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 and Objective-C++ Dialect Options:: Variations on Objective-C
126 * Language Independent Options:: Controlling how diagnostics should be
128 * Warning Options:: How picky should the compiler be?
129 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
130 * Optimize Options:: How much optimization?
131 * Preprocessor Options:: Controlling header files and macro definitions.
132 Also, getting dependency information for Make.
133 * Assembler Options:: Passing options to the assembler.
134 * Link Options:: Specifying libraries and so on.
135 * Directory Options:: Where to find header files and libraries.
136 Where to find the compiler executable files.
137 * Spec Files:: How to pass switches to sub-processes.
138 * Target Options:: Running a cross-compiler, or an old version of GCC.
139 * Submodel Options:: Specifying minor hardware or convention variations,
140 such as 68010 vs 68020.
141 * Code Gen Options:: Specifying conventions for function calls, data layout
143 * Environment Variables:: Env vars that affect GCC.
144 * Precompiled Headers:: Compiling a header once, and using it many times.
145 * Running Protoize:: Automatically adding or removing function prototypes.
151 @section Option Summary
153 Here is a summary of all the options, grouped by type. Explanations are
154 in the following sections.
157 @item Overall Options
158 @xref{Overall Options,,Options Controlling the Kind of Output}.
159 @gccoptlist{-c -S -E -o @var{file} -combine -pipe -pass-exit-codes @gol
160 -x @var{language} -v -### --help --target-help --version}
162 @item C Language Options
163 @xref{C Dialect Options,,Options Controlling C Dialect}.
164 @gccoptlist{-ansi -std=@var{standard} -aux-info @var{filename} @gol
165 -fno-asm -fno-builtin -fno-builtin-@var{function} @gol
166 -fhosted -ffreestanding -fms-extensions @gol
167 -trigraphs -no-integrated-cpp -traditional -traditional-cpp @gol
168 -fallow-single-precision -fcond-mismatch @gol
169 -fsigned-bitfields -fsigned-char @gol
170 -funsigned-bitfields -funsigned-char}
172 @item C++ Language Options
173 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
174 @gccoptlist{-fabi-version=@var{n} -fno-access-control -fcheck-new @gol
175 -fconserve-space -fno-const-strings @gol
176 -fno-elide-constructors @gol
177 -fno-enforce-eh-specs @gol
178 -ffor-scope -fno-for-scope -fno-gnu-keywords @gol
179 -fno-implicit-templates @gol
180 -fno-implicit-inline-templates @gol
181 -fno-implement-inlines -fms-extensions @gol
182 -fno-nonansi-builtins -fno-operator-names @gol
183 -fno-optional-diags -fpermissive @gol
184 -frepo -fno-rtti -fstats -ftemplate-depth-@var{n} @gol
185 -fno-threadsafe-statics -fuse-cxa-atexit -fno-weak -nostdinc++ @gol
186 -fno-default-inline -fvisibility-inlines-hidden @gol
187 -Wabi -Wctor-dtor-privacy @gol
188 -Wnon-virtual-dtor -Wreorder @gol
189 -Weffc++ -Wno-deprecated @gol
190 -Wno-non-template-friend -Wold-style-cast @gol
191 -Woverloaded-virtual -Wno-pmf-conversions @gol
192 -Wsign-promo -Wsynth}
194 @item Objective-C and Objective-C++ Language Options
195 @xref{Objective-C and Objective-C++ Dialect Options,,Options Controlling
196 Objective-C and Objective-C++ Dialects}.
198 -fconstant-string-class=@var{class-name} @gol
199 -fgnu-runtime -fnext-runtime @gol
200 -fno-nil-receivers @gol
201 -fobjc-exceptions @gol
202 -freplace-objc-classes @gol
205 -Wno-protocol -Wselector -Wundeclared-selector}
207 @item Language Independent Options
208 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
209 @gccoptlist{-fmessage-length=@var{n} @gol
210 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]}}
212 @item Warning Options
213 @xref{Warning Options,,Options to Request or Suppress Warnings}.
214 @gccoptlist{-fsyntax-only -pedantic -pedantic-errors @gol
215 -w -Wextra -Wall -Waggregate-return @gol
216 -Wcast-align -Wcast-qual -Wchar-subscripts -Wcomment @gol
217 -Wconversion -Wno-deprecated-declarations @gol
218 -Wdisabled-optimization -Wno-div-by-zero -Wendif-labels @gol
219 -Werror -Werror-implicit-function-declaration @gol
220 -Wfatal-errors -Wfloat-equal -Wformat -Wformat=2 @gol
221 -Wno-format-extra-args -Wformat-nonliteral @gol
222 -Wformat-security -Wformat-y2k @gol
223 -Wimplicit -Wimplicit-function-declaration -Wimplicit-int @gol
224 -Wimport -Wno-import -Winit-self -Winline @gol
225 -Wno-invalid-offsetof -Winvalid-pch @gol
226 -Wlarger-than-@var{len} -Wlong-long @gol
227 -Wmain -Wmissing-braces -Wmissing-field-initializers @gol
228 -Wmissing-format-attribute -Wmissing-include-dirs @gol
229 -Wmissing-noreturn @gol
230 -Wno-multichar -Wnonnull -Wpacked -Wpadded @gol
231 -Wparentheses -Wpointer-arith -Wredundant-decls @gol
232 -Wreturn-type -Wsequence-point -Wshadow @gol
233 -Wsign-compare -Wstrict-aliasing -Wstrict-aliasing=2 @gol
234 -Wswitch -Wswitch-default -Wswitch-enum @gol
235 -Wsystem-headers -Wtrigraphs -Wundef -Wuninitialized @gol
236 -Wunknown-pragmas -Wunreachable-code @gol
237 -Wunused -Wunused-function -Wunused-label -Wunused-parameter @gol
238 -Wunused-value -Wunused-variable -Wwrite-strings @gol
241 @item C-only Warning Options
242 @gccoptlist{-Wbad-function-cast -Wmissing-declarations @gol
243 -Wmissing-prototypes -Wnested-externs -Wold-style-definition @gol
244 -Wstrict-prototypes -Wtraditional @gol
245 -Wdeclaration-after-statement}
247 @item Debugging Options
248 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
249 @gccoptlist{-d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
250 -fdump-unnumbered -fdump-translation-unit@r{[}-@var{n}@r{]} @gol
251 -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
252 -fdump-ipa-all -fdump-ipa-cgraph @gol
254 -fdump-tree-original@r{[}-@var{n}@r{]} @gol
255 -fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
256 -fdump-tree-inlined@r{[}-@var{n}@r{]} @gol
257 -fdump-tree-cfg -fdump-tree-vcg -fdump-tree-alias @gol
259 -fdump-tree-ssa@r{[}-@var{n}@r{]} -fdump-tree-pre@r{[}-@var{n}@r{]} @gol
260 -fdump-tree-ccp@r{[}-@var{n}@r{]} -fdump-tree-dce@r{[}-@var{n}@r{]} @gol
261 -fdump-tree-gimple@r{[}-raw@r{]} -fdump-tree-mudflap@r{[}-@var{n}@r{]} @gol
262 -fdump-tree-dom@r{[}-@var{n}@r{]} @gol
263 -fdump-tree-dse@r{[}-@var{n}@r{]} @gol
264 -fdump-tree-phiopt@r{[}-@var{n}@r{]} @gol
265 -fdump-tree-forwprop@r{[}-@var{n}@r{]} @gol
266 -fdump-tree-copyrename@r{[}-@var{n}@r{]} @gol
267 -fdump-tree-nrv -fdump-tree-vect @gol
268 -fdump-tree-sra@r{[}-@var{n}@r{]} @gol
269 -fdump-tree-fre@r{[}-@var{n}@r{]} @gol
270 -feliminate-dwarf2-dups -feliminate-unused-debug-types @gol
271 -feliminate-unused-debug-symbols -fmem-report -fprofile-arcs -ftree-based-profiling @gol
272 -frandom-seed=@var{string} -fsched-verbose=@var{n} @gol
273 -ftest-coverage -ftime-report -fvar-tracking @gol
274 -g -g@var{level} -gcoff -gdwarf-2 @gol
275 -ggdb -gstabs -gstabs+ -gvms -gxcoff -gxcoff+ @gol
276 -p -pg -print-file-name=@var{library} -print-libgcc-file-name @gol
277 -print-multi-directory -print-multi-lib @gol
278 -print-prog-name=@var{program} -print-search-dirs -Q @gol
281 @item Optimization Options
282 @xref{Optimize Options,,Options that Control Optimization}.
283 @gccoptlist{-falign-functions=@var{n} -falign-jumps=@var{n} @gol
284 -falign-labels=@var{n} -falign-loops=@var{n} @gol
285 -fbounds-check -fmudflap -fmudflapth -fmudflapir @gol
286 -fbranch-probabilities -fprofile-values -fvpt -fbranch-target-load-optimize @gol
287 -fbranch-target-load-optimize2 -fbtr-bb-exclusive @gol
288 -fcaller-saves -fcprop-registers @gol
289 -fcse-follow-jumps -fcse-skip-blocks -fdata-sections @gol
290 -fdelayed-branch -fdelete-null-pointer-checks @gol
291 -fexpensive-optimizations -ffast-math -ffloat-store @gol
292 -fforce-addr -fforce-mem -ffunction-sections @gol
293 -fgcse -fgcse-lm -fgcse-sm -fgcse-las -fgcse-after-reload @gol
294 -floop-optimize -fcrossjumping -fif-conversion -fif-conversion2 @gol
295 -finline-functions -finline-limit=@var{n} -fkeep-inline-functions @gol
296 -fkeep-static-consts -fmerge-constants -fmerge-all-constants @gol
297 -fmodulo-sched -fnew-ra -fno-branch-count-reg @gol
298 -fno-default-inline -fno-defer-pop -floop-optimize2 -fmove-loop-invariants @gol
299 -fno-function-cse -fno-guess-branch-probability @gol
300 -fno-inline -fno-math-errno -fno-peephole -fno-peephole2 @gol
301 -funsafe-math-optimizations -ffinite-math-only @gol
302 -fno-trapping-math -fno-zero-initialized-in-bss @gol
303 -fomit-frame-pointer -foptimize-register-move @gol
304 -foptimize-sibling-calls -fprefetch-loop-arrays @gol
305 -fprofile-generate -fprofile-use @gol
306 -fregmove -frename-registers @gol
307 -freorder-blocks -freorder-blocks-and-partition -freorder-functions @gol
308 -frerun-cse-after-loop -frerun-loop-opt @gol
309 -frounding-math -fschedule-insns -fschedule-insns2 @gol
310 -fno-sched-interblock -fno-sched-spec -fsched-spec-load @gol
311 -fsched-spec-load-dangerous @gol
312 -fsched-stalled-insns=@var{n} -sched-stalled-insns-dep=@var{n} @gol
313 -fsched2-use-superblocks @gol
314 -fsched2-use-traces -freschedule-modulo-scheduled-loops @gol
315 -fsignaling-nans -fsingle-precision-constant -fspeculative-prefetching @gol
316 -fstrength-reduce -fstrict-aliasing -ftracer -fthread-jumps @gol
317 -funroll-all-loops -funroll-loops -fpeel-loops @gol
318 -fsplit-ivs-in-unroller -funswitch-loops @gol
319 -ftree-pre -ftree-ccp -ftree-dce -ftree-loop-optimize @gol
320 -ftree-loop-linear -ftree-loop-im -ftree-loop-ivcanon -fivopts @gol
321 -ftree-dominator-opts -ftree-dse -ftree-copyrename @gol
322 -ftree-ch -ftree-sra -ftree-ter -ftree-lrs -ftree-fre -ftree-vectorize @gol
323 --param @var{name}=@var{value}
324 -O -O0 -O1 -O2 -O3 -Os}
326 @item Preprocessor Options
327 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
328 @gccoptlist{-A@var{question}=@var{answer} @gol
329 -A-@var{question}@r{[}=@var{answer}@r{]} @gol
330 -C -dD -dI -dM -dN @gol
331 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
332 -idirafter @var{dir} @gol
333 -include @var{file} -imacros @var{file} @gol
334 -iprefix @var{file} -iwithprefix @var{dir} @gol
335 -iwithprefixbefore @var{dir} -isystem @var{dir} @gol
336 -M -MM -MF -MG -MP -MQ -MT -nostdinc @gol
337 -P -fworking-directory -remap @gol
338 -trigraphs -undef -U@var{macro} -Wp,@var{option} @gol
339 -Xpreprocessor @var{option}}
341 @item Assembler Option
342 @xref{Assembler Options,,Passing Options to the Assembler}.
343 @gccoptlist{-Wa,@var{option} -Xassembler @var{option}}
346 @xref{Link Options,,Options for Linking}.
347 @gccoptlist{@var{object-file-name} -l@var{library} @gol
348 -nostartfiles -nodefaultlibs -nostdlib -pie @gol
349 -s -static -static-libgcc -shared -shared-libgcc -symbolic @gol
350 -Wl,@var{option} -Xlinker @var{option} @gol
353 @item Directory Options
354 @xref{Directory Options,,Options for Directory Search}.
355 @gccoptlist{-B@var{prefix} -I@var{dir} -iquote@var{dir} -L@var{dir} -specs=@var{file} -I-}
358 @c I wrote this xref this way to avoid overfull hbox. -- rms
359 @xref{Target Options}.
360 @gccoptlist{-V @var{version} -b @var{machine}}
362 @item Machine Dependent Options
363 @xref{Submodel Options,,Hardware Models and Configurations}.
364 @c This list is ordered alphanumerically by subsection name.
365 @c Try and put the significant identifier (CPU or system) first,
366 @c so users have a clue at guessing where the ones they want will be.
369 @gccoptlist{-EB -EL @gol
370 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
371 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
374 @gccoptlist{-mapcs-frame -mno-apcs-frame @gol
375 -mabi=@var{name} @gol
376 -mapcs-stack-check -mno-apcs-stack-check @gol
377 -mapcs-float -mno-apcs-float @gol
378 -mapcs-reentrant -mno-apcs-reentrant @gol
379 -msched-prolog -mno-sched-prolog @gol
380 -mlittle-endian -mbig-endian -mwords-little-endian @gol
381 -mfloat-abi=@var{name} -msoft-float -mhard-float -mfpe @gol
382 -mthumb-interwork -mno-thumb-interwork @gol
383 -mcpu=@var{name} -march=@var{name} -mfpu=@var{name} @gol
384 -mstructure-size-boundary=@var{n} @gol
385 -mabort-on-noreturn @gol
386 -mlong-calls -mno-long-calls @gol
387 -msingle-pic-base -mno-single-pic-base @gol
388 -mpic-register=@var{reg} @gol
389 -mnop-fun-dllimport @gol
390 -mcirrus-fix-invalid-insns -mno-cirrus-fix-invalid-insns @gol
391 -mpoke-function-name @gol
393 -mtpcs-frame -mtpcs-leaf-frame @gol
394 -mcaller-super-interworking -mcallee-super-interworking}
397 @gccoptlist{-mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts @gol
398 -mcall-prologues -mno-tablejump -mtiny-stack -mint8}
401 @gccoptlist{-mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
402 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
403 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
404 -mstack-align -mdata-align -mconst-align @gol
405 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
406 -melf -maout -melinux -mlinux -sim -sim2 @gol
407 -mmul-bug-workaround -mno-mul-bug-workaround}
409 @emph{Darwin Options}
410 @gccoptlist{-all_load -allowable_client -arch -arch_errors_fatal @gol
411 -arch_only -bind_at_load -bundle -bundle_loader @gol
412 -client_name -compatibility_version -current_version @gol
414 -dependency-file -dylib_file -dylinker_install_name @gol
415 -dynamic -dynamiclib -exported_symbols_list @gol
416 -filelist -flat_namespace -force_cpusubtype_ALL @gol
417 -force_flat_namespace -headerpad_max_install_names @gol
418 -image_base -init -install_name -keep_private_externs @gol
419 -multi_module -multiply_defined -multiply_defined_unused @gol
420 -noall_load -no_dead_strip_inits_and_terms @gol
421 -nofixprebinding -nomultidefs -noprebind -noseglinkedit @gol
422 -pagezero_size -prebind -prebind_all_twolevel_modules @gol
423 -private_bundle -read_only_relocs -sectalign @gol
424 -sectobjectsymbols -whyload -seg1addr @gol
425 -sectcreate -sectobjectsymbols -sectorder @gol
426 -segaddr -segs_read_only_addr -segs_read_write_addr @gol
427 -seg_addr_table -seg_addr_table_filename -seglinkedit @gol
428 -segprot -segs_read_only_addr -segs_read_write_addr @gol
429 -single_module -static -sub_library -sub_umbrella @gol
430 -twolevel_namespace -umbrella -undefined @gol
431 -unexported_symbols_list -weak_reference_mismatches @gol
432 -whatsloaded -F -gused -gfull -mone-byte-bool}
434 @emph{DEC Alpha Options}
435 @gccoptlist{-mno-fp-regs -msoft-float -malpha-as -mgas @gol
436 -mieee -mieee-with-inexact -mieee-conformant @gol
437 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
438 -mtrap-precision=@var{mode} -mbuild-constants @gol
439 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} @gol
440 -mbwx -mmax -mfix -mcix @gol
441 -mfloat-vax -mfloat-ieee @gol
442 -mexplicit-relocs -msmall-data -mlarge-data @gol
443 -msmall-text -mlarge-text @gol
444 -mmemory-latency=@var{time}}
446 @emph{DEC Alpha/VMS Options}
447 @gccoptlist{-mvms-return-codes}
450 @gccoptlist{-mgpr-32 -mgpr-64 -mfpr-32 -mfpr-64 @gol
451 -mhard-float -msoft-float @gol
452 -malloc-cc -mfixed-cc -mdword -mno-dword @gol
453 -mdouble -mno-double @gol
454 -mmedia -mno-media -mmuladd -mno-muladd @gol
455 -mfdpic -minline-plt -mgprel-ro -multilib-library-pic @gol
456 -mlinked-fp -mlong-calls -malign-labels @gol
457 -mlibrary-pic -macc-4 -macc-8 @gol
458 -mpack -mno-pack -mno-eflags -mcond-move -mno-cond-move @gol
459 -mscc -mno-scc -mcond-exec -mno-cond-exec @gol
460 -mvliw-branch -mno-vliw-branch @gol
461 -mmulti-cond-exec -mno-multi-cond-exec -mnested-cond-exec @gol
462 -mno-nested-cond-exec -mtomcat-stats @gol
465 @emph{H8/300 Options}
466 @gccoptlist{-mrelax -mh -ms -mn -mint32 -malign-300}
469 @gccoptlist{-march=@var{architecture-type} @gol
470 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
471 -mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
472 -mfixed-range=@var{register-range} @gol
473 -mjump-in-delay -mlinker-opt -mlong-calls @gol
474 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
475 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
476 -mno-jump-in-delay -mno-long-load-store @gol
477 -mno-portable-runtime -mno-soft-float @gol
478 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
479 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
480 -mschedule=@var{cpu-type} -mspace-regs -msio -mwsio @gol
481 -munix=@var{unix-std} -nolibdld -static -threads}
483 @emph{i386 and x86-64 Options}
484 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
485 -mfpmath=@var{unit} @gol
486 -masm=@var{dialect} -mno-fancy-math-387 @gol
487 -mno-fp-ret-in-387 -msoft-float -msvr3-shlib @gol
488 -mno-wide-multiply -mrtd -malign-double @gol
489 -mpreferred-stack-boundary=@var{num} @gol
490 -mmmx -msse -msse2 -msse3 -m3dnow @gol
491 -mthreads -mno-align-stringops -minline-all-stringops @gol
492 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
493 -m96bit-long-double -mregparm=@var{num} -momit-leaf-frame-pointer @gol
494 -mno-red-zone -mno-tls-direct-seg-refs @gol
495 -mcmodel=@var{code-model} @gol
499 @gccoptlist{-mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
500 -mvolatile-asm-stop -mb-step -mregister-names -mno-sdata @gol
501 -mconstant-gp -mauto-pic -minline-float-divide-min-latency @gol
502 -minline-float-divide-max-throughput @gol
503 -minline-int-divide-min-latency @gol
504 -minline-int-divide-max-throughput -mno-dwarf2-asm @gol
505 -mfixed-range=@var{register-range}}
507 @emph{M32R/D Options}
508 @gccoptlist{-m32r2 -m32rx -m32r @gol
510 -malign-loops -mno-align-loops @gol
511 -missue-rate=@var{number} @gol
512 -mbranch-cost=@var{number} @gol
513 -mmodel=@var{code-size-model-type} @gol
514 -msdata=@var{sdata-type} @gol
515 -mno-flush-func -mflush-func=@var{name} @gol
516 -mno-flush-trap -mflush-trap=@var{number} @gol
519 @emph{M680x0 Options}
520 @gccoptlist{-m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
521 -m68060 -mcpu32 -m5200 -m68881 -mbitfield -mc68000 -mc68020 @gol
522 -mnobitfield -mrtd -mshort -msoft-float -mpcrel @gol
523 -malign-int -mstrict-align -msep-data -mno-sep-data @gol
524 -mshared-library-id=n -mid-shared-library -mno-id-shared-library}
526 @emph{M68hc1x Options}
527 @gccoptlist{-m6811 -m6812 -m68hc11 -m68hc12 -m68hcs12 @gol
528 -mauto-incdec -minmax -mlong-calls -mshort @gol
529 -msoft-reg-count=@var{count}}
532 @gccoptlist{-mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
533 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
534 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
535 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
536 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
539 @gccoptlist{-EL -EB -march=@var{arch} -mtune=@var{arch} @gol
540 -mips1 -mips2 -mips3 -mips4 -mips32 -mips32r2 -mips64 @gol
541 -mips16 -mno-mips16 -mabi=@var{abi} -mabicalls -mno-abicalls @gol
542 -mxgot -mno-xgot -mgp32 -mgp64 -mfp32 -mfp64 @gol
543 -mhard-float -msoft-float -msingle-float -mdouble-float @gol
544 -mpaired-single -mips3d @gol
545 -mint64 -mlong64 -mlong32 @gol
546 -G@var{num} -membedded-data -mno-embedded-data @gol
547 -muninit-const-in-rodata -mno-uninit-const-in-rodata @gol
548 -msplit-addresses -mno-split-addresses @gol
549 -mexplicit-relocs -mno-explicit-relocs @gol
550 -mcheck-zero-division -mno-check-zero-division @gol
551 -mdivide-traps -mdivide-breaks @gol
552 -mmemcpy -mno-memcpy -mlong-calls -mno-long-calls @gol
553 -mmad -mno-mad -mfused-madd -mno-fused-madd -nocpp @gol
554 -mfix-r4000 -mno-fix-r4000 -mfix-r4400 -mno-fix-r4400 @gol
555 -mfix-vr4120 -mno-fix-vr4120 -mfix-sb1 -mno-fix-sb1 @gol
556 -mflush-func=@var{func} -mno-flush-func @gol
557 -mbranch-likely -mno-branch-likely @gol
558 -mfp-exceptions -mno-fp-exceptions @gol
559 -mvr4130-align -mno-vr4130-align}
562 @gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
563 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
564 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
565 -mno-base-addresses -msingle-exit -mno-single-exit}
567 @emph{MN10300 Options}
568 @gccoptlist{-mmult-bug -mno-mult-bug @gol
569 -mam33 -mno-am33 @gol
570 -mam33-2 -mno-am33-2 @gol
574 @gccoptlist{-m32032 -m32332 -m32532 -m32081 -m32381 @gol
575 -mmult-add -mnomult-add -msoft-float -mrtd -mnortd @gol
576 -mregparam -mnoregparam -msb -mnosb @gol
577 -mbitfield -mnobitfield -mhimem -mnohimem}
579 @emph{PDP-11 Options}
580 @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
581 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
582 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
583 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
584 -mbranch-expensive -mbranch-cheap @gol
585 -msplit -mno-split -munix-asm -mdec-asm}
587 @emph{PowerPC Options}
588 See RS/6000 and PowerPC Options.
590 @emph{RS/6000 and PowerPC Options}
591 @gccoptlist{-mcpu=@var{cpu-type} @gol
592 -mtune=@var{cpu-type} @gol
593 -mpower -mno-power -mpower2 -mno-power2 @gol
594 -mpowerpc -mpowerpc64 -mno-powerpc @gol
595 -maltivec -mno-altivec @gol
596 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
597 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
598 -mnew-mnemonics -mold-mnemonics @gol
599 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
600 -m64 -m32 -mxl-call -mno-xl-call -mpe @gol
601 -malign-power -malign-natural @gol
602 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
603 -mstring -mno-string -mupdate -mno-update @gol
604 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
605 -mstrict-align -mno-strict-align -mrelocatable @gol
606 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
607 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
608 -mdynamic-no-pic @gol
609 -mprioritize-restricted-insns=@var{priority} @gol
610 -msched-costly-dep=@var{dependence_type} @gol
611 -minsert-sched-nops=@var{scheme} @gol
612 -mcall-sysv -mcall-netbsd @gol
613 -maix-struct-return -msvr4-struct-return @gol
614 -mabi=altivec -mabi=no-altivec @gol
615 -mabi=spe -mabi=no-spe @gol
616 -misel=yes -misel=no @gol
617 -mspe=yes -mspe=no @gol
618 -mfloat-gprs=yes -mfloat-gprs=no @gol
619 -mprototype -mno-prototype @gol
620 -msim -mmvme -mads -myellowknife -memb -msdata @gol
621 -msdata=@var{opt} -mvxworks -mwindiss -G @var{num} -pthread}
623 @emph{S/390 and zSeries Options}
624 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
625 -mhard-float -msoft-float -mbackchain -mno-backchain -mkernel-backchain @gol
626 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
627 -m64 -m31 -mdebug -mno-debug -mesa -mzarch @gol
628 -mtpf-trace -mno-tpf-trace -mfused-madd -mno-fused-madd @gol
629 -mwarn-framesize -mwarn-dynamicstack -mstack-size -mstack-guard}
632 @gccoptlist{-m1 -m2 -m2e -m3 -m3e @gol
633 -m4-nofpu -m4-single-only -m4-single -m4 @gol
634 -m4a-nofpu -m4a-single-only -m4a-single -m4a -m4al @gol
635 -m5-64media -m5-64media-nofpu @gol
636 -m5-32media -m5-32media-nofpu @gol
637 -m5-compact -m5-compact-nofpu @gol
638 -mb -ml -mdalign -mrelax @gol
639 -mbigtable -mfmovd -mhitachi -mrenesas -mno-renesas -mnomacsave @gol
640 -mieee -misize -mpadstruct -mspace @gol
641 -mprefergot -musermode}
644 @gccoptlist{-mcpu=@var{cpu-type} @gol
645 -mtune=@var{cpu-type} @gol
646 -mcmodel=@var{code-model} @gol
647 -m32 -m64 -mapp-regs -mno-app-regs @gol
648 -mfaster-structs -mno-faster-structs @gol
649 -mfpu -mno-fpu -mhard-float -msoft-float @gol
650 -mhard-quad-float -msoft-quad-float @gol
651 -mimpure-text -mno-impure-text -mlittle-endian @gol
652 -mstack-bias -mno-stack-bias @gol
653 -munaligned-doubles -mno-unaligned-doubles @gol
654 -mv8plus -mno-v8plus -mvis -mno-vis
657 @emph{System V Options}
658 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
660 @emph{TMS320C3x/C4x Options}
661 @gccoptlist{-mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm @gol
662 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload @gol
663 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned @gol
664 -mparallel-insns -mparallel-mpy -mpreserve-float}
667 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
668 -mprolog-function -mno-prolog-function -mspace @gol
669 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
670 -mapp-regs -mno-app-regs @gol
671 -mdisable-callt -mno-disable-callt @gol
677 @gccoptlist{-mg -mgnu -munix}
679 @emph{x86-64 Options}
680 See i386 and x86-64 Options.
682 @emph{Xstormy16 Options}
685 @emph{Xtensa Options}
686 @gccoptlist{-mconst16 -mno-const16 @gol
687 -mfused-madd -mno-fused-madd @gol
688 -mtext-section-literals -mno-text-section-literals @gol
689 -mtarget-align -mno-target-align @gol
690 -mlongcalls -mno-longcalls}
692 @emph{zSeries Options}
693 See S/390 and zSeries Options.
695 @item Code Generation Options
696 @xref{Code Gen Options,,Options for Code Generation Conventions}.
697 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
698 -ffixed-@var{reg} -fexceptions @gol
699 -fnon-call-exceptions -funwind-tables @gol
700 -fasynchronous-unwind-tables @gol
701 -finhibit-size-directive -finstrument-functions @gol
702 -fno-common -fno-ident @gol
703 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
704 -freg-struct-return -fshared-data -fshort-enums @gol
705 -fshort-double -fshort-wchar @gol
706 -fverbose-asm -fpack-struct[=@var{n}] -fstack-check @gol
707 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
708 -fargument-alias -fargument-noalias @gol
709 -fargument-noalias-global -fleading-underscore @gol
710 -ftls-model=@var{model} @gol
711 -ftrapv -fwrapv -fbounds-check @gol
716 * Overall Options:: Controlling the kind of output:
717 an executable, object files, assembler files,
718 or preprocessed source.
719 * C Dialect Options:: Controlling the variant of C language compiled.
720 * C++ Dialect Options:: Variations on C++.
721 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
723 * Language Independent Options:: Controlling how diagnostics should be
725 * Warning Options:: How picky should the compiler be?
726 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
727 * Optimize Options:: How much optimization?
728 * Preprocessor Options:: Controlling header files and macro definitions.
729 Also, getting dependency information for Make.
730 * Assembler Options:: Passing options to the assembler.
731 * Link Options:: Specifying libraries and so on.
732 * Directory Options:: Where to find header files and libraries.
733 Where to find the compiler executable files.
734 * Spec Files:: How to pass switches to sub-processes.
735 * Target Options:: Running a cross-compiler, or an old version of GCC.
738 @node Overall Options
739 @section Options Controlling the Kind of Output
741 Compilation can involve up to four stages: preprocessing, compilation
742 proper, assembly and linking, always in that order. GCC is capable of
743 preprocessing and compiling several files either into several
744 assembler input files, or into one assembler input file; then each
745 assembler input file produces an object file, and linking combines all
746 the object files (those newly compiled, and those specified as input)
747 into an executable file.
749 @cindex file name suffix
750 For any given input file, the file name suffix determines what kind of
755 C source code which must be preprocessed.
758 C source code which should not be preprocessed.
761 C++ source code which should not be preprocessed.
764 Objective-C source code. Note that you must link with the @file{libobjc}
765 library to make an Objective-C program work.
768 Objective-C source code which should not be preprocessed.
772 Objective-C++ source code. Note that you must link with the @file{libobjc}
773 library to make an Objective-C++ program work. Note that @samp{.M} refers
774 to a literal capital M@.
777 Objective-C++ source code which should not be preprocessed.
780 C, C++, Objective-C or Objective-C++ header file to be turned into a
785 @itemx @var{file}.cxx
786 @itemx @var{file}.cpp
787 @itemx @var{file}.CPP
788 @itemx @var{file}.c++
790 C++ source code which must be preprocessed. Note that in @samp{.cxx},
791 the last two letters must both be literally @samp{x}. Likewise,
792 @samp{.C} refers to a literal capital C@.
796 C++ header file to be turned into a precompiled header.
799 @itemx @var{file}.for
800 @itemx @var{file}.FOR
801 Fortran source code which should not be preprocessed.
804 @itemx @var{file}.fpp
805 @itemx @var{file}.FPP
806 Fortran source code which must be preprocessed (with the traditional
810 Fortran source code which must be preprocessed with a RATFOR
811 preprocessor (not included with GCC)@.
814 @itemx @var{file}.f95
815 Fortran 90/95 source code which should not be preprocessed.
817 @c FIXME: Descriptions of Java file types.
824 Ada source code file which contains a library unit declaration (a
825 declaration of a package, subprogram, or generic, or a generic
826 instantiation), or a library unit renaming declaration (a package,
827 generic, or subprogram renaming declaration). Such files are also
830 @itemx @var{file}.adb
831 Ada source code file containing a library unit body (a subprogram or
832 package body). Such files are also called @dfn{bodies}.
834 @c GCC also knows about some suffixes for languages not yet included:
843 Assembler code which must be preprocessed.
846 An object file to be fed straight into linking.
847 Any file name with no recognized suffix is treated this way.
851 You can specify the input language explicitly with the @option{-x} option:
854 @item -x @var{language}
855 Specify explicitly the @var{language} for the following input files
856 (rather than letting the compiler choose a default based on the file
857 name suffix). This option applies to all following input files until
858 the next @option{-x} option. Possible values for @var{language} are:
860 c c-header c-cpp-output
861 c++ c++-header c++-cpp-output
862 objective-c objective-c-header objective-c-cpp-output
863 objective-c++ objective-c++-header objective-c++-cpp-output
864 assembler assembler-with-cpp
866 f77 f77-cpp-input ratfor
873 Turn off any specification of a language, so that subsequent files are
874 handled according to their file name suffixes (as they are if @option{-x}
875 has not been used at all).
877 @item -pass-exit-codes
878 @opindex pass-exit-codes
879 Normally the @command{gcc} program will exit with the code of 1 if any
880 phase of the compiler returns a non-success return code. If you specify
881 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
882 numerically highest error produced by any phase that returned an error
886 If you only want some of the stages of compilation, you can use
887 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
888 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
889 @command{gcc} is to stop. Note that some combinations (for example,
890 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
895 Compile or assemble the source files, but do not link. The linking
896 stage simply is not done. The ultimate output is in the form of an
897 object file for each source file.
899 By default, the object file name for a source file is made by replacing
900 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
902 Unrecognized input files, not requiring compilation or assembly, are
907 Stop after the stage of compilation proper; do not assemble. The output
908 is in the form of an assembler code file for each non-assembler input
911 By default, the assembler file name for a source file is made by
912 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
914 Input files that don't require compilation are ignored.
918 Stop after the preprocessing stage; do not run the compiler proper. The
919 output is in the form of preprocessed source code, which is sent to the
922 Input files which don't require preprocessing are ignored.
924 @cindex output file option
927 Place output in file @var{file}. This applies regardless to whatever
928 sort of output is being produced, whether it be an executable file,
929 an object file, an assembler file or preprocessed C code.
931 If @option{-o} is not specified, the default is to put an executable
932 file in @file{a.out}, the object file for
933 @file{@var{source}.@var{suffix}} in @file{@var{source}.o}, its
934 assembler file in @file{@var{source}.s}, a precompiled header file in
935 @file{@var{source}.@var{suffix}.gch}, and all preprocessed C source on
940 Print (on standard error output) the commands executed to run the stages
941 of compilation. Also print the version number of the compiler driver
942 program and of the preprocessor and the compiler proper.
946 Like @option{-v} except the commands are not executed and all command
947 arguments are quoted. This is useful for shell scripts to capture the
948 driver-generated command lines.
952 Use pipes rather than temporary files for communication between the
953 various stages of compilation. This fails to work on some systems where
954 the assembler is unable to read from a pipe; but the GNU assembler has
959 If you are compiling multiple source files, this option tells the driver
960 to pass all the source files to the compiler at once (for those
961 languages for which the compiler can handle this). This will allow
962 intermodule analysis (IMA) to be performed by the compiler. Currently the only
963 language for which this is supported is C. If you pass source files for
964 multiple languages to the driver, using this option, the driver will invoke
965 the compiler(s) that support IMA once each, passing each compiler all the
966 source files appropriate for it. For those languages that do not support
967 IMA this option will be ignored, and the compiler will be invoked once for
968 each source file in that language. If you use this option in conjunction
969 with -save-temps, the compiler will generate multiple pre-processed files
970 (one for each source file), but only one (combined) .o or .s file.
974 Print (on the standard output) a description of the command line options
975 understood by @command{gcc}. If the @option{-v} option is also specified
976 then @option{--help} will also be passed on to the various processes
977 invoked by @command{gcc}, so that they can display the command line options
978 they accept. If the @option{-Wextra} option is also specified then command
979 line options which have no documentation associated with them will also
984 Print (on the standard output) a description of target specific command
985 line options for each tool.
989 Display the version number and copyrights of the invoked GCC.
993 @section Compiling C++ Programs
995 @cindex suffixes for C++ source
996 @cindex C++ source file suffixes
997 C++ source files conventionally use one of the suffixes @samp{.C},
998 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
999 @samp{.cxx}; C++ header files often use @samp{.hh} or @samp{.H}; and
1000 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
1001 files with these names and compiles them as C++ programs even if you
1002 call the compiler the same way as for compiling C programs (usually
1003 with the name @command{gcc}).
1007 However, C++ programs often require class libraries as well as a
1008 compiler that understands the C++ language---and under some
1009 circumstances, you might want to compile programs or header files from
1010 standard input, or otherwise without a suffix that flags them as C++
1011 programs. You might also like to precompile a C header file with a
1012 @samp{.h} extension to be used in C++ compilations. @command{g++} is a
1013 program that calls GCC with the default language set to C++, and
1014 automatically specifies linking against the C++ library. On many
1015 systems, @command{g++} is also installed with the name @command{c++}.
1017 @cindex invoking @command{g++}
1018 When you compile C++ programs, you may specify many of the same
1019 command-line options that you use for compiling programs in any
1020 language; or command-line options meaningful for C and related
1021 languages; or options that are meaningful only for C++ programs.
1022 @xref{C Dialect Options,,Options Controlling C Dialect}, for
1023 explanations of options for languages related to C@.
1024 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
1025 explanations of options that are meaningful only for C++ programs.
1027 @node C Dialect Options
1028 @section Options Controlling C Dialect
1029 @cindex dialect options
1030 @cindex language dialect options
1031 @cindex options, dialect
1033 The following options control the dialect of C (or languages derived
1034 from C, such as C++, Objective-C and Objective-C++) that the compiler
1038 @cindex ANSI support
1042 In C mode, support all ISO C90 programs. In C++ mode,
1043 remove GNU extensions that conflict with ISO C++.
1045 This turns off certain features of GCC that are incompatible with ISO
1046 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
1047 such as the @code{asm} and @code{typeof} keywords, and
1048 predefined macros such as @code{unix} and @code{vax} that identify the
1049 type of system you are using. It also enables the undesirable and
1050 rarely used ISO trigraph feature. For the C compiler,
1051 it disables recognition of C++ style @samp{//} comments as well as
1052 the @code{inline} keyword.
1054 The alternate keywords @code{__asm__}, @code{__extension__},
1055 @code{__inline__} and @code{__typeof__} continue to work despite
1056 @option{-ansi}. You would not want to use them in an ISO C program, of
1057 course, but it is useful to put them in header files that might be included
1058 in compilations done with @option{-ansi}. Alternate predefined macros
1059 such as @code{__unix__} and @code{__vax__} are also available, with or
1060 without @option{-ansi}.
1062 The @option{-ansi} option does not cause non-ISO programs to be
1063 rejected gratuitously. For that, @option{-pedantic} is required in
1064 addition to @option{-ansi}. @xref{Warning Options}.
1066 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1067 option is used. Some header files may notice this macro and refrain
1068 from declaring certain functions or defining certain macros that the
1069 ISO standard doesn't call for; this is to avoid interfering with any
1070 programs that might use these names for other things.
1072 Functions which would normally be built in but do not have semantics
1073 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1074 functions with @option{-ansi} is used. @xref{Other Builtins,,Other
1075 built-in functions provided by GCC}, for details of the functions
1080 Determine the language standard. This option is currently only
1081 supported when compiling C or C++. A value for this option must be
1082 provided; possible values are
1087 ISO C90 (same as @option{-ansi}).
1089 @item iso9899:199409
1090 ISO C90 as modified in amendment 1.
1096 ISO C99. Note that this standard is not yet fully supported; see
1097 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1098 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1101 Default, ISO C90 plus GNU extensions (including some C99 features).
1105 ISO C99 plus GNU extensions. When ISO C99 is fully implemented in GCC,
1106 this will become the default. The name @samp{gnu9x} is deprecated.
1109 The 1998 ISO C++ standard plus amendments.
1112 The same as @option{-std=c++98} plus GNU extensions. This is the
1113 default for C++ code.
1116 Even when this option is not specified, you can still use some of the
1117 features of newer standards in so far as they do not conflict with
1118 previous C standards. For example, you may use @code{__restrict__} even
1119 when @option{-std=c99} is not specified.
1121 The @option{-std} options specifying some version of ISO C have the same
1122 effects as @option{-ansi}, except that features that were not in ISO C90
1123 but are in the specified version (for example, @samp{//} comments and
1124 the @code{inline} keyword in ISO C99) are not disabled.
1126 @xref{Standards,,Language Standards Supported by GCC}, for details of
1127 these standard versions.
1129 @item -aux-info @var{filename}
1131 Output to the given filename prototyped declarations for all functions
1132 declared and/or defined in a translation unit, including those in header
1133 files. This option is silently ignored in any language other than C@.
1135 Besides declarations, the file indicates, in comments, the origin of
1136 each declaration (source file and line), whether the declaration was
1137 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1138 @samp{O} for old, respectively, in the first character after the line
1139 number and the colon), and whether it came from a declaration or a
1140 definition (@samp{C} or @samp{F}, respectively, in the following
1141 character). In the case of function definitions, a K&R-style list of
1142 arguments followed by their declarations is also provided, inside
1143 comments, after the declaration.
1147 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1148 keyword, so that code can use these words as identifiers. You can use
1149 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1150 instead. @option{-ansi} implies @option{-fno-asm}.
1152 In C++, this switch only affects the @code{typeof} keyword, since
1153 @code{asm} and @code{inline} are standard keywords. You may want to
1154 use the @option{-fno-gnu-keywords} flag instead, which has the same
1155 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1156 switch only affects the @code{asm} and @code{typeof} keywords, since
1157 @code{inline} is a standard keyword in ISO C99.
1160 @itemx -fno-builtin-@var{function}
1161 @opindex fno-builtin
1162 @cindex built-in functions
1163 Don't recognize built-in functions that do not begin with
1164 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1165 functions provided by GCC}, for details of the functions affected,
1166 including those which are not built-in functions when @option{-ansi} or
1167 @option{-std} options for strict ISO C conformance are used because they
1168 do not have an ISO standard meaning.
1170 GCC normally generates special code to handle certain built-in functions
1171 more efficiently; for instance, calls to @code{alloca} may become single
1172 instructions that adjust the stack directly, and calls to @code{memcpy}
1173 may become inline copy loops. The resulting code is often both smaller
1174 and faster, but since the function calls no longer appear as such, you
1175 cannot set a breakpoint on those calls, nor can you change the behavior
1176 of the functions by linking with a different library.
1178 With the @option{-fno-builtin-@var{function}} option
1179 only the built-in function @var{function} is
1180 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1181 function is named this is not built-in in this version of GCC, this
1182 option is ignored. There is no corresponding
1183 @option{-fbuiltin-@var{function}} option; if you wish to enable
1184 built-in functions selectively when using @option{-fno-builtin} or
1185 @option{-ffreestanding}, you may define macros such as:
1188 #define abs(n) __builtin_abs ((n))
1189 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1194 @cindex hosted environment
1196 Assert that compilation takes place in a hosted environment. This implies
1197 @option{-fbuiltin}. A hosted environment is one in which the
1198 entire standard library is available, and in which @code{main} has a return
1199 type of @code{int}. Examples are nearly everything except a kernel.
1200 This is equivalent to @option{-fno-freestanding}.
1202 @item -ffreestanding
1203 @opindex ffreestanding
1204 @cindex hosted environment
1206 Assert that compilation takes place in a freestanding environment. This
1207 implies @option{-fno-builtin}. A freestanding environment
1208 is one in which the standard library may not exist, and program startup may
1209 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1210 This is equivalent to @option{-fno-hosted}.
1212 @xref{Standards,,Language Standards Supported by GCC}, for details of
1213 freestanding and hosted environments.
1215 @item -fms-extensions
1216 @opindex fms-extensions
1217 Accept some non-standard constructs used in Microsoft header files.
1221 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1222 options for strict ISO C conformance) implies @option{-trigraphs}.
1224 @item -no-integrated-cpp
1225 @opindex no-integrated-cpp
1226 Performs a compilation in two passes: preprocessing and compiling. This
1227 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1228 @option{-B} option. The user supplied compilation step can then add in
1229 an additional preprocessing step after normal preprocessing but before
1230 compiling. The default is to use the integrated cpp (internal cpp)
1232 The semantics of this option will change if "cc1", "cc1plus", and
1233 "cc1obj" are merged.
1235 @cindex traditional C language
1236 @cindex C language, traditional
1238 @itemx -traditional-cpp
1239 @opindex traditional-cpp
1240 @opindex traditional
1241 Formerly, these options caused GCC to attempt to emulate a pre-standard
1242 C compiler. They are now only supported with the @option{-E} switch.
1243 The preprocessor continues to support a pre-standard mode. See the GNU
1244 CPP manual for details.
1246 @item -fcond-mismatch
1247 @opindex fcond-mismatch
1248 Allow conditional expressions with mismatched types in the second and
1249 third arguments. The value of such an expression is void. This option
1250 is not supported for C++.
1252 @item -funsigned-char
1253 @opindex funsigned-char
1254 Let the type @code{char} be unsigned, like @code{unsigned char}.
1256 Each kind of machine has a default for what @code{char} should
1257 be. It is either like @code{unsigned char} by default or like
1258 @code{signed char} by default.
1260 Ideally, a portable program should always use @code{signed char} or
1261 @code{unsigned char} when it depends on the signedness of an object.
1262 But many programs have been written to use plain @code{char} and
1263 expect it to be signed, or expect it to be unsigned, depending on the
1264 machines they were written for. This option, and its inverse, let you
1265 make such a program work with the opposite default.
1267 The type @code{char} is always a distinct type from each of
1268 @code{signed char} or @code{unsigned char}, even though its behavior
1269 is always just like one of those two.
1272 @opindex fsigned-char
1273 Let the type @code{char} be signed, like @code{signed char}.
1275 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1276 the negative form of @option{-funsigned-char}. Likewise, the option
1277 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1279 @item -fsigned-bitfields
1280 @itemx -funsigned-bitfields
1281 @itemx -fno-signed-bitfields
1282 @itemx -fno-unsigned-bitfields
1283 @opindex fsigned-bitfields
1284 @opindex funsigned-bitfields
1285 @opindex fno-signed-bitfields
1286 @opindex fno-unsigned-bitfields
1287 These options control whether a bit-field is signed or unsigned, when the
1288 declaration does not use either @code{signed} or @code{unsigned}. By
1289 default, such a bit-field is signed, because this is consistent: the
1290 basic integer types such as @code{int} are signed types.
1293 @node C++ Dialect Options
1294 @section Options Controlling C++ Dialect
1296 @cindex compiler options, C++
1297 @cindex C++ options, command line
1298 @cindex options, C++
1299 This section describes the command-line options that are only meaningful
1300 for C++ programs; but you can also use most of the GNU compiler options
1301 regardless of what language your program is in. For example, you
1302 might compile a file @code{firstClass.C} like this:
1305 g++ -g -frepo -O -c firstClass.C
1309 In this example, only @option{-frepo} is an option meant
1310 only for C++ programs; you can use the other options with any
1311 language supported by GCC@.
1313 Here is a list of options that are @emph{only} for compiling C++ programs:
1317 @item -fabi-version=@var{n}
1318 @opindex fabi-version
1319 Use version @var{n} of the C++ ABI. Version 2 is the version of the
1320 C++ ABI that first appeared in G++ 3.4. Version 1 is the version of
1321 the C++ ABI that first appeared in G++ 3.2. Version 0 will always be
1322 the version that conforms most closely to the C++ ABI specification.
1323 Therefore, the ABI obtained using version 0 will change as ABI bugs
1326 The default is version 2.
1328 @item -fno-access-control
1329 @opindex fno-access-control
1330 Turn off all access checking. This switch is mainly useful for working
1331 around bugs in the access control code.
1335 Check that the pointer returned by @code{operator new} is non-null
1336 before attempting to modify the storage allocated. This check is
1337 normally unnecessary because the C++ standard specifies that
1338 @code{operator new} will only return @code{0} if it is declared
1339 @samp{throw()}, in which case the compiler will always check the
1340 return value even without this option. In all other cases, when
1341 @code{operator new} has a non-empty exception specification, memory
1342 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1343 @samp{new (nothrow)}.
1345 @item -fconserve-space
1346 @opindex fconserve-space
1347 Put uninitialized or runtime-initialized global variables into the
1348 common segment, as C does. This saves space in the executable at the
1349 cost of not diagnosing duplicate definitions. If you compile with this
1350 flag and your program mysteriously crashes after @code{main()} has
1351 completed, you may have an object that is being destroyed twice because
1352 two definitions were merged.
1354 This option is no longer useful on most targets, now that support has
1355 been added for putting variables into BSS without making them common.
1357 @item -fno-const-strings
1358 @opindex fno-const-strings
1359 Give string constants type @code{char *} instead of type @code{const
1360 char *}. By default, G++ uses type @code{const char *} as required by
1361 the standard. Even if you use @option{-fno-const-strings}, you cannot
1362 actually modify the value of a string constant.
1364 This option might be removed in a future release of G++. For maximum
1365 portability, you should structure your code so that it works with
1366 string constants that have type @code{const char *}.
1368 @item -fno-elide-constructors
1369 @opindex fno-elide-constructors
1370 The C++ standard allows an implementation to omit creating a temporary
1371 which is only used to initialize another object of the same type.
1372 Specifying this option disables that optimization, and forces G++ to
1373 call the copy constructor in all cases.
1375 @item -fno-enforce-eh-specs
1376 @opindex fno-enforce-eh-specs
1377 Don't check for violation of exception specifications at runtime. This
1378 option violates the C++ standard, but may be useful for reducing code
1379 size in production builds, much like defining @samp{NDEBUG}. The compiler
1380 will still optimize based on the exception specifications.
1383 @itemx -fno-for-scope
1385 @opindex fno-for-scope
1386 If @option{-ffor-scope} is specified, the scope of variables declared in
1387 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1388 as specified by the C++ standard.
1389 If @option{-fno-for-scope} is specified, the scope of variables declared in
1390 a @i{for-init-statement} extends to the end of the enclosing scope,
1391 as was the case in old versions of G++, and other (traditional)
1392 implementations of C++.
1394 The default if neither flag is given to follow the standard,
1395 but to allow and give a warning for old-style code that would
1396 otherwise be invalid, or have different behavior.
1398 @item -fno-gnu-keywords
1399 @opindex fno-gnu-keywords
1400 Do not recognize @code{typeof} as a keyword, so that code can use this
1401 word as an identifier. You can use the keyword @code{__typeof__} instead.
1402 @option{-ansi} implies @option{-fno-gnu-keywords}.
1404 @item -fno-implicit-templates
1405 @opindex fno-implicit-templates
1406 Never emit code for non-inline templates which are instantiated
1407 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1408 @xref{Template Instantiation}, for more information.
1410 @item -fno-implicit-inline-templates
1411 @opindex fno-implicit-inline-templates
1412 Don't emit code for implicit instantiations of inline templates, either.
1413 The default is to handle inlines differently so that compiles with and
1414 without optimization will need the same set of explicit instantiations.
1416 @item -fno-implement-inlines
1417 @opindex fno-implement-inlines
1418 To save space, do not emit out-of-line copies of inline functions
1419 controlled by @samp{#pragma implementation}. This will cause linker
1420 errors if these functions are not inlined everywhere they are called.
1422 @item -fms-extensions
1423 @opindex fms-extensions
1424 Disable pedantic warnings about constructs used in MFC, such as implicit
1425 int and getting a pointer to member function via non-standard syntax.
1427 @item -fno-nonansi-builtins
1428 @opindex fno-nonansi-builtins
1429 Disable built-in declarations of functions that are not mandated by
1430 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1431 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1433 @item -fno-operator-names
1434 @opindex fno-operator-names
1435 Do not treat the operator name keywords @code{and}, @code{bitand},
1436 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1437 synonyms as keywords.
1439 @item -fno-optional-diags
1440 @opindex fno-optional-diags
1441 Disable diagnostics that the standard says a compiler does not need to
1442 issue. Currently, the only such diagnostic issued by G++ is the one for
1443 a name having multiple meanings within a class.
1446 @opindex fpermissive
1447 Downgrade some diagnostics about nonconformant code from errors to
1448 warnings. Thus, using @option{-fpermissive} will allow some
1449 nonconforming code to compile.
1453 Enable automatic template instantiation at link time. This option also
1454 implies @option{-fno-implicit-templates}. @xref{Template
1455 Instantiation}, for more information.
1459 Disable generation of information about every class with virtual
1460 functions for use by the C++ runtime type identification features
1461 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1462 of the language, you can save some space by using this flag. Note that
1463 exception handling uses the same information, but it will generate it as
1468 Emit statistics about front-end processing at the end of the compilation.
1469 This information is generally only useful to the G++ development team.
1471 @item -ftemplate-depth-@var{n}
1472 @opindex ftemplate-depth
1473 Set the maximum instantiation depth for template classes to @var{n}.
1474 A limit on the template instantiation depth is needed to detect
1475 endless recursions during template class instantiation. ANSI/ISO C++
1476 conforming programs must not rely on a maximum depth greater than 17.
1478 @item -fno-threadsafe-statics
1479 @opindex fno-threadsafe-statics
1480 Do not emit the extra code to use the routines specified in the C++
1481 ABI for thread-safe initialization of local statics. You can use this
1482 option to reduce code size slightly in code that doesn't need to be
1485 @item -fuse-cxa-atexit
1486 @opindex fuse-cxa-atexit
1487 Register destructors for objects with static storage duration with the
1488 @code{__cxa_atexit} function rather than the @code{atexit} function.
1489 This option is required for fully standards-compliant handling of static
1490 destructors, but will only work if your C library supports
1491 @code{__cxa_atexit}.
1493 @item -fvisibility-inlines-hidden
1494 @opindex fvisibility-inlines-hidden
1495 Causes all inlined methods to be marked with
1496 @code{__attribute__ ((visibility ("hidden")))} so that they do not
1497 appear in the export table of a DSO and do not require a PLT indirection
1498 when used within the DSO. Enabling this option can have a dramatic effect
1499 on load and link times of a DSO as it massively reduces the size of the
1500 dynamic export table when the library makes heavy use of templates. While
1501 it can cause bloating through duplication of code within each DSO where
1502 it is used, often the wastage is less than the considerable space occupied
1503 by a long symbol name in the export table which is typical when using
1504 templates and namespaces. For even more savings, combine with the
1505 @code{-fvisibility=hidden} switch.
1509 Do not use weak symbol support, even if it is provided by the linker.
1510 By default, G++ will use weak symbols if they are available. This
1511 option exists only for testing, and should not be used by end-users;
1512 it will result in inferior code and has no benefits. This option may
1513 be removed in a future release of G++.
1517 Do not search for header files in the standard directories specific to
1518 C++, but do still search the other standard directories. (This option
1519 is used when building the C++ library.)
1522 In addition, these optimization, warning, and code generation options
1523 have meanings only for C++ programs:
1526 @item -fno-default-inline
1527 @opindex fno-default-inline
1528 Do not assume @samp{inline} for functions defined inside a class scope.
1529 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1530 functions will have linkage like inline functions; they just won't be
1533 @item -Wabi @r{(C++ only)}
1535 Warn when G++ generates code that is probably not compatible with the
1536 vendor-neutral C++ ABI. Although an effort has been made to warn about
1537 all such cases, there are probably some cases that are not warned about,
1538 even though G++ is generating incompatible code. There may also be
1539 cases where warnings are emitted even though the code that is generated
1542 You should rewrite your code to avoid these warnings if you are
1543 concerned about the fact that code generated by G++ may not be binary
1544 compatible with code generated by other compilers.
1546 The known incompatibilities at this point include:
1551 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
1552 pack data into the same byte as a base class. For example:
1555 struct A @{ virtual void f(); int f1 : 1; @};
1556 struct B : public A @{ int f2 : 1; @};
1560 In this case, G++ will place @code{B::f2} into the same byte
1561 as@code{A::f1}; other compilers will not. You can avoid this problem
1562 by explicitly padding @code{A} so that its size is a multiple of the
1563 byte size on your platform; that will cause G++ and other compilers to
1564 layout @code{B} identically.
1567 Incorrect handling of tail-padding for virtual bases. G++ does not use
1568 tail padding when laying out virtual bases. For example:
1571 struct A @{ virtual void f(); char c1; @};
1572 struct B @{ B(); char c2; @};
1573 struct C : public A, public virtual B @{@};
1577 In this case, G++ will not place @code{B} into the tail-padding for
1578 @code{A}; other compilers will. You can avoid this problem by
1579 explicitly padding @code{A} so that its size is a multiple of its
1580 alignment (ignoring virtual base classes); that will cause G++ and other
1581 compilers to layout @code{C} identically.
1584 Incorrect handling of bit-fields with declared widths greater than that
1585 of their underlying types, when the bit-fields appear in a union. For
1589 union U @{ int i : 4096; @};
1593 Assuming that an @code{int} does not have 4096 bits, G++ will make the
1594 union too small by the number of bits in an @code{int}.
1597 Empty classes can be placed at incorrect offsets. For example:
1607 struct C : public B, public A @{@};
1611 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
1612 it should be placed at offset zero. G++ mistakenly believes that the
1613 @code{A} data member of @code{B} is already at offset zero.
1616 Names of template functions whose types involve @code{typename} or
1617 template template parameters can be mangled incorrectly.
1620 template <typename Q>
1621 void f(typename Q::X) @{@}
1623 template <template <typename> class Q>
1624 void f(typename Q<int>::X) @{@}
1628 Instantiations of these templates may be mangled incorrectly.
1632 @item -Wctor-dtor-privacy @r{(C++ only)}
1633 @opindex Wctor-dtor-privacy
1634 Warn when a class seems unusable because all the constructors or
1635 destructors in that class are private, and it has neither friends nor
1636 public static member functions.
1638 @item -Wnon-virtual-dtor @r{(C++ only)}
1639 @opindex Wnon-virtual-dtor
1640 Warn when a class appears to be polymorphic, thereby requiring a virtual
1641 destructor, yet it declares a non-virtual one.
1642 This warning is enabled by @option{-Wall}.
1644 @item -Wreorder @r{(C++ only)}
1646 @cindex reordering, warning
1647 @cindex warning for reordering of member initializers
1648 Warn when the order of member initializers given in the code does not
1649 match the order in which they must be executed. For instance:
1655 A(): j (0), i (1) @{ @}
1659 The compiler will rearrange the member initializers for @samp{i}
1660 and @samp{j} to match the declaration order of the members, emitting
1661 a warning to that effect. This warning is enabled by @option{-Wall}.
1664 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
1667 @item -Weffc++ @r{(C++ only)}
1669 Warn about violations of the following style guidelines from Scott Meyers'
1670 @cite{Effective C++} book:
1674 Item 11: Define a copy constructor and an assignment operator for classes
1675 with dynamically allocated memory.
1678 Item 12: Prefer initialization to assignment in constructors.
1681 Item 14: Make destructors virtual in base classes.
1684 Item 15: Have @code{operator=} return a reference to @code{*this}.
1687 Item 23: Don't try to return a reference when you must return an object.
1691 Also warn about violations of the following style guidelines from
1692 Scott Meyers' @cite{More Effective C++} book:
1696 Item 6: Distinguish between prefix and postfix forms of increment and
1697 decrement operators.
1700 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
1704 When selecting this option, be aware that the standard library
1705 headers do not obey all of these guidelines; use @samp{grep -v}
1706 to filter out those warnings.
1708 @item -Wno-deprecated @r{(C++ only)}
1709 @opindex Wno-deprecated
1710 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1712 @item -Wno-non-template-friend @r{(C++ only)}
1713 @opindex Wno-non-template-friend
1714 Disable warnings when non-templatized friend functions are declared
1715 within a template. Since the advent of explicit template specification
1716 support in G++, if the name of the friend is an unqualified-id (i.e.,
1717 @samp{friend foo(int)}), the C++ language specification demands that the
1718 friend declare or define an ordinary, nontemplate function. (Section
1719 14.5.3). Before G++ implemented explicit specification, unqualified-ids
1720 could be interpreted as a particular specialization of a templatized
1721 function. Because this non-conforming behavior is no longer the default
1722 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
1723 check existing code for potential trouble spots and is on by default.
1724 This new compiler behavior can be turned off with
1725 @option{-Wno-non-template-friend} which keeps the conformant compiler code
1726 but disables the helpful warning.
1728 @item -Wold-style-cast @r{(C++ only)}
1729 @opindex Wold-style-cast
1730 Warn if an old-style (C-style) cast to a non-void type is used within
1731 a C++ program. The new-style casts (@samp{static_cast},
1732 @samp{reinterpret_cast}, and @samp{const_cast}) are less vulnerable to
1733 unintended effects and much easier to search for.
1735 @item -Woverloaded-virtual @r{(C++ only)}
1736 @opindex Woverloaded-virtual
1737 @cindex overloaded virtual fn, warning
1738 @cindex warning for overloaded virtual fn
1739 Warn when a function declaration hides virtual functions from a
1740 base class. For example, in:
1747 struct B: public A @{
1752 the @code{A} class version of @code{f} is hidden in @code{B}, and code
1760 will fail to compile.
1762 @item -Wno-pmf-conversions @r{(C++ only)}
1763 @opindex Wno-pmf-conversions
1764 Disable the diagnostic for converting a bound pointer to member function
1767 @item -Wsign-promo @r{(C++ only)}
1768 @opindex Wsign-promo
1769 Warn when overload resolution chooses a promotion from unsigned or
1770 enumerated type to a signed type, over a conversion to an unsigned type of
1771 the same size. Previous versions of G++ would try to preserve
1772 unsignedness, but the standard mandates the current behavior.
1774 @item -Wsynth @r{(C++ only)}
1776 @cindex warning for synthesized methods
1777 @cindex synthesized methods, warning
1778 Warn when G++'s synthesis behavior does not match that of cfront. For
1784 A& operator = (int);
1794 In this example, G++ will synthesize a default @samp{A& operator =
1795 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1798 @node Objective-C and Objective-C++ Dialect Options
1799 @section Options Controlling Objective-C and Objective-C++ Dialects
1801 @cindex compiler options, Objective-C and Objective-C++
1802 @cindex Objective-C and Objective-C++ options, command line
1803 @cindex options, Objective-C and Objective-C++
1804 (NOTE: This manual does not describe the Objective-C and Objective-C++
1805 languages themselves. See @xref{Standards,,Language Standards
1806 Supported by GCC}, for references.)
1808 This section describes the command-line options that are only meaningful
1809 for Objective-C and Objective-C++ programs, but you can also use most of
1810 the language-independent GNU compiler options.
1811 For example, you might compile a file @code{some_class.m} like this:
1814 gcc -g -fgnu-runtime -O -c some_class.m
1818 In this example, @option{-fgnu-runtime} is an option meant only for
1819 Objective-C and Objective-C++ programs; you can use the other options with
1820 any language supported by GCC@.
1822 Note that since Objective-C is an extension of the C language, Objective-C
1823 compilations may also use options specific to the C front-end (e.g.,
1824 @option{-Wtraditional}). Similarly, Objective-C++ compilations may use
1825 C++-specific options (e.g., @option{-Wabi}).
1827 Here is a list of options that are @emph{only} for compiling Objective-C
1828 and Objective-C++ programs:
1831 @item -fconstant-string-class=@var{class-name}
1832 @opindex fconstant-string-class
1833 Use @var{class-name} as the name of the class to instantiate for each
1834 literal string specified with the syntax @code{@@"@dots{}"}. The default
1835 class name is @code{NXConstantString} if the GNU runtime is being used, and
1836 @code{NSConstantString} if the NeXT runtime is being used (see below). The
1837 @option{-fconstant-cfstrings} option, if also present, will override the
1838 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
1839 to be laid out as constant CoreFoundation strings.
1842 @opindex fgnu-runtime
1843 Generate object code compatible with the standard GNU Objective-C
1844 runtime. This is the default for most types of systems.
1846 @item -fnext-runtime
1847 @opindex fnext-runtime
1848 Generate output compatible with the NeXT runtime. This is the default
1849 for NeXT-based systems, including Darwin and Mac OS X@. The macro
1850 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
1853 @item -fno-nil-receivers
1854 @opindex fno-nil-receivers
1855 Assume that all Objective-C message dispatches (e.g.,
1856 @code{[receiver message:arg]}) in this translation unit ensure that the receiver
1857 is not @code{nil}. This allows for more efficient entry points in the runtime
1858 to be used. Currently, this option is only available in conjunction with
1859 the NeXT runtime on Mac OS X 10.3 and later.
1861 @item -fobjc-exceptions
1862 @opindex fobjc-exceptions
1863 Enable syntactic support for structured exception handling in Objective-C,
1864 similar to what is offered by C++ and Java. Currently, this option is only
1865 available in conjunction with the NeXT runtime on Mac OS X 10.3 and later.
1873 @@catch (AnObjCClass *exc) @{
1880 @@catch (AnotherClass *exc) @{
1883 @@catch (id allOthers) @{
1893 The @code{@@throw} statement may appear anywhere in an Objective-C or
1894 Objective-C++ program; when used inside of a @code{@@catch} block, the
1895 @code{@@throw} may appear without an argument (as shown above), in which case
1896 the object caught by the @code{@@catch} will be rethrown.
1898 Note that only (pointers to) Objective-C objects may be thrown and
1899 caught using this scheme. When an object is thrown, it will be caught
1900 by the nearest @code{@@catch} clause capable of handling objects of that type,
1901 analogously to how @code{catch} blocks work in C++ and Java. A
1902 @code{@@catch(id @dots{})} clause (as shown above) may also be provided to catch
1903 any and all Objective-C exceptions not caught by previous @code{@@catch}
1906 The @code{@@finally} clause, if present, will be executed upon exit from the
1907 immediately preceding @code{@@try @dots{} @@catch} section. This will happen
1908 regardless of whether any exceptions are thrown, caught or rethrown
1909 inside the @code{@@try @dots{} @@catch} section, analogously to the behavior
1910 of the @code{finally} clause in Java.
1912 There are several caveats to using the new exception mechanism:
1916 Although currently designed to be binary compatible with @code{NS_HANDLER}-style
1917 idioms provided by the @code{NSException} class, the new
1918 exceptions can only be used on Mac OS X 10.3 (Panther) and later
1919 systems, due to additional functionality needed in the (NeXT) Objective-C
1923 As mentioned above, the new exceptions do not support handling
1924 types other than Objective-C objects. Furthermore, when used from
1925 Objective-C++, the Objective-C exception model does not interoperate with C++
1926 exceptions at this time. This means you cannot @code{@@throw} an exception
1927 from Objective-C and @code{catch} it in C++, or vice versa
1928 (i.e., @code{throw @dots{} @@catch}).
1931 The @option{-fobjc-exceptions} switch also enables the use of synchronization
1932 blocks for thread-safe execution:
1935 @@synchronized (ObjCClass *guard) @{
1940 Upon entering the @code{@@synchronized} block, a thread of execution shall
1941 first check whether a lock has been placed on the corresponding @code{guard}
1942 object by another thread. If it has, the current thread shall wait until
1943 the other thread relinquishes its lock. Once @code{guard} becomes available,
1944 the current thread will place its own lock on it, execute the code contained in
1945 the @code{@@synchronized} block, and finally relinquish the lock (thereby
1946 making @code{guard} available to other threads).
1948 Unlike Java, Objective-C does not allow for entire methods to be marked
1949 @code{@@synchronized}. Note that throwing exceptions out of
1950 @code{@@synchronized} blocks is allowed, and will cause the guarding object
1951 to be unlocked properly.
1953 @item -freplace-objc-classes
1954 @opindex freplace-objc-classes
1955 Emit a special marker instructing @command{ld(1)} not to statically link in
1956 the resulting object file, and allow @command{dyld(1)} to load it in at
1957 run time instead. This is used in conjunction with the Fix-and-Continue
1958 debugging mode, where the object file in question may be recompiled and
1959 dynamically reloaded in the course of program execution, without the need
1960 to restart the program itself. Currently, Fix-and-Continue functionality
1961 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
1966 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
1967 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
1968 compile time) with static class references that get initialized at load time,
1969 which improves run-time performance. Specifying the @option{-fzero-link} flag
1970 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
1971 to be retained. This is useful in Zero-Link debugging mode, since it allows
1972 for individual class implementations to be modified during program execution.
1976 Dump interface declarations for all classes seen in the source file to a
1977 file named @file{@var{sourcename}.decl}.
1980 @opindex Wno-protocol
1981 If a class is declared to implement a protocol, a warning is issued for
1982 every method in the protocol that is not implemented by the class. The
1983 default behavior is to issue a warning for every method not explicitly
1984 implemented in the class, even if a method implementation is inherited
1985 from the superclass. If you use the @code{-Wno-protocol} option, then
1986 methods inherited from the superclass are considered to be implemented,
1987 and no warning is issued for them.
1991 Warn if multiple methods of different types for the same selector are
1992 found during compilation. The check is performed on the list of methods
1993 in the final stage of compilation. Additionally, a check is performed
1994 for each selector appearing in a @code{@@selector(@dots{})}
1995 expression, and a corresponding method for that selector has been found
1996 during compilation. Because these checks scan the method table only at
1997 the end of compilation, these warnings are not produced if the final
1998 stage of compilation is not reached, for example because an error is
1999 found during compilation, or because the @code{-fsyntax-only} option is
2002 @item -Wundeclared-selector
2003 @opindex Wundeclared-selector
2004 Warn if a @code{@@selector(@dots{})} expression referring to an
2005 undeclared selector is found. A selector is considered undeclared if no
2006 method with that name has been declared before the
2007 @code{@@selector(@dots{})} expression, either explicitly in an
2008 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
2009 an @code{@@implementation} section. This option always performs its
2010 checks as soon as a @code{@@selector(@dots{})} expression is found,
2011 while @code{-Wselector} only performs its checks in the final stage of
2012 compilation. This also enforces the coding style convention
2013 that methods and selectors must be declared before being used.
2015 @item -print-objc-runtime-info
2016 @opindex print-objc-runtime-info
2017 Generate C header describing the largest structure that is passed by
2022 @node Language Independent Options
2023 @section Options to Control Diagnostic Messages Formatting
2024 @cindex options to control diagnostics formatting
2025 @cindex diagnostic messages
2026 @cindex message formatting
2028 Traditionally, diagnostic messages have been formatted irrespective of
2029 the output device's aspect (e.g.@: its width, @dots{}). The options described
2030 below can be used to control the diagnostic messages formatting
2031 algorithm, e.g.@: how many characters per line, how often source location
2032 information should be reported. Right now, only the C++ front end can
2033 honor these options. However it is expected, in the near future, that
2034 the remaining front ends would be able to digest them correctly.
2037 @item -fmessage-length=@var{n}
2038 @opindex fmessage-length
2039 Try to format error messages so that they fit on lines of about @var{n}
2040 characters. The default is 72 characters for @command{g++} and 0 for the rest of
2041 the front ends supported by GCC@. If @var{n} is zero, then no
2042 line-wrapping will be done; each error message will appear on a single
2045 @opindex fdiagnostics-show-location
2046 @item -fdiagnostics-show-location=once
2047 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
2048 reporter to emit @emph{once} source location information; that is, in
2049 case the message is too long to fit on a single physical line and has to
2050 be wrapped, the source location won't be emitted (as prefix) again,
2051 over and over, in subsequent continuation lines. This is the default
2054 @item -fdiagnostics-show-location=every-line
2055 Only meaningful in line-wrapping mode. Instructs the diagnostic
2056 messages reporter to emit the same source location information (as
2057 prefix) for physical lines that result from the process of breaking
2058 a message which is too long to fit on a single line.
2062 @node Warning Options
2063 @section Options to Request or Suppress Warnings
2064 @cindex options to control warnings
2065 @cindex warning messages
2066 @cindex messages, warning
2067 @cindex suppressing warnings
2069 Warnings are diagnostic messages that report constructions which
2070 are not inherently erroneous but which are risky or suggest there
2071 may have been an error.
2073 You can request many specific warnings with options beginning @samp{-W},
2074 for example @option{-Wimplicit} to request warnings on implicit
2075 declarations. Each of these specific warning options also has a
2076 negative form beginning @samp{-Wno-} to turn off warnings;
2077 for example, @option{-Wno-implicit}. This manual lists only one of the
2078 two forms, whichever is not the default.
2080 The following options control the amount and kinds of warnings produced
2081 by GCC; for further, language-specific options also refer to
2082 @ref{C++ Dialect Options} and @ref{Objective-C and Objective-C++ Dialect
2086 @cindex syntax checking
2088 @opindex fsyntax-only
2089 Check the code for syntax errors, but don't do anything beyond that.
2093 Issue all the warnings demanded by strict ISO C and ISO C++;
2094 reject all programs that use forbidden extensions, and some other
2095 programs that do not follow ISO C and ISO C++. For ISO C, follows the
2096 version of the ISO C standard specified by any @option{-std} option used.
2098 Valid ISO C and ISO C++ programs should compile properly with or without
2099 this option (though a rare few will require @option{-ansi} or a
2100 @option{-std} option specifying the required version of ISO C)@. However,
2101 without this option, certain GNU extensions and traditional C and C++
2102 features are supported as well. With this option, they are rejected.
2104 @option{-pedantic} does not cause warning messages for use of the
2105 alternate keywords whose names begin and end with @samp{__}. Pedantic
2106 warnings are also disabled in the expression that follows
2107 @code{__extension__}. However, only system header files should use
2108 these escape routes; application programs should avoid them.
2109 @xref{Alternate Keywords}.
2111 Some users try to use @option{-pedantic} to check programs for strict ISO
2112 C conformance. They soon find that it does not do quite what they want:
2113 it finds some non-ISO practices, but not all---only those for which
2114 ISO C @emph{requires} a diagnostic, and some others for which
2115 diagnostics have been added.
2117 A feature to report any failure to conform to ISO C might be useful in
2118 some instances, but would require considerable additional work and would
2119 be quite different from @option{-pedantic}. We don't have plans to
2120 support such a feature in the near future.
2122 Where the standard specified with @option{-std} represents a GNU
2123 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
2124 corresponding @dfn{base standard}, the version of ISO C on which the GNU
2125 extended dialect is based. Warnings from @option{-pedantic} are given
2126 where they are required by the base standard. (It would not make sense
2127 for such warnings to be given only for features not in the specified GNU
2128 C dialect, since by definition the GNU dialects of C include all
2129 features the compiler supports with the given option, and there would be
2130 nothing to warn about.)
2132 @item -pedantic-errors
2133 @opindex pedantic-errors
2134 Like @option{-pedantic}, except that errors are produced rather than
2139 Inhibit all warning messages.
2143 Inhibit warning messages about the use of @samp{#import}.
2145 @item -Wchar-subscripts
2146 @opindex Wchar-subscripts
2147 Warn if an array subscript has type @code{char}. This is a common cause
2148 of error, as programmers often forget that this type is signed on some
2153 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
2154 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
2156 @item -Wfatal-errors
2157 @opindex Wfatal-errors
2158 This option causes the compiler to abort compilation on the first error
2159 occurred rather than trying to keep going and printing further error
2164 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
2165 the arguments supplied have types appropriate to the format string
2166 specified, and that the conversions specified in the format string make
2167 sense. This includes standard functions, and others specified by format
2168 attributes (@pxref{Function Attributes}), in the @code{printf},
2169 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
2170 not in the C standard) families (or other target-specific families).
2172 The formats are checked against the format features supported by GNU
2173 libc version 2.2. These include all ISO C90 and C99 features, as well
2174 as features from the Single Unix Specification and some BSD and GNU
2175 extensions. Other library implementations may not support all these
2176 features; GCC does not support warning about features that go beyond a
2177 particular library's limitations. However, if @option{-pedantic} is used
2178 with @option{-Wformat}, warnings will be given about format features not
2179 in the selected standard version (but not for @code{strfmon} formats,
2180 since those are not in any version of the C standard). @xref{C Dialect
2181 Options,,Options Controlling C Dialect}.
2183 Since @option{-Wformat} also checks for null format arguments for
2184 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
2186 @option{-Wformat} is included in @option{-Wall}. For more control over some
2187 aspects of format checking, the options @option{-Wformat-y2k},
2188 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
2189 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
2190 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
2193 @opindex Wformat-y2k
2194 If @option{-Wformat} is specified, also warn about @code{strftime}
2195 formats which may yield only a two-digit year.
2197 @item -Wno-format-extra-args
2198 @opindex Wno-format-extra-args
2199 If @option{-Wformat} is specified, do not warn about excess arguments to a
2200 @code{printf} or @code{scanf} format function. The C standard specifies
2201 that such arguments are ignored.
2203 Where the unused arguments lie between used arguments that are
2204 specified with @samp{$} operand number specifications, normally
2205 warnings are still given, since the implementation could not know what
2206 type to pass to @code{va_arg} to skip the unused arguments. However,
2207 in the case of @code{scanf} formats, this option will suppress the
2208 warning if the unused arguments are all pointers, since the Single
2209 Unix Specification says that such unused arguments are allowed.
2211 @item -Wno-format-zero-length
2212 @opindex Wno-format-zero-length
2213 If @option{-Wformat} is specified, do not warn about zero-length formats.
2214 The C standard specifies that zero-length formats are allowed.
2216 @item -Wformat-nonliteral
2217 @opindex Wformat-nonliteral
2218 If @option{-Wformat} is specified, also warn if the format string is not a
2219 string literal and so cannot be checked, unless the format function
2220 takes its format arguments as a @code{va_list}.
2222 @item -Wformat-security
2223 @opindex Wformat-security
2224 If @option{-Wformat} is specified, also warn about uses of format
2225 functions that represent possible security problems. At present, this
2226 warns about calls to @code{printf} and @code{scanf} functions where the
2227 format string is not a string literal and there are no format arguments,
2228 as in @code{printf (foo);}. This may be a security hole if the format
2229 string came from untrusted input and contains @samp{%n}. (This is
2230 currently a subset of what @option{-Wformat-nonliteral} warns about, but
2231 in future warnings may be added to @option{-Wformat-security} that are not
2232 included in @option{-Wformat-nonliteral}.)
2236 Enable @option{-Wformat} plus format checks not included in
2237 @option{-Wformat}. Currently equivalent to @samp{-Wformat
2238 -Wformat-nonliteral -Wformat-security -Wformat-y2k}.
2242 Warn about passing a null pointer for arguments marked as
2243 requiring a non-null value by the @code{nonnull} function attribute.
2245 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
2246 can be disabled with the @option{-Wno-nonnull} option.
2248 @item -Winit-self @r{(C, C++, Objective-C and Objective-C++ only)}
2250 Warn about uninitialized variables which are initialized with themselves.
2251 Note this option can only be used with the @option{-Wuninitialized} option,
2252 which in turn only works with @option{-O1} and above.
2254 For example, GCC will warn about @code{i} being uninitialized in the
2255 following snippet only when @option{-Winit-self} has been specified:
2266 @item -Wimplicit-int
2267 @opindex Wimplicit-int
2268 Warn when a declaration does not specify a type.
2270 @item -Wimplicit-function-declaration
2271 @itemx -Werror-implicit-function-declaration
2272 @opindex Wimplicit-function-declaration
2273 @opindex Werror-implicit-function-declaration
2274 Give a warning (or error) whenever a function is used before being
2279 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
2283 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
2284 function with external linkage, returning int, taking either zero
2285 arguments, two, or three arguments of appropriate types.
2287 @item -Wmissing-braces
2288 @opindex Wmissing-braces
2289 Warn if an aggregate or union initializer is not fully bracketed. In
2290 the following example, the initializer for @samp{a} is not fully
2291 bracketed, but that for @samp{b} is fully bracketed.
2294 int a[2][2] = @{ 0, 1, 2, 3 @};
2295 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
2298 @item -Wmissing-include-dirs @r{(C, C++, Objective-C and Objective-C++ only)}
2299 @opindex Wmissing-include-dirs
2300 Warn if a user-supplied include directory does not exist.
2303 @opindex Wparentheses
2304 Warn if parentheses are omitted in certain contexts, such
2305 as when there is an assignment in a context where a truth value
2306 is expected, or when operators are nested whose precedence people
2307 often get confused about. Only the warning for an assignment used as
2308 a truth value is supported when compiling C++; the other warnings are
2309 only supported when compiling C@.
2311 Also warn if a comparison like @samp{x<=y<=z} appears; this is
2312 equivalent to @samp{(x<=y ? 1 : 0) <= z}, which is a different
2313 interpretation from that of ordinary mathematical notation.
2315 Also warn about constructions where there may be confusion to which
2316 @code{if} statement an @code{else} branch belongs. Here is an example of
2331 In C, every @code{else} branch belongs to the innermost possible @code{if}
2332 statement, which in this example is @code{if (b)}. This is often not
2333 what the programmer expected, as illustrated in the above example by
2334 indentation the programmer chose. When there is the potential for this
2335 confusion, GCC will issue a warning when this flag is specified.
2336 To eliminate the warning, add explicit braces around the innermost
2337 @code{if} statement so there is no way the @code{else} could belong to
2338 the enclosing @code{if}. The resulting code would look like this:
2354 @item -Wsequence-point
2355 @opindex Wsequence-point
2356 Warn about code that may have undefined semantics because of violations
2357 of sequence point rules in the C standard.
2359 The C standard defines the order in which expressions in a C program are
2360 evaluated in terms of @dfn{sequence points}, which represent a partial
2361 ordering between the execution of parts of the program: those executed
2362 before the sequence point, and those executed after it. These occur
2363 after the evaluation of a full expression (one which is not part of a
2364 larger expression), after the evaluation of the first operand of a
2365 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
2366 function is called (but after the evaluation of its arguments and the
2367 expression denoting the called function), and in certain other places.
2368 Other than as expressed by the sequence point rules, the order of
2369 evaluation of subexpressions of an expression is not specified. All
2370 these rules describe only a partial order rather than a total order,
2371 since, for example, if two functions are called within one expression
2372 with no sequence point between them, the order in which the functions
2373 are called is not specified. However, the standards committee have
2374 ruled that function calls do not overlap.
2376 It is not specified when between sequence points modifications to the
2377 values of objects take effect. Programs whose behavior depends on this
2378 have undefined behavior; the C standard specifies that ``Between the
2379 previous and next sequence point an object shall have its stored value
2380 modified at most once by the evaluation of an expression. Furthermore,
2381 the prior value shall be read only to determine the value to be
2382 stored.''. If a program breaks these rules, the results on any
2383 particular implementation are entirely unpredictable.
2385 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
2386 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
2387 diagnosed by this option, and it may give an occasional false positive
2388 result, but in general it has been found fairly effective at detecting
2389 this sort of problem in programs.
2391 The present implementation of this option only works for C programs. A
2392 future implementation may also work for C++ programs.
2394 The C standard is worded confusingly, therefore there is some debate
2395 over the precise meaning of the sequence point rules in subtle cases.
2396 Links to discussions of the problem, including proposed formal
2397 definitions, may be found on the GCC readings page, at
2398 @w{@uref{http://gcc.gnu.org/readings.html}}.
2401 @opindex Wreturn-type
2402 Warn whenever a function is defined with a return-type that defaults to
2403 @code{int}. Also warn about any @code{return} statement with no
2404 return-value in a function whose return-type is not @code{void}.
2406 For C, also warn if the return type of a function has a type qualifier
2407 such as @code{const}. Such a type qualifier has no effect, since the
2408 value returned by a function is not an lvalue. ISO C prohibits
2409 qualified @code{void} return types on function definitions, so such
2410 return types always receive a warning even without this option.
2412 For C++, a function without return type always produces a diagnostic
2413 message, even when @option{-Wno-return-type} is specified. The only
2414 exceptions are @samp{main} and functions defined in system headers.
2418 Warn whenever a @code{switch} statement has an index of enumerated type
2419 and lacks a @code{case} for one or more of the named codes of that
2420 enumeration. (The presence of a @code{default} label prevents this
2421 warning.) @code{case} labels outside the enumeration range also
2422 provoke warnings when this option is used.
2424 @item -Wswitch-default
2425 @opindex Wswitch-switch
2426 Warn whenever a @code{switch} statement does not have a @code{default}
2430 @opindex Wswitch-enum
2431 Warn whenever a @code{switch} statement has an index of enumerated type
2432 and lacks a @code{case} for one or more of the named codes of that
2433 enumeration. @code{case} labels outside the enumeration range also
2434 provoke warnings when this option is used.
2438 Warn if any trigraphs are encountered that might change the meaning of
2439 the program (trigraphs within comments are not warned about).
2441 @item -Wunused-function
2442 @opindex Wunused-function
2443 Warn whenever a static function is declared but not defined or a
2444 non\-inline static function is unused.
2446 @item -Wunused-label
2447 @opindex Wunused-label
2448 Warn whenever a label is declared but not used.
2450 To suppress this warning use the @samp{unused} attribute
2451 (@pxref{Variable Attributes}).
2453 @item -Wunused-parameter
2454 @opindex Wunused-parameter
2455 Warn whenever a function parameter is unused aside from its declaration.
2457 To suppress this warning use the @samp{unused} attribute
2458 (@pxref{Variable Attributes}).
2460 @item -Wunused-variable
2461 @opindex Wunused-variable
2462 Warn whenever a local variable or non-constant static variable is unused
2463 aside from its declaration
2465 To suppress this warning use the @samp{unused} attribute
2466 (@pxref{Variable Attributes}).
2468 @item -Wunused-value
2469 @opindex Wunused-value
2470 Warn whenever a statement computes a result that is explicitly not used.
2472 To suppress this warning cast the expression to @samp{void}.
2476 All the above @option{-Wunused} options combined.
2478 In order to get a warning about an unused function parameter, you must
2479 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
2480 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
2482 @item -Wuninitialized
2483 @opindex Wuninitialized
2484 Warn if an automatic variable is used without first being initialized or
2485 if a variable may be clobbered by a @code{setjmp} call.
2487 These warnings are possible only in optimizing compilation,
2488 because they require data flow information that is computed only
2489 when optimizing. If you don't specify @option{-O}, you simply won't
2492 If you want to warn about code which uses the uninitialized value of the
2493 variable in its own initializer, use the @option{-Winit-self} option.
2495 These warnings occur only for variables that are candidates for
2496 register allocation. Therefore, they do not occur for a variable that
2497 is declared @code{volatile}, or whose address is taken, or whose size
2498 is other than 1, 2, 4 or 8 bytes. Also, they do not occur for
2499 structures, unions or arrays, even when they are in registers.
2501 Note that there may be no warning about a variable that is used only
2502 to compute a value that itself is never used, because such
2503 computations may be deleted by data flow analysis before the warnings
2506 These warnings are made optional because GCC is not smart
2507 enough to see all the reasons why the code might be correct
2508 despite appearing to have an error. Here is one example of how
2529 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
2530 always initialized, but GCC doesn't know this. Here is
2531 another common case:
2536 if (change_y) save_y = y, y = new_y;
2538 if (change_y) y = save_y;
2543 This has no bug because @code{save_y} is used only if it is set.
2545 @cindex @code{longjmp} warnings
2546 This option also warns when a non-volatile automatic variable might be
2547 changed by a call to @code{longjmp}. These warnings as well are possible
2548 only in optimizing compilation.
2550 The compiler sees only the calls to @code{setjmp}. It cannot know
2551 where @code{longjmp} will be called; in fact, a signal handler could
2552 call it at any point in the code. As a result, you may get a warning
2553 even when there is in fact no problem because @code{longjmp} cannot
2554 in fact be called at the place which would cause a problem.
2556 Some spurious warnings can be avoided if you declare all the functions
2557 you use that never return as @code{noreturn}. @xref{Function
2560 @item -Wunknown-pragmas
2561 @opindex Wunknown-pragmas
2562 @cindex warning for unknown pragmas
2563 @cindex unknown pragmas, warning
2564 @cindex pragmas, warning of unknown
2565 Warn when a #pragma directive is encountered which is not understood by
2566 GCC@. If this command line option is used, warnings will even be issued
2567 for unknown pragmas in system header files. This is not the case if
2568 the warnings were only enabled by the @option{-Wall} command line option.
2570 @item -Wstrict-aliasing
2571 @opindex Wstrict-aliasing
2572 This option is only active when @option{-fstrict-aliasing} is active.
2573 It warns about code which might break the strict aliasing rules that the
2574 compiler is using for optimization. The warning does not catch all
2575 cases, but does attempt to catch the more common pitfalls. It is
2576 included in @option{-Wall}.
2578 @item -Wstrict-aliasing=2
2579 @opindex Wstrict-aliasing=2
2580 This option is only active when @option{-fstrict-aliasing} is active.
2581 It warns about all code which might break the strict aliasing rules that the
2582 compiler is using for optimization. This warning catches all cases, but
2583 it will also give a warning for some ambiguous cases that are safe.
2587 All of the above @samp{-W} options combined. This enables all the
2588 warnings about constructions that some users consider questionable, and
2589 that are easy to avoid (or modify to prevent the warning), even in
2590 conjunction with macros. This also enables some language-specific
2591 warnings described in @ref{C++ Dialect Options} and
2592 @ref{Objective-C and Objective-C++ Dialect Options}.
2595 The following @option{-W@dots{}} options are not implied by @option{-Wall}.
2596 Some of them warn about constructions that users generally do not
2597 consider questionable, but which occasionally you might wish to check
2598 for; others warn about constructions that are necessary or hard to avoid
2599 in some cases, and there is no simple way to modify the code to suppress
2606 (This option used to be called @option{-W}. The older name is still
2607 supported, but the newer name is more descriptive.) Print extra warning
2608 messages for these events:
2612 A function can return either with or without a value. (Falling
2613 off the end of the function body is considered returning without
2614 a value.) For example, this function would evoke such a
2628 An expression-statement or the left-hand side of a comma expression
2629 contains no side effects.
2630 To suppress the warning, cast the unused expression to void.
2631 For example, an expression such as @samp{x[i,j]} will cause a warning,
2632 but @samp{x[(void)i,j]} will not.
2635 An unsigned value is compared against zero with @samp{<} or @samp{>=}.
2638 Storage-class specifiers like @code{static} are not the first things in
2639 a declaration. According to the C Standard, this usage is obsolescent.
2642 If @option{-Wall} or @option{-Wunused} is also specified, warn about unused
2646 A comparison between signed and unsigned values could produce an
2647 incorrect result when the signed value is converted to unsigned.
2648 (But don't warn if @option{-Wno-sign-compare} is also specified.)
2651 An aggregate has an initializer which does not initialize all members.
2652 This warning can be independently controlled by
2653 @option{-Wmissing-field-initializers}.
2656 A function parameter is declared without a type specifier in K&R-style
2664 An empty body occurs in an @samp{if} or @samp{else} statement.
2667 A pointer is compared against integer zero with @samp{<}, @samp{<=},
2668 @samp{>}, or @samp{>=}.
2671 A variable might be changed by @samp{longjmp} or @samp{vfork}.
2674 Any of several floating-point events that often indicate errors, such as
2675 overflow, underflow, loss of precision, etc.
2677 @item @r{(C++ only)}
2678 An enumerator and a non-enumerator both appear in a conditional expression.
2680 @item @r{(C++ only)}
2681 A non-static reference or non-static @samp{const} member appears in a
2682 class without constructors.
2684 @item @r{(C++ only)}
2685 Ambiguous virtual bases.
2687 @item @r{(C++ only)}
2688 Subscripting an array which has been declared @samp{register}.
2690 @item @r{(C++ only)}
2691 Taking the address of a variable which has been declared @samp{register}.
2693 @item @r{(C++ only)}
2694 A base class is not initialized in a derived class' copy constructor.
2697 @item -Wno-div-by-zero
2698 @opindex Wno-div-by-zero
2699 @opindex Wdiv-by-zero
2700 Do not warn about compile-time integer division by zero. Floating point
2701 division by zero is not warned about, as it can be a legitimate way of
2702 obtaining infinities and NaNs.
2704 @item -Wsystem-headers
2705 @opindex Wsystem-headers
2706 @cindex warnings from system headers
2707 @cindex system headers, warnings from
2708 Print warning messages for constructs found in system header files.
2709 Warnings from system headers are normally suppressed, on the assumption
2710 that they usually do not indicate real problems and would only make the
2711 compiler output harder to read. Using this command line option tells
2712 GCC to emit warnings from system headers as if they occurred in user
2713 code. However, note that using @option{-Wall} in conjunction with this
2714 option will @emph{not} warn about unknown pragmas in system
2715 headers---for that, @option{-Wunknown-pragmas} must also be used.
2718 @opindex Wfloat-equal
2719 Warn if floating point values are used in equality comparisons.
2721 The idea behind this is that sometimes it is convenient (for the
2722 programmer) to consider floating-point values as approximations to
2723 infinitely precise real numbers. If you are doing this, then you need
2724 to compute (by analyzing the code, or in some other way) the maximum or
2725 likely maximum error that the computation introduces, and allow for it
2726 when performing comparisons (and when producing output, but that's a
2727 different problem). In particular, instead of testing for equality, you
2728 would check to see whether the two values have ranges that overlap; and
2729 this is done with the relational operators, so equality comparisons are
2732 @item -Wtraditional @r{(C only)}
2733 @opindex Wtraditional
2734 Warn about certain constructs that behave differently in traditional and
2735 ISO C@. Also warn about ISO C constructs that have no traditional C
2736 equivalent, and/or problematic constructs which should be avoided.
2740 Macro parameters that appear within string literals in the macro body.
2741 In traditional C macro replacement takes place within string literals,
2742 but does not in ISO C@.
2745 In traditional C, some preprocessor directives did not exist.
2746 Traditional preprocessors would only consider a line to be a directive
2747 if the @samp{#} appeared in column 1 on the line. Therefore
2748 @option{-Wtraditional} warns about directives that traditional C
2749 understands but would ignore because the @samp{#} does not appear as the
2750 first character on the line. It also suggests you hide directives like
2751 @samp{#pragma} not understood by traditional C by indenting them. Some
2752 traditional implementations would not recognize @samp{#elif}, so it
2753 suggests avoiding it altogether.
2756 A function-like macro that appears without arguments.
2759 The unary plus operator.
2762 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
2763 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
2764 constants.) Note, these suffixes appear in macros defined in the system
2765 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
2766 Use of these macros in user code might normally lead to spurious
2767 warnings, however GCC's integrated preprocessor has enough context to
2768 avoid warning in these cases.
2771 A function declared external in one block and then used after the end of
2775 A @code{switch} statement has an operand of type @code{long}.
2778 A non-@code{static} function declaration follows a @code{static} one.
2779 This construct is not accepted by some traditional C compilers.
2782 The ISO type of an integer constant has a different width or
2783 signedness from its traditional type. This warning is only issued if
2784 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
2785 typically represent bit patterns, are not warned about.
2788 Usage of ISO string concatenation is detected.
2791 Initialization of automatic aggregates.
2794 Identifier conflicts with labels. Traditional C lacks a separate
2795 namespace for labels.
2798 Initialization of unions. If the initializer is zero, the warning is
2799 omitted. This is done under the assumption that the zero initializer in
2800 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
2801 initializer warnings and relies on default initialization to zero in the
2805 Conversions by prototypes between fixed/floating point values and vice
2806 versa. The absence of these prototypes when compiling with traditional
2807 C would cause serious problems. This is a subset of the possible
2808 conversion warnings, for the full set use @option{-Wconversion}.
2811 Use of ISO C style function definitions. This warning intentionally is
2812 @emph{not} issued for prototype declarations or variadic functions
2813 because these ISO C features will appear in your code when using
2814 libiberty's traditional C compatibility macros, @code{PARAMS} and
2815 @code{VPARAMS}. This warning is also bypassed for nested functions
2816 because that feature is already a GCC extension and thus not relevant to
2817 traditional C compatibility.
2820 @item -Wdeclaration-after-statement @r{(C only)}
2821 @opindex Wdeclaration-after-statement
2822 Warn when a declaration is found after a statement in a block. This
2823 construct, known from C++, was introduced with ISO C99 and is by default
2824 allowed in GCC@. It is not supported by ISO C90 and was not supported by
2825 GCC versions before GCC 3.0. @xref{Mixed Declarations}.
2829 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
2831 @item -Wendif-labels
2832 @opindex Wendif-labels
2833 Warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
2837 Warn whenever a local variable shadows another local variable, parameter or
2838 global variable or whenever a built-in function is shadowed.
2840 @item -Wlarger-than-@var{len}
2841 @opindex Wlarger-than
2842 Warn whenever an object of larger than @var{len} bytes is defined.
2844 @item -Wpointer-arith
2845 @opindex Wpointer-arith
2846 Warn about anything that depends on the ``size of'' a function type or
2847 of @code{void}. GNU C assigns these types a size of 1, for
2848 convenience in calculations with @code{void *} pointers and pointers
2851 @item -Wbad-function-cast @r{(C only)}
2852 @opindex Wbad-function-cast
2853 Warn whenever a function call is cast to a non-matching type.
2854 For example, warn if @code{int malloc()} is cast to @code{anything *}.
2858 Warn whenever a pointer is cast so as to remove a type qualifier from
2859 the target type. For example, warn if a @code{const char *} is cast
2860 to an ordinary @code{char *}.
2863 @opindex Wcast-align
2864 Warn whenever a pointer is cast such that the required alignment of the
2865 target is increased. For example, warn if a @code{char *} is cast to
2866 an @code{int *} on machines where integers can only be accessed at
2867 two- or four-byte boundaries.
2869 @item -Wwrite-strings
2870 @opindex Wwrite-strings
2871 When compiling C, give string constants the type @code{const
2872 char[@var{length}]} so that
2873 copying the address of one into a non-@code{const} @code{char *}
2874 pointer will get a warning; when compiling C++, warn about the
2875 deprecated conversion from string constants to @code{char *}.
2876 These warnings will help you find at
2877 compile time code that can try to write into a string constant, but
2878 only if you have been very careful about using @code{const} in
2879 declarations and prototypes. Otherwise, it will just be a nuisance;
2880 this is why we did not make @option{-Wall} request these warnings.
2883 @opindex Wconversion
2884 Warn if a prototype causes a type conversion that is different from what
2885 would happen to the same argument in the absence of a prototype. This
2886 includes conversions of fixed point to floating and vice versa, and
2887 conversions changing the width or signedness of a fixed point argument
2888 except when the same as the default promotion.
2890 Also, warn if a negative integer constant expression is implicitly
2891 converted to an unsigned type. For example, warn about the assignment
2892 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
2893 casts like @code{(unsigned) -1}.
2895 @item -Wsign-compare
2896 @opindex Wsign-compare
2897 @cindex warning for comparison of signed and unsigned values
2898 @cindex comparison of signed and unsigned values, warning
2899 @cindex signed and unsigned values, comparison warning
2900 Warn when a comparison between signed and unsigned values could produce
2901 an incorrect result when the signed value is converted to unsigned.
2902 This warning is also enabled by @option{-Wextra}; to get the other warnings
2903 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
2905 @item -Waggregate-return
2906 @opindex Waggregate-return
2907 Warn if any functions that return structures or unions are defined or
2908 called. (In languages where you can return an array, this also elicits
2911 @item -Wstrict-prototypes @r{(C only)}
2912 @opindex Wstrict-prototypes
2913 Warn if a function is declared or defined without specifying the
2914 argument types. (An old-style function definition is permitted without
2915 a warning if preceded by a declaration which specifies the argument
2918 @item -Wold-style-definition @r{(C only)}
2919 @opindex Wold-style-definition
2920 Warn if an old-style function definition is used. A warning is given
2921 even if there is a previous prototype.
2923 @item -Wmissing-prototypes @r{(C only)}
2924 @opindex Wmissing-prototypes
2925 Warn if a global function is defined without a previous prototype
2926 declaration. This warning is issued even if the definition itself
2927 provides a prototype. The aim is to detect global functions that fail
2928 to be declared in header files.
2930 @item -Wmissing-declarations @r{(C only)}
2931 @opindex Wmissing-declarations
2932 Warn if a global function is defined without a previous declaration.
2933 Do so even if the definition itself provides a prototype.
2934 Use this option to detect global functions that are not declared in
2937 @item -Wmissing-field-initializers
2938 @opindex Wmissing-field-initializers
2941 Warn if a structure's initializer has some fields missing. For
2942 example, the following code would cause such a warning, because
2943 @code{x.h} is implicitly zero:
2946 struct s @{ int f, g, h; @};
2947 struct s x = @{ 3, 4 @};
2950 This option does not warn about designated initializers, so the following
2951 modification would not trigger a warning:
2954 struct s @{ int f, g, h; @};
2955 struct s x = @{ .f = 3, .g = 4 @};
2958 This warning is included in @option{-Wextra}. To get other @option{-Wextra}
2959 warnings without this one, use @samp{-Wextra -Wno-missing-field-initializers}.
2961 @item -Wmissing-noreturn
2962 @opindex Wmissing-noreturn
2963 Warn about functions which might be candidates for attribute @code{noreturn}.
2964 Note these are only possible candidates, not absolute ones. Care should
2965 be taken to manually verify functions actually do not ever return before
2966 adding the @code{noreturn} attribute, otherwise subtle code generation
2967 bugs could be introduced. You will not get a warning for @code{main} in
2968 hosted C environments.
2970 @item -Wmissing-format-attribute
2971 @opindex Wmissing-format-attribute
2973 If @option{-Wformat} is enabled, also warn about functions which might be
2974 candidates for @code{format} attributes. Note these are only possible
2975 candidates, not absolute ones. GCC will guess that @code{format}
2976 attributes might be appropriate for any function that calls a function
2977 like @code{vprintf} or @code{vscanf}, but this might not always be the
2978 case, and some functions for which @code{format} attributes are
2979 appropriate may not be detected. This option has no effect unless
2980 @option{-Wformat} is enabled (possibly by @option{-Wall}).
2982 @item -Wno-multichar
2983 @opindex Wno-multichar
2985 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
2986 Usually they indicate a typo in the user's code, as they have
2987 implementation-defined values, and should not be used in portable code.
2989 @item -Wno-deprecated-declarations
2990 @opindex Wno-deprecated-declarations
2991 Do not warn about uses of functions, variables, and types marked as
2992 deprecated by using the @code{deprecated} attribute.
2993 (@pxref{Function Attributes}, @pxref{Variable Attributes},
2994 @pxref{Type Attributes}.)
2998 Warn if a structure is given the packed attribute, but the packed
2999 attribute has no effect on the layout or size of the structure.
3000 Such structures may be mis-aligned for little benefit. For
3001 instance, in this code, the variable @code{f.x} in @code{struct bar}
3002 will be misaligned even though @code{struct bar} does not itself
3003 have the packed attribute:
3010 @} __attribute__((packed));
3020 Warn if padding is included in a structure, either to align an element
3021 of the structure or to align the whole structure. Sometimes when this
3022 happens it is possible to rearrange the fields of the structure to
3023 reduce the padding and so make the structure smaller.
3025 @item -Wredundant-decls
3026 @opindex Wredundant-decls
3027 Warn if anything is declared more than once in the same scope, even in
3028 cases where multiple declaration is valid and changes nothing.
3030 @item -Wnested-externs @r{(C only)}
3031 @opindex Wnested-externs
3032 Warn if an @code{extern} declaration is encountered within a function.
3034 @item -Wunreachable-code
3035 @opindex Wunreachable-code
3036 Warn if the compiler detects that code will never be executed.
3038 This option is intended to warn when the compiler detects that at
3039 least a whole line of source code will never be executed, because
3040 some condition is never satisfied or because it is after a
3041 procedure that never returns.
3043 It is possible for this option to produce a warning even though there
3044 are circumstances under which part of the affected line can be executed,
3045 so care should be taken when removing apparently-unreachable code.
3047 For instance, when a function is inlined, a warning may mean that the
3048 line is unreachable in only one inlined copy of the function.
3050 This option is not made part of @option{-Wall} because in a debugging
3051 version of a program there is often substantial code which checks
3052 correct functioning of the program and is, hopefully, unreachable
3053 because the program does work. Another common use of unreachable
3054 code is to provide behavior which is selectable at compile-time.
3058 Warn if a function can not be inlined and it was declared as inline.
3059 Even with this option, the compiler will not warn about failures to
3060 inline functions declared in system headers.
3062 The compiler uses a variety of heuristics to determine whether or not
3063 to inline a function. For example, the compiler takes into account
3064 the size of the function being inlined and the the amount of inlining
3065 that has already been done in the current function. Therefore,
3066 seemingly insignificant changes in the source program can cause the
3067 warnings produced by @option{-Winline} to appear or disappear.
3069 @item -Wno-invalid-offsetof @r{(C++ only)}
3070 @opindex Wno-invalid-offsetof
3071 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
3072 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
3073 to a non-POD type is undefined. In existing C++ implementations,
3074 however, @samp{offsetof} typically gives meaningful results even when
3075 applied to certain kinds of non-POD types. (Such as a simple
3076 @samp{struct} that fails to be a POD type only by virtue of having a
3077 constructor.) This flag is for users who are aware that they are
3078 writing nonportable code and who have deliberately chosen to ignore the
3081 The restrictions on @samp{offsetof} may be relaxed in a future version
3082 of the C++ standard.
3085 @opindex Winvalid-pch
3086 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
3087 the search path but can't be used.
3091 @opindex Wno-long-long
3092 Warn if @samp{long long} type is used. This is default. To inhibit
3093 the warning messages, use @option{-Wno-long-long}. Flags
3094 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
3095 only when @option{-pedantic} flag is used.
3097 @item -Wvariadic-macros
3098 @opindex Wvariadic-macros
3099 @opindex Wno-variadic-macros
3100 Warn if variadic macros are used in pedantic ISO C90 mode, or the GNU
3101 alternate syntax when in pedantic ISO C99 mode. This is default.
3102 To inhibit the warning messages, use @option{-Wno-variadic-macros}.
3104 @item -Wdisabled-optimization
3105 @opindex Wdisabled-optimization
3106 Warn if a requested optimization pass is disabled. This warning does
3107 not generally indicate that there is anything wrong with your code; it
3108 merely indicates that GCC's optimizers were unable to handle the code
3109 effectively. Often, the problem is that your code is too big or too
3110 complex; GCC will refuse to optimize programs when the optimization
3111 itself is likely to take inordinate amounts of time.
3115 Make all warnings into errors.
3118 @node Debugging Options
3119 @section Options for Debugging Your Program or GCC
3120 @cindex options, debugging
3121 @cindex debugging information options
3123 GCC has various special options that are used for debugging
3124 either your program or GCC:
3129 Produce debugging information in the operating system's native format
3130 (stabs, COFF, XCOFF, or DWARF 2)@. GDB can work with this debugging
3133 On most systems that use stabs format, @option{-g} enables use of extra
3134 debugging information that only GDB can use; this extra information
3135 makes debugging work better in GDB but will probably make other debuggers
3137 refuse to read the program. If you want to control for certain whether
3138 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
3139 @option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
3141 GCC allows you to use @option{-g} with
3142 @option{-O}. The shortcuts taken by optimized code may occasionally
3143 produce surprising results: some variables you declared may not exist
3144 at all; flow of control may briefly move where you did not expect it;
3145 some statements may not be executed because they compute constant
3146 results or their values were already at hand; some statements may
3147 execute in different places because they were moved out of loops.
3149 Nevertheless it proves possible to debug optimized output. This makes
3150 it reasonable to use the optimizer for programs that might have bugs.
3152 The following options are useful when GCC is generated with the
3153 capability for more than one debugging format.
3157 Produce debugging information for use by GDB@. This means to use the
3158 most expressive format available (DWARF 2, stabs, or the native format
3159 if neither of those are supported), including GDB extensions if at all
3164 Produce debugging information in stabs format (if that is supported),
3165 without GDB extensions. This is the format used by DBX on most BSD
3166 systems. On MIPS, Alpha and System V Release 4 systems this option
3167 produces stabs debugging output which is not understood by DBX or SDB@.
3168 On System V Release 4 systems this option requires the GNU assembler.
3170 @item -feliminate-unused-debug-symbols
3171 @opindex feliminate-unused-debug-symbols
3172 Produce debugging information in stabs format (if that is supported),
3173 for only symbols that are actually used.
3177 Produce debugging information in stabs format (if that is supported),
3178 using GNU extensions understood only by the GNU debugger (GDB)@. The
3179 use of these extensions is likely to make other debuggers crash or
3180 refuse to read the program.
3184 Produce debugging information in COFF format (if that is supported).
3185 This is the format used by SDB on most System V systems prior to
3190 Produce debugging information in XCOFF format (if that is supported).
3191 This is the format used by the DBX debugger on IBM RS/6000 systems.
3195 Produce debugging information in XCOFF format (if that is supported),
3196 using GNU extensions understood only by the GNU debugger (GDB)@. The
3197 use of these extensions is likely to make other debuggers crash or
3198 refuse to read the program, and may cause assemblers other than the GNU
3199 assembler (GAS) to fail with an error.
3203 Produce debugging information in DWARF version 2 format (if that is
3204 supported). This is the format used by DBX on IRIX 6. With this
3205 option, GCC uses features of DWARF version 3 when they are useful;
3206 version 3 is upward compatible with version 2, but may still cause
3207 problems for older debuggers.
3211 Produce debugging information in VMS debug format (if that is
3212 supported). This is the format used by DEBUG on VMS systems.
3215 @itemx -ggdb@var{level}
3216 @itemx -gstabs@var{level}
3217 @itemx -gcoff@var{level}
3218 @itemx -gxcoff@var{level}
3219 @itemx -gvms@var{level}
3220 Request debugging information and also use @var{level} to specify how
3221 much information. The default level is 2.
3223 Level 1 produces minimal information, enough for making backtraces in
3224 parts of the program that you don't plan to debug. This includes
3225 descriptions of functions and external variables, but no information
3226 about local variables and no line numbers.
3228 Level 3 includes extra information, such as all the macro definitions
3229 present in the program. Some debuggers support macro expansion when
3230 you use @option{-g3}.
3232 @option{-gdwarf-2} does not accept a concatenated debug level, because
3233 GCC used to support an option @option{-gdwarf} that meant to generate
3234 debug information in version 1 of the DWARF format (which is very
3235 different from version 2), and it would have been too confusing. That
3236 debug format is long obsolete, but the option cannot be changed now.
3237 Instead use an additional @option{-g@var{level}} option to change the
3238 debug level for DWARF2.
3240 @item -feliminate-dwarf2-dups
3241 @opindex feliminate-dwarf2-dups
3242 Compress DWARF2 debugging information by eliminating duplicated
3243 information about each symbol. This option only makes sense when
3244 generating DWARF2 debugging information with @option{-gdwarf-2}.
3246 @cindex @command{prof}
3249 Generate extra code to write profile information suitable for the
3250 analysis program @command{prof}. You must use this option when compiling
3251 the source files you want data about, and you must also use it when
3254 @cindex @command{gprof}
3257 Generate extra code to write profile information suitable for the
3258 analysis program @command{gprof}. You must use this option when compiling
3259 the source files you want data about, and you must also use it when
3264 Makes the compiler print out each function name as it is compiled, and
3265 print some statistics about each pass when it finishes.
3268 @opindex ftime-report
3269 Makes the compiler print some statistics about the time consumed by each
3270 pass when it finishes.
3273 @opindex fmem-report
3274 Makes the compiler print some statistics about permanent memory
3275 allocation when it finishes.
3277 @item -fprofile-arcs
3278 @opindex fprofile-arcs
3279 Add code so that program flow @dfn{arcs} are instrumented. During
3280 execution the program records how many times each branch and call is
3281 executed and how many times it is taken or returns. When the compiled
3282 program exits it saves this data to a file called
3283 @file{@var{auxname}.gcda} for each source file. The data may be used for
3284 profile-directed optimizations (@option{-fbranch-probabilities}), or for
3285 test coverage analysis (@option{-ftest-coverage}). Each object file's
3286 @var{auxname} is generated from the name of the output file, if
3287 explicitly specified and it is not the final executable, otherwise it is
3288 the basename of the source file. In both cases any suffix is removed
3289 (e.g.@: @file{foo.gcda} for input file @file{dir/foo.c}, or
3290 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
3295 Compile the source files with @option{-fprofile-arcs} plus optimization
3296 and code generation options. For test coverage analysis, use the
3297 additional @option{-ftest-coverage} option. You do not need to profile
3298 every source file in a program.
3301 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
3302 (the latter implies the former).
3305 Run the program on a representative workload to generate the arc profile
3306 information. This may be repeated any number of times. You can run
3307 concurrent instances of your program, and provided that the file system
3308 supports locking, the data files will be correctly updated. Also
3309 @code{fork} calls are detected and correctly handled (double counting
3313 For profile-directed optimizations, compile the source files again with
3314 the same optimization and code generation options plus
3315 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
3316 Control Optimization}).
3319 For test coverage analysis, use @command{gcov} to produce human readable
3320 information from the @file{.gcno} and @file{.gcda} files. Refer to the
3321 @command{gcov} documentation for further information.
3325 With @option{-fprofile-arcs}, for each function of your program GCC
3326 creates a program flow graph, then finds a spanning tree for the graph.
3327 Only arcs that are not on the spanning tree have to be instrumented: the
3328 compiler adds code to count the number of times that these arcs are
3329 executed. When an arc is the only exit or only entrance to a block, the
3330 instrumentation code can be added to the block; otherwise, a new basic
3331 block must be created to hold the instrumentation code.
3333 @item -ftree-based-profiling
3334 @opindex ftree-based-profiling
3335 This option is used in addition to @option{-fprofile-arcs} or
3336 @option{-fbranch-probabilities} to control whether those optimizations
3337 are performed on a tree-based or rtl-based internal representation.
3338 If you use this option when compiling with @option{-fprofile-arcs},
3339 you must also use it when compiling later with @option{-fbranch-probabilities}.
3340 Currently the tree-based optimization is in an early stage of
3341 development, and this option is recommended only for those people
3342 working on improving it.
3345 @item -ftest-coverage
3346 @opindex ftest-coverage
3347 Produce a notes file that the @command{gcov} code-coverage utility
3348 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
3349 show program coverage. Each source file's note file is called
3350 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
3351 above for a description of @var{auxname} and instructions on how to
3352 generate test coverage data. Coverage data will match the source files
3353 more closely, if you do not optimize.
3355 @item -d@var{letters}
3356 @item -fdump-rtl-@var{pass}
3358 Says to make debugging dumps during compilation at times specified by
3359 @var{letters}. This is used for debugging the RTL-based passes of the
3360 compiler. The file names for most of the dumps are made by appending a
3361 pass number and a word to the @var{dumpname}. @var{dumpname} is generated
3362 from the name of the output file, if explicitly specified and it is not
3363 an executable, otherwise it is the basename of the source file.
3365 Most debug dumps can be enabled either passing a letter to the @option{-d}
3366 option, or with a long @option{-fdump-rtl} switch; here are the possible
3367 letters for use in @var{letters} and @var{pass}, and their meanings:
3372 Annotate the assembler output with miscellaneous debugging information.
3375 @itemx -fdump-rtl-bp
3377 @opindex fdump-rtl-bp
3378 Dump after computing branch probabilities, to @file{@var{file}.09.bp}.
3381 @itemx -fdump-rtl-bbro
3383 @opindex fdump-rtl-bbro
3384 Dump after block reordering, to @file{@var{file}.30.bbro}.
3387 @itemx -fdump-rtl-combine
3389 @opindex fdump-rtl-combine
3390 Dump after instruction combination, to the file @file{@var{file}.17.combine}.
3393 @itemx -fdump-rtl-ce1
3394 @itemx -fdump-rtl-ce2
3396 @opindex fdump-rtl-ce1
3397 @opindex fdump-rtl-ce2
3398 @option{-dC} and @option{-fdump-rtl-ce1} enable dumping after the
3399 first if conversion, to the file @file{@var{file}.11.ce1}. @option{-dC}
3400 and @option{-fdump-rtl-ce2} enable dumping after the second if
3401 conversion, to the file @file{@var{file}.18.ce2}.
3404 @itemx -fdump-rtl-btl
3405 @itemx -fdump-rtl-dbr
3407 @opindex fdump-rtl-btl
3408 @opindex fdump-rtl-dbr
3409 @option{-dd} and @option{-fdump-rtl-btl} enable dumping after branch
3410 target load optimization, to to @file{@var{file}.31.btl}. @option{-dd}
3411 and @option{-fdump-rtl-dbr} enable dumping after delayed branch
3412 scheduling, to @file{@var{file}.36.dbr}.
3416 Dump all macro definitions, at the end of preprocessing, in addition to
3420 @itemx -fdump-rtl-ce3
3422 @opindex fdump-rtl-ce3
3423 Dump after the third if conversion, to @file{@var{file}.28.ce3}.
3426 @itemx -fdump-rtl-cfg
3427 @itemx -fdump-rtl-life
3429 @opindex fdump-rtl-cfg
3430 @opindex fdump-rtl-life
3431 @option{-df} and @option{-fdump-rtl-cfg} enable dumping after control
3432 and data flow analysis, to @file{@var{file}.08.cfg}. @option{-df}
3433 and @option{-fdump-rtl-cfg} enable dumping dump after life analysis,
3434 to @file{@var{file}.16.life}.
3437 @itemx -fdump-rtl-greg
3439 @opindex fdump-rtl-greg
3440 Dump after global register allocation, to @file{@var{file}.23.greg}.
3443 @itemx -fdump-rtl-gcse
3444 @itemx -fdump-rtl-bypass
3446 @opindex fdump-rtl-gcse
3447 @opindex fdump-rtl-bypass
3448 @option{-dG} and @option{-fdump-rtl-gcse} enable dumping after GCSE, to
3449 @file{@var{file}.05.gcse}. @option{-dG} and @option{-fdump-rtl-bypass}
3450 enable dumping after jump bypassing and control flow optimizations, to
3451 @file{@var{file}.07.bypass}.
3454 @itemx -fdump-rtl-eh
3456 @opindex fdump-rtl-eh
3457 Dump after finalization of EH handling code, to @file{@var{file}.02.eh}.
3460 @itemx -fdump-rtl-sibling
3462 @opindex fdump-rtl-sibling
3463 Dump after sibling call optimizations, to @file{@var{file}.01.sibling}.
3466 @itemx -fdump-rtl-jump
3468 @opindex fdump-rtl-jump
3469 Dump after the first jump optimization, to @file{@var{file}.03.jump}.
3472 @itemx -fdump-rtl-stack
3474 @opindex fdump-rtl-stack
3475 Dump after conversion from registers to stack, to @file{@var{file}.33.stack}.
3478 @itemx -fdump-rtl-lreg
3480 @opindex fdump-rtl-lreg
3481 Dump after local register allocation, to @file{@var{file}.22.lreg}.
3484 @itemx -fdump-rtl-loop
3485 @itemx -fdump-rtl-loop2
3487 @opindex fdump-rtl-loop
3488 @opindex fdump-rtl-loop2
3489 @option{-dL} and @option{-fdump-rtl-loop} enable dumping after the first
3490 loop optimization pass, to @file{@var{file}.06.loop}. @option{-dL} and
3491 @option{-fdump-rtl-loop2} enable dumping after the second pass, to
3492 @file{@var{file}.13.loop2}.
3495 @itemx -fdump-rtl-sms
3497 @opindex fdump-rtl-sms
3498 Dump after modulo scheduling, to @file{@var{file}.20.sms}.
3501 @itemx -fdump-rtl-mach
3503 @opindex fdump-rtl-mach
3504 Dump after performing the machine dependent reorganization pass, to
3505 @file{@var{file}.35.mach}.
3508 @itemx -fdump-rtl-rnreg
3510 @opindex fdump-rtl-rnreg
3511 Dump after register renumbering, to @file{@var{file}.29.rnreg}.
3514 @itemx -fdump-rtl-regmove
3516 @opindex fdump-rtl-regmove
3517 Dump after the register move pass, to @file{@var{file}.19.regmove}.
3520 @itemx -fdump-rtl-postreload
3522 @opindex fdump-rtl-postreload
3523 Dump after post-reload optimizations, to @file{@var{file}.24.postreload}.
3526 @itemx -fdump-rtl-expand
3528 @opindex fdump-rtl-expand
3529 Dump after RTL generation, to @file{@var{file}.00.expand}.
3532 @itemx -fdump-rtl-sched2
3534 @opindex fdump-rtl-sched2
3535 Dump after the second scheduling pass, to @file{@var{file}.32.sched2}.
3538 @itemx -fdump-rtl-cse
3540 @opindex fdump-rtl-cse
3541 Dump after CSE (including the jump optimization that sometimes follows
3542 CSE), to @file{@var{file}.04.cse}.
3545 @itemx -fdump-rtl-sched
3547 @opindex fdump-rtl-sched
3548 Dump after the first scheduling pass, to @file{@var{file}.21.sched}.
3551 @itemx -fdump-rtl-cse2
3553 @opindex fdump-rtl-cse2
3554 Dump after the second CSE pass (including the jump optimization that
3555 sometimes follows CSE), to @file{@var{file}.15.cse2}.
3558 @itemx -fdump-rtl-tracer
3560 @opindex fdump-rtl-tracer
3561 Dump after running tracer, to @file{@var{file}.12.tracer}.
3564 @itemx -fdump-rtl-vpt
3565 @itemx -fdump-rtl-vartrack
3567 @opindex fdump-rtl-vpt
3568 @opindex fdump-rtl-vartrack
3569 @option{-dV} and @option{-fdump-rtl-vpt} enable dumping after the value
3570 profile transformations, to @file{@var{file}.10.vpt}. @option{-dV}
3571 and @option{-fdump-rtl-vartrack} enable dumping after variable tracking,
3572 to @file{@var{file}.34.vartrack}.
3575 @itemx -fdump-rtl-flow2
3577 @opindex fdump-rtl-flow2
3578 Dump after the second flow pass, to @file{@var{file}.26.flow2}.
3581 @itemx -fdump-rtl-peephole2
3583 @opindex fdump-rtl-peephole2
3584 Dump after the peephole pass, to @file{@var{file}.27.peephole2}.
3587 @itemx -fdump-rtl-web
3589 @opindex fdump-rtl-web
3590 Dump after live range splitting, to @file{@var{file}.14.web}.
3593 @itemx -fdump-rtl-all
3595 @opindex fdump-rtl-all
3596 Produce all the dumps listed above.
3600 Produce a core dump whenever an error occurs.
3604 Print statistics on memory usage, at the end of the run, to
3609 Annotate the assembler output with a comment indicating which
3610 pattern and alternative was used. The length of each instruction is
3615 Dump the RTL in the assembler output as a comment before each instruction.
3616 Also turns on @option{-dp} annotation.
3620 For each of the other indicated dump files (either with @option{-d} or
3621 @option{-fdump-rtl-@var{pass}}), dump a representation of the control flow
3622 graph suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
3626 Just generate RTL for a function instead of compiling it. Usually used
3627 with @samp{r} (@option{-fdump-rtl-expand}).
3631 Dump debugging information during parsing, to standard error.
3634 @item -fdump-unnumbered
3635 @opindex fdump-unnumbered
3636 When doing debugging dumps (see @option{-d} option above), suppress instruction
3637 numbers and line number note output. This makes it more feasible to
3638 use diff on debugging dumps for compiler invocations with different
3639 options, in particular with and without @option{-g}.
3641 @item -fdump-translation-unit @r{(C and C++ only)}
3642 @itemx -fdump-translation-unit-@var{options} @r{(C and C++ only)}
3643 @opindex fdump-translation-unit
3644 Dump a representation of the tree structure for the entire translation
3645 unit to a file. The file name is made by appending @file{.tu} to the
3646 source file name. If the @samp{-@var{options}} form is used, @var{options}
3647 controls the details of the dump as described for the
3648 @option{-fdump-tree} options.
3650 @item -fdump-class-hierarchy @r{(C++ only)}
3651 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
3652 @opindex fdump-class-hierarchy
3653 Dump a representation of each class's hierarchy and virtual function
3654 table layout to a file. The file name is made by appending @file{.class}
3655 to the source file name. If the @samp{-@var{options}} form is used,
3656 @var{options} controls the details of the dump as described for the
3657 @option{-fdump-tree} options.
3659 @item -fdump-ipa-@var{switch}
3661 Control the dumping at various stages of inter-procedural analysis
3662 language tree to a file. The file name is generated by appending a switch
3663 specific suffix to the source file name. The following dumps are possible:
3667 Enables all inter-procedural analysis dumps; currently the only produced
3668 dump is the @samp{cgraph} dump.
3671 Dumps information about call-graph optimization, unused function removal,
3672 and inlining decisions.
3675 @item -fdump-tree-@var{switch} @r{(C and C++ only)}
3676 @itemx -fdump-tree-@var{switch}-@var{options} @r{(C and C++ only)}
3678 Control the dumping at various stages of processing the intermediate
3679 language tree to a file. The file name is generated by appending a switch
3680 specific suffix to the source file name. If the @samp{-@var{options}}
3681 form is used, @var{options} is a list of @samp{-} separated options that
3682 control the details of the dump. Not all options are applicable to all
3683 dumps, those which are not meaningful will be ignored. The following
3684 options are available
3688 Print the address of each node. Usually this is not meaningful as it
3689 changes according to the environment and source file. Its primary use
3690 is for tying up a dump file with a debug environment.
3692 Inhibit dumping of members of a scope or body of a function merely
3693 because that scope has been reached. Only dump such items when they
3694 are directly reachable by some other path. When dumping pretty-printed
3695 trees, this option inhibits dumping the bodies of control structures.
3697 Print a raw representation of the tree. By default, trees are
3698 pretty-printed into a C-like representation.
3700 Enable more detailed dumps (not honored by every dump option).
3702 Enable dumping various statistics about the pass (not honored by every dump
3705 Enable showing basic block boundaries (disabled in raw dumps).
3707 Enable showing virtual operands for every statement.
3709 Enable showing line numbers for statements.
3711 Enable showing the unique ID (@code{DECL_UID}) for each variable.
3713 Turn on all options, except @option{raw}, @option{slim} and @option{lineno}.
3716 The following tree dumps are possible:
3720 Dump before any tree based optimization, to @file{@var{file}.original}.
3723 Dump after all tree based optimization, to @file{@var{file}.optimized}.
3726 Dump after function inlining, to @file{@var{file}.inlined}.
3729 @opindex fdump-tree-gimple
3730 Dump each function before and after the gimplification pass to a file. The
3731 file name is made by appending @file{.gimple} to the source file name.
3734 @opindex fdump-tree-cfg
3735 Dump the control flow graph of each function to a file. The file name is
3736 made by appending @file{.cfg} to the source file name.
3739 @opindex fdump-tree-vcg
3740 Dump the control flow graph of each function to a file in VCG format. The
3741 file name is made by appending @file{.vcg} to the source file name. Note
3742 that if the file contains more than one function, the generated file cannot
3743 be used directly by VCG. You will need to cut and paste each function's
3744 graph into its own separate file first.
3747 @opindex fdump-tree-ch
3748 Dump each function after copying loop headers. The file name is made by
3749 appending @file{.ch} to the source file name.
3752 @opindex fdump-tree-ssa
3753 Dump SSA related information to a file. The file name is made by appending
3754 @file{.ssa} to the source file name.
3757 @opindex fdump-tree-alias
3758 Dump aliasing information for each function. The file name is made by
3759 appending @file{.alias} to the source file name.
3762 @opindex fdump-tree-ccp
3763 Dump each function after CCP. The file name is made by appending
3764 @file{.ccp} to the source file name.
3767 @opindex fdump-tree-pre
3768 Dump trees after partial redundancy elimination. The file name is made
3769 by appending @file{.pre} to the source file name.
3772 @opindex fdump-tree-fre
3773 Dump trees after full redundancy elimination. The file name is made
3774 by appending @file{.fre} to the source file name.
3777 @opindex fdump-tree-dce
3778 Dump each function after dead code elimination. The file name is made by
3779 appending @file{.dce} to the source file name.
3782 @opindex fdump-tree-mudflap
3783 Dump each function after adding mudflap instrumentation. The file name is
3784 made by appending @file{.mudflap} to the source file name.
3787 @opindex fdump-tree-sra
3788 Dump each function after performing scalar replacement of aggregates. The
3789 file name is made by appending @file{.sra} to the source file name.
3792 @opindex fdump-tree-dom
3793 Dump each function after applying dominator tree optimizations. The file
3794 name is made by appending @file{.dom} to the source file name.
3797 @opindex fdump-tree-dse
3798 Dump each function after applying dead store elimination. The file
3799 name is made by appending @file{.dse} to the source file name.
3802 @opindex fdump-tree-phiopt
3803 Dump each function after optimizing PHI nodes into straightline code. The file
3804 name is made by appending @file{.phiopt} to the source file name.
3807 @opindex fdump-tree-forwprop
3808 Dump each function after forward propagating single use variables. The file
3809 name is made by appending @file{.forwprop} to the source file name.
3812 @opindex fdump-tree-copyrename
3813 Dump each function after applying the copy rename optimization. The file
3814 name is made by appending @file{.copyrename} to the source file name.
3817 @opindex fdump-tree-nrv
3818 Dump each function after applying the named return value optimization on
3819 generic trees. The file name is made by appending @file{.nrv} to the source
3823 @opindex fdump-tree-vect
3824 Dump each function after applying vectorization of loops. The file name is
3825 made by appending @file{.vect} to the source file name.
3828 @opindex fdump-tree-all
3829 Enable all the available tree dumps with the flags provided in this option.
3832 @item -frandom-seed=@var{string}
3833 @opindex frandom-string
3834 This option provides a seed that GCC uses when it would otherwise use
3835 random numbers. It is used to generate certain symbol names
3836 that have to be different in every compiled file. It is also used to
3837 place unique stamps in coverage data files and the object files that
3838 produce them. You can use the @option{-frandom-seed} option to produce
3839 reproducibly identical object files.
3841 The @var{string} should be different for every file you compile.
3843 @item -fsched-verbose=@var{n}
3844 @opindex fsched-verbose
3845 On targets that use instruction scheduling, this option controls the
3846 amount of debugging output the scheduler prints. This information is
3847 written to standard error, unless @option{-dS} or @option{-dR} is
3848 specified, in which case it is output to the usual dump
3849 listing file, @file{.sched} or @file{.sched2} respectively. However
3850 for @var{n} greater than nine, the output is always printed to standard
3853 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
3854 same information as @option{-dRS}. For @var{n} greater than one, it
3855 also output basic block probabilities, detailed ready list information
3856 and unit/insn info. For @var{n} greater than two, it includes RTL
3857 at abort point, control-flow and regions info. And for @var{n} over
3858 four, @option{-fsched-verbose} also includes dependence info.
3862 Store the usual ``temporary'' intermediate files permanently; place them
3863 in the current directory and name them based on the source file. Thus,
3864 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
3865 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
3866 preprocessed @file{foo.i} output file even though the compiler now
3867 normally uses an integrated preprocessor.
3871 Report the CPU time taken by each subprocess in the compilation
3872 sequence. For C source files, this is the compiler proper and assembler
3873 (plus the linker if linking is done). The output looks like this:
3880 The first number on each line is the ``user time,'' that is time spent
3881 executing the program itself. The second number is ``system time,''
3882 time spent executing operating system routines on behalf of the program.
3883 Both numbers are in seconds.
3885 @item -fvar-tracking
3886 @opindex fvar-tracking
3887 Run variable tracking pass. It computes where variables are stored at each
3888 position in code. Better debugging information is then generated
3889 (if the debugging information format supports this information).
3891 It is enabled by default when compiling with optimization (@option{-Os},
3892 @option{-O}, @option{-O2}, ...), debugging information (@option{-g}) and
3893 the debug info format supports it.
3895 @item -print-file-name=@var{library}
3896 @opindex print-file-name
3897 Print the full absolute name of the library file @var{library} that
3898 would be used when linking---and don't do anything else. With this
3899 option, GCC does not compile or link anything; it just prints the
3902 @item -print-multi-directory
3903 @opindex print-multi-directory
3904 Print the directory name corresponding to the multilib selected by any
3905 other switches present in the command line. This directory is supposed
3906 to exist in @env{GCC_EXEC_PREFIX}.
3908 @item -print-multi-lib
3909 @opindex print-multi-lib
3910 Print the mapping from multilib directory names to compiler switches
3911 that enable them. The directory name is separated from the switches by
3912 @samp{;}, and each switch starts with an @samp{@@} instead of the
3913 @samp{-}, without spaces between multiple switches. This is supposed to
3914 ease shell-processing.
3916 @item -print-prog-name=@var{program}
3917 @opindex print-prog-name
3918 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
3920 @item -print-libgcc-file-name
3921 @opindex print-libgcc-file-name
3922 Same as @option{-print-file-name=libgcc.a}.
3924 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
3925 but you do want to link with @file{libgcc.a}. You can do
3928 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
3931 @item -print-search-dirs
3932 @opindex print-search-dirs
3933 Print the name of the configured installation directory and a list of
3934 program and library directories @command{gcc} will search---and don't do anything else.
3936 This is useful when @command{gcc} prints the error message
3937 @samp{installation problem, cannot exec cpp0: No such file or directory}.
3938 To resolve this you either need to put @file{cpp0} and the other compiler
3939 components where @command{gcc} expects to find them, or you can set the environment
3940 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
3941 Don't forget the trailing '/'.
3942 @xref{Environment Variables}.
3945 @opindex dumpmachine
3946 Print the compiler's target machine (for example,
3947 @samp{i686-pc-linux-gnu})---and don't do anything else.
3950 @opindex dumpversion
3951 Print the compiler version (for example, @samp{3.0})---and don't do
3956 Print the compiler's built-in specs---and don't do anything else. (This
3957 is used when GCC itself is being built.) @xref{Spec Files}.
3959 @item -feliminate-unused-debug-types
3960 @opindex feliminate-unused-debug-types
3961 Normally, when producing DWARF2 output, GCC will emit debugging
3962 information for all types declared in a compilation
3963 unit, regardless of whether or not they are actually used
3964 in that compilation unit. Sometimes this is useful, such as
3965 if, in the debugger, you want to cast a value to a type that is
3966 not actually used in your program (but is declared). More often,
3967 however, this results in a significant amount of wasted space.
3968 With this option, GCC will avoid producing debug symbol output
3969 for types that are nowhere used in the source file being compiled.
3972 @node Optimize Options
3973 @section Options That Control Optimization
3974 @cindex optimize options
3975 @cindex options, optimization
3977 These options control various sorts of optimizations.
3979 Without any optimization option, the compiler's goal is to reduce the
3980 cost of compilation and to make debugging produce the expected
3981 results. Statements are independent: if you stop the program with a
3982 breakpoint between statements, you can then assign a new value to any
3983 variable or change the program counter to any other statement in the
3984 function and get exactly the results you would expect from the source
3987 Turning on optimization flags makes the compiler attempt to improve
3988 the performance and/or code size at the expense of compilation time
3989 and possibly the ability to debug the program.
3991 The compiler performs optimization based on the knowledge it has of
3992 the program. Optimization levels @option{-O2} and above, in
3993 particular, enable @emph{unit-at-a-time} mode, which allows the
3994 compiler to consider information gained from later functions in
3995 the file when compiling a function. Compiling multiple files at
3996 once to a single output file in @emph{unit-at-a-time} mode allows
3997 the compiler to use information gained from all of the files when
3998 compiling each of them.
4000 Not all optimizations are controlled directly by a flag. Only
4001 optimizations that have a flag are listed.
4008 Optimize. Optimizing compilation takes somewhat more time, and a lot
4009 more memory for a large function.
4011 With @option{-O}, the compiler tries to reduce code size and execution
4012 time, without performing any optimizations that take a great deal of
4015 @option{-O} turns on the following optimization flags:
4016 @gccoptlist{-fdefer-pop @gol
4017 -fmerge-constants @gol
4019 -floop-optimize @gol
4020 -fif-conversion @gol
4021 -fif-conversion2 @gol
4022 -fdelayed-branch @gol
4023 -fguess-branch-probability @gol
4026 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
4027 where doing so does not interfere with debugging.
4031 Optimize even more. GCC performs nearly all supported optimizations
4032 that do not involve a space-speed tradeoff. The compiler does not
4033 perform loop unrolling or function inlining when you specify @option{-O2}.
4034 As compared to @option{-O}, this option increases both compilation time
4035 and the performance of the generated code.
4037 @option{-O2} turns on all optimization flags specified by @option{-O}. It
4038 also turns on the following optimization flags:
4039 @gccoptlist{-fforce-mem @gol
4040 -foptimize-sibling-calls @gol
4041 -fstrength-reduce @gol
4042 -fcse-follow-jumps -fcse-skip-blocks @gol
4043 -frerun-cse-after-loop -frerun-loop-opt @gol
4044 -fgcse -fgcse-lm -fgcse-sm -fgcse-las @gol
4045 -fdelete-null-pointer-checks @gol
4046 -fexpensive-optimizations @gol
4048 -fschedule-insns -fschedule-insns2 @gol
4049 -fsched-interblock -fsched-spec @gol
4052 -freorder-blocks -freorder-functions @gol
4053 -fstrict-aliasing @gol
4054 -funit-at-a-time @gol
4055 -falign-functions -falign-jumps @gol
4056 -falign-loops -falign-labels @gol
4059 Please note the warning under @option{-fgcse} about
4060 invoking @option{-O2} on programs that use computed gotos.
4064 Optimize yet more. @option{-O3} turns on all optimizations specified by
4065 @option{-O2} and also turns on the @option{-finline-functions},
4066 @option{-fweb} and @option{-fgcse-after-reload} options.
4070 Do not optimize. This is the default.
4074 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
4075 do not typically increase code size. It also performs further
4076 optimizations designed to reduce code size.
4078 @option{-Os} disables the following optimization flags:
4079 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
4080 -falign-labels -freorder-blocks -freorder-blocks-and-partition -fprefetch-loop-arrays}
4082 If you use multiple @option{-O} options, with or without level numbers,
4083 the last such option is the one that is effective.
4086 Options of the form @option{-f@var{flag}} specify machine-independent
4087 flags. Most flags have both positive and negative forms; the negative
4088 form of @option{-ffoo} would be @option{-fno-foo}. In the table
4089 below, only one of the forms is listed---the one you typically will
4090 use. You can figure out the other form by either removing @samp{no-}
4093 The following options control specific optimizations. They are either
4094 activated by @option{-O} options or are related to ones that are. You
4095 can use the following flags in the rare cases when ``fine-tuning'' of
4096 optimizations to be performed is desired.
4099 @item -fno-default-inline
4100 @opindex fno-default-inline
4101 Do not make member functions inline by default merely because they are
4102 defined inside the class scope (C++ only). Otherwise, when you specify
4103 @w{@option{-O}}, member functions defined inside class scope are compiled
4104 inline by default; i.e., you don't need to add @samp{inline} in front of
4105 the member function name.
4107 @item -fno-defer-pop
4108 @opindex fno-defer-pop
4109 Always pop the arguments to each function call as soon as that function
4110 returns. For machines which must pop arguments after a function call,
4111 the compiler normally lets arguments accumulate on the stack for several
4112 function calls and pops them all at once.
4114 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4118 Force memory operands to be copied into registers before doing
4119 arithmetic on them. This produces better code by making all memory
4120 references potential common subexpressions. When they are not common
4121 subexpressions, instruction combination should eliminate the separate
4124 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4127 @opindex fforce-addr
4128 Force memory address constants to be copied into registers before
4129 doing arithmetic on them. This may produce better code just as
4130 @option{-fforce-mem} may.
4132 @item -fomit-frame-pointer
4133 @opindex fomit-frame-pointer
4134 Don't keep the frame pointer in a register for functions that
4135 don't need one. This avoids the instructions to save, set up and
4136 restore frame pointers; it also makes an extra register available
4137 in many functions. @strong{It also makes debugging impossible on
4140 On some machines, such as the VAX, this flag has no effect, because
4141 the standard calling sequence automatically handles the frame pointer
4142 and nothing is saved by pretending it doesn't exist. The
4143 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
4144 whether a target machine supports this flag. @xref{Registers,,Register
4145 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
4147 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4149 @item -foptimize-sibling-calls
4150 @opindex foptimize-sibling-calls
4151 Optimize sibling and tail recursive calls.
4153 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4157 Don't pay attention to the @code{inline} keyword. Normally this option
4158 is used to keep the compiler from expanding any functions inline.
4159 Note that if you are not optimizing, no functions can be expanded inline.
4161 @item -finline-functions
4162 @opindex finline-functions
4163 Integrate all simple functions into their callers. The compiler
4164 heuristically decides which functions are simple enough to be worth
4165 integrating in this way.
4167 If all calls to a given function are integrated, and the function is
4168 declared @code{static}, then the function is normally not output as
4169 assembler code in its own right.
4171 Enabled at level @option{-O3}.
4173 @item -finline-limit=@var{n}
4174 @opindex finline-limit
4175 By default, GCC limits the size of functions that can be inlined. This flag
4176 allows the control of this limit for functions that are explicitly marked as
4177 inline (i.e., marked with the inline keyword or defined within the class
4178 definition in c++). @var{n} is the size of functions that can be inlined in
4179 number of pseudo instructions (not counting parameter handling). The default
4180 value of @var{n} is 600.
4181 Increasing this value can result in more inlined code at
4182 the cost of compilation time and memory consumption. Decreasing usually makes
4183 the compilation faster and less code will be inlined (which presumably
4184 means slower programs). This option is particularly useful for programs that
4185 use inlining heavily such as those based on recursive templates with C++.
4187 Inlining is actually controlled by a number of parameters, which may be
4188 specified individually by using @option{--param @var{name}=@var{value}}.
4189 The @option{-finline-limit=@var{n}} option sets some of these parameters
4193 @item max-inline-insns-single
4194 is set to @var{n}/2.
4195 @item max-inline-insns-auto
4196 is set to @var{n}/2.
4197 @item min-inline-insns
4198 is set to 130 or @var{n}/4, whichever is smaller.
4199 @item max-inline-insns-rtl
4203 See below for a documentation of the individual
4204 parameters controlling inlining.
4206 @emph{Note:} pseudo instruction represents, in this particular context, an
4207 abstract measurement of function's size. In no way, it represents a count
4208 of assembly instructions and as such its exact meaning might change from one
4209 release to an another.
4211 @item -fkeep-inline-functions
4212 @opindex fkeep-inline-functions
4213 In C, emit @code{static} functions that are declared @code{inline}
4214 into the object file, even if the function has been inlined into all
4215 of its callers. This switch does not affect functions using the
4216 @code{extern inline} extension in GNU C. In C++, emit any and all
4217 inline functions into the object file.
4219 @item -fkeep-static-consts
4220 @opindex fkeep-static-consts
4221 Emit variables declared @code{static const} when optimization isn't turned
4222 on, even if the variables aren't referenced.
4224 GCC enables this option by default. If you want to force the compiler to
4225 check if the variable was referenced, regardless of whether or not
4226 optimization is turned on, use the @option{-fno-keep-static-consts} option.
4228 @item -fmerge-constants
4229 Attempt to merge identical constants (string constants and floating point
4230 constants) across compilation units.
4232 This option is the default for optimized compilation if the assembler and
4233 linker support it. Use @option{-fno-merge-constants} to inhibit this
4236 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4238 @item -fmerge-all-constants
4239 Attempt to merge identical constants and identical variables.
4241 This option implies @option{-fmerge-constants}. In addition to
4242 @option{-fmerge-constants} this considers e.g.@: even constant initialized
4243 arrays or initialized constant variables with integral or floating point
4244 types. Languages like C or C++ require each non-automatic variable to
4245 have distinct location, so using this option will result in non-conforming
4248 @item -fmodulo-sched
4249 @opindex fmodulo-sched
4250 Perform swing modulo scheduling immediately before the first scheduling
4251 pass. This pass looks at innermost loops and reorders their
4252 instructions by overlapping different iterations.
4256 Use a graph coloring register allocator. Currently this option is meant
4257 only for testing. Users should not specify this option, since it is not
4258 yet ready for production use.
4260 @item -fno-branch-count-reg
4261 @opindex fno-branch-count-reg
4262 Do not use ``decrement and branch'' instructions on a count register,
4263 but instead generate a sequence of instructions that decrement a
4264 register, compare it against zero, then branch based upon the result.
4265 This option is only meaningful on architectures that support such
4266 instructions, which include x86, PowerPC, IA-64 and S/390.
4268 The default is @option{-fbranch-count-reg}, enabled when
4269 @option{-fstrength-reduce} is enabled.
4271 @item -fno-function-cse
4272 @opindex fno-function-cse
4273 Do not put function addresses in registers; make each instruction that
4274 calls a constant function contain the function's address explicitly.
4276 This option results in less efficient code, but some strange hacks
4277 that alter the assembler output may be confused by the optimizations
4278 performed when this option is not used.
4280 The default is @option{-ffunction-cse}
4282 @item -fno-zero-initialized-in-bss
4283 @opindex fno-zero-initialized-in-bss
4284 If the target supports a BSS section, GCC by default puts variables that
4285 are initialized to zero into BSS@. This can save space in the resulting
4288 This option turns off this behavior because some programs explicitly
4289 rely on variables going to the data section. E.g., so that the
4290 resulting executable can find the beginning of that section and/or make
4291 assumptions based on that.
4293 The default is @option{-fzero-initialized-in-bss}.
4295 @item -fbounds-check
4296 @opindex fbounds-check
4297 For front-ends that support it, generate additional code to check that
4298 indices used to access arrays are within the declared range. This is
4299 currently only supported by the Java and Fortran front-ends, where
4300 this option defaults to true and false respectively.
4302 @item -fmudflap -fmudflapth -fmudflapir
4306 @cindex bounds checking
4308 For front-ends that support it (C and C++), instrument all risky
4309 pointer/array dereferencing operations, some standard library
4310 string/heap functions, and some other associated constructs with
4311 range/validity tests. Modules so instrumented should be immune to
4312 buffer overflows, invalid heap use, and some other classes of C/C++
4313 programming errors. The instrumentation relies on a separate runtime
4314 library (@file{libmudflap}), which will be linked into a program if
4315 @option{-fmudflap} is given at link time. Run-time behavior of the
4316 instrumented program is controlled by the @env{MUDFLAP_OPTIONS}
4317 environment variable. See @code{env MUDFLAP_OPTIONS=-help a.out}
4320 Use @option{-fmudflapth} instead of @option{-fmudflap} to compile and to
4321 link if your program is multi-threaded. Use @option{-fmudflapir}, in
4322 addition to @option{-fmudflap} or @option{-fmudflapth}, if
4323 instrumentation should ignore pointer reads. This produces less
4324 instrumentation (and therefore faster execution) and still provides
4325 some protection against outright memory corrupting writes, but allows
4326 erroneously read data to propagate within a program.
4328 @item -fstrength-reduce
4329 @opindex fstrength-reduce
4330 Perform the optimizations of loop strength reduction and
4331 elimination of iteration variables.
4333 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4335 @item -fthread-jumps
4336 @opindex fthread-jumps
4337 Perform optimizations where we check to see if a jump branches to a
4338 location where another comparison subsumed by the first is found. If
4339 so, the first branch is redirected to either the destination of the
4340 second branch or a point immediately following it, depending on whether
4341 the condition is known to be true or false.
4343 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4345 @item -fcse-follow-jumps
4346 @opindex fcse-follow-jumps
4347 In common subexpression elimination, scan through jump instructions
4348 when the target of the jump is not reached by any other path. For
4349 example, when CSE encounters an @code{if} statement with an
4350 @code{else} clause, CSE will follow the jump when the condition
4353 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4355 @item -fcse-skip-blocks
4356 @opindex fcse-skip-blocks
4357 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
4358 follow jumps which conditionally skip over blocks. When CSE
4359 encounters a simple @code{if} statement with no else clause,
4360 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
4361 body of the @code{if}.
4363 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4365 @item -frerun-cse-after-loop
4366 @opindex frerun-cse-after-loop
4367 Re-run common subexpression elimination after loop optimizations has been
4370 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4372 @item -frerun-loop-opt
4373 @opindex frerun-loop-opt
4374 Run the loop optimizer twice.
4376 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4380 Perform a global common subexpression elimination pass.
4381 This pass also performs global constant and copy propagation.
4383 @emph{Note:} When compiling a program using computed gotos, a GCC
4384 extension, you may get better runtime performance if you disable
4385 the global common subexpression elimination pass by adding
4386 @option{-fno-gcse} to the command line.
4388 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4392 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
4393 attempt to move loads which are only killed by stores into themselves. This
4394 allows a loop containing a load/store sequence to be changed to a load outside
4395 the loop, and a copy/store within the loop.
4397 Enabled by default when gcse is enabled.
4401 When @option{-fgcse-sm} is enabled, a store motion pass is run after
4402 global common subexpression elimination. This pass will attempt to move
4403 stores out of loops. When used in conjunction with @option{-fgcse-lm},
4404 loops containing a load/store sequence can be changed to a load before
4405 the loop and a store after the loop.
4407 Enabled by default when gcse is enabled.
4411 When @option{-fgcse-las} is enabled, the global common subexpression
4412 elimination pass eliminates redundant loads that come after stores to the
4413 same memory location (both partial and full redundancies).
4415 Enabled by default when gcse is enabled.
4417 @item -fgcse-after-reload
4418 @opindex fgcse-after-reload
4419 When @option{-fgcse-after-reload} is enabled, a redundant load elimination
4420 pass is performed after reload. The purpose of this pass is to cleanup
4423 @item -floop-optimize
4424 @opindex floop-optimize
4425 Perform loop optimizations: move constant expressions out of loops, simplify
4426 exit test conditions and optionally do strength-reduction as well.
4428 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4430 @item -floop-optimize2
4431 @opindex floop-optimize2
4432 Perform loop optimizations using the new loop optimizer. The optimizations
4433 (loop unrolling, peeling and unswitching, loop invariant motion) are enabled
4436 @item -fcrossjumping
4437 @opindex crossjumping
4438 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
4439 resulting code may or may not perform better than without cross-jumping.
4441 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4443 @item -fif-conversion
4444 @opindex if-conversion
4445 Attempt to transform conditional jumps into branch-less equivalents. This
4446 include use of conditional moves, min, max, set flags and abs instructions, and
4447 some tricks doable by standard arithmetics. The use of conditional execution
4448 on chips where it is available is controlled by @code{if-conversion2}.
4450 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4452 @item -fif-conversion2
4453 @opindex if-conversion2
4454 Use conditional execution (where available) to transform conditional jumps into
4455 branch-less equivalents.
4457 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4459 @item -fdelete-null-pointer-checks
4460 @opindex fdelete-null-pointer-checks
4461 Use global dataflow analysis to identify and eliminate useless checks
4462 for null pointers. The compiler assumes that dereferencing a null
4463 pointer would have halted the program. If a pointer is checked after
4464 it has already been dereferenced, it cannot be null.
4466 In some environments, this assumption is not true, and programs can
4467 safely dereference null pointers. Use
4468 @option{-fno-delete-null-pointer-checks} to disable this optimization
4469 for programs which depend on that behavior.
4471 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4473 @item -fexpensive-optimizations
4474 @opindex fexpensive-optimizations
4475 Perform a number of minor optimizations that are relatively expensive.
4477 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4479 @item -foptimize-register-move
4481 @opindex foptimize-register-move
4483 Attempt to reassign register numbers in move instructions and as
4484 operands of other simple instructions in order to maximize the amount of
4485 register tying. This is especially helpful on machines with two-operand
4488 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
4491 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4493 @item -fdelayed-branch
4494 @opindex fdelayed-branch
4495 If supported for the target machine, attempt to reorder instructions
4496 to exploit instruction slots available after delayed branch
4499 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4501 @item -fschedule-insns
4502 @opindex fschedule-insns
4503 If supported for the target machine, attempt to reorder instructions to
4504 eliminate execution stalls due to required data being unavailable. This
4505 helps machines that have slow floating point or memory load instructions
4506 by allowing other instructions to be issued until the result of the load
4507 or floating point instruction is required.
4509 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4511 @item -fschedule-insns2
4512 @opindex fschedule-insns2
4513 Similar to @option{-fschedule-insns}, but requests an additional pass of
4514 instruction scheduling after register allocation has been done. This is
4515 especially useful on machines with a relatively small number of
4516 registers and where memory load instructions take more than one cycle.
4518 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4520 @item -fno-sched-interblock
4521 @opindex fno-sched-interblock
4522 Don't schedule instructions across basic blocks. This is normally
4523 enabled by default when scheduling before register allocation, i.e.@:
4524 with @option{-fschedule-insns} or at @option{-O2} or higher.
4526 @item -fno-sched-spec
4527 @opindex fno-sched-spec
4528 Don't allow speculative motion of non-load instructions. This is normally
4529 enabled by default when scheduling before register allocation, i.e.@:
4530 with @option{-fschedule-insns} or at @option{-O2} or higher.
4532 @item -fsched-spec-load
4533 @opindex fsched-spec-load
4534 Allow speculative motion of some load instructions. This only makes
4535 sense when scheduling before register allocation, i.e.@: with
4536 @option{-fschedule-insns} or at @option{-O2} or higher.
4538 @item -fsched-spec-load-dangerous
4539 @opindex fsched-spec-load-dangerous
4540 Allow speculative motion of more load instructions. This only makes
4541 sense when scheduling before register allocation, i.e.@: with
4542 @option{-fschedule-insns} or at @option{-O2} or higher.
4544 @item -fsched-stalled-insns=@var{n}
4545 @opindex fsched-stalled-insns
4546 Define how many insns (if any) can be moved prematurely from the queue
4547 of stalled insns into the ready list, during the second scheduling pass.
4549 @item -fsched-stalled-insns-dep=@var{n}
4550 @opindex fsched-stalled-insns-dep
4551 Define how many insn groups (cycles) will be examined for a dependency
4552 on a stalled insn that is candidate for premature removal from the queue
4553 of stalled insns. Has an effect only during the second scheduling pass,
4554 and only if @option{-fsched-stalled-insns} is used and its value is not zero.
4556 @item -fsched2-use-superblocks
4557 @opindex fsched2-use-superblocks
4558 When scheduling after register allocation, do use superblock scheduling
4559 algorithm. Superblock scheduling allows motion across basic block boundaries
4560 resulting on faster schedules. This option is experimental, as not all machine
4561 descriptions used by GCC model the CPU closely enough to avoid unreliable
4562 results from the algorithm.
4564 This only makes sense when scheduling after register allocation, i.e.@: with
4565 @option{-fschedule-insns2} or at @option{-O2} or higher.
4567 @item -fsched2-use-traces
4568 @opindex fsched2-use-traces
4569 Use @option{-fsched2-use-superblocks} algorithm when scheduling after register
4570 allocation and additionally perform code duplication in order to increase the
4571 size of superblocks using tracer pass. See @option{-ftracer} for details on
4574 This mode should produce faster but significantly longer programs. Also
4575 without @code{-fbranch-probabilities} the traces constructed may not match the
4576 reality and hurt the performance. This only makes
4577 sense when scheduling after register allocation, i.e.@: with
4578 @option{-fschedule-insns2} or at @option{-O2} or higher.
4580 @item -freschedule-modulo-scheduled-loops
4581 @opindex fscheduling-in-modulo-scheduled-loops
4582 The modulo scheduling comes before the traditional scheduling, if a loop was modulo scheduled
4583 we may want to prevent the later scheduling passes from changing its schedule, we use this
4584 option to control that.
4586 @item -fcaller-saves
4587 @opindex fcaller-saves
4588 Enable values to be allocated in registers that will be clobbered by
4589 function calls, by emitting extra instructions to save and restore the
4590 registers around such calls. Such allocation is done only when it
4591 seems to result in better code than would otherwise be produced.
4593 This option is always enabled by default on certain machines, usually
4594 those which have no call-preserved registers to use instead.
4596 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4599 Perform Partial Redundancy Elimination (PRE) on trees. This flag is
4600 enabled by default at -O and higher.
4603 Perform Full Redundancy Elimination (FRE) on trees. The difference
4604 between FRE and PRE is that FRE only considers expressions
4605 that are computed on all paths leading to the redundant computation.
4606 This analysis faster than PRE, though it exposes fewer redundancies.
4607 This flag is enabled by default at -O and higher.
4610 Perform sparse conditional constant propagation (CCP) on trees. This flag
4611 is enabled by default at -O and higher.
4614 Perform dead code elimination (DCE) on trees. This flag is enabled by
4615 default at -O and higher.
4617 @item -ftree-dominator-opts
4618 Perform dead code elimination (DCE) on trees. This flag is enabled by
4619 default at -O and higher.
4622 Perform loop header copying on trees. This is beneficial since it increases
4623 effectiveness of code motion optimizations. It also saves one jump. This flag
4624 is enabled by default at -O and higher. It is not enabled for -Os, since it
4625 usually increases code size.
4627 @item -ftree-loop-optimize
4628 Perform loop optimizations on trees. This flag is enabled by default at -O
4631 @item -ftree-loop-linear
4632 Perform linear loop transformations on tree. This flag can improve cache
4633 performance and allow further loop optimizations to take place.
4636 Perform loop invariant motion on trees. This pass moves only invartiants that
4637 would be hard to handle on rtl level (function calls, operations that expand to
4638 nontrivial sequences of insns). With @option{-funswitch-loops} it also moves
4639 operands of conditions that are invariant out of the loop, so that we can use
4640 just trivial invariantness analysis in loop unswitching. The pass also includes
4644 Create a canonical counter for number of iterations in the loop for that
4645 determining number of iterations requires complicated analysis. Later
4646 optimizations then may determine the number easily. Useful especially
4647 in connection with unrolling.
4650 Perform induction variable optimizations (strength reduction, induction
4651 variable merging and induction variable elimination) on trees.
4654 Perform scalar replacement of aggregates. This pass replaces structure
4655 references with scalars to prevent committing structures to memory too
4656 early. This flag is enabled by default at -O and higher.
4658 @item -ftree-copyrename
4659 Perform copy renaming on trees. This pass attempts to rename compiler
4660 temporaries to other variables at copy locations, usually resulting in
4661 variable names which more closely resemble the original variables. This flag
4662 is enabled by default at -O and higher.
4665 Perform temporary expression replacement during the SSA->normal phase. Single
4666 use/single def temporaries are replaced at their use location with their
4667 defining expression. This results in non-GIMPLE code, but gives the expanders
4668 much more complex trees to work on resulting in better RTL generation. This is
4669 enabled by default at -O and higher.
4672 Perform live range splitting during the SSA->normal phase. Distinct live
4673 ranges of a variable are split into unique variables, allowing for better
4674 optimization later. This is enabled by default at -O and higher.
4676 @item -ftree-vectorize
4677 Perform loop vectorization on trees.
4681 Perform tail duplication to enlarge superblock size. This transformation
4682 simplifies the control flow of the function allowing other optimizations to do
4685 @item -funroll-loops
4686 @opindex funroll-loops
4687 Unroll loops whose number of iterations can be determined at compile
4688 time or upon entry to the loop. @option{-funroll-loops} implies both
4689 @option{-fstrength-reduce} and @option{-frerun-cse-after-loop}. This
4690 option makes code larger, and may or may not make it run faster.
4692 @item -funroll-all-loops
4693 @opindex funroll-all-loops
4694 Unroll all loops, even if their number of iterations is uncertain when
4695 the loop is entered. This usually makes programs run more slowly.
4696 @option{-funroll-all-loops} implies the same options as
4697 @option{-funroll-loops},
4699 @item -fsplit-ivs-in-unroller
4700 @opindex -fsplit-ivs-in-unroller
4701 Enables expressing of values of induction variables in later iterations
4702 of the unrolled loop using the value in the first iteration. This breaks
4703 long dependency chains, thus improving efficiency of the scheduling passes
4704 (for best results, @option{-fweb} should be used as well).
4706 Combination of @option{-fweb} and CSE is often sufficient to obtain the
4707 same effect. However in cases the loop body is more complicated than
4708 a single basic block, this is not reliable. It also does not work at all
4709 on some of the architectures due to restrictions in the CSE pass.
4711 This optimization is enabled by default.
4713 @item -fprefetch-loop-arrays
4714 @opindex fprefetch-loop-arrays
4715 If supported by the target machine, generate instructions to prefetch
4716 memory to improve the performance of loops that access large arrays.
4718 These options may generate better or worse code; results are highly
4719 dependent on the structure of loops within the source code.
4721 These two options are intended to be removed someday, once
4722 they have helped determine the efficacy of various
4723 approaches to improving loop optimizations.
4725 Please contact @w{@email{gcc@@gcc.gnu.org}}, and describe how use of
4726 these options affects the performance of your production code.
4727 Examples of code that runs @emph{slower} when these options are
4728 @emph{enabled} are very valuable.
4731 @itemx -fno-peephole2
4732 @opindex fno-peephole
4733 @opindex fno-peephole2
4734 Disable any machine-specific peephole optimizations. The difference
4735 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
4736 are implemented in the compiler; some targets use one, some use the
4737 other, a few use both.
4739 @option{-fpeephole} is enabled by default.
4740 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4742 @item -fno-guess-branch-probability
4743 @opindex fno-guess-branch-probability
4744 Do not guess branch probabilities using a randomized model.
4746 Sometimes GCC will opt to use a randomized model to guess branch
4747 probabilities, when none are available from either profiling feedback
4748 (@option{-fprofile-arcs}) or @samp{__builtin_expect}. This means that
4749 different runs of the compiler on the same program may produce different
4752 In a hard real-time system, people don't want different runs of the
4753 compiler to produce code that has different behavior; minimizing
4754 non-determinism is of paramount import. This switch allows users to
4755 reduce non-determinism, possibly at the expense of inferior
4758 The default is @option{-fguess-branch-probability} at levels
4759 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4761 @item -freorder-blocks
4762 @opindex freorder-blocks
4763 Reorder basic blocks in the compiled function in order to reduce number of
4764 taken branches and improve code locality.
4766 Enabled at levels @option{-O2}, @option{-O3}.
4768 @item -freorder-blocks-and-partition
4769 @opindex freorder-blocks-and-partition
4770 In addition to reordering basic blocks in the compiled function, in order
4771 to reduce number of taken branches, partitions hot and cold basic blocks
4772 into separate sections of the assembly and .o files, to improve
4773 paging and cache locality performance.
4775 This optimization is automatically turned off in the presence of
4776 exception handling, for linkonce sections, for functions with a user-defined
4777 section attribute and on any architecture that does not support named
4780 @item -freorder-functions
4781 @opindex freorder-functions
4782 Reorder basic blocks in the compiled function in order to reduce number of
4783 taken branches and improve code locality. This is implemented by using special
4784 subsections @code{.text.hot} for most frequently executed functions and
4785 @code{.text.unlikely} for unlikely executed functions. Reordering is done by
4786 the linker so object file format must support named sections and linker must
4787 place them in a reasonable way.
4789 Also profile feedback must be available in to make this option effective. See
4790 @option{-fprofile-arcs} for details.
4792 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4794 @item -fstrict-aliasing
4795 @opindex fstrict-aliasing
4796 Allows the compiler to assume the strictest aliasing rules applicable to
4797 the language being compiled. For C (and C++), this activates
4798 optimizations based on the type of expressions. In particular, an
4799 object of one type is assumed never to reside at the same address as an
4800 object of a different type, unless the types are almost the same. For
4801 example, an @code{unsigned int} can alias an @code{int}, but not a
4802 @code{void*} or a @code{double}. A character type may alias any other
4805 Pay special attention to code like this:
4818 The practice of reading from a different union member than the one most
4819 recently written to (called ``type-punning'') is common. Even with
4820 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
4821 is accessed through the union type. So, the code above will work as
4822 expected. However, this code might not:
4833 Every language that wishes to perform language-specific alias analysis
4834 should define a function that computes, given an @code{tree}
4835 node, an alias set for the node. Nodes in different alias sets are not
4836 allowed to alias. For an example, see the C front-end function
4837 @code{c_get_alias_set}.
4839 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4841 @item -falign-functions
4842 @itemx -falign-functions=@var{n}
4843 @opindex falign-functions
4844 Align the start of functions to the next power-of-two greater than
4845 @var{n}, skipping up to @var{n} bytes. For instance,
4846 @option{-falign-functions=32} aligns functions to the next 32-byte
4847 boundary, but @option{-falign-functions=24} would align to the next
4848 32-byte boundary only if this can be done by skipping 23 bytes or less.
4850 @option{-fno-align-functions} and @option{-falign-functions=1} are
4851 equivalent and mean that functions will not be aligned.
4853 Some assemblers only support this flag when @var{n} is a power of two;
4854 in that case, it is rounded up.
4856 If @var{n} is not specified or is zero, use a machine-dependent default.
4858 Enabled at levels @option{-O2}, @option{-O3}.
4860 @item -falign-labels
4861 @itemx -falign-labels=@var{n}
4862 @opindex falign-labels
4863 Align all branch targets to a power-of-two boundary, skipping up to
4864 @var{n} bytes like @option{-falign-functions}. This option can easily
4865 make code slower, because it must insert dummy operations for when the
4866 branch target is reached in the usual flow of the code.
4868 @option{-fno-align-labels} and @option{-falign-labels=1} are
4869 equivalent and mean that labels will not be aligned.
4871 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
4872 are greater than this value, then their values are used instead.
4874 If @var{n} is not specified or is zero, use a machine-dependent default
4875 which is very likely to be @samp{1}, meaning no alignment.
4877 Enabled at levels @option{-O2}, @option{-O3}.
4880 @itemx -falign-loops=@var{n}
4881 @opindex falign-loops
4882 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
4883 like @option{-falign-functions}. The hope is that the loop will be
4884 executed many times, which will make up for any execution of the dummy
4887 @option{-fno-align-loops} and @option{-falign-loops=1} are
4888 equivalent and mean that loops will not be aligned.
4890 If @var{n} is not specified or is zero, use a machine-dependent default.
4892 Enabled at levels @option{-O2}, @option{-O3}.
4895 @itemx -falign-jumps=@var{n}
4896 @opindex falign-jumps
4897 Align branch targets to a power-of-two boundary, for branch targets
4898 where the targets can only be reached by jumping, skipping up to @var{n}
4899 bytes like @option{-falign-functions}. In this case, no dummy operations
4902 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
4903 equivalent and mean that loops will not be aligned.
4905 If @var{n} is not specified or is zero, use a machine-dependent default.
4907 Enabled at levels @option{-O2}, @option{-O3}.
4909 @item -funit-at-a-time
4910 @opindex funit-at-a-time
4911 Parse the whole compilation unit before starting to produce code.
4912 This allows some extra optimizations to take place but consumes
4913 more memory (in general). There are some compatibility issues
4914 with @emph{unit-at-at-time} mode:
4917 enabling @emph{unit-at-a-time} mode may change the order
4918 in which functions, variables, and top-level @code{asm} statements
4919 are emitted, and will likely break code relying on some particular
4920 ordering. The majority of such top-level @code{asm} statements,
4921 though, can be replaced by @code{section} attributes.
4924 @emph{unit-at-a-time} mode removes unreferenced static variables
4925 and functions are removed. This may result in undefined references
4926 when an @code{asm} statement refers directly to variables or functions
4927 that are otherwise unused. In that case either the variable/function
4928 shall be listed as an operand of the @code{asm} statement operand or,
4929 in the case of top-level @code{asm} statements the attribute @code{used}
4930 shall be used on the declaration.
4933 Static functions now can use non-standard passing conventions that
4934 may break @code{asm} statements calling functions directly. Again,
4935 attribute @code{used} will prevent this behavior.
4938 As a temporary workaround, @option{-fno-unit-at-a-time} can be used,
4939 but this scheme may not be supported by future releases of GCC.
4941 Enabled at levels @option{-O2}, @option{-O3}.
4945 Constructs webs as commonly used for register allocation purposes and assign
4946 each web individual pseudo register. This allows the register allocation pass
4947 to operate on pseudos directly, but also strengthens several other optimization
4948 passes, such as CSE, loop optimizer and trivial dead code remover. It can,
4949 however, make debugging impossible, since variables will no longer stay in a
4952 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os},
4953 on targets where the default format for debugging information supports
4956 @item -fno-cprop-registers
4957 @opindex fno-cprop-registers
4958 After register allocation and post-register allocation instruction splitting,
4959 we perform a copy-propagation pass to try to reduce scheduling dependencies
4960 and occasionally eliminate the copy.
4962 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4964 @item -fprofile-generate
4965 @opindex fprofile-generate
4967 Enable options usually used for instrumenting application to produce
4968 profile useful for later recompilation with profile feedback based
4969 optimization. You must use @code{-fprofile-generate} both when
4970 compiling and when linking your program.
4972 The following options are enabled: @code{-fprofile-arcs}, @code{-fprofile-values}, @code{-fvpt}.
4975 @opindex fprofile-use
4976 Enable profile feedback directed optimizations, and optimizations
4977 generally profitable only with profile feedback available.
4979 The following options are enabled: @code{-fbranch-probabilities},
4980 @code{-fvpt}, @code{-funroll-loops}, @code{-fpeel-loops}, @code{-ftracer}.
4984 The following options control compiler behavior regarding floating
4985 point arithmetic. These options trade off between speed and
4986 correctness. All must be specifically enabled.
4990 @opindex ffloat-store
4991 Do not store floating point variables in registers, and inhibit other
4992 options that might change whether a floating point value is taken from a
4995 @cindex floating point precision
4996 This option prevents undesirable excess precision on machines such as
4997 the 68000 where the floating registers (of the 68881) keep more
4998 precision than a @code{double} is supposed to have. Similarly for the
4999 x86 architecture. For most programs, the excess precision does only
5000 good, but a few programs rely on the precise definition of IEEE floating
5001 point. Use @option{-ffloat-store} for such programs, after modifying
5002 them to store all pertinent intermediate computations into variables.
5006 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations}, @*
5007 @option{-fno-trapping-math}, @option{-ffinite-math-only},
5008 @option{-fno-rounding-math} and @option{-fno-signaling-nans}.
5010 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
5012 This option should never be turned on by any @option{-O} option since
5013 it can result in incorrect output for programs which depend on
5014 an exact implementation of IEEE or ISO rules/specifications for
5017 @item -fno-math-errno
5018 @opindex fno-math-errno
5019 Do not set ERRNO after calling math functions that are executed
5020 with a single instruction, e.g., sqrt. A program that relies on
5021 IEEE exceptions for math error handling may want to use this flag
5022 for speed while maintaining IEEE arithmetic compatibility.
5024 This option should never be turned on by any @option{-O} option since
5025 it can result in incorrect output for programs which depend on
5026 an exact implementation of IEEE or ISO rules/specifications for
5029 The default is @option{-fmath-errno}.
5031 @item -funsafe-math-optimizations
5032 @opindex funsafe-math-optimizations
5033 Allow optimizations for floating-point arithmetic that (a) assume
5034 that arguments and results are valid and (b) may violate IEEE or
5035 ANSI standards. When used at link-time, it may include libraries
5036 or startup files that change the default FPU control word or other
5037 similar optimizations.
5039 This option should never be turned on by any @option{-O} option since
5040 it can result in incorrect output for programs which depend on
5041 an exact implementation of IEEE or ISO rules/specifications for
5044 The default is @option{-fno-unsafe-math-optimizations}.
5046 @item -ffinite-math-only
5047 @opindex ffinite-math-only
5048 Allow optimizations for floating-point arithmetic that assume
5049 that arguments and results are not NaNs or +-Infs.
5051 This option should never be turned on by any @option{-O} option since
5052 it can result in incorrect output for programs which depend on
5053 an exact implementation of IEEE or ISO rules/specifications.
5055 The default is @option{-fno-finite-math-only}.
5057 @item -fno-trapping-math
5058 @opindex fno-trapping-math
5059 Compile code assuming that floating-point operations cannot generate
5060 user-visible traps. These traps include division by zero, overflow,
5061 underflow, inexact result and invalid operation. This option implies
5062 @option{-fno-signaling-nans}. Setting this option may allow faster
5063 code if one relies on ``non-stop'' IEEE arithmetic, for example.
5065 This option should never be turned on by any @option{-O} option since
5066 it can result in incorrect output for programs which depend on
5067 an exact implementation of IEEE or ISO rules/specifications for
5070 The default is @option{-ftrapping-math}.
5072 @item -frounding-math
5073 @opindex frounding-math
5074 Disable transformations and optimizations that assume default floating
5075 point rounding behavior. This is round-to-zero for all floating point
5076 to integer conversions, and round-to-nearest for all other arithmetic
5077 truncations. This option should be specified for programs that change
5078 the FP rounding mode dynamically, or that may be executed with a
5079 non-default rounding mode. This option disables constant folding of
5080 floating point expressions at compile-time (which may be affected by
5081 rounding mode) and arithmetic transformations that are unsafe in the
5082 presence of sign-dependent rounding modes.
5084 The default is @option{-fno-rounding-math}.
5086 This option is experimental and does not currently guarantee to
5087 disable all GCC optimizations that are affected by rounding mode.
5088 Future versions of GCC may provide finer control of this setting
5089 using C99's @code{FENV_ACCESS} pragma. This command line option
5090 will be used to specify the default state for @code{FENV_ACCESS}.
5092 @item -fsignaling-nans
5093 @opindex fsignaling-nans
5094 Compile code assuming that IEEE signaling NaNs may generate user-visible
5095 traps during floating-point operations. Setting this option disables
5096 optimizations that may change the number of exceptions visible with
5097 signaling NaNs. This option implies @option{-ftrapping-math}.
5099 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
5102 The default is @option{-fno-signaling-nans}.
5104 This option is experimental and does not currently guarantee to
5105 disable all GCC optimizations that affect signaling NaN behavior.
5107 @item -fsingle-precision-constant
5108 @opindex fsingle-precision-constant
5109 Treat floating point constant as single precision constant instead of
5110 implicitly converting it to double precision constant.
5115 The following options control optimizations that may improve
5116 performance, but are not enabled by any @option{-O} options. This
5117 section includes experimental options that may produce broken code.
5120 @item -fbranch-probabilities
5121 @opindex fbranch-probabilities
5122 After running a program compiled with @option{-fprofile-arcs}
5123 (@pxref{Debugging Options,, Options for Debugging Your Program or
5124 @command{gcc}}), you can compile it a second time using
5125 @option{-fbranch-probabilities}, to improve optimizations based on
5126 the number of times each branch was taken. When the program
5127 compiled with @option{-fprofile-arcs} exits it saves arc execution
5128 counts to a file called @file{@var{sourcename}.gcda} for each source
5129 file The information in this data file is very dependent on the
5130 structure of the generated code, so you must use the same source code
5131 and the same optimization options for both compilations.
5133 With @option{-fbranch-probabilities}, GCC puts a
5134 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
5135 These can be used to improve optimization. Currently, they are only
5136 used in one place: in @file{reorg.c}, instead of guessing which path a
5137 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
5138 exactly determine which path is taken more often.
5140 @item -fprofile-values
5141 @opindex fprofile-values
5142 If combined with @option{-fprofile-arcs}, it adds code so that some
5143 data about values of expressions in the program is gathered.
5145 With @option{-fbranch-probabilities}, it reads back the data gathered
5146 from profiling values of expressions and adds @samp{REG_VALUE_PROFILE}
5147 notes to instructions for their later usage in optimizations.
5149 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
5153 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
5154 a code to gather information about values of expressions.
5156 With @option{-fbranch-probabilities}, it reads back the data gathered
5157 and actually performs the optimizations based on them.
5158 Currently the optimizations include specialization of division operation
5159 using the knowledge about the value of the denominator.
5161 @item -fspeculative-prefetching
5162 @opindex fspeculative-prefetching
5163 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
5164 a code to gather information about addresses of memory references in the
5167 With @option{-fbranch-probabilities}, it reads back the data gathered
5168 and issues prefetch instructions according to them. In addition to the opportunities
5169 noticed by @option{-fprefetch-loop-arrays}, it also notices more complicated
5170 memory access patterns -- for example accesses to the data stored in linked
5171 list whose elements are usually allocated sequentially.
5173 In order to prevent issuing double prefetches, usage of
5174 @option{-fspeculative-prefetching} implies @option{-fno-prefetch-loop-arrays}.
5176 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
5178 @item -frename-registers
5179 @opindex frename-registers
5180 Attempt to avoid false dependencies in scheduled code by making use
5181 of registers left over after register allocation. This optimization
5182 will most benefit processors with lots of registers. Depending on the
5183 debug information format adopted by the target, however, it can
5184 make debugging impossible, since variables will no longer stay in
5185 a ``home register''.
5187 Not enabled by default at any level because it has known bugs.
5191 Use a graph coloring register allocator. Currently this option is meant
5192 for testing, so we are interested to hear about miscompilations with
5197 Perform tail duplication to enlarge superblock size. This transformation
5198 simplifies the control flow of the function allowing other optimizations to do
5201 Enabled with @option{-fprofile-use}.
5203 @item -funroll-loops
5204 @opindex funroll-loops
5205 Unroll loops whose number of iterations can be determined at compile time or
5206 upon entry to the loop. @option{-funroll-loops} implies
5207 @option{-frerun-cse-after-loop}. It also turns on complete loop peeling
5208 (i.e.@: complete removal of loops with small constant number of iterations).
5209 This option makes code larger, and may or may not make it run faster.
5211 Enabled with @option{-fprofile-use}.
5213 @item -funroll-all-loops
5214 @opindex funroll-all-loops
5215 Unroll all loops, even if their number of iterations is uncertain when
5216 the loop is entered. This usually makes programs run more slowly.
5217 @option{-funroll-all-loops} implies the same options as
5218 @option{-funroll-loops}.
5221 @opindex fpeel-loops
5222 Peels the loops for that there is enough information that they do not
5223 roll much (from profile feedback). It also turns on complete loop peeling
5224 (i.e.@: complete removal of loops with small constant number of iterations).
5226 Enabled with @option{-fprofile-use}.
5228 @item -fmove-loop-invariants
5229 @opindex fmove-loop-invariants
5230 Enables the loop invariant motion pass in the new loop optimizer. Enabled
5231 at level @option{-O1}
5233 @item -funswitch-loops
5234 @opindex funswitch-loops
5235 Move branches with loop invariant conditions out of the loop, with duplicates
5236 of the loop on both branches (modified according to result of the condition).
5238 @item -fprefetch-loop-arrays
5239 @opindex fprefetch-loop-arrays
5240 If supported by the target machine, generate instructions to prefetch
5241 memory to improve the performance of loops that access large arrays.
5243 Disabled at level @option{-Os}.
5245 @item -ffunction-sections
5246 @itemx -fdata-sections
5247 @opindex ffunction-sections
5248 @opindex fdata-sections
5249 Place each function or data item into its own section in the output
5250 file if the target supports arbitrary sections. The name of the
5251 function or the name of the data item determines the section's name
5254 Use these options on systems where the linker can perform optimizations
5255 to improve locality of reference in the instruction space. Most systems
5256 using the ELF object format and SPARC processors running Solaris 2 have
5257 linkers with such optimizations. AIX may have these optimizations in
5260 Only use these options when there are significant benefits from doing
5261 so. When you specify these options, the assembler and linker will
5262 create larger object and executable files and will also be slower.
5263 You will not be able to use @code{gprof} on all systems if you
5264 specify this option and you may have problems with debugging if
5265 you specify both this option and @option{-g}.
5267 @item -fbranch-target-load-optimize
5268 @opindex fbranch-target-load-optimize
5269 Perform branch target register load optimization before prologue / epilogue
5271 The use of target registers can typically be exposed only during reload,
5272 thus hoisting loads out of loops and doing inter-block scheduling needs
5273 a separate optimization pass.
5275 @item -fbranch-target-load-optimize2
5276 @opindex fbranch-target-load-optimize2
5277 Perform branch target register load optimization after prologue / epilogue
5280 @item -fbtr-bb-exclusive
5281 @opindex fbtr-bb-exclusive
5282 When performing branch target register load optimization, don't reuse
5283 branch target registers in within any basic block.
5285 @item --param @var{name}=@var{value}
5287 In some places, GCC uses various constants to control the amount of
5288 optimization that is done. For example, GCC will not inline functions
5289 that contain more that a certain number of instructions. You can
5290 control some of these constants on the command-line using the
5291 @option{--param} option.
5293 The names of specific parameters, and the meaning of the values, are
5294 tied to the internals of the compiler, and are subject to change
5295 without notice in future releases.
5297 In each case, the @var{value} is an integer. The allowable choices for
5298 @var{name} are given in the following table:
5301 @item max-crossjump-edges
5302 The maximum number of incoming edges to consider for crossjumping.
5303 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
5304 the number of edges incoming to each block. Increasing values mean
5305 more aggressive optimization, making the compile time increase with
5306 probably small improvement in executable size.
5308 @item min-crossjump-insns
5309 The minimum number of instructions which must be matched at the end
5310 of two blocks before crossjumping will be performed on them. This
5311 value is ignored in the case where all instructions in the block being
5312 crossjumped from are matched. The default value is 5.
5314 @item max-delay-slot-insn-search
5315 The maximum number of instructions to consider when looking for an
5316 instruction to fill a delay slot. If more than this arbitrary number of
5317 instructions is searched, the time savings from filling the delay slot
5318 will be minimal so stop searching. Increasing values mean more
5319 aggressive optimization, making the compile time increase with probably
5320 small improvement in executable run time.
5322 @item max-delay-slot-live-search
5323 When trying to fill delay slots, the maximum number of instructions to
5324 consider when searching for a block with valid live register
5325 information. Increasing this arbitrarily chosen value means more
5326 aggressive optimization, increasing the compile time. This parameter
5327 should be removed when the delay slot code is rewritten to maintain the
5330 @item max-gcse-memory
5331 The approximate maximum amount of memory that will be allocated in
5332 order to perform the global common subexpression elimination
5333 optimization. If more memory than specified is required, the
5334 optimization will not be done.
5336 @item max-gcse-passes
5337 The maximum number of passes of GCSE to run. The default is 1.
5339 @item max-pending-list-length
5340 The maximum number of pending dependencies scheduling will allow
5341 before flushing the current state and starting over. Large functions
5342 with few branches or calls can create excessively large lists which
5343 needlessly consume memory and resources.
5345 @item max-inline-insns-single
5346 Several parameters control the tree inliner used in gcc.
5347 This number sets the maximum number of instructions (counted in GCC's
5348 internal representation) in a single function that the tree inliner
5349 will consider for inlining. This only affects functions declared
5350 inline and methods implemented in a class declaration (C++).
5351 The default value is 500.
5353 @item max-inline-insns-auto
5354 When you use @option{-finline-functions} (included in @option{-O3}),
5355 a lot of functions that would otherwise not be considered for inlining
5356 by the compiler will be investigated. To those functions, a different
5357 (more restrictive) limit compared to functions declared inline can
5359 The default value is 120.
5361 @item large-function-insns
5362 The limit specifying really large functions. For functions greater than this
5363 limit inlining is constrained by @option{--param large-function-growth}.
5364 This parameter is useful primarily to avoid extreme compilation time caused by non-linear
5365 algorithms used by the backend.
5366 This parameter is ignored when @option{-funit-at-a-time} is not used.
5367 The default value is 3000.
5369 @item large-function-growth
5370 Specifies maximal growth of large function caused by inlining in percents.
5371 This parameter is ignored when @option{-funit-at-a-time} is not used.
5372 The default value is 200.
5374 @item inline-unit-growth
5375 Specifies maximal overall growth of the compilation unit caused by inlining.
5376 This parameter is ignored when @option{-funit-at-a-time} is not used.
5377 The default value is 150.
5379 @item max-inline-insns-recursive
5380 @itemx max-inline-insns-recursive-auto
5381 Specifies maximum number of instructions out-of-line copy of self recursive inline
5382 function can grow into by performing recursive inlining.
5384 For functions declared inline @option{--param max-inline-insns-recursive} is
5385 taken into acount. For function not declared inline, recursive inlining
5386 happens only when @option{-finline-functions} (included in @option{-O3}) is
5387 enabled and @option{--param max-inline-insns-recursive-auto} is used. The
5388 default value is 500.
5390 @item max-inline-recursive-depth
5391 @itemx max-inline-recursive-depth-auto
5392 Specifies maximum recursion depth used by the recursive inlining.
5394 For functions declared inline @option{--param max-inline-recursive-depth} is
5395 taken into acount. For function not declared inline, recursive inlining
5396 happens only when @option{-finline-functions} (included in @option{-O3}) is
5397 enabled and @option{--param max-inline-recursive-depth-auto} is used. The
5398 default value is 500.
5400 @item max-inline-insns-rtl
5401 For languages that use the RTL inliner (this happens at a later stage
5402 than tree inlining), you can set the maximum allowable size (counted
5403 in RTL instructions) for the RTL inliner with this parameter.
5404 The default value is 600.
5406 @item max-unrolled-insns
5407 The maximum number of instructions that a loop should have if that loop
5408 is unrolled, and if the loop is unrolled, it determines how many times
5409 the loop code is unrolled.
5411 @item max-average-unrolled-insns
5412 The maximum number of instructions biased by probabilities of their execution
5413 that a loop should have if that loop is unrolled, and if the loop is unrolled,
5414 it determines how many times the loop code is unrolled.
5416 @item max-unroll-times
5417 The maximum number of unrollings of a single loop.
5419 @item max-peeled-insns
5420 The maximum number of instructions that a loop should have if that loop
5421 is peeled, and if the loop is peeled, it determines how many times
5422 the loop code is peeled.
5424 @item max-peel-times
5425 The maximum number of peelings of a single loop.
5427 @item max-completely-peeled-insns
5428 The maximum number of insns of a completely peeled loop.
5430 @item max-completely-peel-times
5431 The maximum number of iterations of a loop to be suitable for complete peeling.
5433 @item max-unswitch-insns
5434 The maximum number of insns of an unswitched loop.
5436 @item max-unswitch-level
5437 The maximum number of branches unswitched in a single loop.
5440 The minimum cost of an expensive expression in the loop invariant motion.
5442 @item iv-consider-all-candidates-bound
5443 Bound on number of candidates for induction variables below that
5444 all candidates are considered for each use in induction variable
5445 optimizations. Only the most relevant candidates are considered
5446 if there are more candidates, to avoid quadratic time complexity.
5448 @item iv-max-considered-uses
5449 The induction variable optimizations give up on loops that contain more
5450 induction variable uses.
5452 @item max-iterations-to-track
5454 The maximum number of iterations of a loop the brute force algorithm
5455 for analysis of # of iterations of the loop tries to evaluate.
5457 @item hot-bb-count-fraction
5458 Select fraction of the maximal count of repetitions of basic block in program
5459 given basic block needs to have to be considered hot.
5461 @item hot-bb-frequency-fraction
5462 Select fraction of the maximal frequency of executions of basic block in
5463 function given basic block needs to have to be considered hot
5465 @item tracer-dynamic-coverage
5466 @itemx tracer-dynamic-coverage-feedback
5468 This value is used to limit superblock formation once the given percentage of
5469 executed instructions is covered. This limits unnecessary code size
5472 The @option{tracer-dynamic-coverage-feedback} is used only when profile
5473 feedback is available. The real profiles (as opposed to statically estimated
5474 ones) are much less balanced allowing the threshold to be larger value.
5476 @item tracer-max-code-growth
5477 Stop tail duplication once code growth has reached given percentage. This is
5478 rather hokey argument, as most of the duplicates will be eliminated later in
5479 cross jumping, so it may be set to much higher values than is the desired code
5482 @item tracer-min-branch-ratio
5484 Stop reverse growth when the reverse probability of best edge is less than this
5485 threshold (in percent).
5487 @item tracer-min-branch-ratio
5488 @itemx tracer-min-branch-ratio-feedback
5490 Stop forward growth if the best edge do have probability lower than this
5493 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
5494 compilation for profile feedback and one for compilation without. The value
5495 for compilation with profile feedback needs to be more conservative (higher) in
5496 order to make tracer effective.
5498 @item max-cse-path-length
5500 Maximum number of basic blocks on path that cse considers. The default is 10.
5502 @item global-var-threshold
5504 Counts the number of function calls (N) and the number of
5505 call-clobbered variables (V). If NxV is larger than this limit, a
5506 single artificial variable will be created to represent all the
5507 call-clobbered variables at function call sites. This artificial
5508 variable will then be made to alias every call-clobbered variable.
5509 (done as int * size_t on the host machine; beware overflow).
5511 @item max-aliased-vops
5513 Maximum number of virtual operands allowed to represent aliases
5514 before triggering the alias grouping heuristic. Alias grouping
5515 reduces compile times and memory consumption needed for aliasing at
5516 the expense of precision loss in alias information.
5518 @item ggc-min-expand
5520 GCC uses a garbage collector to manage its own memory allocation. This
5521 parameter specifies the minimum percentage by which the garbage
5522 collector's heap should be allowed to expand between collections.
5523 Tuning this may improve compilation speed; it has no effect on code
5526 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
5527 RAM >= 1GB. If @code{getrlimit} is available, the notion of "RAM" is
5528 the smallest of actual RAM and RLIMIT_DATA or RLIMIT_AS. If
5529 GCC is not able to calculate RAM on a particular platform, the lower
5530 bound of 30% is used. Setting this parameter and
5531 @option{ggc-min-heapsize} to zero causes a full collection to occur at
5532 every opportunity. This is extremely slow, but can be useful for
5535 @item ggc-min-heapsize
5537 Minimum size of the garbage collector's heap before it begins bothering
5538 to collect garbage. The first collection occurs after the heap expands
5539 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
5540 tuning this may improve compilation speed, and has no effect on code
5543 The default is the smaller of RAM/8, RLIMIT_RSS, or a limit which
5544 tries to ensure that RLIMIT_DATA or RLIMIT_AS are not exceeded, but
5545 with a lower bound of 4096 (four megabytes) and an upper bound of
5546 131072 (128 megabytes). If GCC is not able to calculate RAM on a
5547 particular platform, the lower bound is used. Setting this parameter
5548 very large effectively disables garbage collection. Setting this
5549 parameter and @option{ggc-min-expand} to zero causes a full collection
5550 to occur at every opportunity.
5552 @item max-reload-search-insns
5553 The maximum number of instruction reload should look backward for equivalent
5554 register. Increasing values mean more aggressive optimization, making the
5555 compile time increase with probably slightly better performance. The default
5558 @item max-cselib-memory-location
5559 The maximum number of memory locations cselib should take into acount.
5560 Increasing values mean more aggressive optimization, making the compile time
5561 increase with probably slightly better performance. The default value is 500.
5563 @item reorder-blocks-duplicate
5564 @itemx reorder-blocks-duplicate-feedback
5566 Used by basic block reordering pass to decide whether to use unconditional
5567 branch or duplicate the code on its destination. Code is duplicated when its
5568 estimated size is smaller than this value multiplied by the estimated size of
5569 unconditional jump in the hot spots of the program.
5571 The @option{reorder-block-duplicate-feedback} is used only when profile
5572 feedback is available and may be set to higher values than
5573 @option{reorder-block-duplicate} since information about the hot spots is more
5576 @item max-sched-region-blocks
5577 The maximum number of blocks in a region to be considered for
5578 interblock scheduling. The default value is 10.
5580 @item max-sched-region-insns
5581 The maximum number of insns in a region to be considered for
5582 interblock scheduling. The default value is 100.
5584 @item integer-share-limit
5585 Small integer constants can use a shared data structure, reducing the
5586 compiler's memory usage and increasing its speed. This sets the maximum
5587 value of a shared integer constant's. The default value is 256.
5592 @node Preprocessor Options
5593 @section Options Controlling the Preprocessor
5594 @cindex preprocessor options
5595 @cindex options, preprocessor
5597 These options control the C preprocessor, which is run on each C source
5598 file before actual compilation.
5600 If you use the @option{-E} option, nothing is done except preprocessing.
5601 Some of these options make sense only together with @option{-E} because
5602 they cause the preprocessor output to be unsuitable for actual
5607 You can use @option{-Wp,@var{option}} to bypass the compiler driver
5608 and pass @var{option} directly through to the preprocessor. If
5609 @var{option} contains commas, it is split into multiple options at the
5610 commas. However, many options are modified, translated or interpreted
5611 by the compiler driver before being passed to the preprocessor, and
5612 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
5613 interface is undocumented and subject to change, so whenever possible
5614 you should avoid using @option{-Wp} and let the driver handle the
5617 @item -Xpreprocessor @var{option}
5618 @opindex preprocessor
5619 Pass @var{option} as an option to the preprocessor. You can use this to
5620 supply system-specific preprocessor options which GCC does not know how to
5623 If you want to pass an option that takes an argument, you must use
5624 @option{-Xpreprocessor} twice, once for the option and once for the argument.
5627 @include cppopts.texi
5629 @node Assembler Options
5630 @section Passing Options to the Assembler
5632 @c prevent bad page break with this line
5633 You can pass options to the assembler.
5636 @item -Wa,@var{option}
5638 Pass @var{option} as an option to the assembler. If @var{option}
5639 contains commas, it is split into multiple options at the commas.
5641 @item -Xassembler @var{option}
5643 Pass @var{option} as an option to the assembler. You can use this to
5644 supply system-specific assembler options which GCC does not know how to
5647 If you want to pass an option that takes an argument, you must use
5648 @option{-Xassembler} twice, once for the option and once for the argument.
5653 @section Options for Linking
5654 @cindex link options
5655 @cindex options, linking
5657 These options come into play when the compiler links object files into
5658 an executable output file. They are meaningless if the compiler is
5659 not doing a link step.
5663 @item @var{object-file-name}
5664 A file name that does not end in a special recognized suffix is
5665 considered to name an object file or library. (Object files are
5666 distinguished from libraries by the linker according to the file
5667 contents.) If linking is done, these object files are used as input
5676 If any of these options is used, then the linker is not run, and
5677 object file names should not be used as arguments. @xref{Overall
5681 @item -l@var{library}
5682 @itemx -l @var{library}
5684 Search the library named @var{library} when linking. (The second
5685 alternative with the library as a separate argument is only for
5686 POSIX compliance and is not recommended.)
5688 It makes a difference where in the command you write this option; the
5689 linker searches and processes libraries and object files in the order they
5690 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
5691 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
5692 to functions in @samp{z}, those functions may not be loaded.
5694 The linker searches a standard list of directories for the library,
5695 which is actually a file named @file{lib@var{library}.a}. The linker
5696 then uses this file as if it had been specified precisely by name.
5698 The directories searched include several standard system directories
5699 plus any that you specify with @option{-L}.
5701 Normally the files found this way are library files---archive files
5702 whose members are object files. The linker handles an archive file by
5703 scanning through it for members which define symbols that have so far
5704 been referenced but not defined. But if the file that is found is an
5705 ordinary object file, it is linked in the usual fashion. The only
5706 difference between using an @option{-l} option and specifying a file name
5707 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
5708 and searches several directories.
5712 You need this special case of the @option{-l} option in order to
5713 link an Objective-C or Objective-C++ program.
5716 @opindex nostartfiles
5717 Do not use the standard system startup files when linking.
5718 The standard system libraries are used normally, unless @option{-nostdlib}
5719 or @option{-nodefaultlibs} is used.
5721 @item -nodefaultlibs
5722 @opindex nodefaultlibs
5723 Do not use the standard system libraries when linking.
5724 Only the libraries you specify will be passed to the linker.
5725 The standard startup files are used normally, unless @option{-nostartfiles}
5726 is used. The compiler may generate calls to @code{memcmp},
5727 @code{memset}, @code{memcpy} and @code{memmove}.
5728 These entries are usually resolved by entries in
5729 libc. These entry points should be supplied through some other
5730 mechanism when this option is specified.
5734 Do not use the standard system startup files or libraries when linking.
5735 No startup files and only the libraries you specify will be passed to
5736 the linker. The compiler may generate calls to @code{memcmp}, @code{memset},
5737 @code{memcpy} and @code{memmove}.
5738 These entries are usually resolved by entries in
5739 libc. These entry points should be supplied through some other
5740 mechanism when this option is specified.
5742 @cindex @option{-lgcc}, use with @option{-nostdlib}
5743 @cindex @option{-nostdlib} and unresolved references
5744 @cindex unresolved references and @option{-nostdlib}
5745 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
5746 @cindex @option{-nodefaultlibs} and unresolved references
5747 @cindex unresolved references and @option{-nodefaultlibs}
5748 One of the standard libraries bypassed by @option{-nostdlib} and
5749 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
5750 that GCC uses to overcome shortcomings of particular machines, or special
5751 needs for some languages.
5752 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
5753 Collection (GCC) Internals},
5754 for more discussion of @file{libgcc.a}.)
5755 In most cases, you need @file{libgcc.a} even when you want to avoid
5756 other standard libraries. In other words, when you specify @option{-nostdlib}
5757 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
5758 This ensures that you have no unresolved references to internal GCC
5759 library subroutines. (For example, @samp{__main}, used to ensure C++
5760 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
5761 GNU Compiler Collection (GCC) Internals}.)
5765 Produce a position independent executable on targets which support it.
5766 For predictable results, you must also specify the same set of options
5767 that were used to generate code (@option{-fpie}, @option{-fPIE},
5768 or model suboptions) when you specify this option.
5772 Remove all symbol table and relocation information from the executable.
5776 On systems that support dynamic linking, this prevents linking with the shared
5777 libraries. On other systems, this option has no effect.
5781 Produce a shared object which can then be linked with other objects to
5782 form an executable. Not all systems support this option. For predictable
5783 results, you must also specify the same set of options that were used to
5784 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
5785 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
5786 needs to build supplementary stub code for constructors to work. On
5787 multi-libbed systems, @samp{gcc -shared} must select the correct support
5788 libraries to link against. Failing to supply the correct flags may lead
5789 to subtle defects. Supplying them in cases where they are not necessary
5792 @item -shared-libgcc
5793 @itemx -static-libgcc
5794 @opindex shared-libgcc
5795 @opindex static-libgcc
5796 On systems that provide @file{libgcc} as a shared library, these options
5797 force the use of either the shared or static version respectively.
5798 If no shared version of @file{libgcc} was built when the compiler was
5799 configured, these options have no effect.
5801 There are several situations in which an application should use the
5802 shared @file{libgcc} instead of the static version. The most common
5803 of these is when the application wishes to throw and catch exceptions
5804 across different shared libraries. In that case, each of the libraries
5805 as well as the application itself should use the shared @file{libgcc}.
5807 Therefore, the G++ and GCJ drivers automatically add
5808 @option{-shared-libgcc} whenever you build a shared library or a main
5809 executable, because C++ and Java programs typically use exceptions, so
5810 this is the right thing to do.
5812 If, instead, you use the GCC driver to create shared libraries, you may
5813 find that they will not always be linked with the shared @file{libgcc}.
5814 If GCC finds, at its configuration time, that you have a non-GNU linker
5815 or a GNU linker that does not support option @option{--eh-frame-hdr},
5816 it will link the shared version of @file{libgcc} into shared libraries
5817 by default. Otherwise, it will take advantage of the linker and optimize
5818 away the linking with the shared version of @file{libgcc}, linking with
5819 the static version of libgcc by default. This allows exceptions to
5820 propagate through such shared libraries, without incurring relocation
5821 costs at library load time.
5823 However, if a library or main executable is supposed to throw or catch
5824 exceptions, you must link it using the G++ or GCJ driver, as appropriate
5825 for the languages used in the program, or using the option
5826 @option{-shared-libgcc}, such that it is linked with the shared
5831 Bind references to global symbols when building a shared object. Warn
5832 about any unresolved references (unless overridden by the link editor
5833 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
5836 @item -Xlinker @var{option}
5838 Pass @var{option} as an option to the linker. You can use this to
5839 supply system-specific linker options which GCC does not know how to
5842 If you want to pass an option that takes an argument, you must use
5843 @option{-Xlinker} twice, once for the option and once for the argument.
5844 For example, to pass @option{-assert definitions}, you must write
5845 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
5846 @option{-Xlinker "-assert definitions"}, because this passes the entire
5847 string as a single argument, which is not what the linker expects.
5849 @item -Wl,@var{option}
5851 Pass @var{option} as an option to the linker. If @var{option} contains
5852 commas, it is split into multiple options at the commas.
5854 @item -u @var{symbol}
5856 Pretend the symbol @var{symbol} is undefined, to force linking of
5857 library modules to define it. You can use @option{-u} multiple times with
5858 different symbols to force loading of additional library modules.
5861 @node Directory Options
5862 @section Options for Directory Search
5863 @cindex directory options
5864 @cindex options, directory search
5867 These options specify directories to search for header files, for
5868 libraries and for parts of the compiler:
5873 Add the directory @var{dir} to the head of the list of directories to be
5874 searched for header files. This can be used to override a system header
5875 file, substituting your own version, since these directories are
5876 searched before the system header file directories. However, you should
5877 not use this option to add directories that contain vendor-supplied
5878 system header files (use @option{-isystem} for that). If you use more than
5879 one @option{-I} option, the directories are scanned in left-to-right
5880 order; the standard system directories come after.
5882 If a standard system include directory, or a directory specified with
5883 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
5884 option will be ignored. The directory will still be searched but as a
5885 system directory at its normal position in the system include chain.
5886 This is to ensure that GCC's procedure to fix buggy system headers and
5887 the ordering for the include_next directive are not inadvertently changed.
5888 If you really need to change the search order for system directories,
5889 use the @option{-nostdinc} and/or @option{-isystem} options.
5891 @item -iquote@var{dir}
5893 Add the directory @var{dir} to the head of the list of directories to
5894 be searched for header files only for the case of @samp{#include
5895 "@var{file}"}; they are not searched for @samp{#include <@var{file}>},
5896 otherwise just like @option{-I}.
5900 Add directory @var{dir} to the list of directories to be searched
5903 @item -B@var{prefix}
5905 This option specifies where to find the executables, libraries,
5906 include files, and data files of the compiler itself.
5908 The compiler driver program runs one or more of the subprograms
5909 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
5910 @var{prefix} as a prefix for each program it tries to run, both with and
5911 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
5913 For each subprogram to be run, the compiler driver first tries the
5914 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
5915 was not specified, the driver tries two standard prefixes, which are
5916 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
5917 those results in a file name that is found, the unmodified program
5918 name is searched for using the directories specified in your
5919 @env{PATH} environment variable.
5921 The compiler will check to see if the path provided by the @option{-B}
5922 refers to a directory, and if necessary it will add a directory
5923 separator character at the end of the path.
5925 @option{-B} prefixes that effectively specify directory names also apply
5926 to libraries in the linker, because the compiler translates these
5927 options into @option{-L} options for the linker. They also apply to
5928 includes files in the preprocessor, because the compiler translates these
5929 options into @option{-isystem} options for the preprocessor. In this case,
5930 the compiler appends @samp{include} to the prefix.
5932 The run-time support file @file{libgcc.a} can also be searched for using
5933 the @option{-B} prefix, if needed. If it is not found there, the two
5934 standard prefixes above are tried, and that is all. The file is left
5935 out of the link if it is not found by those means.
5937 Another way to specify a prefix much like the @option{-B} prefix is to use
5938 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
5941 As a special kludge, if the path provided by @option{-B} is
5942 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
5943 9, then it will be replaced by @file{[dir/]include}. This is to help
5944 with boot-strapping the compiler.
5946 @item -specs=@var{file}
5948 Process @var{file} after the compiler reads in the standard @file{specs}
5949 file, in order to override the defaults that the @file{gcc} driver
5950 program uses when determining what switches to pass to @file{cc1},
5951 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
5952 @option{-specs=@var{file}} can be specified on the command line, and they
5953 are processed in order, from left to right.
5957 This option has been deprecated. Please use @option{-iquote} instead for
5958 @option{-I} directories before the @option{-I-} and remove the @option{-I-}.
5959 Any directories you specify with @option{-I} options before the @option{-I-}
5960 option are searched only for the case of @samp{#include "@var{file}"};
5961 they are not searched for @samp{#include <@var{file}>}.
5963 If additional directories are specified with @option{-I} options after
5964 the @option{-I-}, these directories are searched for all @samp{#include}
5965 directives. (Ordinarily @emph{all} @option{-I} directories are used
5968 In addition, the @option{-I-} option inhibits the use of the current
5969 directory (where the current input file came from) as the first search
5970 directory for @samp{#include "@var{file}"}. There is no way to
5971 override this effect of @option{-I-}. With @option{-I.} you can specify
5972 searching the directory which was current when the compiler was
5973 invoked. That is not exactly the same as what the preprocessor does
5974 by default, but it is often satisfactory.
5976 @option{-I-} does not inhibit the use of the standard system directories
5977 for header files. Thus, @option{-I-} and @option{-nostdinc} are
5984 @section Specifying subprocesses and the switches to pass to them
5987 @command{gcc} is a driver program. It performs its job by invoking a
5988 sequence of other programs to do the work of compiling, assembling and
5989 linking. GCC interprets its command-line parameters and uses these to
5990 deduce which programs it should invoke, and which command-line options
5991 it ought to place on their command lines. This behavior is controlled
5992 by @dfn{spec strings}. In most cases there is one spec string for each
5993 program that GCC can invoke, but a few programs have multiple spec
5994 strings to control their behavior. The spec strings built into GCC can
5995 be overridden by using the @option{-specs=} command-line switch to specify
5998 @dfn{Spec files} are plaintext files that are used to construct spec
5999 strings. They consist of a sequence of directives separated by blank
6000 lines. The type of directive is determined by the first non-whitespace
6001 character on the line and it can be one of the following:
6004 @item %@var{command}
6005 Issues a @var{command} to the spec file processor. The commands that can
6009 @item %include <@var{file}>
6011 Search for @var{file} and insert its text at the current point in the
6014 @item %include_noerr <@var{file}>
6015 @cindex %include_noerr
6016 Just like @samp{%include}, but do not generate an error message if the include
6017 file cannot be found.
6019 @item %rename @var{old_name} @var{new_name}
6021 Rename the spec string @var{old_name} to @var{new_name}.
6025 @item *[@var{spec_name}]:
6026 This tells the compiler to create, override or delete the named spec
6027 string. All lines after this directive up to the next directive or
6028 blank line are considered to be the text for the spec string. If this
6029 results in an empty string then the spec will be deleted. (Or, if the
6030 spec did not exist, then nothing will happened.) Otherwise, if the spec
6031 does not currently exist a new spec will be created. If the spec does
6032 exist then its contents will be overridden by the text of this
6033 directive, unless the first character of that text is the @samp{+}
6034 character, in which case the text will be appended to the spec.
6036 @item [@var{suffix}]:
6037 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
6038 and up to the next directive or blank line are considered to make up the
6039 spec string for the indicated suffix. When the compiler encounters an
6040 input file with the named suffix, it will processes the spec string in
6041 order to work out how to compile that file. For example:
6048 This says that any input file whose name ends in @samp{.ZZ} should be
6049 passed to the program @samp{z-compile}, which should be invoked with the
6050 command-line switch @option{-input} and with the result of performing the
6051 @samp{%i} substitution. (See below.)
6053 As an alternative to providing a spec string, the text that follows a
6054 suffix directive can be one of the following:
6057 @item @@@var{language}
6058 This says that the suffix is an alias for a known @var{language}. This is
6059 similar to using the @option{-x} command-line switch to GCC to specify a
6060 language explicitly. For example:
6067 Says that .ZZ files are, in fact, C++ source files.
6070 This causes an error messages saying:
6073 @var{name} compiler not installed on this system.
6077 GCC already has an extensive list of suffixes built into it.
6078 This directive will add an entry to the end of the list of suffixes, but
6079 since the list is searched from the end backwards, it is effectively
6080 possible to override earlier entries using this technique.
6084 GCC has the following spec strings built into it. Spec files can
6085 override these strings or create their own. Note that individual
6086 targets can also add their own spec strings to this list.
6089 asm Options to pass to the assembler
6090 asm_final Options to pass to the assembler post-processor
6091 cpp Options to pass to the C preprocessor
6092 cc1 Options to pass to the C compiler
6093 cc1plus Options to pass to the C++ compiler
6094 endfile Object files to include at the end of the link
6095 link Options to pass to the linker
6096 lib Libraries to include on the command line to the linker
6097 libgcc Decides which GCC support library to pass to the linker
6098 linker Sets the name of the linker
6099 predefines Defines to be passed to the C preprocessor
6100 signed_char Defines to pass to CPP to say whether @code{char} is signed
6102 startfile Object files to include at the start of the link
6105 Here is a small example of a spec file:
6111 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
6114 This example renames the spec called @samp{lib} to @samp{old_lib} and
6115 then overrides the previous definition of @samp{lib} with a new one.
6116 The new definition adds in some extra command-line options before
6117 including the text of the old definition.
6119 @dfn{Spec strings} are a list of command-line options to be passed to their
6120 corresponding program. In addition, the spec strings can contain
6121 @samp{%}-prefixed sequences to substitute variable text or to
6122 conditionally insert text into the command line. Using these constructs
6123 it is possible to generate quite complex command lines.
6125 Here is a table of all defined @samp{%}-sequences for spec
6126 strings. Note that spaces are not generated automatically around the
6127 results of expanding these sequences. Therefore you can concatenate them
6128 together or combine them with constant text in a single argument.
6132 Substitute one @samp{%} into the program name or argument.
6135 Substitute the name of the input file being processed.
6138 Substitute the basename of the input file being processed.
6139 This is the substring up to (and not including) the last period
6140 and not including the directory.
6143 This is the same as @samp{%b}, but include the file suffix (text after
6147 Marks the argument containing or following the @samp{%d} as a
6148 temporary file name, so that that file will be deleted if GCC exits
6149 successfully. Unlike @samp{%g}, this contributes no text to the
6152 @item %g@var{suffix}
6153 Substitute a file name that has suffix @var{suffix} and is chosen
6154 once per compilation, and mark the argument in the same way as
6155 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
6156 name is now chosen in a way that is hard to predict even when previously
6157 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
6158 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
6159 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
6160 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
6161 was simply substituted with a file name chosen once per compilation,
6162 without regard to any appended suffix (which was therefore treated
6163 just like ordinary text), making such attacks more likely to succeed.
6165 @item %u@var{suffix}
6166 Like @samp{%g}, but generates a new temporary file name even if
6167 @samp{%u@var{suffix}} was already seen.
6169 @item %U@var{suffix}
6170 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
6171 new one if there is no such last file name. In the absence of any
6172 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
6173 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
6174 would involve the generation of two distinct file names, one
6175 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
6176 simply substituted with a file name chosen for the previous @samp{%u},
6177 without regard to any appended suffix.
6179 @item %j@var{suffix}
6180 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
6181 writable, and if save-temps is off; otherwise, substitute the name
6182 of a temporary file, just like @samp{%u}. This temporary file is not
6183 meant for communication between processes, but rather as a junk
6186 @item %|@var{suffix}
6187 @itemx %m@var{suffix}
6188 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
6189 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
6190 all. These are the two most common ways to instruct a program that it
6191 should read from standard input or write to standard output. If you
6192 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
6193 construct: see for example @file{f/lang-specs.h}.
6195 @item %.@var{SUFFIX}
6196 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
6197 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
6198 terminated by the next space or %.
6201 Marks the argument containing or following the @samp{%w} as the
6202 designated output file of this compilation. This puts the argument
6203 into the sequence of arguments that @samp{%o} will substitute later.
6206 Substitutes the names of all the output files, with spaces
6207 automatically placed around them. You should write spaces
6208 around the @samp{%o} as well or the results are undefined.
6209 @samp{%o} is for use in the specs for running the linker.
6210 Input files whose names have no recognized suffix are not compiled
6211 at all, but they are included among the output files, so they will
6215 Substitutes the suffix for object files. Note that this is
6216 handled specially when it immediately follows @samp{%g, %u, or %U},
6217 because of the need for those to form complete file names. The
6218 handling is such that @samp{%O} is treated exactly as if it had already
6219 been substituted, except that @samp{%g, %u, and %U} do not currently
6220 support additional @var{suffix} characters following @samp{%O} as they would
6221 following, for example, @samp{.o}.
6224 Substitutes the standard macro predefinitions for the
6225 current target machine. Use this when running @code{cpp}.
6228 Like @samp{%p}, but puts @samp{__} before and after the name of each
6229 predefined macro, except for macros that start with @samp{__} or with
6230 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
6234 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
6235 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}), and
6236 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
6240 Current argument is the name of a library or startup file of some sort.
6241 Search for that file in a standard list of directories and substitute
6242 the full name found.
6245 Print @var{str} as an error message. @var{str} is terminated by a newline.
6246 Use this when inconsistent options are detected.
6249 Substitute the contents of spec string @var{name} at this point.
6252 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
6254 @item %x@{@var{option}@}
6255 Accumulate an option for @samp{%X}.
6258 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
6262 Output the accumulated assembler options specified by @option{-Wa}.
6265 Output the accumulated preprocessor options specified by @option{-Wp}.
6268 Process the @code{asm} spec. This is used to compute the
6269 switches to be passed to the assembler.
6272 Process the @code{asm_final} spec. This is a spec string for
6273 passing switches to an assembler post-processor, if such a program is
6277 Process the @code{link} spec. This is the spec for computing the
6278 command line passed to the linker. Typically it will make use of the
6279 @samp{%L %G %S %D and %E} sequences.
6282 Dump out a @option{-L} option for each directory that GCC believes might
6283 contain startup files. If the target supports multilibs then the
6284 current multilib directory will be prepended to each of these paths.
6287 Output the multilib directory with directory separators replaced with
6288 @samp{_}. If multilib directories are not set, or the multilib directory is
6289 @file{.} then this option emits nothing.
6292 Process the @code{lib} spec. This is a spec string for deciding which
6293 libraries should be included on the command line to the linker.
6296 Process the @code{libgcc} spec. This is a spec string for deciding
6297 which GCC support library should be included on the command line to the linker.
6300 Process the @code{startfile} spec. This is a spec for deciding which
6301 object files should be the first ones passed to the linker. Typically
6302 this might be a file named @file{crt0.o}.
6305 Process the @code{endfile} spec. This is a spec string that specifies
6306 the last object files that will be passed to the linker.
6309 Process the @code{cpp} spec. This is used to construct the arguments
6310 to be passed to the C preprocessor.
6313 Process the @code{cc1} spec. This is used to construct the options to be
6314 passed to the actual C compiler (@samp{cc1}).
6317 Process the @code{cc1plus} spec. This is used to construct the options to be
6318 passed to the actual C++ compiler (@samp{cc1plus}).
6321 Substitute the variable part of a matched option. See below.
6322 Note that each comma in the substituted string is replaced by
6326 Remove all occurrences of @code{-S} from the command line. Note---this
6327 command is position dependent. @samp{%} commands in the spec string
6328 before this one will see @code{-S}, @samp{%} commands in the spec string
6329 after this one will not.
6331 @item %:@var{function}(@var{args})
6332 Call the named function @var{function}, passing it @var{args}.
6333 @var{args} is first processed as a nested spec string, then split
6334 into an argument vector in the usual fashion. The function returns
6335 a string which is processed as if it had appeared literally as part
6336 of the current spec.
6338 The following built-in spec functions are provided:
6341 @item @code{if-exists}
6342 The @code{if-exists} spec function takes one argument, an absolute
6343 pathname to a file. If the file exists, @code{if-exists} returns the
6344 pathname. Here is a small example of its usage:
6348 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
6351 @item @code{if-exists-else}
6352 The @code{if-exists-else} spec function is similar to the @code{if-exists}
6353 spec function, except that it takes two arguments. The first argument is
6354 an absolute pathname to a file. If the file exists, @code{if-exists-else}
6355 returns the pathname. If it does not exist, it returns the second argument.
6356 This way, @code{if-exists-else} can be used to select one file or another,
6357 based on the existence of the first. Here is a small example of its usage:
6361 crt0%O%s %:if-exists(crti%O%s) \
6362 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
6365 @item @code{replace-outfile}
6366 The @code{replace-outfile} spec function takes two arguments. It looks for the
6367 first argument in the outfiles array and replaces it with the second argument. Here
6368 is a small example of its usage:
6371 %@{fgnu-runtime:%:replace-outfile(-lobjc -lobjc-gnu)@}
6377 Substitutes the @code{-S} switch, if that switch was given to GCC@.
6378 If that switch was not specified, this substitutes nothing. Note that
6379 the leading dash is omitted when specifying this option, and it is
6380 automatically inserted if the substitution is performed. Thus the spec
6381 string @samp{%@{foo@}} would match the command-line option @option{-foo}
6382 and would output the command line option @option{-foo}.
6384 @item %W@{@code{S}@}
6385 Like %@{@code{S}@} but mark last argument supplied within as a file to be
6388 @item %@{@code{S}*@}
6389 Substitutes all the switches specified to GCC whose names start
6390 with @code{-S}, but which also take an argument. This is used for
6391 switches like @option{-o}, @option{-D}, @option{-I}, etc.
6392 GCC considers @option{-o foo} as being
6393 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
6394 text, including the space. Thus two arguments would be generated.
6396 @item %@{@code{S}*&@code{T}*@}
6397 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
6398 (the order of @code{S} and @code{T} in the spec is not significant).
6399 There can be any number of ampersand-separated variables; for each the
6400 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
6402 @item %@{@code{S}:@code{X}@}
6403 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
6405 @item %@{!@code{S}:@code{X}@}
6406 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
6408 @item %@{@code{S}*:@code{X}@}
6409 Substitutes @code{X} if one or more switches whose names start with
6410 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
6411 once, no matter how many such switches appeared. However, if @code{%*}
6412 appears somewhere in @code{X}, then @code{X} will be substituted once
6413 for each matching switch, with the @code{%*} replaced by the part of
6414 that switch that matched the @code{*}.
6416 @item %@{.@code{S}:@code{X}@}
6417 Substitutes @code{X}, if processing a file with suffix @code{S}.
6419 @item %@{!.@code{S}:@code{X}@}
6420 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
6422 @item %@{@code{S}|@code{P}:@code{X}@}
6423 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC@.
6424 This may be combined with @samp{!}, @samp{.}, and @code{*} sequences as well,
6425 although they have a stronger binding than the @samp{|}. If @code{%*}
6426 appears in @code{X}, all of the alternatives must be starred, and only
6427 the first matching alternative is substituted.
6429 For example, a spec string like this:
6432 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
6435 will output the following command-line options from the following input
6436 command-line options:
6441 -d fred.c -foo -baz -boggle
6442 -d jim.d -bar -baz -boggle
6445 @item %@{S:X; T:Y; :D@}
6447 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
6448 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
6449 be as many clauses as you need. This may be combined with @code{.},
6450 @code{!}, @code{|}, and @code{*} as needed.
6455 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
6456 construct may contain other nested @samp{%} constructs or spaces, or
6457 even newlines. They are processed as usual, as described above.
6458 Trailing white space in @code{X} is ignored. White space may also
6459 appear anywhere on the left side of the colon in these constructs,
6460 except between @code{.} or @code{*} and the corresponding word.
6462 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
6463 handled specifically in these constructs. If another value of
6464 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
6465 @option{-W} switch is found later in the command line, the earlier
6466 switch value is ignored, except with @{@code{S}*@} where @code{S} is
6467 just one letter, which passes all matching options.
6469 The character @samp{|} at the beginning of the predicate text is used to
6470 indicate that a command should be piped to the following command, but
6471 only if @option{-pipe} is specified.
6473 It is built into GCC which switches take arguments and which do not.
6474 (You might think it would be useful to generalize this to allow each
6475 compiler's spec to say which switches take arguments. But this cannot
6476 be done in a consistent fashion. GCC cannot even decide which input
6477 files have been specified without knowing which switches take arguments,
6478 and it must know which input files to compile in order to tell which
6481 GCC also knows implicitly that arguments starting in @option{-l} are to be
6482 treated as compiler output files, and passed to the linker in their
6483 proper position among the other output files.
6485 @c man begin OPTIONS
6487 @node Target Options
6488 @section Specifying Target Machine and Compiler Version
6489 @cindex target options
6490 @cindex cross compiling
6491 @cindex specifying machine version
6492 @cindex specifying compiler version and target machine
6493 @cindex compiler version, specifying
6494 @cindex target machine, specifying
6496 The usual way to run GCC is to run the executable called @file{gcc}, or
6497 @file{<machine>-gcc} when cross-compiling, or
6498 @file{<machine>-gcc-<version>} to run a version other than the one that
6499 was installed last. Sometimes this is inconvenient, so GCC provides
6500 options that will switch to another cross-compiler or version.
6503 @item -b @var{machine}
6505 The argument @var{machine} specifies the target machine for compilation.
6507 The value to use for @var{machine} is the same as was specified as the
6508 machine type when configuring GCC as a cross-compiler. For
6509 example, if a cross-compiler was configured with @samp{configure
6510 i386v}, meaning to compile for an 80386 running System V, then you
6511 would specify @option{-b i386v} to run that cross compiler.
6513 @item -V @var{version}
6515 The argument @var{version} specifies which version of GCC to run.
6516 This is useful when multiple versions are installed. For example,
6517 @var{version} might be @samp{2.0}, meaning to run GCC version 2.0.
6520 The @option{-V} and @option{-b} options work by running the
6521 @file{<machine>-gcc-<version>} executable, so there's no real reason to
6522 use them if you can just run that directly.
6524 @node Submodel Options
6525 @section Hardware Models and Configurations
6526 @cindex submodel options
6527 @cindex specifying hardware config
6528 @cindex hardware models and configurations, specifying
6529 @cindex machine dependent options
6531 Earlier we discussed the standard option @option{-b} which chooses among
6532 different installed compilers for completely different target
6533 machines, such as VAX vs.@: 68000 vs.@: 80386.
6535 In addition, each of these target machine types can have its own
6536 special options, starting with @samp{-m}, to choose among various
6537 hardware models or configurations---for example, 68010 vs 68020,
6538 floating coprocessor or none. A single installed version of the
6539 compiler can compile for any model or configuration, according to the
6542 Some configurations of the compiler also support additional special
6543 options, usually for compatibility with other compilers on the same
6546 These options are defined by the macro @code{TARGET_SWITCHES} in the
6547 machine description. The default for the options is also defined by
6548 that macro, which enables you to change the defaults.
6550 @c This list is ordered alphanumerically by subsection name.
6551 @c It should be the same order and spelling as these options are listed
6552 @c in Machine Dependent Options
6560 * DEC Alpha Options::
6561 * DEC Alpha/VMS Options::
6565 * i386 and x86-64 Options::
6577 * RS/6000 and PowerPC Options::
6578 * S/390 and zSeries Options::
6581 * System V Options::
6582 * TMS320C3x/C4x Options::
6586 * Xstormy16 Options::
6592 @subsection ARC Options
6595 These options are defined for ARC implementations:
6600 Compile code for little endian mode. This is the default.
6604 Compile code for big endian mode.
6607 @opindex mmangle-cpu
6608 Prepend the name of the cpu to all public symbol names.
6609 In multiple-processor systems, there are many ARC variants with different
6610 instruction and register set characteristics. This flag prevents code
6611 compiled for one cpu to be linked with code compiled for another.
6612 No facility exists for handling variants that are ``almost identical''.
6613 This is an all or nothing option.
6615 @item -mcpu=@var{cpu}
6617 Compile code for ARC variant @var{cpu}.
6618 Which variants are supported depend on the configuration.
6619 All variants support @option{-mcpu=base}, this is the default.
6621 @item -mtext=@var{text-section}
6622 @itemx -mdata=@var{data-section}
6623 @itemx -mrodata=@var{readonly-data-section}
6627 Put functions, data, and readonly data in @var{text-section},
6628 @var{data-section}, and @var{readonly-data-section} respectively
6629 by default. This can be overridden with the @code{section} attribute.
6630 @xref{Variable Attributes}.
6635 @subsection ARM Options
6638 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
6642 @item -mabi=@var{name}
6644 Generate code for the specified ABI. Permissible values are: @samp{apcs-gnu},
6645 @samp{atpcs}, @samp{aapcs} and @samp{iwmmxt}.
6648 @opindex mapcs-frame
6649 Generate a stack frame that is compliant with the ARM Procedure Call
6650 Standard for all functions, even if this is not strictly necessary for
6651 correct execution of the code. Specifying @option{-fomit-frame-pointer}
6652 with this option will cause the stack frames not to be generated for
6653 leaf functions. The default is @option{-mno-apcs-frame}.
6657 This is a synonym for @option{-mapcs-frame}.
6660 @c not currently implemented
6661 @item -mapcs-stack-check
6662 @opindex mapcs-stack-check
6663 Generate code to check the amount of stack space available upon entry to
6664 every function (that actually uses some stack space). If there is
6665 insufficient space available then either the function
6666 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
6667 called, depending upon the amount of stack space required. The run time
6668 system is required to provide these functions. The default is
6669 @option{-mno-apcs-stack-check}, since this produces smaller code.
6671 @c not currently implemented
6673 @opindex mapcs-float
6674 Pass floating point arguments using the float point registers. This is
6675 one of the variants of the APCS@. This option is recommended if the
6676 target hardware has a floating point unit or if a lot of floating point
6677 arithmetic is going to be performed by the code. The default is
6678 @option{-mno-apcs-float}, since integer only code is slightly increased in
6679 size if @option{-mapcs-float} is used.
6681 @c not currently implemented
6682 @item -mapcs-reentrant
6683 @opindex mapcs-reentrant
6684 Generate reentrant, position independent code. The default is
6685 @option{-mno-apcs-reentrant}.
6688 @item -mthumb-interwork
6689 @opindex mthumb-interwork
6690 Generate code which supports calling between the ARM and Thumb
6691 instruction sets. Without this option the two instruction sets cannot
6692 be reliably used inside one program. The default is
6693 @option{-mno-thumb-interwork}, since slightly larger code is generated
6694 when @option{-mthumb-interwork} is specified.
6696 @item -mno-sched-prolog
6697 @opindex mno-sched-prolog
6698 Prevent the reordering of instructions in the function prolog, or the
6699 merging of those instruction with the instructions in the function's
6700 body. This means that all functions will start with a recognizable set
6701 of instructions (or in fact one of a choice from a small set of
6702 different function prologues), and this information can be used to
6703 locate the start if functions inside an executable piece of code. The
6704 default is @option{-msched-prolog}.
6707 @opindex mhard-float
6708 Generate output containing floating point instructions. This is the
6712 @opindex msoft-float
6713 Generate output containing library calls for floating point.
6714 @strong{Warning:} the requisite libraries are not available for all ARM
6715 targets. Normally the facilities of the machine's usual C compiler are
6716 used, but this cannot be done directly in cross-compilation. You must make
6717 your own arrangements to provide suitable library functions for
6720 @option{-msoft-float} changes the calling convention in the output file;
6721 therefore, it is only useful if you compile @emph{all} of a program with
6722 this option. In particular, you need to compile @file{libgcc.a}, the
6723 library that comes with GCC, with @option{-msoft-float} in order for
6726 @item -mfloat-abi=@var{name}
6728 Specifies which ABI to use for floating point values. Permissible values
6729 are: @samp{soft}, @samp{softfp} and @samp{hard}.
6731 @samp{soft} and @samp{hard} are equivalent to @option{-msoft-float}
6732 and @option{-mhard-float} respectively. @samp{softfp} allows the generation
6733 of floating point instructions, but still uses the soft-float calling
6736 @item -mlittle-endian
6737 @opindex mlittle-endian
6738 Generate code for a processor running in little-endian mode. This is
6739 the default for all standard configurations.
6742 @opindex mbig-endian
6743 Generate code for a processor running in big-endian mode; the default is
6744 to compile code for a little-endian processor.
6746 @item -mwords-little-endian
6747 @opindex mwords-little-endian
6748 This option only applies when generating code for big-endian processors.
6749 Generate code for a little-endian word order but a big-endian byte
6750 order. That is, a byte order of the form @samp{32107654}. Note: this
6751 option should only be used if you require compatibility with code for
6752 big-endian ARM processors generated by versions of the compiler prior to
6755 @item -mcpu=@var{name}
6757 This specifies the name of the target ARM processor. GCC uses this name
6758 to determine what kind of instructions it can emit when generating
6759 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
6760 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
6761 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
6762 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
6763 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
6764 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm7tdmi-s},
6765 @samp{arm8}, @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
6766 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
6767 @samp{arm920t}, @samp{arm922t}, @samp{arm946e-s}, @samp{arm966e-s},
6768 @samp{arm968e-s}, @samp{arm926ej-s}, @samp{arm940t}, @samp{arm9tdmi},
6769 @samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ej-s},
6770 @samp{arm10e}, @samp{arm1020e}, @samp{arm1022e},
6771 @samp{arm1136j-s}, @samp{arm1136jf-s} ,@samp{xscale}, @samp{iwmmxt},
6774 @itemx -mtune=@var{name}
6776 This option is very similar to the @option{-mcpu=} option, except that
6777 instead of specifying the actual target processor type, and hence
6778 restricting which instructions can be used, it specifies that GCC should
6779 tune the performance of the code as if the target were of the type
6780 specified in this option, but still choosing the instructions that it
6781 will generate based on the cpu specified by a @option{-mcpu=} option.
6782 For some ARM implementations better performance can be obtained by using
6785 @item -march=@var{name}
6787 This specifies the name of the target ARM architecture. GCC uses this
6788 name to determine what kind of instructions it can emit when generating
6789 assembly code. This option can be used in conjunction with or instead
6790 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
6791 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
6792 @samp{armv5}, @samp{armv5t}, @samp{armv5te}, @samp{armv6}, @samp{armv6j},
6793 @samp{iwmmxt}, @samp{ep9312}.
6795 @item -mfpu=@var{name}
6796 @itemx -mfpe=@var{number}
6797 @itemx -mfp=@var{number}
6801 This specifies what floating point hardware (or hardware emulation) is
6802 available on the target. Permissible names are: @samp{fpa}, @samp{fpe2},
6803 @samp{fpe3}, @samp{maverick}, @samp{vfp}. @option{-mfp} and @option{-mfpe}
6804 are synonyms for @option{-mfpu}=@samp{fpe}@var{number}, for compatibility
6805 with older versions of GCC@.
6807 If @option{-msoft-float} is specified this specifies the format of
6808 floating point values.
6810 @item -mstructure-size-boundary=@var{n}
6811 @opindex mstructure-size-boundary
6812 The size of all structures and unions will be rounded up to a multiple
6813 of the number of bits set by this option. Permissible values are 8, 32
6814 and 64. The default value varies for different toolchains. For the COFF
6815 targeted toolchain the default value is 8. A value of 64 is only allowed
6816 if the underlying ABI supports it.
6818 Specifying the larger number can produce faster, more efficient code, but
6819 can also increase the size of the program. Different values are potentially
6820 incompatible. Code compiled with one value cannot necessarily expect to
6821 work with code or libraries compiled with another value, if they exchange
6822 information using structures or unions.
6824 @item -mabort-on-noreturn
6825 @opindex mabort-on-noreturn
6826 Generate a call to the function @code{abort} at the end of a
6827 @code{noreturn} function. It will be executed if the function tries to
6831 @itemx -mno-long-calls
6832 @opindex mlong-calls
6833 @opindex mno-long-calls
6834 Tells the compiler to perform function calls by first loading the
6835 address of the function into a register and then performing a subroutine
6836 call on this register. This switch is needed if the target function
6837 will lie outside of the 64 megabyte addressing range of the offset based
6838 version of subroutine call instruction.
6840 Even if this switch is enabled, not all function calls will be turned
6841 into long calls. The heuristic is that static functions, functions
6842 which have the @samp{short-call} attribute, functions that are inside
6843 the scope of a @samp{#pragma no_long_calls} directive and functions whose
6844 definitions have already been compiled within the current compilation
6845 unit, will not be turned into long calls. The exception to this rule is
6846 that weak function definitions, functions with the @samp{long-call}
6847 attribute or the @samp{section} attribute, and functions that are within
6848 the scope of a @samp{#pragma long_calls} directive, will always be
6849 turned into long calls.
6851 This feature is not enabled by default. Specifying
6852 @option{-mno-long-calls} will restore the default behavior, as will
6853 placing the function calls within the scope of a @samp{#pragma
6854 long_calls_off} directive. Note these switches have no effect on how
6855 the compiler generates code to handle function calls via function
6858 @item -mnop-fun-dllimport
6859 @opindex mnop-fun-dllimport
6860 Disable support for the @code{dllimport} attribute.
6862 @item -msingle-pic-base
6863 @opindex msingle-pic-base
6864 Treat the register used for PIC addressing as read-only, rather than
6865 loading it in the prologue for each function. The run-time system is
6866 responsible for initializing this register with an appropriate value
6867 before execution begins.
6869 @item -mpic-register=@var{reg}
6870 @opindex mpic-register
6871 Specify the register to be used for PIC addressing. The default is R10
6872 unless stack-checking is enabled, when R9 is used.
6874 @item -mcirrus-fix-invalid-insns
6875 @opindex mcirrus-fix-invalid-insns
6876 @opindex mno-cirrus-fix-invalid-insns
6877 Insert NOPs into the instruction stream to in order to work around
6878 problems with invalid Maverick instruction combinations. This option
6879 is only valid if the @option{-mcpu=ep9312} option has been used to
6880 enable generation of instructions for the Cirrus Maverick floating
6881 point co-processor. This option is not enabled by default, since the
6882 problem is only present in older Maverick implementations. The default
6883 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
6886 @item -mpoke-function-name
6887 @opindex mpoke-function-name
6888 Write the name of each function into the text section, directly
6889 preceding the function prologue. The generated code is similar to this:
6893 .ascii "arm_poke_function_name", 0
6896 .word 0xff000000 + (t1 - t0)
6897 arm_poke_function_name
6899 stmfd sp!, @{fp, ip, lr, pc@}
6903 When performing a stack backtrace, code can inspect the value of
6904 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
6905 location @code{pc - 12} and the top 8 bits are set, then we know that
6906 there is a function name embedded immediately preceding this location
6907 and has length @code{((pc[-3]) & 0xff000000)}.
6911 Generate code for the 16-bit Thumb instruction set. The default is to
6912 use the 32-bit ARM instruction set.
6915 @opindex mtpcs-frame
6916 Generate a stack frame that is compliant with the Thumb Procedure Call
6917 Standard for all non-leaf functions. (A leaf function is one that does
6918 not call any other functions.) The default is @option{-mno-tpcs-frame}.
6920 @item -mtpcs-leaf-frame
6921 @opindex mtpcs-leaf-frame
6922 Generate a stack frame that is compliant with the Thumb Procedure Call
6923 Standard for all leaf functions. (A leaf function is one that does
6924 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
6926 @item -mcallee-super-interworking
6927 @opindex mcallee-super-interworking
6928 Gives all externally visible functions in the file being compiled an ARM
6929 instruction set header which switches to Thumb mode before executing the
6930 rest of the function. This allows these functions to be called from
6931 non-interworking code.
6933 @item -mcaller-super-interworking
6934 @opindex mcaller-super-interworking
6935 Allows calls via function pointers (including virtual functions) to
6936 execute correctly regardless of whether the target code has been
6937 compiled for interworking or not. There is a small overhead in the cost
6938 of executing a function pointer if this option is enabled.
6943 @subsection AVR Options
6946 These options are defined for AVR implementations:
6949 @item -mmcu=@var{mcu}
6951 Specify ATMEL AVR instruction set or MCU type.
6953 Instruction set avr1 is for the minimal AVR core, not supported by the C
6954 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
6955 attiny11, attiny12, attiny15, attiny28).
6957 Instruction set avr2 (default) is for the classic AVR core with up to
6958 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
6959 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
6960 at90c8534, at90s8535).
6962 Instruction set avr3 is for the classic AVR core with up to 128K program
6963 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
6965 Instruction set avr4 is for the enhanced AVR core with up to 8K program
6966 memory space (MCU types: atmega8, atmega83, atmega85).
6968 Instruction set avr5 is for the enhanced AVR core with up to 128K program
6969 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
6970 atmega64, atmega128, at43usb355, at94k).
6974 Output instruction sizes to the asm file.
6976 @item -minit-stack=@var{N}
6977 @opindex minit-stack
6978 Specify the initial stack address, which may be a symbol or numeric value,
6979 @samp{__stack} is the default.
6981 @item -mno-interrupts
6982 @opindex mno-interrupts
6983 Generated code is not compatible with hardware interrupts.
6984 Code size will be smaller.
6986 @item -mcall-prologues
6987 @opindex mcall-prologues
6988 Functions prologues/epilogues expanded as call to appropriate
6989 subroutines. Code size will be smaller.
6991 @item -mno-tablejump
6992 @opindex mno-tablejump
6993 Do not generate tablejump insns which sometimes increase code size.
6996 @opindex mtiny-stack
6997 Change only the low 8 bits of the stack pointer.
7001 Assume int to be 8 bit integer. This affects the sizes of all types: A
7002 char will be 1 byte, an int will be 1 byte, an long will be 2 bytes
7003 and long long will be 4 bytes. Please note that this option does not
7004 comply to the C standards, but it will provide you with smaller code
7009 @subsection CRIS Options
7010 @cindex CRIS Options
7012 These options are defined specifically for the CRIS ports.
7015 @item -march=@var{architecture-type}
7016 @itemx -mcpu=@var{architecture-type}
7019 Generate code for the specified architecture. The choices for
7020 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
7021 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX.
7022 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
7025 @item -mtune=@var{architecture-type}
7027 Tune to @var{architecture-type} everything applicable about the generated
7028 code, except for the ABI and the set of available instructions. The
7029 choices for @var{architecture-type} are the same as for
7030 @option{-march=@var{architecture-type}}.
7032 @item -mmax-stack-frame=@var{n}
7033 @opindex mmax-stack-frame
7034 Warn when the stack frame of a function exceeds @var{n} bytes.
7036 @item -melinux-stacksize=@var{n}
7037 @opindex melinux-stacksize
7038 Only available with the @samp{cris-axis-aout} target. Arranges for
7039 indications in the program to the kernel loader that the stack of the
7040 program should be set to @var{n} bytes.
7046 The options @option{-metrax4} and @option{-metrax100} are synonyms for
7047 @option{-march=v3} and @option{-march=v8} respectively.
7049 @item -mmul-bug-workaround
7050 @itemx -mno-mul-bug-workaround
7051 @opindex mmul-bug-workaround
7052 @opindex mno-mul-bug-workaround
7053 Work around a bug in the @code{muls} and @code{mulu} instructions for CPU
7054 models where it applies. This option is active by default.
7058 Enable CRIS-specific verbose debug-related information in the assembly
7059 code. This option also has the effect to turn off the @samp{#NO_APP}
7060 formatted-code indicator to the assembler at the beginning of the
7065 Do not use condition-code results from previous instruction; always emit
7066 compare and test instructions before use of condition codes.
7068 @item -mno-side-effects
7069 @opindex mno-side-effects
7070 Do not emit instructions with side-effects in addressing modes other than
7074 @itemx -mno-stack-align
7076 @itemx -mno-data-align
7077 @itemx -mconst-align
7078 @itemx -mno-const-align
7079 @opindex mstack-align
7080 @opindex mno-stack-align
7081 @opindex mdata-align
7082 @opindex mno-data-align
7083 @opindex mconst-align
7084 @opindex mno-const-align
7085 These options (no-options) arranges (eliminate arrangements) for the
7086 stack-frame, individual data and constants to be aligned for the maximum
7087 single data access size for the chosen CPU model. The default is to
7088 arrange for 32-bit alignment. ABI details such as structure layout are
7089 not affected by these options.
7097 Similar to the stack- data- and const-align options above, these options
7098 arrange for stack-frame, writable data and constants to all be 32-bit,
7099 16-bit or 8-bit aligned. The default is 32-bit alignment.
7101 @item -mno-prologue-epilogue
7102 @itemx -mprologue-epilogue
7103 @opindex mno-prologue-epilogue
7104 @opindex mprologue-epilogue
7105 With @option{-mno-prologue-epilogue}, the normal function prologue and
7106 epilogue that sets up the stack-frame are omitted and no return
7107 instructions or return sequences are generated in the code. Use this
7108 option only together with visual inspection of the compiled code: no
7109 warnings or errors are generated when call-saved registers must be saved,
7110 or storage for local variable needs to be allocated.
7116 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
7117 instruction sequences that load addresses for functions from the PLT part
7118 of the GOT rather than (traditional on other architectures) calls to the
7119 PLT. The default is @option{-mgotplt}.
7123 Legacy no-op option only recognized with the cris-axis-aout target.
7127 Legacy no-op option only recognized with the cris-axis-elf and
7128 cris-axis-linux-gnu targets.
7132 Only recognized with the cris-axis-aout target, where it selects a
7133 GNU/linux-like multilib, include files and instruction set for
7138 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
7142 This option, recognized for the cris-axis-aout and cris-axis-elf arranges
7143 to link with input-output functions from a simulator library. Code,
7144 initialized data and zero-initialized data are allocated consecutively.
7148 Like @option{-sim}, but pass linker options to locate initialized data at
7149 0x40000000 and zero-initialized data at 0x80000000.
7152 @node Darwin Options
7153 @subsection Darwin Options
7154 @cindex Darwin options
7156 These options are defined for all architectures running the Darwin operating
7157 system. They are useful for compatibility with other Mac OS compilers.
7162 Add the framework directory @var{dir} to the head of the list of
7163 directories to be searched for header files. These directories are
7164 interleaved with those specified by @option{-I} options and are
7165 scanned in a left-to-right order.
7167 A framework directory is a directory with frameworks in it. A
7168 framework is a directory with a @samp{"Headers"} and/or
7169 @samp{"PrivateHeaders"} directory contained directly in it that ends
7170 in @samp{".framework"}. The name of a framework is the name of this
7171 directory excluding the @samp{".framework"}. Headers associated with
7172 the framework are found in one of those two directories, with
7173 @samp{"Headers"} being searched first. A subframework is a framework
7174 directory that is in a framework's @samp{"Frameworks"} directory.
7175 Includes of subframework headers can only appear in a header of a
7176 framework that contains the subframework, or in a sibling subframework
7177 header. Two subframeworks are siblings if they occur in the same
7178 framework. A subframework should not have the same name as a
7179 framework, a warning will be issued if this is violated. Currently a
7180 subframework cannot have subframeworks, in the future, the mechanism
7181 may be extended to support this. The standard frameworks can be found
7182 in @samp{"/System/Library/Frameworks"} and
7183 @samp{"/Library/Frameworks"}. An example include looks like
7184 @code{#include <Framework/header.h>}, where @samp{Framework} denotes
7185 the name of the framework and header.h is found in the
7186 @samp{"PrivateHeaders"} or @samp{"Headers"} directory.
7190 Emit debugging information for symbols that are used. For STABS
7191 debugging format, this enables @option{-feliminate-unused-debug-symbols}.
7192 This is by default ON.
7196 Emit debugging information for all symbols and types.
7198 @item -mone-byte-bool
7199 @opindex -mone-byte-bool
7200 Override the defaults for @samp{bool} so that @samp{sizeof(bool)==1}.
7201 By default @samp{sizeof(bool)} is @samp{4} when compiling for
7202 Darwin/PowerPC and @samp{1} when compiling for Darwin/x86, so this
7203 option has no effect on x86.
7205 @strong{Warning:} The @option{-mone-byte-bool} switch causes GCC
7206 to generate code that is not binary compatible with code generated
7207 without that switch. Using this switch may require recompiling all
7208 other modules in a program, including system libraries. Use this
7209 switch to conform to a non-default data model.
7211 @item -mfix-and-continue
7212 @itemx -ffix-and-continue
7213 @itemx -findirect-data
7214 @opindex mfix-and-continue
7215 @opindex ffix-and-continue
7216 @opindex findirect-data
7217 Generate code suitable for fast turn around development. Needed to
7218 enable gdb to dynamically load @code{.o} files into already running
7219 programs. @option{-findirect-data} and @option{-ffix-and-continue}
7220 are provided for backwards compatibility.
7224 Loads all members of static archive libraries.
7225 See man ld(1) for more information.
7227 @item -arch_errors_fatal
7228 @opindex arch_errors_fatal
7229 Cause the errors having to do with files that have the wrong architecture
7233 @opindex bind_at_load
7234 Causes the output file to be marked such that the dynamic linker will
7235 bind all undefined references when the file is loaded or launched.
7239 Produce a Mach-o bundle format file.
7240 See man ld(1) for more information.
7242 @item -bundle_loader @var{executable}
7243 @opindex bundle_loader
7244 This specifies the @var{executable} that will be loading the build
7245 output file being linked. See man ld(1) for more information.
7247 @item -allowable_client @var{client_name}
7251 @itemx -compatibility_version
7252 @itemx -current_version
7254 @itemx -dependency-file
7256 @itemx -dylinker_install_name
7259 @itemx -exported_symbols_list
7261 @itemx -flat_namespace
7262 @itemx -force_cpusubtype_ALL
7263 @itemx -force_flat_namespace
7264 @itemx -headerpad_max_install_names
7267 @itemx -install_name
7268 @itemx -keep_private_externs
7269 @itemx -multi_module
7270 @itemx -multiply_defined
7271 @itemx -multiply_defined_unused
7273 @itemx -no_dead_strip_inits_and_terms
7274 @itemx -nofixprebinding
7277 @itemx -noseglinkedit
7278 @itemx -pagezero_size
7280 @itemx -prebind_all_twolevel_modules
7281 @itemx -private_bundle
7282 @itemx -read_only_relocs
7284 @itemx -sectobjectsymbols
7288 @itemx -sectobjectsymbols
7291 @itemx -segs_read_only_addr
7292 @itemx -segs_read_write_addr
7293 @itemx -seg_addr_table
7294 @itemx -seg_addr_table_filename
7297 @itemx -segs_read_only_addr
7298 @itemx -segs_read_write_addr
7299 @itemx -single_module
7302 @itemx -sub_umbrella
7303 @itemx -twolevel_namespace
7306 @itemx -unexported_symbols_list
7307 @itemx -weak_reference_mismatches
7310 @opindex allowable_client
7312 @opindex client_name
7313 @opindex compatibility_version
7314 @opindex current_version
7316 @opindex dependency-file
7318 @opindex dylinker_install_name
7321 @opindex exported_symbols_list
7323 @opindex flat_namespace
7324 @opindex force_cpusubtype_ALL
7325 @opindex force_flat_namespace
7326 @opindex headerpad_max_install_names
7329 @opindex install_name
7330 @opindex keep_private_externs
7331 @opindex multi_module
7332 @opindex multiply_defined
7333 @opindex multiply_defined_unused
7335 @opindex no_dead_strip_inits_and_terms
7336 @opindex nofixprebinding
7337 @opindex nomultidefs
7339 @opindex noseglinkedit
7340 @opindex pagezero_size
7342 @opindex prebind_all_twolevel_modules
7343 @opindex private_bundle
7344 @opindex read_only_relocs
7346 @opindex sectobjectsymbols
7350 @opindex sectobjectsymbols
7353 @opindex segs_read_only_addr
7354 @opindex segs_read_write_addr
7355 @opindex seg_addr_table
7356 @opindex seg_addr_table_filename
7357 @opindex seglinkedit
7359 @opindex segs_read_only_addr
7360 @opindex segs_read_write_addr
7361 @opindex single_module
7363 @opindex sub_library
7364 @opindex sub_umbrella
7365 @opindex twolevel_namespace
7368 @opindex unexported_symbols_list
7369 @opindex weak_reference_mismatches
7370 @opindex whatsloaded
7372 These options are available for Darwin linker. Darwin linker man page
7373 describes them in detail.
7376 @node DEC Alpha Options
7377 @subsection DEC Alpha Options
7379 These @samp{-m} options are defined for the DEC Alpha implementations:
7382 @item -mno-soft-float
7384 @opindex mno-soft-float
7385 @opindex msoft-float
7386 Use (do not use) the hardware floating-point instructions for
7387 floating-point operations. When @option{-msoft-float} is specified,
7388 functions in @file{libgcc.a} will be used to perform floating-point
7389 operations. Unless they are replaced by routines that emulate the
7390 floating-point operations, or compiled in such a way as to call such
7391 emulations routines, these routines will issue floating-point
7392 operations. If you are compiling for an Alpha without floating-point
7393 operations, you must ensure that the library is built so as not to call
7396 Note that Alpha implementations without floating-point operations are
7397 required to have floating-point registers.
7402 @opindex mno-fp-regs
7403 Generate code that uses (does not use) the floating-point register set.
7404 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
7405 register set is not used, floating point operands are passed in integer
7406 registers as if they were integers and floating-point results are passed
7407 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
7408 so any function with a floating-point argument or return value called by code
7409 compiled with @option{-mno-fp-regs} must also be compiled with that
7412 A typical use of this option is building a kernel that does not use,
7413 and hence need not save and restore, any floating-point registers.
7417 The Alpha architecture implements floating-point hardware optimized for
7418 maximum performance. It is mostly compliant with the IEEE floating
7419 point standard. However, for full compliance, software assistance is
7420 required. This option generates code fully IEEE compliant code
7421 @emph{except} that the @var{inexact-flag} is not maintained (see below).
7422 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
7423 defined during compilation. The resulting code is less efficient but is
7424 able to correctly support denormalized numbers and exceptional IEEE
7425 values such as not-a-number and plus/minus infinity. Other Alpha
7426 compilers call this option @option{-ieee_with_no_inexact}.
7428 @item -mieee-with-inexact
7429 @opindex mieee-with-inexact
7430 This is like @option{-mieee} except the generated code also maintains
7431 the IEEE @var{inexact-flag}. Turning on this option causes the
7432 generated code to implement fully-compliant IEEE math. In addition to
7433 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
7434 macro. On some Alpha implementations the resulting code may execute
7435 significantly slower than the code generated by default. Since there is
7436 very little code that depends on the @var{inexact-flag}, you should
7437 normally not specify this option. Other Alpha compilers call this
7438 option @option{-ieee_with_inexact}.
7440 @item -mfp-trap-mode=@var{trap-mode}
7441 @opindex mfp-trap-mode
7442 This option controls what floating-point related traps are enabled.
7443 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
7444 The trap mode can be set to one of four values:
7448 This is the default (normal) setting. The only traps that are enabled
7449 are the ones that cannot be disabled in software (e.g., division by zero
7453 In addition to the traps enabled by @samp{n}, underflow traps are enabled
7457 Like @samp{su}, but the instructions are marked to be safe for software
7458 completion (see Alpha architecture manual for details).
7461 Like @samp{su}, but inexact traps are enabled as well.
7464 @item -mfp-rounding-mode=@var{rounding-mode}
7465 @opindex mfp-rounding-mode
7466 Selects the IEEE rounding mode. Other Alpha compilers call this option
7467 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
7472 Normal IEEE rounding mode. Floating point numbers are rounded towards
7473 the nearest machine number or towards the even machine number in case
7477 Round towards minus infinity.
7480 Chopped rounding mode. Floating point numbers are rounded towards zero.
7483 Dynamic rounding mode. A field in the floating point control register
7484 (@var{fpcr}, see Alpha architecture reference manual) controls the
7485 rounding mode in effect. The C library initializes this register for
7486 rounding towards plus infinity. Thus, unless your program modifies the
7487 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
7490 @item -mtrap-precision=@var{trap-precision}
7491 @opindex mtrap-precision
7492 In the Alpha architecture, floating point traps are imprecise. This
7493 means without software assistance it is impossible to recover from a
7494 floating trap and program execution normally needs to be terminated.
7495 GCC can generate code that can assist operating system trap handlers
7496 in determining the exact location that caused a floating point trap.
7497 Depending on the requirements of an application, different levels of
7498 precisions can be selected:
7502 Program precision. This option is the default and means a trap handler
7503 can only identify which program caused a floating point exception.
7506 Function precision. The trap handler can determine the function that
7507 caused a floating point exception.
7510 Instruction precision. The trap handler can determine the exact
7511 instruction that caused a floating point exception.
7514 Other Alpha compilers provide the equivalent options called
7515 @option{-scope_safe} and @option{-resumption_safe}.
7517 @item -mieee-conformant
7518 @opindex mieee-conformant
7519 This option marks the generated code as IEEE conformant. You must not
7520 use this option unless you also specify @option{-mtrap-precision=i} and either
7521 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
7522 is to emit the line @samp{.eflag 48} in the function prologue of the
7523 generated assembly file. Under DEC Unix, this has the effect that
7524 IEEE-conformant math library routines will be linked in.
7526 @item -mbuild-constants
7527 @opindex mbuild-constants
7528 Normally GCC examines a 32- or 64-bit integer constant to
7529 see if it can construct it from smaller constants in two or three
7530 instructions. If it cannot, it will output the constant as a literal and
7531 generate code to load it from the data segment at runtime.
7533 Use this option to require GCC to construct @emph{all} integer constants
7534 using code, even if it takes more instructions (the maximum is six).
7536 You would typically use this option to build a shared library dynamic
7537 loader. Itself a shared library, it must relocate itself in memory
7538 before it can find the variables and constants in its own data segment.
7544 Select whether to generate code to be assembled by the vendor-supplied
7545 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
7563 Indicate whether GCC should generate code to use the optional BWX,
7564 CIX, FIX and MAX instruction sets. The default is to use the instruction
7565 sets supported by the CPU type specified via @option{-mcpu=} option or that
7566 of the CPU on which GCC was built if none was specified.
7571 @opindex mfloat-ieee
7572 Generate code that uses (does not use) VAX F and G floating point
7573 arithmetic instead of IEEE single and double precision.
7575 @item -mexplicit-relocs
7576 @itemx -mno-explicit-relocs
7577 @opindex mexplicit-relocs
7578 @opindex mno-explicit-relocs
7579 Older Alpha assemblers provided no way to generate symbol relocations
7580 except via assembler macros. Use of these macros does not allow
7581 optimal instruction scheduling. GNU binutils as of version 2.12
7582 supports a new syntax that allows the compiler to explicitly mark
7583 which relocations should apply to which instructions. This option
7584 is mostly useful for debugging, as GCC detects the capabilities of
7585 the assembler when it is built and sets the default accordingly.
7589 @opindex msmall-data
7590 @opindex mlarge-data
7591 When @option{-mexplicit-relocs} is in effect, static data is
7592 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
7593 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
7594 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
7595 16-bit relocations off of the @code{$gp} register. This limits the
7596 size of the small data area to 64KB, but allows the variables to be
7597 directly accessed via a single instruction.
7599 The default is @option{-mlarge-data}. With this option the data area
7600 is limited to just below 2GB. Programs that require more than 2GB of
7601 data must use @code{malloc} or @code{mmap} to allocate the data in the
7602 heap instead of in the program's data segment.
7604 When generating code for shared libraries, @option{-fpic} implies
7605 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
7609 @opindex msmall-text
7610 @opindex mlarge-text
7611 When @option{-msmall-text} is used, the compiler assumes that the
7612 code of the entire program (or shared library) fits in 4MB, and is
7613 thus reachable with a branch instruction. When @option{-msmall-data}
7614 is used, the compiler can assume that all local symbols share the
7615 same @code{$gp} value, and thus reduce the number of instructions
7616 required for a function call from 4 to 1.
7618 The default is @option{-mlarge-text}.
7620 @item -mcpu=@var{cpu_type}
7622 Set the instruction set and instruction scheduling parameters for
7623 machine type @var{cpu_type}. You can specify either the @samp{EV}
7624 style name or the corresponding chip number. GCC supports scheduling
7625 parameters for the EV4, EV5 and EV6 family of processors and will
7626 choose the default values for the instruction set from the processor
7627 you specify. If you do not specify a processor type, GCC will default
7628 to the processor on which the compiler was built.
7630 Supported values for @var{cpu_type} are
7636 Schedules as an EV4 and has no instruction set extensions.
7640 Schedules as an EV5 and has no instruction set extensions.
7644 Schedules as an EV5 and supports the BWX extension.
7649 Schedules as an EV5 and supports the BWX and MAX extensions.
7653 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
7657 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
7660 @item -mtune=@var{cpu_type}
7662 Set only the instruction scheduling parameters for machine type
7663 @var{cpu_type}. The instruction set is not changed.
7665 @item -mmemory-latency=@var{time}
7666 @opindex mmemory-latency
7667 Sets the latency the scheduler should assume for typical memory
7668 references as seen by the application. This number is highly
7669 dependent on the memory access patterns used by the application
7670 and the size of the external cache on the machine.
7672 Valid options for @var{time} are
7676 A decimal number representing clock cycles.
7682 The compiler contains estimates of the number of clock cycles for
7683 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
7684 (also called Dcache, Scache, and Bcache), as well as to main memory.
7685 Note that L3 is only valid for EV5.
7690 @node DEC Alpha/VMS Options
7691 @subsection DEC Alpha/VMS Options
7693 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
7696 @item -mvms-return-codes
7697 @opindex mvms-return-codes
7698 Return VMS condition codes from main. The default is to return POSIX
7699 style condition (e.g.@ error) codes.
7703 @subsection FRV Options
7710 Only use the first 32 general purpose registers.
7715 Use all 64 general purpose registers.
7720 Use only the first 32 floating point registers.
7725 Use all 64 floating point registers
7728 @opindex mhard-float
7730 Use hardware instructions for floating point operations.
7733 @opindex msoft-float
7735 Use library routines for floating point operations.
7740 Dynamically allocate condition code registers.
7745 Do not try to dynamically allocate condition code registers, only
7746 use @code{icc0} and @code{fcc0}.
7751 Change ABI to use double word insns.
7756 Do not use double word instructions.
7761 Use floating point double instructions.
7766 Do not use floating point double instructions.
7771 Use media instructions.
7776 Do not use media instructions.
7781 Use multiply and add/subtract instructions.
7786 Do not use multiply and add/subtract instructions.
7791 Select the FDPIC ABI, that uses function descriptors to represent
7792 pointers to functions. Without any PIC/PIE-related options, it
7793 implies @option{-fPIE}. With @option{-fpic} or @option{-fpie}, it
7794 assumes GOT entries and small data are within a 12-bit range from the
7795 GOT base address; with @option{-fPIC} or @option{-fPIE}, GOT offsets
7796 are computed with 32 bits.
7799 @opindex minline-plt
7801 Enable inlining of PLT entries in function calls to functions that are
7802 not known to bind locally. It has no effect without @option{-mfdpic}.
7803 It's enabled by default if optimizing for speed and compiling for
7804 shared libraries (i.e., @option{-fPIC} or @option{-fpic}), or when an
7805 optimization option such as @option{-O3} or above is present in the
7811 Enable the use of @code{GPREL} relocations in the FDPIC ABI for data
7812 that is known to be in read-only sections. It's enabled by default,
7813 except for @option{-fpic} or @option{-fpie}: even though it may help
7814 make the global offset table smaller, it trades 1 instruction for 4.
7815 With @option{-fPIC} or @option{-fPIE}, it trades 3 instructions for 4,
7816 one of which may be shared by multiple symbols, and it avoids the need
7817 for a GOT entry for the referenced symbol, so it's more likely to be a
7818 win. If it is not, @option{-mno-gprel-ro} can be used to disable it.
7820 @item -multilib-library-pic
7821 @opindex multilib-library-pic
7823 Link with the (library, not FD) pic libraries. It's implied by
7824 @option{-mlibrary-pic}, as well as by @option{-fPIC} and
7825 @option{-fpic} without @option{-mfdpic}. You should never have to use
7831 Follow the EABI requirement of always creating a frame pointer whenever
7832 a stack frame is allocated. This option is enabled by default and can
7833 be disabled with @option{-mno-linked-fp}.
7836 @opindex mlong-calls
7838 Use indirect addressing to call functions outside the current
7839 compilation unit. This allows the functions to be placed anywhere
7840 within the 32-bit address space.
7842 @item -malign-labels
7843 @opindex malign-labels
7845 Try to align labels to an 8-byte boundary by inserting nops into the
7846 previous packet. This option only has an effect when VLIW packing
7847 is enabled. It doesn't create new packets; it merely adds nops to
7851 @opindex mlibrary-pic
7853 Generate position-independent EABI code.
7858 Use only the first four media accumulator registers.
7863 Use all eight media accumulator registers.
7868 Pack VLIW instructions.
7873 Do not pack VLIW instructions.
7878 Do not mark ABI switches in e_flags.
7883 Enable the use of conditional-move instructions (default).
7885 This switch is mainly for debugging the compiler and will likely be removed
7886 in a future version.
7888 @item -mno-cond-move
7889 @opindex mno-cond-move
7891 Disable the use of conditional-move instructions.
7893 This switch is mainly for debugging the compiler and will likely be removed
7894 in a future version.
7899 Enable the use of conditional set instructions (default).
7901 This switch is mainly for debugging the compiler and will likely be removed
7902 in a future version.
7907 Disable the use of conditional set instructions.
7909 This switch is mainly for debugging the compiler and will likely be removed
7910 in a future version.
7915 Enable the use of conditional execution (default).
7917 This switch is mainly for debugging the compiler and will likely be removed
7918 in a future version.
7920 @item -mno-cond-exec
7921 @opindex mno-cond-exec
7923 Disable the use of conditional execution.
7925 This switch is mainly for debugging the compiler and will likely be removed
7926 in a future version.
7929 @opindex mvliw-branch
7931 Run a pass to pack branches into VLIW instructions (default).
7933 This switch is mainly for debugging the compiler and will likely be removed
7934 in a future version.
7936 @item -mno-vliw-branch
7937 @opindex mno-vliw-branch
7939 Do not run a pass to pack branches into VLIW instructions.
7941 This switch is mainly for debugging the compiler and will likely be removed
7942 in a future version.
7944 @item -mmulti-cond-exec
7945 @opindex mmulti-cond-exec
7947 Enable optimization of @code{&&} and @code{||} in conditional execution
7950 This switch is mainly for debugging the compiler and will likely be removed
7951 in a future version.
7953 @item -mno-multi-cond-exec
7954 @opindex mno-multi-cond-exec
7956 Disable optimization of @code{&&} and @code{||} in conditional execution.
7958 This switch is mainly for debugging the compiler and will likely be removed
7959 in a future version.
7961 @item -mnested-cond-exec
7962 @opindex mnested-cond-exec
7964 Enable nested conditional execution optimizations (default).
7966 This switch is mainly for debugging the compiler and will likely be removed
7967 in a future version.
7969 @item -mno-nested-cond-exec
7970 @opindex mno-nested-cond-exec
7972 Disable nested conditional execution optimizations.
7974 This switch is mainly for debugging the compiler and will likely be removed
7975 in a future version.
7977 @item -mtomcat-stats
7978 @opindex mtomcat-stats
7980 Cause gas to print out tomcat statistics.
7982 @item -mcpu=@var{cpu}
7985 Select the processor type for which to generate code. Possible values are
7986 @samp{frv}, @samp{fr550}, @samp{tomcat}, @samp{fr500}, @samp{fr450},
7987 @samp{fr405}, @samp{fr400}, @samp{fr300} and @samp{simple}.
7991 @node H8/300 Options
7992 @subsection H8/300 Options
7994 These @samp{-m} options are defined for the H8/300 implementations:
7999 Shorten some address references at link time, when possible; uses the
8000 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
8001 ld, Using ld}, for a fuller description.
8005 Generate code for the H8/300H@.
8009 Generate code for the H8S@.
8013 Generate code for the H8S and H8/300H in the normal mode. This switch
8014 must be used either with -mh or -ms.
8018 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
8022 Make @code{int} data 32 bits by default.
8026 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
8027 The default for the H8/300H and H8S is to align longs and floats on 4
8029 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
8030 This option has no effect on the H8/300.
8034 @subsection HPPA Options
8035 @cindex HPPA Options
8037 These @samp{-m} options are defined for the HPPA family of computers:
8040 @item -march=@var{architecture-type}
8042 Generate code for the specified architecture. The choices for
8043 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
8044 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
8045 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
8046 architecture option for your machine. Code compiled for lower numbered
8047 architectures will run on higher numbered architectures, but not the
8051 @itemx -mpa-risc-1-1
8052 @itemx -mpa-risc-2-0
8053 @opindex mpa-risc-1-0
8054 @opindex mpa-risc-1-1
8055 @opindex mpa-risc-2-0
8056 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
8059 @opindex mbig-switch
8060 Generate code suitable for big switch tables. Use this option only if
8061 the assembler/linker complain about out of range branches within a switch
8064 @item -mjump-in-delay
8065 @opindex mjump-in-delay
8066 Fill delay slots of function calls with unconditional jump instructions
8067 by modifying the return pointer for the function call to be the target
8068 of the conditional jump.
8070 @item -mdisable-fpregs
8071 @opindex mdisable-fpregs
8072 Prevent floating point registers from being used in any manner. This is
8073 necessary for compiling kernels which perform lazy context switching of
8074 floating point registers. If you use this option and attempt to perform
8075 floating point operations, the compiler will abort.
8077 @item -mdisable-indexing
8078 @opindex mdisable-indexing
8079 Prevent the compiler from using indexing address modes. This avoids some
8080 rather obscure problems when compiling MIG generated code under MACH@.
8082 @item -mno-space-regs
8083 @opindex mno-space-regs
8084 Generate code that assumes the target has no space registers. This allows
8085 GCC to generate faster indirect calls and use unscaled index address modes.
8087 Such code is suitable for level 0 PA systems and kernels.
8089 @item -mfast-indirect-calls
8090 @opindex mfast-indirect-calls
8091 Generate code that assumes calls never cross space boundaries. This
8092 allows GCC to emit code which performs faster indirect calls.
8094 This option will not work in the presence of shared libraries or nested
8097 @item -mfixed-range=@var{register-range}
8098 @opindex mfixed-range
8099 Generate code treating the given register range as fixed registers.
8100 A fixed register is one that the register allocator can not use. This is
8101 useful when compiling kernel code. A register range is specified as
8102 two registers separated by a dash. Multiple register ranges can be
8103 specified separated by a comma.
8105 @item -mlong-load-store
8106 @opindex mlong-load-store
8107 Generate 3-instruction load and store sequences as sometimes required by
8108 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
8111 @item -mportable-runtime
8112 @opindex mportable-runtime
8113 Use the portable calling conventions proposed by HP for ELF systems.
8117 Enable the use of assembler directives only GAS understands.
8119 @item -mschedule=@var{cpu-type}
8121 Schedule code according to the constraints for the machine type
8122 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
8123 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
8124 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
8125 proper scheduling option for your machine. The default scheduling is
8129 @opindex mlinker-opt
8130 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
8131 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
8132 linkers in which they give bogus error messages when linking some programs.
8135 @opindex msoft-float
8136 Generate output containing library calls for floating point.
8137 @strong{Warning:} the requisite libraries are not available for all HPPA
8138 targets. Normally the facilities of the machine's usual C compiler are
8139 used, but this cannot be done directly in cross-compilation. You must make
8140 your own arrangements to provide suitable library functions for
8141 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
8142 does provide software floating point support.
8144 @option{-msoft-float} changes the calling convention in the output file;
8145 therefore, it is only useful if you compile @emph{all} of a program with
8146 this option. In particular, you need to compile @file{libgcc.a}, the
8147 library that comes with GCC, with @option{-msoft-float} in order for
8152 Generate the predefine, @code{_SIO}, for server IO. The default is
8153 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
8154 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO. These
8155 options are available under HP-UX and HI-UX.
8159 Use GNU ld specific options. This passes @option{-shared} to ld when
8160 building a shared library. It is the default when GCC is configured,
8161 explicitly or implicitly, with the GNU linker. This option does not
8162 have any affect on which ld is called, it only changes what parameters
8163 are passed to that ld. The ld that is called is determined by the
8164 @option{--with-ld} configure option, GCC's program search path, and
8165 finally by the user's @env{PATH}. The linker used by GCC can be printed
8166 using @samp{which `gcc -print-prog-name=ld`}.
8170 Use HP ld specific options. This passes @option{-b} to ld when building
8171 a shared library and passes @option{+Accept TypeMismatch} to ld on all
8172 links. It is the default when GCC is configured, explicitly or
8173 implicitly, with the HP linker. This option does not have any affect on
8174 which ld is called, it only changes what parameters are passed to that
8175 ld. The ld that is called is determined by the @option{--with-ld}
8176 configure option, GCC's program search path, and finally by the user's
8177 @env{PATH}. The linker used by GCC can be printed using @samp{which
8178 `gcc -print-prog-name=ld`}.
8181 @opindex mno-long-calls
8182 Generate code that uses long call sequences. This ensures that a call
8183 is always able to reach linker generated stubs. The default is to generate
8184 long calls only when the distance from the call site to the beginning
8185 of the function or translation unit, as the case may be, exceeds a
8186 predefined limit set by the branch type being used. The limits for
8187 normal calls are 7,600,000 and 240,000 bytes, respectively for the
8188 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
8191 Distances are measured from the beginning of functions when using the
8192 @option{-ffunction-sections} option, or when using the @option{-mgas}
8193 and @option{-mno-portable-runtime} options together under HP-UX with
8196 It is normally not desirable to use this option as it will degrade
8197 performance. However, it may be useful in large applications,
8198 particularly when partial linking is used to build the application.
8200 The types of long calls used depends on the capabilities of the
8201 assembler and linker, and the type of code being generated. The
8202 impact on systems that support long absolute calls, and long pic
8203 symbol-difference or pc-relative calls should be relatively small.
8204 However, an indirect call is used on 32-bit ELF systems in pic code
8205 and it is quite long.
8207 @item -munix=@var{unix-std}
8209 Generate compiler predefines and select a startfile for the specified
8210 UNIX standard. The choices for @var{unix-std} are @samp{93}, @samp{95}
8211 and @samp{98}. @samp{93} is supported on all HP-UX versions. @samp{95}
8212 is available on HP-UX 10.10 and later. @samp{98} is available on HP-UX
8213 11.11 and later. The default values are @samp{93} for HP-UX 10.00,
8214 @samp{95} for HP-UX 10.10 though to 11.00, and @samp{98} for HP-UX 11.11
8217 @option{-munix=93} provides the same predefines as GCC 3.3 and 3.4.
8218 @option{-munix=95} provides additional predefines for @code{XOPEN_UNIX}
8219 and @code{_XOPEN_SOURCE_EXTENDED}, and the startfile @file{unix95.o}.
8220 @option{-munix=98} provides additional predefines for @code{_XOPEN_UNIX},
8221 @code{_XOPEN_SOURCE_EXTENDED}, @code{_INCLUDE__STDC_A1_SOURCE} and
8222 @code{_INCLUDE_XOPEN_SOURCE_500}, and the startfile @file{unix98.o}.
8224 It is @emph{important} to note that this option changes the interfaces
8225 for various library routines. It also affects the operational behavior
8226 of the C library. Thus, @emph{extreme} care is needed in using this
8229 Library code that is intended to operate with more than one UNIX
8230 standard must test, set and restore the variable @var{__xpg4_extended_mask}
8231 as appropriate. Most GNU software doesn't provide this capability.
8235 Suppress the generation of link options to search libdld.sl when the
8236 @option{-static} option is specified on HP-UX 10 and later.
8240 The HP-UX implementation of setlocale in libc has a dependency on
8241 libdld.sl. There isn't an archive version of libdld.sl. Thus,
8242 when the @option{-static} option is specified, special link options
8243 are needed to resolve this dependency.
8245 On HP-UX 10 and later, the GCC driver adds the necessary options to
8246 link with libdld.sl when the @option{-static} option is specified.
8247 This causes the resulting binary to be dynamic. On the 64-bit port,
8248 the linkers generate dynamic binaries by default in any case. The
8249 @option{-nolibdld} option can be used to prevent the GCC driver from
8250 adding these link options.
8254 Add support for multithreading with the @dfn{dce thread} library
8255 under HP-UX. This option sets flags for both the preprocessor and
8259 @node i386 and x86-64 Options
8260 @subsection Intel 386 and AMD x86-64 Options
8261 @cindex i386 Options
8262 @cindex x86-64 Options
8263 @cindex Intel 386 Options
8264 @cindex AMD x86-64 Options
8266 These @samp{-m} options are defined for the i386 and x86-64 family of
8270 @item -mtune=@var{cpu-type}
8272 Tune to @var{cpu-type} everything applicable about the generated code, except
8273 for the ABI and the set of available instructions. The choices for
8277 Original Intel's i386 CPU.
8279 Intel's i486 CPU. (No scheduling is implemented for this chip.)
8281 Intel Pentium CPU with no MMX support.
8283 Intel PentiumMMX CPU based on Pentium core with MMX instruction set support.
8284 @item i686, pentiumpro
8285 Intel PentiumPro CPU.
8287 Intel Pentium2 CPU based on PentiumPro core with MMX instruction set support.
8288 @item pentium3, pentium3m
8289 Intel Pentium3 CPU based on PentiumPro core with MMX and SSE instruction set
8292 Low power version of Intel Pentium3 CPU with MMX, SSE and SSE2 instruction set
8293 support. Used by Centrino notebooks.
8294 @item pentium4, pentium4m
8295 Intel Pentium4 CPU with MMX, SSE and SSE2 instruction set support.
8297 Improved version of Intel Pentium4 CPU with MMX, SSE, SSE2 and SSE3 instruction
8300 Improved version of Intel Pentium4 CPU with 64-bit extensions, MMX, SSE,
8301 SSE2 and SSE3 instruction set support.
8303 AMD K6 CPU with MMX instruction set support.
8305 Improved versions of AMD K6 CPU with MMX and 3dNOW! instruction set support.
8306 @item athlon, athlon-tbird
8307 AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and SSE prefetch instructions
8309 @item athlon-4, athlon-xp, athlon-mp
8310 Improved AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and full SSE
8311 instruction set support.
8312 @item k8, opteron, athlon64, athlon-fx
8313 AMD K8 core based CPUs with x86-64 instruction set support. (This supersets
8314 MMX, SSE, SSE2, 3dNOW!, enhanced 3dNOW! and 64-bit instruction set extensions.)
8316 IDT Winchip C6 CPU, dealt in same way as i486 with additional MMX instruction
8319 IDT Winchip2 CPU, dealt in same way as i486 with additional MMX and 3dNOW!
8320 instruction set support.
8322 Via C3 CPU with MMX and 3dNOW! instruction set support. (No scheduling is
8323 implemented for this chip.)
8325 Via C3-2 CPU with MMX and SSE instruction set support. (No scheduling is
8326 implemented for this chip.)
8329 While picking a specific @var{cpu-type} will schedule things appropriately
8330 for that particular chip, the compiler will not generate any code that
8331 does not run on the i386 without the @option{-march=@var{cpu-type}} option
8334 @item -march=@var{cpu-type}
8336 Generate instructions for the machine type @var{cpu-type}. The choices
8337 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
8338 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
8340 @item -mcpu=@var{cpu-type}
8342 A deprecated synonym for @option{-mtune}.
8351 @opindex mpentiumpro
8352 These options are synonyms for @option{-mtune=i386}, @option{-mtune=i486},
8353 @option{-mtune=pentium}, and @option{-mtune=pentiumpro} respectively.
8354 These synonyms are deprecated.
8356 @item -mfpmath=@var{unit}
8358 Generate floating point arithmetics for selected unit @var{unit}. The choices
8363 Use the standard 387 floating point coprocessor present majority of chips and
8364 emulated otherwise. Code compiled with this option will run almost everywhere.
8365 The temporary results are computed in 80bit precision instead of precision
8366 specified by the type resulting in slightly different results compared to most
8367 of other chips. See @option{-ffloat-store} for more detailed description.
8369 This is the default choice for i386 compiler.
8372 Use scalar floating point instructions present in the SSE instruction set.
8373 This instruction set is supported by Pentium3 and newer chips, in the AMD line
8374 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
8375 instruction set supports only single precision arithmetics, thus the double and
8376 extended precision arithmetics is still done using 387. Later version, present
8377 only in Pentium4 and the future AMD x86-64 chips supports double precision
8380 For i387 you need to use @option{-march=@var{cpu-type}}, @option{-msse} or
8381 @option{-msse2} switches to enable SSE extensions and make this option
8382 effective. For x86-64 compiler, these extensions are enabled by default.
8384 The resulting code should be considerably faster in the majority of cases and avoid
8385 the numerical instability problems of 387 code, but may break some existing
8386 code that expects temporaries to be 80bit.
8388 This is the default choice for the x86-64 compiler.
8391 Attempt to utilize both instruction sets at once. This effectively double the
8392 amount of available registers and on chips with separate execution units for
8393 387 and SSE the execution resources too. Use this option with care, as it is
8394 still experimental, because the GCC register allocator does not model separate
8395 functional units well resulting in instable performance.
8398 @item -masm=@var{dialect}
8399 @opindex masm=@var{dialect}
8400 Output asm instructions using selected @var{dialect}. Supported choices are
8401 @samp{intel} or @samp{att} (the default one).
8406 @opindex mno-ieee-fp
8407 Control whether or not the compiler uses IEEE floating point
8408 comparisons. These handle correctly the case where the result of a
8409 comparison is unordered.
8412 @opindex msoft-float
8413 Generate output containing library calls for floating point.
8414 @strong{Warning:} the requisite libraries are not part of GCC@.
8415 Normally the facilities of the machine's usual C compiler are used, but
8416 this can't be done directly in cross-compilation. You must make your
8417 own arrangements to provide suitable library functions for
8420 On machines where a function returns floating point results in the 80387
8421 register stack, some floating point opcodes may be emitted even if
8422 @option{-msoft-float} is used.
8424 @item -mno-fp-ret-in-387
8425 @opindex mno-fp-ret-in-387
8426 Do not use the FPU registers for return values of functions.
8428 The usual calling convention has functions return values of types
8429 @code{float} and @code{double} in an FPU register, even if there
8430 is no FPU@. The idea is that the operating system should emulate
8433 The option @option{-mno-fp-ret-in-387} causes such values to be returned
8434 in ordinary CPU registers instead.
8436 @item -mno-fancy-math-387
8437 @opindex mno-fancy-math-387
8438 Some 387 emulators do not support the @code{sin}, @code{cos} and
8439 @code{sqrt} instructions for the 387. Specify this option to avoid
8440 generating those instructions. This option is the default on FreeBSD,
8441 OpenBSD and NetBSD@. This option is overridden when @option{-march}
8442 indicates that the target cpu will always have an FPU and so the
8443 instruction will not need emulation. As of revision 2.6.1, these
8444 instructions are not generated unless you also use the
8445 @option{-funsafe-math-optimizations} switch.
8447 @item -malign-double
8448 @itemx -mno-align-double
8449 @opindex malign-double
8450 @opindex mno-align-double
8451 Control whether GCC aligns @code{double}, @code{long double}, and
8452 @code{long long} variables on a two word boundary or a one word
8453 boundary. Aligning @code{double} variables on a two word boundary will
8454 produce code that runs somewhat faster on a @samp{Pentium} at the
8455 expense of more memory.
8457 @strong{Warning:} if you use the @option{-malign-double} switch,
8458 structures containing the above types will be aligned differently than
8459 the published application binary interface specifications for the 386
8460 and will not be binary compatible with structures in code compiled
8461 without that switch.
8463 @item -m96bit-long-double
8464 @itemx -m128bit-long-double
8465 @opindex m96bit-long-double
8466 @opindex m128bit-long-double
8467 These switches control the size of @code{long double} type. The i386
8468 application binary interface specifies the size to be 96 bits,
8469 so @option{-m96bit-long-double} is the default in 32 bit mode.
8471 Modern architectures (Pentium and newer) would prefer @code{long double}
8472 to be aligned to an 8 or 16 byte boundary. In arrays or structures
8473 conforming to the ABI, this would not be possible. So specifying a
8474 @option{-m128bit-long-double} will align @code{long double}
8475 to a 16 byte boundary by padding the @code{long double} with an additional
8478 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
8479 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
8481 Notice that neither of these options enable any extra precision over the x87
8482 standard of 80 bits for a @code{long double}.
8484 @strong{Warning:} if you override the default value for your target ABI, the
8485 structures and arrays containing @code{long double} variables will change
8486 their size as well as function calling convention for function taking
8487 @code{long double} will be modified. Hence they will not be binary
8488 compatible with arrays or structures in code compiled without that switch.
8492 @itemx -mno-svr3-shlib
8493 @opindex msvr3-shlib
8494 @opindex mno-svr3-shlib
8495 Control whether GCC places uninitialized local variables into the
8496 @code{bss} or @code{data} segments. @option{-msvr3-shlib} places them
8497 into @code{bss}. These options are meaningful only on System V Release 3.
8501 Use a different function-calling convention, in which functions that
8502 take a fixed number of arguments return with the @code{ret} @var{num}
8503 instruction, which pops their arguments while returning. This saves one
8504 instruction in the caller since there is no need to pop the arguments
8507 You can specify that an individual function is called with this calling
8508 sequence with the function attribute @samp{stdcall}. You can also
8509 override the @option{-mrtd} option by using the function attribute
8510 @samp{cdecl}. @xref{Function Attributes}.
8512 @strong{Warning:} this calling convention is incompatible with the one
8513 normally used on Unix, so you cannot use it if you need to call
8514 libraries compiled with the Unix compiler.
8516 Also, you must provide function prototypes for all functions that
8517 take variable numbers of arguments (including @code{printf});
8518 otherwise incorrect code will be generated for calls to those
8521 In addition, seriously incorrect code will result if you call a
8522 function with too many arguments. (Normally, extra arguments are
8523 harmlessly ignored.)
8525 @item -mregparm=@var{num}
8527 Control how many registers are used to pass integer arguments. By
8528 default, no registers are used to pass arguments, and at most 3
8529 registers can be used. You can control this behavior for a specific
8530 function by using the function attribute @samp{regparm}.
8531 @xref{Function Attributes}.
8533 @strong{Warning:} if you use this switch, and
8534 @var{num} is nonzero, then you must build all modules with the same
8535 value, including any libraries. This includes the system libraries and
8538 @item -mpreferred-stack-boundary=@var{num}
8539 @opindex mpreferred-stack-boundary
8540 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
8541 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
8542 the default is 4 (16 bytes or 128 bits), except when optimizing for code
8543 size (@option{-Os}), in which case the default is the minimum correct
8544 alignment (4 bytes for x86, and 8 bytes for x86-64).
8546 On Pentium and PentiumPro, @code{double} and @code{long double} values
8547 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
8548 suffer significant run time performance penalties. On Pentium III, the
8549 Streaming SIMD Extension (SSE) data type @code{__m128} suffers similar
8550 penalties if it is not 16 byte aligned.
8552 To ensure proper alignment of this values on the stack, the stack boundary
8553 must be as aligned as that required by any value stored on the stack.
8554 Further, every function must be generated such that it keeps the stack
8555 aligned. Thus calling a function compiled with a higher preferred
8556 stack boundary from a function compiled with a lower preferred stack
8557 boundary will most likely misalign the stack. It is recommended that
8558 libraries that use callbacks always use the default setting.
8560 This extra alignment does consume extra stack space, and generally
8561 increases code size. Code that is sensitive to stack space usage, such
8562 as embedded systems and operating system kernels, may want to reduce the
8563 preferred alignment to @option{-mpreferred-stack-boundary=2}.
8581 These switches enable or disable the use of built-in functions that allow
8582 direct access to the MMX, SSE, SSE2, SSE3 and 3Dnow extensions of the
8585 @xref{X86 Built-in Functions}, for details of the functions enabled
8586 and disabled by these switches.
8588 To have SSE/SSE2 instructions generated automatically from floating-point
8589 code, see @option{-mfpmath=sse}.
8592 @itemx -mno-push-args
8594 @opindex mno-push-args
8595 Use PUSH operations to store outgoing parameters. This method is shorter
8596 and usually equally fast as method using SUB/MOV operations and is enabled
8597 by default. In some cases disabling it may improve performance because of
8598 improved scheduling and reduced dependencies.
8600 @item -maccumulate-outgoing-args
8601 @opindex maccumulate-outgoing-args
8602 If enabled, the maximum amount of space required for outgoing arguments will be
8603 computed in the function prologue. This is faster on most modern CPUs
8604 because of reduced dependencies, improved scheduling and reduced stack usage
8605 when preferred stack boundary is not equal to 2. The drawback is a notable
8606 increase in code size. This switch implies @option{-mno-push-args}.
8610 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
8611 on thread-safe exception handling must compile and link all code with the
8612 @option{-mthreads} option. When compiling, @option{-mthreads} defines
8613 @option{-D_MT}; when linking, it links in a special thread helper library
8614 @option{-lmingwthrd} which cleans up per thread exception handling data.
8616 @item -mno-align-stringops
8617 @opindex mno-align-stringops
8618 Do not align destination of inlined string operations. This switch reduces
8619 code size and improves performance in case the destination is already aligned,
8620 but GCC doesn't know about it.
8622 @item -minline-all-stringops
8623 @opindex minline-all-stringops
8624 By default GCC inlines string operations only when destination is known to be
8625 aligned at least to 4 byte boundary. This enables more inlining, increase code
8626 size, but may improve performance of code that depends on fast memcpy, strlen
8627 and memset for short lengths.
8629 @item -momit-leaf-frame-pointer
8630 @opindex momit-leaf-frame-pointer
8631 Don't keep the frame pointer in a register for leaf functions. This
8632 avoids the instructions to save, set up and restore frame pointers and
8633 makes an extra register available in leaf functions. The option
8634 @option{-fomit-frame-pointer} removes the frame pointer for all functions
8635 which might make debugging harder.
8637 @item -mtls-direct-seg-refs
8638 @itemx -mno-tls-direct-seg-refs
8639 @opindex mtls-direct-seg-refs
8640 Controls whether TLS variables may be accessed with offsets from the
8641 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
8642 or whether the thread base pointer must be added. Whether or not this
8643 is legal depends on the operating system, and whether it maps the
8644 segment to cover the entire TLS area.
8646 For systems that use GNU libc, the default is on.
8649 These @samp{-m} switches are supported in addition to the above
8650 on AMD x86-64 processors in 64-bit environments.
8657 Generate code for a 32-bit or 64-bit environment.
8658 The 32-bit environment sets int, long and pointer to 32 bits and
8659 generates code that runs on any i386 system.
8660 The 64-bit environment sets int to 32 bits and long and pointer
8661 to 64 bits and generates code for AMD's x86-64 architecture.
8664 @opindex no-red-zone
8665 Do not use a so called red zone for x86-64 code. The red zone is mandated
8666 by the x86-64 ABI, it is a 128-byte area beyond the location of the
8667 stack pointer that will not be modified by signal or interrupt handlers
8668 and therefore can be used for temporary data without adjusting the stack
8669 pointer. The flag @option{-mno-red-zone} disables this red zone.
8671 @item -mcmodel=small
8672 @opindex mcmodel=small
8673 Generate code for the small code model: the program and its symbols must
8674 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
8675 Programs can be statically or dynamically linked. This is the default
8678 @item -mcmodel=kernel
8679 @opindex mcmodel=kernel
8680 Generate code for the kernel code model. The kernel runs in the
8681 negative 2 GB of the address space.
8682 This model has to be used for Linux kernel code.
8684 @item -mcmodel=medium
8685 @opindex mcmodel=medium
8686 Generate code for the medium model: The program is linked in the lower 2
8687 GB of the address space but symbols can be located anywhere in the
8688 address space. Programs can be statically or dynamically linked, but
8689 building of shared libraries are not supported with the medium model.
8691 @item -mcmodel=large
8692 @opindex mcmodel=large
8693 Generate code for the large model: This model makes no assumptions
8694 about addresses and sizes of sections. Currently GCC does not implement
8699 @subsection IA-64 Options
8700 @cindex IA-64 Options
8702 These are the @samp{-m} options defined for the Intel IA-64 architecture.
8706 @opindex mbig-endian
8707 Generate code for a big endian target. This is the default for HP-UX@.
8709 @item -mlittle-endian
8710 @opindex mlittle-endian
8711 Generate code for a little endian target. This is the default for AIX5
8718 Generate (or don't) code for the GNU assembler. This is the default.
8719 @c Also, this is the default if the configure option @option{--with-gnu-as}
8726 Generate (or don't) code for the GNU linker. This is the default.
8727 @c Also, this is the default if the configure option @option{--with-gnu-ld}
8732 Generate code that does not use a global pointer register. The result
8733 is not position independent code, and violates the IA-64 ABI@.
8735 @item -mvolatile-asm-stop
8736 @itemx -mno-volatile-asm-stop
8737 @opindex mvolatile-asm-stop
8738 @opindex mno-volatile-asm-stop
8739 Generate (or don't) a stop bit immediately before and after volatile asm
8744 Generate code that works around Itanium B step errata.
8746 @item -mregister-names
8747 @itemx -mno-register-names
8748 @opindex mregister-names
8749 @opindex mno-register-names
8750 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
8751 the stacked registers. This may make assembler output more readable.
8757 Disable (or enable) optimizations that use the small data section. This may
8758 be useful for working around optimizer bugs.
8761 @opindex mconstant-gp
8762 Generate code that uses a single constant global pointer value. This is
8763 useful when compiling kernel code.
8767 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
8768 This is useful when compiling firmware code.
8770 @item -minline-float-divide-min-latency
8771 @opindex minline-float-divide-min-latency
8772 Generate code for inline divides of floating point values
8773 using the minimum latency algorithm.
8775 @item -minline-float-divide-max-throughput
8776 @opindex minline-float-divide-max-throughput
8777 Generate code for inline divides of floating point values
8778 using the maximum throughput algorithm.
8780 @item -minline-int-divide-min-latency
8781 @opindex minline-int-divide-min-latency
8782 Generate code for inline divides of integer values
8783 using the minimum latency algorithm.
8785 @item -minline-int-divide-max-throughput
8786 @opindex minline-int-divide-max-throughput
8787 Generate code for inline divides of integer values
8788 using the maximum throughput algorithm.
8790 @item -mno-dwarf2-asm
8792 @opindex mno-dwarf2-asm
8793 @opindex mdwarf2-asm
8794 Don't (or do) generate assembler code for the DWARF2 line number debugging
8795 info. This may be useful when not using the GNU assembler.
8797 @item -mfixed-range=@var{register-range}
8798 @opindex mfixed-range
8799 Generate code treating the given register range as fixed registers.
8800 A fixed register is one that the register allocator can not use. This is
8801 useful when compiling kernel code. A register range is specified as
8802 two registers separated by a dash. Multiple register ranges can be
8803 specified separated by a comma.
8805 @item -mearly-stop-bits
8806 @itemx -mno-early-stop-bits
8807 @opindex mearly-stop-bits
8808 @opindex mno-early-stop-bits
8809 Allow stop bits to be placed earlier than immediately preceding the
8810 instruction that triggered the stop bit. This can improve instruction
8811 scheduling, but does not always do so.
8814 @node M32R/D Options
8815 @subsection M32R/D Options
8816 @cindex M32R/D options
8818 These @option{-m} options are defined for Renesas M32R/D architectures:
8823 Generate code for the M32R/2@.
8827 Generate code for the M32R/X@.
8831 Generate code for the M32R@. This is the default.
8834 @opindex mmodel=small
8835 Assume all objects live in the lower 16MB of memory (so that their addresses
8836 can be loaded with the @code{ld24} instruction), and assume all subroutines
8837 are reachable with the @code{bl} instruction.
8838 This is the default.
8840 The addressability of a particular object can be set with the
8841 @code{model} attribute.
8843 @item -mmodel=medium
8844 @opindex mmodel=medium
8845 Assume objects may be anywhere in the 32-bit address space (the compiler
8846 will generate @code{seth/add3} instructions to load their addresses), and
8847 assume all subroutines are reachable with the @code{bl} instruction.
8850 @opindex mmodel=large
8851 Assume objects may be anywhere in the 32-bit address space (the compiler
8852 will generate @code{seth/add3} instructions to load their addresses), and
8853 assume subroutines may not be reachable with the @code{bl} instruction
8854 (the compiler will generate the much slower @code{seth/add3/jl}
8855 instruction sequence).
8858 @opindex msdata=none
8859 Disable use of the small data area. Variables will be put into
8860 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
8861 @code{section} attribute has been specified).
8862 This is the default.
8864 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
8865 Objects may be explicitly put in the small data area with the
8866 @code{section} attribute using one of these sections.
8869 @opindex msdata=sdata
8870 Put small global and static data in the small data area, but do not
8871 generate special code to reference them.
8875 Put small global and static data in the small data area, and generate
8876 special instructions to reference them.
8880 @cindex smaller data references
8881 Put global and static objects less than or equal to @var{num} bytes
8882 into the small data or bss sections instead of the normal data or bss
8883 sections. The default value of @var{num} is 8.
8884 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
8885 for this option to have any effect.
8887 All modules should be compiled with the same @option{-G @var{num}} value.
8888 Compiling with different values of @var{num} may or may not work; if it
8889 doesn't the linker will give an error message---incorrect code will not be
8894 Makes the M32R specific code in the compiler display some statistics
8895 that might help in debugging programs.
8898 @opindex malign-loops
8899 Align all loops to a 32-byte boundary.
8901 @item -mno-align-loops
8902 @opindex mno-align-loops
8903 Do not enforce a 32-byte alignment for loops. This is the default.
8905 @item -missue-rate=@var{number}
8906 @opindex missue-rate=@var{number}
8907 Issue @var{number} instructions per cycle. @var{number} can only be 1
8910 @item -mbranch-cost=@var{number}
8911 @opindex mbranch-cost=@var{number}
8912 @var{number} can only be 1 or 2. If it is 1 then branches will be
8913 preferred over conditional code, if it is 2, then the opposite will
8916 @item -mflush-trap=@var{number}
8917 @opindex mflush-trap=@var{number}
8918 Specifies the trap number to use to flush the cache. The default is
8919 12. Valid numbers are between 0 and 15 inclusive.
8921 @item -mno-flush-trap
8922 @opindex mno-flush-trap
8923 Specifies that the cache cannot be flushed by using a trap.
8925 @item -mflush-func=@var{name}
8926 @opindex mflush-func=@var{name}
8927 Specifies the name of the operating system function to call to flush
8928 the cache. The default is @emph{_flush_cache}, but a function call
8929 will only be used if a trap is not available.
8931 @item -mno-flush-func
8932 @opindex mno-flush-func
8933 Indicates that there is no OS function for flushing the cache.
8937 @node M680x0 Options
8938 @subsection M680x0 Options
8939 @cindex M680x0 options
8941 These are the @samp{-m} options defined for the 68000 series. The default
8942 values for these options depends on which style of 68000 was selected when
8943 the compiler was configured; the defaults for the most common choices are
8951 Generate output for a 68000. This is the default
8952 when the compiler is configured for 68000-based systems.
8954 Use this option for microcontrollers with a 68000 or EC000 core,
8955 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
8961 Generate output for a 68020. This is the default
8962 when the compiler is configured for 68020-based systems.
8966 Generate output containing 68881 instructions for floating point.
8967 This is the default for most 68020 systems unless @option{--nfp} was
8968 specified when the compiler was configured.
8972 Generate output for a 68030. This is the default when the compiler is
8973 configured for 68030-based systems.
8977 Generate output for a 68040. This is the default when the compiler is
8978 configured for 68040-based systems.
8980 This option inhibits the use of 68881/68882 instructions that have to be
8981 emulated by software on the 68040. Use this option if your 68040 does not
8982 have code to emulate those instructions.
8986 Generate output for a 68060. This is the default when the compiler is
8987 configured for 68060-based systems.
8989 This option inhibits the use of 68020 and 68881/68882 instructions that
8990 have to be emulated by software on the 68060. Use this option if your 68060
8991 does not have code to emulate those instructions.
8995 Generate output for a CPU32. This is the default
8996 when the compiler is configured for CPU32-based systems.
8998 Use this option for microcontrollers with a
8999 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
9000 68336, 68340, 68341, 68349 and 68360.
9004 Generate output for a 520X ``coldfire'' family cpu. This is the default
9005 when the compiler is configured for 520X-based systems.
9007 Use this option for microcontroller with a 5200 core, including
9008 the MCF5202, MCF5203, MCF5204 and MCF5202.
9013 Generate output for a 68040, without using any of the new instructions.
9014 This results in code which can run relatively efficiently on either a
9015 68020/68881 or a 68030 or a 68040. The generated code does use the
9016 68881 instructions that are emulated on the 68040.
9020 Generate output for a 68060, without using any of the new instructions.
9021 This results in code which can run relatively efficiently on either a
9022 68020/68881 or a 68030 or a 68040. The generated code does use the
9023 68881 instructions that are emulated on the 68060.
9026 @opindex msoft-float
9027 Generate output containing library calls for floating point.
9028 @strong{Warning:} the requisite libraries are not available for all m68k
9029 targets. Normally the facilities of the machine's usual C compiler are
9030 used, but this can't be done directly in cross-compilation. You must
9031 make your own arrangements to provide suitable library functions for
9032 cross-compilation. The embedded targets @samp{m68k-*-aout} and
9033 @samp{m68k-*-coff} do provide software floating point support.
9037 Consider type @code{int} to be 16 bits wide, like @code{short int}.
9038 Additionally, parameters passed on the stack are also aligned to a
9039 16-bit boundary even on targets whose API mandates promotion to 32-bit.
9042 @opindex mnobitfield
9043 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
9044 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
9048 Do use the bit-field instructions. The @option{-m68020} option implies
9049 @option{-mbitfield}. This is the default if you use a configuration
9050 designed for a 68020.
9054 Use a different function-calling convention, in which functions
9055 that take a fixed number of arguments return with the @code{rtd}
9056 instruction, which pops their arguments while returning. This
9057 saves one instruction in the caller since there is no need to pop
9058 the arguments there.
9060 This calling convention is incompatible with the one normally
9061 used on Unix, so you cannot use it if you need to call libraries
9062 compiled with the Unix compiler.
9064 Also, you must provide function prototypes for all functions that
9065 take variable numbers of arguments (including @code{printf});
9066 otherwise incorrect code will be generated for calls to those
9069 In addition, seriously incorrect code will result if you call a
9070 function with too many arguments. (Normally, extra arguments are
9071 harmlessly ignored.)
9073 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
9074 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
9077 @itemx -mno-align-int
9079 @opindex mno-align-int
9080 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
9081 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
9082 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
9083 Aligning variables on 32-bit boundaries produces code that runs somewhat
9084 faster on processors with 32-bit busses at the expense of more memory.
9086 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
9087 align structures containing the above types differently than
9088 most published application binary interface specifications for the m68k.
9092 Use the pc-relative addressing mode of the 68000 directly, instead of
9093 using a global offset table. At present, this option implies @option{-fpic},
9094 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
9095 not presently supported with @option{-mpcrel}, though this could be supported for
9096 68020 and higher processors.
9098 @item -mno-strict-align
9099 @itemx -mstrict-align
9100 @opindex mno-strict-align
9101 @opindex mstrict-align
9102 Do not (do) assume that unaligned memory references will be handled by
9106 Generate code that allows the data segment to be located in a different
9107 area of memory from the text segment. This allows for execute in place in
9108 an environment without virtual memory management. This option implies -fPIC.
9111 Generate code that assumes that the data segment follows the text segment.
9112 This is the default.
9114 @item -mid-shared-library
9115 Generate code that supports shared libraries via the library ID method.
9116 This allows for execute in place and shared libraries in an environment
9117 without virtual memory management. This option implies -fPIC.
9119 @item -mno-id-shared-library
9120 Generate code that doesn't assume ID based shared libraries are being used.
9121 This is the default.
9123 @item -mshared-library-id=n
9124 Specified the identification number of the ID based shared library being
9125 compiled. Specifying a value of 0 will generate more compact code, specifying
9126 other values will force the allocation of that number to the current
9127 library but is no more space or time efficient than omitting this option.
9131 @node M68hc1x Options
9132 @subsection M68hc1x Options
9133 @cindex M68hc1x options
9135 These are the @samp{-m} options defined for the 68hc11 and 68hc12
9136 microcontrollers. The default values for these options depends on
9137 which style of microcontroller was selected when the compiler was configured;
9138 the defaults for the most common choices are given below.
9145 Generate output for a 68HC11. This is the default
9146 when the compiler is configured for 68HC11-based systems.
9152 Generate output for a 68HC12. This is the default
9153 when the compiler is configured for 68HC12-based systems.
9159 Generate output for a 68HCS12.
9162 @opindex mauto-incdec
9163 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
9170 Enable the use of 68HC12 min and max instructions.
9173 @itemx -mno-long-calls
9174 @opindex mlong-calls
9175 @opindex mno-long-calls
9176 Treat all calls as being far away (near). If calls are assumed to be
9177 far away, the compiler will use the @code{call} instruction to
9178 call a function and the @code{rtc} instruction for returning.
9182 Consider type @code{int} to be 16 bits wide, like @code{short int}.
9184 @item -msoft-reg-count=@var{count}
9185 @opindex msoft-reg-count
9186 Specify the number of pseudo-soft registers which are used for the
9187 code generation. The maximum number is 32. Using more pseudo-soft
9188 register may or may not result in better code depending on the program.
9189 The default is 4 for 68HC11 and 2 for 68HC12.
9194 @subsection MCore Options
9195 @cindex MCore options
9197 These are the @samp{-m} options defined for the Motorola M*Core
9205 @opindex mno-hardlit
9206 Inline constants into the code stream if it can be done in two
9207 instructions or less.
9213 Use the divide instruction. (Enabled by default).
9215 @item -mrelax-immediate
9216 @itemx -mno-relax-immediate
9217 @opindex mrelax-immediate
9218 @opindex mno-relax-immediate
9219 Allow arbitrary sized immediates in bit operations.
9221 @item -mwide-bitfields
9222 @itemx -mno-wide-bitfields
9223 @opindex mwide-bitfields
9224 @opindex mno-wide-bitfields
9225 Always treat bit-fields as int-sized.
9227 @item -m4byte-functions
9228 @itemx -mno-4byte-functions
9229 @opindex m4byte-functions
9230 @opindex mno-4byte-functions
9231 Force all functions to be aligned to a four byte boundary.
9233 @item -mcallgraph-data
9234 @itemx -mno-callgraph-data
9235 @opindex mcallgraph-data
9236 @opindex mno-callgraph-data
9237 Emit callgraph information.
9240 @itemx -mno-slow-bytes
9241 @opindex mslow-bytes
9242 @opindex mno-slow-bytes
9243 Prefer word access when reading byte quantities.
9245 @item -mlittle-endian
9247 @opindex mlittle-endian
9248 @opindex mbig-endian
9249 Generate code for a little endian target.
9255 Generate code for the 210 processor.
9259 @subsection MIPS Options
9260 @cindex MIPS options
9266 Generate big-endian code.
9270 Generate little-endian code. This is the default for @samp{mips*el-*-*}
9273 @item -march=@var{arch}
9275 Generate code that will run on @var{arch}, which can be the name of a
9276 generic MIPS ISA, or the name of a particular processor.
9278 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
9279 @samp{mips32}, @samp{mips32r2}, and @samp{mips64}.
9280 The processor names are:
9281 @samp{4kc}, @samp{4kp}, @samp{5kc}, @samp{20kc},
9283 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
9284 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000}, @samp{rm7000},
9288 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4130}, @samp{vr4300},
9289 @samp{vr5000}, @samp{vr5400} and @samp{vr5500}.
9290 The special value @samp{from-abi} selects the
9291 most compatible architecture for the selected ABI (that is,
9292 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
9294 In processor names, a final @samp{000} can be abbreviated as @samp{k}
9295 (for example, @samp{-march=r2k}). Prefixes are optional, and
9296 @samp{vr} may be written @samp{r}.
9298 GCC defines two macros based on the value of this option. The first
9299 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
9300 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
9301 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
9302 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
9303 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
9305 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
9306 above. In other words, it will have the full prefix and will not
9307 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
9308 the macro names the resolved architecture (either @samp{"mips1"} or
9309 @samp{"mips3"}). It names the default architecture when no
9310 @option{-march} option is given.
9312 @item -mtune=@var{arch}
9314 Optimize for @var{arch}. Among other things, this option controls
9315 the way instructions are scheduled, and the perceived cost of arithmetic
9316 operations. The list of @var{arch} values is the same as for
9319 When this option is not used, GCC will optimize for the processor
9320 specified by @option{-march}. By using @option{-march} and
9321 @option{-mtune} together, it is possible to generate code that will
9322 run on a family of processors, but optimize the code for one
9323 particular member of that family.
9325 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
9326 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
9327 @samp{-march} ones described above.
9331 Equivalent to @samp{-march=mips1}.
9335 Equivalent to @samp{-march=mips2}.
9339 Equivalent to @samp{-march=mips3}.
9343 Equivalent to @samp{-march=mips4}.
9347 Equivalent to @samp{-march=mips32}.
9351 Equivalent to @samp{-march=mips32r2}.
9355 Equivalent to @samp{-march=mips64}.
9361 Use (do not use) the MIPS16 ISA.
9373 Generate code for the given ABI@.
9375 Note that the EABI has a 32-bit and a 64-bit variant. GCC normally
9376 generates 64-bit code when you select a 64-bit architecture, but you
9377 can use @option{-mgp32} to get 32-bit code instead.
9379 For information about the O64 ABI, see
9380 @w{@uref{http://gcc.gnu.org/projects/mipso64-abi.html}}.
9383 @itemx -mno-abicalls
9385 @opindex mno-abicalls
9386 Generate (do not generate) SVR4-style position-independent code.
9387 @option{-mabicalls} is the default for SVR4-based systems.
9393 Lift (do not lift) the usual restrictions on the size of the global
9396 GCC normally uses a single instruction to load values from the GOT.
9397 While this is relatively efficient, it will only work if the GOT
9398 is smaller than about 64k. Anything larger will cause the linker
9399 to report an error such as:
9401 @cindex relocation truncated to fit (MIPS)
9403 relocation truncated to fit: R_MIPS_GOT16 foobar
9406 If this happens, you should recompile your code with @option{-mxgot}.
9407 It should then work with very large GOTs, although it will also be
9408 less efficient, since it will take three instructions to fetch the
9409 value of a global symbol.
9411 Note that some linkers can create multiple GOTs. If you have such a
9412 linker, you should only need to use @option{-mxgot} when a single object
9413 file accesses more than 64k's worth of GOT entries. Very few do.
9415 These options have no effect unless GCC is generating position
9420 Assume that general-purpose registers are 32 bits wide.
9424 Assume that general-purpose registers are 64 bits wide.
9428 Assume that floating-point registers are 32 bits wide.
9432 Assume that floating-point registers are 64 bits wide.
9435 @opindex mhard-float
9436 Use floating-point coprocessor instructions.
9439 @opindex msoft-float
9440 Do not use floating-point coprocessor instructions. Implement
9441 floating-point calculations using library calls instead.
9443 @item -msingle-float
9444 @opindex msingle-float
9445 Assume that the floating-point coprocessor only supports single-precision
9448 @itemx -mdouble-float
9449 @opindex mdouble-float
9450 Assume that the floating-point coprocessor supports double-precision
9451 operations. This is the default.
9453 @itemx -mpaired-single
9454 @itemx -mno-paired-single
9455 @opindex mpaired-single
9456 @opindex mno-paired-single
9457 Use (do not use) the paired single instructions.
9463 Use (do not use) the MIPS-3D ASE. The option @option{-mips3d} implies
9464 @option{-mpaired-single}.
9468 Force @code{int} and @code{long} types to be 64 bits wide. See
9469 @option{-mlong32} for an explanation of the default and the way
9470 that the pointer size is determined.
9474 Force @code{long} types to be 64 bits wide. See @option{-mlong32} for
9475 an explanation of the default and the way that the pointer size is
9480 Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
9482 The default size of @code{int}s, @code{long}s and pointers depends on
9483 the ABI@. All the supported ABIs use 32-bit @code{int}s. The n64 ABI
9484 uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
9485 32-bit @code{long}s. Pointers are the same size as @code{long}s,
9486 or the same size as integer registers, whichever is smaller.
9490 @cindex smaller data references (MIPS)
9491 @cindex gp-relative references (MIPS)
9492 Put global and static items less than or equal to @var{num} bytes into
9493 the small data or bss section instead of the normal data or bss section.
9494 This allows the data to be accessed using a single instruction.
9496 All modules should be compiled with the same @option{-G @var{num}}
9499 @item -membedded-data
9500 @itemx -mno-embedded-data
9501 @opindex membedded-data
9502 @opindex mno-embedded-data
9503 Allocate variables to the read-only data section first if possible, then
9504 next in the small data section if possible, otherwise in data. This gives
9505 slightly slower code than the default, but reduces the amount of RAM required
9506 when executing, and thus may be preferred for some embedded systems.
9508 @item -muninit-const-in-rodata
9509 @itemx -mno-uninit-const-in-rodata
9510 @opindex muninit-const-in-rodata
9511 @opindex mno-uninit-const-in-rodata
9512 Put uninitialized @code{const} variables in the read-only data section.
9513 This option is only meaningful in conjunction with @option{-membedded-data}.
9515 @item -msplit-addresses
9516 @itemx -mno-split-addresses
9517 @opindex msplit-addresses
9518 @opindex mno-split-addresses
9519 Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
9520 relocation operators. This option has been superceded by
9521 @option{-mexplicit-relocs} but is retained for backwards compatibility.
9523 @item -mexplicit-relocs
9524 @itemx -mno-explicit-relocs
9525 @opindex mexplicit-relocs
9526 @opindex mno-explicit-relocs
9527 Use (do not use) assembler relocation operators when dealing with symbolic
9528 addresses. The alternative, selected by @option{-mno-explicit-relocs},
9529 is to use assembler macros instead.
9531 @option{-mexplicit-relocs} is the default if GCC was configured
9532 to use an assembler that supports relocation operators.
9534 @item -mcheck-zero-division
9535 @itemx -mno-check-zero-division
9536 @opindex mcheck-zero-division
9537 @opindex mno-check-zero-division
9538 Trap (do not trap) on integer division by zero. The default is
9539 @option{-mcheck-zero-division}.
9541 @item -mdivide-traps
9542 @itemx -mdivide-breaks
9543 @opindex mdivide-traps
9544 @opindex mdivide-breaks
9545 MIPS systems check for division by zero by generating either a
9546 conditional trap or a break instruction. Using traps results in
9547 smaller code, but is only supported on MIPS II and later. Also, some
9548 versions of the Linux kernel have a bug that prevents trap from
9549 generating the proper signal (SIGFPE). Use @option{-mdivide-traps} to
9550 allow conditional traps on architectures that support them and
9551 @option{-mdivide-breaks} to force the use of breaks.
9553 The default is usually @option{-mdivide-traps}, but this can be
9554 overridden at configure time using @option{--with-divide=breaks}.
9555 Divide-by-zero checks can be completely disabled using
9556 @option{-mno-check-zero-division}.
9562 Force (do not force) the use of @code{memcpy()} for non-trivial block
9563 moves. The default is @option{-mno-memcpy}, which allows GCC to inline
9564 most constant-sized copies.
9567 @itemx -mno-long-calls
9568 @opindex mlong-calls
9569 @opindex mno-long-calls
9570 Disable (do not disable) use of the @code{jal} instruction. Calling
9571 functions using @code{jal} is more efficient but requires the caller
9572 and callee to be in the same 256 megabyte segment.
9574 This option has no effect on abicalls code. The default is
9575 @option{-mno-long-calls}.
9581 Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
9582 instructions, as provided by the R4650 ISA.
9585 @itemx -mno-fused-madd
9586 @opindex mfused-madd
9587 @opindex mno-fused-madd
9588 Enable (disable) use of the floating point multiply-accumulate
9589 instructions, when they are available. The default is
9590 @option{-mfused-madd}.
9592 When multiply-accumulate instructions are used, the intermediate
9593 product is calculated to infinite precision and is not subject to
9594 the FCSR Flush to Zero bit. This may be undesirable in some
9599 Tell the MIPS assembler to not run its preprocessor over user
9600 assembler files (with a @samp{.s} suffix) when assembling them.
9603 @itemx -mno-fix-r4000
9605 @opindex mno-fix-r4000
9606 Work around certain R4000 CPU errata:
9609 A double-word or a variable shift may give an incorrect result if executed
9610 immediately after starting an integer division.
9612 A double-word or a variable shift may give an incorrect result if executed
9613 while an integer multiplication is in progress.
9615 An integer division may give an incorrect result if started in a delay slot
9616 of a taken branch or a jump.
9620 @itemx -mno-fix-r4400
9622 @opindex mno-fix-r4400
9623 Work around certain R4400 CPU errata:
9626 A double-word or a variable shift may give an incorrect result if executed
9627 immediately after starting an integer division.
9631 @itemx -mno-fix-vr4120
9632 @opindex mfix-vr4120
9633 Work around certain VR4120 errata:
9636 @code{dmultu} does not always produce the correct result.
9638 @code{div} and @code{ddiv} do not always produce the correct result if one
9639 of the operands is negative.
9641 The workarounds for the division errata rely on special functions in
9642 @file{libgcc.a}. At present, these functions are only provided by
9643 the @code{mips64vr*-elf} configurations.
9645 Other VR4120 errata require a nop to be inserted between certain pairs of
9646 instructions. These errata are handled by the assembler, not by GCC itself.
9651 Work around certain SB-1 CPU core errata.
9652 (This flag currently works around the SB-1 revision 2
9653 ``F1'' and ``F2'' floating point errata.)
9655 @item -mflush-func=@var{func}
9656 @itemx -mno-flush-func
9657 @opindex mflush-func
9658 Specifies the function to call to flush the I and D caches, or to not
9659 call any such function. If called, the function must take the same
9660 arguments as the common @code{_flush_func()}, that is, the address of the
9661 memory range for which the cache is being flushed, the size of the
9662 memory range, and the number 3 (to flush both caches). The default
9663 depends on the target GCC was configured for, but commonly is either
9664 @samp{_flush_func} or @samp{__cpu_flush}.
9666 @item -mbranch-likely
9667 @itemx -mno-branch-likely
9668 @opindex mbranch-likely
9669 @opindex mno-branch-likely
9670 Enable or disable use of Branch Likely instructions, regardless of the
9671 default for the selected architecture. By default, Branch Likely
9672 instructions may be generated if they are supported by the selected
9673 architecture. An exception is for the MIPS32 and MIPS64 architectures
9674 and processors which implement those architectures; for those, Branch
9675 Likely instructions will not be generated by default because the MIPS32
9676 and MIPS64 architectures specifically deprecate their use.
9678 @item -mfp-exceptions
9679 @itemx -mno-fp-exceptions
9680 @opindex mfp-exceptions
9681 Specifies whether FP exceptions are enabled. This affects how we schedule
9682 FP instructions for some processors. The default is that FP exceptions are
9685 For instance, on the SB-1, if FP exceptions are disabled, and we are emitting
9686 64-bit code, then we can use both FP pipes. Otherwise, we can only use one
9689 @item -mvr4130-align
9690 @itemx -mno-vr4130-align
9691 @opindex mvr4130-align
9692 The VR4130 pipeline is two-way superscalar, but can only issue two
9693 instructions together if the first one is 8-byte aligned. When this
9694 option is enabled, GCC will align pairs of instructions that it
9695 thinks should execute in parallel.
9697 This option only has an effect when optimizing for the VR4130.
9698 It normally makes code faster, but at the expense of making it bigger.
9699 It is enabled by default at optimization level @option{-O3}.
9703 @subsection MMIX Options
9704 @cindex MMIX Options
9706 These options are defined for the MMIX:
9710 @itemx -mno-libfuncs
9712 @opindex mno-libfuncs
9713 Specify that intrinsic library functions are being compiled, passing all
9714 values in registers, no matter the size.
9719 @opindex mno-epsilon
9720 Generate floating-point comparison instructions that compare with respect
9721 to the @code{rE} epsilon register.
9723 @item -mabi=mmixware
9725 @opindex mabi-mmixware
9727 Generate code that passes function parameters and return values that (in
9728 the called function) are seen as registers @code{$0} and up, as opposed to
9729 the GNU ABI which uses global registers @code{$231} and up.
9732 @itemx -mno-zero-extend
9733 @opindex mzero-extend
9734 @opindex mno-zero-extend
9735 When reading data from memory in sizes shorter than 64 bits, use (do not
9736 use) zero-extending load instructions by default, rather than
9737 sign-extending ones.
9740 @itemx -mno-knuthdiv
9742 @opindex mno-knuthdiv
9743 Make the result of a division yielding a remainder have the same sign as
9744 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
9745 remainder follows the sign of the dividend. Both methods are
9746 arithmetically valid, the latter being almost exclusively used.
9748 @item -mtoplevel-symbols
9749 @itemx -mno-toplevel-symbols
9750 @opindex mtoplevel-symbols
9751 @opindex mno-toplevel-symbols
9752 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
9753 code can be used with the @code{PREFIX} assembly directive.
9757 Generate an executable in the ELF format, rather than the default
9758 @samp{mmo} format used by the @command{mmix} simulator.
9760 @item -mbranch-predict
9761 @itemx -mno-branch-predict
9762 @opindex mbranch-predict
9763 @opindex mno-branch-predict
9764 Use (do not use) the probable-branch instructions, when static branch
9765 prediction indicates a probable branch.
9767 @item -mbase-addresses
9768 @itemx -mno-base-addresses
9769 @opindex mbase-addresses
9770 @opindex mno-base-addresses
9771 Generate (do not generate) code that uses @emph{base addresses}. Using a
9772 base address automatically generates a request (handled by the assembler
9773 and the linker) for a constant to be set up in a global register. The
9774 register is used for one or more base address requests within the range 0
9775 to 255 from the value held in the register. The generally leads to short
9776 and fast code, but the number of different data items that can be
9777 addressed is limited. This means that a program that uses lots of static
9778 data may require @option{-mno-base-addresses}.
9781 @itemx -mno-single-exit
9782 @opindex msingle-exit
9783 @opindex mno-single-exit
9784 Force (do not force) generated code to have a single exit point in each
9788 @node MN10300 Options
9789 @subsection MN10300 Options
9790 @cindex MN10300 options
9792 These @option{-m} options are defined for Matsushita MN10300 architectures:
9797 Generate code to avoid bugs in the multiply instructions for the MN10300
9798 processors. This is the default.
9801 @opindex mno-mult-bug
9802 Do not generate code to avoid bugs in the multiply instructions for the
9807 Generate code which uses features specific to the AM33 processor.
9811 Do not generate code which uses features specific to the AM33 processor. This
9816 Do not link in the C run-time initialization object file.
9820 Indicate to the linker that it should perform a relaxation optimization pass
9821 to shorten branches, calls and absolute memory addresses. This option only
9822 has an effect when used on the command line for the final link step.
9824 This option makes symbolic debugging impossible.
9828 @subsection NS32K Options
9829 @cindex NS32K options
9831 These are the @samp{-m} options defined for the 32000 series. The default
9832 values for these options depends on which style of 32000 was selected when
9833 the compiler was configured; the defaults for the most common choices are
9841 Generate output for a 32032. This is the default
9842 when the compiler is configured for 32032 and 32016 based systems.
9848 Generate output for a 32332. This is the default
9849 when the compiler is configured for 32332-based systems.
9855 Generate output for a 32532. This is the default
9856 when the compiler is configured for 32532-based systems.
9860 Generate output containing 32081 instructions for floating point.
9861 This is the default for all systems.
9865 Generate output containing 32381 instructions for floating point. This
9866 also implies @option{-m32081}. The 32381 is only compatible with the 32332
9867 and 32532 cpus. This is the default for the pc532-netbsd configuration.
9871 Try and generate multiply-add floating point instructions @code{polyF}
9872 and @code{dotF}. This option is only available if the @option{-m32381}
9873 option is in effect. Using these instructions requires changes to
9874 register allocation which generally has a negative impact on
9875 performance. This option should only be enabled when compiling code
9876 particularly likely to make heavy use of multiply-add instructions.
9879 @opindex mnomulti-add
9880 Do not try and generate multiply-add floating point instructions
9881 @code{polyF} and @code{dotF}. This is the default on all platforms.
9884 @opindex msoft-float
9885 Generate output containing library calls for floating point.
9886 @strong{Warning:} the requisite libraries may not be available.
9888 @item -mieee-compare
9889 @itemx -mno-ieee-compare
9890 @opindex mieee-compare
9891 @opindex mno-ieee-compare
9892 Control whether or not the compiler uses IEEE floating point
9893 comparisons. These handle correctly the case where the result of a
9894 comparison is unordered.
9895 @strong{Warning:} the requisite kernel support may not be available.
9898 @opindex mnobitfield
9899 Do not use the bit-field instructions. On some machines it is faster to
9900 use shifting and masking operations. This is the default for the pc532.
9904 Do use the bit-field instructions. This is the default for all platforms
9909 Use a different function-calling convention, in which functions
9910 that take a fixed number of arguments return pop their
9911 arguments on return with the @code{ret} instruction.
9913 This calling convention is incompatible with the one normally
9914 used on Unix, so you cannot use it if you need to call libraries
9915 compiled with the Unix compiler.
9917 Also, you must provide function prototypes for all functions that
9918 take variable numbers of arguments (including @code{printf});
9919 otherwise incorrect code will be generated for calls to those
9922 In addition, seriously incorrect code will result if you call a
9923 function with too many arguments. (Normally, extra arguments are
9924 harmlessly ignored.)
9926 This option takes its name from the 680x0 @code{rtd} instruction.
9931 Use a different function-calling convention where the first two arguments
9932 are passed in registers.
9934 This calling convention is incompatible with the one normally
9935 used on Unix, so you cannot use it if you need to call libraries
9936 compiled with the Unix compiler.
9939 @opindex mnoregparam
9940 Do not pass any arguments in registers. This is the default for all
9945 It is OK to use the sb as an index register which is always loaded with
9946 zero. This is the default for the pc532-netbsd target.
9950 The sb register is not available for use or has not been initialized to
9951 zero by the run time system. This is the default for all targets except
9952 the pc532-netbsd. It is also implied whenever @option{-mhimem} or
9953 @option{-fpic} is set.
9957 Many ns32000 series addressing modes use displacements of up to 512MB@.
9958 If an address is above 512MB then displacements from zero can not be used.
9959 This option causes code to be generated which can be loaded above 512MB@.
9960 This may be useful for operating systems or ROM code.
9964 Assume code will be loaded in the first 512MB of virtual address space.
9965 This is the default for all platforms.
9969 @node PDP-11 Options
9970 @subsection PDP-11 Options
9971 @cindex PDP-11 Options
9973 These options are defined for the PDP-11:
9978 Use hardware FPP floating point. This is the default. (FIS floating
9979 point on the PDP-11/40 is not supported.)
9982 @opindex msoft-float
9983 Do not use hardware floating point.
9987 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
9991 Return floating-point results in memory. This is the default.
9995 Generate code for a PDP-11/40.
9999 Generate code for a PDP-11/45. This is the default.
10003 Generate code for a PDP-11/10.
10005 @item -mbcopy-builtin
10006 @opindex bcopy-builtin
10007 Use inline @code{movmemhi} patterns for copying memory. This is the
10012 Do not use inline @code{movmemhi} patterns for copying memory.
10018 Use 16-bit @code{int}. This is the default.
10024 Use 32-bit @code{int}.
10027 @itemx -mno-float32
10029 @opindex mno-float32
10030 Use 64-bit @code{float}. This is the default.
10033 @itemx -mno-float64
10035 @opindex mno-float64
10036 Use 32-bit @code{float}.
10040 Use @code{abshi2} pattern. This is the default.
10044 Do not use @code{abshi2} pattern.
10046 @item -mbranch-expensive
10047 @opindex mbranch-expensive
10048 Pretend that branches are expensive. This is for experimenting with
10049 code generation only.
10051 @item -mbranch-cheap
10052 @opindex mbranch-cheap
10053 Do not pretend that branches are expensive. This is the default.
10057 Generate code for a system with split I&D.
10061 Generate code for a system without split I&D. This is the default.
10065 Use Unix assembler syntax. This is the default when configured for
10066 @samp{pdp11-*-bsd}.
10070 Use DEC assembler syntax. This is the default when configured for any
10071 PDP-11 target other than @samp{pdp11-*-bsd}.
10074 @node PowerPC Options
10075 @subsection PowerPC Options
10076 @cindex PowerPC options
10078 These are listed under @xref{RS/6000 and PowerPC Options}.
10080 @node RS/6000 and PowerPC Options
10081 @subsection IBM RS/6000 and PowerPC Options
10082 @cindex RS/6000 and PowerPC Options
10083 @cindex IBM RS/6000 and PowerPC Options
10085 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
10092 @itemx -mno-powerpc
10093 @itemx -mpowerpc-gpopt
10094 @itemx -mno-powerpc-gpopt
10095 @itemx -mpowerpc-gfxopt
10096 @itemx -mno-powerpc-gfxopt
10098 @itemx -mno-powerpc64
10102 @opindex mno-power2
10104 @opindex mno-powerpc
10105 @opindex mpowerpc-gpopt
10106 @opindex mno-powerpc-gpopt
10107 @opindex mpowerpc-gfxopt
10108 @opindex mno-powerpc-gfxopt
10109 @opindex mpowerpc64
10110 @opindex mno-powerpc64
10111 GCC supports two related instruction set architectures for the
10112 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
10113 instructions supported by the @samp{rios} chip set used in the original
10114 RS/6000 systems and the @dfn{PowerPC} instruction set is the
10115 architecture of the Motorola MPC5xx, MPC6xx, MPC8xx microprocessors, and
10116 the IBM 4xx microprocessors.
10118 Neither architecture is a subset of the other. However there is a
10119 large common subset of instructions supported by both. An MQ
10120 register is included in processors supporting the POWER architecture.
10122 You use these options to specify which instructions are available on the
10123 processor you are using. The default value of these options is
10124 determined when configuring GCC@. Specifying the
10125 @option{-mcpu=@var{cpu_type}} overrides the specification of these
10126 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
10127 rather than the options listed above.
10129 The @option{-mpower} option allows GCC to generate instructions that
10130 are found only in the POWER architecture and to use the MQ register.
10131 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
10132 to generate instructions that are present in the POWER2 architecture but
10133 not the original POWER architecture.
10135 The @option{-mpowerpc} option allows GCC to generate instructions that
10136 are found only in the 32-bit subset of the PowerPC architecture.
10137 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
10138 GCC to use the optional PowerPC architecture instructions in the
10139 General Purpose group, including floating-point square root. Specifying
10140 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
10141 use the optional PowerPC architecture instructions in the Graphics
10142 group, including floating-point select.
10144 The @option{-mpowerpc64} option allows GCC to generate the additional
10145 64-bit instructions that are found in the full PowerPC64 architecture
10146 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
10147 @option{-mno-powerpc64}.
10149 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
10150 will use only the instructions in the common subset of both
10151 architectures plus some special AIX common-mode calls, and will not use
10152 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
10153 permits GCC to use any instruction from either architecture and to
10154 allow use of the MQ register; specify this for the Motorola MPC601.
10156 @item -mnew-mnemonics
10157 @itemx -mold-mnemonics
10158 @opindex mnew-mnemonics
10159 @opindex mold-mnemonics
10160 Select which mnemonics to use in the generated assembler code. With
10161 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
10162 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
10163 assembler mnemonics defined for the POWER architecture. Instructions
10164 defined in only one architecture have only one mnemonic; GCC uses that
10165 mnemonic irrespective of which of these options is specified.
10167 GCC defaults to the mnemonics appropriate for the architecture in
10168 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
10169 value of these option. Unless you are building a cross-compiler, you
10170 should normally not specify either @option{-mnew-mnemonics} or
10171 @option{-mold-mnemonics}, but should instead accept the default.
10173 @item -mcpu=@var{cpu_type}
10175 Set architecture type, register usage, choice of mnemonics, and
10176 instruction scheduling parameters for machine type @var{cpu_type}.
10177 Supported values for @var{cpu_type} are @samp{401}, @samp{403},
10178 @samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{505},
10179 @samp{601}, @samp{602}, @samp{603}, @samp{603e}, @samp{604},
10180 @samp{604e}, @samp{620}, @samp{630}, @samp{740}, @samp{7400},
10181 @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
10182 @samp{860}, @samp{970}, @samp{common}, @samp{ec603e}, @samp{G3},
10183 @samp{G4}, @samp{G5}, @samp{power}, @samp{power2}, @samp{power3},
10184 @samp{power4}, @samp{power5}, @samp{powerpc}, @samp{powerpc64},
10185 @samp{rios}, @samp{rios1}, @samp{rios2}, @samp{rsc}, and @samp{rs64a}.
10187 @option{-mcpu=common} selects a completely generic processor. Code
10188 generated under this option will run on any POWER or PowerPC processor.
10189 GCC will use only the instructions in the common subset of both
10190 architectures, and will not use the MQ register. GCC assumes a generic
10191 processor model for scheduling purposes.
10193 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
10194 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
10195 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
10196 types, with an appropriate, generic processor model assumed for
10197 scheduling purposes.
10199 The other options specify a specific processor. Code generated under
10200 those options will run best on that processor, and may not run at all on
10203 The @option{-mcpu} options automatically enable or disable the
10204 following options: @option{-maltivec}, @option{-mhard-float},
10205 @option{-mmfcrf}, @option{-mmultiple}, @option{-mnew-mnemonics},
10206 @option{-mpower}, @option{-mpower2}, @option{-mpowerpc64},
10207 @option{-mpowerpc-gpopt}, @option{-mpowerpc-gfxopt},
10208 @option{-mstring}. The particular options set for any particular CPU
10209 will vary between compiler versions, depending on what setting seems
10210 to produce optimal code for that CPU; it doesn't necessarily reflect
10211 the actual hardware's capabilities. If you wish to set an individual
10212 option to a particular value, you may specify it after the
10213 @option{-mcpu} option, like @samp{-mcpu=970 -mno-altivec}.
10215 On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
10216 not enabled or disabled by the @option{-mcpu} option at present, since
10217 AIX does not have full support for these options. You may still
10218 enable or disable them individually if you're sure it'll work in your
10221 @item -mtune=@var{cpu_type}
10223 Set the instruction scheduling parameters for machine type
10224 @var{cpu_type}, but do not set the architecture type, register usage, or
10225 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
10226 values for @var{cpu_type} are used for @option{-mtune} as for
10227 @option{-mcpu}. If both are specified, the code generated will use the
10228 architecture, registers, and mnemonics set by @option{-mcpu}, but the
10229 scheduling parameters set by @option{-mtune}.
10232 @itemx -mno-altivec
10234 @opindex mno-altivec
10235 These switches enable or disable the use of built-in functions that
10236 allow access to the AltiVec instruction set. You may also need to set
10237 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
10242 Extend the current ABI with SPE ABI extensions. This does not change
10243 the default ABI, instead it adds the SPE ABI extensions to the current
10247 @opindex mabi=no-spe
10248 Disable Booke SPE ABI extensions for the current ABI.
10250 @item -misel=@var{yes/no}
10253 This switch enables or disables the generation of ISEL instructions.
10255 @item -mspe=@var{yes/no}
10258 This switch enables or disables the generation of SPE simd
10261 @item -mfloat-gprs=@var{yes/no}
10262 @itemx -mfloat-gprs
10263 @opindex mfloat-gprs
10264 This switch enables or disables the generation of floating point
10265 operations on the general purpose registers for architectures that
10266 support it. This option is currently only available on the MPC8540.
10272 Generate code for 32-bit or 64-bit environments of Darwin and SVR4
10273 targets (including GNU/Linux). The 32-bit environment sets int, long
10274 and pointer to 32 bits and generates code that runs on any PowerPC
10275 variant. The 64-bit environment sets int to 32 bits and long and
10276 pointer to 64 bits, and generates code for PowerPC64, as for
10277 @option{-mpowerpc64}.
10280 @itemx -mno-fp-in-toc
10281 @itemx -mno-sum-in-toc
10282 @itemx -mminimal-toc
10284 @opindex mno-fp-in-toc
10285 @opindex mno-sum-in-toc
10286 @opindex mminimal-toc
10287 Modify generation of the TOC (Table Of Contents), which is created for
10288 every executable file. The @option{-mfull-toc} option is selected by
10289 default. In that case, GCC will allocate at least one TOC entry for
10290 each unique non-automatic variable reference in your program. GCC
10291 will also place floating-point constants in the TOC@. However, only
10292 16,384 entries are available in the TOC@.
10294 If you receive a linker error message that saying you have overflowed
10295 the available TOC space, you can reduce the amount of TOC space used
10296 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
10297 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
10298 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
10299 generate code to calculate the sum of an address and a constant at
10300 run-time instead of putting that sum into the TOC@. You may specify one
10301 or both of these options. Each causes GCC to produce very slightly
10302 slower and larger code at the expense of conserving TOC space.
10304 If you still run out of space in the TOC even when you specify both of
10305 these options, specify @option{-mminimal-toc} instead. This option causes
10306 GCC to make only one TOC entry for every file. When you specify this
10307 option, GCC will produce code that is slower and larger but which
10308 uses extremely little TOC space. You may wish to use this option
10309 only on files that contain less frequently executed code.
10315 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
10316 @code{long} type, and the infrastructure needed to support them.
10317 Specifying @option{-maix64} implies @option{-mpowerpc64} and
10318 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
10319 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
10322 @itemx -mno-xl-call
10324 @opindex mno-xl-call
10325 On AIX, pass floating-point arguments to prototyped functions beyond the
10326 register save area (RSA) on the stack in addition to argument FPRs. The
10327 AIX calling convention was extended but not initially documented to
10328 handle an obscure K&R C case of calling a function that takes the
10329 address of its arguments with fewer arguments than declared. AIX XL
10330 compilers access floating point arguments which do not fit in the
10331 RSA from the stack when a subroutine is compiled without
10332 optimization. Because always storing floating-point arguments on the
10333 stack is inefficient and rarely needed, this option is not enabled by
10334 default and only is necessary when calling subroutines compiled by AIX
10335 XL compilers without optimization.
10339 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
10340 application written to use message passing with special startup code to
10341 enable the application to run. The system must have PE installed in the
10342 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
10343 must be overridden with the @option{-specs=} option to specify the
10344 appropriate directory location. The Parallel Environment does not
10345 support threads, so the @option{-mpe} option and the @option{-pthread}
10346 option are incompatible.
10348 @item -malign-natural
10349 @itemx -malign-power
10350 @opindex malign-natural
10351 @opindex malign-power
10352 On AIX, Darwin, and 64-bit PowerPC GNU/Linux, the option
10353 @option{-malign-natural} overrides the ABI-defined alignment of larger
10354 types, such as floating-point doubles, on their natural size-based boundary.
10355 The option @option{-malign-power} instructs GCC to follow the ABI-specified
10356 alignment rules. GCC defaults to the standard alignment defined in the ABI.
10359 @itemx -mhard-float
10360 @opindex msoft-float
10361 @opindex mhard-float
10362 Generate code that does not use (uses) the floating-point register set.
10363 Software floating point emulation is provided if you use the
10364 @option{-msoft-float} option, and pass the option to GCC when linking.
10367 @itemx -mno-multiple
10369 @opindex mno-multiple
10370 Generate code that uses (does not use) the load multiple word
10371 instructions and the store multiple word instructions. These
10372 instructions are generated by default on POWER systems, and not
10373 generated on PowerPC systems. Do not use @option{-mmultiple} on little
10374 endian PowerPC systems, since those instructions do not work when the
10375 processor is in little endian mode. The exceptions are PPC740 and
10376 PPC750 which permit the instructions usage in little endian mode.
10381 @opindex mno-string
10382 Generate code that uses (does not use) the load string instructions
10383 and the store string word instructions to save multiple registers and
10384 do small block moves. These instructions are generated by default on
10385 POWER systems, and not generated on PowerPC systems. Do not use
10386 @option{-mstring} on little endian PowerPC systems, since those
10387 instructions do not work when the processor is in little endian mode.
10388 The exceptions are PPC740 and PPC750 which permit the instructions
10389 usage in little endian mode.
10394 @opindex mno-update
10395 Generate code that uses (does not use) the load or store instructions
10396 that update the base register to the address of the calculated memory
10397 location. These instructions are generated by default. If you use
10398 @option{-mno-update}, there is a small window between the time that the
10399 stack pointer is updated and the address of the previous frame is
10400 stored, which means code that walks the stack frame across interrupts or
10401 signals may get corrupted data.
10404 @itemx -mno-fused-madd
10405 @opindex mfused-madd
10406 @opindex mno-fused-madd
10407 Generate code that uses (does not use) the floating point multiply and
10408 accumulate instructions. These instructions are generated by default if
10409 hardware floating is used.
10411 @item -mno-bit-align
10413 @opindex mno-bit-align
10414 @opindex mbit-align
10415 On System V.4 and embedded PowerPC systems do not (do) force structures
10416 and unions that contain bit-fields to be aligned to the base type of the
10419 For example, by default a structure containing nothing but 8
10420 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
10421 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
10422 the structure would be aligned to a 1 byte boundary and be one byte in
10425 @item -mno-strict-align
10426 @itemx -mstrict-align
10427 @opindex mno-strict-align
10428 @opindex mstrict-align
10429 On System V.4 and embedded PowerPC systems do not (do) assume that
10430 unaligned memory references will be handled by the system.
10432 @item -mrelocatable
10433 @itemx -mno-relocatable
10434 @opindex mrelocatable
10435 @opindex mno-relocatable
10436 On embedded PowerPC systems generate code that allows (does not allow)
10437 the program to be relocated to a different address at runtime. If you
10438 use @option{-mrelocatable} on any module, all objects linked together must
10439 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
10441 @item -mrelocatable-lib
10442 @itemx -mno-relocatable-lib
10443 @opindex mrelocatable-lib
10444 @opindex mno-relocatable-lib
10445 On embedded PowerPC systems generate code that allows (does not allow)
10446 the program to be relocated to a different address at runtime. Modules
10447 compiled with @option{-mrelocatable-lib} can be linked with either modules
10448 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
10449 with modules compiled with the @option{-mrelocatable} options.
10455 On System V.4 and embedded PowerPC systems do not (do) assume that
10456 register 2 contains a pointer to a global area pointing to the addresses
10457 used in the program.
10460 @itemx -mlittle-endian
10462 @opindex mlittle-endian
10463 On System V.4 and embedded PowerPC systems compile code for the
10464 processor in little endian mode. The @option{-mlittle-endian} option is
10465 the same as @option{-mlittle}.
10468 @itemx -mbig-endian
10470 @opindex mbig-endian
10471 On System V.4 and embedded PowerPC systems compile code for the
10472 processor in big endian mode. The @option{-mbig-endian} option is
10473 the same as @option{-mbig}.
10475 @item -mdynamic-no-pic
10476 @opindex mdynamic-no-pic
10477 On Darwin and Mac OS X systems, compile code so that it is not
10478 relocatable, but that its external references are relocatable. The
10479 resulting code is suitable for applications, but not shared
10482 @item -mprioritize-restricted-insns=@var{priority}
10483 @opindex mprioritize-restricted-insns
10484 This option controls the priority that is assigned to
10485 dispatch-slot restricted instructions during the second scheduling
10486 pass. The argument @var{priority} takes the value @var{0/1/2} to assign
10487 @var{no/highest/second-highest} priority to dispatch slot restricted
10490 @item -msched-costly-dep=@var{dependence_type}
10491 @opindex msched-costly-dep
10492 This option controls which dependences are considered costly
10493 by the target during instruction scheduling. The argument
10494 @var{dependence_type} takes one of the following values:
10495 @var{no}: no dependence is costly,
10496 @var{all}: all dependences are costly,
10497 @var{true_store_to_load}: a true dependence from store to load is costly,
10498 @var{store_to_load}: any dependence from store to load is costly,
10499 @var{number}: any dependence which latency >= @var{number} is costly.
10501 @item -minsert-sched-nops=@var{scheme}
10502 @opindex minsert-sched-nops
10503 This option controls which nop insertion scheme will be used during
10504 the second scheduling pass. The argument @var{scheme} takes one of the
10506 @var{no}: Don't insert nops.
10507 @var{pad}: Pad with nops any dispatch group which has vacant issue slots,
10508 according to the scheduler's grouping.
10509 @var{regroup_exact}: Insert nops to force costly dependent insns into
10510 separate groups. Insert exactly as many nops as needed to force an insn
10511 to a new group, according to the estimated processor grouping.
10512 @var{number}: Insert nops to force costly dependent insns into
10513 separate groups. Insert @var{number} nops to force an insn to a new group.
10516 @opindex mcall-sysv
10517 On System V.4 and embedded PowerPC systems compile code using calling
10518 conventions that adheres to the March 1995 draft of the System V
10519 Application Binary Interface, PowerPC processor supplement. This is the
10520 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
10522 @item -mcall-sysv-eabi
10523 @opindex mcall-sysv-eabi
10524 Specify both @option{-mcall-sysv} and @option{-meabi} options.
10526 @item -mcall-sysv-noeabi
10527 @opindex mcall-sysv-noeabi
10528 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
10530 @item -mcall-solaris
10531 @opindex mcall-solaris
10532 On System V.4 and embedded PowerPC systems compile code for the Solaris
10536 @opindex mcall-linux
10537 On System V.4 and embedded PowerPC systems compile code for the
10538 Linux-based GNU system.
10542 On System V.4 and embedded PowerPC systems compile code for the
10543 Hurd-based GNU system.
10545 @item -mcall-netbsd
10546 @opindex mcall-netbsd
10547 On System V.4 and embedded PowerPC systems compile code for the
10548 NetBSD operating system.
10550 @item -maix-struct-return
10551 @opindex maix-struct-return
10552 Return all structures in memory (as specified by the AIX ABI)@.
10554 @item -msvr4-struct-return
10555 @opindex msvr4-struct-return
10556 Return structures smaller than 8 bytes in registers (as specified by the
10559 @item -mabi=altivec
10560 @opindex mabi=altivec
10561 Extend the current ABI with AltiVec ABI extensions. This does not
10562 change the default ABI, instead it adds the AltiVec ABI extensions to
10565 @item -mabi=no-altivec
10566 @opindex mabi=no-altivec
10567 Disable AltiVec ABI extensions for the current ABI.
10570 @itemx -mno-prototype
10571 @opindex mprototype
10572 @opindex mno-prototype
10573 On System V.4 and embedded PowerPC systems assume that all calls to
10574 variable argument functions are properly prototyped. Otherwise, the
10575 compiler must insert an instruction before every non prototyped call to
10576 set or clear bit 6 of the condition code register (@var{CR}) to
10577 indicate whether floating point values were passed in the floating point
10578 registers in case the function takes a variable arguments. With
10579 @option{-mprototype}, only calls to prototyped variable argument functions
10580 will set or clear the bit.
10584 On embedded PowerPC systems, assume that the startup module is called
10585 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
10586 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
10591 On embedded PowerPC systems, assume that the startup module is called
10592 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
10597 On embedded PowerPC systems, assume that the startup module is called
10598 @file{crt0.o} and the standard C libraries are @file{libads.a} and
10601 @item -myellowknife
10602 @opindex myellowknife
10603 On embedded PowerPC systems, assume that the startup module is called
10604 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
10609 On System V.4 and embedded PowerPC systems, specify that you are
10610 compiling for a VxWorks system.
10614 Specify that you are compiling for the WindISS simulation environment.
10618 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
10619 header to indicate that @samp{eabi} extended relocations are used.
10625 On System V.4 and embedded PowerPC systems do (do not) adhere to the
10626 Embedded Applications Binary Interface (eabi) which is a set of
10627 modifications to the System V.4 specifications. Selecting @option{-meabi}
10628 means that the stack is aligned to an 8 byte boundary, a function
10629 @code{__eabi} is called to from @code{main} to set up the eabi
10630 environment, and the @option{-msdata} option can use both @code{r2} and
10631 @code{r13} to point to two separate small data areas. Selecting
10632 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
10633 do not call an initialization function from @code{main}, and the
10634 @option{-msdata} option will only use @code{r13} to point to a single
10635 small data area. The @option{-meabi} option is on by default if you
10636 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
10639 @opindex msdata=eabi
10640 On System V.4 and embedded PowerPC systems, put small initialized
10641 @code{const} global and static data in the @samp{.sdata2} section, which
10642 is pointed to by register @code{r2}. Put small initialized
10643 non-@code{const} global and static data in the @samp{.sdata} section,
10644 which is pointed to by register @code{r13}. Put small uninitialized
10645 global and static data in the @samp{.sbss} section, which is adjacent to
10646 the @samp{.sdata} section. The @option{-msdata=eabi} option is
10647 incompatible with the @option{-mrelocatable} option. The
10648 @option{-msdata=eabi} option also sets the @option{-memb} option.
10651 @opindex msdata=sysv
10652 On System V.4 and embedded PowerPC systems, put small global and static
10653 data in the @samp{.sdata} section, which is pointed to by register
10654 @code{r13}. Put small uninitialized global and static data in the
10655 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
10656 The @option{-msdata=sysv} option is incompatible with the
10657 @option{-mrelocatable} option.
10659 @item -msdata=default
10661 @opindex msdata=default
10663 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
10664 compile code the same as @option{-msdata=eabi}, otherwise compile code the
10665 same as @option{-msdata=sysv}.
10668 @opindex msdata-data
10669 On System V.4 and embedded PowerPC systems, put small global and static
10670 data in the @samp{.sdata} section. Put small uninitialized global and
10671 static data in the @samp{.sbss} section. Do not use register @code{r13}
10672 to address small data however. This is the default behavior unless
10673 other @option{-msdata} options are used.
10677 @opindex msdata=none
10679 On embedded PowerPC systems, put all initialized global and static data
10680 in the @samp{.data} section, and all uninitialized data in the
10681 @samp{.bss} section.
10685 @cindex smaller data references (PowerPC)
10686 @cindex .sdata/.sdata2 references (PowerPC)
10687 On embedded PowerPC systems, put global and static items less than or
10688 equal to @var{num} bytes into the small data or bss sections instead of
10689 the normal data or bss section. By default, @var{num} is 8. The
10690 @option{-G @var{num}} switch is also passed to the linker.
10691 All modules should be compiled with the same @option{-G @var{num}} value.
10694 @itemx -mno-regnames
10696 @opindex mno-regnames
10697 On System V.4 and embedded PowerPC systems do (do not) emit register
10698 names in the assembly language output using symbolic forms.
10701 @itemx -mno-longcall
10703 @opindex mno-longcall
10704 Default to making all function calls indirectly, using a register, so
10705 that functions which reside further than 32 megabytes (33,554,432
10706 bytes) from the current location can be called. This setting can be
10707 overridden by the @code{shortcall} function attribute, or by
10708 @code{#pragma longcall(0)}.
10710 Some linkers are capable of detecting out-of-range calls and generating
10711 glue code on the fly. On these systems, long calls are unnecessary and
10712 generate slower code. As of this writing, the AIX linker can do this,
10713 as can the GNU linker for PowerPC/64. It is planned to add this feature
10714 to the GNU linker for 32-bit PowerPC systems as well.
10716 On Darwin/PPC systems, @code{#pragma longcall} will generate ``jbsr
10717 callee, L42'', plus a ``branch island'' (glue code). The two target
10718 addresses represent the callee and the ``branch island.'' The
10719 Darwin/PPC linker will prefer the first address and generate a ``bl
10720 callee'' if the PPC ``bl'' instruction will reach the callee directly;
10721 otherwise, the linker will generate ``bl L42'' to call the ``branch
10722 island.'' The ``branch island'' is appended to the body of the
10723 calling function; it computes the full 32-bit address of the callee
10726 On Mach-O (Darwin) systems, this option directs the compiler emit to
10727 the glue for every direct call, and the Darwin linker decides whether
10728 to use or discard it.
10730 In the future, we may cause GCC to ignore all longcall specifications
10731 when the linker is known to generate glue.
10735 Adds support for multithreading with the @dfn{pthreads} library.
10736 This option sets flags for both the preprocessor and linker.
10740 @node S/390 and zSeries Options
10741 @subsection S/390 and zSeries Options
10742 @cindex S/390 and zSeries Options
10744 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
10748 @itemx -msoft-float
10749 @opindex mhard-float
10750 @opindex msoft-float
10751 Use (do not use) the hardware floating-point instructions and registers
10752 for floating-point operations. When @option{-msoft-float} is specified,
10753 functions in @file{libgcc.a} will be used to perform floating-point
10754 operations. When @option{-mhard-float} is specified, the compiler
10755 generates IEEE floating-point instructions. This is the default.
10758 @itemx -mno-backchain
10759 @itemx -mkernel-backchain
10760 @opindex mbackchain
10761 @opindex mno-backchain
10762 @opindex mkernel-backchain
10763 In order to provide a backchain the address of the caller's frame
10764 is stored within the callee's stack frame.
10765 A backchain may be needed to allow debugging using tools that do not understand
10766 DWARF-2 call frame information.
10767 For @option{-mno-backchain} no backchain is maintained at all which is the
10769 If one of the other options is present the backchain pointer is placed either
10770 on top of the stack frame (@option{-mkernel-backchain}) or on
10771 the bottom (@option{-mbackchain}).
10772 Beside the different backchain location @option{-mkernel-backchain}
10773 also changes stack frame layout breaking the ABI. This option
10774 is intended to be used for code which internally needs a backchain but has
10775 to get by with a limited stack size e.g.@: the linux kernel.
10776 Internal unwinding code not using DWARF-2 info has to be able to locate the
10777 return address of a function. That will be eased be the fact that
10778 the return address of a function is placed two words below the backchain
10782 @itemx -mno-small-exec
10783 @opindex msmall-exec
10784 @opindex mno-small-exec
10785 Generate (or do not generate) code using the @code{bras} instruction
10786 to do subroutine calls.
10787 This only works reliably if the total executable size does not
10788 exceed 64k. The default is to use the @code{basr} instruction instead,
10789 which does not have this limitation.
10795 When @option{-m31} is specified, generate code compliant to the
10796 GNU/Linux for S/390 ABI@. When @option{-m64} is specified, generate
10797 code compliant to the GNU/Linux for zSeries ABI@. This allows GCC in
10798 particular to generate 64-bit instructions. For the @samp{s390}
10799 targets, the default is @option{-m31}, while the @samp{s390x}
10800 targets default to @option{-m64}.
10806 When @option{-mzarch} is specified, generate code using the
10807 instructions available on z/Architecture.
10808 When @option{-mesa} is specified, generate code using the
10809 instructions available on ESA/390. Note that @option{-mesa} is
10810 not possible with @option{-m64}.
10811 When generating code compliant to the GNU/Linux for S/390 ABI,
10812 the default is @option{-mesa}. When generating code compliant
10813 to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
10819 Generate (or do not generate) code using the @code{mvcle} instruction
10820 to perform block moves. When @option{-mno-mvcle} is specified,
10821 use a @code{mvc} loop instead. This is the default.
10827 Print (or do not print) additional debug information when compiling.
10828 The default is to not print debug information.
10830 @item -march=@var{cpu-type}
10832 Generate code that will run on @var{cpu-type}, which is the name of a system
10833 representing a certain processor type. Possible values for
10834 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, and @samp{z990}.
10835 When generating code using the instructions available on z/Architecture,
10836 the default is @option{-march=z900}. Otherwise, the default is
10837 @option{-march=g5}.
10839 @item -mtune=@var{cpu-type}
10841 Tune to @var{cpu-type} everything applicable about the generated code,
10842 except for the ABI and the set of available instructions.
10843 The list of @var{cpu-type} values is the same as for @option{-march}.
10844 The default is the value used for @option{-march}.
10847 @itemx -mno-tpf-trace
10848 @opindex mtpf-trace
10849 @opindex mno-tpf-trace
10850 Generate code that adds (does not add) in TPF OS specific branches to trace
10851 routines in the operating system. This option is off by default, even
10852 when compiling for the TPF OS.
10855 @itemx -mno-fused-madd
10856 @opindex mfused-madd
10857 @opindex mno-fused-madd
10858 Generate code that uses (does not use) the floating point multiply and
10859 accumulate instructions. These instructions are generated by default if
10860 hardware floating point is used.
10862 @item -mwarn-framesize=@var{framesize}
10863 @opindex mwarn-framesize
10864 Emit a warning if the current function exceeds the given frame size. Because
10865 this is a compile time check it doesn't need to be a real problem when the program
10866 runs. It is intended to identify functions which most probably cause
10867 a stack overflow. It is useful to be used in an environment with limited stack
10868 size e.g.@: the linux kernel.
10870 @item -mwarn-dynamicstack
10871 @opindex mwarn-dynamicstack
10872 Emit a warning if the function calls alloca or uses dynamically
10873 sized arrays. This is generally a bad idea with a limited stack size.
10875 @item -mstack-guard=@var{stack-guard}
10876 @item -mstack-size=@var{stack-size}
10877 @opindex mstack-guard
10878 @opindex mstack-size
10879 These arguments always have to be used in conjunction. If they are present the s390
10880 back end emits additional instructions in the function prologue which trigger a trap
10881 if the stack size is @var{stack-guard} bytes above the @var{stack-size}
10882 (remember that the stack on s390 grows downward). These options are intended to
10883 be used to help debugging stack overflow problems. The additionally emitted code
10884 cause only little overhead and hence can also be used in production like systems
10885 without greater performance degradation. The given values have to be exact
10886 powers of 2 and @var{stack-size} has to be greater than @var{stack-guard}.
10887 In order to be efficient the extra code makes the assumption that the stack starts
10888 at an address aligned to the value given by @var{stack-size}.
10892 @subsection SH Options
10894 These @samp{-m} options are defined for the SH implementations:
10899 Generate code for the SH1.
10903 Generate code for the SH2.
10906 Generate code for the SH2e.
10910 Generate code for the SH3.
10914 Generate code for the SH3e.
10918 Generate code for the SH4 without a floating-point unit.
10920 @item -m4-single-only
10921 @opindex m4-single-only
10922 Generate code for the SH4 with a floating-point unit that only
10923 supports single-precision arithmetic.
10927 Generate code for the SH4 assuming the floating-point unit is in
10928 single-precision mode by default.
10932 Generate code for the SH4.
10936 Generate code for the SH4al-dsp, or for a SH4a in such a way that the
10937 floating-point unit is not used.
10939 @item -m4a-single-only
10940 @opindex m4a-single-only
10941 Generate code for the SH4a, in such a way that no double-precision
10942 floating point operations are used.
10945 @opindex m4a-single
10946 Generate code for the SH4a assuming the floating-point unit is in
10947 single-precision mode by default.
10951 Generate code for the SH4a.
10955 Same as @option{-m4a-nofpu}, except that it implicitly passes
10956 @option{-dsp} to the assembler. GCC doesn't generate any DSP
10957 instructions at the moment.
10961 Compile code for the processor in big endian mode.
10965 Compile code for the processor in little endian mode.
10969 Align doubles at 64-bit boundaries. Note that this changes the calling
10970 conventions, and thus some functions from the standard C library will
10971 not work unless you recompile it first with @option{-mdalign}.
10975 Shorten some address references at link time, when possible; uses the
10976 linker option @option{-relax}.
10980 Use 32-bit offsets in @code{switch} tables. The default is to use
10985 Enable the use of the instruction @code{fmovd}.
10989 Comply with the calling conventions defined by Renesas.
10993 Comply with the calling conventions defined by Renesas.
10997 Comply with the calling conventions defined for GCC before the Renesas
10998 conventions were available. This option is the default for all
10999 targets of the SH toolchain except for @samp{sh-symbianelf}.
11002 @opindex mnomacsave
11003 Mark the @code{MAC} register as call-clobbered, even if
11004 @option{-mhitachi} is given.
11008 Increase IEEE-compliance of floating-point code.
11012 Dump instruction size and location in the assembly code.
11015 @opindex mpadstruct
11016 This option is deprecated. It pads structures to multiple of 4 bytes,
11017 which is incompatible with the SH ABI@.
11021 Optimize for space instead of speed. Implied by @option{-Os}.
11024 @opindex mprefergot
11025 When generating position-independent code, emit function calls using
11026 the Global Offset Table instead of the Procedure Linkage Table.
11030 Generate a library function call to invalidate instruction cache
11031 entries, after fixing up a trampoline. This library function call
11032 doesn't assume it can write to the whole memory address space. This
11033 is the default when the target is @code{sh-*-linux*}.
11036 @node SPARC Options
11037 @subsection SPARC Options
11038 @cindex SPARC options
11040 These @samp{-m} options are supported on the SPARC:
11043 @item -mno-app-regs
11045 @opindex mno-app-regs
11047 Specify @option{-mapp-regs} to generate output using the global registers
11048 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
11051 To be fully SVR4 ABI compliant at the cost of some performance loss,
11052 specify @option{-mno-app-regs}. You should compile libraries and system
11053 software with this option.
11056 @itemx -mhard-float
11058 @opindex mhard-float
11059 Generate output containing floating point instructions. This is the
11063 @itemx -msoft-float
11065 @opindex msoft-float
11066 Generate output containing library calls for floating point.
11067 @strong{Warning:} the requisite libraries are not available for all SPARC
11068 targets. Normally the facilities of the machine's usual C compiler are
11069 used, but this cannot be done directly in cross-compilation. You must make
11070 your own arrangements to provide suitable library functions for
11071 cross-compilation. The embedded targets @samp{sparc-*-aout} and
11072 @samp{sparclite-*-*} do provide software floating point support.
11074 @option{-msoft-float} changes the calling convention in the output file;
11075 therefore, it is only useful if you compile @emph{all} of a program with
11076 this option. In particular, you need to compile @file{libgcc.a}, the
11077 library that comes with GCC, with @option{-msoft-float} in order for
11080 @item -mhard-quad-float
11081 @opindex mhard-quad-float
11082 Generate output containing quad-word (long double) floating point
11085 @item -msoft-quad-float
11086 @opindex msoft-quad-float
11087 Generate output containing library calls for quad-word (long double)
11088 floating point instructions. The functions called are those specified
11089 in the SPARC ABI@. This is the default.
11091 As of this writing, there are no SPARC implementations that have hardware
11092 support for the quad-word floating point instructions. They all invoke
11093 a trap handler for one of these instructions, and then the trap handler
11094 emulates the effect of the instruction. Because of the trap handler overhead,
11095 this is much slower than calling the ABI library routines. Thus the
11096 @option{-msoft-quad-float} option is the default.
11098 @item -mno-unaligned-doubles
11099 @itemx -munaligned-doubles
11100 @opindex mno-unaligned-doubles
11101 @opindex munaligned-doubles
11102 Assume that doubles have 8 byte alignment. This is the default.
11104 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
11105 alignment only if they are contained in another type, or if they have an
11106 absolute address. Otherwise, it assumes they have 4 byte alignment.
11107 Specifying this option avoids some rare compatibility problems with code
11108 generated by other compilers. It is not the default because it results
11109 in a performance loss, especially for floating point code.
11111 @item -mno-faster-structs
11112 @itemx -mfaster-structs
11113 @opindex mno-faster-structs
11114 @opindex mfaster-structs
11115 With @option{-mfaster-structs}, the compiler assumes that structures
11116 should have 8 byte alignment. This enables the use of pairs of
11117 @code{ldd} and @code{std} instructions for copies in structure
11118 assignment, in place of twice as many @code{ld} and @code{st} pairs.
11119 However, the use of this changed alignment directly violates the SPARC
11120 ABI@. Thus, it's intended only for use on targets where the developer
11121 acknowledges that their resulting code will not be directly in line with
11122 the rules of the ABI@.
11124 @item -mimpure-text
11125 @opindex mimpure-text
11126 @option{-mimpure-text}, used in addition to @option{-shared}, tells
11127 the compiler to not pass @option{-z text} to the linker when linking a
11128 shared object. Using this option, you can link position-dependent
11129 code into a shared object.
11131 @option{-mimpure-text} suppresses the ``relocations remain against
11132 allocatable but non-writable sections'' linker error message.
11133 However, the necessary relocations will trigger copy-on-write, and the
11134 shared object is not actually shared across processes. Instead of
11135 using @option{-mimpure-text}, you should compile all source code with
11136 @option{-fpic} or @option{-fPIC}.
11138 This option is only available on SunOS and Solaris.
11140 @item -mcpu=@var{cpu_type}
11142 Set the instruction set, register set, and instruction scheduling parameters
11143 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
11144 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
11145 @samp{f930}, @samp{f934}, @samp{hypersparc}, @samp{sparclite86x},
11146 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc}, and
11147 @samp{ultrasparc3}.
11149 Default instruction scheduling parameters are used for values that select
11150 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
11151 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
11153 Here is a list of each supported architecture and their supported
11158 v8: supersparc, hypersparc
11159 sparclite: f930, f934, sparclite86x
11161 v9: ultrasparc, ultrasparc3
11164 By default (unless configured otherwise), GCC generates code for the V7
11165 variant of the SPARC architecture. With @option{-mcpu=cypress}, the compiler
11166 additionally optimizes it for the Cypress CY7C602 chip, as used in the
11167 SPARCStation/SPARCServer 3xx series. This is also appropriate for the older
11168 SPARCStation 1, 2, IPX etc.
11170 With @option{-mcpu=v8}, GCC generates code for the V8 variant of the SPARC
11171 architecture. The only difference from V7 code is that the compiler emits
11172 the integer multiply and integer divide instructions which exist in SPARC-V8
11173 but not in SPARC-V7. With @option{-mcpu=supersparc}, the compiler additionally
11174 optimizes it for the SuperSPARC chip, as used in the SPARCStation 10, 1000 and
11177 With @option{-mcpu=sparclite}, GCC generates code for the SPARClite variant of
11178 the SPARC architecture. This adds the integer multiply, integer divide step
11179 and scan (@code{ffs}) instructions which exist in SPARClite but not in SPARC-V7.
11180 With @option{-mcpu=f930}, the compiler additionally optimizes it for the
11181 Fujitsu MB86930 chip, which is the original SPARClite, with no FPU. With
11182 @option{-mcpu=f934}, the compiler additionally optimizes it for the Fujitsu
11183 MB86934 chip, which is the more recent SPARClite with FPU.
11185 With @option{-mcpu=sparclet}, GCC generates code for the SPARClet variant of
11186 the SPARC architecture. This adds the integer multiply, multiply/accumulate,
11187 integer divide step and scan (@code{ffs}) instructions which exist in SPARClet
11188 but not in SPARC-V7. With @option{-mcpu=tsc701}, the compiler additionally
11189 optimizes it for the TEMIC SPARClet chip.
11191 With @option{-mcpu=v9}, GCC generates code for the V9 variant of the SPARC
11192 architecture. This adds 64-bit integer and floating-point move instructions,
11193 3 additional floating-point condition code registers and conditional move
11194 instructions. With @option{-mcpu=ultrasparc}, the compiler additionally
11195 optimizes it for the Sun UltraSPARC I/II chips. With
11196 @option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
11197 Sun UltraSPARC III chip.
11199 @item -mtune=@var{cpu_type}
11201 Set the instruction scheduling parameters for machine type
11202 @var{cpu_type}, but do not set the instruction set or register set that the
11203 option @option{-mcpu=@var{cpu_type}} would.
11205 The same values for @option{-mcpu=@var{cpu_type}} can be used for
11206 @option{-mtune=@var{cpu_type}}, but the only useful values are those
11207 that select a particular cpu implementation. Those are @samp{cypress},
11208 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
11209 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc}, and
11210 @samp{ultrasparc3}.
11215 @opindex mno-v8plus
11216 With @option{-mv8plus}, GCC generates code for the SPARC-V8+ ABI. The
11217 difference from the V8 ABI is that the global and out registers are
11218 considered 64-bit wide. This is enabled by default on Solaris in 32-bit
11219 mode for all SPARC-V9 processors.
11225 With @option{-mvis}, GCC generates code that takes advantage of the UltraSPARC
11226 Visual Instruction Set extensions. The default is @option{-mno-vis}.
11229 These @samp{-m} options are supported in addition to the above
11230 on SPARC-V9 processors in 64-bit environments:
11233 @item -mlittle-endian
11234 @opindex mlittle-endian
11235 Generate code for a processor running in little-endian mode. It is only
11236 available for a few configurations and most notably not on Solaris.
11242 Generate code for a 32-bit or 64-bit environment.
11243 The 32-bit environment sets int, long and pointer to 32 bits.
11244 The 64-bit environment sets int to 32 bits and long and pointer
11247 @item -mcmodel=medlow
11248 @opindex mcmodel=medlow
11249 Generate code for the Medium/Low code model: 64-bit addresses, programs
11250 must be linked in the low 32 bits of memory. Programs can be statically
11251 or dynamically linked.
11253 @item -mcmodel=medmid
11254 @opindex mcmodel=medmid
11255 Generate code for the Medium/Middle code model: 64-bit addresses, programs
11256 must be linked in the low 44 bits of memory, the text and data segments must
11257 be less than 2GB in size and the data segment must be located within 2GB of
11260 @item -mcmodel=medany
11261 @opindex mcmodel=medany
11262 Generate code for the Medium/Anywhere code model: 64-bit addresses, programs
11263 may be linked anywhere in memory, the text and data segments must be less
11264 than 2GB in size and the data segment must be located within 2GB of the
11267 @item -mcmodel=embmedany
11268 @opindex mcmodel=embmedany
11269 Generate code for the Medium/Anywhere code model for embedded systems:
11270 64-bit addresses, the text and data segments must be less than 2GB in
11271 size, both starting anywhere in memory (determined at link time). The
11272 global register %g4 points to the base of the data segment. Programs
11273 are statically linked and PIC is not supported.
11276 @itemx -mno-stack-bias
11277 @opindex mstack-bias
11278 @opindex mno-stack-bias
11279 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
11280 frame pointer if present, are offset by @minus{}2047 which must be added back
11281 when making stack frame references. This is the default in 64-bit mode.
11282 Otherwise, assume no such offset is present.
11285 These switches are supported in addition to the above on Solaris:
11290 Add support for multithreading using the Solaris threads library. This
11291 option sets flags for both the preprocessor and linker. This option does
11292 not affect the thread safety of object code produced by the compiler or
11293 that of libraries supplied with it.
11297 Add support for multithreading using the POSIX threads library. This
11298 option sets flags for both the preprocessor and linker. This option does
11299 not affect the thread safety of object code produced by the compiler or
11300 that of libraries supplied with it.
11303 @node System V Options
11304 @subsection Options for System V
11306 These additional options are available on System V Release 4 for
11307 compatibility with other compilers on those systems:
11312 Create a shared object.
11313 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
11317 Identify the versions of each tool used by the compiler, in a
11318 @code{.ident} assembler directive in the output.
11322 Refrain from adding @code{.ident} directives to the output file (this is
11325 @item -YP,@var{dirs}
11327 Search the directories @var{dirs}, and no others, for libraries
11328 specified with @option{-l}.
11330 @item -Ym,@var{dir}
11332 Look in the directory @var{dir} to find the M4 preprocessor.
11333 The assembler uses this option.
11334 @c This is supposed to go with a -Yd for predefined M4 macro files, but
11335 @c the generic assembler that comes with Solaris takes just -Ym.
11338 @node TMS320C3x/C4x Options
11339 @subsection TMS320C3x/C4x Options
11340 @cindex TMS320C3x/C4x Options
11342 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
11346 @item -mcpu=@var{cpu_type}
11348 Set the instruction set, register set, and instruction scheduling
11349 parameters for machine type @var{cpu_type}. Supported values for
11350 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
11351 @samp{c44}. The default is @samp{c40} to generate code for the
11356 @itemx -msmall-memory
11358 @opindex mbig-memory
11360 @opindex msmall-memory
11362 Generates code for the big or small memory model. The small memory
11363 model assumed that all data fits into one 64K word page. At run-time
11364 the data page (DP) register must be set to point to the 64K page
11365 containing the .bss and .data program sections. The big memory model is
11366 the default and requires reloading of the DP register for every direct
11373 Allow (disallow) allocation of general integer operands into the block
11374 count register BK@.
11380 Enable (disable) generation of code using decrement and branch,
11381 DBcond(D), instructions. This is enabled by default for the C4x. To be
11382 on the safe side, this is disabled for the C3x, since the maximum
11383 iteration count on the C3x is @math{2^{23} + 1} (but who iterates loops more than
11384 @math{2^{23}} times on the C3x?). Note that GCC will try to reverse a loop so
11385 that it can utilize the decrement and branch instruction, but will give
11386 up if there is more than one memory reference in the loop. Thus a loop
11387 where the loop counter is decremented can generate slightly more
11388 efficient code, in cases where the RPTB instruction cannot be utilized.
11390 @item -mdp-isr-reload
11392 @opindex mdp-isr-reload
11394 Force the DP register to be saved on entry to an interrupt service
11395 routine (ISR), reloaded to point to the data section, and restored on
11396 exit from the ISR@. This should not be required unless someone has
11397 violated the small memory model by modifying the DP register, say within
11404 For the C3x use the 24-bit MPYI instruction for integer multiplies
11405 instead of a library call to guarantee 32-bit results. Note that if one
11406 of the operands is a constant, then the multiplication will be performed
11407 using shifts and adds. If the @option{-mmpyi} option is not specified for the C3x,
11408 then squaring operations are performed inline instead of a library call.
11411 @itemx -mno-fast-fix
11413 @opindex mno-fast-fix
11414 The C3x/C4x FIX instruction to convert a floating point value to an
11415 integer value chooses the nearest integer less than or equal to the
11416 floating point value rather than to the nearest integer. Thus if the
11417 floating point number is negative, the result will be incorrectly
11418 truncated an additional code is necessary to detect and correct this
11419 case. This option can be used to disable generation of the additional
11420 code required to correct the result.
11426 Enable (disable) generation of repeat block sequences using the RPTB
11427 instruction for zero overhead looping. The RPTB construct is only used
11428 for innermost loops that do not call functions or jump across the loop
11429 boundaries. There is no advantage having nested RPTB loops due to the
11430 overhead required to save and restore the RC, RS, and RE registers.
11431 This is enabled by default with @option{-O2}.
11433 @item -mrpts=@var{count}
11437 Enable (disable) the use of the single instruction repeat instruction
11438 RPTS@. If a repeat block contains a single instruction, and the loop
11439 count can be guaranteed to be less than the value @var{count}, GCC will
11440 emit a RPTS instruction instead of a RPTB@. If no value is specified,
11441 then a RPTS will be emitted even if the loop count cannot be determined
11442 at compile time. Note that the repeated instruction following RPTS does
11443 not have to be reloaded from memory each iteration, thus freeing up the
11444 CPU buses for operands. However, since interrupts are blocked by this
11445 instruction, it is disabled by default.
11447 @item -mloop-unsigned
11448 @itemx -mno-loop-unsigned
11449 @opindex mloop-unsigned
11450 @opindex mno-loop-unsigned
11451 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
11452 is @math{2^{31} + 1} since these instructions test if the iteration count is
11453 negative to terminate the loop. If the iteration count is unsigned
11454 there is a possibility than the @math{2^{31} + 1} maximum iteration count may be
11455 exceeded. This switch allows an unsigned iteration count.
11459 Try to emit an assembler syntax that the TI assembler (asm30) is happy
11460 with. This also enforces compatibility with the API employed by the TI
11461 C3x C compiler. For example, long doubles are passed as structures
11462 rather than in floating point registers.
11468 Generate code that uses registers (stack) for passing arguments to functions.
11469 By default, arguments are passed in registers where possible rather
11470 than by pushing arguments on to the stack.
11472 @item -mparallel-insns
11473 @itemx -mno-parallel-insns
11474 @opindex mparallel-insns
11475 @opindex mno-parallel-insns
11476 Allow the generation of parallel instructions. This is enabled by
11477 default with @option{-O2}.
11479 @item -mparallel-mpy
11480 @itemx -mno-parallel-mpy
11481 @opindex mparallel-mpy
11482 @opindex mno-parallel-mpy
11483 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
11484 provided @option{-mparallel-insns} is also specified. These instructions have
11485 tight register constraints which can pessimize the code generation
11486 of large functions.
11491 @subsection V850 Options
11492 @cindex V850 Options
11494 These @samp{-m} options are defined for V850 implementations:
11498 @itemx -mno-long-calls
11499 @opindex mlong-calls
11500 @opindex mno-long-calls
11501 Treat all calls as being far away (near). If calls are assumed to be
11502 far away, the compiler will always load the functions address up into a
11503 register, and call indirect through the pointer.
11509 Do not optimize (do optimize) basic blocks that use the same index
11510 pointer 4 or more times to copy pointer into the @code{ep} register, and
11511 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
11512 option is on by default if you optimize.
11514 @item -mno-prolog-function
11515 @itemx -mprolog-function
11516 @opindex mno-prolog-function
11517 @opindex mprolog-function
11518 Do not use (do use) external functions to save and restore registers
11519 at the prologue and epilogue of a function. The external functions
11520 are slower, but use less code space if more than one function saves
11521 the same number of registers. The @option{-mprolog-function} option
11522 is on by default if you optimize.
11526 Try to make the code as small as possible. At present, this just turns
11527 on the @option{-mep} and @option{-mprolog-function} options.
11529 @item -mtda=@var{n}
11531 Put static or global variables whose size is @var{n} bytes or less into
11532 the tiny data area that register @code{ep} points to. The tiny data
11533 area can hold up to 256 bytes in total (128 bytes for byte references).
11535 @item -msda=@var{n}
11537 Put static or global variables whose size is @var{n} bytes or less into
11538 the small data area that register @code{gp} points to. The small data
11539 area can hold up to 64 kilobytes.
11541 @item -mzda=@var{n}
11543 Put static or global variables whose size is @var{n} bytes or less into
11544 the first 32 kilobytes of memory.
11548 Specify that the target processor is the V850.
11551 @opindex mbig-switch
11552 Generate code suitable for big switch tables. Use this option only if
11553 the assembler/linker complain about out of range branches within a switch
11558 This option will cause r2 and r5 to be used in the code generated by
11559 the compiler. This setting is the default.
11561 @item -mno-app-regs
11562 @opindex mno-app-regs
11563 This option will cause r2 and r5 to be treated as fixed registers.
11567 Specify that the target processor is the V850E1. The preprocessor
11568 constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
11569 this option is used.
11573 Specify that the target processor is the V850E. The preprocessor
11574 constant @samp{__v850e__} will be defined if this option is used.
11576 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
11577 are defined then a default target processor will be chosen and the
11578 relevant @samp{__v850*__} preprocessor constant will be defined.
11580 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
11581 defined, regardless of which processor variant is the target.
11583 @item -mdisable-callt
11584 @opindex mdisable-callt
11585 This option will suppress generation of the CALLT instruction for the
11586 v850e and v850e1 flavors of the v850 architecture. The default is
11587 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
11592 @subsection VAX Options
11593 @cindex VAX options
11595 These @samp{-m} options are defined for the VAX:
11600 Do not output certain jump instructions (@code{aobleq} and so on)
11601 that the Unix assembler for the VAX cannot handle across long
11606 Do output those jump instructions, on the assumption that you
11607 will assemble with the GNU assembler.
11611 Output code for g-format floating point numbers instead of d-format.
11614 @node x86-64 Options
11615 @subsection x86-64 Options
11616 @cindex x86-64 options
11618 These are listed under @xref{i386 and x86-64 Options}.
11620 @node Xstormy16 Options
11621 @subsection Xstormy16 Options
11622 @cindex Xstormy16 Options
11624 These options are defined for Xstormy16:
11629 Choose startup files and linker script suitable for the simulator.
11632 @node Xtensa Options
11633 @subsection Xtensa Options
11634 @cindex Xtensa Options
11636 These options are supported for Xtensa targets:
11640 @itemx -mno-const16
11642 @opindex mno-const16
11643 Enable or disable use of @code{CONST16} instructions for loading
11644 constant values. The @code{CONST16} instruction is currently not a
11645 standard option from Tensilica. When enabled, @code{CONST16}
11646 instructions are always used in place of the standard @code{L32R}
11647 instructions. The use of @code{CONST16} is enabled by default only if
11648 the @code{L32R} instruction is not available.
11651 @itemx -mno-fused-madd
11652 @opindex mfused-madd
11653 @opindex mno-fused-madd
11654 Enable or disable use of fused multiply/add and multiply/subtract
11655 instructions in the floating-point option. This has no effect if the
11656 floating-point option is not also enabled. Disabling fused multiply/add
11657 and multiply/subtract instructions forces the compiler to use separate
11658 instructions for the multiply and add/subtract operations. This may be
11659 desirable in some cases where strict IEEE 754-compliant results are
11660 required: the fused multiply add/subtract instructions do not round the
11661 intermediate result, thereby producing results with @emph{more} bits of
11662 precision than specified by the IEEE standard. Disabling fused multiply
11663 add/subtract instructions also ensures that the program output is not
11664 sensitive to the compiler's ability to combine multiply and add/subtract
11667 @item -mtext-section-literals
11668 @itemx -mno-text-section-literals
11669 @opindex mtext-section-literals
11670 @opindex mno-text-section-literals
11671 Control the treatment of literal pools. The default is
11672 @option{-mno-text-section-literals}, which places literals in a separate
11673 section in the output file. This allows the literal pool to be placed
11674 in a data RAM/ROM, and it also allows the linker to combine literal
11675 pools from separate object files to remove redundant literals and
11676 improve code size. With @option{-mtext-section-literals}, the literals
11677 are interspersed in the text section in order to keep them as close as
11678 possible to their references. This may be necessary for large assembly
11681 @item -mtarget-align
11682 @itemx -mno-target-align
11683 @opindex mtarget-align
11684 @opindex mno-target-align
11685 When this option is enabled, GCC instructs the assembler to
11686 automatically align instructions to reduce branch penalties at the
11687 expense of some code density. The assembler attempts to widen density
11688 instructions to align branch targets and the instructions following call
11689 instructions. If there are not enough preceding safe density
11690 instructions to align a target, no widening will be performed. The
11691 default is @option{-mtarget-align}. These options do not affect the
11692 treatment of auto-aligned instructions like @code{LOOP}, which the
11693 assembler will always align, either by widening density instructions or
11694 by inserting no-op instructions.
11697 @itemx -mno-longcalls
11698 @opindex mlongcalls
11699 @opindex mno-longcalls
11700 When this option is enabled, GCC instructs the assembler to translate
11701 direct calls to indirect calls unless it can determine that the target
11702 of a direct call is in the range allowed by the call instruction. This
11703 translation typically occurs for calls to functions in other source
11704 files. Specifically, the assembler translates a direct @code{CALL}
11705 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
11706 The default is @option{-mno-longcalls}. This option should be used in
11707 programs where the call target can potentially be out of range. This
11708 option is implemented in the assembler, not the compiler, so the
11709 assembly code generated by GCC will still show direct call
11710 instructions---look at the disassembled object code to see the actual
11711 instructions. Note that the assembler will use an indirect call for
11712 every cross-file call, not just those that really will be out of range.
11715 @node zSeries Options
11716 @subsection zSeries Options
11717 @cindex zSeries options
11719 These are listed under @xref{S/390 and zSeries Options}.
11721 @node Code Gen Options
11722 @section Options for Code Generation Conventions
11723 @cindex code generation conventions
11724 @cindex options, code generation
11725 @cindex run-time options
11727 These machine-independent options control the interface conventions
11728 used in code generation.
11730 Most of them have both positive and negative forms; the negative form
11731 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
11732 one of the forms is listed---the one which is not the default. You
11733 can figure out the other form by either removing @samp{no-} or adding
11737 @item -fbounds-check
11738 @opindex fbounds-check
11739 For front-ends that support it, generate additional code to check that
11740 indices used to access arrays are within the declared range. This is
11741 currently only supported by the Java and Fortran 77 front-ends, where
11742 this option defaults to true and false respectively.
11746 This option generates traps for signed overflow on addition, subtraction,
11747 multiplication operations.
11751 This option instructs the compiler to assume that signed arithmetic
11752 overflow of addition, subtraction and multiplication wraps around
11753 using twos-complement representation. This flag enables some optimizations
11754 and disables other. This option is enabled by default for the Java
11755 front-end, as required by the Java language specification.
11758 @opindex fexceptions
11759 Enable exception handling. Generates extra code needed to propagate
11760 exceptions. For some targets, this implies GCC will generate frame
11761 unwind information for all functions, which can produce significant data
11762 size overhead, although it does not affect execution. If you do not
11763 specify this option, GCC will enable it by default for languages like
11764 C++ which normally require exception handling, and disable it for
11765 languages like C that do not normally require it. However, you may need
11766 to enable this option when compiling C code that needs to interoperate
11767 properly with exception handlers written in C++. You may also wish to
11768 disable this option if you are compiling older C++ programs that don't
11769 use exception handling.
11771 @item -fnon-call-exceptions
11772 @opindex fnon-call-exceptions
11773 Generate code that allows trapping instructions to throw exceptions.
11774 Note that this requires platform-specific runtime support that does
11775 not exist everywhere. Moreover, it only allows @emph{trapping}
11776 instructions to throw exceptions, i.e.@: memory references or floating
11777 point instructions. It does not allow exceptions to be thrown from
11778 arbitrary signal handlers such as @code{SIGALRM}.
11780 @item -funwind-tables
11781 @opindex funwind-tables
11782 Similar to @option{-fexceptions}, except that it will just generate any needed
11783 static data, but will not affect the generated code in any other way.
11784 You will normally not enable this option; instead, a language processor
11785 that needs this handling would enable it on your behalf.
11787 @item -fasynchronous-unwind-tables
11788 @opindex fasynchronous-unwind-tables
11789 Generate unwind table in dwarf2 format, if supported by target machine. The
11790 table is exact at each instruction boundary, so it can be used for stack
11791 unwinding from asynchronous events (such as debugger or garbage collector).
11793 @item -fpcc-struct-return
11794 @opindex fpcc-struct-return
11795 Return ``short'' @code{struct} and @code{union} values in memory like
11796 longer ones, rather than in registers. This convention is less
11797 efficient, but it has the advantage of allowing intercallability between
11798 GCC-compiled files and files compiled with other compilers, particularly
11799 the Portable C Compiler (pcc).
11801 The precise convention for returning structures in memory depends
11802 on the target configuration macros.
11804 Short structures and unions are those whose size and alignment match
11805 that of some integer type.
11807 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
11808 switch is not binary compatible with code compiled with the
11809 @option{-freg-struct-return} switch.
11810 Use it to conform to a non-default application binary interface.
11812 @item -freg-struct-return
11813 @opindex freg-struct-return
11814 Return @code{struct} and @code{union} values in registers when possible.
11815 This is more efficient for small structures than
11816 @option{-fpcc-struct-return}.
11818 If you specify neither @option{-fpcc-struct-return} nor
11819 @option{-freg-struct-return}, GCC defaults to whichever convention is
11820 standard for the target. If there is no standard convention, GCC
11821 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
11822 the principal compiler. In those cases, we can choose the standard, and
11823 we chose the more efficient register return alternative.
11825 @strong{Warning:} code compiled with the @option{-freg-struct-return}
11826 switch is not binary compatible with code compiled with the
11827 @option{-fpcc-struct-return} switch.
11828 Use it to conform to a non-default application binary interface.
11830 @item -fshort-enums
11831 @opindex fshort-enums
11832 Allocate to an @code{enum} type only as many bytes as it needs for the
11833 declared range of possible values. Specifically, the @code{enum} type
11834 will be equivalent to the smallest integer type which has enough room.
11836 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
11837 code that is not binary compatible with code generated without that switch.
11838 Use it to conform to a non-default application binary interface.
11840 @item -fshort-double
11841 @opindex fshort-double
11842 Use the same size for @code{double} as for @code{float}.
11844 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
11845 code that is not binary compatible with code generated without that switch.
11846 Use it to conform to a non-default application binary interface.
11848 @item -fshort-wchar
11849 @opindex fshort-wchar
11850 Override the underlying type for @samp{wchar_t} to be @samp{short
11851 unsigned int} instead of the default for the target. This option is
11852 useful for building programs to run under WINE@.
11854 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
11855 code that is not binary compatible with code generated without that switch.
11856 Use it to conform to a non-default application binary interface.
11858 @item -fshared-data
11859 @opindex fshared-data
11860 Requests that the data and non-@code{const} variables of this
11861 compilation be shared data rather than private data. The distinction
11862 makes sense only on certain operating systems, where shared data is
11863 shared between processes running the same program, while private data
11864 exists in one copy per process.
11867 @opindex fno-common
11868 In C, allocate even uninitialized global variables in the data section of the
11869 object file, rather than generating them as common blocks. This has the
11870 effect that if the same variable is declared (without @code{extern}) in
11871 two different compilations, you will get an error when you link them.
11872 The only reason this might be useful is if you wish to verify that the
11873 program will work on other systems which always work this way.
11877 Ignore the @samp{#ident} directive.
11879 @item -finhibit-size-directive
11880 @opindex finhibit-size-directive
11881 Don't output a @code{.size} assembler directive, or anything else that
11882 would cause trouble if the function is split in the middle, and the
11883 two halves are placed at locations far apart in memory. This option is
11884 used when compiling @file{crtstuff.c}; you should not need to use it
11887 @item -fverbose-asm
11888 @opindex fverbose-asm
11889 Put extra commentary information in the generated assembly code to
11890 make it more readable. This option is generally only of use to those
11891 who actually need to read the generated assembly code (perhaps while
11892 debugging the compiler itself).
11894 @option{-fno-verbose-asm}, the default, causes the
11895 extra information to be omitted and is useful when comparing two assembler
11900 @cindex global offset table
11902 Generate position-independent code (PIC) suitable for use in a shared
11903 library, if supported for the target machine. Such code accesses all
11904 constant addresses through a global offset table (GOT)@. The dynamic
11905 loader resolves the GOT entries when the program starts (the dynamic
11906 loader is not part of GCC; it is part of the operating system). If
11907 the GOT size for the linked executable exceeds a machine-specific
11908 maximum size, you get an error message from the linker indicating that
11909 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
11910 instead. (These maximums are 8k on the SPARC and 32k
11911 on the m68k and RS/6000. The 386 has no such limit.)
11913 Position-independent code requires special support, and therefore works
11914 only on certain machines. For the 386, GCC supports PIC for System V
11915 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
11916 position-independent.
11920 If supported for the target machine, emit position-independent code,
11921 suitable for dynamic linking and avoiding any limit on the size of the
11922 global offset table. This option makes a difference on the m68k
11925 Position-independent code requires special support, and therefore works
11926 only on certain machines.
11932 These options are similar to @option{-fpic} and @option{-fPIC}, but
11933 generated position independent code can be only linked into executables.
11934 Usually these options are used when @option{-pie} GCC option will be
11935 used during linking.
11937 @item -ffixed-@var{reg}
11939 Treat the register named @var{reg} as a fixed register; generated code
11940 should never refer to it (except perhaps as a stack pointer, frame
11941 pointer or in some other fixed role).
11943 @var{reg} must be the name of a register. The register names accepted
11944 are machine-specific and are defined in the @code{REGISTER_NAMES}
11945 macro in the machine description macro file.
11947 This flag does not have a negative form, because it specifies a
11950 @item -fcall-used-@var{reg}
11951 @opindex fcall-used
11952 Treat the register named @var{reg} as an allocable register that is
11953 clobbered by function calls. It may be allocated for temporaries or
11954 variables that do not live across a call. Functions compiled this way
11955 will not save and restore the register @var{reg}.
11957 It is an error to used this flag with the frame pointer or stack pointer.
11958 Use of this flag for other registers that have fixed pervasive roles in
11959 the machine's execution model will produce disastrous results.
11961 This flag does not have a negative form, because it specifies a
11964 @item -fcall-saved-@var{reg}
11965 @opindex fcall-saved
11966 Treat the register named @var{reg} as an allocable register saved by
11967 functions. It may be allocated even for temporaries or variables that
11968 live across a call. Functions compiled this way will save and restore
11969 the register @var{reg} if they use it.
11971 It is an error to used this flag with the frame pointer or stack pointer.
11972 Use of this flag for other registers that have fixed pervasive roles in
11973 the machine's execution model will produce disastrous results.
11975 A different sort of disaster will result from the use of this flag for
11976 a register in which function values may be returned.
11978 This flag does not have a negative form, because it specifies a
11981 @item -fpack-struct[=@var{n}]
11982 @opindex fpack-struct
11983 Without a value specified, pack all structure members together without
11984 holes. When a value is specified (which must be a small power of two), pack
11985 structure members according to this value, representing the maximum
11986 alignment (that is, objects with default alignment requirements larger than
11987 this will be output potentially unaligned at the next fitting location.
11989 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
11990 code that is not binary compatible with code generated without that switch.
11991 Additionally, it makes the code suboptimal.
11992 Use it to conform to a non-default application binary interface.
11994 @item -finstrument-functions
11995 @opindex finstrument-functions
11996 Generate instrumentation calls for entry and exit to functions. Just
11997 after function entry and just before function exit, the following
11998 profiling functions will be called with the address of the current
11999 function and its call site. (On some platforms,
12000 @code{__builtin_return_address} does not work beyond the current
12001 function, so the call site information may not be available to the
12002 profiling functions otherwise.)
12005 void __cyg_profile_func_enter (void *this_fn,
12007 void __cyg_profile_func_exit (void *this_fn,
12011 The first argument is the address of the start of the current function,
12012 which may be looked up exactly in the symbol table.
12014 This instrumentation is also done for functions expanded inline in other
12015 functions. The profiling calls will indicate where, conceptually, the
12016 inline function is entered and exited. This means that addressable
12017 versions of such functions must be available. If all your uses of a
12018 function are expanded inline, this may mean an additional expansion of
12019 code size. If you use @samp{extern inline} in your C code, an
12020 addressable version of such functions must be provided. (This is
12021 normally the case anyways, but if you get lucky and the optimizer always
12022 expands the functions inline, you might have gotten away without
12023 providing static copies.)
12025 A function may be given the attribute @code{no_instrument_function}, in
12026 which case this instrumentation will not be done. This can be used, for
12027 example, for the profiling functions listed above, high-priority
12028 interrupt routines, and any functions from which the profiling functions
12029 cannot safely be called (perhaps signal handlers, if the profiling
12030 routines generate output or allocate memory).
12032 @item -fstack-check
12033 @opindex fstack-check
12034 Generate code to verify that you do not go beyond the boundary of the
12035 stack. You should specify this flag if you are running in an
12036 environment with multiple threads, but only rarely need to specify it in
12037 a single-threaded environment since stack overflow is automatically
12038 detected on nearly all systems if there is only one stack.
12040 Note that this switch does not actually cause checking to be done; the
12041 operating system must do that. The switch causes generation of code
12042 to ensure that the operating system sees the stack being extended.
12044 @item -fstack-limit-register=@var{reg}
12045 @itemx -fstack-limit-symbol=@var{sym}
12046 @itemx -fno-stack-limit
12047 @opindex fstack-limit-register
12048 @opindex fstack-limit-symbol
12049 @opindex fno-stack-limit
12050 Generate code to ensure that the stack does not grow beyond a certain value,
12051 either the value of a register or the address of a symbol. If the stack
12052 would grow beyond the value, a signal is raised. For most targets,
12053 the signal is raised before the stack overruns the boundary, so
12054 it is possible to catch the signal without taking special precautions.
12056 For instance, if the stack starts at absolute address @samp{0x80000000}
12057 and grows downwards, you can use the flags
12058 @option{-fstack-limit-symbol=__stack_limit} and
12059 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
12060 of 128KB@. Note that this may only work with the GNU linker.
12062 @cindex aliasing of parameters
12063 @cindex parameters, aliased
12064 @item -fargument-alias
12065 @itemx -fargument-noalias
12066 @itemx -fargument-noalias-global
12067 @opindex fargument-alias
12068 @opindex fargument-noalias
12069 @opindex fargument-noalias-global
12070 Specify the possible relationships among parameters and between
12071 parameters and global data.
12073 @option{-fargument-alias} specifies that arguments (parameters) may
12074 alias each other and may alias global storage.@*
12075 @option{-fargument-noalias} specifies that arguments do not alias
12076 each other, but may alias global storage.@*
12077 @option{-fargument-noalias-global} specifies that arguments do not
12078 alias each other and do not alias global storage.
12080 Each language will automatically use whatever option is required by
12081 the language standard. You should not need to use these options yourself.
12083 @item -fleading-underscore
12084 @opindex fleading-underscore
12085 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
12086 change the way C symbols are represented in the object file. One use
12087 is to help link with legacy assembly code.
12089 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
12090 generate code that is not binary compatible with code generated without that
12091 switch. Use it to conform to a non-default application binary interface.
12092 Not all targets provide complete support for this switch.
12094 @item -ftls-model=@var{model}
12095 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
12096 The @var{model} argument should be one of @code{global-dynamic},
12097 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
12099 The default without @option{-fpic} is @code{initial-exec}; with
12100 @option{-fpic} the default is @code{global-dynamic}.
12102 @item -fvisibility=@var{default|internal|hidden|protected}
12103 @opindex fvisibility
12104 Set the default ELF image symbol visibility to the specified option - all
12105 symbols will be marked with this unless overridden within the code.
12106 Using this feature can very substantially improve linking and
12107 load times of shared object libraries, produce more optimized
12108 code, provide near-perfect API export and prevent symbol clashes.
12109 It is @strong{strongly} recommended that you use this in any shared objects
12112 Despite the nomenclature, @code{default} always means public ie;
12113 available to be linked against from outside the shared object.
12114 @code{protected} and @code{internal} are pretty useless in real-world
12115 usage so the only other commonly used option will be @code{hidden}.
12116 The default if -fvisibility isn't specified is @code{default} ie; make every
12117 symbol public - this causes the same behavior as previous versions of
12120 A good explanation of the benefits offered by ensuring ELF
12121 symbols have the correct visibility is given by ``How To Write
12122 Shared Libraries'' by Ulrich Drepper (which can be found at
12123 @w{@uref{http://people.redhat.com/~drepper/}}) - however a superior
12124 solution made possible by this option to marking things hidden when
12125 the default is public is to make the default hidden and mark things
12126 public. This is the norm with DLL's on Windows and with @option{-fvisibility=hidden}
12127 and @code{__attribute__ ((visibility("default")))} instead of
12128 @code{__declspec(dllexport)} you get almost identical semantics with
12129 identical syntax. This is a great boon to those working with
12130 cross-platform projects.
12132 For those adding visibility support to existing code, you may find
12133 @samp{#pragma GCC visibility} of use. This works by you enclosing
12134 the declarations you wish to set visibility for with (for example)
12135 @samp{#pragma GCC visibility push(hidden)} and
12136 @samp{#pragma GCC visibility pop}. These can be nested up to sixteen
12137 times. Bear in mind that symbol visibility should be viewed @strong{as
12138 part of the API interface contract} and thus all new code should
12139 always specify visibility when it is not the default ie; declarations
12140 only for use within the local DSO should @strong{always} be marked explicitly
12141 as hidden as so to avoid PLT indirection overheads - making this
12142 abundantly clear also aids readability and self-documentation of the code.
12143 Note that due to ISO C++ specification requirements, operator new and
12144 operator delete must always be of default visibility.
12146 An overview of these techniques, their benefits and how to use them
12147 is at @w{@uref{http://www.nedprod.com/programs/gccvisibility.html}}.
12153 @node Environment Variables
12154 @section Environment Variables Affecting GCC
12155 @cindex environment variables
12157 @c man begin ENVIRONMENT
12158 This section describes several environment variables that affect how GCC
12159 operates. Some of them work by specifying directories or prefixes to use
12160 when searching for various kinds of files. Some are used to specify other
12161 aspects of the compilation environment.
12163 Note that you can also specify places to search using options such as
12164 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
12165 take precedence over places specified using environment variables, which
12166 in turn take precedence over those specified by the configuration of GCC@.
12167 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
12168 GNU Compiler Collection (GCC) Internals}.
12173 @c @itemx LC_COLLATE
12175 @c @itemx LC_MONETARY
12176 @c @itemx LC_NUMERIC
12181 @c @findex LC_COLLATE
12182 @findex LC_MESSAGES
12183 @c @findex LC_MONETARY
12184 @c @findex LC_NUMERIC
12188 These environment variables control the way that GCC uses
12189 localization information that allow GCC to work with different
12190 national conventions. GCC inspects the locale categories
12191 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
12192 so. These locale categories can be set to any value supported by your
12193 installation. A typical value is @samp{en_GB.UTF-8} for English in the United
12194 Kingdom encoded in UTF-8.
12196 The @env{LC_CTYPE} environment variable specifies character
12197 classification. GCC uses it to determine the character boundaries in
12198 a string; this is needed for some multibyte encodings that contain quote
12199 and escape characters that would otherwise be interpreted as a string
12202 The @env{LC_MESSAGES} environment variable specifies the language to
12203 use in diagnostic messages.
12205 If the @env{LC_ALL} environment variable is set, it overrides the value
12206 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
12207 and @env{LC_MESSAGES} default to the value of the @env{LANG}
12208 environment variable. If none of these variables are set, GCC
12209 defaults to traditional C English behavior.
12213 If @env{TMPDIR} is set, it specifies the directory to use for temporary
12214 files. GCC uses temporary files to hold the output of one stage of
12215 compilation which is to be used as input to the next stage: for example,
12216 the output of the preprocessor, which is the input to the compiler
12219 @item GCC_EXEC_PREFIX
12220 @findex GCC_EXEC_PREFIX
12221 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
12222 names of the subprograms executed by the compiler. No slash is added
12223 when this prefix is combined with the name of a subprogram, but you can
12224 specify a prefix that ends with a slash if you wish.
12226 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
12227 an appropriate prefix to use based on the pathname it was invoked with.
12229 If GCC cannot find the subprogram using the specified prefix, it
12230 tries looking in the usual places for the subprogram.
12232 The default value of @env{GCC_EXEC_PREFIX} is
12233 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the value
12234 of @code{prefix} when you ran the @file{configure} script.
12236 Other prefixes specified with @option{-B} take precedence over this prefix.
12238 This prefix is also used for finding files such as @file{crt0.o} that are
12241 In addition, the prefix is used in an unusual way in finding the
12242 directories to search for header files. For each of the standard
12243 directories whose name normally begins with @samp{/usr/local/lib/gcc}
12244 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
12245 replacing that beginning with the specified prefix to produce an
12246 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
12247 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
12248 These alternate directories are searched first; the standard directories
12251 @item COMPILER_PATH
12252 @findex COMPILER_PATH
12253 The value of @env{COMPILER_PATH} is a colon-separated list of
12254 directories, much like @env{PATH}. GCC tries the directories thus
12255 specified when searching for subprograms, if it can't find the
12256 subprograms using @env{GCC_EXEC_PREFIX}.
12259 @findex LIBRARY_PATH
12260 The value of @env{LIBRARY_PATH} is a colon-separated list of
12261 directories, much like @env{PATH}. When configured as a native compiler,
12262 GCC tries the directories thus specified when searching for special
12263 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
12264 using GCC also uses these directories when searching for ordinary
12265 libraries for the @option{-l} option (but directories specified with
12266 @option{-L} come first).
12270 @cindex locale definition
12271 This variable is used to pass locale information to the compiler. One way in
12272 which this information is used is to determine the character set to be used
12273 when character literals, string literals and comments are parsed in C and C++.
12274 When the compiler is configured to allow multibyte characters,
12275 the following values for @env{LANG} are recognized:
12279 Recognize JIS characters.
12281 Recognize SJIS characters.
12283 Recognize EUCJP characters.
12286 If @env{LANG} is not defined, or if it has some other value, then the
12287 compiler will use mblen and mbtowc as defined by the default locale to
12288 recognize and translate multibyte characters.
12292 Some additional environments variables affect the behavior of the
12295 @include cppenv.texi
12299 @node Precompiled Headers
12300 @section Using Precompiled Headers
12301 @cindex precompiled headers
12302 @cindex speed of compilation
12304 Often large projects have many header files that are included in every
12305 source file. The time the compiler takes to process these header files
12306 over and over again can account for nearly all of the time required to
12307 build the project. To make builds faster, GCC allows users to
12308 `precompile' a header file; then, if builds can use the precompiled
12309 header file they will be much faster.
12311 @strong{Caution:} There are a few known situations where GCC will
12312 crash when trying to use a precompiled header. If you have trouble
12313 with a precompiled header, you should remove the precompiled header
12314 and compile without it. In addition, please use GCC's on-line
12315 defect-tracking system to report any problems you encounter with
12316 precompiled headers. @xref{Bugs}.
12318 To create a precompiled header file, simply compile it as you would any
12319 other file, if necessary using the @option{-x} option to make the driver
12320 treat it as a C or C++ header file. You will probably want to use a
12321 tool like @command{make} to keep the precompiled header up-to-date when
12322 the headers it contains change.
12324 A precompiled header file will be searched for when @code{#include} is
12325 seen in the compilation. As it searches for the included file
12326 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
12327 compiler looks for a precompiled header in each directory just before it
12328 looks for the include file in that directory. The name searched for is
12329 the name specified in the @code{#include} with @samp{.gch} appended. If
12330 the precompiled header file can't be used, it is ignored.
12332 For instance, if you have @code{#include "all.h"}, and you have
12333 @file{all.h.gch} in the same directory as @file{all.h}, then the
12334 precompiled header file will be used if possible, and the original
12335 header will be used otherwise.
12337 Alternatively, you might decide to put the precompiled header file in a
12338 directory and use @option{-I} to ensure that directory is searched
12339 before (or instead of) the directory containing the original header.
12340 Then, if you want to check that the precompiled header file is always
12341 used, you can put a file of the same name as the original header in this
12342 directory containing an @code{#error} command.
12344 This also works with @option{-include}. So yet another way to use
12345 precompiled headers, good for projects not designed with precompiled
12346 header files in mind, is to simply take most of the header files used by
12347 a project, include them from another header file, precompile that header
12348 file, and @option{-include} the precompiled header. If the header files
12349 have guards against multiple inclusion, they will be skipped because
12350 they've already been included (in the precompiled header).
12352 If you need to precompile the same header file for different
12353 languages, targets, or compiler options, you can instead make a
12354 @emph{directory} named like @file{all.h.gch}, and put each precompiled
12355 header in the directory, perhaps using @option{-o}. It doesn't matter
12356 what you call the files in the directory, every precompiled header in
12357 the directory will be considered. The first precompiled header
12358 encountered in the directory that is valid for this compilation will
12359 be used; they're searched in no particular order.
12361 There are many other possibilities, limited only by your imagination,
12362 good sense, and the constraints of your build system.
12364 A precompiled header file can be used only when these conditions apply:
12368 Only one precompiled header can be used in a particular compilation.
12371 A precompiled header can't be used once the first C token is seen. You
12372 can have preprocessor directives before a precompiled header; you can
12373 even include a precompiled header from inside another header, so long as
12374 there are no C tokens before the @code{#include}.
12377 The precompiled header file must be produced for the same language as
12378 the current compilation. You can't use a C precompiled header for a C++
12382 The precompiled header file must be produced by the same compiler
12383 version and configuration as the current compilation is using.
12384 The easiest way to guarantee this is to use the same compiler binary
12385 for creating and using precompiled headers.
12388 Any macros defined before the precompiled header is included must
12389 either be defined in the same way as when the precompiled header was
12390 generated, or must not affect the precompiled header, which usually
12391 means that the they don't appear in the precompiled header at all.
12393 The @option{-D} option is one way to define a macro before a
12394 precompiled header is included; using a @code{#define} can also do it.
12395 There are also some options that define macros implicitly, like
12396 @option{-O} and @option{-Wdeprecated}; the same rule applies to macros
12399 @item If debugging information is output when using the precompiled
12400 header, using @option{-g} or similar, the same kind of debugging information
12401 must have been output when building the precompiled header. However,
12402 a precompiled header built using @option{-g} can be used in a compilation
12403 when no debugging information is being output.
12405 @item The same @option{-m} options must generally be used when building
12406 and using the precompiled header. @xref{Submodel Options},
12407 for any cases where this rule is relaxed.
12409 @item Each of the following options must be the same when building and using
12410 the precompiled header:
12412 @gccoptlist{-fexceptions -funit-at-a-time}
12415 Some other command-line options starting with @option{-f},
12416 @option{-p}, or @option{-O} must be defined in the same way as when
12417 the precompiled header was generated. At present, it's not clear
12418 which options are safe to change and which are not; the safest choice
12419 is to use exactly the same options when generating and using the
12420 precompiled header. The following are known to be safe:
12422 @gccoptlist{-fpreprocessed -pedantic-errors}
12426 For all of these except the last, the compiler will automatically
12427 ignore the precompiled header if the conditions aren't met. If you
12428 find an option combination that doesn't work and doesn't cause the
12429 precompiled header to be ignored, please consider filing a bug report,
12432 If you do use differing options when generating and using the
12433 precompiled header, the actual behavior will be a mixture of the
12434 behavior for the options. For instance, if you use @option{-g} to
12435 generate the precompiled header but not when using it, you may or may
12436 not get debugging information for routines in the precompiled header.
12438 @node Running Protoize
12439 @section Running Protoize
12441 The program @code{protoize} is an optional part of GCC@. You can use
12442 it to add prototypes to a program, thus converting the program to ISO
12443 C in one respect. The companion program @code{unprotoize} does the
12444 reverse: it removes argument types from any prototypes that are found.
12446 When you run these programs, you must specify a set of source files as
12447 command line arguments. The conversion programs start out by compiling
12448 these files to see what functions they define. The information gathered
12449 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
12451 After scanning comes actual conversion. The specified files are all
12452 eligible to be converted; any files they include (whether sources or
12453 just headers) are eligible as well.
12455 But not all the eligible files are converted. By default,
12456 @code{protoize} and @code{unprotoize} convert only source and header
12457 files in the current directory. You can specify additional directories
12458 whose files should be converted with the @option{-d @var{directory}}
12459 option. You can also specify particular files to exclude with the
12460 @option{-x @var{file}} option. A file is converted if it is eligible, its
12461 directory name matches one of the specified directory names, and its
12462 name within the directory has not been excluded.
12464 Basic conversion with @code{protoize} consists of rewriting most
12465 function definitions and function declarations to specify the types of
12466 the arguments. The only ones not rewritten are those for varargs
12469 @code{protoize} optionally inserts prototype declarations at the
12470 beginning of the source file, to make them available for any calls that
12471 precede the function's definition. Or it can insert prototype
12472 declarations with block scope in the blocks where undeclared functions
12475 Basic conversion with @code{unprotoize} consists of rewriting most
12476 function declarations to remove any argument types, and rewriting
12477 function definitions to the old-style pre-ISO form.
12479 Both conversion programs print a warning for any function declaration or
12480 definition that they can't convert. You can suppress these warnings
12483 The output from @code{protoize} or @code{unprotoize} replaces the
12484 original source file. The original file is renamed to a name ending
12485 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
12486 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
12487 for DOS) file already exists, then the source file is simply discarded.
12489 @code{protoize} and @code{unprotoize} both depend on GCC itself to
12490 scan the program and collect information about the functions it uses.
12491 So neither of these programs will work until GCC is installed.
12493 Here is a table of the options you can use with @code{protoize} and
12494 @code{unprotoize}. Each option works with both programs unless
12498 @item -B @var{directory}
12499 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
12500 usual directory (normally @file{/usr/local/lib}). This file contains
12501 prototype information about standard system functions. This option
12502 applies only to @code{protoize}.
12504 @item -c @var{compilation-options}
12505 Use @var{compilation-options} as the options when running @command{gcc} to
12506 produce the @samp{.X} files. The special option @option{-aux-info} is
12507 always passed in addition, to tell @command{gcc} to write a @samp{.X} file.
12509 Note that the compilation options must be given as a single argument to
12510 @code{protoize} or @code{unprotoize}. If you want to specify several
12511 @command{gcc} options, you must quote the entire set of compilation options
12512 to make them a single word in the shell.
12514 There are certain @command{gcc} arguments that you cannot use, because they
12515 would produce the wrong kind of output. These include @option{-g},
12516 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
12517 the @var{compilation-options}, they are ignored.
12520 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
12521 systems) instead of @samp{.c}. This is convenient if you are converting
12522 a C program to C++. This option applies only to @code{protoize}.
12525 Add explicit global declarations. This means inserting explicit
12526 declarations at the beginning of each source file for each function
12527 that is called in the file and was not declared. These declarations
12528 precede the first function definition that contains a call to an
12529 undeclared function. This option applies only to @code{protoize}.
12531 @item -i @var{string}
12532 Indent old-style parameter declarations with the string @var{string}.
12533 This option applies only to @code{protoize}.
12535 @code{unprotoize} converts prototyped function definitions to old-style
12536 function definitions, where the arguments are declared between the
12537 argument list and the initial @samp{@{}. By default, @code{unprotoize}
12538 uses five spaces as the indentation. If you want to indent with just
12539 one space instead, use @option{-i " "}.
12542 Keep the @samp{.X} files. Normally, they are deleted after conversion
12546 Add explicit local declarations. @code{protoize} with @option{-l} inserts
12547 a prototype declaration for each function in each block which calls the
12548 function without any declaration. This option applies only to
12552 Make no real changes. This mode just prints information about the conversions
12553 that would have been done without @option{-n}.
12556 Make no @samp{.save} files. The original files are simply deleted.
12557 Use this option with caution.
12559 @item -p @var{program}
12560 Use the program @var{program} as the compiler. Normally, the name
12561 @file{gcc} is used.
12564 Work quietly. Most warnings are suppressed.
12567 Print the version number, just like @option{-v} for @command{gcc}.
12570 If you need special compiler options to compile one of your program's
12571 source files, then you should generate that file's @samp{.X} file
12572 specially, by running @command{gcc} on that source file with the
12573 appropriate options and the option @option{-aux-info}. Then run
12574 @code{protoize} on the entire set of files. @code{protoize} will use
12575 the existing @samp{.X} file because it is newer than the source file.
12579 gcc -Dfoo=bar file1.c -aux-info file1.X
12584 You need to include the special files along with the rest in the
12585 @code{protoize} command, even though their @samp{.X} files already
12586 exist, because otherwise they won't get converted.
12588 @xref{Protoize Caveats}, for more information on how to use
12589 @code{protoize} successfully.