
9861 This switch enables or disables the generation of SPE simd
9864 @item -mfloat-gprs=@var{yes/no}
9866 @opindex mfloat-gprs
9867 This switch enables or disables the generation of floating point
9868 operations on the general purpose registers for architectures that
9869 support it. This option is currently only available on the MPC8540.
9872 @itemx -mno-fp-in-toc
9873 @itemx -mno-sum-in-toc
9874 @itemx -mminimal-toc
9876 @opindex mno-fp-in-toc
9877 @opindex mno-sum-in-toc
9878 @opindex mminimal-toc
9879 Modify generation of the TOC (Table Of Contents), which is created for
9880 every executable file. The @option{-mfull-toc} option is selected by
9881 default. In that case, GCC will allocate at least one TOC entry for
9882 each unique non-automatic variable reference in your program. GCC
9883 will also place floating-point constants in the TOC@. However, only
9884 16,384 entries are available in the TOC@.
9886 If you receive a linker error message that saying you have overflowed
9887 the available TOC space, you can reduce the amount of TOC space used
9888 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
9889 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
9890 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
9891 generate code to calculate the sum of an address and a constant at
9892 run-time instead of putting that sum into the TOC@. You may specify one
9893 or both of these options. Each causes GCC to produce very slightly
9894 slower and larger code at the expense of conserving TOC space.
9896 If you still run out of space in the TOC even when you specify both of
9897 these options, specify @option{-mminimal-toc} instead. This option causes
9898 GCC to make only one TOC entry for every file. When you specify this
9899 option, GCC will produce code that is slower and larger but which
9900 uses extremely little TOC space. You may wish to use this option
9901 only on files that contain less frequently executed code.
9907 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
9908 @code{long} type, and the infrastructure needed to support them.
9909 Specifying @option{-maix64} implies @option{-mpowerpc64} and
9910 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
9911 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
9916 @opindex mno-xl-call
9917 On AIX, pass floating-point arguments to prototyped functions beyond the
9918 register save area (RSA) on the stack in addition to argument FPRs. The
9919 AIX calling convention was extended but not initially documented to
9920 handle an obscure K&R C case of calling a function that takes the
9921 address of its arguments with fewer arguments than declared. AIX XL
9922 compilers access floating point arguments which do not fit in the
9923 RSA from the stack when a subroutine is compiled without
9924 optimization. Because always storing floating-point arguments on the
9925 stack is inefficient and rarely needed, this option is not enabled by
9926 default and only is necessary when calling subroutines compiled by AIX
9927 XL compilers without optimization.
9931 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
9932 application written to use message passing with special startup code to
9933 enable the application to run. The system must have PE installed in the
9934 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
9935 must be overridden with the @option{-specs=} option to specify the
9936 appropriate directory location. The Parallel Environment does not
9937 support threads, so the @option{-mpe} option and the @option{-pthread}
9938 option are incompatible.
9940 @item -malign-natural
9941 @itemx -malign-power
9942 @opindex malign-natural
9943 @opindex malign-power
9944 On AIX, Darwin, and 64-bit PowerPC GNU/Linux, the option
9945 @option{-malign-natural} overrides the ABI-defined alignment of larger
9946 types, such as floating-point doubles, on their natural size-based boundary.
9947 The option @option{-malign-power} instructs GCC to follow the ABI-specified
9948 alignment rules. GCC defaults to the standard alignment defined in the ABI.
9952 @opindex msoft-float
9953 @opindex mhard-float
9954 Generate code that does not use (uses) the floating-point register set.
9955 Software floating point emulation is provided if you use the
9956 @option{-msoft-float} option, and pass the option to GCC when linking.
9959 @itemx -mno-multiple
9961 @opindex mno-multiple
9962 Generate code that uses (does not use) the load multiple word
9963 instructions and the store multiple word instructions. These
9964 instructions are generated by default on POWER systems, and not
9965 generated on PowerPC systems. Do not use @option{-mmultiple} on little
9966 endian PowerPC systems, since those instructions do not work when the
9967 processor is in little endian mode. The exceptions are PPC740 and
9968 PPC750 which permit the instructions usage in little endian mode.
9974 Generate code that uses (does not use) the load string instructions
9975 and the store string word instructions to save multiple registers and
9976 do small block moves. These instructions are generated by default on
9977 POWER systems, and not generated on PowerPC systems. Do not use
9978 @option{-mstring} on little endian PowerPC systems, since those
9979 instructions do not work when the processor is in little endian mode.
9980 The exceptions are PPC740 and PPC750 which permit the instructions
9981 usage in little endian mode.
9987 Generate code that uses (does not use) the load or store instructions
9988 that update the base register to the address of the calculated memory
9989 location. These instructions are generated by default. If you use
9990 @option{-mno-update}, there is a small window between the time that the
9991 stack pointer is updated and the address of the previous frame is
9992 stored, which means code that walks the stack frame across interrupts or
9993 signals may get corrupted data.
9996 @itemx -mno-fused-madd
9997 @opindex mfused-madd
9998 @opindex mno-fused-madd
9999 Generate code that uses (does not use) the floating point multiply and
10000 accumulate instructions. These instructions are generated by default if
10001 hardware floating is used.
10003 @item -mno-bit-align
10005 @opindex mno-bit-align
10006 @opindex mbit-align
10007 On System V.4 and embedded PowerPC systems do not (do) force structures
10008 and unions that contain bit-fields to be aligned to the base type of the
10011 For example, by default a structure containing nothing but 8
10012 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
10013 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
10014 the structure would be aligned to a 1 byte boundary and be one byte in
10017 @item -mno-strict-align
10018 @itemx -mstrict-align
10019 @opindex mno-strict-align
10020 @opindex mstrict-align
10021 On System V.4 and embedded PowerPC systems do not (do) assume that
10022 unaligned memory references will be handled by the system.
10024 @item -mrelocatable
10025 @itemx -mno-relocatable
10026 @opindex mrelocatable
10027 @opindex mno-relocatable
10028 On embedded PowerPC systems generate code that allows (does not allow)
10029 the program to be relocated to a different address at runtime. If you
10030 use @option{-mrelocatable} on any module, all objects linked together must
10031 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
10033 @item -mrelocatable-lib
10034 @itemx -mno-relocatable-lib
10035 @opindex mrelocatable-lib
10036 @opindex mno-relocatable-lib
10037 On embedded PowerPC systems generate code that allows (does not allow)
10038 the program to be relocated to a different address at runtime. Modules
10039 compiled with @option{-mrelocatable-lib} can be linked with either modules
10040 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
10041 with modules compiled with the @option{-mrelocatable} options.
10047 On System V.4 and embedded PowerPC systems do not (do) assume that
10048 register 2 contains a pointer to a global area pointing to the addresses
10049 used in the program.
10052 @itemx -mlittle-endian
10054 @opindex mlittle-endian
10055 On System V.4 and embedded PowerPC systems compile code for the
10056 processor in little endian mode. The @option{-mlittle-endian} option is
10057 the same as @option{-mlittle}.
10060 @itemx -mbig-endian
10062 @opindex mbig-endian
10063 On System V.4 and embedded PowerPC systems compile code for the
10064 processor in big endian mode. The @option{-mbig-endian} option is
10065 the same as @option{-mbig}.
10067 @item -mdynamic-no-pic
10068 @opindex mdynamic-no-pic
10069 On Darwin and Mac OS X systems, compile code so that it is not
10070 relocatable, but that its external references are relocatable. The
10071 resulting code is suitable for applications, but not shared
10074 @item -mprioritize-restricted-insns=@var{priority}
10075 @opindex mprioritize-restricted-insns
10076 This option controls the priority that is assigned to
10077 dispatch-slot restricted instructions during the second scheduling
10078 pass. The argument @var{priority} takes the value @var{0/1/2} to assign
10079 @var{no/highest/second-highest} priority to dispatch slot restricted
10082 @item -msched-costly-dep=@var{dependence_type}
10083 @opindex msched-costly-dep
10084 This option controls which dependences are considered costly
10085 by the target during instruction scheduling. The argument
10086 @var{dependence_type} takes one of the following values:
10087 @var{no}: no dependence is costly,
10088 @var{all}: all dependences are costly,
10089 @var{true_store_to_load}: a true dependence from store to load is costly,
10090 @var{store_to_load}: any dependence from store to load is costly,
10091 @var{number}: any dependence which latency >= @var{number} is costly.
10093 @item -minsert-sched-nops=@var{scheme}
10094 @opindex minsert-sched-nops
10095 This option controls which nop insertion scheme will be used during
10096 the second scheduling pass. The argument @var{scheme} takes one of the
10098 @var{no}: Don't insert nops.
10099 @var{pad}: Pad with nops any dispatch group which has vacant issue slots,
10100 according to the scheduler's grouping.
10101 @var{regroup_exact}: Insert nops to force costly dependent insns into
10102 separate groups. Insert exactly as many nops as needed to force an insn
10103 to a new group, according to the estimated processor grouping.
10104 @var{number}: Insert nops to force costly dependent insns into
10105 separate groups. Insert @var{number} nops to force an insn to a new group.
10108 @opindex mcall-sysv
10109 On System V.4 and embedded PowerPC systems compile code using calling
10110 conventions that adheres to the March 1995 draft of the System V
10111 Application Binary Interface, PowerPC processor supplement. This is the
10112 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
10114 @item -mcall-sysv-eabi
10115 @opindex mcall-sysv-eabi
10116 Specify both @option{-mcall-sysv} and @option{-meabi} options.
10118 @item -mcall-sysv-noeabi
10119 @opindex mcall-sysv-noeabi
10120 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
10122 @item -mcall-solaris
10123 @opindex mcall-solaris
10124 On System V.4 and embedded PowerPC systems compile code for the Solaris
10128 @opindex mcall-linux
10129 On System V.4 and embedded PowerPC systems compile code for the
10130 Linux-based GNU system.
10134 On System V.4 and embedded PowerPC systems compile code for the
10135 Hurd-based GNU system.
10137 @item -mcall-netbsd
10138 @opindex mcall-netbsd
10139 On System V.4 and embedded PowerPC systems compile code for the
10140 NetBSD operating system.
10142 @item -maix-struct-return
10143 @opindex maix-struct-return
10144 Return all structures in memory (as specified by the AIX ABI)@.
10146 @item -msvr4-struct-return
10147 @opindex msvr4-struct-return
10148 Return structures smaller than 8 bytes in registers (as specified by the
10151 @item -mabi=altivec
10152 @opindex mabi=altivec
10153 Extend the current ABI with AltiVec ABI extensions. This does not
10154 change the default ABI, instead it adds the AltiVec ABI extensions to
10157 @item -mabi=no-altivec
10158 @opindex mabi=no-altivec
10159 Disable AltiVec ABI extensions for the current ABI.
10162 @itemx -mno-prototype
10163 @opindex mprototype
10164 @opindex mno-prototype
10165 On System V.4 and embedded PowerPC systems assume that all calls to
10166 variable argument functions are properly prototyped. Otherwise, the
10167 compiler must insert an instruction before every non prototyped call to
10168 set or clear bit 6 of the condition code register (@var{CR}) to
10169 indicate whether floating point values were passed in the floating point
10170 registers in case the function takes a variable arguments. With
10171 @option{-mprototype}, only calls to prototyped variable argument functions
10172 will set or clear the bit.
10176 On embedded PowerPC systems, assume that the startup module is called
10177 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
10178 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
10183 On embedded PowerPC systems, assume that the startup module is called
10184 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
10189 On embedded PowerPC systems, assume that the startup module is called
10190 @file{crt0.o} and the standard C libraries are @file{libads.a} and
10193 @item -myellowknife
10194 @opindex myellowknife
10195 On embedded PowerPC systems, assume that the startup module is called
10196 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
10201 On System V.4 and embedded PowerPC systems, specify that you are
10202 compiling for a VxWorks system.
10206 Specify that you are compiling for the WindISS simulation environment.
10210 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
10211 header to indicate that @samp{eabi} extended relocations are used.
10217 On System V.4 and embedded PowerPC systems do (do not) adhere to the
10218 Embedded Applications Binary Interface (eabi) which is a set of
10219 modifications to the System V.4 specifications. Selecting @option{-meabi}
10220 means that the stack is aligned to an 8 byte boundary, a function
10221 @code{__eabi} is called to from @code{main} to set up the eabi
10222 environment, and the @option{-msdata} option can use both @code{r2} and
10223 @code{r13} to point to two separate small data areas. Selecting
10224 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
10225 do not call an initialization function from @code{main}, and the
10226 @option{-msdata} option will only use @code{r13} to point to a single
10227 small data area. The @option{-meabi} option is on by default if you
10228 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
10231 @opindex msdata=eabi
10232 On System V.4 and embedded PowerPC systems, put small initialized
10233 @code{const} global and static data in the @samp{.sdata2} section, which
10234 is pointed to by register @code{r2}. Put small initialized
10235 non-@code{const} global and static data in the @samp{.sdata} section,
10236 which is pointed to by register @code{r13}. Put small uninitialized
10237 global and static data in the @samp{.sbss} section, which is adjacent to
10238 the @samp{.sdata} section. The @option{-msdata=eabi} option is
10239 incompatible with the @option{-mrelocatable} option. The
10240 @option{-msdata=eabi} option also sets the @option{-memb} option.
10243 @opindex msdata=sysv
10244 On System V.4 and embedded PowerPC systems, put small global and static
10245 data in the @samp{.sdata} section, which is pointed to by register
10246 @code{r13}. Put small uninitialized global and static data in the
10247 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
10248 The @option{-msdata=sysv} option is incompatible with the
10249 @option{-mrelocatable} option.
10251 @item -msdata=default
10253 @opindex msdata=default
10255 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
10256 compile code the same as @option{-msdata=eabi}, otherwise compile code the
10257 same as @option{-msdata=sysv}.
10260 @opindex msdata-data
10261 On System V.4 and embedded PowerPC systems, put small global and static
10262 data in the @samp{.sdata} section. Put small uninitialized global and
10263 static data in the @samp{.sbss} section. Do not use register @code{r13}
10264 to address small data however. This is the default behavior unless
10265 other @option{-msdata} options are used.
10269 @opindex msdata=none
10271 On embedded PowerPC systems, put all initialized global and static data
10272 in the @samp{.data} section, and all uninitialized data in the
10273 @samp{.bss} section.
10277 @cindex smaller data references (PowerPC)
10278 @cindex .sdata/.sdata2 references (PowerPC)
10279 On embedded PowerPC systems, put global and static items less than or
10280 equal to @var{num} bytes into the small data or bss sections instead of
10281 the normal data or bss section. By default, @var{num} is 8. The
10282 @option{-G @var{num}} switch is also passed to the linker.
10283 All modules should be compiled with the same @option{-G @var{num}} value.
10286 @itemx -mno-regnames
10288 @opindex mno-regnames
10289 On System V.4 and embedded PowerPC systems do (do not) emit register
10290 names in the assembly language output using symbolic forms.
10293 @itemx -mno-longcall
10295 @opindex mno-longcall
10296 Default to making all function calls indirectly, using a register, so
10297 that functions which reside further than 32 megabytes (33,554,432
10298 bytes) from the current location can be called. This setting can be
10299 overridden by the @code{shortcall} function attribute, or by
10300 @code{#pragma longcall(0)}.
10302 Some linkers are capable of detecting out-of-range calls and generating
10303 glue code on the fly. On these systems, long calls are unnecessary and
10304 generate slower code. As of this writing, the AIX linker can do this,
10305 as can the GNU linker for PowerPC/64. It is planned to add this feature
10306 to the GNU linker for 32-bit PowerPC systems as well.
10308 On Darwin/PPC systems, @code{#pragma longcall} will generate ``jbsr
10309 callee, L42'', plus a ``branch island'' (glue code). The two target
10310 addresses represent the callee and the ``branch island.'' The
10311 Darwin/PPC linker will prefer the first address and generate a ``bl
10312 callee'' if the PPC ``bl'' instruction will reach the callee directly;
10313 otherwise, the linker will generate ``bl L42'' to call the ``branch
10314 island.'' The ``branch island'' is appended to the body of the
10315 calling function; it computes the full 32-bit address of the callee
10318 On Mach-O (Darwin) systems, this option directs the compiler emit to
10319 the glue for every direct call, and the Darwin linker decides whether
10320 to use or discard it.
10322 In the future, we may cause GCC to ignore all longcall specifications
10323 when the linker is known to generate glue.
10327 Adds support for multithreading with the @dfn{pthreads} library.
10328 This option sets flags for both the preprocessor and linker.
10332 @node S/390 and zSeries Options
10333 @subsection S/390 and zSeries Options
10334 @cindex S/390 and zSeries Options
10336 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
10340 @itemx -msoft-float
10341 @opindex mhard-float
10342 @opindex msoft-float
10343 Use (do not use) the hardware floating-point instructions and registers
10344 for floating-point operations. When @option{-msoft-float} is specified,
10345 functions in @file{libgcc.a} will be used to perform floating-point
10346 operations. When @option{-mhard-float} is specified, the compiler
10347 generates IEEE floating-point instructions. This is the default.
10350 @itemx -mno-backchain
10351 @opindex mbackchain
10352 @opindex mno-backchain
10353 Generate (or do not generate) code which maintains an explicit
10354 backchain within the stack frame that points to the caller's frame.
10355 This may be needed to allow debugging using tools that do not understand
10356 DWARF-2 call frame information. The default is not to generate the
10360 @itemx -mno-small-exec
10361 @opindex msmall-exec
10362 @opindex mno-small-exec
10363 Generate (or do not generate) code using the @code{bras} instruction
10364 to do subroutine calls.
10365 This only works reliably if the total executable size does not
10366 exceed 64k. The default is to use the @code{basr} instruction instead,
10367 which does not have this limitation.
10373 When @option{-m31} is specified, generate code compliant to the
10374 GNU/Linux for S/390 ABI@. When @option{-m64} is specified, generate
10375 code compliant to the GNU/Linux for zSeries ABI@. This allows GCC in
10376 particular to generate 64-bit instructions. For the @samp{s390}
10377 targets, the default is @option{-m31}, while the @samp{s390x}
10378 targets default to @option{-m64}.
10384 When @option{-mzarch} is specified, generate code using the
10385 instructions available on z/Architecture.
10386 When @option{-mesa} is specified, generate code using the
10387 instructions available on ESA/390. Note that @option{-mesa} is
10388 not possible with @option{-m64}.
10389 When generating code compliant to the GNU/Linux for S/390 ABI,
10390 the default is @option{-mesa}. When generating code compliant
10391 to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
10397 Generate (or do not generate) code using the @code{mvcle} instruction
10398 to perform block moves. When @option{-mno-mvcle} is specified,
10399 use a @code{mvc} loop instead. This is the default.
10405 Print (or do not print) additional debug information when compiling.
10406 The default is to not print debug information.
10408 @item -march=@var{cpu-type}
10410 Generate code that will run on @var{cpu-type}, which is the name of a system
10411 representing a certain processor type. Possible values for
10412 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, and @samp{z990}.
10413 When generating code using the instructions available on z/Architecture,
10414 the default is @option{-march=z900}. Otherwise, the default is
10415 @option{-march=g5}.
10417 @item -mtune=@var{cpu-type}
10419 Tune to @var{cpu-type} everything applicable about the generated code,
10420 except for the ABI and the set of available instructions.
10421 The list of @var{cpu-type} values is the same as for @option{-march}.
10422 The default is the value used for @option{-march}.
10425 @itemx -mno-tpf-trace
10426 @opindex mtpf-trace
10427 @opindex mno-tpf-trace
10428 Generate code that adds (does not add) in TPF OS specific branches to trace
10429 routines in the operating system. This option is off by default, even
10430 when compiling for the TPF OS.
10433 @itemx -mno-fused-madd
10434 @opindex mfused-madd
10435 @opindex mno-fused-madd
10436 Generate code that uses (does not use) the floating point multiply and
10437 accumulate instructions. These instructions are generated by default if
10438 hardware floating point is used.
10442 @subsection SH Options
10444 These @samp{-m} options are defined for the SH implementations:
10449 Generate code for the SH1.
10453 Generate code for the SH2.
10456 Generate code for the SH2e.
10460 Generate code for the SH3.
10464 Generate code for the SH3e.
10468 Generate code for the SH4 without a floating-point unit.
10470 @item -m4-single-only
10471 @opindex m4-single-only
10472 Generate code for the SH4 with a floating-point unit that only
10473 supports single-precision arithmetic.
10477 Generate code for the SH4 assuming the floating-point unit is in
10478 single-precision mode by default.
10482 Generate code for the SH4.
10486 Compile code for the processor in big endian mode.
10490 Compile code for the processor in little endian mode.
10494 Align doubles at 64-bit boundaries. Note that this changes the calling
10495 conventions, and thus some functions from the standard C library will
10496 not work unless you recompile it first with @option{-mdalign}.
10500 Shorten some address references at link time, when possible; uses the
10501 linker option @option{-relax}.
10505 Use 32-bit offsets in @code{switch} tables. The default is to use
10510 Enable the use of the instruction @code{fmovd}.
10514 Comply with the calling conventions defined by Renesas.
10517 @opindex mnomacsave
10518 Mark the @code{MAC} register as call-clobbered, even if
10519 @option{-mhitachi} is given.
10523 Increase IEEE-compliance of floating-point code.
10527 Dump instruction size and location in the assembly code.
10530 @opindex mpadstruct
10531 This option is deprecated. It pads structures to multiple of 4 bytes,
10532 which is incompatible with the SH ABI@.
10536 Optimize for space instead of speed. Implied by @option{-Os}.
10539 @opindex mprefergot
10540 When generating position-independent code, emit function calls using
10541 the Global Offset Table instead of the Procedure Linkage Table.
10545 Generate a library function call to invalidate instruction cache
10546 entries, after fixing up a trampoline. This library function call
10547 doesn't assume it can write to the whole memory address space. This
10548 is the default when the target is @code{sh-*-linux*}.
10551 @node SPARC Options
10552 @subsection SPARC Options
10553 @cindex SPARC options
10555 These @samp{-m} options are supported on the SPARC:
10558 @item -mno-app-regs
10560 @opindex mno-app-regs
10562 Specify @option{-mapp-regs} to generate output using the global registers
10563 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
10566 To be fully SVR4 ABI compliant at the cost of some performance loss,
10567 specify @option{-mno-app-regs}. You should compile libraries and system
10568 software with this option.
10571 @itemx -mhard-float
10573 @opindex mhard-float
10574 Generate output containing floating point instructions. This is the
10578 @itemx -msoft-float
10580 @opindex msoft-float
10581 Generate output containing library calls for floating point.
10582 @strong{Warning:} the requisite libraries are not available for all SPARC
10583 targets. Normally the facilities of the machine's usual C compiler are
10584 used, but this cannot be done directly in cross-compilation. You must make
10585 your own arrangements to provide suitable library functions for
10586 cross-compilation. The embedded targets @samp{sparc-*-aout} and
10587 @samp{sparclite-*-*} do provide software floating point support.
10589 @option{-msoft-float} changes the calling convention in the output file;
10590 therefore, it is only useful if you compile @emph{all} of a program with
10591 this option. In particular, you need to compile @file{libgcc.a}, the
10592 library that comes with GCC, with @option{-msoft-float} in order for
10595 @item -mhard-quad-float
10596 @opindex mhard-quad-float
10597 Generate output containing quad-word (long double) floating point
10600 @item -msoft-quad-float
10601 @opindex msoft-quad-float
10602 Generate output containing library calls for quad-word (long double)
10603 floating point instructions. The functions called are those specified
10604 in the SPARC ABI@. This is the default.
10606 As of this writing, there are no SPARC implementations that have hardware
10607 support for the quad-word floating point instructions. They all invoke
10608 a trap handler for one of these instructions, and then the trap handler
10609 emulates the effect of the instruction. Because of the trap handler overhead,
10610 this is much slower than calling the ABI library routines. Thus the
10611 @option{-msoft-quad-float} option is the default.
10613 @item -mno-unaligned-doubles
10614 @itemx -munaligned-doubles
10615 @opindex mno-unaligned-doubles
10616 @opindex munaligned-doubles
10617 Assume that doubles have 8 byte alignment. This is the default.
10619 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
10620 alignment only if they are contained in another type, or if they have an
10621 absolute address. Otherwise, it assumes they have 4 byte alignment.
10622 Specifying this option avoids some rare compatibility problems with code
10623 generated by other compilers. It is not the default because it results
10624 in a performance loss, especially for floating point code.
10626 @item -mno-faster-structs
10627 @itemx -mfaster-structs
10628 @opindex mno-faster-structs
10629 @opindex mfaster-structs
10630 With @option{-mfaster-structs}, the compiler assumes that structures
10631 should have 8 byte alignment. This enables the use of pairs of
10632 @code{ldd} and @code{std} instructions for copies in structure
10633 assignment, in place of twice as many @code{ld} and @code{st} pairs.
10634 However, the use of this changed alignment directly violates the SPARC
10635 ABI@. Thus, it's intended only for use on targets where the developer
10636 acknowledges that their resulting code will not be directly in line with
10637 the rules of the ABI@.
10639 @item -mimpure-text
10640 @opindex mimpure-text
10641 @option{-mimpure-text}, used in addition to @option{-shared}, tells
10642 the compiler to not pass @option{-z text} to the linker when linking a
10643 shared object. Using this option, you can link position-dependent
10644 code into a shared object.
10646 @option{-mimpure-text} suppresses the ``relocations remain against
10647 allocatable but non-writable sections'' linker error message.
10648 However, the necessary relocations will trigger copy-on-write, and the
10649 shared object is not actually shared across processes. Instead of
10650 using @option{-mimpure-text}, you should compile all source code with
10651 @option{-fpic} or @option{-fPIC}.
10653 This option is only available on SunOS and Solaris.
10655 @item -mcpu=@var{cpu_type}
10657 Set the instruction set, register set, and instruction scheduling parameters
10658 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
10659 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
10660 @samp{f930}, @samp{f934}, @samp{hypersparc}, @samp{sparclite86x},
10661 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc}, and
10662 @samp{ultrasparc3}.
10664 Default instruction scheduling parameters are used for values that select
10665 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
10666 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
10668 Here is a list of each supported architecture and their supported
10673 v8: supersparc, hypersparc
10674 sparclite: f930, f934, sparclite86x
10676 v9: ultrasparc, ultrasparc3
10679 By default (unless configured otherwise), GCC generates code for the V7
10680 variant of the SPARC architecture. With @option{-mcpu=cypress}, the compiler
10681 additionally optimizes it for the Cypress CY7C602 chip, as used in the
10682 SPARCStation/SPARCServer 3xx series. This is also appropriate for the older
10683 SPARCStation 1, 2, IPX etc.
10685 With @option{-mcpu=v8}, GCC generates code for the V8 variant of the SPARC
10686 architecture. The only difference from V7 code is that the compiler emits
10687 the integer multiply and integer divide instructions which exist in SPARC-V8
10688 but not in SPARC-V7. With @option{-mcpu=supersparc}, the compiler additionally
10689 optimizes it for the SuperSPARC chip, as used in the SPARCStation 10, 1000 and
10692 With @option{-mcpu=sparclite}, GCC generates code for the SPARClite variant of
10693 the SPARC architecture. This adds the integer multiply, integer divide step
10694 and scan (@code{ffs}) instructions which exist in SPARClite but not in SPARC-V7.
10695 With @option{-mcpu=f930}, the compiler additionally optimizes it for the
10696 Fujitsu MB86930 chip, which is the original SPARClite, with no FPU. With
10697 @option{-mcpu=f934}, the compiler additionally optimizes it for the Fujitsu
10698 MB86934 chip, which is the more recent SPARClite with FPU.
10700 With @option{-mcpu=sparclet}, GCC generates code for the SPARClet variant of
10701 the SPARC architecture. This adds the integer multiply, multiply/accumulate,
10702 integer divide step and scan (@code{ffs}) instructions which exist in SPARClet
10703 but not in SPARC-V7. With @option{-mcpu=tsc701}, the compiler additionally
10704 optimizes it for the TEMIC SPARClet chip.
10706 With @option{-mcpu=v9}, GCC generates code for the V9 variant of the SPARC
10707 architecture. This adds 64-bit integer and floating-point move instructions,
10708 3 additional floating-point condition code registers and conditional move
10709 instructions. With @option{-mcpu=ultrasparc}, the compiler additionally
10710 optimizes it for the Sun UltraSPARC I/II chips. With
10711 @option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
10712 Sun UltraSPARC III chip.
10714 @item -mtune=@var{cpu_type}
10716 Set the instruction scheduling parameters for machine type
10717 @var{cpu_type}, but do not set the instruction set or register set that the
10718 option @option{-mcpu=@var{cpu_type}} would.
10720 The same values for @option{-mcpu=@var{cpu_type}} can be used for
10721 @option{-mtune=@var{cpu_type}}, but the only useful values are those
10722 that select a particular cpu implementation. Those are @samp{cypress},
10723 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
10724 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc}, and
10725 @samp{ultrasparc3}.
10730 @opindex mno-v8plus
10731 With @option{-mv8plus}, GCC generates code for the SPARC-V8+ ABI. The
10732 difference from the V8 ABI is that the global and out registers are
10733 considered 64-bit wide. This is enabled by default on Solaris in 32-bit
10734 mode for all SPARC-V9 processors.
10740 With @option{-mvis}, GCC generates code that takes advantage of the UltraSPARC
10741 Visual Instruction Set extensions. The default is @option{-mno-vis}.
10744 These @samp{-m} options are supported in addition to the above
10745 on SPARC-V9 processors in 64-bit environments:
10748 @item -mlittle-endian
10749 @opindex mlittle-endian
10750 Generate code for a processor running in little-endian mode. It is only
10751 available for a few configurations and most notably not on Solaris.
10757 Generate code for a 32-bit or 64-bit environment.
10758 The 32-bit environment sets int, long and pointer to 32 bits.
10759 The 64-bit environment sets int to 32 bits and long and pointer
10762 @item -mcmodel=medlow
10763 @opindex mcmodel=medlow
10764 Generate code for the Medium/Low code model: 64-bit addresses, programs
10765 must be linked in the low 32 bits of memory. Programs can be statically
10766 or dynamically linked.
10768 @item -mcmodel=medmid
10769 @opindex mcmodel=medmid
10770 Generate code for the Medium/Middle code model: 64-bit addresses, programs
10771 must be linked in the low 44 bits of memory, the text and data segments must
10772 be less than 2GB in size and the data segment must be located within 2GB of
10775 @item -mcmodel=medany
10776 @opindex mcmodel=medany
10777 Generate code for the Medium/Anywhere code model: 64-bit addresses, programs
10778 may be linked anywhere in memory, the text and data segments must be less
10779 than 2GB in size and the data segment must be located within 2GB of the
10782 @item -mcmodel=embmedany
10783 @opindex mcmodel=embmedany
10784 Generate code for the Medium/Anywhere code model for embedded systems:
10785 64-bit addresses, the text and data segments must be less than 2GB in
10786 size, both starting anywhere in memory (determined at link time). The
10787 global register %g4 points to the base of the data segment. Programs
10788 are statically linked and PIC is not supported.
10791 @itemx -mno-stack-bias
10792 @opindex mstack-bias
10793 @opindex mno-stack-bias
10794 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
10795 frame pointer if present, are offset by @minus{}2047 which must be added back
10796 when making stack frame references. This is the default in 64-bit mode.
10797 Otherwise, assume no such offset is present.
10800 @node System V Options
10801 @subsection Options for System V
10803 These additional options are available on System V Release 4 for
10804 compatibility with other compilers on those systems:
10809 Create a shared object.
10810 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
10814 Identify the versions of each tool used by the compiler, in a
10815 @code{.ident} assembler directive in the output.
10819 Refrain from adding @code{.ident} directives to the output file (this is
10822 @item -YP,@var{dirs}
10824 Search the directories @var{dirs}, and no others, for libraries
10825 specified with @option{-l}.
10827 @item -Ym,@var{dir}
10829 Look in the directory @var{dir} to find the M4 preprocessor.
10830 The assembler uses this option.
10831 @c This is supposed to go with a -Yd for predefined M4 macro files, but
10832 @c the generic assembler that comes with Solaris takes just -Ym.
10835 @node TMS320C3x/C4x Options
10836 @subsection TMS320C3x/C4x Options
10837 @cindex TMS320C3x/C4x Options
10839 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
10843 @item -mcpu=@var{cpu_type}
10845 Set the instruction set, register set, and instruction scheduling
10846 parameters for machine type @var{cpu_type}. Supported values for
10847 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
10848 @samp{c44}. The default is @samp{c40} to generate code for the
10853 @itemx -msmall-memory
10855 @opindex mbig-memory
10857 @opindex msmall-memory
10859 Generates code for the big or small memory model. The small memory
10860 model assumed that all data fits into one 64K word page. At run-time
10861 the data page (DP) register must be set to point to the 64K page
10862 containing the .bss and .data program sections. The big memory model is
10863 the default and requires reloading of the DP register for every direct
10870 Allow (disallow) allocation of general integer operands into the block
10871 count register BK@.
10877 Enable (disable) generation of code using decrement and branch,
10878 DBcond(D), instructions. This is enabled by default for the C4x. To be
10879 on the safe side, this is disabled for the C3x, since the maximum
10880 iteration count on the C3x is @math{2^{23} + 1} (but who iterates loops more than
10881 @math{2^{23}} times on the C3x?). Note that GCC will try to reverse a loop so
10882 that it can utilize the decrement and branch instruction, but will give
10883 up if there is more than one memory reference in the loop. Thus a loop
10884 where the loop counter is decremented can generate slightly more
10885 efficient code, in cases where the RPTB instruction cannot be utilized.
10887 @item -mdp-isr-reload
10889 @opindex mdp-isr-reload
10891 Force the DP register to be saved on entry to an interrupt service
10892 routine (ISR), reloaded to point to the data section, and restored on
10893 exit from the ISR@. This should not be required unless someone has
10894 violated the small memory model by modifying the DP register, say within
10901 For the C3x use the 24-bit MPYI instruction for integer multiplies
10902 instead of a library call to guarantee 32-bit results. Note that if one
10903 of the operands is a constant, then the multiplication will be performed
10904 using shifts and adds. If the @option{-mmpyi} option is not specified for the C3x,
10905 then squaring operations are performed inline instead of a library call.
10908 @itemx -mno-fast-fix
10910 @opindex mno-fast-fix
10911 The C3x/C4x FIX instruction to convert a floating point value to an
10912 integer value chooses the nearest integer less than or equal to the
10913 floating point value rather than to the nearest integer. Thus if the
10914 floating point number is negative, the result will be incorrectly
10915 truncated an additional code is necessary to detect and correct this
10916 case. This option can be used to disable generation of the additional
10917 code required to correct the result.
10923 Enable (disable) generation of repeat block sequences using the RPTB
10924 instruction for zero overhead looping. The RPTB construct is only used
10925 for innermost loops that do not call functions or jump across the loop
10926 boundaries. There is no advantage having nested RPTB loops due to the
10927 overhead required to save and restore the RC, RS, and RE registers.
10928 This is enabled by default with @option{-O2}.
10930 @item -mrpts=@var{count}
10934 Enable (disable) the use of the single instruction repeat instruction
10935 RPTS@. If a repeat block contains a single instruction, and the loop
10936 count can be guaranteed to be less than the value @var{count}, GCC will
10937 emit a RPTS instruction instead of a RPTB@. If no value is specified,
10938 then a RPTS will be emitted even if the loop count cannot be determined
10939 at compile time. Note that the repeated instruction following RPTS does
10940 not have to be reloaded from memory each iteration, thus freeing up the
10941 CPU buses for operands. However, since interrupts are blocked by this
10942 instruction, it is disabled by default.
10944 @item -mloop-unsigned
10945 @itemx -mno-loop-unsigned
10946 @opindex mloop-unsigned
10947 @opindex mno-loop-unsigned
10948 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
10949 is @math{2^{31} + 1} since these instructions test if the iteration count is
10950 negative to terminate the loop. If the iteration count is unsigned
10951 there is a possibility than the @math{2^{31} + 1} maximum iteration count may be
10952 exceeded. This switch allows an unsigned iteration count.
10956 Try to emit an assembler syntax that the TI assembler (asm30) is happy
10957 with. This also enforces compatibility with the API employed by the TI
10958 C3x C compiler. For example, long doubles are passed as structures
10959 rather than in floating point registers.
10965 Generate code that uses registers (stack) for passing arguments to functions.
10966 By default, arguments are passed in registers where possible rather
10967 than by pushing arguments on to the stack.
10969 @item -mparallel-insns
10970 @itemx -mno-parallel-insns
10971 @opindex mparallel-insns
10972 @opindex mno-parallel-insns
10973 Allow the generation of parallel instructions. This is enabled by
10974 default with @option{-O2}.
10976 @item -mparallel-mpy
10977 @itemx -mno-parallel-mpy
10978 @opindex mparallel-mpy
10979 @opindex mno-parallel-mpy
10980 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
10981 provided @option{-mparallel-insns} is also specified. These instructions have
10982 tight register constraints which can pessimize the code generation
10983 of large functions.
10988 @subsection V850 Options
10989 @cindex V850 Options
10991 These @samp{-m} options are defined for V850 implementations:
10995 @itemx -mno-long-calls
10996 @opindex mlong-calls
10997 @opindex mno-long-calls
10998 Treat all calls as being far away (near). If calls are assumed to be
10999 far away, the compiler will always load the functions address up into a
11000 register, and call indirect through the pointer.
11006 Do not optimize (do optimize) basic blocks that use the same index
11007 pointer 4 or more times to copy pointer into the @code{ep} register, and
11008 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
11009 option is on by default if you optimize.
11011 @item -mno-prolog-function
11012 @itemx -mprolog-function
11013 @opindex mno-prolog-function
11014 @opindex mprolog-function
11015 Do not use (do use) external functions to save and restore registers
11016 at the prologue and epilogue of a function. The external functions
11017 are slower, but use less code space if more than one function saves
11018 the same number of registers. The @option{-mprolog-function} option
11019 is on by default if you optimize.
11023 Try to make the code as small as possible. At present, this just turns
11024 on the @option{-mep} and @option{-mprolog-function} options.
11026 @item -mtda=@var{n}
11028 Put static or global variables whose size is @var{n} bytes or less into
11029 the tiny data area that register @code{ep} points to. The tiny data
11030 area can hold up to 256 bytes in total (128 bytes for byte references).
11032 @item -msda=@var{n}
11034 Put static or global variables whose size is @var{n} bytes or less into
11035 the small data area that register @code{gp} points to. The small data
11036 area can hold up to 64 kilobytes.
11038 @item -mzda=@var{n}
11040 Put static or global variables whose size is @var{n} bytes or less into
11041 the first 32 kilobytes of memory.
11045 Specify that the target processor is the V850.
11048 @opindex mbig-switch
11049 Generate code suitable for big switch tables. Use this option only if
11050 the assembler/linker complain about out of range branches within a switch
11055 This option will cause r2 and r5 to be used in the code generated by
11056 the compiler. This setting is the default.
11058 @item -mno-app-regs
11059 @opindex mno-app-regs
11060 This option will cause r2 and r5 to be treated as fixed registers.
11064 Specify that the target processor is the V850E1. The preprocessor
11065 constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
11066 this option is used.
11070 Specify that the target processor is the V850E. The preprocessor
11071 constant @samp{__v850e__} will be defined if this option is used.
11073 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
11074 are defined then a default target processor will be chosen and the
11075 relevant @samp{__v850*__} preprocessor constant will be defined.
11077 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
11078 defined, regardless of which processor variant is the target.
11080 @item -mdisable-callt
11081 @opindex mdisable-callt
11082 This option will suppress generation of the CALLT instruction for the
11083 v850e and v850e1 flavors of the v850 architecture. The default is
11084 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
11089 @subsection VAX Options
11090 @cindex VAX options
11092 These @samp{-m} options are defined for the VAX:
11097 Do not output certain jump instructions (@code{aobleq} and so on)
11098 that the Unix assembler for the VAX cannot handle across long
11103 Do output those jump instructions, on the assumption that you
11104 will assemble with the GNU assembler.
11108 Output code for g-format floating point numbers instead of d-format.
11111 @node x86-64 Options
11112 @subsection x86-64 Options
11113 @cindex x86-64 options
11115 These are listed under @xref{i386 and x86-64 Options}.
11117 @node Xstormy16 Options
11118 @subsection Xstormy16 Options
11119 @cindex Xstormy16 Options
11121 These options are defined for Xstormy16:
11126 Choose startup files and linker script suitable for the simulator.
11129 @node Xtensa Options
11130 @subsection Xtensa Options
11131 @cindex Xtensa Options
11133 These options are supported for Xtensa targets:
11137 @itemx -mno-const16
11139 @opindex mno-const16
11140 Enable or disable use of @code{CONST16} instructions for loading
11141 constant values. The @code{CONST16} instruction is currently not a
11142 standard option from Tensilica. When enabled, @code{CONST16}
11143 instructions are always used in place of the standard @code{L32R}
11144 instructions. The use of @code{CONST16} is enabled by default only if
11145 the @code{L32R} instruction is not available.
11148 @itemx -mno-fused-madd
11149 @opindex mfused-madd
11150 @opindex mno-fused-madd
11151 Enable or disable use of fused multiply/add and multiply/subtract
11152 instructions in the floating-point option. This has no effect if the
11153 floating-point option is not also enabled. Disabling fused multiply/add
11154 and multiply/subtract instructions forces the compiler to use separate
11155 instructions for the multiply and add/subtract operations. This may be
11156 desirable in some cases where strict IEEE 754-compliant results are
11157 required: the fused multiply add/subtract instructions do not round the
11158 intermediate result, thereby producing results with @emph{more} bits of
11159 precision than specified by the IEEE standard. Disabling fused multiply
11160 add/subtract instructions also ensures that the program output is not
11161 sensitive to the compiler's ability to combine multiply and add/subtract
11164 @item -mtext-section-literals
11165 @itemx -mno-text-section-literals
11166 @opindex mtext-section-literals
11167 @opindex mno-text-section-literals
11168 Control the treatment of literal pools. The default is
11169 @option{-mno-text-section-literals}, which places literals in a separate
11170 section in the output file. This allows the literal pool to be placed
11171 in a data RAM/ROM, and it also allows the linker to combine literal
11172 pools from separate object files to remove redundant literals and
11173 improve code size. With @option{-mtext-section-literals}, the literals
11174 are interspersed in the text section in order to keep them as close as
11175 possible to their references. This may be necessary for large assembly
11178 @item -mtarget-align
11179 @itemx -mno-target-align
11180 @opindex mtarget-align
11181 @opindex mno-target-align
11182 When this option is enabled, GCC instructs the assembler to
11183 automatically align instructions to reduce branch penalties at the
11184 expense of some code density. The assembler attempts to widen density
11185 instructions to align branch targets and the instructions following call
11186 instructions. If there are not enough preceding safe density
11187 instructions to align a target, no widening will be performed. The
11188 default is @option{-mtarget-align}. These options do not affect the
11189 treatment of auto-aligned instructions like @code{LOOP}, which the
11190 assembler will always align, either by widening density instructions or
11191 by inserting no-op instructions.
11194 @itemx -mno-longcalls
11195 @opindex mlongcalls
11196 @opindex mno-longcalls
11197 When this option is enabled, GCC instructs the assembler to translate
11198 direct calls to indirect calls unless it can determine that the target
11199 of a direct call is in the range allowed by the call instruction. This
11200 translation typically occurs for calls to functions in other source
11201 files. Specifically, the assembler translates a direct @code{CALL}
11202 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
11203 The default is @option{-mno-longcalls}. This option should be used in
11204 programs where the call target can potentially be out of range. This
11205 option is implemented in the assembler, not the compiler, so the
11206 assembly code generated by GCC will still show direct call
11207 instructions---look at the disassembled object code to see the actual
11208 instructions. Note that the assembler will use an indirect call for
11209 every cross-file call, not just those that really will be out of range.
11212 @node zSeries Options
11213 @subsection zSeries Options
11214 @cindex zSeries options
11216 These are listed under @xref{S/390 and zSeries Options}.
11218 @node Code Gen Options
11219 @section Options for Code Generation Conventions
11220 @cindex code generation conventions
11221 @cindex options, code generation
11222 @cindex run-time options
11224 These machine-independent options control the interface conventions
11225 used in code generation.
11227 Most of them have both positive and negative forms; the negative form
11228 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
11229 one of the forms is listed---the one which is not the default. You
11230 can figure out the other form by either removing @samp{no-} or adding
11234 @item -fbounds-check
11235 @opindex fbounds-check
11236 For front-ends that support it, generate additional code to check that
11237 indices used to access arrays are within the declared range. This is
11238 currently only supported by the Java and Fortran 77 front-ends, where
11239 this option defaults to true and false respectively.
11243 This option generates traps for signed overflow on addition, subtraction,
11244 multiplication operations.
11248 This option instructs the compiler to assume that signed arithmetic
11249 overflow of addition, subtraction and multiplication wraps around
11250 using twos-complement representation. This flag enables some optimizations
11251 and disables other. This option is enabled by default for the Java
11252 front-end, as required by the Java language specification.
11255 @opindex fexceptions
11256 Enable exception handling. Generates extra code needed to propagate
11257 exceptions. For some targets, this implies GCC will generate frame
11258 unwind information for all functions, which can produce significant data
11259 size overhead, although it does not affect execution. If you do not
11260 specify this option, GCC will enable it by default for languages like
11261 C++ which normally require exception handling, and disable it for
11262 languages like C that do not normally require it. However, you may need
11263 to enable this option when compiling C code that needs to interoperate
11264 properly with exception handlers written in C++. You may also wish to
11265 disable this option if you are compiling older C++ programs that don't
11266 use exception handling.
11268 @item -fnon-call-exceptions
11269 @opindex fnon-call-exceptions
11270 Generate code that allows trapping instructions to throw exceptions.
11271 Note that this requires platform-specific runtime support that does
11272 not exist everywhere. Moreover, it only allows @emph{trapping}
11273 instructions to throw exceptions, i.e.@: memory references or floating
11274 point instructions. It does not allow exceptions to be thrown from
11275 arbitrary signal handlers such as @code{SIGALRM}.
11277 @item -funwind-tables
11278 @opindex funwind-tables
11279 Similar to @option{-fexceptions}, except that it will just generate any needed
11280 static data, but will not affect the generated code in any other way.
11281 You will normally not enable this option; instead, a language processor
11282 that needs this handling would enable it on your behalf.
11284 @item -fasynchronous-unwind-tables
11285 @opindex funwind-tables
11286 Generate unwind table in dwarf2 format, if supported by target machine. The
11287 table is exact at each instruction boundary, so it can be used for stack
11288 unwinding from asynchronous events (such as debugger or garbage collector).
11290 @item -fpcc-struct-return
11291 @opindex fpcc-struct-return
11292 Return ``short'' @code{struct} and @code{union} values in memory like
11293 longer ones, rather than in registers. This convention is less
11294 efficient, but it has the advantage of allowing intercallability between
11295 GCC-compiled files and files compiled with other compilers, particularly
11296 the Portable C Compiler (pcc).
11298 The precise convention for returning structures in memory depends
11299 on the target configuration macros.
11301 Short structures and unions are those whose size and alignment match
11302 that of some integer type.
11304 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
11305 switch is not binary compatible with code compiled with the
11306 @option{-freg-struct-return} switch.
11307 Use it to conform to a non-default application binary interface.
11309 @item -freg-struct-return
11310 @opindex freg-struct-return
11311 Return @code{struct} and @code{union} values in registers when possible.
11312 This is more efficient for small structures than
11313 @option{-fpcc-struct-return}.
11315 If you specify neither @option{-fpcc-struct-return} nor
11316 @option{-freg-struct-return}, GCC defaults to whichever convention is
11317 standard for the target. If there is no standard convention, GCC
11318 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
11319 the principal compiler. In those cases, we can choose the standard, and
11320 we chose the more efficient register return alternative.
11322 @strong{Warning:} code compiled with the @option{-freg-struct-return}
11323 switch is not binary compatible with code compiled with the
11324 @option{-fpcc-struct-return} switch.
11325 Use it to conform to a non-default application binary interface.
11327 @item -fshort-enums
11328 @opindex fshort-enums
11329 Allocate to an @code{enum} type only as many bytes as it needs for the
11330 declared range of possible values. Specifically, the @code{enum} type
11331 will be equivalent to the smallest integer type which has enough room.
11333 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
11334 code that is not binary compatible with code generated without that switch.
11335 Use it to conform to a non-default application binary interface.
11337 @item -fshort-double
11338 @opindex fshort-double
11339 Use the same size for @code{double} as for @code{float}.
11341 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
11342 code that is not binary compatible with code generated without that switch.
11343 Use it to conform to a non-default application binary interface.
11345 @item -fshort-wchar
11346 @opindex fshort-wchar
11347 Override the underlying type for @samp{wchar_t} to be @samp{short
11348 unsigned int} instead of the default for the target. This option is
11349 useful for building programs to run under WINE@.
11351 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
11352 code that is not binary compatible with code generated without that switch.
11353 Use it to conform to a non-default application binary interface.
11355 @item -fshared-data
11356 @opindex fshared-data
11357 Requests that the data and non-@code{const} variables of this
11358 compilation be shared data rather than private data. The distinction
11359 makes sense only on certain operating systems, where shared data is
11360 shared between processes running the same program, while private data
11361 exists in one copy per process.
11364 @opindex fno-common
11365 In C, allocate even uninitialized global variables in the data section of the
11366 object file, rather than generating them as common blocks. This has the
11367 effect that if the same variable is declared (without @code{extern}) in
11368 two different compilations, you will get an error when you link them.
11369 The only reason this might be useful is if you wish to verify that the
11370 program will work on other systems which always work this way.
11374 Ignore the @samp{#ident} directive.
11376 @item -finhibit-size-directive
11377 @opindex finhibit-size-directive
11378 Don't output a @code{.size} assembler directive, or anything else that
11379 would cause trouble if the function is split in the middle, and the
11380 two halves are placed at locations far apart in memory. This option is
11381 used when compiling @file{crtstuff.c}; you should not need to use it
11384 @item -fverbose-asm
11385 @opindex fverbose-asm
11386 Put extra commentary information in the generated assembly code to
11387 make it more readable. This option is generally only of use to those
11388 who actually need to read the generated assembly code (perhaps while
11389 debugging the compiler itself).
11391 @option{-fno-verbose-asm}, the default, causes the
11392 extra information to be omitted and is useful when comparing two assembler
11397 @cindex global offset table
11399 Generate position-independent code (PIC) suitable for use in a shared
11400 library, if supported for the target machine. Such code accesses all
11401 constant addresses through a global offset table (GOT)@. The dynamic
11402 loader resolves the GOT entries when the program starts (the dynamic
11403 loader is not part of GCC; it is part of the operating system). If
11404 the GOT size for the linked executable exceeds a machine-specific
11405 maximum size, you get an error message from the linker indicating that
11406 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
11407 instead. (These maximums are 8k on the SPARC and 32k
11408 on the m68k and RS/6000. The 386 has no such limit.)
11410 Position-independent code requires special support, and therefore works
11411 only on certain machines. For the 386, GCC supports PIC for System V
11412 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
11413 position-independent.
11417 If supported for the target machine, emit position-independent code,
11418 suitable for dynamic linking and avoiding any limit on the size of the
11419 global offset table. This option makes a difference on the m68k
11422 Position-independent code requires special support, and therefore works
11423 only on certain machines.
11429 These options are similar to @option{-fpic} and @option{-fPIC}, but
11430 generated position independent code can be only linked into executables.
11431 Usually these options are used when @option{-pie} GCC option will be
11432 used during linking.
11434 @item -ffixed-@var{reg}
11436 Treat the register named @var{reg} as a fixed register; generated code
11437 should never refer to it (except perhaps as a stack pointer, frame
11438 pointer or in some other fixed role).
11440 @var{reg} must be the name of a register. The register names accepted
11441 are machine-specific and are defined in the @code{REGISTER_NAMES}
11442 macro in the machine description macro file.
11444 This flag does not have a negative form, because it specifies a
11447 @item -fcall-used-@var{reg}
11448 @opindex fcall-used
11449 Treat the register named @var{reg} as an allocable register that is
11450 clobbered by function calls. It may be allocated for temporaries or
11451 variables that do not live across a call. Functions compiled this way
11452 will not save and restore the register @var{reg}.
11454 It is an error to used this flag with the frame pointer or stack pointer.
11455 Use of this flag for other registers that have fixed pervasive roles in
11456 the machine's execution model will produce disastrous results.
11458 This flag does not have a negative form, because it specifies a
11461 @item -fcall-saved-@var{reg}
11462 @opindex fcall-saved
11463 Treat the register named @var{reg} as an allocable register saved by
11464 functions. It may be allocated even for temporaries or variables that
11465 live across a call. Functions compiled this way will save and restore
11466 the register @var{reg} if they use it.
11468 It is an error to used this flag with the frame pointer or stack pointer.
11469 Use of this flag for other registers that have fixed pervasive roles in
11470 the machine's execution model will produce disastrous results.
11472 A different sort of disaster will result from the use of this flag for
11473 a register in which function values may be returned.
11475 This flag does not have a negative form, because it specifies a
11478 @item -fpack-struct
11479 @opindex fpack-struct
11480 Pack all structure members together without holes.
11482 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
11483 code that is not binary compatible with code generated without that switch.
11484 Additionally, it makes the code suboptimal.
11485 Use it to conform to a non-default application binary interface.
11487 @item -finstrument-functions
11488 @opindex finstrument-functions
11489 Generate instrumentation calls for entry and exit to functions. Just
11490 after function entry and just before function exit, the following
11491 profiling functions will be called with the address of the current
11492 function and its call site. (On some platforms,
11493 @code{__builtin_return_address} does not work beyond the current
11494 function, so the call site information may not be available to the
11495 profiling functions otherwise.)
11498 void __cyg_profile_func_enter (void *this_fn,
11500 void __cyg_profile_func_exit (void *this_fn,
11504 The first argument is the address of the start of the current function,
11505 which may be looked up exactly in the symbol table.
11507 This instrumentation is also done for functions expanded inline in other
11508 functions. The profiling calls will indicate where, conceptually, the
11509 inline function is entered and exited. This means that addressable
11510 versions of such functions must be available. If all your uses of a
11511 function are expanded inline, this may mean an additional expansion of
11512 code size. If you use @samp{extern inline} in your C code, an
11513 addressable version of such functions must be provided. (This is
11514 normally the case anyways, but if you get lucky and the optimizer always
11515 expands the functions inline, you might have gotten away without
11516 providing static copies.)
11518 A function may be given the attribute @code{no_instrument_function}, in
11519 which case this instrumentation will not be done. This can be used, for
11520 example, for the profiling functions listed above, high-priority
11521 interrupt routines, and any functions from which the profiling functions
11522 cannot safely be called (perhaps signal handlers, if the profiling
11523 routines generate output or allocate memory).
11525 @item -fstack-check
11526 @opindex fstack-check
11527 Generate code to verify that you do not go beyond the boundary of the
11528 stack. You should specify this flag if you are running in an
11529 environment with multiple threads, but only rarely need to specify it in
11530 a single-threaded environment since stack overflow is automatically
11531 detected on nearly all systems if there is only one stack.
11533 Note that this switch does not actually cause checking to be done; the
11534 operating system must do that. The switch causes generation of code
11535 to ensure that the operating system sees the stack being extended.
11537 @item -fstack-limit-register=@var{reg}
11538 @itemx -fstack-limit-symbol=@var{sym}
11539 @itemx -fno-stack-limit
11540 @opindex fstack-limit-register
11541 @opindex fstack-limit-symbol
11542 @opindex fno-stack-limit
11543 Generate code to ensure that the stack does not grow beyond a certain value,
11544 either the value of a register or the address of a symbol. If the stack
11545 would grow beyond the value, a signal is raised. For most targets,
11546 the signal is raised before the stack overruns the boundary, so
11547 it is possible to catch the signal without taking special precautions.
11549 For instance, if the stack starts at absolute address @samp{0x80000000}
11550 and grows downwards, you can use the flags
11551 @option{-fstack-limit-symbol=__stack_limit} and
11552 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
11553 of 128KB@. Note that this may only work with the GNU linker.
11555 @cindex aliasing of parameters
11556 @cindex parameters, aliased
11557 @item -fargument-alias
11558 @itemx -fargument-noalias
11559 @itemx -fargument-noalias-global
11560 @opindex fargument-alias
11561 @opindex fargument-noalias
11562 @opindex fargument-noalias-global
11563 Specify the possible relationships among parameters and between
11564 parameters and global data.
11566 @option{-fargument-alias} specifies that arguments (parameters) may
11567 alias each other and may alias global storage.@*
11568 @option{-fargument-noalias} specifies that arguments do not alias
11569 each other, but may alias global storage.@*
11570 @option{-fargument-noalias-global} specifies that arguments do not
11571 alias each other and do not alias global storage.
11573 Each language will automatically use whatever option is required by
11574 the language standard. You should not need to use these options yourself.
11576 @item -fleading-underscore
11577 @opindex fleading-underscore
11578 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
11579 change the way C symbols are represented in the object file. One use
11580 is to help link with legacy assembly code.
11582 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
11583 generate code that is not binary compatible with code generated without that
11584 switch. Use it to conform to a non-default application binary interface.
11585 Not all targets provide complete support for this switch.
11587 @item -ftls-model=@var{model}
11588 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
11589 The @var{model} argument should be one of @code{global-dynamic},
11590 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
11592 The default without @option{-fpic} is @code{initial-exec}; with
11593 @option{-fpic} the default is @code{global-dynamic}.
11598 @node Environment Variables
11599 @section Environment Variables Affecting GCC
11600 @cindex environment variables
11602 @c man begin ENVIRONMENT
11603 This section describes several environment variables that affect how GCC
11604 operates. Some of them work by specifying directories or prefixes to use
11605 when searching for various kinds of files. Some are used to specify other
11606 aspects of the compilation environment.
11608 Note that you can also specify places to search using options such as
11609 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
11610 take precedence over places specified using environment variables, which
11611 in turn take precedence over those specified by the configuration of GCC@.
11612 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
11613 GNU Compiler Collection (GCC) Internals}.
11618 @c @itemx LC_COLLATE
11620 @c @itemx LC_MONETARY
11621 @c @itemx LC_NUMERIC
11626 @c @findex LC_COLLATE
11627 @findex LC_MESSAGES
11628 @c @findex LC_MONETARY
11629 @c @findex LC_NUMERIC
11633 These environment variables control the way that GCC uses
11634 localization information that allow GCC to work with different
11635 national conventions. GCC inspects the locale categories
11636 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
11637 so. These locale categories can be set to any value supported by your
11638 installation. A typical value is @samp{en_UK} for English in the United
11641 The @env{LC_CTYPE} environment variable specifies character
11642 classification. GCC uses it to determine the character boundaries in
11643 a string; this is needed for some multibyte encodings that contain quote
11644 and escape characters that would otherwise be interpreted as a string
11647 The @env{LC_MESSAGES} environment variable specifies the language to
11648 use in diagnostic messages.
11650 If the @env{LC_ALL} environment variable is set, it overrides the value
11651 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
11652 and @env{LC_MESSAGES} default to the value of the @env{LANG}
11653 environment variable. If none of these variables are set, GCC
11654 defaults to traditional C English behavior.
11658 If @env{TMPDIR} is set, it specifies the directory to use for temporary
11659 files. GCC uses temporary files to hold the output of one stage of
11660 compilation which is to be used as input to the next stage: for example,
11661 the output of the preprocessor, which is the input to the compiler
11664 @item GCC_EXEC_PREFIX
11665 @findex GCC_EXEC_PREFIX
11666 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
11667 names of the subprograms executed by the compiler. No slash is added
11668 when this prefix is combined with the name of a subprogram, but you can
11669 specify a prefix that ends with a slash if you wish.
11671 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
11672 an appropriate prefix to use based on the pathname it was invoked with.
11674 If GCC cannot find the subprogram using the specified prefix, it
11675 tries looking in the usual places for the subprogram.
11677 The default value of @env{GCC_EXEC_PREFIX} is
11678 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the value
11679 of @code{prefix} when you ran the @file{configure} script.
11681 Other prefixes specified with @option{-B} take precedence over this prefix.
11683 This prefix is also used for finding files such as @file{crt0.o} that are
11686 In addition, the prefix is used in an unusual way in finding the
11687 directories to search for header files. For each of the standard
11688 directories whose name normally begins with @samp{/usr/local/lib/gcc}
11689 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
11690 replacing that beginning with the specified prefix to produce an
11691 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
11692 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
11693 These alternate directories are searched first; the standard directories
11696 @item COMPILER_PATH
11697 @findex COMPILER_PATH
11698 The value of @env{COMPILER_PATH} is a colon-separated list of
11699 directories, much like @env{PATH}. GCC tries the directories thus
11700 specified when searching for subprograms, if it can't find the
11701 subprograms using @env{GCC_EXEC_PREFIX}.
11704 @findex LIBRARY_PATH
11705 The value of @env{LIBRARY_PATH} is a colon-separated list of
11706 directories, much like @env{PATH}. When configured as a native compiler,
11707 GCC tries the directories thus specified when searching for special
11708 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
11709 using GCC also uses these directories when searching for ordinary
11710 libraries for the @option{-l} option (but directories specified with
11711 @option{-L} come first).
11715 @cindex locale definition
11716 This variable is used to pass locale information to the compiler. One way in
11717 which this information is used is to determine the character set to be used
11718 when character literals, string literals and comments are parsed in C and C++.
11719 When the compiler is configured to allow multibyte characters,
11720 the following values for @env{LANG} are recognized:
11724 Recognize JIS characters.
11726 Recognize SJIS characters.
11728 Recognize EUCJP characters.
11731 If @env{LANG} is not defined, or if it has some other value, then the
11732 compiler will use mblen and mbtowc as defined by the default locale to
11733 recognize and translate multibyte characters.
11737 Some additional environments variables affect the behavior of the
11740 @include cppenv.texi
11744 @node Precompiled Headers
11745 @section Using Precompiled Headers
11746 @cindex precompiled headers
11747 @cindex speed of compilation
11749 Often large projects have many header files that are included in every
11750 source file. The time the compiler takes to process these header files
11751 over and over again can account for nearly all of the time required to
11752 build the project. To make builds faster, GCC allows users to
11753 `precompile' a header file; then, if builds can use the precompiled
11754 header file they will be much faster.
11756 @strong{Caution:} There are a few known situations where GCC will
11757 crash when trying to use a precompiled header. If you have trouble
11758 with a precompiled header, you should remove the precompiled header
11759 and compile without it. In addition, please use GCC's on-line
11760 defect-tracking system to report any problems you encounter with
11761 precompiled headers. @xref{Bugs}.
11763 To create a precompiled header file, simply compile it as you would any
11764 other file, if necessary using the @option{-x} option to make the driver
11765 treat it as a C or C++ header file. You will probably want to use a
11766 tool like @command{make} to keep the precompiled header up-to-date when
11767 the headers it contains change.
11769 A precompiled header file will be searched for when @code{#include} is
11770 seen in the compilation. As it searches for the included file
11771 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
11772 compiler looks for a precompiled header in each directory just before it
11773 looks for the include file in that directory. The name searched for is
11774 the name specified in the @code{#include} with @samp{.gch} appended. If
11775 the precompiled header file can't be used, it is ignored.
11777 For instance, if you have @code{#include "all.h"}, and you have
11778 @file{all.h.gch} in the same directory as @file{all.h}, then the
11779 precompiled header file will be used if possible, and the original
11780 header will be used otherwise.
11782 Alternatively, you might decide to put the precompiled header file in a
11783 directory and use @option{-I} to ensure that directory is searched
11784 before (or instead of) the directory containing the original header.
11785 Then, if you want to check that the precompiled header file is always
11786 used, you can put a file of the same name as the original header in this
11787 directory containing an @code{#error} command.
11789 This also works with @option{-include}. So yet another way to use
11790 precompiled headers, good for projects not designed with precompiled
11791 header files in mind, is to simply take most of the header files used by
11792 a project, include them from another header file, precompile that header
11793 file, and @option{-include} the precompiled header. If the header files
11794 have guards against multiple inclusion, they will be skipped because
11795 they've already been included (in the precompiled header).
11797 If you need to precompile the same header file for different
11798 languages, targets, or compiler options, you can instead make a
11799 @emph{directory} named like @file{all.h.gch}, and put each precompiled
11800 header in the directory, perhaps using @option{-o}. It doesn't matter
11801 what you call the files in the directory, every precompiled header in
11802 the directory will be considered. The first precompiled header
11803 encountered in the directory that is valid for this compilation will
11804 be used; they're searched in no particular order.
11806 There are many other possibilities, limited only by your imagination,
11807 good sense, and the constraints of your build system.
11809 A precompiled header file can be used only when these conditions apply:
11813 Only one precompiled header can be used in a particular compilation.
11816 A precompiled header can't be used once the first C token is seen. You
11817 can have preprocessor directives before a precompiled header; you can
11818 even include a precompiled header from inside another header, so long as
11819 there are no C tokens before the @code{#include}.
11822 The precompiled header file must be produced for the same language as
11823 the current compilation. You can't use a C precompiled header for a C++
11827 The precompiled header file must be produced by the same compiler
11828 version and configuration as the current compilation is using.
11829 The easiest way to guarantee this is to use the same compiler binary
11830 for creating and using precompiled headers.
11833 Any macros defined before the precompiled header is included must
11834 either be defined in the same way as when the precompiled header was
11835 generated, or must not affect the precompiled header, which usually
11836 means that the they don't appear in the precompiled header at all.
11838 The @option{-D} option is one way to define a macro before a
11839 precompiled header is included; using a @code{#define} can also do it.
11840 There are also some options that define macros implicitly, like
11841 @option{-O} and @option{-Wdeprecated}; the same rule applies to macros
11844 @item If debugging information is output when using the precompiled
11845 header, using @option{-g} or similar, the same kind of debugging information
11846 must have been output when building the precompiled header. However,
11847 a precompiled header built using @option{-g} can be used in a compilation
11848 when no debugging information is being output.
11850 @item The same @option{-m} options must generally be used when building
11851 and using the precompiled header. @xref{Submodel Options},
11852 for any cases where this rule is relaxed.
11854 @item Each of the following options must be the same when building and using
11855 the precompiled header:
11857 @gccoptlist{-fexceptions -funit-at-a-time}
11860 Some other command-line options starting with @option{-f},
11861 @option{-p}, or @option{-O} must be defined in the same way as when
11862 the precompiled header was generated. At present, it's not clear
11863 which options are safe to change and which are not; the safest choice
11864 is to use exactly the same options when generating and using the
11865 precompiled header. The following are known to be safe:
11867 @gccoptlist{-fpreprocessed -pedantic-errors}
11871 For all of these except the last, the compiler will automatically
11872 ignore the precompiled header if the conditions aren't met. If you
11873 find an option combination that doesn't work and doesn't cause the
11874 precompiled header to be ignored, please consider filing a bug report,
11877 If you do use differing options when generating and using the
11878 precompiled header, the actual behaviour will be a mixture of the
11879 behaviour for the options. For instance, if you use @option{-g} to
11880 generate the precompiled header but not when using it, you may or may
11881 not get debugging information for routines in the precompiled header.
11883 @node Running Protoize
11884 @section Running Protoize
11886 The program @code{protoize} is an optional part of GCC@. You can use
11887 it to add prototypes to a program, thus converting the program to ISO
11888 C in one respect. The companion program @code{unprotoize} does the
11889 reverse: it removes argument types from any prototypes that are found.
11891 When you run these programs, you must specify a set of source files as
11892 command line arguments. The conversion programs start out by compiling
11893 these files to see what functions they define. The information gathered
11894 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
11896 After scanning comes actual conversion. The specified files are all
11897 eligible to be converted; any files they include (whether sources or
11898 just headers) are eligible as well.
11900 But not all the eligible files are converted. By default,
11901 @code{protoize} and @code{unprotoize} convert only source and header
11902 files in the current directory. You can specify additional directories
11903 whose files should be converted with the @option{-d @var{directory}}
11904 option. You can also specify particular files to exclude with the
11905 @option{-x @var{file}} option. A file is converted if it is eligible, its
11906 directory name matches one of the specified directory names, and its
11907 name within the directory has not been excluded.
11909 Basic conversion with @code{protoize} consists of rewriting most
11910 function definitions and function declarations to specify the types of
11911 the arguments. The only ones not rewritten are those for varargs
11914 @code{protoize} optionally inserts prototype declarations at the
11915 beginning of the source file, to make them available for any calls that
11916 precede the function's definition. Or it can insert prototype
11917 declarations with block scope in the blocks where undeclared functions
11920 Basic conversion with @code{unprotoize} consists of rewriting most
11921 function declarations to remove any argument types, and rewriting
11922 function definitions to the old-style pre-ISO form.
11924 Both conversion programs print a warning for any function declaration or
11925 definition that they can't convert. You can suppress these warnings
11928 The output from @code{protoize} or @code{unprotoize} replaces the
11929 original source file. The original file is renamed to a name ending
11930 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
11931 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
11932 for DOS) file already exists, then the source file is simply discarded.
11934 @code{protoize} and @code{unprotoize} both depend on GCC itself to
11935 scan the program and collect information about the functions it uses.
11936 So neither of these programs will work until GCC is installed.
11938 Here is a table of the options you can use with @code{protoize} and
11939 @code{unprotoize}. Each option works with both programs unless
11943 @item -B @var{directory}
11944 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
11945 usual directory (normally @file{/usr/local/lib}). This file contains
11946 prototype information about standard system functions. This option
11947 applies only to @code{protoize}.
11949 @item -c @var{compilation-options}
11950 Use @var{compilation-options} as the options when running @command{gcc} to
11951 produce the @samp{.X} files. The special option @option{-aux-info} is
11952 always passed in addition, to tell @command{gcc} to write a @samp{.X} file.
11954 Note that the compilation options must be given as a single argument to
11955 @code{protoize} or @code{unprotoize}. If you want to specify several
11956 @command{gcc} options, you must quote the entire set of compilation options
11957 to make them a single word in the shell.
11959 There are certain @command{gcc} arguments that you cannot use, because they
11960 would produce the wrong kind of output. These include @option{-g},
11961 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
11962 the @var{compilation-options}, they are ignored.
11965 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
11966 systems) instead of @samp{.c}. This is convenient if you are converting
11967 a C program to C++. This option applies only to @code{protoize}.
11970 Add explicit global declarations. This means inserting explicit
11971 declarations at the beginning of each source file for each function
11972 that is called in the file and was not declared. These declarations
11973 precede the first function definition that contains a call to an
11974 undeclared function. This option applies only to @code{protoize}.
11976 @item -i @var{string}
11977 Indent old-style parameter declarations with the string @var{string}.
11978 This option applies only to @code{protoize}.
11980 @code{unprotoize} converts prototyped function definitions to old-style
11981 function definitions, where the arguments are declared between the
11982 argument list and the initial @samp{@{}. By default, @code{unprotoize}
11983 uses five spaces as the indentation. If you want to indent with just
11984 one space instead, use @option{-i " "}.
11987 Keep the @samp{.X} files. Normally, they are deleted after conversion
11991 Add explicit local declarations. @code{protoize} with @option{-l} inserts
11992 a prototype declaration for each function in each block which calls the
11993 function without any declaration. This option applies only to
11997 Make no real changes. This mode just prints information about the conversions
11998 that would have been done without @option{-n}.
12001 Make no @samp{.save} files. The original files are simply deleted.
12002 Use this option with caution.
12004 @item -p @var{program}
12005 Use the program @var{program} as the compiler. Normally, the name
12006 @file{gcc} is used.
12009 Work quietly. Most warnings are suppressed.
12012 Print the version number, just like @option{-v} for @command{gcc}.
12015 If you need special compiler options to compile one of your program's
12016 source files, then you should generate that file's @samp{.X} file
12017 specially, by running @command{gcc} on that source file with the
12018 appropriate options and the option @option{-aux-info}. Then run
12019 @code{protoize} on the entire set of files. @code{protoize} will use
12020 the existing @samp{.X} file because it is newer than the source file.
12024 gcc -Dfoo=bar file1.c -aux-info file1.X
12029 You need to include the special files along with the rest in the
12030 @code{protoize} command, even though their @samp{.X} files already
12031 exist, because otherwise they won't get converted.
12033 @xref{Protoize Caveats}, for more information on how to use
12034 @code{protoize} successfully.