1 @c Copyright (C) 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
2 @c 2000, 2001, 2002, 2003, 2004, 2005 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, 2005 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}}.
55 See the Info entry for @command{gcc}, or
56 @w{@uref{http://gcc.gnu.org/onlinedocs/gcc/Contributors.html}},
57 for contributors to GCC@.
62 @chapter GCC Command Options
63 @cindex GCC command options
64 @cindex command options
65 @cindex options, GCC command
67 @c man begin DESCRIPTION
68 When you invoke GCC, it normally does preprocessing, compilation,
69 assembly and linking. The ``overall options'' allow you to stop this
70 process at an intermediate stage. For example, the @option{-c} option
71 says not to run the linker. Then the output consists of object files
72 output by the assembler.
74 Other options are passed on to one stage of processing. Some options
75 control the preprocessor and others the compiler itself. Yet other
76 options control the assembler and linker; most of these are not
77 documented here, since you rarely need to use any of them.
79 @cindex C compilation options
80 Most of the command line options that you can use with GCC are useful
81 for C programs; when an option is only useful with another language
82 (usually C++), the explanation says so explicitly. If the description
83 for a particular option does not mention a source language, you can use
84 that option with all supported languages.
86 @cindex C++ compilation options
87 @xref{Invoking G++,,Compiling C++ Programs}, for a summary of special
88 options for compiling C++ programs.
90 @cindex grouping options
91 @cindex options, grouping
92 The @command{gcc} program accepts options and file names as operands. Many
93 options have multi-letter names; therefore multiple single-letter options
94 may @emph{not} be grouped: @option{-dr} is very different from @w{@samp{-d
97 @cindex order of options
98 @cindex options, order
99 You can mix options and other arguments. For the most part, the order
100 you use doesn't matter. Order does matter when you use several options
101 of the same kind; for example, if you specify @option{-L} more than once,
102 the directories are searched in the order specified.
104 Many options have long names starting with @samp{-f} or with
105 @samp{-W}---for example, @option{-fforce-mem},
106 @option{-fstrength-reduce}, @option{-Wformat} and so on. Most of
107 these have both positive and negative forms; the negative form of
108 @option{-ffoo} would be @option{-fno-foo}. This manual documents
109 only one of these two forms, whichever one is not the default.
113 @xref{Option Index}, for an index to GCC's options.
116 * Option Summary:: Brief list of all options, without explanations.
117 * Overall Options:: Controlling the kind of output:
118 an executable, object files, assembler files,
119 or preprocessed source.
120 * Invoking G++:: Compiling C++ programs.
121 * C Dialect Options:: Controlling the variant of C language compiled.
122 * C++ Dialect Options:: Variations on C++.
123 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
125 * Language Independent Options:: Controlling how diagnostics should be
127 * Warning Options:: How picky should the compiler be?
128 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
129 * Optimize Options:: How much optimization?
130 * Preprocessor Options:: Controlling header files and macro definitions.
131 Also, getting dependency information for Make.
132 * Assembler Options:: Passing options to the assembler.
133 * Link Options:: Specifying libraries and so on.
134 * Directory Options:: Where to find header files and libraries.
135 Where to find the compiler executable files.
136 * Spec Files:: How to pass switches to sub-processes.
137 * Target Options:: Running a cross-compiler, or an old version of GCC.
138 * Submodel Options:: Specifying minor hardware or convention variations,
139 such as 68010 vs 68020.
140 * Code Gen Options:: Specifying conventions for function calls, data layout
142 * Environment Variables:: Env vars that affect GCC.
143 * Precompiled Headers:: Compiling a header once, and using it many times.
144 * Running Protoize:: Automatically adding or removing function prototypes.
150 @section Option Summary
152 Here is a summary of all the options, grouped by type. Explanations are
153 in the following sections.
156 @item Overall Options
157 @xref{Overall Options,,Options Controlling the Kind of Output}.
158 @gccoptlist{-c -S -E -o @var{file} -combine -pipe -pass-exit-codes @gol
159 -x @var{language} -v -### --help --target-help --version}
161 @item C Language Options
162 @xref{C Dialect Options,,Options Controlling C Dialect}.
163 @gccoptlist{-ansi -std=@var{standard} -aux-info @var{filename} @gol
164 -fno-asm -fno-builtin -fno-builtin-@var{function} @gol
165 -fhosted -ffreestanding -fms-extensions @gol
166 -trigraphs -no-integrated-cpp -traditional -traditional-cpp @gol
167 -fallow-single-precision -fcond-mismatch @gol
168 -fsigned-bitfields -fsigned-char @gol
169 -funsigned-bitfields -funsigned-char}
171 @item C++ Language Options
172 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
173 @gccoptlist{-fabi-version=@var{n} -fno-access-control -fcheck-new @gol
174 -fconserve-space -fno-const-strings @gol
175 -fno-elide-constructors @gol
176 -fno-enforce-eh-specs @gol
177 -ffor-scope -fno-for-scope -fno-gnu-keywords @gol
178 -fno-implicit-templates @gol
179 -fno-implicit-inline-templates @gol
180 -fno-implement-inlines -fms-extensions @gol
181 -fno-nonansi-builtins -fno-operator-names @gol
182 -fno-optional-diags -fpermissive @gol
183 -frepo -fno-rtti -fstats -ftemplate-depth-@var{n} @gol
184 -fno-threadsafe-statics -fuse-cxa-atexit -fno-weak -nostdinc++ @gol
185 -fno-default-inline -fvisibility-inlines-hidden @gol
186 -Wabi -Wctor-dtor-privacy @gol
187 -Wnon-virtual-dtor -Wreorder @gol
188 -Weffc++ -Wno-deprecated @gol
189 -Wno-non-template-friend -Wold-style-cast @gol
190 -Woverloaded-virtual -Wno-pmf-conversions @gol
191 -Wsign-promo -Wsynth}
193 @item Objective-C and Objective-C++ Language Options
194 @xref{Objective-C and Objective-C++ Dialect Options,,Options Controlling
195 Objective-C and Objective-C++ Dialects}.
197 -fconstant-string-class=@var{class-name} @gol
198 -fgnu-runtime -fnext-runtime @gol
199 -fno-nil-receivers @gol
200 -fobjc-exceptions @gol
201 -freplace-objc-classes @gol
204 -Wno-protocol -Wselector -Wundeclared-selector}
206 @item Language Independent Options
207 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
208 @gccoptlist{-fmessage-length=@var{n} @gol
209 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]}}
211 @item Warning Options
212 @xref{Warning Options,,Options to Request or Suppress Warnings}.
213 @gccoptlist{-fsyntax-only -pedantic -pedantic-errors @gol
214 -w -Wextra -Wall -Waggregate-return @gol
215 -Wcast-align -Wcast-qual -Wchar-subscripts -Wcomment @gol
216 -Wconversion -Wno-deprecated-declarations @gol
217 -Wdisabled-optimization -Wno-div-by-zero -Wno-endif-labels @gol
218 -Werror -Werror-implicit-function-declaration @gol
219 -Wfatal-errors -Wfloat-equal -Wformat -Wformat=2 @gol
220 -Wno-format-extra-args -Wformat-nonliteral @gol
221 -Wformat-security -Wformat-y2k @gol
222 -Wimplicit -Wimplicit-function-declaration -Wimplicit-int @gol
223 -Wimport -Wno-import -Winit-self -Winline @gol
224 -Wno-invalid-offsetof -Winvalid-pch @gol
225 -Wlarger-than-@var{len} -Wlong-long @gol
226 -Wmain -Wmissing-braces -Wmissing-field-initializers @gol
227 -Wmissing-format-attribute -Wmissing-include-dirs @gol
228 -Wmissing-noreturn @gol
229 -Wno-multichar -Wnonnull -Wpacked -Wpadded @gol
230 -Wparentheses -Wpointer-arith -Wredundant-decls @gol
231 -Wreturn-type -Wsequence-point -Wshadow @gol
232 -Wsign-compare -Wstrict-aliasing -Wstrict-aliasing=2 @gol
233 -Wswitch -Wswitch-default -Wswitch-enum @gol
234 -Wsystem-headers -Wtrigraphs -Wundef -Wuninitialized @gol
235 -Wunknown-pragmas -Wunreachable-code @gol
236 -Wunused -Wunused-function -Wunused-label -Wunused-parameter @gol
237 -Wunused-value -Wunused-variable -Wwrite-strings @gol
240 @item C-only Warning Options
241 @gccoptlist{-Wbad-function-cast -Wmissing-declarations @gol
242 -Wmissing-prototypes -Wnested-externs -Wold-style-definition @gol
243 -Wstrict-prototypes -Wtraditional @gol
244 -Wdeclaration-after-statement -Wno-pointer-sign}
246 @item Debugging Options
247 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
248 @gccoptlist{-d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
249 -fdump-unnumbered -fdump-translation-unit@r{[}-@var{n}@r{]} @gol
250 -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
251 -fdump-ipa-all -fdump-ipa-cgraph @gol
253 -fdump-tree-original@r{[}-@var{n}@r{]} @gol
254 -fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
255 -fdump-tree-inlined@r{[}-@var{n}@r{]} @gol
256 -fdump-tree-cfg -fdump-tree-vcg -fdump-tree-alias @gol
258 -fdump-tree-ssa@r{[}-@var{n}@r{]} -fdump-tree-pre@r{[}-@var{n}@r{]} @gol
259 -fdump-tree-ccp@r{[}-@var{n}@r{]} -fdump-tree-dce@r{[}-@var{n}@r{]} @gol
260 -fdump-tree-gimple@r{[}-raw@r{]} -fdump-tree-mudflap@r{[}-@var{n}@r{]} @gol
261 -fdump-tree-dom@r{[}-@var{n}@r{]} @gol
262 -fdump-tree-dse@r{[}-@var{n}@r{]} @gol
263 -fdump-tree-phiopt@r{[}-@var{n}@r{]} @gol
264 -fdump-tree-forwprop@r{[}-@var{n}@r{]} @gol
265 -fdump-tree-copyrename@r{[}-@var{n}@r{]} @gol
266 -fdump-tree-nrv -fdump-tree-vect @gol
267 -fdump-tree-sra@r{[}-@var{n}@r{]} @gol
268 -fdump-tree-fre@r{[}-@var{n}@r{]} @gol
269 -ftree-vectorizer-verbose=@var{n} @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 -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 -fvariable-expansion-in-unroller @gol
320 -ftree-pre -ftree-ccp -ftree-dce -ftree-loop-optimize @gol
321 -ftree-loop-linear -ftree-loop-im -ftree-loop-ivcanon -fivopts @gol
322 -ftree-dominator-opts -ftree-dse -ftree-copyrename @gol
323 -ftree-ch -ftree-sra -ftree-ter -ftree-lrs -ftree-fre -ftree-vectorize @gol
324 --param @var{name}=@var{value}
325 -O -O0 -O1 -O2 -O3 -Os}
327 @item Preprocessor Options
328 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
329 @gccoptlist{-A@var{question}=@var{answer} @gol
330 -A-@var{question}@r{[}=@var{answer}@r{]} @gol
331 -C -dD -dI -dM -dN @gol
332 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
333 -idirafter @var{dir} @gol
334 -include @var{file} -imacros @var{file} @gol
335 -iprefix @var{file} -iwithprefix @var{dir} @gol
336 -iwithprefixbefore @var{dir} -isystem @var{dir} @gol
337 -M -MM -MF -MG -MP -MQ -MT -nostdinc @gol
338 -P -fworking-directory -remap @gol
339 -trigraphs -undef -U@var{macro} -Wp,@var{option} @gol
340 -Xpreprocessor @var{option}}
342 @item Assembler Option
343 @xref{Assembler Options,,Passing Options to the Assembler}.
344 @gccoptlist{-Wa,@var{option} -Xassembler @var{option}}
347 @xref{Link Options,,Options for Linking}.
348 @gccoptlist{@var{object-file-name} -l@var{library} @gol
349 -nostartfiles -nodefaultlibs -nostdlib -pie @gol
350 -s -static -static-libgcc -shared -shared-libgcc -symbolic @gol
351 -Wl,@var{option} -Xlinker @var{option} @gol
354 @item Directory Options
355 @xref{Directory Options,,Options for Directory Search}.
356 @gccoptlist{-B@var{prefix} -I@var{dir} -iquote@var{dir} -L@var{dir} -specs=@var{file} -I-}
359 @c I wrote this xref this way to avoid overfull hbox. -- rms
360 @xref{Target Options}.
361 @gccoptlist{-V @var{version} -b @var{machine}}
363 @item Machine Dependent Options
364 @xref{Submodel Options,,Hardware Models and Configurations}.
365 @c This list is ordered alphanumerically by subsection name.
366 @c Try and put the significant identifier (CPU or system) first,
367 @c so users have a clue at guessing where the ones they want will be.
370 @gccoptlist{-EB -EL @gol
371 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
372 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
375 @gccoptlist{-mapcs-frame -mno-apcs-frame @gol
376 -mabi=@var{name} @gol
377 -mapcs-stack-check -mno-apcs-stack-check @gol
378 -mapcs-float -mno-apcs-float @gol
379 -mapcs-reentrant -mno-apcs-reentrant @gol
380 -msched-prolog -mno-sched-prolog @gol
381 -mlittle-endian -mbig-endian -mwords-little-endian @gol
382 -mfloat-abi=@var{name} -msoft-float -mhard-float -mfpe @gol
383 -mthumb-interwork -mno-thumb-interwork @gol
384 -mcpu=@var{name} -march=@var{name} -mfpu=@var{name} @gol
385 -mstructure-size-boundary=@var{n} @gol
386 -mabort-on-noreturn @gol
387 -mlong-calls -mno-long-calls @gol
388 -msingle-pic-base -mno-single-pic-base @gol
389 -mpic-register=@var{reg} @gol
390 -mnop-fun-dllimport @gol
391 -mcirrus-fix-invalid-insns -mno-cirrus-fix-invalid-insns @gol
392 -mpoke-function-name @gol
394 -mtpcs-frame -mtpcs-leaf-frame @gol
395 -mcaller-super-interworking -mcallee-super-interworking}
398 @gccoptlist{-mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts @gol
399 -mcall-prologues -mno-tablejump -mtiny-stack -mint8}
402 @gccoptlist{-mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
403 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
404 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
405 -mstack-align -mdata-align -mconst-align @gol
406 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
407 -melf -maout -melinux -mlinux -sim -sim2 @gol
408 -mmul-bug-workaround -mno-mul-bug-workaround}
410 @emph{Darwin Options}
411 @gccoptlist{-all_load -allowable_client -arch -arch_errors_fatal @gol
412 -arch_only -bind_at_load -bundle -bundle_loader @gol
413 -client_name -compatibility_version -current_version @gol
415 -dependency-file -dylib_file -dylinker_install_name @gol
416 -dynamic -dynamiclib -exported_symbols_list @gol
417 -filelist -flat_namespace -force_cpusubtype_ALL @gol
418 -force_flat_namespace -headerpad_max_install_names @gol
419 -image_base -init -install_name -keep_private_externs @gol
420 -multi_module -multiply_defined -multiply_defined_unused @gol
421 -noall_load -no_dead_strip_inits_and_terms @gol
422 -nofixprebinding -nomultidefs -noprebind -noseglinkedit @gol
423 -pagezero_size -prebind -prebind_all_twolevel_modules @gol
424 -private_bundle -read_only_relocs -sectalign @gol
425 -sectobjectsymbols -whyload -seg1addr @gol
426 -sectcreate -sectobjectsymbols -sectorder @gol
427 -segaddr -segs_read_only_addr -segs_read_write_addr @gol
428 -seg_addr_table -seg_addr_table_filename -seglinkedit @gol
429 -segprot -segs_read_only_addr -segs_read_write_addr @gol
430 -single_module -static -sub_library -sub_umbrella @gol
431 -twolevel_namespace -umbrella -undefined @gol
432 -unexported_symbols_list -weak_reference_mismatches @gol
433 -whatsloaded -F -gused -gfull -mone-byte-bool}
435 @emph{DEC Alpha Options}
436 @gccoptlist{-mno-fp-regs -msoft-float -malpha-as -mgas @gol
437 -mieee -mieee-with-inexact -mieee-conformant @gol
438 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
439 -mtrap-precision=@var{mode} -mbuild-constants @gol
440 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} @gol
441 -mbwx -mmax -mfix -mcix @gol
442 -mfloat-vax -mfloat-ieee @gol
443 -mexplicit-relocs -msmall-data -mlarge-data @gol
444 -msmall-text -mlarge-text @gol
445 -mmemory-latency=@var{time}}
447 @emph{DEC Alpha/VMS Options}
448 @gccoptlist{-mvms-return-codes}
451 @gccoptlist{-mgpr-32 -mgpr-64 -mfpr-32 -mfpr-64 @gol
452 -mhard-float -msoft-float @gol
453 -malloc-cc -mfixed-cc -mdword -mno-dword @gol
454 -mdouble -mno-double @gol
455 -mmedia -mno-media -mmuladd -mno-muladd @gol
456 -mfdpic -minline-plt -mgprel-ro -multilib-library-pic @gol
457 -mlinked-fp -mlong-calls -malign-labels @gol
458 -mlibrary-pic -macc-4 -macc-8 @gol
459 -mpack -mno-pack -mno-eflags -mcond-move -mno-cond-move @gol
460 -mscc -mno-scc -mcond-exec -mno-cond-exec @gol
461 -mvliw-branch -mno-vliw-branch @gol
462 -mmulti-cond-exec -mno-multi-cond-exec -mnested-cond-exec @gol
463 -mno-nested-cond-exec -mtomcat-stats @gol
467 @emph{H8/300 Options}
468 @gccoptlist{-mrelax -mh -ms -mn -mint32 -malign-300}
471 @gccoptlist{-march=@var{architecture-type} @gol
472 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
473 -mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
474 -mfixed-range=@var{register-range} @gol
475 -mjump-in-delay -mlinker-opt -mlong-calls @gol
476 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
477 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
478 -mno-jump-in-delay -mno-long-load-store @gol
479 -mno-portable-runtime -mno-soft-float @gol
480 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
481 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
482 -mschedule=@var{cpu-type} -mspace-regs -msio -mwsio @gol
483 -munix=@var{unix-std} -nolibdld -static -threads}
485 @emph{i386 and x86-64 Options}
486 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
487 -mfpmath=@var{unit} @gol
488 -masm=@var{dialect} -mno-fancy-math-387 @gol
489 -mno-fp-ret-in-387 -msoft-float -msvr3-shlib @gol
490 -mno-wide-multiply -mrtd -malign-double @gol
491 -mpreferred-stack-boundary=@var{num} @gol
492 -mmmx -msse -msse2 -msse3 -m3dnow @gol
493 -mthreads -mno-align-stringops -minline-all-stringops @gol
494 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
495 -m96bit-long-double -mregparm=@var{num} -momit-leaf-frame-pointer @gol
496 -mno-red-zone -mno-tls-direct-seg-refs @gol
497 -mcmodel=@var{code-model} @gol
501 @gccoptlist{-mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
502 -mvolatile-asm-stop -mregister-names -mno-sdata @gol
503 -mconstant-gp -mauto-pic -minline-float-divide-min-latency @gol
504 -minline-float-divide-max-throughput @gol
505 -minline-int-divide-min-latency @gol
506 -minline-int-divide-max-throughput @gol
507 -minline-sqrt-min-latency -minline-sqrt-max-throughput @gol
508 -mno-dwarf2-asm -mearly-stop-bits @gol
509 -mfixed-range=@var{register-range} -mtls-size=@var{tls-size} @gol
510 -mtune=@var{cpu-type} -mt -pthread -milp32 -mlp64}
512 @emph{M32R/D Options}
513 @gccoptlist{-m32r2 -m32rx -m32r @gol
515 -malign-loops -mno-align-loops @gol
516 -missue-rate=@var{number} @gol
517 -mbranch-cost=@var{number} @gol
518 -mmodel=@var{code-size-model-type} @gol
519 -msdata=@var{sdata-type} @gol
520 -mno-flush-func -mflush-func=@var{name} @gol
521 -mno-flush-trap -mflush-trap=@var{number} @gol
524 @emph{M680x0 Options}
525 @gccoptlist{-m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
526 -m68060 -mcpu32 -m5200 -m68881 -mbitfield -mc68000 -mc68020 @gol
527 -mnobitfield -mrtd -mshort -msoft-float -mpcrel @gol
528 -malign-int -mstrict-align -msep-data -mno-sep-data @gol
529 -mshared-library-id=n -mid-shared-library -mno-id-shared-library}
531 @emph{M68hc1x Options}
532 @gccoptlist{-m6811 -m6812 -m68hc11 -m68hc12 -m68hcs12 @gol
533 -mauto-incdec -minmax -mlong-calls -mshort @gol
534 -msoft-reg-count=@var{count}}
537 @gccoptlist{-mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
538 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
539 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
540 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
541 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
544 @gccoptlist{-EL -EB -march=@var{arch} -mtune=@var{arch} @gol
545 -mips1 -mips2 -mips3 -mips4 -mips32 -mips32r2 -mips64 @gol
546 -mips16 -mno-mips16 -mabi=@var{abi} -mabicalls -mno-abicalls @gol
547 -mxgot -mno-xgot -mgp32 -mgp64 -mfp32 -mfp64 @gol
548 -mhard-float -msoft-float -msingle-float -mdouble-float @gol
549 -mpaired-single -mips3d @gol
550 -mint64 -mlong64 -mlong32 @gol
551 -G@var{num} -membedded-data -mno-embedded-data @gol
552 -muninit-const-in-rodata -mno-uninit-const-in-rodata @gol
553 -msplit-addresses -mno-split-addresses @gol
554 -mexplicit-relocs -mno-explicit-relocs @gol
555 -mcheck-zero-division -mno-check-zero-division @gol
556 -mdivide-traps -mdivide-breaks @gol
557 -mmemcpy -mno-memcpy -mlong-calls -mno-long-calls @gol
558 -mmad -mno-mad -mfused-madd -mno-fused-madd -nocpp @gol
559 -mfix-r4000 -mno-fix-r4000 -mfix-r4400 -mno-fix-r4400 @gol
560 -mfix-vr4120 -mno-fix-vr4120 -mfix-sb1 -mno-fix-sb1 @gol
561 -mflush-func=@var{func} -mno-flush-func @gol
562 -mbranch-likely -mno-branch-likely @gol
563 -mfp-exceptions -mno-fp-exceptions @gol
564 -mvr4130-align -mno-vr4130-align}
567 @gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
568 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
569 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
570 -mno-base-addresses -msingle-exit -mno-single-exit}
572 @emph{MN10300 Options}
573 @gccoptlist{-mmult-bug -mno-mult-bug @gol
574 -mam33 -mno-am33 @gol
575 -mam33-2 -mno-am33-2 @gol
579 @gccoptlist{-m32032 -m32332 -m32532 -m32081 -m32381 @gol
580 -mmult-add -mnomult-add -msoft-float -mrtd -mnortd @gol
581 -mregparam -mnoregparam -msb -mnosb @gol
582 -mbitfield -mnobitfield -mhimem -mnohimem}
584 @emph{PDP-11 Options}
585 @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
586 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
587 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
588 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
589 -mbranch-expensive -mbranch-cheap @gol
590 -msplit -mno-split -munix-asm -mdec-asm}
592 @emph{PowerPC Options}
593 See RS/6000 and PowerPC Options.
595 @emph{RS/6000 and PowerPC Options}
596 @gccoptlist{-mcpu=@var{cpu-type} @gol
597 -mtune=@var{cpu-type} @gol
598 -mpower -mno-power -mpower2 -mno-power2 @gol
599 -mpowerpc -mpowerpc64 -mno-powerpc @gol
600 -maltivec -mno-altivec @gol
601 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
602 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
603 -mnew-mnemonics -mold-mnemonics @gol
604 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
605 -m64 -m32 -mxl-compat -mno-xl-compat -mpe @gol
606 -malign-power -malign-natural @gol
607 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
608 -mstring -mno-string -mupdate -mno-update @gol
609 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
610 -mstrict-align -mno-strict-align -mrelocatable @gol
611 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
612 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
613 -mdynamic-no-pic @gol
614 -mprioritize-restricted-insns=@var{priority} @gol
615 -msched-costly-dep=@var{dependence_type} @gol
616 -minsert-sched-nops=@var{scheme} @gol
617 -mcall-sysv -mcall-netbsd @gol
618 -maix-struct-return -msvr4-struct-return @gol
619 -mabi=altivec -mabi=no-altivec @gol
620 -mabi=spe -mabi=no-spe @gol
621 -misel=yes -misel=no @gol
622 -mspe=yes -mspe=no @gol
623 -mfloat-gprs=yes -mfloat-gprs=no -mfloat-gprs=single -mfloat-gprs=double @gol
624 -mprototype -mno-prototype @gol
625 -msim -mmvme -mads -myellowknife -memb -msdata @gol
626 -msdata=@var{opt} -mvxworks -mwindiss -G @var{num} -pthread}
628 @emph{S/390 and zSeries Options}
629 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
630 -mhard-float -msoft-float -mbackchain -mno-backchain @gol
631 -mpacked-stack -mno-packed-stack @gol
632 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
633 -m64 -m31 -mdebug -mno-debug -mesa -mzarch @gol
634 -mtpf-trace -mno-tpf-trace -mfused-madd -mno-fused-madd @gol
635 -mwarn-framesize -mwarn-dynamicstack -mstack-size -mstack-guard}
638 @gccoptlist{-m1 -m2 -m2e -m3 -m3e @gol
639 -m4-nofpu -m4-single-only -m4-single -m4 @gol
640 -m4a-nofpu -m4a-single-only -m4a-single -m4a -m4al @gol
641 -m5-64media -m5-64media-nofpu @gol
642 -m5-32media -m5-32media-nofpu @gol
643 -m5-compact -m5-compact-nofpu @gol
644 -mb -ml -mdalign -mrelax @gol
645 -mbigtable -mfmovd -mhitachi -mrenesas -mno-renesas -mnomacsave @gol
646 -mieee -misize -mpadstruct -mspace @gol
647 -mprefergot -musermode}
650 @gccoptlist{-mcpu=@var{cpu-type} @gol
651 -mtune=@var{cpu-type} @gol
652 -mcmodel=@var{code-model} @gol
653 -m32 -m64 -mapp-regs -mno-app-regs @gol
654 -mfaster-structs -mno-faster-structs @gol
655 -mfpu -mno-fpu -mhard-float -msoft-float @gol
656 -mhard-quad-float -msoft-quad-float @gol
657 -mimpure-text -mno-impure-text -mlittle-endian @gol
658 -mstack-bias -mno-stack-bias @gol
659 -munaligned-doubles -mno-unaligned-doubles @gol
660 -mv8plus -mno-v8plus -mvis -mno-vis
663 @emph{System V Options}
664 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
666 @emph{TMS320C3x/C4x Options}
667 @gccoptlist{-mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm @gol
668 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload @gol
669 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned @gol
670 -mparallel-insns -mparallel-mpy -mpreserve-float}
673 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
674 -mprolog-function -mno-prolog-function -mspace @gol
675 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
676 -mapp-regs -mno-app-regs @gol
677 -mdisable-callt -mno-disable-callt @gol
683 @gccoptlist{-mg -mgnu -munix}
685 @emph{x86-64 Options}
686 See i386 and x86-64 Options.
688 @emph{Xstormy16 Options}
691 @emph{Xtensa Options}
692 @gccoptlist{-mconst16 -mno-const16 @gol
693 -mfused-madd -mno-fused-madd @gol
694 -mtext-section-literals -mno-text-section-literals @gol
695 -mtarget-align -mno-target-align @gol
696 -mlongcalls -mno-longcalls}
698 @emph{zSeries Options}
699 See S/390 and zSeries Options.
701 @item Code Generation Options
702 @xref{Code Gen Options,,Options for Code Generation Conventions}.
703 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
704 -ffixed-@var{reg} -fexceptions @gol
705 -fnon-call-exceptions -funwind-tables @gol
706 -fasynchronous-unwind-tables @gol
707 -finhibit-size-directive -finstrument-functions @gol
708 -fno-common -fno-ident @gol
709 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
710 -freg-struct-return -fshared-data -fshort-enums @gol
711 -fshort-double -fshort-wchar @gol
712 -fverbose-asm -fpack-struct[=@var{n}] -fstack-check @gol
713 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
714 -fargument-alias -fargument-noalias @gol
715 -fargument-noalias-global -fleading-underscore @gol
716 -ftls-model=@var{model} @gol
717 -ftrapv -fwrapv -fbounds-check @gol
722 * Overall Options:: Controlling the kind of output:
723 an executable, object files, assembler files,
724 or preprocessed source.
725 * C Dialect Options:: Controlling the variant of C language compiled.
726 * C++ Dialect Options:: Variations on C++.
727 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
729 * Language Independent Options:: Controlling how diagnostics should be
731 * Warning Options:: How picky should the compiler be?
732 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
733 * Optimize Options:: How much optimization?
734 * Preprocessor Options:: Controlling header files and macro definitions.
735 Also, getting dependency information for Make.
736 * Assembler Options:: Passing options to the assembler.
737 * Link Options:: Specifying libraries and so on.
738 * Directory Options:: Where to find header files and libraries.
739 Where to find the compiler executable files.
740 * Spec Files:: How to pass switches to sub-processes.
741 * Target Options:: Running a cross-compiler, or an old version of GCC.
744 @node Overall Options
745 @section Options Controlling the Kind of Output
747 Compilation can involve up to four stages: preprocessing, compilation
748 proper, assembly and linking, always in that order. GCC is capable of
749 preprocessing and compiling several files either into several
750 assembler input files, or into one assembler input file; then each
751 assembler input file produces an object file, and linking combines all
752 the object files (those newly compiled, and those specified as input)
753 into an executable file.
755 @cindex file name suffix
756 For any given input file, the file name suffix determines what kind of
761 C source code which must be preprocessed.
764 C source code which should not be preprocessed.
767 C++ source code which should not be preprocessed.
770 Objective-C source code. Note that you must link with the @file{libobjc}
771 library to make an Objective-C program work.
774 Objective-C source code which should not be preprocessed.
778 Objective-C++ source code. Note that you must link with the @file{libobjc}
779 library to make an Objective-C++ program work. Note that @samp{.M} refers
780 to a literal capital M@.
783 Objective-C++ source code which should not be preprocessed.
786 C, C++, Objective-C or Objective-C++ header file to be turned into a
791 @itemx @var{file}.cxx
792 @itemx @var{file}.cpp
793 @itemx @var{file}.CPP
794 @itemx @var{file}.c++
796 C++ source code which must be preprocessed. Note that in @samp{.cxx},
797 the last two letters must both be literally @samp{x}. Likewise,
798 @samp{.C} refers to a literal capital C@.
802 C++ header file to be turned into a precompiled header.
805 @itemx @var{file}.for
806 @itemx @var{file}.FOR
807 Fortran source code which should not be preprocessed.
810 @itemx @var{file}.fpp
811 @itemx @var{file}.FPP
812 Fortran source code which must be preprocessed (with the traditional
816 Fortran source code which must be preprocessed with a RATFOR
817 preprocessor (not included with GCC)@.
820 @itemx @var{file}.f95
821 Fortran 90/95 source code which should not be preprocessed.
823 @c FIXME: Descriptions of Java file types.
830 Ada source code file which contains a library unit declaration (a
831 declaration of a package, subprogram, or generic, or a generic
832 instantiation), or a library unit renaming declaration (a package,
833 generic, or subprogram renaming declaration). Such files are also
836 @itemx @var{file}.adb
837 Ada source code file containing a library unit body (a subprogram or
838 package body). Such files are also called @dfn{bodies}.
840 @c GCC also knows about some suffixes for languages not yet included:
849 Assembler code which must be preprocessed.
852 An object file to be fed straight into linking.
853 Any file name with no recognized suffix is treated this way.
857 You can specify the input language explicitly with the @option{-x} option:
860 @item -x @var{language}
861 Specify explicitly the @var{language} for the following input files
862 (rather than letting the compiler choose a default based on the file
863 name suffix). This option applies to all following input files until
864 the next @option{-x} option. Possible values for @var{language} are:
866 c c-header c-cpp-output
867 c++ c++-header c++-cpp-output
868 objective-c objective-c-header objective-c-cpp-output
869 objective-c++ objective-c++-header objective-c++-cpp-output
870 assembler assembler-with-cpp
872 f77 f77-cpp-input ratfor
879 Turn off any specification of a language, so that subsequent files are
880 handled according to their file name suffixes (as they are if @option{-x}
881 has not been used at all).
883 @item -pass-exit-codes
884 @opindex pass-exit-codes
885 Normally the @command{gcc} program will exit with the code of 1 if any
886 phase of the compiler returns a non-success return code. If you specify
887 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
888 numerically highest error produced by any phase that returned an error
892 If you only want some of the stages of compilation, you can use
893 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
894 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
895 @command{gcc} is to stop. Note that some combinations (for example,
896 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
901 Compile or assemble the source files, but do not link. The linking
902 stage simply is not done. The ultimate output is in the form of an
903 object file for each source file.
905 By default, the object file name for a source file is made by replacing
906 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
908 Unrecognized input files, not requiring compilation or assembly, are
913 Stop after the stage of compilation proper; do not assemble. The output
914 is in the form of an assembler code file for each non-assembler input
917 By default, the assembler file name for a source file is made by
918 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
920 Input files that don't require compilation are ignored.
924 Stop after the preprocessing stage; do not run the compiler proper. The
925 output is in the form of preprocessed source code, which is sent to the
928 Input files which don't require preprocessing are ignored.
930 @cindex output file option
933 Place output in file @var{file}. This applies regardless to whatever
934 sort of output is being produced, whether it be an executable file,
935 an object file, an assembler file or preprocessed C code.
937 If @option{-o} is not specified, the default is to put an executable
938 file in @file{a.out}, the object file for
939 @file{@var{source}.@var{suffix}} in @file{@var{source}.o}, its
940 assembler file in @file{@var{source}.s}, a precompiled header file in
941 @file{@var{source}.@var{suffix}.gch}, and all preprocessed C source on
946 Print (on standard error output) the commands executed to run the stages
947 of compilation. Also print the version number of the compiler driver
948 program and of the preprocessor and the compiler proper.
952 Like @option{-v} except the commands are not executed and all command
953 arguments are quoted. This is useful for shell scripts to capture the
954 driver-generated command lines.
958 Use pipes rather than temporary files for communication between the
959 various stages of compilation. This fails to work on some systems where
960 the assembler is unable to read from a pipe; but the GNU assembler has
965 If you are compiling multiple source files, this option tells the driver
966 to pass all the source files to the compiler at once (for those
967 languages for which the compiler can handle this). This will allow
968 intermodule analysis (IMA) to be performed by the compiler. Currently the only
969 language for which this is supported is C@. If you pass source files for
970 multiple languages to the driver, using this option, the driver will invoke
971 the compiler(s) that support IMA once each, passing each compiler all the
972 source files appropriate for it. For those languages that do not support
973 IMA this option will be ignored, and the compiler will be invoked once for
974 each source file in that language. If you use this option in conjunction
975 with @option{-save-temps}, the compiler will generate multiple
977 (one for each source file), but only one (combined) @file{.o} or
982 Print (on the standard output) a description of the command line options
983 understood by @command{gcc}. If the @option{-v} option is also specified
984 then @option{--help} will also be passed on to the various processes
985 invoked by @command{gcc}, so that they can display the command line options
986 they accept. If the @option{-Wextra} option is also specified then command
987 line options which have no documentation associated with them will also
992 Print (on the standard output) a description of target specific command
993 line options for each tool.
997 Display the version number and copyrights of the invoked GCC@.
1001 @section Compiling C++ Programs
1003 @cindex suffixes for C++ source
1004 @cindex C++ source file suffixes
1005 C++ source files conventionally use one of the suffixes @samp{.C},
1006 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
1007 @samp{.cxx}; C++ header files often use @samp{.hh} or @samp{.H}; and
1008 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
1009 files with these names and compiles them as C++ programs even if you
1010 call the compiler the same way as for compiling C programs (usually
1011 with the name @command{gcc}).
1015 However, C++ programs often require class libraries as well as a
1016 compiler that understands the C++ language---and under some
1017 circumstances, you might want to compile programs or header files from
1018 standard input, or otherwise without a suffix that flags them as C++
1019 programs. You might also like to precompile a C header file with a
1020 @samp{.h} extension to be used in C++ compilations. @command{g++} is a
1021 program that calls GCC with the default language set to C++, and
1022 automatically specifies linking against the C++ library. On many
1023 systems, @command{g++} is also installed with the name @command{c++}.
1025 @cindex invoking @command{g++}
1026 When you compile C++ programs, you may specify many of the same
1027 command-line options that you use for compiling programs in any
1028 language; or command-line options meaningful for C and related
1029 languages; or options that are meaningful only for C++ programs.
1030 @xref{C Dialect Options,,Options Controlling C Dialect}, for
1031 explanations of options for languages related to C@.
1032 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
1033 explanations of options that are meaningful only for C++ programs.
1035 @node C Dialect Options
1036 @section Options Controlling C Dialect
1037 @cindex dialect options
1038 @cindex language dialect options
1039 @cindex options, dialect
1041 The following options control the dialect of C (or languages derived
1042 from C, such as C++, Objective-C and Objective-C++) that the compiler
1046 @cindex ANSI support
1050 In C mode, support all ISO C90 programs. In C++ mode,
1051 remove GNU extensions that conflict with ISO C++.
1053 This turns off certain features of GCC that are incompatible with ISO
1054 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
1055 such as the @code{asm} and @code{typeof} keywords, and
1056 predefined macros such as @code{unix} and @code{vax} that identify the
1057 type of system you are using. It also enables the undesirable and
1058 rarely used ISO trigraph feature. For the C compiler,
1059 it disables recognition of C++ style @samp{//} comments as well as
1060 the @code{inline} keyword.
1062 The alternate keywords @code{__asm__}, @code{__extension__},
1063 @code{__inline__} and @code{__typeof__} continue to work despite
1064 @option{-ansi}. You would not want to use them in an ISO C program, of
1065 course, but it is useful to put them in header files that might be included
1066 in compilations done with @option{-ansi}. Alternate predefined macros
1067 such as @code{__unix__} and @code{__vax__} are also available, with or
1068 without @option{-ansi}.
1070 The @option{-ansi} option does not cause non-ISO programs to be
1071 rejected gratuitously. For that, @option{-pedantic} is required in
1072 addition to @option{-ansi}. @xref{Warning Options}.
1074 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1075 option is used. Some header files may notice this macro and refrain
1076 from declaring certain functions or defining certain macros that the
1077 ISO standard doesn't call for; this is to avoid interfering with any
1078 programs that might use these names for other things.
1080 Functions which would normally be built in but do not have semantics
1081 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1082 functions with @option{-ansi} is used. @xref{Other Builtins,,Other
1083 built-in functions provided by GCC}, for details of the functions
1088 Determine the language standard. This option is currently only
1089 supported when compiling C or C++. A value for this option must be
1090 provided; possible values are
1095 ISO C90 (same as @option{-ansi}).
1097 @item iso9899:199409
1098 ISO C90 as modified in amendment 1.
1104 ISO C99. Note that this standard is not yet fully supported; see
1105 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1106 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1109 Default, ISO C90 plus GNU extensions (including some C99 features).
1113 ISO C99 plus GNU extensions. When ISO C99 is fully implemented in GCC,
1114 this will become the default. The name @samp{gnu9x} is deprecated.
1117 The 1998 ISO C++ standard plus amendments.
1120 The same as @option{-std=c++98} plus GNU extensions. This is the
1121 default for C++ code.
1124 Even when this option is not specified, you can still use some of the
1125 features of newer standards in so far as they do not conflict with
1126 previous C standards. For example, you may use @code{__restrict__} even
1127 when @option{-std=c99} is not specified.
1129 The @option{-std} options specifying some version of ISO C have the same
1130 effects as @option{-ansi}, except that features that were not in ISO C90
1131 but are in the specified version (for example, @samp{//} comments and
1132 the @code{inline} keyword in ISO C99) are not disabled.
1134 @xref{Standards,,Language Standards Supported by GCC}, for details of
1135 these standard versions.
1137 @item -aux-info @var{filename}
1139 Output to the given filename prototyped declarations for all functions
1140 declared and/or defined in a translation unit, including those in header
1141 files. This option is silently ignored in any language other than C@.
1143 Besides declarations, the file indicates, in comments, the origin of
1144 each declaration (source file and line), whether the declaration was
1145 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1146 @samp{O} for old, respectively, in the first character after the line
1147 number and the colon), and whether it came from a declaration or a
1148 definition (@samp{C} or @samp{F}, respectively, in the following
1149 character). In the case of function definitions, a K&R-style list of
1150 arguments followed by their declarations is also provided, inside
1151 comments, after the declaration.
1155 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1156 keyword, so that code can use these words as identifiers. You can use
1157 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1158 instead. @option{-ansi} implies @option{-fno-asm}.
1160 In C++, this switch only affects the @code{typeof} keyword, since
1161 @code{asm} and @code{inline} are standard keywords. You may want to
1162 use the @option{-fno-gnu-keywords} flag instead, which has the same
1163 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1164 switch only affects the @code{asm} and @code{typeof} keywords, since
1165 @code{inline} is a standard keyword in ISO C99.
1168 @itemx -fno-builtin-@var{function}
1169 @opindex fno-builtin
1170 @cindex built-in functions
1171 Don't recognize built-in functions that do not begin with
1172 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1173 functions provided by GCC}, for details of the functions affected,
1174 including those which are not built-in functions when @option{-ansi} or
1175 @option{-std} options for strict ISO C conformance are used because they
1176 do not have an ISO standard meaning.
1178 GCC normally generates special code to handle certain built-in functions
1179 more efficiently; for instance, calls to @code{alloca} may become single
1180 instructions that adjust the stack directly, and calls to @code{memcpy}
1181 may become inline copy loops. The resulting code is often both smaller
1182 and faster, but since the function calls no longer appear as such, you
1183 cannot set a breakpoint on those calls, nor can you change the behavior
1184 of the functions by linking with a different library. In addition,
1185 when a function is recognized as a built-in function, GCC may use
1186 information about that function to warn about problems with calls to
1187 that function, or to generate more efficient code, even if the
1188 resulting code still contains calls to that function. For example,
1189 warnings are given with @option{-Wformat} for bad calls to
1190 @code{printf}, when @code{printf} is built in, and @code{strlen} is
1191 known not to modify global memory.
1193 With the @option{-fno-builtin-@var{function}} option
1194 only the built-in function @var{function} is
1195 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1196 function is named this is not built-in in this version of GCC, this
1197 option is ignored. There is no corresponding
1198 @option{-fbuiltin-@var{function}} option; if you wish to enable
1199 built-in functions selectively when using @option{-fno-builtin} or
1200 @option{-ffreestanding}, you may define macros such as:
1203 #define abs(n) __builtin_abs ((n))
1204 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1209 @cindex hosted environment
1211 Assert that compilation takes place in a hosted environment. This implies
1212 @option{-fbuiltin}. A hosted environment is one in which the
1213 entire standard library is available, and in which @code{main} has a return
1214 type of @code{int}. Examples are nearly everything except a kernel.
1215 This is equivalent to @option{-fno-freestanding}.
1217 @item -ffreestanding
1218 @opindex ffreestanding
1219 @cindex hosted environment
1221 Assert that compilation takes place in a freestanding environment. This
1222 implies @option{-fno-builtin}. A freestanding environment
1223 is one in which the standard library may not exist, and program startup may
1224 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1225 This is equivalent to @option{-fno-hosted}.
1227 @xref{Standards,,Language Standards Supported by GCC}, for details of
1228 freestanding and hosted environments.
1230 @item -fms-extensions
1231 @opindex fms-extensions
1232 Accept some non-standard constructs used in Microsoft header files.
1234 Some cases of unnamed fields in structures and unions are only
1235 accepted with this option. @xref{Unnamed Fields,,Unnamed struct/union
1236 fields within structs/unions}, for details.
1240 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1241 options for strict ISO C conformance) implies @option{-trigraphs}.
1243 @item -no-integrated-cpp
1244 @opindex no-integrated-cpp
1245 Performs a compilation in two passes: preprocessing and compiling. This
1246 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1247 @option{-B} option. The user supplied compilation step can then add in
1248 an additional preprocessing step after normal preprocessing but before
1249 compiling. The default is to use the integrated cpp (internal cpp)
1251 The semantics of this option will change if "cc1", "cc1plus", and
1252 "cc1obj" are merged.
1254 @cindex traditional C language
1255 @cindex C language, traditional
1257 @itemx -traditional-cpp
1258 @opindex traditional-cpp
1259 @opindex traditional
1260 Formerly, these options caused GCC to attempt to emulate a pre-standard
1261 C compiler. They are now only supported with the @option{-E} switch.
1262 The preprocessor continues to support a pre-standard mode. See the GNU
1263 CPP manual for details.
1265 @item -fcond-mismatch
1266 @opindex fcond-mismatch
1267 Allow conditional expressions with mismatched types in the second and
1268 third arguments. The value of such an expression is void. This option
1269 is not supported for C++.
1271 @item -funsigned-char
1272 @opindex funsigned-char
1273 Let the type @code{char} be unsigned, like @code{unsigned char}.
1275 Each kind of machine has a default for what @code{char} should
1276 be. It is either like @code{unsigned char} by default or like
1277 @code{signed char} by default.
1279 Ideally, a portable program should always use @code{signed char} or
1280 @code{unsigned char} when it depends on the signedness of an object.
1281 But many programs have been written to use plain @code{char} and
1282 expect it to be signed, or expect it to be unsigned, depending on the
1283 machines they were written for. This option, and its inverse, let you
1284 make such a program work with the opposite default.
1286 The type @code{char} is always a distinct type from each of
1287 @code{signed char} or @code{unsigned char}, even though its behavior
1288 is always just like one of those two.
1291 @opindex fsigned-char
1292 Let the type @code{char} be signed, like @code{signed char}.
1294 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1295 the negative form of @option{-funsigned-char}. Likewise, the option
1296 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1298 @item -fsigned-bitfields
1299 @itemx -funsigned-bitfields
1300 @itemx -fno-signed-bitfields
1301 @itemx -fno-unsigned-bitfields
1302 @opindex fsigned-bitfields
1303 @opindex funsigned-bitfields
1304 @opindex fno-signed-bitfields
1305 @opindex fno-unsigned-bitfields
1306 These options control whether a bit-field is signed or unsigned, when the
1307 declaration does not use either @code{signed} or @code{unsigned}. By
1308 default, such a bit-field is signed, because this is consistent: the
1309 basic integer types such as @code{int} are signed types.
1312 @node C++ Dialect Options
1313 @section Options Controlling C++ Dialect
1315 @cindex compiler options, C++
1316 @cindex C++ options, command line
1317 @cindex options, C++
1318 This section describes the command-line options that are only meaningful
1319 for C++ programs; but you can also use most of the GNU compiler options
1320 regardless of what language your program is in. For example, you
1321 might compile a file @code{firstClass.C} like this:
1324 g++ -g -frepo -O -c firstClass.C
1328 In this example, only @option{-frepo} is an option meant
1329 only for C++ programs; you can use the other options with any
1330 language supported by GCC@.
1332 Here is a list of options that are @emph{only} for compiling C++ programs:
1336 @item -fabi-version=@var{n}
1337 @opindex fabi-version
1338 Use version @var{n} of the C++ ABI@. Version 2 is the version of the
1339 C++ ABI that first appeared in G++ 3.4. Version 1 is the version of
1340 the C++ ABI that first appeared in G++ 3.2. Version 0 will always be
1341 the version that conforms most closely to the C++ ABI specification.
1342 Therefore, the ABI obtained using version 0 will change as ABI bugs
1345 The default is version 2.
1347 @item -fno-access-control
1348 @opindex fno-access-control
1349 Turn off all access checking. This switch is mainly useful for working
1350 around bugs in the access control code.
1354 Check that the pointer returned by @code{operator new} is non-null
1355 before attempting to modify the storage allocated. This check is
1356 normally unnecessary because the C++ standard specifies that
1357 @code{operator new} will only return @code{0} if it is declared
1358 @samp{throw()}, in which case the compiler will always check the
1359 return value even without this option. In all other cases, when
1360 @code{operator new} has a non-empty exception specification, memory
1361 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1362 @samp{new (nothrow)}.
1364 @item -fconserve-space
1365 @opindex fconserve-space
1366 Put uninitialized or runtime-initialized global variables into the
1367 common segment, as C does. This saves space in the executable at the
1368 cost of not diagnosing duplicate definitions. If you compile with this
1369 flag and your program mysteriously crashes after @code{main()} has
1370 completed, you may have an object that is being destroyed twice because
1371 two definitions were merged.
1373 This option is no longer useful on most targets, now that support has
1374 been added for putting variables into BSS without making them common.
1376 @item -fno-const-strings
1377 @opindex fno-const-strings
1378 Give string constants type @code{char *} instead of type @code{const
1379 char *}. By default, G++ uses type @code{const char *} as required by
1380 the standard. Even if you use @option{-fno-const-strings}, you cannot
1381 actually modify the value of a string constant.
1383 This option might be removed in a future release of G++. For maximum
1384 portability, you should structure your code so that it works with
1385 string constants that have type @code{const char *}.
1387 @item -fno-elide-constructors
1388 @opindex fno-elide-constructors
1389 The C++ standard allows an implementation to omit creating a temporary
1390 which is only used to initialize another object of the same type.
1391 Specifying this option disables that optimization, and forces G++ to
1392 call the copy constructor in all cases.
1394 @item -fno-enforce-eh-specs
1395 @opindex fno-enforce-eh-specs
1396 Don't check for violation of exception specifications at runtime. This
1397 option violates the C++ standard, but may be useful for reducing code
1398 size in production builds, much like defining @samp{NDEBUG}. The compiler
1399 will still optimize based on the exception specifications.
1402 @itemx -fno-for-scope
1404 @opindex fno-for-scope
1405 If @option{-ffor-scope} is specified, the scope of variables declared in
1406 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1407 as specified by the C++ standard.
1408 If @option{-fno-for-scope} is specified, the scope of variables declared in
1409 a @i{for-init-statement} extends to the end of the enclosing scope,
1410 as was the case in old versions of G++, and other (traditional)
1411 implementations of C++.
1413 The default if neither flag is given to follow the standard,
1414 but to allow and give a warning for old-style code that would
1415 otherwise be invalid, or have different behavior.
1417 @item -fno-gnu-keywords
1418 @opindex fno-gnu-keywords
1419 Do not recognize @code{typeof} as a keyword, so that code can use this
1420 word as an identifier. You can use the keyword @code{__typeof__} instead.
1421 @option{-ansi} implies @option{-fno-gnu-keywords}.
1423 @item -fno-implicit-templates
1424 @opindex fno-implicit-templates
1425 Never emit code for non-inline templates which are instantiated
1426 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1427 @xref{Template Instantiation}, for more information.
1429 @item -fno-implicit-inline-templates
1430 @opindex fno-implicit-inline-templates
1431 Don't emit code for implicit instantiations of inline templates, either.
1432 The default is to handle inlines differently so that compiles with and
1433 without optimization will need the same set of explicit instantiations.
1435 @item -fno-implement-inlines
1436 @opindex fno-implement-inlines
1437 To save space, do not emit out-of-line copies of inline functions
1438 controlled by @samp{#pragma implementation}. This will cause linker
1439 errors if these functions are not inlined everywhere they are called.
1441 @item -fms-extensions
1442 @opindex fms-extensions
1443 Disable pedantic warnings about constructs used in MFC, such as implicit
1444 int and getting a pointer to member function via non-standard syntax.
1446 @item -fno-nonansi-builtins
1447 @opindex fno-nonansi-builtins
1448 Disable built-in declarations of functions that are not mandated by
1449 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1450 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1452 @item -fno-operator-names
1453 @opindex fno-operator-names
1454 Do not treat the operator name keywords @code{and}, @code{bitand},
1455 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1456 synonyms as keywords.
1458 @item -fno-optional-diags
1459 @opindex fno-optional-diags
1460 Disable diagnostics that the standard says a compiler does not need to
1461 issue. Currently, the only such diagnostic issued by G++ is the one for
1462 a name having multiple meanings within a class.
1465 @opindex fpermissive
1466 Downgrade some diagnostics about nonconformant code from errors to
1467 warnings. Thus, using @option{-fpermissive} will allow some
1468 nonconforming code to compile.
1472 Enable automatic template instantiation at link time. This option also
1473 implies @option{-fno-implicit-templates}. @xref{Template
1474 Instantiation}, for more information.
1478 Disable generation of information about every class with virtual
1479 functions for use by the C++ runtime type identification features
1480 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1481 of the language, you can save some space by using this flag. Note that
1482 exception handling uses the same information, but it will generate it as
1487 Emit statistics about front-end processing at the end of the compilation.
1488 This information is generally only useful to the G++ development team.
1490 @item -ftemplate-depth-@var{n}
1491 @opindex ftemplate-depth
1492 Set the maximum instantiation depth for template classes to @var{n}.
1493 A limit on the template instantiation depth is needed to detect
1494 endless recursions during template class instantiation. ANSI/ISO C++
1495 conforming programs must not rely on a maximum depth greater than 17.
1497 @item -fno-threadsafe-statics
1498 @opindex fno-threadsafe-statics
1499 Do not emit the extra code to use the routines specified in the C++
1500 ABI for thread-safe initialization of local statics. You can use this
1501 option to reduce code size slightly in code that doesn't need to be
1504 @item -fuse-cxa-atexit
1505 @opindex fuse-cxa-atexit
1506 Register destructors for objects with static storage duration with the
1507 @code{__cxa_atexit} function rather than the @code{atexit} function.
1508 This option is required for fully standards-compliant handling of static
1509 destructors, but will only work if your C library supports
1510 @code{__cxa_atexit}.
1512 @item -fvisibility-inlines-hidden
1513 @opindex fvisibility-inlines-hidden
1514 Causes all inlined methods to be marked with
1515 @code{__attribute__ ((visibility ("hidden")))} so that they do not
1516 appear in the export table of a DSO and do not require a PLT indirection
1517 when used within the DSO@. Enabling this option can have a dramatic effect
1518 on load and link times of a DSO as it massively reduces the size of the
1519 dynamic export table when the library makes heavy use of templates. While
1520 it can cause bloating through duplication of code within each DSO where
1521 it is used, often the wastage is less than the considerable space occupied
1522 by a long symbol name in the export table which is typical when using
1523 templates and namespaces. For even more savings, combine with the
1524 @option{-fvisibility=hidden} switch.
1528 Do not use weak symbol support, even if it is provided by the linker.
1529 By default, G++ will use weak symbols if they are available. This
1530 option exists only for testing, and should not be used by end-users;
1531 it will result in inferior code and has no benefits. This option may
1532 be removed in a future release of G++.
1536 Do not search for header files in the standard directories specific to
1537 C++, but do still search the other standard directories. (This option
1538 is used when building the C++ library.)
1541 In addition, these optimization, warning, and code generation options
1542 have meanings only for C++ programs:
1545 @item -fno-default-inline
1546 @opindex fno-default-inline
1547 Do not assume @samp{inline} for functions defined inside a class scope.
1548 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1549 functions will have linkage like inline functions; they just won't be
1552 @item -Wabi @r{(C++ only)}
1554 Warn when G++ generates code that is probably not compatible with the
1555 vendor-neutral C++ ABI@. Although an effort has been made to warn about
1556 all such cases, there are probably some cases that are not warned about,
1557 even though G++ is generating incompatible code. There may also be
1558 cases where warnings are emitted even though the code that is generated
1561 You should rewrite your code to avoid these warnings if you are
1562 concerned about the fact that code generated by G++ may not be binary
1563 compatible with code generated by other compilers.
1565 The known incompatibilities at this point include:
1570 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
1571 pack data into the same byte as a base class. For example:
1574 struct A @{ virtual void f(); int f1 : 1; @};
1575 struct B : public A @{ int f2 : 1; @};
1579 In this case, G++ will place @code{B::f2} into the same byte
1580 as@code{A::f1}; other compilers will not. You can avoid this problem
1581 by explicitly padding @code{A} so that its size is a multiple of the
1582 byte size on your platform; that will cause G++ and other compilers to
1583 layout @code{B} identically.
1586 Incorrect handling of tail-padding for virtual bases. G++ does not use
1587 tail padding when laying out virtual bases. For example:
1590 struct A @{ virtual void f(); char c1; @};
1591 struct B @{ B(); char c2; @};
1592 struct C : public A, public virtual B @{@};
1596 In this case, G++ will not place @code{B} into the tail-padding for
1597 @code{A}; other compilers will. You can avoid this problem by
1598 explicitly padding @code{A} so that its size is a multiple of its
1599 alignment (ignoring virtual base classes); that will cause G++ and other
1600 compilers to layout @code{C} identically.
1603 Incorrect handling of bit-fields with declared widths greater than that
1604 of their underlying types, when the bit-fields appear in a union. For
1608 union U @{ int i : 4096; @};
1612 Assuming that an @code{int} does not have 4096 bits, G++ will make the
1613 union too small by the number of bits in an @code{int}.
1616 Empty classes can be placed at incorrect offsets. For example:
1626 struct C : public B, public A @{@};
1630 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
1631 it should be placed at offset zero. G++ mistakenly believes that the
1632 @code{A} data member of @code{B} is already at offset zero.
1635 Names of template functions whose types involve @code{typename} or
1636 template template parameters can be mangled incorrectly.
1639 template <typename Q>
1640 void f(typename Q::X) @{@}
1642 template <template <typename> class Q>
1643 void f(typename Q<int>::X) @{@}
1647 Instantiations of these templates may be mangled incorrectly.
1651 @item -Wctor-dtor-privacy @r{(C++ only)}
1652 @opindex Wctor-dtor-privacy
1653 Warn when a class seems unusable because all the constructors or
1654 destructors in that class are private, and it has neither friends nor
1655 public static member functions.
1657 @item -Wnon-virtual-dtor @r{(C++ only)}
1658 @opindex Wnon-virtual-dtor
1659 Warn when a class appears to be polymorphic, thereby requiring a virtual
1660 destructor, yet it declares a non-virtual one.
1661 This warning is enabled by @option{-Wall}.
1663 @item -Wreorder @r{(C++ only)}
1665 @cindex reordering, warning
1666 @cindex warning for reordering of member initializers
1667 Warn when the order of member initializers given in the code does not
1668 match the order in which they must be executed. For instance:
1674 A(): j (0), i (1) @{ @}
1678 The compiler will rearrange the member initializers for @samp{i}
1679 and @samp{j} to match the declaration order of the members, emitting
1680 a warning to that effect. This warning is enabled by @option{-Wall}.
1683 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
1686 @item -Weffc++ @r{(C++ only)}
1688 Warn about violations of the following style guidelines from Scott Meyers'
1689 @cite{Effective C++} book:
1693 Item 11: Define a copy constructor and an assignment operator for classes
1694 with dynamically allocated memory.
1697 Item 12: Prefer initialization to assignment in constructors.
1700 Item 14: Make destructors virtual in base classes.
1703 Item 15: Have @code{operator=} return a reference to @code{*this}.
1706 Item 23: Don't try to return a reference when you must return an object.
1710 Also warn about violations of the following style guidelines from
1711 Scott Meyers' @cite{More Effective C++} book:
1715 Item 6: Distinguish between prefix and postfix forms of increment and
1716 decrement operators.
1719 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
1723 When selecting this option, be aware that the standard library
1724 headers do not obey all of these guidelines; use @samp{grep -v}
1725 to filter out those warnings.
1727 @item -Wno-deprecated @r{(C++ only)}
1728 @opindex Wno-deprecated
1729 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1731 @item -Wno-non-template-friend @r{(C++ only)}
1732 @opindex Wno-non-template-friend
1733 Disable warnings when non-templatized friend functions are declared
1734 within a template. Since the advent of explicit template specification
1735 support in G++, if the name of the friend is an unqualified-id (i.e.,
1736 @samp{friend foo(int)}), the C++ language specification demands that the
1737 friend declare or define an ordinary, nontemplate function. (Section
1738 14.5.3). Before G++ implemented explicit specification, unqualified-ids
1739 could be interpreted as a particular specialization of a templatized
1740 function. Because this non-conforming behavior is no longer the default
1741 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
1742 check existing code for potential trouble spots and is on by default.
1743 This new compiler behavior can be turned off with
1744 @option{-Wno-non-template-friend} which keeps the conformant compiler code
1745 but disables the helpful warning.
1747 @item -Wold-style-cast @r{(C++ only)}
1748 @opindex Wold-style-cast
1749 Warn if an old-style (C-style) cast to a non-void type is used within
1750 a C++ program. The new-style casts (@samp{static_cast},
1751 @samp{reinterpret_cast}, and @samp{const_cast}) are less vulnerable to
1752 unintended effects and much easier to search for.
1754 @item -Woverloaded-virtual @r{(C++ only)}
1755 @opindex Woverloaded-virtual
1756 @cindex overloaded virtual fn, warning
1757 @cindex warning for overloaded virtual fn
1758 Warn when a function declaration hides virtual functions from a
1759 base class. For example, in:
1766 struct B: public A @{
1771 the @code{A} class version of @code{f} is hidden in @code{B}, and code
1779 will fail to compile.
1781 @item -Wno-pmf-conversions @r{(C++ only)}
1782 @opindex Wno-pmf-conversions
1783 Disable the diagnostic for converting a bound pointer to member function
1786 @item -Wsign-promo @r{(C++ only)}
1787 @opindex Wsign-promo
1788 Warn when overload resolution chooses a promotion from unsigned or
1789 enumerated type to a signed type, over a conversion to an unsigned type of
1790 the same size. Previous versions of G++ would try to preserve
1791 unsignedness, but the standard mandates the current behavior.
1793 @item -Wsynth @r{(C++ only)}
1795 @cindex warning for synthesized methods
1796 @cindex synthesized methods, warning
1797 Warn when G++'s synthesis behavior does not match that of cfront. For
1803 A& operator = (int);
1813 In this example, G++ will synthesize a default @samp{A& operator =
1814 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1817 @node Objective-C and Objective-C++ Dialect Options
1818 @section Options Controlling Objective-C and Objective-C++ Dialects
1820 @cindex compiler options, Objective-C and Objective-C++
1821 @cindex Objective-C and Objective-C++ options, command line
1822 @cindex options, Objective-C and Objective-C++
1823 (NOTE: This manual does not describe the Objective-C and Objective-C++
1824 languages themselves. See @xref{Standards,,Language Standards
1825 Supported by GCC}, for references.)
1827 This section describes the command-line options that are only meaningful
1828 for Objective-C and Objective-C++ programs, but you can also use most of
1829 the language-independent GNU compiler options.
1830 For example, you might compile a file @code{some_class.m} like this:
1833 gcc -g -fgnu-runtime -O -c some_class.m
1837 In this example, @option{-fgnu-runtime} is an option meant only for
1838 Objective-C and Objective-C++ programs; you can use the other options with
1839 any language supported by GCC@.
1841 Note that since Objective-C is an extension of the C language, Objective-C
1842 compilations may also use options specific to the C front-end (e.g.,
1843 @option{-Wtraditional}). Similarly, Objective-C++ compilations may use
1844 C++-specific options (e.g., @option{-Wabi}).
1846 Here is a list of options that are @emph{only} for compiling Objective-C
1847 and Objective-C++ programs:
1850 @item -fconstant-string-class=@var{class-name}
1851 @opindex fconstant-string-class
1852 Use @var{class-name} as the name of the class to instantiate for each
1853 literal string specified with the syntax @code{@@"@dots{}"}. The default
1854 class name is @code{NXConstantString} if the GNU runtime is being used, and
1855 @code{NSConstantString} if the NeXT runtime is being used (see below). The
1856 @option{-fconstant-cfstrings} option, if also present, will override the
1857 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
1858 to be laid out as constant CoreFoundation strings.
1861 @opindex fgnu-runtime
1862 Generate object code compatible with the standard GNU Objective-C
1863 runtime. This is the default for most types of systems.
1865 @item -fnext-runtime
1866 @opindex fnext-runtime
1867 Generate output compatible with the NeXT runtime. This is the default
1868 for NeXT-based systems, including Darwin and Mac OS X@. The macro
1869 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
1872 @item -fno-nil-receivers
1873 @opindex fno-nil-receivers
1874 Assume that all Objective-C message dispatches (e.g.,
1875 @code{[receiver message:arg]}) in this translation unit ensure that the receiver
1876 is not @code{nil}. This allows for more efficient entry points in the runtime
1877 to be used. Currently, this option is only available in conjunction with
1878 the NeXT runtime on Mac OS X 10.3 and later.
1880 @item -fobjc-exceptions
1881 @opindex fobjc-exceptions
1882 Enable syntactic support for structured exception handling in Objective-C,
1883 similar to what is offered by C++ and Java. Currently, this option is only
1884 available in conjunction with the NeXT runtime on Mac OS X 10.3 and later.
1892 @@catch (AnObjCClass *exc) @{
1899 @@catch (AnotherClass *exc) @{
1902 @@catch (id allOthers) @{
1912 The @code{@@throw} statement may appear anywhere in an Objective-C or
1913 Objective-C++ program; when used inside of a @code{@@catch} block, the
1914 @code{@@throw} may appear without an argument (as shown above), in which case
1915 the object caught by the @code{@@catch} will be rethrown.
1917 Note that only (pointers to) Objective-C objects may be thrown and
1918 caught using this scheme. When an object is thrown, it will be caught
1919 by the nearest @code{@@catch} clause capable of handling objects of that type,
1920 analogously to how @code{catch} blocks work in C++ and Java. A
1921 @code{@@catch(id @dots{})} clause (as shown above) may also be provided to catch
1922 any and all Objective-C exceptions not caught by previous @code{@@catch}
1925 The @code{@@finally} clause, if present, will be executed upon exit from the
1926 immediately preceding @code{@@try @dots{} @@catch} section. This will happen
1927 regardless of whether any exceptions are thrown, caught or rethrown
1928 inside the @code{@@try @dots{} @@catch} section, analogously to the behavior
1929 of the @code{finally} clause in Java.
1931 There are several caveats to using the new exception mechanism:
1935 Although currently designed to be binary compatible with @code{NS_HANDLER}-style
1936 idioms provided by the @code{NSException} class, the new
1937 exceptions can only be used on Mac OS X 10.3 (Panther) and later
1938 systems, due to additional functionality needed in the (NeXT) Objective-C
1942 As mentioned above, the new exceptions do not support handling
1943 types other than Objective-C objects. Furthermore, when used from
1944 Objective-C++, the Objective-C exception model does not interoperate with C++
1945 exceptions at this time. This means you cannot @code{@@throw} an exception
1946 from Objective-C and @code{catch} it in C++, or vice versa
1947 (i.e., @code{throw @dots{} @@catch}).
1950 The @option{-fobjc-exceptions} switch also enables the use of synchronization
1951 blocks for thread-safe execution:
1954 @@synchronized (ObjCClass *guard) @{
1959 Upon entering the @code{@@synchronized} block, a thread of execution shall
1960 first check whether a lock has been placed on the corresponding @code{guard}
1961 object by another thread. If it has, the current thread shall wait until
1962 the other thread relinquishes its lock. Once @code{guard} becomes available,
1963 the current thread will place its own lock on it, execute the code contained in
1964 the @code{@@synchronized} block, and finally relinquish the lock (thereby
1965 making @code{guard} available to other threads).
1967 Unlike Java, Objective-C does not allow for entire methods to be marked
1968 @code{@@synchronized}. Note that throwing exceptions out of
1969 @code{@@synchronized} blocks is allowed, and will cause the guarding object
1970 to be unlocked properly.
1972 @item -freplace-objc-classes
1973 @opindex freplace-objc-classes
1974 Emit a special marker instructing @command{ld(1)} not to statically link in
1975 the resulting object file, and allow @command{dyld(1)} to load it in at
1976 run time instead. This is used in conjunction with the Fix-and-Continue
1977 debugging mode, where the object file in question may be recompiled and
1978 dynamically reloaded in the course of program execution, without the need
1979 to restart the program itself. Currently, Fix-and-Continue functionality
1980 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
1985 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
1986 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
1987 compile time) with static class references that get initialized at load time,
1988 which improves run-time performance. Specifying the @option{-fzero-link} flag
1989 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
1990 to be retained. This is useful in Zero-Link debugging mode, since it allows
1991 for individual class implementations to be modified during program execution.
1995 Dump interface declarations for all classes seen in the source file to a
1996 file named @file{@var{sourcename}.decl}.
1999 @opindex Wno-protocol
2000 If a class is declared to implement a protocol, a warning is issued for
2001 every method in the protocol that is not implemented by the class. The
2002 default behavior is to issue a warning for every method not explicitly
2003 implemented in the class, even if a method implementation is inherited
2004 from the superclass. If you use the @option{-Wno-protocol} option, then
2005 methods inherited from the superclass are considered to be implemented,
2006 and no warning is issued for them.
2010 Warn if multiple methods of different types for the same selector are
2011 found during compilation. The check is performed on the list of methods
2012 in the final stage of compilation. Additionally, a check is performed
2013 for each selector appearing in a @code{@@selector(@dots{})}
2014 expression, and a corresponding method for that selector has been found
2015 during compilation. Because these checks scan the method table only at
2016 the end of compilation, these warnings are not produced if the final
2017 stage of compilation is not reached, for example because an error is
2018 found during compilation, or because the @option{-fsyntax-only} option is
2021 @item -Wundeclared-selector
2022 @opindex Wundeclared-selector
2023 Warn if a @code{@@selector(@dots{})} expression referring to an
2024 undeclared selector is found. A selector is considered undeclared if no
2025 method with that name has been declared before the
2026 @code{@@selector(@dots{})} expression, either explicitly in an
2027 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
2028 an @code{@@implementation} section. This option always performs its
2029 checks as soon as a @code{@@selector(@dots{})} expression is found,
2030 while @option{-Wselector} only performs its checks in the final stage of
2031 compilation. This also enforces the coding style convention
2032 that methods and selectors must be declared before being used.
2034 @item -print-objc-runtime-info
2035 @opindex print-objc-runtime-info
2036 Generate C header describing the largest structure that is passed by
2041 @node Language Independent Options
2042 @section Options to Control Diagnostic Messages Formatting
2043 @cindex options to control diagnostics formatting
2044 @cindex diagnostic messages
2045 @cindex message formatting
2047 Traditionally, diagnostic messages have been formatted irrespective of
2048 the output device's aspect (e.g.@: its width, @dots{}). The options described
2049 below can be used to control the diagnostic messages formatting
2050 algorithm, e.g.@: how many characters per line, how often source location
2051 information should be reported. Right now, only the C++ front end can
2052 honor these options. However it is expected, in the near future, that
2053 the remaining front ends would be able to digest them correctly.
2056 @item -fmessage-length=@var{n}
2057 @opindex fmessage-length
2058 Try to format error messages so that they fit on lines of about @var{n}
2059 characters. The default is 72 characters for @command{g++} and 0 for the rest of
2060 the front ends supported by GCC@. If @var{n} is zero, then no
2061 line-wrapping will be done; each error message will appear on a single
2064 @opindex fdiagnostics-show-location
2065 @item -fdiagnostics-show-location=once
2066 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
2067 reporter to emit @emph{once} source location information; that is, in
2068 case the message is too long to fit on a single physical line and has to
2069 be wrapped, the source location won't be emitted (as prefix) again,
2070 over and over, in subsequent continuation lines. This is the default
2073 @item -fdiagnostics-show-location=every-line
2074 Only meaningful in line-wrapping mode. Instructs the diagnostic
2075 messages reporter to emit the same source location information (as
2076 prefix) for physical lines that result from the process of breaking
2077 a message which is too long to fit on a single line.
2081 @node Warning Options
2082 @section Options to Request or Suppress Warnings
2083 @cindex options to control warnings
2084 @cindex warning messages
2085 @cindex messages, warning
2086 @cindex suppressing warnings
2088 Warnings are diagnostic messages that report constructions which
2089 are not inherently erroneous but which are risky or suggest there
2090 may have been an error.
2092 You can request many specific warnings with options beginning @samp{-W},
2093 for example @option{-Wimplicit} to request warnings on implicit
2094 declarations. Each of these specific warning options also has a
2095 negative form beginning @samp{-Wno-} to turn off warnings;
2096 for example, @option{-Wno-implicit}. This manual lists only one of the
2097 two forms, whichever is not the default.
2099 The following options control the amount and kinds of warnings produced
2100 by GCC; for further, language-specific options also refer to
2101 @ref{C++ Dialect Options} and @ref{Objective-C and Objective-C++ Dialect
2105 @cindex syntax checking
2107 @opindex fsyntax-only
2108 Check the code for syntax errors, but don't do anything beyond that.
2112 Issue all the warnings demanded by strict ISO C and ISO C++;
2113 reject all programs that use forbidden extensions, and some other
2114 programs that do not follow ISO C and ISO C++. For ISO C, follows the
2115 version of the ISO C standard specified by any @option{-std} option used.
2117 Valid ISO C and ISO C++ programs should compile properly with or without
2118 this option (though a rare few will require @option{-ansi} or a
2119 @option{-std} option specifying the required version of ISO C)@. However,
2120 without this option, certain GNU extensions and traditional C and C++
2121 features are supported as well. With this option, they are rejected.
2123 @option{-pedantic} does not cause warning messages for use of the
2124 alternate keywords whose names begin and end with @samp{__}. Pedantic
2125 warnings are also disabled in the expression that follows
2126 @code{__extension__}. However, only system header files should use
2127 these escape routes; application programs should avoid them.
2128 @xref{Alternate Keywords}.
2130 Some users try to use @option{-pedantic} to check programs for strict ISO
2131 C conformance. They soon find that it does not do quite what they want:
2132 it finds some non-ISO practices, but not all---only those for which
2133 ISO C @emph{requires} a diagnostic, and some others for which
2134 diagnostics have been added.
2136 A feature to report any failure to conform to ISO C might be useful in
2137 some instances, but would require considerable additional work and would
2138 be quite different from @option{-pedantic}. We don't have plans to
2139 support such a feature in the near future.
2141 Where the standard specified with @option{-std} represents a GNU
2142 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
2143 corresponding @dfn{base standard}, the version of ISO C on which the GNU
2144 extended dialect is based. Warnings from @option{-pedantic} are given
2145 where they are required by the base standard. (It would not make sense
2146 for such warnings to be given only for features not in the specified GNU
2147 C dialect, since by definition the GNU dialects of C include all
2148 features the compiler supports with the given option, and there would be
2149 nothing to warn about.)
2151 @item -pedantic-errors
2152 @opindex pedantic-errors
2153 Like @option{-pedantic}, except that errors are produced rather than
2158 Inhibit all warning messages.
2162 Inhibit warning messages about the use of @samp{#import}.
2164 @item -Wchar-subscripts
2165 @opindex Wchar-subscripts
2166 Warn if an array subscript has type @code{char}. This is a common cause
2167 of error, as programmers often forget that this type is signed on some
2169 This warning is enabled by @option{-Wall}.
2173 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
2174 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
2175 This warning is enabled by @option{-Wall}.
2177 @item -Wfatal-errors
2178 @opindex Wfatal-errors
2179 This option causes the compiler to abort compilation on the first error
2180 occurred rather than trying to keep going and printing further error
2185 @opindex ffreestanding
2186 @opindex fno-builtin
2187 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
2188 the arguments supplied have types appropriate to the format string
2189 specified, and that the conversions specified in the format string make
2190 sense. This includes standard functions, and others specified by format
2191 attributes (@pxref{Function Attributes}), in the @code{printf},
2192 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
2193 not in the C standard) families (or other target-specific families).
2194 Which functions are checked without format attributes having been
2195 specified depends on the standard version selected, and such checks of
2196 functions without the attribute specified are disabled by
2197 @option{-ffreestanding} or @option{-fno-builtin}.
2199 The formats are checked against the format features supported by GNU
2200 libc version 2.2. These include all ISO C90 and C99 features, as well
2201 as features from the Single Unix Specification and some BSD and GNU
2202 extensions. Other library implementations may not support all these
2203 features; GCC does not support warning about features that go beyond a
2204 particular library's limitations. However, if @option{-pedantic} is used
2205 with @option{-Wformat}, warnings will be given about format features not
2206 in the selected standard version (but not for @code{strfmon} formats,
2207 since those are not in any version of the C standard). @xref{C Dialect
2208 Options,,Options Controlling C Dialect}.
2210 Since @option{-Wformat} also checks for null format arguments for
2211 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
2213 @option{-Wformat} is included in @option{-Wall}. For more control over some
2214 aspects of format checking, the options @option{-Wformat-y2k},
2215 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
2216 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
2217 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
2220 @opindex Wformat-y2k
2221 If @option{-Wformat} is specified, also warn about @code{strftime}
2222 formats which may yield only a two-digit year.
2224 @item -Wno-format-extra-args
2225 @opindex Wno-format-extra-args
2226 If @option{-Wformat} is specified, do not warn about excess arguments to a
2227 @code{printf} or @code{scanf} format function. The C standard specifies
2228 that such arguments are ignored.
2230 Where the unused arguments lie between used arguments that are
2231 specified with @samp{$} operand number specifications, normally
2232 warnings are still given, since the implementation could not know what
2233 type to pass to @code{va_arg} to skip the unused arguments. However,
2234 in the case of @code{scanf} formats, this option will suppress the
2235 warning if the unused arguments are all pointers, since the Single
2236 Unix Specification says that such unused arguments are allowed.
2238 @item -Wno-format-zero-length
2239 @opindex Wno-format-zero-length
2240 If @option{-Wformat} is specified, do not warn about zero-length formats.
2241 The C standard specifies that zero-length formats are allowed.
2243 @item -Wformat-nonliteral
2244 @opindex Wformat-nonliteral
2245 If @option{-Wformat} is specified, also warn if the format string is not a
2246 string literal and so cannot be checked, unless the format function
2247 takes its format arguments as a @code{va_list}.
2249 @item -Wformat-security
2250 @opindex Wformat-security
2251 If @option{-Wformat} is specified, also warn about uses of format
2252 functions that represent possible security problems. At present, this
2253 warns about calls to @code{printf} and @code{scanf} functions where the
2254 format string is not a string literal and there are no format arguments,
2255 as in @code{printf (foo);}. This may be a security hole if the format
2256 string came from untrusted input and contains @samp{%n}. (This is
2257 currently a subset of what @option{-Wformat-nonliteral} warns about, but
2258 in future warnings may be added to @option{-Wformat-security} that are not
2259 included in @option{-Wformat-nonliteral}.)
2263 Enable @option{-Wformat} plus format checks not included in
2264 @option{-Wformat}. Currently equivalent to @samp{-Wformat
2265 -Wformat-nonliteral -Wformat-security -Wformat-y2k}.
2269 Warn about passing a null pointer for arguments marked as
2270 requiring a non-null value by the @code{nonnull} function attribute.
2272 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
2273 can be disabled with the @option{-Wno-nonnull} option.
2275 @item -Winit-self @r{(C, C++, Objective-C and Objective-C++ only)}
2277 Warn about uninitialized variables which are initialized with themselves.
2278 Note this option can only be used with the @option{-Wuninitialized} option,
2279 which in turn only works with @option{-O1} and above.
2281 For example, GCC will warn about @code{i} being uninitialized in the
2282 following snippet only when @option{-Winit-self} has been specified:
2293 @item -Wimplicit-int
2294 @opindex Wimplicit-int
2295 Warn when a declaration does not specify a type.
2296 This warning is enabled by @option{-Wall}.
2298 @item -Wimplicit-function-declaration
2299 @itemx -Werror-implicit-function-declaration
2300 @opindex Wimplicit-function-declaration
2301 @opindex Werror-implicit-function-declaration
2302 Give a warning (or error) whenever a function is used before being
2303 declared. The form @option{-Wno-error-implicit-function-declaration}
2305 This warning is enabled by @option{-Wall} (as a warning, not an error).
2309 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
2310 This warning is enabled by @option{-Wall}.
2314 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
2315 function with external linkage, returning int, taking either zero
2316 arguments, two, or three arguments of appropriate types.
2317 This warning is enabled by @option{-Wall}.
2319 @item -Wmissing-braces
2320 @opindex Wmissing-braces
2321 Warn if an aggregate or union initializer is not fully bracketed. In
2322 the following example, the initializer for @samp{a} is not fully
2323 bracketed, but that for @samp{b} is fully bracketed.
2326 int a[2][2] = @{ 0, 1, 2, 3 @};
2327 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
2330 This warning is enabled by @option{-Wall}.
2332 @item -Wmissing-include-dirs @r{(C, C++, Objective-C and Objective-C++ only)}
2333 @opindex Wmissing-include-dirs
2334 Warn if a user-supplied include directory does not exist.
2337 @opindex Wparentheses
2338 Warn if parentheses are omitted in certain contexts, such
2339 as when there is an assignment in a context where a truth value
2340 is expected, or when operators are nested whose precedence people
2341 often get confused about. Only the warning for an assignment used as
2342 a truth value is supported when compiling C++; the other warnings are
2343 only supported when compiling C@.
2345 Also warn if a comparison like @samp{x<=y<=z} appears; this is
2346 equivalent to @samp{(x<=y ? 1 : 0) <= z}, which is a different
2347 interpretation from that of ordinary mathematical notation.
2349 Also warn about constructions where there may be confusion to which
2350 @code{if} statement an @code{else} branch belongs. Here is an example of
2365 In C, every @code{else} branch belongs to the innermost possible @code{if}
2366 statement, which in this example is @code{if (b)}. This is often not
2367 what the programmer expected, as illustrated in the above example by
2368 indentation the programmer chose. When there is the potential for this
2369 confusion, GCC will issue a warning when this flag is specified.
2370 To eliminate the warning, add explicit braces around the innermost
2371 @code{if} statement so there is no way the @code{else} could belong to
2372 the enclosing @code{if}. The resulting code would look like this:
2388 This warning is enabled by @option{-Wall}.
2390 @item -Wsequence-point
2391 @opindex Wsequence-point
2392 Warn about code that may have undefined semantics because of violations
2393 of sequence point rules in the C standard.
2395 The C standard defines the order in which expressions in a C program are
2396 evaluated in terms of @dfn{sequence points}, which represent a partial
2397 ordering between the execution of parts of the program: those executed
2398 before the sequence point, and those executed after it. These occur
2399 after the evaluation of a full expression (one which is not part of a
2400 larger expression), after the evaluation of the first operand of a
2401 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
2402 function is called (but after the evaluation of its arguments and the
2403 expression denoting the called function), and in certain other places.
2404 Other than as expressed by the sequence point rules, the order of
2405 evaluation of subexpressions of an expression is not specified. All
2406 these rules describe only a partial order rather than a total order,
2407 since, for example, if two functions are called within one expression
2408 with no sequence point between them, the order in which the functions
2409 are called is not specified. However, the standards committee have
2410 ruled that function calls do not overlap.
2412 It is not specified when between sequence points modifications to the
2413 values of objects take effect. Programs whose behavior depends on this
2414 have undefined behavior; the C standard specifies that ``Between the
2415 previous and next sequence point an object shall have its stored value
2416 modified at most once by the evaluation of an expression. Furthermore,
2417 the prior value shall be read only to determine the value to be
2418 stored.''. If a program breaks these rules, the results on any
2419 particular implementation are entirely unpredictable.
2421 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
2422 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
2423 diagnosed by this option, and it may give an occasional false positive
2424 result, but in general it has been found fairly effective at detecting
2425 this sort of problem in programs.
2427 The present implementation of this option only works for C programs. A
2428 future implementation may also work for C++ programs.
2430 The C standard is worded confusingly, therefore there is some debate
2431 over the precise meaning of the sequence point rules in subtle cases.
2432 Links to discussions of the problem, including proposed formal
2433 definitions, may be found on the GCC readings page, at
2434 @w{@uref{http://gcc.gnu.org/readings.html}}.
2436 This warning is enabled by @option{-Wall}.
2439 @opindex Wreturn-type
2440 Warn whenever a function is defined with a return-type that defaults to
2441 @code{int}. Also warn about any @code{return} statement with no
2442 return-value in a function whose return-type is not @code{void}.
2444 For C, also warn if the return type of a function has a type qualifier
2445 such as @code{const}. Such a type qualifier has no effect, since the
2446 value returned by a function is not an lvalue. ISO C prohibits
2447 qualified @code{void} return types on function definitions, so such
2448 return types always receive a warning even without this option.
2450 For C++, a function without return type always produces a diagnostic
2451 message, even when @option{-Wno-return-type} is specified. The only
2452 exceptions are @samp{main} and functions defined in system headers.
2454 This warning is enabled by @option{-Wall}.
2458 Warn whenever a @code{switch} statement has an index of enumerated type
2459 and lacks a @code{case} for one or more of the named codes of that
2460 enumeration. (The presence of a @code{default} label prevents this
2461 warning.) @code{case} labels outside the enumeration range also
2462 provoke warnings when this option is used.
2463 This warning is enabled by @option{-Wall}.
2465 @item -Wswitch-default
2466 @opindex Wswitch-switch
2467 Warn whenever a @code{switch} statement does not have a @code{default}
2471 @opindex Wswitch-enum
2472 Warn whenever a @code{switch} statement has an index of enumerated type
2473 and lacks a @code{case} for one or more of the named codes of that
2474 enumeration. @code{case} labels outside the enumeration range also
2475 provoke warnings when this option is used.
2479 Warn if any trigraphs are encountered that might change the meaning of
2480 the program (trigraphs within comments are not warned about).
2481 This warning is enabled by @option{-Wall}.
2483 @item -Wunused-function
2484 @opindex Wunused-function
2485 Warn whenever a static function is declared but not defined or a
2486 non\-inline static function is unused.
2487 This warning is enabled by @option{-Wall}.
2489 @item -Wunused-label
2490 @opindex Wunused-label
2491 Warn whenever a label is declared but not used.
2492 This warning is enabled by @option{-Wall}.
2494 To suppress this warning use the @samp{unused} attribute
2495 (@pxref{Variable Attributes}).
2497 @item -Wunused-parameter
2498 @opindex Wunused-parameter
2499 Warn whenever a function parameter is unused aside from its declaration.
2501 To suppress this warning use the @samp{unused} attribute
2502 (@pxref{Variable Attributes}).
2504 @item -Wunused-variable
2505 @opindex Wunused-variable
2506 Warn whenever a local variable or non-constant static variable is unused
2507 aside from its declaration
2508 This warning is enabled by @option{-Wall}.
2510 To suppress this warning use the @samp{unused} attribute
2511 (@pxref{Variable Attributes}).
2513 @item -Wunused-value
2514 @opindex Wunused-value
2515 Warn whenever a statement computes a result that is explicitly not used.
2516 This warning is enabled by @option{-Wall}.
2518 To suppress this warning cast the expression to @samp{void}.
2522 All the above @option{-Wunused} options combined.
2524 In order to get a warning about an unused function parameter, you must
2525 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
2526 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
2528 @item -Wuninitialized
2529 @opindex Wuninitialized
2530 Warn if an automatic variable is used without first being initialized or
2531 if a variable may be clobbered by a @code{setjmp} call.
2533 These warnings are possible only in optimizing compilation,
2534 because they require data flow information that is computed only
2535 when optimizing. If you don't specify @option{-O}, you simply won't
2538 If you want to warn about code which uses the uninitialized value of the
2539 variable in its own initializer, use the @option{-Winit-self} option.
2541 These warnings occur only for variables that are candidates for
2542 register allocation. Therefore, they do not occur for a variable that
2543 is declared @code{volatile}, or whose address is taken, or whose size
2544 is other than 1, 2, 4 or 8 bytes. Also, they do not occur for
2545 structures, unions or arrays, even when they are in registers.
2547 Note that there may be no warning about a variable that is used only
2548 to compute a value that itself is never used, because such
2549 computations may be deleted by data flow analysis before the warnings
2552 These warnings are made optional because GCC is not smart
2553 enough to see all the reasons why the code might be correct
2554 despite appearing to have an error. Here is one example of how
2575 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
2576 always initialized, but GCC doesn't know this. Here is
2577 another common case:
2582 if (change_y) save_y = y, y = new_y;
2584 if (change_y) y = save_y;
2589 This has no bug because @code{save_y} is used only if it is set.
2591 @cindex @code{longjmp} warnings
2592 This option also warns when a non-volatile automatic variable might be
2593 changed by a call to @code{longjmp}. These warnings as well are possible
2594 only in optimizing compilation.
2596 The compiler sees only the calls to @code{setjmp}. It cannot know
2597 where @code{longjmp} will be called; in fact, a signal handler could
2598 call it at any point in the code. As a result, you may get a warning
2599 even when there is in fact no problem because @code{longjmp} cannot
2600 in fact be called at the place which would cause a problem.
2602 Some spurious warnings can be avoided if you declare all the functions
2603 you use that never return as @code{noreturn}. @xref{Function
2606 This warning is enabled by @option{-Wall}.
2608 @item -Wunknown-pragmas
2609 @opindex Wunknown-pragmas
2610 @cindex warning for unknown pragmas
2611 @cindex unknown pragmas, warning
2612 @cindex pragmas, warning of unknown
2613 Warn when a #pragma directive is encountered which is not understood by
2614 GCC@. If this command line option is used, warnings will even be issued
2615 for unknown pragmas in system header files. This is not the case if
2616 the warnings were only enabled by the @option{-Wall} command line option.
2618 @item -Wstrict-aliasing
2619 @opindex Wstrict-aliasing
2620 This option is only active when @option{-fstrict-aliasing} is active.
2621 It warns about code which might break the strict aliasing rules that the
2622 compiler is using for optimization. The warning does not catch all
2623 cases, but does attempt to catch the more common pitfalls. It is
2624 included in @option{-Wall}.
2626 @item -Wstrict-aliasing=2
2627 @opindex Wstrict-aliasing=2
2628 This option is only active when @option{-fstrict-aliasing} is active.
2629 It warns about all code which might break the strict aliasing rules that the
2630 compiler is using for optimization. This warning catches all cases, but
2631 it will also give a warning for some ambiguous cases that are safe.
2635 All of the above @samp{-W} options combined. This enables all the
2636 warnings about constructions that some users consider questionable, and
2637 that are easy to avoid (or modify to prevent the warning), even in
2638 conjunction with macros. This also enables some language-specific
2639 warnings described in @ref{C++ Dialect Options} and
2640 @ref{Objective-C and Objective-C++ Dialect Options}.
2643 The following @option{-W@dots{}} options are not implied by @option{-Wall}.
2644 Some of them warn about constructions that users generally do not
2645 consider questionable, but which occasionally you might wish to check
2646 for; others warn about constructions that are necessary or hard to avoid
2647 in some cases, and there is no simple way to modify the code to suppress
2654 (This option used to be called @option{-W}. The older name is still
2655 supported, but the newer name is more descriptive.) Print extra warning
2656 messages for these events:
2660 A function can return either with or without a value. (Falling
2661 off the end of the function body is considered returning without
2662 a value.) For example, this function would evoke such a
2676 An expression-statement or the left-hand side of a comma expression
2677 contains no side effects.
2678 To suppress the warning, cast the unused expression to void.
2679 For example, an expression such as @samp{x[i,j]} will cause a warning,
2680 but @samp{x[(void)i,j]} will not.
2683 An unsigned value is compared against zero with @samp{<} or @samp{>=}.
2686 Storage-class specifiers like @code{static} are not the first things in
2687 a declaration. According to the C Standard, this usage is obsolescent.
2690 If @option{-Wall} or @option{-Wunused} is also specified, warn about unused
2694 A comparison between signed and unsigned values could produce an
2695 incorrect result when the signed value is converted to unsigned.
2696 (But don't warn if @option{-Wno-sign-compare} is also specified.)
2699 An aggregate has an initializer which does not initialize all members.
2700 This warning can be independently controlled by
2701 @option{-Wmissing-field-initializers}.
2704 A function parameter is declared without a type specifier in K&R-style
2712 An empty body occurs in an @samp{if} or @samp{else} statement.
2715 A pointer is compared against integer zero with @samp{<}, @samp{<=},
2716 @samp{>}, or @samp{>=}.
2719 A variable might be changed by @samp{longjmp} or @samp{vfork}.
2722 Any of several floating-point events that often indicate errors, such as
2723 overflow, underflow, loss of precision, etc.
2725 @item @r{(C++ only)}
2726 An enumerator and a non-enumerator both appear in a conditional expression.
2728 @item @r{(C++ only)}
2729 A non-static reference or non-static @samp{const} member appears in a
2730 class without constructors.
2732 @item @r{(C++ only)}
2733 Ambiguous virtual bases.
2735 @item @r{(C++ only)}
2736 Subscripting an array which has been declared @samp{register}.
2738 @item @r{(C++ only)}
2739 Taking the address of a variable which has been declared @samp{register}.
2741 @item @r{(C++ only)}
2742 A base class is not initialized in a derived class' copy constructor.
2745 @item -Wno-div-by-zero
2746 @opindex Wno-div-by-zero
2747 @opindex Wdiv-by-zero
2748 Do not warn about compile-time integer division by zero. Floating point
2749 division by zero is not warned about, as it can be a legitimate way of
2750 obtaining infinities and NaNs.
2752 @item -Wsystem-headers
2753 @opindex Wsystem-headers
2754 @cindex warnings from system headers
2755 @cindex system headers, warnings from
2756 Print warning messages for constructs found in system header files.
2757 Warnings from system headers are normally suppressed, on the assumption
2758 that they usually do not indicate real problems and would only make the
2759 compiler output harder to read. Using this command line option tells
2760 GCC to emit warnings from system headers as if they occurred in user
2761 code. However, note that using @option{-Wall} in conjunction with this
2762 option will @emph{not} warn about unknown pragmas in system
2763 headers---for that, @option{-Wunknown-pragmas} must also be used.
2766 @opindex Wfloat-equal
2767 Warn if floating point values are used in equality comparisons.
2769 The idea behind this is that sometimes it is convenient (for the
2770 programmer) to consider floating-point values as approximations to
2771 infinitely precise real numbers. If you are doing this, then you need
2772 to compute (by analyzing the code, or in some other way) the maximum or
2773 likely maximum error that the computation introduces, and allow for it
2774 when performing comparisons (and when producing output, but that's a
2775 different problem). In particular, instead of testing for equality, you
2776 would check to see whether the two values have ranges that overlap; and
2777 this is done with the relational operators, so equality comparisons are
2780 @item -Wtraditional @r{(C only)}
2781 @opindex Wtraditional
2782 Warn about certain constructs that behave differently in traditional and
2783 ISO C@. Also warn about ISO C constructs that have no traditional C
2784 equivalent, and/or problematic constructs which should be avoided.
2788 Macro parameters that appear within string literals in the macro body.
2789 In traditional C macro replacement takes place within string literals,
2790 but does not in ISO C@.
2793 In traditional C, some preprocessor directives did not exist.
2794 Traditional preprocessors would only consider a line to be a directive
2795 if the @samp{#} appeared in column 1 on the line. Therefore
2796 @option{-Wtraditional} warns about directives that traditional C
2797 understands but would ignore because the @samp{#} does not appear as the
2798 first character on the line. It also suggests you hide directives like
2799 @samp{#pragma} not understood by traditional C by indenting them. Some
2800 traditional implementations would not recognize @samp{#elif}, so it
2801 suggests avoiding it altogether.
2804 A function-like macro that appears without arguments.
2807 The unary plus operator.
2810 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
2811 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
2812 constants.) Note, these suffixes appear in macros defined in the system
2813 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
2814 Use of these macros in user code might normally lead to spurious
2815 warnings, however GCC's integrated preprocessor has enough context to
2816 avoid warning in these cases.
2819 A function declared external in one block and then used after the end of
2823 A @code{switch} statement has an operand of type @code{long}.
2826 A non-@code{static} function declaration follows a @code{static} one.
2827 This construct is not accepted by some traditional C compilers.
2830 The ISO type of an integer constant has a different width or
2831 signedness from its traditional type. This warning is only issued if
2832 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
2833 typically represent bit patterns, are not warned about.
2836 Usage of ISO string concatenation is detected.
2839 Initialization of automatic aggregates.
2842 Identifier conflicts with labels. Traditional C lacks a separate
2843 namespace for labels.
2846 Initialization of unions. If the initializer is zero, the warning is
2847 omitted. This is done under the assumption that the zero initializer in
2848 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
2849 initializer warnings and relies on default initialization to zero in the
2853 Conversions by prototypes between fixed/floating point values and vice
2854 versa. The absence of these prototypes when compiling with traditional
2855 C would cause serious problems. This is a subset of the possible
2856 conversion warnings, for the full set use @option{-Wconversion}.
2859 Use of ISO C style function definitions. This warning intentionally is
2860 @emph{not} issued for prototype declarations or variadic functions
2861 because these ISO C features will appear in your code when using
2862 libiberty's traditional C compatibility macros, @code{PARAMS} and
2863 @code{VPARAMS}. This warning is also bypassed for nested functions
2864 because that feature is already a GCC extension and thus not relevant to
2865 traditional C compatibility.
2868 @item -Wdeclaration-after-statement @r{(C only)}
2869 @opindex Wdeclaration-after-statement
2870 Warn when a declaration is found after a statement in a block. This
2871 construct, known from C++, was introduced with ISO C99 and is by default
2872 allowed in GCC@. It is not supported by ISO C90 and was not supported by
2873 GCC versions before GCC 3.0. @xref{Mixed Declarations}.
2877 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
2879 @item -Wno-endif-labels
2880 @opindex Wno-endif-labels
2881 @opindex Wendif-labels
2882 Do not warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
2886 Warn whenever a local variable shadows another local variable, parameter or
2887 global variable or whenever a built-in function is shadowed.
2889 @item -Wlarger-than-@var{len}
2890 @opindex Wlarger-than
2891 Warn whenever an object of larger than @var{len} bytes is defined.
2893 @item -Wpointer-arith
2894 @opindex Wpointer-arith
2895 Warn about anything that depends on the ``size of'' a function type or
2896 of @code{void}. GNU C assigns these types a size of 1, for
2897 convenience in calculations with @code{void *} pointers and pointers
2900 @item -Wbad-function-cast @r{(C only)}
2901 @opindex Wbad-function-cast
2902 Warn whenever a function call is cast to a non-matching type.
2903 For example, warn if @code{int malloc()} is cast to @code{anything *}.
2907 Warn whenever a pointer is cast so as to remove a type qualifier from
2908 the target type. For example, warn if a @code{const char *} is cast
2909 to an ordinary @code{char *}.
2912 @opindex Wcast-align
2913 Warn whenever a pointer is cast such that the required alignment of the
2914 target is increased. For example, warn if a @code{char *} is cast to
2915 an @code{int *} on machines where integers can only be accessed at
2916 two- or four-byte boundaries.
2918 @item -Wwrite-strings
2919 @opindex Wwrite-strings
2920 When compiling C, give string constants the type @code{const
2921 char[@var{length}]} so that
2922 copying the address of one into a non-@code{const} @code{char *}
2923 pointer will get a warning; when compiling C++, warn about the
2924 deprecated conversion from string constants to @code{char *}.
2925 These warnings will help you find at
2926 compile time code that can try to write into a string constant, but
2927 only if you have been very careful about using @code{const} in
2928 declarations and prototypes. Otherwise, it will just be a nuisance;
2929 this is why we did not make @option{-Wall} request these warnings.
2932 @opindex Wconversion
2933 Warn if a prototype causes a type conversion that is different from what
2934 would happen to the same argument in the absence of a prototype. This
2935 includes conversions of fixed point to floating and vice versa, and
2936 conversions changing the width or signedness of a fixed point argument
2937 except when the same as the default promotion.
2939 Also, warn if a negative integer constant expression is implicitly
2940 converted to an unsigned type. For example, warn about the assignment
2941 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
2942 casts like @code{(unsigned) -1}.
2944 @item -Wsign-compare
2945 @opindex Wsign-compare
2946 @cindex warning for comparison of signed and unsigned values
2947 @cindex comparison of signed and unsigned values, warning
2948 @cindex signed and unsigned values, comparison warning
2949 Warn when a comparison between signed and unsigned values could produce
2950 an incorrect result when the signed value is converted to unsigned.
2951 This warning is also enabled by @option{-Wextra}; to get the other warnings
2952 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
2954 @item -Waggregate-return
2955 @opindex Waggregate-return
2956 Warn if any functions that return structures or unions are defined or
2957 called. (In languages where you can return an array, this also elicits
2960 @item -Wstrict-prototypes @r{(C only)}
2961 @opindex Wstrict-prototypes
2962 Warn if a function is declared or defined without specifying the
2963 argument types. (An old-style function definition is permitted without
2964 a warning if preceded by a declaration which specifies the argument
2967 @item -Wold-style-definition @r{(C only)}
2968 @opindex Wold-style-definition
2969 Warn if an old-style function definition is used. A warning is given
2970 even if there is a previous prototype.
2972 @item -Wmissing-prototypes @r{(C only)}
2973 @opindex Wmissing-prototypes
2974 Warn if a global function is defined without a previous prototype
2975 declaration. This warning is issued even if the definition itself
2976 provides a prototype. The aim is to detect global functions that fail
2977 to be declared in header files.
2979 @item -Wmissing-declarations @r{(C only)}
2980 @opindex Wmissing-declarations
2981 Warn if a global function is defined without a previous declaration.
2982 Do so even if the definition itself provides a prototype.
2983 Use this option to detect global functions that are not declared in
2986 @item -Wmissing-field-initializers
2987 @opindex Wmissing-field-initializers
2990 Warn if a structure's initializer has some fields missing. For
2991 example, the following code would cause such a warning, because
2992 @code{x.h} is implicitly zero:
2995 struct s @{ int f, g, h; @};
2996 struct s x = @{ 3, 4 @};
2999 This option does not warn about designated initializers, so the following
3000 modification would not trigger a warning:
3003 struct s @{ int f, g, h; @};
3004 struct s x = @{ .f = 3, .g = 4 @};
3007 This warning is included in @option{-Wextra}. To get other @option{-Wextra}
3008 warnings without this one, use @samp{-Wextra -Wno-missing-field-initializers}.
3010 @item -Wmissing-noreturn
3011 @opindex Wmissing-noreturn
3012 Warn about functions which might be candidates for attribute @code{noreturn}.
3013 Note these are only possible candidates, not absolute ones. Care should
3014 be taken to manually verify functions actually do not ever return before
3015 adding the @code{noreturn} attribute, otherwise subtle code generation
3016 bugs could be introduced. You will not get a warning for @code{main} in
3017 hosted C environments.
3019 @item -Wmissing-format-attribute
3020 @opindex Wmissing-format-attribute
3022 If @option{-Wformat} is enabled, also warn about functions which might be
3023 candidates for @code{format} attributes. Note these are only possible
3024 candidates, not absolute ones. GCC will guess that @code{format}
3025 attributes might be appropriate for any function that calls a function
3026 like @code{vprintf} or @code{vscanf}, but this might not always be the
3027 case, and some functions for which @code{format} attributes are
3028 appropriate may not be detected. This option has no effect unless
3029 @option{-Wformat} is enabled (possibly by @option{-Wall}).
3031 @item -Wno-multichar
3032 @opindex Wno-multichar
3034 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
3035 Usually they indicate a typo in the user's code, as they have
3036 implementation-defined values, and should not be used in portable code.
3038 @item -Wno-deprecated-declarations
3039 @opindex Wno-deprecated-declarations
3040 Do not warn about uses of functions, variables, and types marked as
3041 deprecated by using the @code{deprecated} attribute.
3042 (@pxref{Function Attributes}, @pxref{Variable Attributes},
3043 @pxref{Type Attributes}.)
3047 Warn if a structure is given the packed attribute, but the packed
3048 attribute has no effect on the layout or size of the structure.
3049 Such structures may be mis-aligned for little benefit. For
3050 instance, in this code, the variable @code{f.x} in @code{struct bar}
3051 will be misaligned even though @code{struct bar} does not itself
3052 have the packed attribute:
3059 @} __attribute__((packed));
3069 Warn if padding is included in a structure, either to align an element
3070 of the structure or to align the whole structure. Sometimes when this
3071 happens it is possible to rearrange the fields of the structure to
3072 reduce the padding and so make the structure smaller.
3074 @item -Wredundant-decls
3075 @opindex Wredundant-decls
3076 Warn if anything is declared more than once in the same scope, even in
3077 cases where multiple declaration is valid and changes nothing.
3079 @item -Wnested-externs @r{(C only)}
3080 @opindex Wnested-externs
3081 Warn if an @code{extern} declaration is encountered within a function.
3083 @item -Wunreachable-code
3084 @opindex Wunreachable-code
3085 Warn if the compiler detects that code will never be executed.
3087 This option is intended to warn when the compiler detects that at
3088 least a whole line of source code will never be executed, because
3089 some condition is never satisfied or because it is after a
3090 procedure that never returns.
3092 It is possible for this option to produce a warning even though there
3093 are circumstances under which part of the affected line can be executed,
3094 so care should be taken when removing apparently-unreachable code.
3096 For instance, when a function is inlined, a warning may mean that the
3097 line is unreachable in only one inlined copy of the function.
3099 This option is not made part of @option{-Wall} because in a debugging
3100 version of a program there is often substantial code which checks
3101 correct functioning of the program and is, hopefully, unreachable
3102 because the program does work. Another common use of unreachable
3103 code is to provide behavior which is selectable at compile-time.
3107 Warn if a function can not be inlined and it was declared as inline.
3108 Even with this option, the compiler will not warn about failures to
3109 inline functions declared in system headers.
3111 The compiler uses a variety of heuristics to determine whether or not
3112 to inline a function. For example, the compiler takes into account
3113 the size of the function being inlined and the the amount of inlining
3114 that has already been done in the current function. Therefore,
3115 seemingly insignificant changes in the source program can cause the
3116 warnings produced by @option{-Winline} to appear or disappear.
3118 @item -Wno-invalid-offsetof @r{(C++ only)}
3119 @opindex Wno-invalid-offsetof
3120 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
3121 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
3122 to a non-POD type is undefined. In existing C++ implementations,
3123 however, @samp{offsetof} typically gives meaningful results even when
3124 applied to certain kinds of non-POD types. (Such as a simple
3125 @samp{struct} that fails to be a POD type only by virtue of having a
3126 constructor.) This flag is for users who are aware that they are
3127 writing nonportable code and who have deliberately chosen to ignore the
3130 The restrictions on @samp{offsetof} may be relaxed in a future version
3131 of the C++ standard.
3134 @opindex Winvalid-pch
3135 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
3136 the search path but can't be used.
3140 @opindex Wno-long-long
3141 Warn if @samp{long long} type is used. This is default. To inhibit
3142 the warning messages, use @option{-Wno-long-long}. Flags
3143 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
3144 only when @option{-pedantic} flag is used.
3146 @item -Wvariadic-macros
3147 @opindex Wvariadic-macros
3148 @opindex Wno-variadic-macros
3149 Warn if variadic macros are used in pedantic ISO C90 mode, or the GNU
3150 alternate syntax when in pedantic ISO C99 mode. This is default.
3151 To inhibit the warning messages, use @option{-Wno-variadic-macros}.
3153 @item -Wdisabled-optimization
3154 @opindex Wdisabled-optimization
3155 Warn if a requested optimization pass is disabled. This warning does
3156 not generally indicate that there is anything wrong with your code; it
3157 merely indicates that GCC's optimizers were unable to handle the code
3158 effectively. Often, the problem is that your code is too big or too
3159 complex; GCC will refuse to optimize programs when the optimization
3160 itself is likely to take inordinate amounts of time.
3162 @item -Wno-pointer-sign
3163 @opindex Wno-pointer-sign
3164 Don't warn for pointer argument passing or assignment with different signedness.
3165 Only useful in the negative form since this warning is enabled by default.
3166 This option is only supported for C and Objective-C@.
3170 Make all warnings into errors.
3173 @node Debugging Options
3174 @section Options for Debugging Your Program or GCC
3175 @cindex options, debugging
3176 @cindex debugging information options
3178 GCC has various special options that are used for debugging
3179 either your program or GCC:
3184 Produce debugging information in the operating system's native format
3185 (stabs, COFF, XCOFF, or DWARF 2)@. GDB can work with this debugging
3188 On most systems that use stabs format, @option{-g} enables use of extra
3189 debugging information that only GDB can use; this extra information
3190 makes debugging work better in GDB but will probably make other debuggers
3192 refuse to read the program. If you want to control for certain whether
3193 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
3194 @option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
3196 GCC allows you to use @option{-g} with
3197 @option{-O}. The shortcuts taken by optimized code may occasionally
3198 produce surprising results: some variables you declared may not exist
3199 at all; flow of control may briefly move where you did not expect it;
3200 some statements may not be executed because they compute constant
3201 results or their values were already at hand; some statements may
3202 execute in different places because they were moved out of loops.
3204 Nevertheless it proves possible to debug optimized output. This makes
3205 it reasonable to use the optimizer for programs that might have bugs.
3207 The following options are useful when GCC is generated with the
3208 capability for more than one debugging format.
3212 Produce debugging information for use by GDB@. This means to use the
3213 most expressive format available (DWARF 2, stabs, or the native format
3214 if neither of those are supported), including GDB extensions if at all
3219 Produce debugging information in stabs format (if that is supported),
3220 without GDB extensions. This is the format used by DBX on most BSD
3221 systems. On MIPS, Alpha and System V Release 4 systems this option
3222 produces stabs debugging output which is not understood by DBX or SDB@.
3223 On System V Release 4 systems this option requires the GNU assembler.
3225 @item -feliminate-unused-debug-symbols
3226 @opindex feliminate-unused-debug-symbols
3227 Produce debugging information in stabs format (if that is supported),
3228 for only symbols that are actually used.
3232 Produce debugging information in stabs format (if that is supported),
3233 using GNU extensions understood only by the GNU debugger (GDB)@. The
3234 use of these extensions is likely to make other debuggers crash or
3235 refuse to read the program.
3239 Produce debugging information in COFF format (if that is supported).
3240 This is the format used by SDB on most System V systems prior to
3245 Produce debugging information in XCOFF format (if that is supported).
3246 This is the format used by the DBX debugger on IBM RS/6000 systems.
3250 Produce debugging information in XCOFF format (if that is supported),
3251 using GNU extensions understood only by the GNU debugger (GDB)@. The
3252 use of these extensions is likely to make other debuggers crash or
3253 refuse to read the program, and may cause assemblers other than the GNU
3254 assembler (GAS) to fail with an error.
3258 Produce debugging information in DWARF version 2 format (if that is
3259 supported). This is the format used by DBX on IRIX 6. With this
3260 option, GCC uses features of DWARF version 3 when they are useful;
3261 version 3 is upward compatible with version 2, but may still cause
3262 problems for older debuggers.
3266 Produce debugging information in VMS debug format (if that is
3267 supported). This is the format used by DEBUG on VMS systems.
3270 @itemx -ggdb@var{level}
3271 @itemx -gstabs@var{level}
3272 @itemx -gcoff@var{level}
3273 @itemx -gxcoff@var{level}
3274 @itemx -gvms@var{level}
3275 Request debugging information and also use @var{level} to specify how
3276 much information. The default level is 2.
3278 Level 1 produces minimal information, enough for making backtraces in
3279 parts of the program that you don't plan to debug. This includes
3280 descriptions of functions and external variables, but no information
3281 about local variables and no line numbers.
3283 Level 3 includes extra information, such as all the macro definitions
3284 present in the program. Some debuggers support macro expansion when
3285 you use @option{-g3}.
3287 @option{-gdwarf-2} does not accept a concatenated debug level, because
3288 GCC used to support an option @option{-gdwarf} that meant to generate
3289 debug information in version 1 of the DWARF format (which is very
3290 different from version 2), and it would have been too confusing. That
3291 debug format is long obsolete, but the option cannot be changed now.
3292 Instead use an additional @option{-g@var{level}} option to change the
3293 debug level for DWARF2.
3295 @item -feliminate-dwarf2-dups
3296 @opindex feliminate-dwarf2-dups
3297 Compress DWARF2 debugging information by eliminating duplicated
3298 information about each symbol. This option only makes sense when
3299 generating DWARF2 debugging information with @option{-gdwarf-2}.
3301 @cindex @command{prof}
3304 Generate extra code to write profile information suitable for the
3305 analysis program @command{prof}. You must use this option when compiling
3306 the source files you want data about, and you must also use it when
3309 @cindex @command{gprof}
3312 Generate extra code to write profile information suitable for the
3313 analysis program @command{gprof}. You must use this option when compiling
3314 the source files you want data about, and you must also use it when
3319 Makes the compiler print out each function name as it is compiled, and
3320 print some statistics about each pass when it finishes.
3323 @opindex ftime-report
3324 Makes the compiler print some statistics about the time consumed by each
3325 pass when it finishes.
3328 @opindex fmem-report
3329 Makes the compiler print some statistics about permanent memory
3330 allocation when it finishes.
3332 @item -fprofile-arcs
3333 @opindex fprofile-arcs
3334 Add code so that program flow @dfn{arcs} are instrumented. During
3335 execution the program records how many times each branch and call is
3336 executed and how many times it is taken or returns. When the compiled
3337 program exits it saves this data to a file called
3338 @file{@var{auxname}.gcda} for each source file. The data may be used for
3339 profile-directed optimizations (@option{-fbranch-probabilities}), or for
3340 test coverage analysis (@option{-ftest-coverage}). Each object file's
3341 @var{auxname} is generated from the name of the output file, if
3342 explicitly specified and it is not the final executable, otherwise it is
3343 the basename of the source file. In both cases any suffix is removed
3344 (e.g.@: @file{foo.gcda} for input file @file{dir/foo.c}, or
3345 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
3350 Compile the source files with @option{-fprofile-arcs} plus optimization
3351 and code generation options. For test coverage analysis, use the
3352 additional @option{-ftest-coverage} option. You do not need to profile
3353 every source file in a program.
3356 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
3357 (the latter implies the former).
3360 Run the program on a representative workload to generate the arc profile
3361 information. This may be repeated any number of times. You can run
3362 concurrent instances of your program, and provided that the file system
3363 supports locking, the data files will be correctly updated. Also
3364 @code{fork} calls are detected and correctly handled (double counting
3368 For profile-directed optimizations, compile the source files again with
3369 the same optimization and code generation options plus
3370 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
3371 Control Optimization}).
3374 For test coverage analysis, use @command{gcov} to produce human readable
3375 information from the @file{.gcno} and @file{.gcda} files. Refer to the
3376 @command{gcov} documentation for further information.
3380 With @option{-fprofile-arcs}, for each function of your program GCC
3381 creates a program flow graph, then finds a spanning tree for the graph.
3382 Only arcs that are not on the spanning tree have to be instrumented: the
3383 compiler adds code to count the number of times that these arcs are
3384 executed. When an arc is the only exit or only entrance to a block, the
3385 instrumentation code can be added to the block; otherwise, a new basic
3386 block must be created to hold the instrumentation code.
3388 @item -ftree-based-profiling
3389 @opindex ftree-based-profiling
3390 This option is used in addition to @option{-fprofile-arcs} or
3391 @option{-fbranch-probabilities} to control whether those optimizations
3392 are performed on a tree-based or rtl-based internal representation.
3393 If you use this option when compiling with @option{-fprofile-arcs},
3394 you must also use it when compiling later with @option{-fbranch-probabilities}.
3395 Currently the tree-based optimization is in an early stage of
3396 development, and this option is recommended only for those people
3397 working on improving it.
3400 @item -ftest-coverage
3401 @opindex ftest-coverage
3402 Produce a notes file that the @command{gcov} code-coverage utility
3403 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
3404 show program coverage. Each source file's note file is called
3405 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
3406 above for a description of @var{auxname} and instructions on how to
3407 generate test coverage data. Coverage data will match the source files
3408 more closely, if you do not optimize.
3410 @item -d@var{letters}
3411 @item -fdump-rtl-@var{pass}
3413 Says to make debugging dumps during compilation at times specified by
3414 @var{letters}. This is used for debugging the RTL-based passes of the
3415 compiler. The file names for most of the dumps are made by appending a
3416 pass number and a word to the @var{dumpname}. @var{dumpname} is generated
3417 from the name of the output file, if explicitly specified and it is not
3418 an executable, otherwise it is the basename of the source file.
3420 Most debug dumps can be enabled either passing a letter to the @option{-d}
3421 option, or with a long @option{-fdump-rtl} switch; here are the possible
3422 letters for use in @var{letters} and @var{pass}, and their meanings:
3427 Annotate the assembler output with miscellaneous debugging information.
3430 @itemx -fdump-rtl-bp
3432 @opindex fdump-rtl-bp
3433 Dump after computing branch probabilities, to @file{@var{file}.09.bp}.
3436 @itemx -fdump-rtl-bbro
3438 @opindex fdump-rtl-bbro
3439 Dump after block reordering, to @file{@var{file}.30.bbro}.
3442 @itemx -fdump-rtl-combine
3444 @opindex fdump-rtl-combine
3445 Dump after instruction combination, to the file @file{@var{file}.17.combine}.
3448 @itemx -fdump-rtl-ce1
3449 @itemx -fdump-rtl-ce2
3451 @opindex fdump-rtl-ce1
3452 @opindex fdump-rtl-ce2
3453 @option{-dC} and @option{-fdump-rtl-ce1} enable dumping after the
3454 first if conversion, to the file @file{@var{file}.11.ce1}. @option{-dC}
3455 and @option{-fdump-rtl-ce2} enable dumping after the second if
3456 conversion, to the file @file{@var{file}.18.ce2}.
3459 @itemx -fdump-rtl-btl
3460 @itemx -fdump-rtl-dbr
3462 @opindex fdump-rtl-btl
3463 @opindex fdump-rtl-dbr
3464 @option{-dd} and @option{-fdump-rtl-btl} enable dumping after branch
3465 target load optimization, to to @file{@var{file}.31.btl}. @option{-dd}
3466 and @option{-fdump-rtl-dbr} enable dumping after delayed branch
3467 scheduling, to @file{@var{file}.36.dbr}.
3471 Dump all macro definitions, at the end of preprocessing, in addition to
3475 @itemx -fdump-rtl-ce3
3477 @opindex fdump-rtl-ce3
3478 Dump after the third if conversion, to @file{@var{file}.28.ce3}.
3481 @itemx -fdump-rtl-cfg
3482 @itemx -fdump-rtl-life
3484 @opindex fdump-rtl-cfg
3485 @opindex fdump-rtl-life
3486 @option{-df} and @option{-fdump-rtl-cfg} enable dumping after control
3487 and data flow analysis, to @file{@var{file}.08.cfg}. @option{-df}
3488 and @option{-fdump-rtl-cfg} enable dumping dump after life analysis,
3489 to @file{@var{file}.16.life}.
3492 @itemx -fdump-rtl-greg
3494 @opindex fdump-rtl-greg
3495 Dump after global register allocation, to @file{@var{file}.23.greg}.
3498 @itemx -fdump-rtl-gcse
3499 @itemx -fdump-rtl-bypass
3501 @opindex fdump-rtl-gcse
3502 @opindex fdump-rtl-bypass
3503 @option{-dG} and @option{-fdump-rtl-gcse} enable dumping after GCSE, to
3504 @file{@var{file}.05.gcse}. @option{-dG} and @option{-fdump-rtl-bypass}
3505 enable dumping after jump bypassing and control flow optimizations, to
3506 @file{@var{file}.07.bypass}.
3509 @itemx -fdump-rtl-eh
3511 @opindex fdump-rtl-eh
3512 Dump after finalization of EH handling code, to @file{@var{file}.02.eh}.
3515 @itemx -fdump-rtl-sibling
3517 @opindex fdump-rtl-sibling
3518 Dump after sibling call optimizations, to @file{@var{file}.01.sibling}.
3521 @itemx -fdump-rtl-jump
3523 @opindex fdump-rtl-jump
3524 Dump after the first jump optimization, to @file{@var{file}.03.jump}.
3527 @itemx -fdump-rtl-stack
3529 @opindex fdump-rtl-stack
3530 Dump after conversion from registers to stack, to @file{@var{file}.33.stack}.
3533 @itemx -fdump-rtl-lreg
3535 @opindex fdump-rtl-lreg
3536 Dump after local register allocation, to @file{@var{file}.22.lreg}.
3539 @itemx -fdump-rtl-loop
3540 @itemx -fdump-rtl-loop2
3542 @opindex fdump-rtl-loop
3543 @opindex fdump-rtl-loop2
3544 @option{-dL} and @option{-fdump-rtl-loop} enable dumping after the first
3545 loop optimization pass, to @file{@var{file}.06.loop}. @option{-dL} and
3546 @option{-fdump-rtl-loop2} enable dumping after the second pass, to
3547 @file{@var{file}.13.loop2}.
3550 @itemx -fdump-rtl-sms
3552 @opindex fdump-rtl-sms
3553 Dump after modulo scheduling, to @file{@var{file}.20.sms}.
3556 @itemx -fdump-rtl-mach
3558 @opindex fdump-rtl-mach
3559 Dump after performing the machine dependent reorganization pass, to
3560 @file{@var{file}.35.mach}.
3563 @itemx -fdump-rtl-rnreg
3565 @opindex fdump-rtl-rnreg
3566 Dump after register renumbering, to @file{@var{file}.29.rnreg}.
3569 @itemx -fdump-rtl-regmove
3571 @opindex fdump-rtl-regmove
3572 Dump after the register move pass, to @file{@var{file}.19.regmove}.
3575 @itemx -fdump-rtl-postreload
3577 @opindex fdump-rtl-postreload
3578 Dump after post-reload optimizations, to @file{@var{file}.24.postreload}.
3581 @itemx -fdump-rtl-expand
3583 @opindex fdump-rtl-expand
3584 Dump after RTL generation, to @file{@var{file}.00.expand}.
3587 @itemx -fdump-rtl-sched2
3589 @opindex fdump-rtl-sched2
3590 Dump after the second scheduling pass, to @file{@var{file}.32.sched2}.
3593 @itemx -fdump-rtl-cse
3595 @opindex fdump-rtl-cse
3596 Dump after CSE (including the jump optimization that sometimes follows
3597 CSE), to @file{@var{file}.04.cse}.
3600 @itemx -fdump-rtl-sched
3602 @opindex fdump-rtl-sched
3603 Dump after the first scheduling pass, to @file{@var{file}.21.sched}.
3606 @itemx -fdump-rtl-cse2
3608 @opindex fdump-rtl-cse2
3609 Dump after the second CSE pass (including the jump optimization that
3610 sometimes follows CSE), to @file{@var{file}.15.cse2}.
3613 @itemx -fdump-rtl-tracer
3615 @opindex fdump-rtl-tracer
3616 Dump after running tracer, to @file{@var{file}.12.tracer}.
3619 @itemx -fdump-rtl-vpt
3620 @itemx -fdump-rtl-vartrack
3622 @opindex fdump-rtl-vpt
3623 @opindex fdump-rtl-vartrack
3624 @option{-dV} and @option{-fdump-rtl-vpt} enable dumping after the value
3625 profile transformations, to @file{@var{file}.10.vpt}. @option{-dV}
3626 and @option{-fdump-rtl-vartrack} enable dumping after variable tracking,
3627 to @file{@var{file}.34.vartrack}.
3630 @itemx -fdump-rtl-flow2
3632 @opindex fdump-rtl-flow2
3633 Dump after the second flow pass, to @file{@var{file}.26.flow2}.
3636 @itemx -fdump-rtl-peephole2
3638 @opindex fdump-rtl-peephole2
3639 Dump after the peephole pass, to @file{@var{file}.27.peephole2}.
3642 @itemx -fdump-rtl-web
3644 @opindex fdump-rtl-web
3645 Dump after live range splitting, to @file{@var{file}.14.web}.
3648 @itemx -fdump-rtl-all
3650 @opindex fdump-rtl-all
3651 Produce all the dumps listed above.
3655 Produce a core dump whenever an error occurs.
3659 Print statistics on memory usage, at the end of the run, to
3664 Annotate the assembler output with a comment indicating which
3665 pattern and alternative was used. The length of each instruction is
3670 Dump the RTL in the assembler output as a comment before each instruction.
3671 Also turns on @option{-dp} annotation.
3675 For each of the other indicated dump files (either with @option{-d} or
3676 @option{-fdump-rtl-@var{pass}}), dump a representation of the control flow
3677 graph suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
3681 Just generate RTL for a function instead of compiling it. Usually used
3682 with @samp{r} (@option{-fdump-rtl-expand}).
3686 Dump debugging information during parsing, to standard error.
3689 @item -fdump-unnumbered
3690 @opindex fdump-unnumbered
3691 When doing debugging dumps (see @option{-d} option above), suppress instruction
3692 numbers and line number note output. This makes it more feasible to
3693 use diff on debugging dumps for compiler invocations with different
3694 options, in particular with and without @option{-g}.
3696 @item -fdump-translation-unit @r{(C and C++ only)}
3697 @itemx -fdump-translation-unit-@var{options} @r{(C and C++ only)}
3698 @opindex fdump-translation-unit
3699 Dump a representation of the tree structure for the entire translation
3700 unit to a file. The file name is made by appending @file{.tu} to the
3701 source file name. If the @samp{-@var{options}} form is used, @var{options}
3702 controls the details of the dump as described for the
3703 @option{-fdump-tree} options.
3705 @item -fdump-class-hierarchy @r{(C++ only)}
3706 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
3707 @opindex fdump-class-hierarchy
3708 Dump a representation of each class's hierarchy and virtual function
3709 table layout to a file. The file name is made by appending @file{.class}
3710 to the source file name. If the @samp{-@var{options}} form is used,
3711 @var{options} controls the details of the dump as described for the
3712 @option{-fdump-tree} options.
3714 @item -fdump-ipa-@var{switch}
3716 Control the dumping at various stages of inter-procedural analysis
3717 language tree to a file. The file name is generated by appending a switch
3718 specific suffix to the source file name. The following dumps are possible:
3722 Enables all inter-procedural analysis dumps; currently the only produced
3723 dump is the @samp{cgraph} dump.
3726 Dumps information about call-graph optimization, unused function removal,
3727 and inlining decisions.
3730 @item -fdump-tree-@var{switch} @r{(C and C++ only)}
3731 @itemx -fdump-tree-@var{switch}-@var{options} @r{(C and C++ only)}
3733 Control the dumping at various stages of processing the intermediate
3734 language tree to a file. The file name is generated by appending a switch
3735 specific suffix to the source file name. If the @samp{-@var{options}}
3736 form is used, @var{options} is a list of @samp{-} separated options that
3737 control the details of the dump. Not all options are applicable to all
3738 dumps, those which are not meaningful will be ignored. The following
3739 options are available
3743 Print the address of each node. Usually this is not meaningful as it
3744 changes according to the environment and source file. Its primary use
3745 is for tying up a dump file with a debug environment.
3747 Inhibit dumping of members of a scope or body of a function merely
3748 because that scope has been reached. Only dump such items when they
3749 are directly reachable by some other path. When dumping pretty-printed
3750 trees, this option inhibits dumping the bodies of control structures.
3752 Print a raw representation of the tree. By default, trees are
3753 pretty-printed into a C-like representation.
3755 Enable more detailed dumps (not honored by every dump option).
3757 Enable dumping various statistics about the pass (not honored by every dump
3760 Enable showing basic block boundaries (disabled in raw dumps).
3762 Enable showing virtual operands for every statement.
3764 Enable showing line numbers for statements.
3766 Enable showing the unique ID (@code{DECL_UID}) for each variable.
3768 Turn on all options, except @option{raw}, @option{slim} and @option{lineno}.
3771 The following tree dumps are possible:
3775 Dump before any tree based optimization, to @file{@var{file}.original}.
3778 Dump after all tree based optimization, to @file{@var{file}.optimized}.
3781 Dump after function inlining, to @file{@var{file}.inlined}.
3784 @opindex fdump-tree-gimple
3785 Dump each function before and after the gimplification pass to a file. The
3786 file name is made by appending @file{.gimple} to the source file name.
3789 @opindex fdump-tree-cfg
3790 Dump the control flow graph of each function to a file. The file name is
3791 made by appending @file{.cfg} to the source file name.
3794 @opindex fdump-tree-vcg
3795 Dump the control flow graph of each function to a file in VCG format. The
3796 file name is made by appending @file{.vcg} to the source file name. Note
3797 that if the file contains more than one function, the generated file cannot
3798 be used directly by VCG@. You will need to cut and paste each function's
3799 graph into its own separate file first.
3802 @opindex fdump-tree-ch
3803 Dump each function after copying loop headers. The file name is made by
3804 appending @file{.ch} to the source file name.
3807 @opindex fdump-tree-ssa
3808 Dump SSA related information to a file. The file name is made by appending
3809 @file{.ssa} to the source file name.
3812 @opindex fdump-tree-alias
3813 Dump aliasing information for each function. The file name is made by
3814 appending @file{.alias} to the source file name.
3817 @opindex fdump-tree-ccp
3818 Dump each function after CCP@. The file name is made by appending
3819 @file{.ccp} to the source file name.
3822 @opindex fdump-tree-pre
3823 Dump trees after partial redundancy elimination. The file name is made
3824 by appending @file{.pre} to the source file name.
3827 @opindex fdump-tree-fre
3828 Dump trees after full redundancy elimination. The file name is made
3829 by appending @file{.fre} to the source file name.
3832 @opindex fdump-tree-dce
3833 Dump each function after dead code elimination. The file name is made by
3834 appending @file{.dce} to the source file name.
3837 @opindex fdump-tree-mudflap
3838 Dump each function after adding mudflap instrumentation. The file name is
3839 made by appending @file{.mudflap} to the source file name.
3842 @opindex fdump-tree-sra
3843 Dump each function after performing scalar replacement of aggregates. The
3844 file name is made by appending @file{.sra} to the source file name.
3847 @opindex fdump-tree-dom
3848 Dump each function after applying dominator tree optimizations. The file
3849 name is made by appending @file{.dom} to the source file name.
3852 @opindex fdump-tree-dse
3853 Dump each function after applying dead store elimination. The file
3854 name is made by appending @file{.dse} to the source file name.
3857 @opindex fdump-tree-phiopt
3858 Dump each function after optimizing PHI nodes into straightline code. The file
3859 name is made by appending @file{.phiopt} to the source file name.
3862 @opindex fdump-tree-forwprop
3863 Dump each function after forward propagating single use variables. The file
3864 name is made by appending @file{.forwprop} to the source file name.
3867 @opindex fdump-tree-copyrename
3868 Dump each function after applying the copy rename optimization. The file
3869 name is made by appending @file{.copyrename} to the source file name.
3872 @opindex fdump-tree-nrv
3873 Dump each function after applying the named return value optimization on
3874 generic trees. The file name is made by appending @file{.nrv} to the source
3878 @opindex fdump-tree-vect
3879 Dump each function after applying vectorization of loops. The file name is
3880 made by appending @file{.vect} to the source file name.
3883 @opindex fdump-tree-all
3884 Enable all the available tree dumps with the flags provided in this option.
3887 @item -ftree-vectorizer-verbose=@var{n}
3888 @opindex ftree-vectorizer-verbose
3889 This option controls the amount of debugging output the vectorizer prints.
3890 This information is written to standard error, unless @option{-fdump-tree-all}
3891 or @option{-fdump-tree-vect} is specified, in which case it is output to the
3892 usual dump listing file, @file{.vect}.
3894 @item -frandom-seed=@var{string}
3895 @opindex frandom-string
3896 This option provides a seed that GCC uses when it would otherwise use
3897 random numbers. It is used to generate certain symbol names
3898 that have to be different in every compiled file. It is also used to
3899 place unique stamps in coverage data files and the object files that
3900 produce them. You can use the @option{-frandom-seed} option to produce
3901 reproducibly identical object files.
3903 The @var{string} should be different for every file you compile.
3905 @item -fsched-verbose=@var{n}
3906 @opindex fsched-verbose
3907 On targets that use instruction scheduling, this option controls the
3908 amount of debugging output the scheduler prints. This information is
3909 written to standard error, unless @option{-dS} or @option{-dR} is
3910 specified, in which case it is output to the usual dump
3911 listing file, @file{.sched} or @file{.sched2} respectively. However
3912 for @var{n} greater than nine, the output is always printed to standard
3915 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
3916 same information as @option{-dRS}. For @var{n} greater than one, it
3917 also output basic block probabilities, detailed ready list information
3918 and unit/insn info. For @var{n} greater than two, it includes RTL
3919 at abort point, control-flow and regions info. And for @var{n} over
3920 four, @option{-fsched-verbose} also includes dependence info.
3924 Store the usual ``temporary'' intermediate files permanently; place them
3925 in the current directory and name them based on the source file. Thus,
3926 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
3927 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
3928 preprocessed @file{foo.i} output file even though the compiler now
3929 normally uses an integrated preprocessor.
3931 When used in combination with the @option{-x} command line option,
3932 @option{-save-temps} is sensible enough to avoid over writing an
3933 input source file with the same extension as an intermediate file.
3934 The corresponding intermediate file may be obtained by renaming the
3935 source file before using @option{-save-temps}.
3939 Report the CPU time taken by each subprocess in the compilation
3940 sequence. For C source files, this is the compiler proper and assembler
3941 (plus the linker if linking is done). The output looks like this:
3948 The first number on each line is the ``user time'', that is time spent
3949 executing the program itself. The second number is ``system time'',
3950 time spent executing operating system routines on behalf of the program.
3951 Both numbers are in seconds.
3953 @item -fvar-tracking
3954 @opindex fvar-tracking
3955 Run variable tracking pass. It computes where variables are stored at each
3956 position in code. Better debugging information is then generated
3957 (if the debugging information format supports this information).
3959 It is enabled by default when compiling with optimization (@option{-Os},
3960 @option{-O}, @option{-O2}, ...), debugging information (@option{-g}) and
3961 the debug info format supports it.
3963 @item -print-file-name=@var{library}
3964 @opindex print-file-name
3965 Print the full absolute name of the library file @var{library} that
3966 would be used when linking---and don't do anything else. With this
3967 option, GCC does not compile or link anything; it just prints the
3970 @item -print-multi-directory
3971 @opindex print-multi-directory
3972 Print the directory name corresponding to the multilib selected by any
3973 other switches present in the command line. This directory is supposed
3974 to exist in @env{GCC_EXEC_PREFIX}.
3976 @item -print-multi-lib
3977 @opindex print-multi-lib
3978 Print the mapping from multilib directory names to compiler switches
3979 that enable them. The directory name is separated from the switches by
3980 @samp{;}, and each switch starts with an @samp{@@} instead of the
3981 @samp{-}, without spaces between multiple switches. This is supposed to
3982 ease shell-processing.
3984 @item -print-prog-name=@var{program}
3985 @opindex print-prog-name
3986 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
3988 @item -print-libgcc-file-name
3989 @opindex print-libgcc-file-name
3990 Same as @option{-print-file-name=libgcc.a}.
3992 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
3993 but you do want to link with @file{libgcc.a}. You can do
3996 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
3999 @item -print-search-dirs
4000 @opindex print-search-dirs
4001 Print the name of the configured installation directory and a list of
4002 program and library directories @command{gcc} will search---and don't do anything else.
4004 This is useful when @command{gcc} prints the error message
4005 @samp{installation problem, cannot exec cpp0: No such file or directory}.
4006 To resolve this you either need to put @file{cpp0} and the other compiler
4007 components where @command{gcc} expects to find them, or you can set the environment
4008 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
4009 Don't forget the trailing @samp{/}.
4010 @xref{Environment Variables}.
4013 @opindex dumpmachine
4014 Print the compiler's target machine (for example,
4015 @samp{i686-pc-linux-gnu})---and don't do anything else.
4018 @opindex dumpversion
4019 Print the compiler version (for example, @samp{3.0})---and don't do
4024 Print the compiler's built-in specs---and don't do anything else. (This
4025 is used when GCC itself is being built.) @xref{Spec Files}.
4027 @item -feliminate-unused-debug-types
4028 @opindex feliminate-unused-debug-types
4029 Normally, when producing DWARF2 output, GCC will emit debugging
4030 information for all types declared in a compilation
4031 unit, regardless of whether or not they are actually used
4032 in that compilation unit. Sometimes this is useful, such as
4033 if, in the debugger, you want to cast a value to a type that is
4034 not actually used in your program (but is declared). More often,
4035 however, this results in a significant amount of wasted space.
4036 With this option, GCC will avoid producing debug symbol output
4037 for types that are nowhere used in the source file being compiled.
4040 @node Optimize Options
4041 @section Options That Control Optimization
4042 @cindex optimize options
4043 @cindex options, optimization
4045 These options control various sorts of optimizations.
4047 Without any optimization option, the compiler's goal is to reduce the
4048 cost of compilation and to make debugging produce the expected
4049 results. Statements are independent: if you stop the program with a
4050 breakpoint between statements, you can then assign a new value to any
4051 variable or change the program counter to any other statement in the
4052 function and get exactly the results you would expect from the source
4055 Turning on optimization flags makes the compiler attempt to improve
4056 the performance and/or code size at the expense of compilation time
4057 and possibly the ability to debug the program.
4059 The compiler performs optimization based on the knowledge it has of
4060 the program. Optimization levels @option{-O2} and above, in
4061 particular, enable @emph{unit-at-a-time} mode, which allows the
4062 compiler to consider information gained from later functions in
4063 the file when compiling a function. Compiling multiple files at
4064 once to a single output file in @emph{unit-at-a-time} mode allows
4065 the compiler to use information gained from all of the files when
4066 compiling each of them.
4068 Not all optimizations are controlled directly by a flag. Only
4069 optimizations that have a flag are listed.
4076 Optimize. Optimizing compilation takes somewhat more time, and a lot
4077 more memory for a large function.
4079 With @option{-O}, the compiler tries to reduce code size and execution
4080 time, without performing any optimizations that take a great deal of
4083 @option{-O} turns on the following optimization flags:
4084 @gccoptlist{-fdefer-pop @gol
4085 -fdelayed-branch @gol
4086 -fguess-branch-probability @gol
4087 -fcprop-registers @gol
4088 -floop-optimize @gol
4089 -fif-conversion @gol
4090 -fif-conversion2 @gol
4096 -ftree-live_range_split @gol
4098 -ftree-copyrename @gol
4103 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
4104 where doing so does not interfere with debugging.
4108 Optimize even more. GCC performs nearly all supported optimizations
4109 that do not involve a space-speed tradeoff. The compiler does not
4110 perform loop unrolling or function inlining when you specify @option{-O2}.
4111 As compared to @option{-O}, this option increases both compilation time
4112 and the performance of the generated code.
4114 @option{-O2} turns on all optimization flags specified by @option{-O}. It
4115 also turns on the following optimization flags:
4116 @gccoptlist{-fthread-jumps @gol
4118 -foptimize-sibling-calls @gol
4119 -fcse-follow-jumps -fcse-skip-blocks @gol
4120 -fgcse -fgcse-lm @gol
4121 -fexpensive-optimizations @gol
4122 -fstrength-reduce @gol
4123 -frerun-cse-after-loop -frerun-loop-opt @gol
4127 -fschedule-insns -fschedule-insns2 @gol
4128 -fsched-interblock -fsched-spec @gol
4130 -fstrict-aliasing @gol
4131 -fdelete-null-pointer-checks @gol
4132 -freorder-blocks -freorder-functions @gol
4133 -funit-at-a-time @gol
4134 -falign-functions -falign-jumps @gol
4135 -falign-loops -falign-labels @gol
4138 Please note the warning under @option{-fgcse} about
4139 invoking @option{-O2} on programs that use computed gotos.
4143 Optimize yet more. @option{-O3} turns on all optimizations specified by
4144 @option{-O2} and also turns on the @option{-finline-functions},
4145 @option{-funswitch-loops} and @option{-fgcse-after-reload} options.
4149 Do not optimize. This is the default.
4153 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
4154 do not typically increase code size. It also performs further
4155 optimizations designed to reduce code size.
4157 @option{-Os} disables the following optimization flags:
4158 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
4159 -falign-labels -freorder-blocks -freorder-blocks-and-partition -fprefetch-loop-arrays}
4161 If you use multiple @option{-O} options, with or without level numbers,
4162 the last such option is the one that is effective.
4165 Options of the form @option{-f@var{flag}} specify machine-independent
4166 flags. Most flags have both positive and negative forms; the negative
4167 form of @option{-ffoo} would be @option{-fno-foo}. In the table
4168 below, only one of the forms is listed---the one you typically will
4169 use. You can figure out the other form by either removing @samp{no-}
4172 The following options control specific optimizations. They are either
4173 activated by @option{-O} options or are related to ones that are. You
4174 can use the following flags in the rare cases when ``fine-tuning'' of
4175 optimizations to be performed is desired.
4178 @item -fno-default-inline
4179 @opindex fno-default-inline
4180 Do not make member functions inline by default merely because they are
4181 defined inside the class scope (C++ only). Otherwise, when you specify
4182 @w{@option{-O}}, member functions defined inside class scope are compiled
4183 inline by default; i.e., you don't need to add @samp{inline} in front of
4184 the member function name.
4186 @item -fno-defer-pop
4187 @opindex fno-defer-pop
4188 Always pop the arguments to each function call as soon as that function
4189 returns. For machines which must pop arguments after a function call,
4190 the compiler normally lets arguments accumulate on the stack for several
4191 function calls and pops them all at once.
4193 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4197 Force memory operands to be copied into registers before doing
4198 arithmetic on them. This produces better code by making all memory
4199 references potential common subexpressions. When they are not common
4200 subexpressions, instruction combination should eliminate the separate
4203 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4206 @opindex fforce-addr
4207 Force memory address constants to be copied into registers before
4208 doing arithmetic on them. This may produce better code just as
4209 @option{-fforce-mem} may.
4211 @item -fomit-frame-pointer
4212 @opindex fomit-frame-pointer
4213 Don't keep the frame pointer in a register for functions that
4214 don't need one. This avoids the instructions to save, set up and
4215 restore frame pointers; it also makes an extra register available
4216 in many functions. @strong{It also makes debugging impossible on
4219 On some machines, such as the VAX, this flag has no effect, because
4220 the standard calling sequence automatically handles the frame pointer
4221 and nothing is saved by pretending it doesn't exist. The
4222 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
4223 whether a target machine supports this flag. @xref{Registers,,Register
4224 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
4226 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4228 @item -foptimize-sibling-calls
4229 @opindex foptimize-sibling-calls
4230 Optimize sibling and tail recursive calls.
4232 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4236 Don't pay attention to the @code{inline} keyword. Normally this option
4237 is used to keep the compiler from expanding any functions inline.
4238 Note that if you are not optimizing, no functions can be expanded inline.
4240 @item -finline-functions
4241 @opindex finline-functions
4242 Integrate all simple functions into their callers. The compiler
4243 heuristically decides which functions are simple enough to be worth
4244 integrating in this way.
4246 If all calls to a given function are integrated, and the function is
4247 declared @code{static}, then the function is normally not output as
4248 assembler code in its own right.
4250 Enabled at level @option{-O3}.
4252 @item -finline-limit=@var{n}
4253 @opindex finline-limit
4254 By default, GCC limits the size of functions that can be inlined. This flag
4255 allows the control of this limit for functions that are explicitly marked as
4256 inline (i.e., marked with the inline keyword or defined within the class
4257 definition in c++). @var{n} is the size of functions that can be inlined in
4258 number of pseudo instructions (not counting parameter handling). The default
4259 value of @var{n} is 600.
4260 Increasing this value can result in more inlined code at
4261 the cost of compilation time and memory consumption. Decreasing usually makes
4262 the compilation faster and less code will be inlined (which presumably
4263 means slower programs). This option is particularly useful for programs that
4264 use inlining heavily such as those based on recursive templates with C++.
4266 Inlining is actually controlled by a number of parameters, which may be
4267 specified individually by using @option{--param @var{name}=@var{value}}.
4268 The @option{-finline-limit=@var{n}} option sets some of these parameters
4272 @item max-inline-insns-single
4273 is set to @var{n}/2.
4274 @item max-inline-insns-auto
4275 is set to @var{n}/2.
4276 @item min-inline-insns
4277 is set to 130 or @var{n}/4, whichever is smaller.
4278 @item max-inline-insns-rtl
4282 See below for a documentation of the individual
4283 parameters controlling inlining.
4285 @emph{Note:} pseudo instruction represents, in this particular context, an
4286 abstract measurement of function's size. In no way, it represents a count
4287 of assembly instructions and as such its exact meaning might change from one
4288 release to an another.
4290 @item -fkeep-inline-functions
4291 @opindex fkeep-inline-functions
4292 In C, emit @code{static} functions that are declared @code{inline}
4293 into the object file, even if the function has been inlined into all
4294 of its callers. This switch does not affect functions using the
4295 @code{extern inline} extension in GNU C@. In C++, emit any and all
4296 inline functions into the object file.
4298 @item -fkeep-static-consts
4299 @opindex fkeep-static-consts
4300 Emit variables declared @code{static const} when optimization isn't turned
4301 on, even if the variables aren't referenced.
4303 GCC enables this option by default. If you want to force the compiler to
4304 check if the variable was referenced, regardless of whether or not
4305 optimization is turned on, use the @option{-fno-keep-static-consts} option.
4307 @item -fmerge-constants
4308 Attempt to merge identical constants (string constants and floating point
4309 constants) across compilation units.
4311 This option is the default for optimized compilation if the assembler and
4312 linker support it. Use @option{-fno-merge-constants} to inhibit this
4315 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4317 @item -fmerge-all-constants
4318 Attempt to merge identical constants and identical variables.
4320 This option implies @option{-fmerge-constants}. In addition to
4321 @option{-fmerge-constants} this considers e.g.@: even constant initialized
4322 arrays or initialized constant variables with integral or floating point
4323 types. Languages like C or C++ require each non-automatic variable to
4324 have distinct location, so using this option will result in non-conforming
4327 @item -fmodulo-sched
4328 @opindex fmodulo-sched
4329 Perform swing modulo scheduling immediately before the first scheduling
4330 pass. This pass looks at innermost loops and reorders their
4331 instructions by overlapping different iterations.
4333 @item -fno-branch-count-reg
4334 @opindex fno-branch-count-reg
4335 Do not use ``decrement and branch'' instructions on a count register,
4336 but instead generate a sequence of instructions that decrement a
4337 register, compare it against zero, then branch based upon the result.
4338 This option is only meaningful on architectures that support such
4339 instructions, which include x86, PowerPC, IA-64 and S/390.
4341 The default is @option{-fbranch-count-reg}, enabled when
4342 @option{-fstrength-reduce} is enabled.
4344 @item -fno-function-cse
4345 @opindex fno-function-cse
4346 Do not put function addresses in registers; make each instruction that
4347 calls a constant function contain the function's address explicitly.
4349 This option results in less efficient code, but some strange hacks
4350 that alter the assembler output may be confused by the optimizations
4351 performed when this option is not used.
4353 The default is @option{-ffunction-cse}
4355 @item -fno-zero-initialized-in-bss
4356 @opindex fno-zero-initialized-in-bss
4357 If the target supports a BSS section, GCC by default puts variables that
4358 are initialized to zero into BSS@. This can save space in the resulting
4361 This option turns off this behavior because some programs explicitly
4362 rely on variables going to the data section. E.g., so that the
4363 resulting executable can find the beginning of that section and/or make
4364 assumptions based on that.
4366 The default is @option{-fzero-initialized-in-bss}.
4368 @item -fbounds-check
4369 @opindex fbounds-check
4370 For front-ends that support it, generate additional code to check that
4371 indices used to access arrays are within the declared range. This is
4372 currently only supported by the Java and Fortran front-ends, where
4373 this option defaults to true and false respectively.
4375 @item -fmudflap -fmudflapth -fmudflapir
4379 @cindex bounds checking
4381 For front-ends that support it (C and C++), instrument all risky
4382 pointer/array dereferencing operations, some standard library
4383 string/heap functions, and some other associated constructs with
4384 range/validity tests. Modules so instrumented should be immune to
4385 buffer overflows, invalid heap use, and some other classes of C/C++
4386 programming errors. The instrumentation relies on a separate runtime
4387 library (@file{libmudflap}), which will be linked into a program if
4388 @option{-fmudflap} is given at link time. Run-time behavior of the
4389 instrumented program is controlled by the @env{MUDFLAP_OPTIONS}
4390 environment variable. See @code{env MUDFLAP_OPTIONS=-help a.out}
4393 Use @option{-fmudflapth} instead of @option{-fmudflap} to compile and to
4394 link if your program is multi-threaded. Use @option{-fmudflapir}, in
4395 addition to @option{-fmudflap} or @option{-fmudflapth}, if
4396 instrumentation should ignore pointer reads. This produces less
4397 instrumentation (and therefore faster execution) and still provides
4398 some protection against outright memory corrupting writes, but allows
4399 erroneously read data to propagate within a program.
4401 @item -fstrength-reduce
4402 @opindex fstrength-reduce
4403 Perform the optimizations of loop strength reduction and
4404 elimination of iteration variables.
4406 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4408 @item -fthread-jumps
4409 @opindex fthread-jumps
4410 Perform optimizations where we check to see if a jump branches to a
4411 location where another comparison subsumed by the first is found. If
4412 so, the first branch is redirected to either the destination of the
4413 second branch or a point immediately following it, depending on whether
4414 the condition is known to be true or false.
4416 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4418 @item -fcse-follow-jumps
4419 @opindex fcse-follow-jumps
4420 In common subexpression elimination, scan through jump instructions
4421 when the target of the jump is not reached by any other path. For
4422 example, when CSE encounters an @code{if} statement with an
4423 @code{else} clause, CSE will follow the jump when the condition
4426 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4428 @item -fcse-skip-blocks
4429 @opindex fcse-skip-blocks
4430 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
4431 follow jumps which conditionally skip over blocks. When CSE
4432 encounters a simple @code{if} statement with no else clause,
4433 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
4434 body of the @code{if}.
4436 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4438 @item -frerun-cse-after-loop
4439 @opindex frerun-cse-after-loop
4440 Re-run common subexpression elimination after loop optimizations has been
4443 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4445 @item -frerun-loop-opt
4446 @opindex frerun-loop-opt
4447 Run the loop optimizer twice.
4449 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4453 Perform a global common subexpression elimination pass.
4454 This pass also performs global constant and copy propagation.
4456 @emph{Note:} When compiling a program using computed gotos, a GCC
4457 extension, you may get better runtime performance if you disable
4458 the global common subexpression elimination pass by adding
4459 @option{-fno-gcse} to the command line.
4461 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4465 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
4466 attempt to move loads which are only killed by stores into themselves. This
4467 allows a loop containing a load/store sequence to be changed to a load outside
4468 the loop, and a copy/store within the loop.
4470 Enabled by default when gcse is enabled.
4474 When @option{-fgcse-sm} is enabled, a store motion pass is run after
4475 global common subexpression elimination. This pass will attempt to move
4476 stores out of loops. When used in conjunction with @option{-fgcse-lm},
4477 loops containing a load/store sequence can be changed to a load before
4478 the loop and a store after the loop.
4480 Not enabled at any optimization level.
4484 When @option{-fgcse-las} is enabled, the global common subexpression
4485 elimination pass eliminates redundant loads that come after stores to the
4486 same memory location (both partial and full redundancies).
4488 Not enabled at any optimization level.
4490 @item -fgcse-after-reload
4491 @opindex fgcse-after-reload
4492 When @option{-fgcse-after-reload} is enabled, a redundant load elimination
4493 pass is performed after reload. The purpose of this pass is to cleanup
4496 @item -floop-optimize
4497 @opindex floop-optimize
4498 Perform loop optimizations: move constant expressions out of loops, simplify
4499 exit test conditions and optionally do strength-reduction as well.
4501 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4503 @item -floop-optimize2
4504 @opindex floop-optimize2
4505 Perform loop optimizations using the new loop optimizer. The optimizations
4506 (loop unrolling, peeling and unswitching, loop invariant motion) are enabled
4509 @item -fcrossjumping
4510 @opindex crossjumping
4511 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
4512 resulting code may or may not perform better than without cross-jumping.
4514 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4516 @item -fif-conversion
4517 @opindex if-conversion
4518 Attempt to transform conditional jumps into branch-less equivalents. This
4519 include use of conditional moves, min, max, set flags and abs instructions, and
4520 some tricks doable by standard arithmetics. The use of conditional execution
4521 on chips where it is available is controlled by @code{if-conversion2}.
4523 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4525 @item -fif-conversion2
4526 @opindex if-conversion2
4527 Use conditional execution (where available) to transform conditional jumps into
4528 branch-less equivalents.
4530 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4532 @item -fdelete-null-pointer-checks
4533 @opindex fdelete-null-pointer-checks
4534 Use global dataflow analysis to identify and eliminate useless checks
4535 for null pointers. The compiler assumes that dereferencing a null
4536 pointer would have halted the program. If a pointer is checked after
4537 it has already been dereferenced, it cannot be null.
4539 In some environments, this assumption is not true, and programs can
4540 safely dereference null pointers. Use
4541 @option{-fno-delete-null-pointer-checks} to disable this optimization
4542 for programs which depend on that behavior.
4544 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4546 @item -fexpensive-optimizations
4547 @opindex fexpensive-optimizations
4548 Perform a number of minor optimizations that are relatively expensive.
4550 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4552 @item -foptimize-register-move
4554 @opindex foptimize-register-move
4556 Attempt to reassign register numbers in move instructions and as
4557 operands of other simple instructions in order to maximize the amount of
4558 register tying. This is especially helpful on machines with two-operand
4561 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
4564 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4566 @item -fdelayed-branch
4567 @opindex fdelayed-branch
4568 If supported for the target machine, attempt to reorder instructions
4569 to exploit instruction slots available after delayed branch
4572 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4574 @item -fschedule-insns
4575 @opindex fschedule-insns
4576 If supported for the target machine, attempt to reorder instructions to
4577 eliminate execution stalls due to required data being unavailable. This
4578 helps machines that have slow floating point or memory load instructions
4579 by allowing other instructions to be issued until the result of the load
4580 or floating point instruction is required.
4582 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4584 @item -fschedule-insns2
4585 @opindex fschedule-insns2
4586 Similar to @option{-fschedule-insns}, but requests an additional pass of
4587 instruction scheduling after register allocation has been done. This is
4588 especially useful on machines with a relatively small number of
4589 registers and where memory load instructions take more than one cycle.
4591 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4593 @item -fno-sched-interblock
4594 @opindex fno-sched-interblock
4595 Don't schedule instructions across basic blocks. This is normally
4596 enabled by default when scheduling before register allocation, i.e.@:
4597 with @option{-fschedule-insns} or at @option{-O2} or higher.
4599 @item -fno-sched-spec
4600 @opindex fno-sched-spec
4601 Don't allow speculative motion of non-load instructions. This is normally
4602 enabled by default when scheduling before register allocation, i.e.@:
4603 with @option{-fschedule-insns} or at @option{-O2} or higher.
4605 @item -fsched-spec-load
4606 @opindex fsched-spec-load
4607 Allow speculative motion of some load instructions. This only makes
4608 sense when scheduling before register allocation, i.e.@: with
4609 @option{-fschedule-insns} or at @option{-O2} or higher.
4611 @item -fsched-spec-load-dangerous
4612 @opindex fsched-spec-load-dangerous
4613 Allow speculative motion of more load instructions. This only makes
4614 sense when scheduling before register allocation, i.e.@: with
4615 @option{-fschedule-insns} or at @option{-O2} or higher.
4617 @item -fsched-stalled-insns=@var{n}
4618 @opindex fsched-stalled-insns
4619 Define how many insns (if any) can be moved prematurely from the queue
4620 of stalled insns into the ready list, during the second scheduling pass.
4622 @item -fsched-stalled-insns-dep=@var{n}
4623 @opindex fsched-stalled-insns-dep
4624 Define how many insn groups (cycles) will be examined for a dependency
4625 on a stalled insn that is candidate for premature removal from the queue
4626 of stalled insns. Has an effect only during the second scheduling pass,
4627 and only if @option{-fsched-stalled-insns} is used and its value is not zero.
4629 @item -fsched2-use-superblocks
4630 @opindex fsched2-use-superblocks
4631 When scheduling after register allocation, do use superblock scheduling
4632 algorithm. Superblock scheduling allows motion across basic block boundaries
4633 resulting on faster schedules. This option is experimental, as not all machine
4634 descriptions used by GCC model the CPU closely enough to avoid unreliable
4635 results from the algorithm.
4637 This only makes sense when scheduling after register allocation, i.e.@: with
4638 @option{-fschedule-insns2} or at @option{-O2} or higher.
4640 @item -fsched2-use-traces
4641 @opindex fsched2-use-traces
4642 Use @option{-fsched2-use-superblocks} algorithm when scheduling after register
4643 allocation and additionally perform code duplication in order to increase the
4644 size of superblocks using tracer pass. See @option{-ftracer} for details on
4647 This mode should produce faster but significantly longer programs. Also
4648 without @option{-fbranch-probabilities} the traces constructed may not
4649 match the reality and hurt the performance. This only makes
4650 sense when scheduling after register allocation, i.e.@: with
4651 @option{-fschedule-insns2} or at @option{-O2} or higher.
4653 @item -freschedule-modulo-scheduled-loops
4654 @opindex fscheduling-in-modulo-scheduled-loops
4655 The modulo scheduling comes before the traditional scheduling, if a loop was modulo scheduled
4656 we may want to prevent the later scheduling passes from changing its schedule, we use this
4657 option to control that.
4659 @item -fcaller-saves
4660 @opindex fcaller-saves
4661 Enable values to be allocated in registers that will be clobbered by
4662 function calls, by emitting extra instructions to save and restore the
4663 registers around such calls. Such allocation is done only when it
4664 seems to result in better code than would otherwise be produced.
4666 This option is always enabled by default on certain machines, usually
4667 those which have no call-preserved registers to use instead.
4669 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4672 Perform Partial Redundancy Elimination (PRE) on trees. This flag is
4673 enabled by default at @option{-O2} and @option{-O3}.
4676 Perform Full Redundancy Elimination (FRE) on trees. The difference
4677 between FRE and PRE is that FRE only considers expressions
4678 that are computed on all paths leading to the redundant computation.
4679 This analysis faster than PRE, though it exposes fewer redundancies.
4680 This flag is enabled by default at @option{-O} and higher.
4683 Perform sparse conditional constant propagation (CCP) on trees. This flag
4684 is enabled by default at @option{-O} and higher.
4687 Perform dead code elimination (DCE) on trees. This flag is enabled by
4688 default at @option{-O} and higher.
4690 @item -ftree-dominator-opts
4691 Perform dead code elimination (DCE) on trees. This flag is enabled by
4692 default at @option{-O} and higher.
4695 Perform loop header copying on trees. This is beneficial since it increases
4696 effectiveness of code motion optimizations. It also saves one jump. This flag
4697 is enabled by default at @option{-O} and higher. It is not enabled
4698 for @option{-Os}, since it usually increases code size.
4700 @item -ftree-loop-optimize
4701 Perform loop optimizations on trees. This flag is enabled by default
4702 at @option{-O} and higher.
4704 @item -ftree-loop-linear
4705 Perform linear loop transformations on tree. This flag can improve cache
4706 performance and allow further loop optimizations to take place.
4709 Perform loop invariant motion on trees. This pass moves only invartiants that
4710 would be hard to handle on rtl level (function calls, operations that expand to
4711 nontrivial sequences of insns). With @option{-funswitch-loops} it also moves
4712 operands of conditions that are invariant out of the loop, so that we can use
4713 just trivial invariantness analysis in loop unswitching. The pass also includes
4717 Create a canonical counter for number of iterations in the loop for that
4718 determining number of iterations requires complicated analysis. Later
4719 optimizations then may determine the number easily. Useful especially
4720 in connection with unrolling.
4723 Perform induction variable optimizations (strength reduction, induction
4724 variable merging and induction variable elimination) on trees.
4727 Perform scalar replacement of aggregates. This pass replaces structure
4728 references with scalars to prevent committing structures to memory too
4729 early. This flag is enabled by default at @option{-O} and higher.
4731 @item -ftree-copyrename
4732 Perform copy renaming on trees. This pass attempts to rename compiler
4733 temporaries to other variables at copy locations, usually resulting in
4734 variable names which more closely resemble the original variables. This flag
4735 is enabled by default at @option{-O} and higher.
4738 Perform temporary expression replacement during the SSA->normal phase. Single
4739 use/single def temporaries are replaced at their use location with their
4740 defining expression. This results in non-GIMPLE code, but gives the expanders
4741 much more complex trees to work on resulting in better RTL generation. This is
4742 enabled by default at @option{-O} and higher.
4745 Perform live range splitting during the SSA->normal phase. Distinct live
4746 ranges of a variable are split into unique variables, allowing for better
4747 optimization later. This is enabled by default at @option{-O} and higher.
4749 @item -ftree-vectorize
4750 Perform loop vectorization on trees.
4754 Perform tail duplication to enlarge superblock size. This transformation
4755 simplifies the control flow of the function allowing other optimizations to do
4758 @item -funroll-loops
4759 @opindex funroll-loops
4760 Unroll loops whose number of iterations can be determined at compile
4761 time or upon entry to the loop. @option{-funroll-loops} implies both
4762 @option{-fstrength-reduce} and @option{-frerun-cse-after-loop}. This
4763 option makes code larger, and may or may not make it run faster.
4765 @item -funroll-all-loops
4766 @opindex funroll-all-loops
4767 Unroll all loops, even if their number of iterations is uncertain when
4768 the loop is entered. This usually makes programs run more slowly.
4769 @option{-funroll-all-loops} implies the same options as
4770 @option{-funroll-loops},
4772 @item -fsplit-ivs-in-unroller
4773 @opindex -fsplit-ivs-in-unroller
4774 Enables expressing of values of induction variables in later iterations
4775 of the unrolled loop using the value in the first iteration. This breaks
4776 long dependency chains, thus improving efficiency of the scheduling passes
4777 (for best results, @option{-fweb} should be used as well).
4779 Combination of @option{-fweb} and CSE is often sufficient to obtain the
4780 same effect. However in cases the loop body is more complicated than
4781 a single basic block, this is not reliable. It also does not work at all
4782 on some of the architectures due to restrictions in the CSE pass.
4784 This optimization is enabled by default.
4786 @item -fvariable-expansion-in-unroller
4787 @opindex -fvariable-expansion-in-unroller
4788 With this option, the compiler will create multiple copies of some
4789 local variables when unrolling a loop which can result in superior code.
4791 @item -fprefetch-loop-arrays
4792 @opindex fprefetch-loop-arrays
4793 If supported by the target machine, generate instructions to prefetch
4794 memory to improve the performance of loops that access large arrays.
4796 These options may generate better or worse code; results are highly
4797 dependent on the structure of loops within the source code.
4800 @itemx -fno-peephole2
4801 @opindex fno-peephole
4802 @opindex fno-peephole2
4803 Disable any machine-specific peephole optimizations. The difference
4804 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
4805 are implemented in the compiler; some targets use one, some use the
4806 other, a few use both.
4808 @option{-fpeephole} is enabled by default.
4809 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4811 @item -fno-guess-branch-probability
4812 @opindex fno-guess-branch-probability
4813 Do not guess branch probabilities using heuristics.
4815 GCC will use heuristics to guess branch probabilities if they are
4816 not provided by profiling feedback (@option{-fprofile-arcs}). These
4817 heuristics are based on the control flow graph. If some branch probabilities
4818 are specified by @samp{__builtin_expect}, then the heuristics will be
4819 used to guess branch probabilities for the rest of the control flow graph,
4820 taking the @samp{__builtin_expect} info into account. The interactions
4821 between the heuristics and @samp{__builtin_expect} can be complex, and in
4822 some cases, it may be useful to disable the heuristics so that the effects
4823 of @samp{__builtin_expect} are easier to understand.
4825 The default is @option{-fguess-branch-probability} at levels
4826 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
4828 @item -freorder-blocks
4829 @opindex freorder-blocks
4830 Reorder basic blocks in the compiled function in order to reduce number of
4831 taken branches and improve code locality.
4833 Enabled at levels @option{-O2}, @option{-O3}.
4835 @item -freorder-blocks-and-partition
4836 @opindex freorder-blocks-and-partition
4837 In addition to reordering basic blocks in the compiled function, in order
4838 to reduce number of taken branches, partitions hot and cold basic blocks
4839 into separate sections of the assembly and .o files, to improve
4840 paging and cache locality performance.
4842 This optimization is automatically turned off in the presence of
4843 exception handling, for linkonce sections, for functions with a user-defined
4844 section attribute and on any architecture that does not support named
4847 @item -freorder-functions
4848 @opindex freorder-functions
4849 Reorder functions in the object file in order to
4850 improve code locality. This is implemented by using special
4851 subsections @code{.text.hot} for most frequently executed functions and
4852 @code{.text.unlikely} for unlikely executed functions. Reordering is done by
4853 the linker so object file format must support named sections and linker must
4854 place them in a reasonable way.
4856 Also profile feedback must be available in to make this option effective. See
4857 @option{-fprofile-arcs} for details.
4859 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4861 @item -fstrict-aliasing
4862 @opindex fstrict-aliasing
4863 Allows the compiler to assume the strictest aliasing rules applicable to
4864 the language being compiled. For C (and C++), this activates
4865 optimizations based on the type of expressions. In particular, an
4866 object of one type is assumed never to reside at the same address as an
4867 object of a different type, unless the types are almost the same. For
4868 example, an @code{unsigned int} can alias an @code{int}, but not a
4869 @code{void*} or a @code{double}. A character type may alias any other
4872 Pay special attention to code like this:
4885 The practice of reading from a different union member than the one most
4886 recently written to (called ``type-punning'') is common. Even with
4887 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
4888 is accessed through the union type. So, the code above will work as
4889 expected. However, this code might not:
4900 Every language that wishes to perform language-specific alias analysis
4901 should define a function that computes, given an @code{tree}
4902 node, an alias set for the node. Nodes in different alias sets are not
4903 allowed to alias. For an example, see the C front-end function
4904 @code{c_get_alias_set}.
4906 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
4908 @item -falign-functions
4909 @itemx -falign-functions=@var{n}
4910 @opindex falign-functions
4911 Align the start of functions to the next power-of-two greater than
4912 @var{n}, skipping up to @var{n} bytes. For instance,
4913 @option{-falign-functions=32} aligns functions to the next 32-byte
4914 boundary, but @option{-falign-functions=24} would align to the next
4915 32-byte boundary only if this can be done by skipping 23 bytes or less.
4917 @option{-fno-align-functions} and @option{-falign-functions=1} are
4918 equivalent and mean that functions will not be aligned.
4920 Some assemblers only support this flag when @var{n} is a power of two;
4921 in that case, it is rounded up.
4923 If @var{n} is not specified or is zero, use a machine-dependent default.
4925 Enabled at levels @option{-O2}, @option{-O3}.
4927 @item -falign-labels
4928 @itemx -falign-labels=@var{n}
4929 @opindex falign-labels
4930 Align all branch targets to a power-of-two boundary, skipping up to
4931 @var{n} bytes like @option{-falign-functions}. This option can easily
4932 make code slower, because it must insert dummy operations for when the
4933 branch target is reached in the usual flow of the code.
4935 @option{-fno-align-labels} and @option{-falign-labels=1} are
4936 equivalent and mean that labels will not be aligned.
4938 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
4939 are greater than this value, then their values are used instead.
4941 If @var{n} is not specified or is zero, use a machine-dependent default
4942 which is very likely to be @samp{1}, meaning no alignment.
4944 Enabled at levels @option{-O2}, @option{-O3}.
4947 @itemx -falign-loops=@var{n}
4948 @opindex falign-loops
4949 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
4950 like @option{-falign-functions}. The hope is that the loop will be
4951 executed many times, which will make up for any execution of the dummy
4954 @option{-fno-align-loops} and @option{-falign-loops=1} are
4955 equivalent and mean that loops will not be aligned.
4957 If @var{n} is not specified or is zero, use a machine-dependent default.
4959 Enabled at levels @option{-O2}, @option{-O3}.
4962 @itemx -falign-jumps=@var{n}
4963 @opindex falign-jumps
4964 Align branch targets to a power-of-two boundary, for branch targets
4965 where the targets can only be reached by jumping, skipping up to @var{n}
4966 bytes like @option{-falign-functions}. In this case, no dummy operations
4969 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
4970 equivalent and mean that loops will not be aligned.
4972 If @var{n} is not specified or is zero, use a machine-dependent default.
4974 Enabled at levels @option{-O2}, @option{-O3}.
4976 @item -funit-at-a-time
4977 @opindex funit-at-a-time
4978 Parse the whole compilation unit before starting to produce code.
4979 This allows some extra optimizations to take place but consumes
4980 more memory (in general). There are some compatibility issues
4981 with @emph{unit-at-at-time} mode:
4984 enabling @emph{unit-at-a-time} mode may change the order
4985 in which functions, variables, and top-level @code{asm} statements
4986 are emitted, and will likely break code relying on some particular
4987 ordering. The majority of such top-level @code{asm} statements,
4988 though, can be replaced by @code{section} attributes.
4991 @emph{unit-at-a-time} mode removes unreferenced static variables
4992 and functions are removed. This may result in undefined references
4993 when an @code{asm} statement refers directly to variables or functions
4994 that are otherwise unused. In that case either the variable/function
4995 shall be listed as an operand of the @code{asm} statement operand or,
4996 in the case of top-level @code{asm} statements the attribute @code{used}
4997 shall be used on the declaration.
5000 Static functions now can use non-standard passing conventions that
5001 may break @code{asm} statements calling functions directly. Again,
5002 attribute @code{used} will prevent this behavior.
5005 As a temporary workaround, @option{-fno-unit-at-a-time} can be used,
5006 but this scheme may not be supported by future releases of GCC@.
5008 Enabled at levels @option{-O2}, @option{-O3}.
5012 Constructs webs as commonly used for register allocation purposes and assign
5013 each web individual pseudo register. This allows the register allocation pass
5014 to operate on pseudos directly, but also strengthens several other optimization
5015 passes, such as CSE, loop optimizer and trivial dead code remover. It can,
5016 however, make debugging impossible, since variables will no longer stay in a
5019 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os},
5020 on targets where the default format for debugging information supports
5023 @item -fno-cprop-registers
5024 @opindex fno-cprop-registers
5025 After register allocation and post-register allocation instruction splitting,
5026 we perform a copy-propagation pass to try to reduce scheduling dependencies
5027 and occasionally eliminate the copy.
5029 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5031 @item -fprofile-generate
5032 @opindex fprofile-generate
5034 Enable options usually used for instrumenting application to produce
5035 profile useful for later recompilation with profile feedback based
5036 optimization. You must use @option{-fprofile-generate} both when
5037 compiling and when linking your program.
5039 The following options are enabled: @code{-fprofile-arcs}, @code{-fprofile-values}, @code{-fvpt}.
5042 @opindex fprofile-use
5043 Enable profile feedback directed optimizations, and optimizations
5044 generally profitable only with profile feedback available.
5046 The following options are enabled: @code{-fbranch-probabilities},
5047 @code{-fvpt}, @code{-funroll-loops}, @code{-fpeel-loops}, @code{-ftracer}.
5051 The following options control compiler behavior regarding floating
5052 point arithmetic. These options trade off between speed and
5053 correctness. All must be specifically enabled.
5057 @opindex ffloat-store
5058 Do not store floating point variables in registers, and inhibit other
5059 options that might change whether a floating point value is taken from a
5062 @cindex floating point precision
5063 This option prevents undesirable excess precision on machines such as
5064 the 68000 where the floating registers (of the 68881) keep more
5065 precision than a @code{double} is supposed to have. Similarly for the
5066 x86 architecture. For most programs, the excess precision does only
5067 good, but a few programs rely on the precise definition of IEEE floating
5068 point. Use @option{-ffloat-store} for such programs, after modifying
5069 them to store all pertinent intermediate computations into variables.
5073 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations}, @*
5074 @option{-fno-trapping-math}, @option{-ffinite-math-only},
5075 @option{-fno-rounding-math} and @option{-fno-signaling-nans}.
5077 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
5079 This option should never be turned on by any @option{-O} option since
5080 it can result in incorrect output for programs which depend on
5081 an exact implementation of IEEE or ISO rules/specifications for
5084 @item -fno-math-errno
5085 @opindex fno-math-errno
5086 Do not set ERRNO after calling math functions that are executed
5087 with a single instruction, e.g., sqrt. A program that relies on
5088 IEEE exceptions for math error handling may want to use this flag
5089 for speed while maintaining IEEE arithmetic compatibility.
5091 This option should never be turned on by any @option{-O} option since
5092 it can result in incorrect output for programs which depend on
5093 an exact implementation of IEEE or ISO rules/specifications for
5096 The default is @option{-fmath-errno}.
5098 @item -funsafe-math-optimizations
5099 @opindex funsafe-math-optimizations
5100 Allow optimizations for floating-point arithmetic that (a) assume
5101 that arguments and results are valid and (b) may violate IEEE or
5102 ANSI standards. When used at link-time, it may include libraries
5103 or startup files that change the default FPU control word or other
5104 similar optimizations.
5106 This option should never be turned on by any @option{-O} option since
5107 it can result in incorrect output for programs which depend on
5108 an exact implementation of IEEE or ISO rules/specifications for
5111 The default is @option{-fno-unsafe-math-optimizations}.
5113 @item -ffinite-math-only
5114 @opindex ffinite-math-only
5115 Allow optimizations for floating-point arithmetic that assume
5116 that arguments and results are not NaNs or +-Infs.
5118 This option should never be turned on by any @option{-O} option since
5119 it can result in incorrect output for programs which depend on
5120 an exact implementation of IEEE or ISO rules/specifications.
5122 The default is @option{-fno-finite-math-only}.
5124 @item -fno-trapping-math
5125 @opindex fno-trapping-math
5126 Compile code assuming that floating-point operations cannot generate
5127 user-visible traps. These traps include division by zero, overflow,
5128 underflow, inexact result and invalid operation. This option implies
5129 @option{-fno-signaling-nans}. Setting this option may allow faster
5130 code if one relies on ``non-stop'' IEEE arithmetic, for example.
5132 This option should never be turned on by any @option{-O} option since
5133 it can result in incorrect output for programs which depend on
5134 an exact implementation of IEEE or ISO rules/specifications for
5137 The default is @option{-ftrapping-math}.
5139 @item -frounding-math
5140 @opindex frounding-math
5141 Disable transformations and optimizations that assume default floating
5142 point rounding behavior. This is round-to-zero for all floating point
5143 to integer conversions, and round-to-nearest for all other arithmetic
5144 truncations. This option should be specified for programs that change
5145 the FP rounding mode dynamically, or that may be executed with a
5146 non-default rounding mode. This option disables constant folding of
5147 floating point expressions at compile-time (which may be affected by
5148 rounding mode) and arithmetic transformations that are unsafe in the
5149 presence of sign-dependent rounding modes.
5151 The default is @option{-fno-rounding-math}.
5153 This option is experimental and does not currently guarantee to
5154 disable all GCC optimizations that are affected by rounding mode.
5155 Future versions of GCC may provide finer control of this setting
5156 using C99's @code{FENV_ACCESS} pragma. This command line option
5157 will be used to specify the default state for @code{FENV_ACCESS}.
5159 @item -fsignaling-nans
5160 @opindex fsignaling-nans
5161 Compile code assuming that IEEE signaling NaNs may generate user-visible
5162 traps during floating-point operations. Setting this option disables
5163 optimizations that may change the number of exceptions visible with
5164 signaling NaNs. This option implies @option{-ftrapping-math}.
5166 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
5169 The default is @option{-fno-signaling-nans}.
5171 This option is experimental and does not currently guarantee to
5172 disable all GCC optimizations that affect signaling NaN behavior.
5174 @item -fsingle-precision-constant
5175 @opindex fsingle-precision-constant
5176 Treat floating point constant as single precision constant instead of
5177 implicitly converting it to double precision constant.
5182 The following options control optimizations that may improve
5183 performance, but are not enabled by any @option{-O} options. This
5184 section includes experimental options that may produce broken code.
5187 @item -fbranch-probabilities
5188 @opindex fbranch-probabilities
5189 After running a program compiled with @option{-fprofile-arcs}
5190 (@pxref{Debugging Options,, Options for Debugging Your Program or
5191 @command{gcc}}), you can compile it a second time using
5192 @option{-fbranch-probabilities}, to improve optimizations based on
5193 the number of times each branch was taken. When the program
5194 compiled with @option{-fprofile-arcs} exits it saves arc execution
5195 counts to a file called @file{@var{sourcename}.gcda} for each source
5196 file The information in this data file is very dependent on the
5197 structure of the generated code, so you must use the same source code
5198 and the same optimization options for both compilations.
5200 With @option{-fbranch-probabilities}, GCC puts a
5201 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
5202 These can be used to improve optimization. Currently, they are only
5203 used in one place: in @file{reorg.c}, instead of guessing which path a
5204 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
5205 exactly determine which path is taken more often.
5207 @item -fprofile-values
5208 @opindex fprofile-values
5209 If combined with @option{-fprofile-arcs}, it adds code so that some
5210 data about values of expressions in the program is gathered.
5212 With @option{-fbranch-probabilities}, it reads back the data gathered
5213 from profiling values of expressions and adds @samp{REG_VALUE_PROFILE}
5214 notes to instructions for their later usage in optimizations.
5216 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
5220 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
5221 a code to gather information about values of expressions.
5223 With @option{-fbranch-probabilities}, it reads back the data gathered
5224 and actually performs the optimizations based on them.
5225 Currently the optimizations include specialization of division operation
5226 using the knowledge about the value of the denominator.
5228 @item -fspeculative-prefetching
5229 @opindex fspeculative-prefetching
5230 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
5231 a code to gather information about addresses of memory references in the
5234 With @option{-fbranch-probabilities}, it reads back the data gathered
5235 and issues prefetch instructions according to them. In addition to the opportunities
5236 noticed by @option{-fprefetch-loop-arrays}, it also notices more complicated
5237 memory access patterns---for example accesses to the data stored in linked
5238 list whose elements are usually allocated sequentially.
5240 In order to prevent issuing double prefetches, usage of
5241 @option{-fspeculative-prefetching} implies @option{-fno-prefetch-loop-arrays}.
5243 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
5245 @item -frename-registers
5246 @opindex frename-registers
5247 Attempt to avoid false dependencies in scheduled code by making use
5248 of registers left over after register allocation. This optimization
5249 will most benefit processors with lots of registers. Depending on the
5250 debug information format adopted by the target, however, it can
5251 make debugging impossible, since variables will no longer stay in
5252 a ``home register''.
5254 Not enabled by default at any level because it has known bugs.
5258 Perform tail duplication to enlarge superblock size. This transformation
5259 simplifies the control flow of the function allowing other optimizations to do
5262 Enabled with @option{-fprofile-use}.
5264 @item -funroll-loops
5265 @opindex funroll-loops
5266 Unroll loops whose number of iterations can be determined at compile time or
5267 upon entry to the loop. @option{-funroll-loops} implies
5268 @option{-frerun-cse-after-loop}. It also turns on complete loop peeling
5269 (i.e.@: complete removal of loops with small constant number of iterations).
5270 This option makes code larger, and may or may not make it run faster.
5272 Enabled with @option{-fprofile-use}.
5274 @item -funroll-all-loops
5275 @opindex funroll-all-loops
5276 Unroll all loops, even if their number of iterations is uncertain when
5277 the loop is entered. This usually makes programs run more slowly.
5278 @option{-funroll-all-loops} implies the same options as
5279 @option{-funroll-loops}.
5282 @opindex fpeel-loops
5283 Peels the loops for that there is enough information that they do not
5284 roll much (from profile feedback). It also turns on complete loop peeling
5285 (i.e.@: complete removal of loops with small constant number of iterations).
5287 Enabled with @option{-fprofile-use}.
5289 @item -fmove-loop-invariants
5290 @opindex fmove-loop-invariants
5291 Enables the loop invariant motion pass in the new loop optimizer. Enabled
5292 at level @option{-O1}
5294 @item -funswitch-loops
5295 @opindex funswitch-loops
5296 Move branches with loop invariant conditions out of the loop, with duplicates
5297 of the loop on both branches (modified according to result of the condition).
5299 @item -fprefetch-loop-arrays
5300 @opindex fprefetch-loop-arrays
5301 If supported by the target machine, generate instructions to prefetch
5302 memory to improve the performance of loops that access large arrays.
5304 Disabled at level @option{-Os}.
5306 @item -ffunction-sections
5307 @itemx -fdata-sections
5308 @opindex ffunction-sections
5309 @opindex fdata-sections
5310 Place each function or data item into its own section in the output
5311 file if the target supports arbitrary sections. The name of the
5312 function or the name of the data item determines the section's name
5315 Use these options on systems where the linker can perform optimizations
5316 to improve locality of reference in the instruction space. Most systems
5317 using the ELF object format and SPARC processors running Solaris 2 have
5318 linkers with such optimizations. AIX may have these optimizations in
5321 Only use these options when there are significant benefits from doing
5322 so. When you specify these options, the assembler and linker will
5323 create larger object and executable files and will also be slower.
5324 You will not be able to use @code{gprof} on all systems if you
5325 specify this option and you may have problems with debugging if
5326 you specify both this option and @option{-g}.
5328 @item -fbranch-target-load-optimize
5329 @opindex fbranch-target-load-optimize
5330 Perform branch target register load optimization before prologue / epilogue
5332 The use of target registers can typically be exposed only during reload,
5333 thus hoisting loads out of loops and doing inter-block scheduling needs
5334 a separate optimization pass.
5336 @item -fbranch-target-load-optimize2
5337 @opindex fbranch-target-load-optimize2
5338 Perform branch target register load optimization after prologue / epilogue
5341 @item -fbtr-bb-exclusive
5342 @opindex fbtr-bb-exclusive
5343 When performing branch target register load optimization, don't reuse
5344 branch target registers in within any basic block.
5346 @item --param @var{name}=@var{value}
5348 In some places, GCC uses various constants to control the amount of
5349 optimization that is done. For example, GCC will not inline functions
5350 that contain more that a certain number of instructions. You can
5351 control some of these constants on the command-line using the
5352 @option{--param} option.
5354 The names of specific parameters, and the meaning of the values, are
5355 tied to the internals of the compiler, and are subject to change
5356 without notice in future releases.
5358 In each case, the @var{value} is an integer. The allowable choices for
5359 @var{name} are given in the following table:
5362 @item sra-max-structure-size
5363 The maximum structure size, in bytes, at which the scalar replacement
5364 of aggregates (SRA) optimization will perform block copies. The
5365 default value, 0, implies that GCC will select the most appropriate
5368 @item sra-field-structure-ratio
5369 The threshold ratio (as a percentage) between instantiated fields and
5370 the complete structure size. We say that if the ratio of the number
5371 of bytes in instantiated fields to the number of bytes in the complete
5372 structure exceeds this parameter, then block copies are not used. The
5375 @item max-crossjump-edges
5376 The maximum number of incoming edges to consider for crossjumping.
5377 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
5378 the number of edges incoming to each block. Increasing values mean
5379 more aggressive optimization, making the compile time increase with
5380 probably small improvement in executable size.
5382 @item min-crossjump-insns
5383 The minimum number of instructions which must be matched at the end
5384 of two blocks before crossjumping will be performed on them. This
5385 value is ignored in the case where all instructions in the block being
5386 crossjumped from are matched. The default value is 5.
5388 @item max-goto-duplication-insns
5389 The maximum number of instructions to duplicate to a block that jumps
5390 to a computed goto. To avoid @math{O(N^2)} behavior in a number of
5391 passes, GCC factors computed gotos early in the compilation process,
5392 and unfactors them as late as possible. Only computed jumps at the
5393 end of a basic blocks with no more than max-goto-duplication-insns are
5394 unfactored. The default value is 8.
5396 @item max-delay-slot-insn-search
5397 The maximum number of instructions to consider when looking for an
5398 instruction to fill a delay slot. If more than this arbitrary number of
5399 instructions is searched, the time savings from filling the delay slot
5400 will be minimal so stop searching. Increasing values mean more
5401 aggressive optimization, making the compile time increase with probably
5402 small improvement in executable run time.
5404 @item max-delay-slot-live-search
5405 When trying to fill delay slots, the maximum number of instructions to
5406 consider when searching for a block with valid live register
5407 information. Increasing this arbitrarily chosen value means more
5408 aggressive optimization, increasing the compile time. This parameter
5409 should be removed when the delay slot code is rewritten to maintain the
5412 @item max-gcse-memory
5413 The approximate maximum amount of memory that will be allocated in
5414 order to perform the global common subexpression elimination
5415 optimization. If more memory than specified is required, the
5416 optimization will not be done.
5418 @item max-gcse-passes
5419 The maximum number of passes of GCSE to run. The default is 1.
5421 @item max-pending-list-length
5422 The maximum number of pending dependencies scheduling will allow
5423 before flushing the current state and starting over. Large functions
5424 with few branches or calls can create excessively large lists which
5425 needlessly consume memory and resources.
5427 @item max-inline-insns-single
5428 Several parameters control the tree inliner used in gcc.
5429 This number sets the maximum number of instructions (counted in GCC's
5430 internal representation) in a single function that the tree inliner
5431 will consider for inlining. This only affects functions declared
5432 inline and methods implemented in a class declaration (C++).
5433 The default value is 500.
5435 @item max-inline-insns-auto
5436 When you use @option{-finline-functions} (included in @option{-O3}),
5437 a lot of functions that would otherwise not be considered for inlining
5438 by the compiler will be investigated. To those functions, a different
5439 (more restrictive) limit compared to functions declared inline can
5441 The default value is 120.
5443 @item large-function-insns
5444 The limit specifying really large functions. For functions larger than this
5445 limit after inlining inlining is constrained by
5446 @option{--param large-function-growth}. This parameter is useful primarily
5447 to avoid extreme compilation time caused by non-linear algorithms used by the
5449 This parameter is ignored when @option{-funit-at-a-time} is not used.
5450 The default value is 3000.
5452 @item large-function-growth
5453 Specifies maximal growth of large function caused by inlining in percents.
5454 This parameter is ignored when @option{-funit-at-a-time} is not used.
5455 The default value is 100 which limits large function growth to 2.0 times
5458 @item inline-unit-growth
5459 Specifies maximal overall growth of the compilation unit caused by inlining.
5460 This parameter is ignored when @option{-funit-at-a-time} is not used.
5461 The default value is 50 which limits unit growth to 1.5 times the original
5464 @item max-inline-insns-recursive
5465 @itemx max-inline-insns-recursive-auto
5466 Specifies maximum number of instructions out-of-line copy of self recursive inline
5467 function can grow into by performing recursive inlining.
5469 For functions declared inline @option{--param max-inline-insns-recursive} is
5470 taken into acount. For function not declared inline, recursive inlining
5471 happens only when @option{-finline-functions} (included in @option{-O3}) is
5472 enabled and @option{--param max-inline-insns-recursive-auto} is used. The
5473 default value is 500.
5475 @item max-inline-recursive-depth
5476 @itemx max-inline-recursive-depth-auto
5477 Specifies maximum recursion depth used by the recursive inlining.
5479 For functions declared inline @option{--param max-inline-recursive-depth} is
5480 taken into acount. For function not declared inline, recursive inlining
5481 happens only when @option{-finline-functions} (included in @option{-O3}) is
5482 enabled and @option{--param max-inline-recursive-depth-auto} is used. The
5483 default value is 500.
5485 @item max-unrolled-insns
5486 The maximum number of instructions that a loop should have if that loop
5487 is unrolled, and if the loop is unrolled, it determines how many times
5488 the loop code is unrolled.
5490 @item max-average-unrolled-insns
5491 The maximum number of instructions biased by probabilities of their execution
5492 that a loop should have if that loop is unrolled, and if the loop is unrolled,
5493 it determines how many times the loop code is unrolled.
5495 @item max-unroll-times
5496 The maximum number of unrollings of a single loop.
5498 @item max-peeled-insns
5499 The maximum number of instructions that a loop should have if that loop
5500 is peeled, and if the loop is peeled, it determines how many times
5501 the loop code is peeled.
5503 @item max-peel-times
5504 The maximum number of peelings of a single loop.
5506 @item max-completely-peeled-insns
5507 The maximum number of insns of a completely peeled loop.
5509 @item max-completely-peel-times
5510 The maximum number of iterations of a loop to be suitable for complete peeling.
5512 @item max-unswitch-insns
5513 The maximum number of insns of an unswitched loop.
5515 @item max-unswitch-level
5516 The maximum number of branches unswitched in a single loop.
5519 The minimum cost of an expensive expression in the loop invariant motion.
5521 @item iv-consider-all-candidates-bound
5522 Bound on number of candidates for induction variables below that
5523 all candidates are considered for each use in induction variable
5524 optimizations. Only the most relevant candidates are considered
5525 if there are more candidates, to avoid quadratic time complexity.
5527 @item iv-max-considered-uses
5528 The induction variable optimizations give up on loops that contain more
5529 induction variable uses.
5531 @item iv-always-prune-cand-set-bound
5532 If number of candidates in the set is smaller than this value,
5533 we always try to remove unnecessary ivs from the set during its
5534 optimization when a new iv is added to the set.
5536 @item max-iterations-to-track
5538 The maximum number of iterations of a loop the brute force algorithm
5539 for analysis of # of iterations of the loop tries to evaluate.
5541 @item hot-bb-count-fraction
5542 Select fraction of the maximal count of repetitions of basic block in program
5543 given basic block needs to have to be considered hot.
5545 @item hot-bb-frequency-fraction
5546 Select fraction of the maximal frequency of executions of basic block in
5547 function given basic block needs to have to be considered hot
5549 @item tracer-dynamic-coverage
5550 @itemx tracer-dynamic-coverage-feedback
5552 This value is used to limit superblock formation once the given percentage of
5553 executed instructions is covered. This limits unnecessary code size
5556 The @option{tracer-dynamic-coverage-feedback} is used only when profile
5557 feedback is available. The real profiles (as opposed to statically estimated
5558 ones) are much less balanced allowing the threshold to be larger value.
5560 @item tracer-max-code-growth
5561 Stop tail duplication once code growth has reached given percentage. This is
5562 rather hokey argument, as most of the duplicates will be eliminated later in
5563 cross jumping, so it may be set to much higher values than is the desired code
5566 @item tracer-min-branch-ratio
5568 Stop reverse growth when the reverse probability of best edge is less than this
5569 threshold (in percent).
5571 @item tracer-min-branch-ratio
5572 @itemx tracer-min-branch-ratio-feedback
5574 Stop forward growth if the best edge do have probability lower than this
5577 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
5578 compilation for profile feedback and one for compilation without. The value
5579 for compilation with profile feedback needs to be more conservative (higher) in
5580 order to make tracer effective.
5582 @item max-cse-path-length
5584 Maximum number of basic blocks on path that cse considers. The default is 10.
5586 @item global-var-threshold
5588 Counts the number of function calls (@var{n}) and the number of
5589 call-clobbered variables (@var{v}). If @var{n}x@var{v} is larger than this limit, a
5590 single artificial variable will be created to represent all the
5591 call-clobbered variables at function call sites. This artificial
5592 variable will then be made to alias every call-clobbered variable.
5593 (done as @code{int * size_t} on the host machine; beware overflow).
5595 @item max-aliased-vops
5597 Maximum number of virtual operands allowed to represent aliases
5598 before triggering the alias grouping heuristic. Alias grouping
5599 reduces compile times and memory consumption needed for aliasing at
5600 the expense of precision loss in alias information.
5602 @item ggc-min-expand
5604 GCC uses a garbage collector to manage its own memory allocation. This
5605 parameter specifies the minimum percentage by which the garbage
5606 collector's heap should be allowed to expand between collections.
5607 Tuning this may improve compilation speed; it has no effect on code
5610 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
5611 RAM >= 1GB@. If @code{getrlimit} is available, the notion of "RAM" is
5612 the smallest of actual RAM and @code{RLIMIT_DATA} or @code{RLIMIT_AS}. If
5613 GCC is not able to calculate RAM on a particular platform, the lower
5614 bound of 30% is used. Setting this parameter and
5615 @option{ggc-min-heapsize} to zero causes a full collection to occur at
5616 every opportunity. This is extremely slow, but can be useful for
5619 @item ggc-min-heapsize
5621 Minimum size of the garbage collector's heap before it begins bothering
5622 to collect garbage. The first collection occurs after the heap expands
5623 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
5624 tuning this may improve compilation speed, and has no effect on code
5627 The default is the smaller of RAM/8, RLIMIT_RSS, or a limit which
5628 tries to ensure that RLIMIT_DATA or RLIMIT_AS are not exceeded, but
5629 with a lower bound of 4096 (four megabytes) and an upper bound of
5630 131072 (128 megabytes). If GCC is not able to calculate RAM on a
5631 particular platform, the lower bound is used. Setting this parameter
5632 very large effectively disables garbage collection. Setting this
5633 parameter and @option{ggc-min-expand} to zero causes a full collection
5634 to occur at every opportunity.
5636 @item max-reload-search-insns
5637 The maximum number of instruction reload should look backward for equivalent
5638 register. Increasing values mean more aggressive optimization, making the
5639 compile time increase with probably slightly better performance. The default
5642 @item max-cselib-memory-location
5643 The maximum number of memory locations cselib should take into acount.
5644 Increasing values mean more aggressive optimization, making the compile time
5645 increase with probably slightly better performance. The default value is 500.
5647 @item reorder-blocks-duplicate
5648 @itemx reorder-blocks-duplicate-feedback
5650 Used by basic block reordering pass to decide whether to use unconditional
5651 branch or duplicate the code on its destination. Code is duplicated when its
5652 estimated size is smaller than this value multiplied by the estimated size of
5653 unconditional jump in the hot spots of the program.
5655 The @option{reorder-block-duplicate-feedback} is used only when profile
5656 feedback is available and may be set to higher values than
5657 @option{reorder-block-duplicate} since information about the hot spots is more
5660 @item max-sched-region-blocks
5661 The maximum number of blocks in a region to be considered for
5662 interblock scheduling. The default value is 10.
5664 @item max-sched-region-insns
5665 The maximum number of insns in a region to be considered for
5666 interblock scheduling. The default value is 100.
5668 @item max-last-value-rtl
5670 The maximum size measured as number of RTLs that can be recorded in an expression
5671 in combiner for a pseudo register as last known value of that register. The default
5674 @item integer-share-limit
5675 Small integer constants can use a shared data structure, reducing the
5676 compiler's memory usage and increasing its speed. This sets the maximum
5677 value of a shared integer constant's. The default value is 256.
5682 @node Preprocessor Options
5683 @section Options Controlling the Preprocessor
5684 @cindex preprocessor options
5685 @cindex options, preprocessor
5687 These options control the C preprocessor, which is run on each C source
5688 file before actual compilation.
5690 If you use the @option{-E} option, nothing is done except preprocessing.
5691 Some of these options make sense only together with @option{-E} because
5692 they cause the preprocessor output to be unsuitable for actual
5697 You can use @option{-Wp,@var{option}} to bypass the compiler driver
5698 and pass @var{option} directly through to the preprocessor. If
5699 @var{option} contains commas, it is split into multiple options at the
5700 commas. However, many options are modified, translated or interpreted
5701 by the compiler driver before being passed to the preprocessor, and
5702 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
5703 interface is undocumented and subject to change, so whenever possible
5704 you should avoid using @option{-Wp} and let the driver handle the
5707 @item -Xpreprocessor @var{option}
5708 @opindex preprocessor
5709 Pass @var{option} as an option to the preprocessor. You can use this to
5710 supply system-specific preprocessor options which GCC does not know how to
5713 If you want to pass an option that takes an argument, you must use
5714 @option{-Xpreprocessor} twice, once for the option and once for the argument.
5717 @include cppopts.texi
5719 @node Assembler Options
5720 @section Passing Options to the Assembler
5722 @c prevent bad page break with this line
5723 You can pass options to the assembler.
5726 @item -Wa,@var{option}
5728 Pass @var{option} as an option to the assembler. If @var{option}
5729 contains commas, it is split into multiple options at the commas.
5731 @item -Xassembler @var{option}
5733 Pass @var{option} as an option to the assembler. You can use this to
5734 supply system-specific assembler options which GCC does not know how to
5737 If you want to pass an option that takes an argument, you must use
5738 @option{-Xassembler} twice, once for the option and once for the argument.
5743 @section Options for Linking
5744 @cindex link options
5745 @cindex options, linking
5747 These options come into play when the compiler links object files into
5748 an executable output file. They are meaningless if the compiler is
5749 not doing a link step.
5753 @item @var{object-file-name}
5754 A file name that does not end in a special recognized suffix is
5755 considered to name an object file or library. (Object files are
5756 distinguished from libraries by the linker according to the file
5757 contents.) If linking is done, these object files are used as input
5766 If any of these options is used, then the linker is not run, and
5767 object file names should not be used as arguments. @xref{Overall
5771 @item -l@var{library}
5772 @itemx -l @var{library}
5774 Search the library named @var{library} when linking. (The second
5775 alternative with the library as a separate argument is only for
5776 POSIX compliance and is not recommended.)
5778 It makes a difference where in the command you write this option; the
5779 linker searches and processes libraries and object files in the order they
5780 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
5781 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
5782 to functions in @samp{z}, those functions may not be loaded.
5784 The linker searches a standard list of directories for the library,
5785 which is actually a file named @file{lib@var{library}.a}. The linker
5786 then uses this file as if it had been specified precisely by name.
5788 The directories searched include several standard system directories
5789 plus any that you specify with @option{-L}.
5791 Normally the files found this way are library files---archive files
5792 whose members are object files. The linker handles an archive file by
5793 scanning through it for members which define symbols that have so far
5794 been referenced but not defined. But if the file that is found is an
5795 ordinary object file, it is linked in the usual fashion. The only
5796 difference between using an @option{-l} option and specifying a file name
5797 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
5798 and searches several directories.
5802 You need this special case of the @option{-l} option in order to
5803 link an Objective-C or Objective-C++ program.
5806 @opindex nostartfiles
5807 Do not use the standard system startup files when linking.
5808 The standard system libraries are used normally, unless @option{-nostdlib}
5809 or @option{-nodefaultlibs} is used.
5811 @item -nodefaultlibs
5812 @opindex nodefaultlibs
5813 Do not use the standard system libraries when linking.
5814 Only the libraries you specify will be passed to the linker.
5815 The standard startup files are used normally, unless @option{-nostartfiles}
5816 is used. The compiler may generate calls to @code{memcmp},
5817 @code{memset}, @code{memcpy} and @code{memmove}.
5818 These entries are usually resolved by entries in
5819 libc. These entry points should be supplied through some other
5820 mechanism when this option is specified.
5824 Do not use the standard system startup files or libraries when linking.
5825 No startup files and only the libraries you specify will be passed to
5826 the linker. The compiler may generate calls to @code{memcmp}, @code{memset},
5827 @code{memcpy} and @code{memmove}.
5828 These entries are usually resolved by entries in
5829 libc. These entry points should be supplied through some other
5830 mechanism when this option is specified.
5832 @cindex @option{-lgcc}, use with @option{-nostdlib}
5833 @cindex @option{-nostdlib} and unresolved references
5834 @cindex unresolved references and @option{-nostdlib}
5835 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
5836 @cindex @option{-nodefaultlibs} and unresolved references
5837 @cindex unresolved references and @option{-nodefaultlibs}
5838 One of the standard libraries bypassed by @option{-nostdlib} and
5839 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
5840 that GCC uses to overcome shortcomings of particular machines, or special
5841 needs for some languages.
5842 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
5843 Collection (GCC) Internals},
5844 for more discussion of @file{libgcc.a}.)
5845 In most cases, you need @file{libgcc.a} even when you want to avoid
5846 other standard libraries. In other words, when you specify @option{-nostdlib}
5847 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
5848 This ensures that you have no unresolved references to internal GCC
5849 library subroutines. (For example, @samp{__main}, used to ensure C++
5850 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
5851 GNU Compiler Collection (GCC) Internals}.)
5855 Produce a position independent executable on targets which support it.
5856 For predictable results, you must also specify the same set of options
5857 that were used to generate code (@option{-fpie}, @option{-fPIE},
5858 or model suboptions) when you specify this option.
5862 Remove all symbol table and relocation information from the executable.
5866 On systems that support dynamic linking, this prevents linking with the shared
5867 libraries. On other systems, this option has no effect.
5871 Produce a shared object which can then be linked with other objects to
5872 form an executable. Not all systems support this option. For predictable
5873 results, you must also specify the same set of options that were used to
5874 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
5875 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
5876 needs to build supplementary stub code for constructors to work. On
5877 multi-libbed systems, @samp{gcc -shared} must select the correct support
5878 libraries to link against. Failing to supply the correct flags may lead
5879 to subtle defects. Supplying them in cases where they are not necessary
5882 @item -shared-libgcc
5883 @itemx -static-libgcc
5884 @opindex shared-libgcc
5885 @opindex static-libgcc
5886 On systems that provide @file{libgcc} as a shared library, these options
5887 force the use of either the shared or static version respectively.
5888 If no shared version of @file{libgcc} was built when the compiler was
5889 configured, these options have no effect.
5891 There are several situations in which an application should use the
5892 shared @file{libgcc} instead of the static version. The most common
5893 of these is when the application wishes to throw and catch exceptions
5894 across different shared libraries. In that case, each of the libraries
5895 as well as the application itself should use the shared @file{libgcc}.
5897 Therefore, the G++ and GCJ drivers automatically add
5898 @option{-shared-libgcc} whenever you build a shared library or a main
5899 executable, because C++ and Java programs typically use exceptions, so
5900 this is the right thing to do.
5902 If, instead, you use the GCC driver to create shared libraries, you may
5903 find that they will not always be linked with the shared @file{libgcc}.
5904 If GCC finds, at its configuration time, that you have a non-GNU linker
5905 or a GNU linker that does not support option @option{--eh-frame-hdr},
5906 it will link the shared version of @file{libgcc} into shared libraries
5907 by default. Otherwise, it will take advantage of the linker and optimize
5908 away the linking with the shared version of @file{libgcc}, linking with
5909 the static version of libgcc by default. This allows exceptions to
5910 propagate through such shared libraries, without incurring relocation
5911 costs at library load time.
5913 However, if a library or main executable is supposed to throw or catch
5914 exceptions, you must link it using the G++ or GCJ driver, as appropriate
5915 for the languages used in the program, or using the option
5916 @option{-shared-libgcc}, such that it is linked with the shared
5921 Bind references to global symbols when building a shared object. Warn
5922 about any unresolved references (unless overridden by the link editor
5923 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
5926 @item -Xlinker @var{option}
5928 Pass @var{option} as an option to the linker. You can use this to
5929 supply system-specific linker options which GCC does not know how to
5932 If you want to pass an option that takes an argument, you must use
5933 @option{-Xlinker} twice, once for the option and once for the argument.
5934 For example, to pass @option{-assert definitions}, you must write
5935 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
5936 @option{-Xlinker "-assert definitions"}, because this passes the entire
5937 string as a single argument, which is not what the linker expects.
5939 @item -Wl,@var{option}
5941 Pass @var{option} as an option to the linker. If @var{option} contains
5942 commas, it is split into multiple options at the commas.
5944 @item -u @var{symbol}
5946 Pretend the symbol @var{symbol} is undefined, to force linking of
5947 library modules to define it. You can use @option{-u} multiple times with
5948 different symbols to force loading of additional library modules.
5951 @node Directory Options
5952 @section Options for Directory Search
5953 @cindex directory options
5954 @cindex options, directory search
5957 These options specify directories to search for header files, for
5958 libraries and for parts of the compiler:
5963 Add the directory @var{dir} to the head of the list of directories to be
5964 searched for header files. This can be used to override a system header
5965 file, substituting your own version, since these directories are
5966 searched before the system header file directories. However, you should
5967 not use this option to add directories that contain vendor-supplied
5968 system header files (use @option{-isystem} for that). If you use more than
5969 one @option{-I} option, the directories are scanned in left-to-right
5970 order; the standard system directories come after.
5972 If a standard system include directory, or a directory specified with
5973 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
5974 option will be ignored. The directory will still be searched but as a
5975 system directory at its normal position in the system include chain.
5976 This is to ensure that GCC's procedure to fix buggy system headers and
5977 the ordering for the include_next directive are not inadvertently changed.
5978 If you really need to change the search order for system directories,
5979 use the @option{-nostdinc} and/or @option{-isystem} options.
5981 @item -iquote@var{dir}
5983 Add the directory @var{dir} to the head of the list of directories to
5984 be searched for header files only for the case of @samp{#include
5985 "@var{file}"}; they are not searched for @samp{#include <@var{file}>},
5986 otherwise just like @option{-I}.
5990 Add directory @var{dir} to the list of directories to be searched
5993 @item -B@var{prefix}
5995 This option specifies where to find the executables, libraries,
5996 include files, and data files of the compiler itself.
5998 The compiler driver program runs one or more of the subprograms
5999 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
6000 @var{prefix} as a prefix for each program it tries to run, both with and
6001 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
6003 For each subprogram to be run, the compiler driver first tries the
6004 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
6005 was not specified, the driver tries two standard prefixes, which are
6006 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
6007 those results in a file name that is found, the unmodified program
6008 name is searched for using the directories specified in your
6009 @env{PATH} environment variable.
6011 The compiler will check to see if the path provided by the @option{-B}
6012 refers to a directory, and if necessary it will add a directory
6013 separator character at the end of the path.
6015 @option{-B} prefixes that effectively specify directory names also apply
6016 to libraries in the linker, because the compiler translates these
6017 options into @option{-L} options for the linker. They also apply to
6018 includes files in the preprocessor, because the compiler translates these
6019 options into @option{-isystem} options for the preprocessor. In this case,
6020 the compiler appends @samp{include} to the prefix.
6022 The run-time support file @file{libgcc.a} can also be searched for using
6023 the @option{-B} prefix, if needed. If it is not found there, the two
6024 standard prefixes above are tried, and that is all. The file is left
6025 out of the link if it is not found by those means.
6027 Another way to specify a prefix much like the @option{-B} prefix is to use
6028 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
6031 As a special kludge, if the path provided by @option{-B} is
6032 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
6033 9, then it will be replaced by @file{[dir/]include}. This is to help
6034 with boot-strapping the compiler.
6036 @item -specs=@var{file}
6038 Process @var{file} after the compiler reads in the standard @file{specs}
6039 file, in order to override the defaults that the @file{gcc} driver
6040 program uses when determining what switches to pass to @file{cc1},
6041 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
6042 @option{-specs=@var{file}} can be specified on the command line, and they
6043 are processed in order, from left to right.
6047 This option has been deprecated. Please use @option{-iquote} instead for
6048 @option{-I} directories before the @option{-I-} and remove the @option{-I-}.
6049 Any directories you specify with @option{-I} options before the @option{-I-}
6050 option are searched only for the case of @samp{#include "@var{file}"};
6051 they are not searched for @samp{#include <@var{file}>}.
6053 If additional directories are specified with @option{-I} options after
6054 the @option{-I-}, these directories are searched for all @samp{#include}
6055 directives. (Ordinarily @emph{all} @option{-I} directories are used
6058 In addition, the @option{-I-} option inhibits the use of the current
6059 directory (where the current input file came from) as the first search
6060 directory for @samp{#include "@var{file}"}. There is no way to
6061 override this effect of @option{-I-}. With @option{-I.} you can specify
6062 searching the directory which was current when the compiler was
6063 invoked. That is not exactly the same as what the preprocessor does
6064 by default, but it is often satisfactory.
6066 @option{-I-} does not inhibit the use of the standard system directories
6067 for header files. Thus, @option{-I-} and @option{-nostdinc} are
6074 @section Specifying subprocesses and the switches to pass to them
6077 @command{gcc} is a driver program. It performs its job by invoking a
6078 sequence of other programs to do the work of compiling, assembling and
6079 linking. GCC interprets its command-line parameters and uses these to
6080 deduce which programs it should invoke, and which command-line options
6081 it ought to place on their command lines. This behavior is controlled
6082 by @dfn{spec strings}. In most cases there is one spec string for each
6083 program that GCC can invoke, but a few programs have multiple spec
6084 strings to control their behavior. The spec strings built into GCC can
6085 be overridden by using the @option{-specs=} command-line switch to specify
6088 @dfn{Spec files} are plaintext files that are used to construct spec
6089 strings. They consist of a sequence of directives separated by blank
6090 lines. The type of directive is determined by the first non-whitespace
6091 character on the line and it can be one of the following:
6094 @item %@var{command}
6095 Issues a @var{command} to the spec file processor. The commands that can
6099 @item %include <@var{file}>
6101 Search for @var{file} and insert its text at the current point in the
6104 @item %include_noerr <@var{file}>
6105 @cindex %include_noerr
6106 Just like @samp{%include}, but do not generate an error message if the include
6107 file cannot be found.
6109 @item %rename @var{old_name} @var{new_name}
6111 Rename the spec string @var{old_name} to @var{new_name}.
6115 @item *[@var{spec_name}]:
6116 This tells the compiler to create, override or delete the named spec
6117 string. All lines after this directive up to the next directive or
6118 blank line are considered to be the text for the spec string. If this
6119 results in an empty string then the spec will be deleted. (Or, if the
6120 spec did not exist, then nothing will happened.) Otherwise, if the spec
6121 does not currently exist a new spec will be created. If the spec does
6122 exist then its contents will be overridden by the text of this
6123 directive, unless the first character of that text is the @samp{+}
6124 character, in which case the text will be appended to the spec.
6126 @item [@var{suffix}]:
6127 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
6128 and up to the next directive or blank line are considered to make up the
6129 spec string for the indicated suffix. When the compiler encounters an
6130 input file with the named suffix, it will processes the spec string in
6131 order to work out how to compile that file. For example:
6138 This says that any input file whose name ends in @samp{.ZZ} should be
6139 passed to the program @samp{z-compile}, which should be invoked with the
6140 command-line switch @option{-input} and with the result of performing the
6141 @samp{%i} substitution. (See below.)
6143 As an alternative to providing a spec string, the text that follows a
6144 suffix directive can be one of the following:
6147 @item @@@var{language}
6148 This says that the suffix is an alias for a known @var{language}. This is
6149 similar to using the @option{-x} command-line switch to GCC to specify a
6150 language explicitly. For example:
6157 Says that .ZZ files are, in fact, C++ source files.
6160 This causes an error messages saying:
6163 @var{name} compiler not installed on this system.
6167 GCC already has an extensive list of suffixes built into it.
6168 This directive will add an entry to the end of the list of suffixes, but
6169 since the list is searched from the end backwards, it is effectively
6170 possible to override earlier entries using this technique.
6174 GCC has the following spec strings built into it. Spec files can
6175 override these strings or create their own. Note that individual
6176 targets can also add their own spec strings to this list.
6179 asm Options to pass to the assembler
6180 asm_final Options to pass to the assembler post-processor
6181 cpp Options to pass to the C preprocessor
6182 cc1 Options to pass to the C compiler
6183 cc1plus Options to pass to the C++ compiler
6184 endfile Object files to include at the end of the link
6185 link Options to pass to the linker
6186 lib Libraries to include on the command line to the linker
6187 libgcc Decides which GCC support library to pass to the linker
6188 linker Sets the name of the linker
6189 predefines Defines to be passed to the C preprocessor
6190 signed_char Defines to pass to CPP to say whether @code{char} is signed
6192 startfile Object files to include at the start of the link
6195 Here is a small example of a spec file:
6201 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
6204 This example renames the spec called @samp{lib} to @samp{old_lib} and
6205 then overrides the previous definition of @samp{lib} with a new one.
6206 The new definition adds in some extra command-line options before
6207 including the text of the old definition.
6209 @dfn{Spec strings} are a list of command-line options to be passed to their
6210 corresponding program. In addition, the spec strings can contain
6211 @samp{%}-prefixed sequences to substitute variable text or to
6212 conditionally insert text into the command line. Using these constructs
6213 it is possible to generate quite complex command lines.
6215 Here is a table of all defined @samp{%}-sequences for spec
6216 strings. Note that spaces are not generated automatically around the
6217 results of expanding these sequences. Therefore you can concatenate them
6218 together or combine them with constant text in a single argument.
6222 Substitute one @samp{%} into the program name or argument.
6225 Substitute the name of the input file being processed.
6228 Substitute the basename of the input file being processed.
6229 This is the substring up to (and not including) the last period
6230 and not including the directory.
6233 This is the same as @samp{%b}, but include the file suffix (text after
6237 Marks the argument containing or following the @samp{%d} as a
6238 temporary file name, so that that file will be deleted if GCC exits
6239 successfully. Unlike @samp{%g}, this contributes no text to the
6242 @item %g@var{suffix}
6243 Substitute a file name that has suffix @var{suffix} and is chosen
6244 once per compilation, and mark the argument in the same way as
6245 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
6246 name is now chosen in a way that is hard to predict even when previously
6247 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
6248 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
6249 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
6250 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
6251 was simply substituted with a file name chosen once per compilation,
6252 without regard to any appended suffix (which was therefore treated
6253 just like ordinary text), making such attacks more likely to succeed.
6255 @item %u@var{suffix}
6256 Like @samp{%g}, but generates a new temporary file name even if
6257 @samp{%u@var{suffix}} was already seen.
6259 @item %U@var{suffix}
6260 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
6261 new one if there is no such last file name. In the absence of any
6262 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
6263 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
6264 would involve the generation of two distinct file names, one
6265 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
6266 simply substituted with a file name chosen for the previous @samp{%u},
6267 without regard to any appended suffix.
6269 @item %j@var{suffix}
6270 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
6271 writable, and if save-temps is off; otherwise, substitute the name
6272 of a temporary file, just like @samp{%u}. This temporary file is not
6273 meant for communication between processes, but rather as a junk
6276 @item %|@var{suffix}
6277 @itemx %m@var{suffix}
6278 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
6279 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
6280 all. These are the two most common ways to instruct a program that it
6281 should read from standard input or write to standard output. If you
6282 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
6283 construct: see for example @file{f/lang-specs.h}.
6285 @item %.@var{SUFFIX}
6286 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
6287 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
6288 terminated by the next space or %.
6291 Marks the argument containing or following the @samp{%w} as the
6292 designated output file of this compilation. This puts the argument
6293 into the sequence of arguments that @samp{%o} will substitute later.
6296 Substitutes the names of all the output files, with spaces
6297 automatically placed around them. You should write spaces
6298 around the @samp{%o} as well or the results are undefined.
6299 @samp{%o} is for use in the specs for running the linker.
6300 Input files whose names have no recognized suffix are not compiled
6301 at all, but they are included among the output files, so they will
6305 Substitutes the suffix for object files. Note that this is
6306 handled specially when it immediately follows @samp{%g, %u, or %U},
6307 because of the need for those to form complete file names. The
6308 handling is such that @samp{%O} is treated exactly as if it had already
6309 been substituted, except that @samp{%g, %u, and %U} do not currently
6310 support additional @var{suffix} characters following @samp{%O} as they would
6311 following, for example, @samp{.o}.
6314 Substitutes the standard macro predefinitions for the
6315 current target machine. Use this when running @code{cpp}.
6318 Like @samp{%p}, but puts @samp{__} before and after the name of each
6319 predefined macro, except for macros that start with @samp{__} or with
6320 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
6324 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
6325 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}), and
6326 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
6330 Current argument is the name of a library or startup file of some sort.
6331 Search for that file in a standard list of directories and substitute
6332 the full name found.
6335 Print @var{str} as an error message. @var{str} is terminated by a newline.
6336 Use this when inconsistent options are detected.
6339 Substitute the contents of spec string @var{name} at this point.
6342 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
6344 @item %x@{@var{option}@}
6345 Accumulate an option for @samp{%X}.
6348 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
6352 Output the accumulated assembler options specified by @option{-Wa}.
6355 Output the accumulated preprocessor options specified by @option{-Wp}.
6358 Process the @code{asm} spec. This is used to compute the
6359 switches to be passed to the assembler.
6362 Process the @code{asm_final} spec. This is a spec string for
6363 passing switches to an assembler post-processor, if such a program is
6367 Process the @code{link} spec. This is the spec for computing the
6368 command line passed to the linker. Typically it will make use of the
6369 @samp{%L %G %S %D and %E} sequences.
6372 Dump out a @option{-L} option for each directory that GCC believes might
6373 contain startup files. If the target supports multilibs then the
6374 current multilib directory will be prepended to each of these paths.
6377 Output the multilib directory with directory separators replaced with
6378 @samp{_}. If multilib directories are not set, or the multilib directory is
6379 @file{.} then this option emits nothing.
6382 Process the @code{lib} spec. This is a spec string for deciding which
6383 libraries should be included on the command line to the linker.
6386 Process the @code{libgcc} spec. This is a spec string for deciding
6387 which GCC support library should be included on the command line to the linker.
6390 Process the @code{startfile} spec. This is a spec for deciding which
6391 object files should be the first ones passed to the linker. Typically
6392 this might be a file named @file{crt0.o}.
6395 Process the @code{endfile} spec. This is a spec string that specifies
6396 the last object files that will be passed to the linker.
6399 Process the @code{cpp} spec. This is used to construct the arguments
6400 to be passed to the C preprocessor.
6403 Process the @code{cc1} spec. This is used to construct the options to be
6404 passed to the actual C compiler (@samp{cc1}).
6407 Process the @code{cc1plus} spec. This is used to construct the options to be
6408 passed to the actual C++ compiler (@samp{cc1plus}).
6411 Substitute the variable part of a matched option. See below.
6412 Note that each comma in the substituted string is replaced by
6416 Remove all occurrences of @code{-S} from the command line. Note---this
6417 command is position dependent. @samp{%} commands in the spec string
6418 before this one will see @code{-S}, @samp{%} commands in the spec string
6419 after this one will not.
6421 @item %:@var{function}(@var{args})
6422 Call the named function @var{function}, passing it @var{args}.
6423 @var{args} is first processed as a nested spec string, then split
6424 into an argument vector in the usual fashion. The function returns
6425 a string which is processed as if it had appeared literally as part
6426 of the current spec.
6428 The following built-in spec functions are provided:
6431 @item @code{if-exists}
6432 The @code{if-exists} spec function takes one argument, an absolute
6433 pathname to a file. If the file exists, @code{if-exists} returns the
6434 pathname. Here is a small example of its usage:
6438 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
6441 @item @code{if-exists-else}
6442 The @code{if-exists-else} spec function is similar to the @code{if-exists}
6443 spec function, except that it takes two arguments. The first argument is
6444 an absolute pathname to a file. If the file exists, @code{if-exists-else}
6445 returns the pathname. If it does not exist, it returns the second argument.
6446 This way, @code{if-exists-else} can be used to select one file or another,
6447 based on the existence of the first. Here is a small example of its usage:
6451 crt0%O%s %:if-exists(crti%O%s) \
6452 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
6455 @item @code{replace-outfile}
6456 The @code{replace-outfile} spec function takes two arguments. It looks for the
6457 first argument in the outfiles array and replaces it with the second argument. Here
6458 is a small example of its usage:
6461 %@{fgnu-runtime:%:replace-outfile(-lobjc -lobjc-gnu)@}
6467 Substitutes the @code{-S} switch, if that switch was given to GCC@.
6468 If that switch was not specified, this substitutes nothing. Note that
6469 the leading dash is omitted when specifying this option, and it is
6470 automatically inserted if the substitution is performed. Thus the spec
6471 string @samp{%@{foo@}} would match the command-line option @option{-foo}
6472 and would output the command line option @option{-foo}.
6474 @item %W@{@code{S}@}
6475 Like %@{@code{S}@} but mark last argument supplied within as a file to be
6478 @item %@{@code{S}*@}
6479 Substitutes all the switches specified to GCC whose names start
6480 with @code{-S}, but which also take an argument. This is used for
6481 switches like @option{-o}, @option{-D}, @option{-I}, etc.
6482 GCC considers @option{-o foo} as being
6483 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
6484 text, including the space. Thus two arguments would be generated.
6486 @item %@{@code{S}*&@code{T}*@}
6487 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
6488 (the order of @code{S} and @code{T} in the spec is not significant).
6489 There can be any number of ampersand-separated variables; for each the
6490 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
6492 @item %@{@code{S}:@code{X}@}
6493 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
6495 @item %@{!@code{S}:@code{X}@}
6496 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
6498 @item %@{@code{S}*:@code{X}@}
6499 Substitutes @code{X} if one or more switches whose names start with
6500 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
6501 once, no matter how many such switches appeared. However, if @code{%*}
6502 appears somewhere in @code{X}, then @code{X} will be substituted once
6503 for each matching switch, with the @code{%*} replaced by the part of
6504 that switch that matched the @code{*}.
6506 @item %@{.@code{S}:@code{X}@}
6507 Substitutes @code{X}, if processing a file with suffix @code{S}.
6509 @item %@{!.@code{S}:@code{X}@}
6510 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
6512 @item %@{@code{S}|@code{P}:@code{X}@}
6513 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC@.
6514 This may be combined with @samp{!}, @samp{.}, and @code{*} sequences as well,
6515 although they have a stronger binding than the @samp{|}. If @code{%*}
6516 appears in @code{X}, all of the alternatives must be starred, and only
6517 the first matching alternative is substituted.
6519 For example, a spec string like this:
6522 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
6525 will output the following command-line options from the following input
6526 command-line options:
6531 -d fred.c -foo -baz -boggle
6532 -d jim.d -bar -baz -boggle
6535 @item %@{S:X; T:Y; :D@}
6537 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
6538 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
6539 be as many clauses as you need. This may be combined with @code{.},
6540 @code{!}, @code{|}, and @code{*} as needed.
6545 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
6546 construct may contain other nested @samp{%} constructs or spaces, or
6547 even newlines. They are processed as usual, as described above.
6548 Trailing white space in @code{X} is ignored. White space may also
6549 appear anywhere on the left side of the colon in these constructs,
6550 except between @code{.} or @code{*} and the corresponding word.
6552 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
6553 handled specifically in these constructs. If another value of
6554 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
6555 @option{-W} switch is found later in the command line, the earlier
6556 switch value is ignored, except with @{@code{S}*@} where @code{S} is
6557 just one letter, which passes all matching options.
6559 The character @samp{|} at the beginning of the predicate text is used to
6560 indicate that a command should be piped to the following command, but
6561 only if @option{-pipe} is specified.
6563 It is built into GCC which switches take arguments and which do not.
6564 (You might think it would be useful to generalize this to allow each
6565 compiler's spec to say which switches take arguments. But this cannot
6566 be done in a consistent fashion. GCC cannot even decide which input
6567 files have been specified without knowing which switches take arguments,
6568 and it must know which input files to compile in order to tell which
6571 GCC also knows implicitly that arguments starting in @option{-l} are to be
6572 treated as compiler output files, and passed to the linker in their
6573 proper position among the other output files.
6575 @c man begin OPTIONS
6577 @node Target Options
6578 @section Specifying Target Machine and Compiler Version
6579 @cindex target options
6580 @cindex cross compiling
6581 @cindex specifying machine version
6582 @cindex specifying compiler version and target machine
6583 @cindex compiler version, specifying
6584 @cindex target machine, specifying
6586 The usual way to run GCC is to run the executable called @file{gcc}, or
6587 @file{<machine>-gcc} when cross-compiling, or
6588 @file{<machine>-gcc-<version>} to run a version other than the one that
6589 was installed last. Sometimes this is inconvenient, so GCC provides
6590 options that will switch to another cross-compiler or version.
6593 @item -b @var{machine}
6595 The argument @var{machine} specifies the target machine for compilation.
6597 The value to use for @var{machine} is the same as was specified as the
6598 machine type when configuring GCC as a cross-compiler. For
6599 example, if a cross-compiler was configured with @samp{configure
6600 i386v}, meaning to compile for an 80386 running System V, then you
6601 would specify @option{-b i386v} to run that cross compiler.
6603 @item -V @var{version}
6605 The argument @var{version} specifies which version of GCC to run.
6606 This is useful when multiple versions are installed. For example,
6607 @var{version} might be @samp{2.0}, meaning to run GCC version 2.0.
6610 The @option{-V} and @option{-b} options work by running the
6611 @file{<machine>-gcc-<version>} executable, so there's no real reason to
6612 use them if you can just run that directly.
6614 @node Submodel Options
6615 @section Hardware Models and Configurations
6616 @cindex submodel options
6617 @cindex specifying hardware config
6618 @cindex hardware models and configurations, specifying
6619 @cindex machine dependent options
6621 Earlier we discussed the standard option @option{-b} which chooses among
6622 different installed compilers for completely different target
6623 machines, such as VAX vs.@: 68000 vs.@: 80386.
6625 In addition, each of these target machine types can have its own
6626 special options, starting with @samp{-m}, to choose among various
6627 hardware models or configurations---for example, 68010 vs 68020,
6628 floating coprocessor or none. A single installed version of the
6629 compiler can compile for any model or configuration, according to the
6632 Some configurations of the compiler also support additional special
6633 options, usually for compatibility with other compilers on the same
6636 These options are defined by the macro @code{TARGET_SWITCHES} in the
6637 machine description. The default for the options is also defined by
6638 that macro, which enables you to change the defaults.
6640 @c This list is ordered alphanumerically by subsection name.
6641 @c It should be the same order and spelling as these options are listed
6642 @c in Machine Dependent Options
6650 * DEC Alpha Options::
6651 * DEC Alpha/VMS Options::
6655 * i386 and x86-64 Options::
6667 * RS/6000 and PowerPC Options::
6668 * S/390 and zSeries Options::
6671 * System V Options::
6672 * TMS320C3x/C4x Options::
6676 * Xstormy16 Options::
6682 @subsection ARC Options
6685 These options are defined for ARC implementations:
6690 Compile code for little endian mode. This is the default.
6694 Compile code for big endian mode.
6697 @opindex mmangle-cpu
6698 Prepend the name of the cpu to all public symbol names.
6699 In multiple-processor systems, there are many ARC variants with different
6700 instruction and register set characteristics. This flag prevents code
6701 compiled for one cpu to be linked with code compiled for another.
6702 No facility exists for handling variants that are ``almost identical''.
6703 This is an all or nothing option.
6705 @item -mcpu=@var{cpu}
6707 Compile code for ARC variant @var{cpu}.
6708 Which variants are supported depend on the configuration.
6709 All variants support @option{-mcpu=base}, this is the default.
6711 @item -mtext=@var{text-section}
6712 @itemx -mdata=@var{data-section}
6713 @itemx -mrodata=@var{readonly-data-section}
6717 Put functions, data, and readonly data in @var{text-section},
6718 @var{data-section}, and @var{readonly-data-section} respectively
6719 by default. This can be overridden with the @code{section} attribute.
6720 @xref{Variable Attributes}.
6725 @subsection ARM Options
6728 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
6732 @item -mabi=@var{name}
6734 Generate code for the specified ABI@. Permissible values are: @samp{apcs-gnu},
6735 @samp{atpcs}, @samp{aapcs} and @samp{iwmmxt}.
6738 @opindex mapcs-frame
6739 Generate a stack frame that is compliant with the ARM Procedure Call
6740 Standard for all functions, even if this is not strictly necessary for
6741 correct execution of the code. Specifying @option{-fomit-frame-pointer}
6742 with this option will cause the stack frames not to be generated for
6743 leaf functions. The default is @option{-mno-apcs-frame}.
6747 This is a synonym for @option{-mapcs-frame}.
6750 @c not currently implemented
6751 @item -mapcs-stack-check
6752 @opindex mapcs-stack-check
6753 Generate code to check the amount of stack space available upon entry to
6754 every function (that actually uses some stack space). If there is
6755 insufficient space available then either the function
6756 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
6757 called, depending upon the amount of stack space required. The run time
6758 system is required to provide these functions. The default is
6759 @option{-mno-apcs-stack-check}, since this produces smaller code.
6761 @c not currently implemented
6763 @opindex mapcs-float
6764 Pass floating point arguments using the float point registers. This is
6765 one of the variants of the APCS@. This option is recommended if the
6766 target hardware has a floating point unit or if a lot of floating point
6767 arithmetic is going to be performed by the code. The default is
6768 @option{-mno-apcs-float}, since integer only code is slightly increased in
6769 size if @option{-mapcs-float} is used.
6771 @c not currently implemented
6772 @item -mapcs-reentrant
6773 @opindex mapcs-reentrant
6774 Generate reentrant, position independent code. The default is
6775 @option{-mno-apcs-reentrant}.
6778 @item -mthumb-interwork
6779 @opindex mthumb-interwork
6780 Generate code which supports calling between the ARM and Thumb
6781 instruction sets. Without this option the two instruction sets cannot
6782 be reliably used inside one program. The default is
6783 @option{-mno-thumb-interwork}, since slightly larger code is generated
6784 when @option{-mthumb-interwork} is specified.
6786 @item -mno-sched-prolog
6787 @opindex mno-sched-prolog
6788 Prevent the reordering of instructions in the function prolog, or the
6789 merging of those instruction with the instructions in the function's
6790 body. This means that all functions will start with a recognizable set
6791 of instructions (or in fact one of a choice from a small set of
6792 different function prologues), and this information can be used to
6793 locate the start if functions inside an executable piece of code. The
6794 default is @option{-msched-prolog}.
6797 @opindex mhard-float
6798 Generate output containing floating point instructions. This is the
6802 @opindex msoft-float
6803 Generate output containing library calls for floating point.
6804 @strong{Warning:} the requisite libraries are not available for all ARM
6805 targets. Normally the facilities of the machine's usual C compiler are
6806 used, but this cannot be done directly in cross-compilation. You must make
6807 your own arrangements to provide suitable library functions for
6810 @option{-msoft-float} changes the calling convention in the output file;
6811 therefore, it is only useful if you compile @emph{all} of a program with
6812 this option. In particular, you need to compile @file{libgcc.a}, the
6813 library that comes with GCC, with @option{-msoft-float} in order for
6816 @item -mfloat-abi=@var{name}
6818 Specifies which ABI to use for floating point values. Permissible values
6819 are: @samp{soft}, @samp{softfp} and @samp{hard}.
6821 @samp{soft} and @samp{hard} are equivalent to @option{-msoft-float}
6822 and @option{-mhard-float} respectively. @samp{softfp} allows the generation
6823 of floating point instructions, but still uses the soft-float calling
6826 @item -mlittle-endian
6827 @opindex mlittle-endian
6828 Generate code for a processor running in little-endian mode. This is
6829 the default for all standard configurations.
6832 @opindex mbig-endian
6833 Generate code for a processor running in big-endian mode; the default is
6834 to compile code for a little-endian processor.
6836 @item -mwords-little-endian
6837 @opindex mwords-little-endian
6838 This option only applies when generating code for big-endian processors.
6839 Generate code for a little-endian word order but a big-endian byte
6840 order. That is, a byte order of the form @samp{32107654}. Note: this
6841 option should only be used if you require compatibility with code for
6842 big-endian ARM processors generated by versions of the compiler prior to
6845 @item -mcpu=@var{name}
6847 This specifies the name of the target ARM processor. GCC uses this name
6848 to determine what kind of instructions it can emit when generating
6849 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
6850 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
6851 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
6852 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
6853 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
6854 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm7tdmi-s},
6855 @samp{arm8}, @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
6856 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
6857 @samp{arm920t}, @samp{arm922t}, @samp{arm946e-s}, @samp{arm966e-s},
6858 @samp{arm968e-s}, @samp{arm926ej-s}, @samp{arm940t}, @samp{arm9tdmi},
6859 @samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ej-s},
6860 @samp{arm10e}, @samp{arm1020e}, @samp{arm1022e},
6861 @samp{arm1136j-s}, @samp{arm1136jf-s}, @samp{mpcore}, @samp{mpcorenovfp},
6862 @samp{arm1176jz-s}, @samp{arm1176jzf-s}, @samp{xscale}, @samp{iwmmxt},
6865 @itemx -mtune=@var{name}
6867 This option is very similar to the @option{-mcpu=} option, except that
6868 instead of specifying the actual target processor type, and hence
6869 restricting which instructions can be used, it specifies that GCC should
6870 tune the performance of the code as if the target were of the type
6871 specified in this option, but still choosing the instructions that it
6872 will generate based on the cpu specified by a @option{-mcpu=} option.
6873 For some ARM implementations better performance can be obtained by using
6876 @item -march=@var{name}
6878 This specifies the name of the target ARM architecture. GCC uses this
6879 name to determine what kind of instructions it can emit when generating
6880 assembly code. This option can be used in conjunction with or instead
6881 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
6882 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
6883 @samp{armv5}, @samp{armv5t}, @samp{armv5te}, @samp{armv6}, @samp{armv6j},
6884 @samp{iwmmxt}, @samp{ep9312}.
6886 @item -mfpu=@var{name}
6887 @itemx -mfpe=@var{number}
6888 @itemx -mfp=@var{number}
6892 This specifies what floating point hardware (or hardware emulation) is
6893 available on the target. Permissible names are: @samp{fpa}, @samp{fpe2},
6894 @samp{fpe3}, @samp{maverick}, @samp{vfp}. @option{-mfp} and @option{-mfpe}
6895 are synonyms for @option{-mfpu}=@samp{fpe}@var{number}, for compatibility
6896 with older versions of GCC@.
6898 If @option{-msoft-float} is specified this specifies the format of
6899 floating point values.
6901 @item -mstructure-size-boundary=@var{n}
6902 @opindex mstructure-size-boundary
6903 The size of all structures and unions will be rounded up to a multiple
6904 of the number of bits set by this option. Permissible values are 8, 32
6905 and 64. The default value varies for different toolchains. For the COFF
6906 targeted toolchain the default value is 8. A value of 64 is only allowed
6907 if the underlying ABI supports it.
6909 Specifying the larger number can produce faster, more efficient code, but
6910 can also increase the size of the program. Different values are potentially
6911 incompatible. Code compiled with one value cannot necessarily expect to
6912 work with code or libraries compiled with another value, if they exchange
6913 information using structures or unions.
6915 @item -mabort-on-noreturn
6916 @opindex mabort-on-noreturn
6917 Generate a call to the function @code{abort} at the end of a
6918 @code{noreturn} function. It will be executed if the function tries to
6922 @itemx -mno-long-calls
6923 @opindex mlong-calls
6924 @opindex mno-long-calls
6925 Tells the compiler to perform function calls by first loading the
6926 address of the function into a register and then performing a subroutine
6927 call on this register. This switch is needed if the target function
6928 will lie outside of the 64 megabyte addressing range of the offset based
6929 version of subroutine call instruction.
6931 Even if this switch is enabled, not all function calls will be turned
6932 into long calls. The heuristic is that static functions, functions
6933 which have the @samp{short-call} attribute, functions that are inside
6934 the scope of a @samp{#pragma no_long_calls} directive and functions whose
6935 definitions have already been compiled within the current compilation
6936 unit, will not be turned into long calls. The exception to this rule is
6937 that weak function definitions, functions with the @samp{long-call}
6938 attribute or the @samp{section} attribute, and functions that are within
6939 the scope of a @samp{#pragma long_calls} directive, will always be
6940 turned into long calls.
6942 This feature is not enabled by default. Specifying
6943 @option{-mno-long-calls} will restore the default behavior, as will
6944 placing the function calls within the scope of a @samp{#pragma
6945 long_calls_off} directive. Note these switches have no effect on how
6946 the compiler generates code to handle function calls via function
6949 @item -mnop-fun-dllimport
6950 @opindex mnop-fun-dllimport
6951 Disable support for the @code{dllimport} attribute.
6953 @item -msingle-pic-base
6954 @opindex msingle-pic-base
6955 Treat the register used for PIC addressing as read-only, rather than
6956 loading it in the prologue for each function. The run-time system is
6957 responsible for initializing this register with an appropriate value
6958 before execution begins.
6960 @item -mpic-register=@var{reg}
6961 @opindex mpic-register
6962 Specify the register to be used for PIC addressing. The default is R10
6963 unless stack-checking is enabled, when R9 is used.
6965 @item -mcirrus-fix-invalid-insns
6966 @opindex mcirrus-fix-invalid-insns
6967 @opindex mno-cirrus-fix-invalid-insns
6968 Insert NOPs into the instruction stream to in order to work around
6969 problems with invalid Maverick instruction combinations. This option
6970 is only valid if the @option{-mcpu=ep9312} option has been used to
6971 enable generation of instructions for the Cirrus Maverick floating
6972 point co-processor. This option is not enabled by default, since the
6973 problem is only present in older Maverick implementations. The default
6974 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
6977 @item -mpoke-function-name
6978 @opindex mpoke-function-name
6979 Write the name of each function into the text section, directly
6980 preceding the function prologue. The generated code is similar to this:
6984 .ascii "arm_poke_function_name", 0
6987 .word 0xff000000 + (t1 - t0)
6988 arm_poke_function_name
6990 stmfd sp!, @{fp, ip, lr, pc@}
6994 When performing a stack backtrace, code can inspect the value of
6995 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
6996 location @code{pc - 12} and the top 8 bits are set, then we know that
6997 there is a function name embedded immediately preceding this location
6998 and has length @code{((pc[-3]) & 0xff000000)}.
7002 Generate code for the 16-bit Thumb instruction set. The default is to
7003 use the 32-bit ARM instruction set.
7006 @opindex mtpcs-frame
7007 Generate a stack frame that is compliant with the Thumb Procedure Call
7008 Standard for all non-leaf functions. (A leaf function is one that does
7009 not call any other functions.) The default is @option{-mno-tpcs-frame}.
7011 @item -mtpcs-leaf-frame
7012 @opindex mtpcs-leaf-frame
7013 Generate a stack frame that is compliant with the Thumb Procedure Call
7014 Standard for all leaf functions. (A leaf function is one that does
7015 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
7017 @item -mcallee-super-interworking
7018 @opindex mcallee-super-interworking
7019 Gives all externally visible functions in the file being compiled an ARM
7020 instruction set header which switches to Thumb mode before executing the
7021 rest of the function. This allows these functions to be called from
7022 non-interworking code.
7024 @item -mcaller-super-interworking
7025 @opindex mcaller-super-interworking
7026 Allows calls via function pointers (including virtual functions) to
7027 execute correctly regardless of whether the target code has been
7028 compiled for interworking or not. There is a small overhead in the cost
7029 of executing a function pointer if this option is enabled.
7034 @subsection AVR Options
7037 These options are defined for AVR implementations:
7040 @item -mmcu=@var{mcu}
7042 Specify ATMEL AVR instruction set or MCU type.
7044 Instruction set avr1 is for the minimal AVR core, not supported by the C
7045 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
7046 attiny11, attiny12, attiny15, attiny28).
7048 Instruction set avr2 (default) is for the classic AVR core with up to
7049 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
7050 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
7051 at90c8534, at90s8535).
7053 Instruction set avr3 is for the classic AVR core with up to 128K program
7054 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
7056 Instruction set avr4 is for the enhanced AVR core with up to 8K program
7057 memory space (MCU types: atmega8, atmega83, atmega85).
7059 Instruction set avr5 is for the enhanced AVR core with up to 128K program
7060 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
7061 atmega64, atmega128, at43usb355, at94k).
7065 Output instruction sizes to the asm file.
7067 @item -minit-stack=@var{N}
7068 @opindex minit-stack
7069 Specify the initial stack address, which may be a symbol or numeric value,
7070 @samp{__stack} is the default.
7072 @item -mno-interrupts
7073 @opindex mno-interrupts
7074 Generated code is not compatible with hardware interrupts.
7075 Code size will be smaller.
7077 @item -mcall-prologues
7078 @opindex mcall-prologues
7079 Functions prologues/epilogues expanded as call to appropriate
7080 subroutines. Code size will be smaller.
7082 @item -mno-tablejump
7083 @opindex mno-tablejump
7084 Do not generate tablejump insns which sometimes increase code size.
7087 @opindex mtiny-stack
7088 Change only the low 8 bits of the stack pointer.
7092 Assume int to be 8 bit integer. This affects the sizes of all types: A
7093 char will be 1 byte, an int will be 1 byte, an long will be 2 bytes
7094 and long long will be 4 bytes. Please note that this option does not
7095 comply to the C standards, but it will provide you with smaller code
7100 @subsection CRIS Options
7101 @cindex CRIS Options
7103 These options are defined specifically for the CRIS ports.
7106 @item -march=@var{architecture-type}
7107 @itemx -mcpu=@var{architecture-type}
7110 Generate code for the specified architecture. The choices for
7111 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
7112 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX@.
7113 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
7116 @item -mtune=@var{architecture-type}
7118 Tune to @var{architecture-type} everything applicable about the generated
7119 code, except for the ABI and the set of available instructions. The
7120 choices for @var{architecture-type} are the same as for
7121 @option{-march=@var{architecture-type}}.
7123 @item -mmax-stack-frame=@var{n}
7124 @opindex mmax-stack-frame
7125 Warn when the stack frame of a function exceeds @var{n} bytes.
7127 @item -melinux-stacksize=@var{n}
7128 @opindex melinux-stacksize
7129 Only available with the @samp{cris-axis-aout} target. Arranges for
7130 indications in the program to the kernel loader that the stack of the
7131 program should be set to @var{n} bytes.
7137 The options @option{-metrax4} and @option{-metrax100} are synonyms for
7138 @option{-march=v3} and @option{-march=v8} respectively.
7140 @item -mmul-bug-workaround
7141 @itemx -mno-mul-bug-workaround
7142 @opindex mmul-bug-workaround
7143 @opindex mno-mul-bug-workaround
7144 Work around a bug in the @code{muls} and @code{mulu} instructions for CPU
7145 models where it applies. This option is active by default.
7149 Enable CRIS-specific verbose debug-related information in the assembly
7150 code. This option also has the effect to turn off the @samp{#NO_APP}
7151 formatted-code indicator to the assembler at the beginning of the
7156 Do not use condition-code results from previous instruction; always emit
7157 compare and test instructions before use of condition codes.
7159 @item -mno-side-effects
7160 @opindex mno-side-effects
7161 Do not emit instructions with side-effects in addressing modes other than
7165 @itemx -mno-stack-align
7167 @itemx -mno-data-align
7168 @itemx -mconst-align
7169 @itemx -mno-const-align
7170 @opindex mstack-align
7171 @opindex mno-stack-align
7172 @opindex mdata-align
7173 @opindex mno-data-align
7174 @opindex mconst-align
7175 @opindex mno-const-align
7176 These options (no-options) arranges (eliminate arrangements) for the
7177 stack-frame, individual data and constants to be aligned for the maximum
7178 single data access size for the chosen CPU model. The default is to
7179 arrange for 32-bit alignment. ABI details such as structure layout are
7180 not affected by these options.
7188 Similar to the stack- data- and const-align options above, these options
7189 arrange for stack-frame, writable data and constants to all be 32-bit,
7190 16-bit or 8-bit aligned. The default is 32-bit alignment.
7192 @item -mno-prologue-epilogue
7193 @itemx -mprologue-epilogue
7194 @opindex mno-prologue-epilogue
7195 @opindex mprologue-epilogue
7196 With @option{-mno-prologue-epilogue}, the normal function prologue and
7197 epilogue that sets up the stack-frame are omitted and no return
7198 instructions or return sequences are generated in the code. Use this
7199 option only together with visual inspection of the compiled code: no
7200 warnings or errors are generated when call-saved registers must be saved,
7201 or storage for local variable needs to be allocated.
7207 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
7208 instruction sequences that load addresses for functions from the PLT part
7209 of the GOT rather than (traditional on other architectures) calls to the
7210 PLT@. The default is @option{-mgotplt}.
7214 Legacy no-op option only recognized with the cris-axis-aout target.
7218 Legacy no-op option only recognized with the cris-axis-elf and
7219 cris-axis-linux-gnu targets.
7223 Only recognized with the cris-axis-aout target, where it selects a
7224 GNU/linux-like multilib, include files and instruction set for
7229 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
7233 This option, recognized for the cris-axis-aout and cris-axis-elf arranges
7234 to link with input-output functions from a simulator library. Code,
7235 initialized data and zero-initialized data are allocated consecutively.
7239 Like @option{-sim}, but pass linker options to locate initialized data at
7240 0x40000000 and zero-initialized data at 0x80000000.
7243 @node Darwin Options
7244 @subsection Darwin Options
7245 @cindex Darwin options
7247 These options are defined for all architectures running the Darwin operating
7250 FSF GCC on Darwin does not create ``fat'' object files; it will create
7251 an object file for the single architecture that it was built to
7252 target. Apple's GCC on Darwin does create ``fat'' files if multiple
7253 @option{-arch} options are used; it does so by running the compiler or
7254 linker multiple times and joining the results together with
7257 The subtype of the file created (like @samp{ppc7400} or @samp{ppc970} or
7258 @samp{i686}) is determined by the flags that specify the ISA
7259 that GCC is targetting, like @option{-mcpu} or @option{-march}. The
7260 @option{-force_cpusubtype_ALL} option can be used to override this.
7262 The Darwin tools vary in their behaviour when presented with an ISA
7263 mismatch. The assembler, @file{as}, will only permit instructions to
7264 be used that are valid for the subtype of the file it is generating,
7265 so you cannot put 64-bit instructions in an @samp{ppc750} object file.
7266 The linker for shared libraries, @file{/usr/bin/libtool}, will fail
7267 and print an error if asked to create a shared library with a less
7268 restrictive subtype than its input files (for instance, trying to put
7269 a @samp{ppc970} object file in a @samp{ppc7400} library). The linker
7270 for executables, @file{ld}, will quietly give the executable the most
7271 restrictive subtype of any of its input files.
7276 Add the framework directory @var{dir} to the head of the list of
7277 directories to be searched for header files. These directories are
7278 interleaved with those specified by @option{-I} options and are
7279 scanned in a left-to-right order.
7281 A framework directory is a directory with frameworks in it. A
7282 framework is a directory with a @samp{"Headers"} and/or
7283 @samp{"PrivateHeaders"} directory contained directly in it that ends
7284 in @samp{".framework"}. The name of a framework is the name of this
7285 directory excluding the @samp{".framework"}. Headers associated with
7286 the framework are found in one of those two directories, with
7287 @samp{"Headers"} being searched first. A subframework is a framework
7288 directory that is in a framework's @samp{"Frameworks"} directory.
7289 Includes of subframework headers can only appear in a header of a
7290 framework that contains the subframework, or in a sibling subframework
7291 header. Two subframeworks are siblings if they occur in the same
7292 framework. A subframework should not have the same name as a
7293 framework, a warning will be issued if this is violated. Currently a
7294 subframework cannot have subframeworks, in the future, the mechanism
7295 may be extended to support this. The standard frameworks can be found
7296 in @samp{"/System/Library/Frameworks"} and
7297 @samp{"/Library/Frameworks"}. An example include looks like
7298 @code{#include <Framework/header.h>}, where @samp{Framework} denotes
7299 the name of the framework and header.h is found in the
7300 @samp{"PrivateHeaders"} or @samp{"Headers"} directory.
7304 Emit debugging information for symbols that are used. For STABS
7305 debugging format, this enables @option{-feliminate-unused-debug-symbols}.
7306 This is by default ON@.
7310 Emit debugging information for all symbols and types.
7312 @item -mone-byte-bool
7313 @opindex -mone-byte-bool
7314 Override the defaults for @samp{bool} so that @samp{sizeof(bool)==1}.
7315 By default @samp{sizeof(bool)} is @samp{4} when compiling for
7316 Darwin/PowerPC and @samp{1} when compiling for Darwin/x86, so this
7317 option has no effect on x86.
7319 @strong{Warning:} The @option{-mone-byte-bool} switch causes GCC
7320 to generate code that is not binary compatible with code generated
7321 without that switch. Using this switch may require recompiling all
7322 other modules in a program, including system libraries. Use this
7323 switch to conform to a non-default data model.
7325 @item -mfix-and-continue
7326 @itemx -ffix-and-continue
7327 @itemx -findirect-data
7328 @opindex mfix-and-continue
7329 @opindex ffix-and-continue
7330 @opindex findirect-data
7331 Generate code suitable for fast turn around development. Needed to
7332 enable gdb to dynamically load @code{.o} files into already running
7333 programs. @option{-findirect-data} and @option{-ffix-and-continue}
7334 are provided for backwards compatibility.
7338 Loads all members of static archive libraries.
7339 See man ld(1) for more information.
7341 @item -arch_errors_fatal
7342 @opindex arch_errors_fatal
7343 Cause the errors having to do with files that have the wrong architecture
7347 @opindex bind_at_load
7348 Causes the output file to be marked such that the dynamic linker will
7349 bind all undefined references when the file is loaded or launched.
7353 Produce a Mach-o bundle format file.
7354 See man ld(1) for more information.
7356 @item -bundle_loader @var{executable}
7357 @opindex bundle_loader
7358 This option specifies the @var{executable} that will be loading the build
7359 output file being linked. See man ld(1) for more information.
7362 @opindex -dynamiclib
7363 When passed this option, GCC will produce a dynamic library instead of
7364 an executable when linking, using the Darwin @file{libtool} command.
7366 @item -force_cpusubtype_ALL
7367 @opindex -force_cpusubtype_ALL
7368 This causes GCC's output file to have the @var{ALL} subtype, instead of
7369 one controlled by the @option{-mcpu} or @option{-march} option.
7371 @item -allowable_client @var{client_name}
7373 @itemx -compatibility_version
7374 @itemx -current_version
7376 @itemx -dependency-file
7378 @itemx -dylinker_install_name
7380 @itemx -exported_symbols_list
7382 @itemx -flat_namespace
7383 @itemx -force_flat_namespace
7384 @itemx -headerpad_max_install_names
7387 @itemx -install_name
7388 @itemx -keep_private_externs
7389 @itemx -multi_module
7390 @itemx -multiply_defined
7391 @itemx -multiply_defined_unused
7393 @itemx -no_dead_strip_inits_and_terms
7394 @itemx -nofixprebinding
7397 @itemx -noseglinkedit
7398 @itemx -pagezero_size
7400 @itemx -prebind_all_twolevel_modules
7401 @itemx -private_bundle
7402 @itemx -read_only_relocs
7404 @itemx -sectobjectsymbols
7408 @itemx -sectobjectsymbols
7411 @itemx -segs_read_only_addr
7412 @itemx -segs_read_write_addr
7413 @itemx -seg_addr_table
7414 @itemx -seg_addr_table_filename
7417 @itemx -segs_read_only_addr
7418 @itemx -segs_read_write_addr
7419 @itemx -single_module
7422 @itemx -sub_umbrella
7423 @itemx -twolevel_namespace
7426 @itemx -unexported_symbols_list
7427 @itemx -weak_reference_mismatches
7430 @opindex allowable_client
7431 @opindex client_name
7432 @opindex compatibility_version
7433 @opindex current_version
7435 @opindex dependency-file
7437 @opindex dylinker_install_name
7439 @opindex exported_symbols_list
7441 @opindex flat_namespace
7442 @opindex force_flat_namespace
7443 @opindex headerpad_max_install_names
7446 @opindex install_name
7447 @opindex keep_private_externs
7448 @opindex multi_module
7449 @opindex multiply_defined
7450 @opindex multiply_defined_unused
7452 @opindex no_dead_strip_inits_and_terms
7453 @opindex nofixprebinding
7454 @opindex nomultidefs
7456 @opindex noseglinkedit
7457 @opindex pagezero_size
7459 @opindex prebind_all_twolevel_modules
7460 @opindex private_bundle
7461 @opindex read_only_relocs
7463 @opindex sectobjectsymbols
7467 @opindex sectobjectsymbols
7470 @opindex segs_read_only_addr
7471 @opindex segs_read_write_addr
7472 @opindex seg_addr_table
7473 @opindex seg_addr_table_filename
7474 @opindex seglinkedit
7476 @opindex segs_read_only_addr
7477 @opindex segs_read_write_addr
7478 @opindex single_module
7480 @opindex sub_library
7481 @opindex sub_umbrella
7482 @opindex twolevel_namespace
7485 @opindex unexported_symbols_list
7486 @opindex weak_reference_mismatches
7487 @opindex whatsloaded
7489 These options are passed to the Darwin linker. The Darwin linker man page
7490 describes them in detail.
7493 @node DEC Alpha Options
7494 @subsection DEC Alpha Options
7496 These @samp{-m} options are defined for the DEC Alpha implementations:
7499 @item -mno-soft-float
7501 @opindex mno-soft-float
7502 @opindex msoft-float
7503 Use (do not use) the hardware floating-point instructions for
7504 floating-point operations. When @option{-msoft-float} is specified,
7505 functions in @file{libgcc.a} will be used to perform floating-point
7506 operations. Unless they are replaced by routines that emulate the
7507 floating-point operations, or compiled in such a way as to call such
7508 emulations routines, these routines will issue floating-point
7509 operations. If you are compiling for an Alpha without floating-point
7510 operations, you must ensure that the library is built so as not to call
7513 Note that Alpha implementations without floating-point operations are
7514 required to have floating-point registers.
7519 @opindex mno-fp-regs
7520 Generate code that uses (does not use) the floating-point register set.
7521 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
7522 register set is not used, floating point operands are passed in integer
7523 registers as if they were integers and floating-point results are passed
7524 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
7525 so any function with a floating-point argument or return value called by code
7526 compiled with @option{-mno-fp-regs} must also be compiled with that
7529 A typical use of this option is building a kernel that does not use,
7530 and hence need not save and restore, any floating-point registers.
7534 The Alpha architecture implements floating-point hardware optimized for
7535 maximum performance. It is mostly compliant with the IEEE floating
7536 point standard. However, for full compliance, software assistance is
7537 required. This option generates code fully IEEE compliant code
7538 @emph{except} that the @var{inexact-flag} is not maintained (see below).
7539 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
7540 defined during compilation. The resulting code is less efficient but is
7541 able to correctly support denormalized numbers and exceptional IEEE
7542 values such as not-a-number and plus/minus infinity. Other Alpha
7543 compilers call this option @option{-ieee_with_no_inexact}.
7545 @item -mieee-with-inexact
7546 @opindex mieee-with-inexact
7547 This is like @option{-mieee} except the generated code also maintains
7548 the IEEE @var{inexact-flag}. Turning on this option causes the
7549 generated code to implement fully-compliant IEEE math. In addition to
7550 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
7551 macro. On some Alpha implementations the resulting code may execute
7552 significantly slower than the code generated by default. Since there is
7553 very little code that depends on the @var{inexact-flag}, you should
7554 normally not specify this option. Other Alpha compilers call this
7555 option @option{-ieee_with_inexact}.
7557 @item -mfp-trap-mode=@var{trap-mode}
7558 @opindex mfp-trap-mode
7559 This option controls what floating-point related traps are enabled.
7560 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
7561 The trap mode can be set to one of four values:
7565 This is the default (normal) setting. The only traps that are enabled
7566 are the ones that cannot be disabled in software (e.g., division by zero
7570 In addition to the traps enabled by @samp{n}, underflow traps are enabled
7574 Like @samp{su}, but the instructions are marked to be safe for software
7575 completion (see Alpha architecture manual for details).
7578 Like @samp{su}, but inexact traps are enabled as well.
7581 @item -mfp-rounding-mode=@var{rounding-mode}
7582 @opindex mfp-rounding-mode
7583 Selects the IEEE rounding mode. Other Alpha compilers call this option
7584 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
7589 Normal IEEE rounding mode. Floating point numbers are rounded towards
7590 the nearest machine number or towards the even machine number in case
7594 Round towards minus infinity.
7597 Chopped rounding mode. Floating point numbers are rounded towards zero.
7600 Dynamic rounding mode. A field in the floating point control register
7601 (@var{fpcr}, see Alpha architecture reference manual) controls the
7602 rounding mode in effect. The C library initializes this register for
7603 rounding towards plus infinity. Thus, unless your program modifies the
7604 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
7607 @item -mtrap-precision=@var{trap-precision}
7608 @opindex mtrap-precision
7609 In the Alpha architecture, floating point traps are imprecise. This
7610 means without software assistance it is impossible to recover from a
7611 floating trap and program execution normally needs to be terminated.
7612 GCC can generate code that can assist operating system trap handlers
7613 in determining the exact location that caused a floating point trap.
7614 Depending on the requirements of an application, different levels of
7615 precisions can be selected:
7619 Program precision. This option is the default and means a trap handler
7620 can only identify which program caused a floating point exception.
7623 Function precision. The trap handler can determine the function that
7624 caused a floating point exception.
7627 Instruction precision. The trap handler can determine the exact
7628 instruction that caused a floating point exception.
7631 Other Alpha compilers provide the equivalent options called
7632 @option{-scope_safe} and @option{-resumption_safe}.
7634 @item -mieee-conformant
7635 @opindex mieee-conformant
7636 This option marks the generated code as IEEE conformant. You must not
7637 use this option unless you also specify @option{-mtrap-precision=i} and either
7638 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
7639 is to emit the line @samp{.eflag 48} in the function prologue of the
7640 generated assembly file. Under DEC Unix, this has the effect that
7641 IEEE-conformant math library routines will be linked in.
7643 @item -mbuild-constants
7644 @opindex mbuild-constants
7645 Normally GCC examines a 32- or 64-bit integer constant to
7646 see if it can construct it from smaller constants in two or three
7647 instructions. If it cannot, it will output the constant as a literal and
7648 generate code to load it from the data segment at runtime.
7650 Use this option to require GCC to construct @emph{all} integer constants
7651 using code, even if it takes more instructions (the maximum is six).
7653 You would typically use this option to build a shared library dynamic
7654 loader. Itself a shared library, it must relocate itself in memory
7655 before it can find the variables and constants in its own data segment.
7661 Select whether to generate code to be assembled by the vendor-supplied
7662 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
7680 Indicate whether GCC should generate code to use the optional BWX,
7681 CIX, FIX and MAX instruction sets. The default is to use the instruction
7682 sets supported by the CPU type specified via @option{-mcpu=} option or that
7683 of the CPU on which GCC was built if none was specified.
7688 @opindex mfloat-ieee
7689 Generate code that uses (does not use) VAX F and G floating point
7690 arithmetic instead of IEEE single and double precision.
7692 @item -mexplicit-relocs
7693 @itemx -mno-explicit-relocs
7694 @opindex mexplicit-relocs
7695 @opindex mno-explicit-relocs
7696 Older Alpha assemblers provided no way to generate symbol relocations
7697 except via assembler macros. Use of these macros does not allow
7698 optimal instruction scheduling. GNU binutils as of version 2.12
7699 supports a new syntax that allows the compiler to explicitly mark
7700 which relocations should apply to which instructions. This option
7701 is mostly useful for debugging, as GCC detects the capabilities of
7702 the assembler when it is built and sets the default accordingly.
7706 @opindex msmall-data
7707 @opindex mlarge-data
7708 When @option{-mexplicit-relocs} is in effect, static data is
7709 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
7710 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
7711 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
7712 16-bit relocations off of the @code{$gp} register. This limits the
7713 size of the small data area to 64KB, but allows the variables to be
7714 directly accessed via a single instruction.
7716 The default is @option{-mlarge-data}. With this option the data area
7717 is limited to just below 2GB@. Programs that require more than 2GB of
7718 data must use @code{malloc} or @code{mmap} to allocate the data in the
7719 heap instead of in the program's data segment.
7721 When generating code for shared libraries, @option{-fpic} implies
7722 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
7726 @opindex msmall-text
7727 @opindex mlarge-text
7728 When @option{-msmall-text} is used, the compiler assumes that the
7729 code of the entire program (or shared library) fits in 4MB, and is
7730 thus reachable with a branch instruction. When @option{-msmall-data}
7731 is used, the compiler can assume that all local symbols share the
7732 same @code{$gp} value, and thus reduce the number of instructions
7733 required for a function call from 4 to 1.
7735 The default is @option{-mlarge-text}.
7737 @item -mcpu=@var{cpu_type}
7739 Set the instruction set and instruction scheduling parameters for
7740 machine type @var{cpu_type}. You can specify either the @samp{EV}
7741 style name or the corresponding chip number. GCC supports scheduling
7742 parameters for the EV4, EV5 and EV6 family of processors and will
7743 choose the default values for the instruction set from the processor
7744 you specify. If you do not specify a processor type, GCC will default
7745 to the processor on which the compiler was built.
7747 Supported values for @var{cpu_type} are
7753 Schedules as an EV4 and has no instruction set extensions.
7757 Schedules as an EV5 and has no instruction set extensions.
7761 Schedules as an EV5 and supports the BWX extension.
7766 Schedules as an EV5 and supports the BWX and MAX extensions.
7770 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
7774 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
7777 @item -mtune=@var{cpu_type}
7779 Set only the instruction scheduling parameters for machine type
7780 @var{cpu_type}. The instruction set is not changed.
7782 @item -mmemory-latency=@var{time}
7783 @opindex mmemory-latency
7784 Sets the latency the scheduler should assume for typical memory
7785 references as seen by the application. This number is highly
7786 dependent on the memory access patterns used by the application
7787 and the size of the external cache on the machine.
7789 Valid options for @var{time} are
7793 A decimal number representing clock cycles.
7799 The compiler contains estimates of the number of clock cycles for
7800 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
7801 (also called Dcache, Scache, and Bcache), as well as to main memory.
7802 Note that L3 is only valid for EV5.
7807 @node DEC Alpha/VMS Options
7808 @subsection DEC Alpha/VMS Options
7810 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
7813 @item -mvms-return-codes
7814 @opindex mvms-return-codes
7815 Return VMS condition codes from main. The default is to return POSIX
7816 style condition (e.g.@ error) codes.
7820 @subsection FRV Options
7827 Only use the first 32 general purpose registers.
7832 Use all 64 general purpose registers.
7837 Use only the first 32 floating point registers.
7842 Use all 64 floating point registers
7845 @opindex mhard-float
7847 Use hardware instructions for floating point operations.
7850 @opindex msoft-float
7852 Use library routines for floating point operations.
7857 Dynamically allocate condition code registers.
7862 Do not try to dynamically allocate condition code registers, only
7863 use @code{icc0} and @code{fcc0}.
7868 Change ABI to use double word insns.
7873 Do not use double word instructions.
7878 Use floating point double instructions.
7883 Do not use floating point double instructions.
7888 Use media instructions.
7893 Do not use media instructions.
7898 Use multiply and add/subtract instructions.
7903 Do not use multiply and add/subtract instructions.
7908 Select the FDPIC ABI, that uses function descriptors to represent
7909 pointers to functions. Without any PIC/PIE-related options, it
7910 implies @option{-fPIE}. With @option{-fpic} or @option{-fpie}, it
7911 assumes GOT entries and small data are within a 12-bit range from the
7912 GOT base address; with @option{-fPIC} or @option{-fPIE}, GOT offsets
7913 are computed with 32 bits.
7916 @opindex minline-plt
7918 Enable inlining of PLT entries in function calls to functions that are
7919 not known to bind locally. It has no effect without @option{-mfdpic}.
7920 It's enabled by default if optimizing for speed and compiling for
7921 shared libraries (i.e., @option{-fPIC} or @option{-fpic}), or when an
7922 optimization option such as @option{-O3} or above is present in the
7928 Assume a large TLS segment when generating thread-local code.
7933 Do not assume a large TLS segment when generating thread-local code.
7938 Enable the use of @code{GPREL} relocations in the FDPIC ABI for data
7939 that is known to be in read-only sections. It's enabled by default,
7940 except for @option{-fpic} or @option{-fpie}: even though it may help
7941 make the global offset table smaller, it trades 1 instruction for 4.
7942 With @option{-fPIC} or @option{-fPIE}, it trades 3 instructions for 4,
7943 one of which may be shared by multiple symbols, and it avoids the need
7944 for a GOT entry for the referenced symbol, so it's more likely to be a
7945 win. If it is not, @option{-mno-gprel-ro} can be used to disable it.
7947 @item -multilib-library-pic
7948 @opindex multilib-library-pic
7950 Link with the (library, not FD) pic libraries. It's implied by
7951 @option{-mlibrary-pic}, as well as by @option{-fPIC} and
7952 @option{-fpic} without @option{-mfdpic}. You should never have to use
7958 Follow the EABI requirement of always creating a frame pointer whenever
7959 a stack frame is allocated. This option is enabled by default and can
7960 be disabled with @option{-mno-linked-fp}.
7963 @opindex mlong-calls
7965 Use indirect addressing to call functions outside the current
7966 compilation unit. This allows the functions to be placed anywhere
7967 within the 32-bit address space.
7969 @item -malign-labels
7970 @opindex malign-labels
7972 Try to align labels to an 8-byte boundary by inserting nops into the
7973 previous packet. This option only has an effect when VLIW packing
7974 is enabled. It doesn't create new packets; it merely adds nops to
7978 @opindex mlibrary-pic
7980 Generate position-independent EABI code.
7985 Use only the first four media accumulator registers.
7990 Use all eight media accumulator registers.
7995 Pack VLIW instructions.
8000 Do not pack VLIW instructions.
8005 Do not mark ABI switches in e_flags.
8010 Enable the use of conditional-move instructions (default).
8012 This switch is mainly for debugging the compiler and will likely be removed
8013 in a future version.
8015 @item -mno-cond-move
8016 @opindex mno-cond-move
8018 Disable the use of conditional-move instructions.
8020 This switch is mainly for debugging the compiler and will likely be removed
8021 in a future version.
8026 Enable the use of conditional set instructions (default).
8028 This switch is mainly for debugging the compiler and will likely be removed
8029 in a future version.
8034 Disable the use of conditional set instructions.
8036 This switch is mainly for debugging the compiler and will likely be removed
8037 in a future version.
8042 Enable the use of conditional execution (default).
8044 This switch is mainly for debugging the compiler and will likely be removed
8045 in a future version.
8047 @item -mno-cond-exec
8048 @opindex mno-cond-exec
8050 Disable the use of conditional execution.
8052 This switch is mainly for debugging the compiler and will likely be removed
8053 in a future version.
8056 @opindex mvliw-branch
8058 Run a pass to pack branches into VLIW instructions (default).
8060 This switch is mainly for debugging the compiler and will likely be removed
8061 in a future version.
8063 @item -mno-vliw-branch
8064 @opindex mno-vliw-branch
8066 Do not run a pass to pack branches into VLIW instructions.
8068 This switch is mainly for debugging the compiler and will likely be removed
8069 in a future version.
8071 @item -mmulti-cond-exec
8072 @opindex mmulti-cond-exec
8074 Enable optimization of @code{&&} and @code{||} in conditional execution
8077 This switch is mainly for debugging the compiler and will likely be removed
8078 in a future version.
8080 @item -mno-multi-cond-exec
8081 @opindex mno-multi-cond-exec
8083 Disable optimization of @code{&&} and @code{||} in conditional execution.
8085 This switch is mainly for debugging the compiler and will likely be removed
8086 in a future version.
8088 @item -mnested-cond-exec
8089 @opindex mnested-cond-exec
8091 Enable nested conditional execution optimizations (default).
8093 This switch is mainly for debugging the compiler and will likely be removed
8094 in a future version.
8096 @item -mno-nested-cond-exec
8097 @opindex mno-nested-cond-exec
8099 Disable nested conditional execution optimizations.
8101 This switch is mainly for debugging the compiler and will likely be removed
8102 in a future version.
8104 @item -mtomcat-stats
8105 @opindex mtomcat-stats
8107 Cause gas to print out tomcat statistics.
8109 @item -mcpu=@var{cpu}
8112 Select the processor type for which to generate code. Possible values are
8113 @samp{frv}, @samp{fr550}, @samp{tomcat}, @samp{fr500}, @samp{fr450},
8114 @samp{fr405}, @samp{fr400}, @samp{fr300} and @samp{simple}.
8118 @node H8/300 Options
8119 @subsection H8/300 Options
8121 These @samp{-m} options are defined for the H8/300 implementations:
8126 Shorten some address references at link time, when possible; uses the
8127 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
8128 ld, Using ld}, for a fuller description.
8132 Generate code for the H8/300H@.
8136 Generate code for the H8S@.
8140 Generate code for the H8S and H8/300H in the normal mode. This switch
8141 must be used either with @option{-mh} or @option{-ms}.
8145 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
8149 Make @code{int} data 32 bits by default.
8153 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
8154 The default for the H8/300H and H8S is to align longs and floats on 4
8156 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
8157 This option has no effect on the H8/300.
8161 @subsection HPPA Options
8162 @cindex HPPA Options
8164 These @samp{-m} options are defined for the HPPA family of computers:
8167 @item -march=@var{architecture-type}
8169 Generate code for the specified architecture. The choices for
8170 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
8171 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
8172 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
8173 architecture option for your machine. Code compiled for lower numbered
8174 architectures will run on higher numbered architectures, but not the
8178 @itemx -mpa-risc-1-1
8179 @itemx -mpa-risc-2-0
8180 @opindex mpa-risc-1-0
8181 @opindex mpa-risc-1-1
8182 @opindex mpa-risc-2-0
8183 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
8186 @opindex mbig-switch
8187 Generate code suitable for big switch tables. Use this option only if
8188 the assembler/linker complain about out of range branches within a switch
8191 @item -mjump-in-delay
8192 @opindex mjump-in-delay
8193 Fill delay slots of function calls with unconditional jump instructions
8194 by modifying the return pointer for the function call to be the target
8195 of the conditional jump.
8197 @item -mdisable-fpregs
8198 @opindex mdisable-fpregs
8199 Prevent floating point registers from being used in any manner. This is
8200 necessary for compiling kernels which perform lazy context switching of
8201 floating point registers. If you use this option and attempt to perform
8202 floating point operations, the compiler will abort.
8204 @item -mdisable-indexing
8205 @opindex mdisable-indexing
8206 Prevent the compiler from using indexing address modes. This avoids some
8207 rather obscure problems when compiling MIG generated code under MACH@.
8209 @item -mno-space-regs
8210 @opindex mno-space-regs
8211 Generate code that assumes the target has no space registers. This allows
8212 GCC to generate faster indirect calls and use unscaled index address modes.
8214 Such code is suitable for level 0 PA systems and kernels.
8216 @item -mfast-indirect-calls
8217 @opindex mfast-indirect-calls
8218 Generate code that assumes calls never cross space boundaries. This
8219 allows GCC to emit code which performs faster indirect calls.
8221 This option will not work in the presence of shared libraries or nested
8224 @item -mfixed-range=@var{register-range}
8225 @opindex mfixed-range
8226 Generate code treating the given register range as fixed registers.
8227 A fixed register is one that the register allocator can not use. This is
8228 useful when compiling kernel code. A register range is specified as
8229 two registers separated by a dash. Multiple register ranges can be
8230 specified separated by a comma.
8232 @item -mlong-load-store
8233 @opindex mlong-load-store
8234 Generate 3-instruction load and store sequences as sometimes required by
8235 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
8238 @item -mportable-runtime
8239 @opindex mportable-runtime
8240 Use the portable calling conventions proposed by HP for ELF systems.
8244 Enable the use of assembler directives only GAS understands.
8246 @item -mschedule=@var{cpu-type}
8248 Schedule code according to the constraints for the machine type
8249 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
8250 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
8251 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
8252 proper scheduling option for your machine. The default scheduling is
8256 @opindex mlinker-opt
8257 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
8258 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
8259 linkers in which they give bogus error messages when linking some programs.
8262 @opindex msoft-float
8263 Generate output containing library calls for floating point.
8264 @strong{Warning:} the requisite libraries are not available for all HPPA
8265 targets. Normally the facilities of the machine's usual C compiler are
8266 used, but this cannot be done directly in cross-compilation. You must make
8267 your own arrangements to provide suitable library functions for
8268 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
8269 does provide software floating point support.
8271 @option{-msoft-float} changes the calling convention in the output file;
8272 therefore, it is only useful if you compile @emph{all} of a program with
8273 this option. In particular, you need to compile @file{libgcc.a}, the
8274 library that comes with GCC, with @option{-msoft-float} in order for
8279 Generate the predefine, @code{_SIO}, for server IO@. The default is
8280 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
8281 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO@. These
8282 options are available under HP-UX and HI-UX@.
8286 Use GNU ld specific options. This passes @option{-shared} to ld when
8287 building a shared library. It is the default when GCC is configured,
8288 explicitly or implicitly, with the GNU linker. This option does not
8289 have any affect on which ld is called, it only changes what parameters
8290 are passed to that ld. The ld that is called is determined by the
8291 @option{--with-ld} configure option, GCC's program search path, and
8292 finally by the user's @env{PATH}. The linker used by GCC can be printed
8293 using @samp{which `gcc -print-prog-name=ld`}.
8297 Use HP ld specific options. This passes @option{-b} to ld when building
8298 a shared library and passes @option{+Accept TypeMismatch} to ld on all
8299 links. It is the default when GCC is configured, explicitly or
8300 implicitly, with the HP linker. This option does not have any affect on
8301 which ld is called, it only changes what parameters are passed to that
8302 ld. The ld that is called is determined by the @option{--with-ld}
8303 configure option, GCC's program search path, and finally by the user's
8304 @env{PATH}. The linker used by GCC can be printed using @samp{which
8305 `gcc -print-prog-name=ld`}.
8308 @opindex mno-long-calls
8309 Generate code that uses long call sequences. This ensures that a call
8310 is always able to reach linker generated stubs. The default is to generate
8311 long calls only when the distance from the call site to the beginning
8312 of the function or translation unit, as the case may be, exceeds a
8313 predefined limit set by the branch type being used. The limits for
8314 normal calls are 7,600,000 and 240,000 bytes, respectively for the
8315 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
8318 Distances are measured from the beginning of functions when using the
8319 @option{-ffunction-sections} option, or when using the @option{-mgas}
8320 and @option{-mno-portable-runtime} options together under HP-UX with
8323 It is normally not desirable to use this option as it will degrade
8324 performance. However, it may be useful in large applications,
8325 particularly when partial linking is used to build the application.
8327 The types of long calls used depends on the capabilities of the
8328 assembler and linker, and the type of code being generated. The
8329 impact on systems that support long absolute calls, and long pic
8330 symbol-difference or pc-relative calls should be relatively small.
8331 However, an indirect call is used on 32-bit ELF systems in pic code
8332 and it is quite long.
8334 @item -munix=@var{unix-std}
8336 Generate compiler predefines and select a startfile for the specified
8337 UNIX standard. The choices for @var{unix-std} are @samp{93}, @samp{95}
8338 and @samp{98}. @samp{93} is supported on all HP-UX versions. @samp{95}
8339 is available on HP-UX 10.10 and later. @samp{98} is available on HP-UX
8340 11.11 and later. The default values are @samp{93} for HP-UX 10.00,
8341 @samp{95} for HP-UX 10.10 though to 11.00, and @samp{98} for HP-UX 11.11
8344 @option{-munix=93} provides the same predefines as GCC 3.3 and 3.4.
8345 @option{-munix=95} provides additional predefines for @code{XOPEN_UNIX}
8346 and @code{_XOPEN_SOURCE_EXTENDED}, and the startfile @file{unix95.o}.
8347 @option{-munix=98} provides additional predefines for @code{_XOPEN_UNIX},
8348 @code{_XOPEN_SOURCE_EXTENDED}, @code{_INCLUDE__STDC_A1_SOURCE} and
8349 @code{_INCLUDE_XOPEN_SOURCE_500}, and the startfile @file{unix98.o}.
8351 It is @emph{important} to note that this option changes the interfaces
8352 for various library routines. It also affects the operational behavior
8353 of the C library. Thus, @emph{extreme} care is needed in using this
8356 Library code that is intended to operate with more than one UNIX
8357 standard must test, set and restore the variable @var{__xpg4_extended_mask}
8358 as appropriate. Most GNU software doesn't provide this capability.
8362 Suppress the generation of link options to search libdld.sl when the
8363 @option{-static} option is specified on HP-UX 10 and later.
8367 The HP-UX implementation of setlocale in libc has a dependency on
8368 libdld.sl. There isn't an archive version of libdld.sl. Thus,
8369 when the @option{-static} option is specified, special link options
8370 are needed to resolve this dependency.
8372 On HP-UX 10 and later, the GCC driver adds the necessary options to
8373 link with libdld.sl when the @option{-static} option is specified.
8374 This causes the resulting binary to be dynamic. On the 64-bit port,
8375 the linkers generate dynamic binaries by default in any case. The
8376 @option{-nolibdld} option can be used to prevent the GCC driver from
8377 adding these link options.
8381 Add support for multithreading with the @dfn{dce thread} library
8382 under HP-UX@. This option sets flags for both the preprocessor and
8386 @node i386 and x86-64 Options
8387 @subsection Intel 386 and AMD x86-64 Options
8388 @cindex i386 Options
8389 @cindex x86-64 Options
8390 @cindex Intel 386 Options
8391 @cindex AMD x86-64 Options
8393 These @samp{-m} options are defined for the i386 and x86-64 family of
8397 @item -mtune=@var{cpu-type}
8399 Tune to @var{cpu-type} everything applicable about the generated code, except
8400 for the ABI and the set of available instructions. The choices for
8404 Original Intel's i386 CPU@.
8406 Intel's i486 CPU@. (No scheduling is implemented for this chip.)
8408 Intel Pentium CPU with no MMX support.
8410 Intel PentiumMMX CPU based on Pentium core with MMX instruction set support.
8411 @item i686, pentiumpro
8412 Intel PentiumPro CPU@.
8414 Intel Pentium2 CPU based on PentiumPro core with MMX instruction set support.
8415 @item pentium3, pentium3m
8416 Intel Pentium3 CPU based on PentiumPro core with MMX and SSE instruction set
8419 Low power version of Intel Pentium3 CPU with MMX, SSE and SSE2 instruction set
8420 support. Used by Centrino notebooks.
8421 @item pentium4, pentium4m
8422 Intel Pentium4 CPU with MMX, SSE and SSE2 instruction set support.
8424 Improved version of Intel Pentium4 CPU with MMX, SSE, SSE2 and SSE3 instruction
8427 Improved version of Intel Pentium4 CPU with 64-bit extensions, MMX, SSE,
8428 SSE2 and SSE3 instruction set support.
8430 AMD K6 CPU with MMX instruction set support.
8432 Improved versions of AMD K6 CPU with MMX and 3dNOW! instruction set support.
8433 @item athlon, athlon-tbird
8434 AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and SSE prefetch instructions
8436 @item athlon-4, athlon-xp, athlon-mp
8437 Improved AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW! and full SSE
8438 instruction set support.
8439 @item k8, opteron, athlon64, athlon-fx
8440 AMD K8 core based CPUs with x86-64 instruction set support. (This supersets
8441 MMX, SSE, SSE2, 3dNOW!, enhanced 3dNOW! and 64-bit instruction set extensions.)
8443 IDT Winchip C6 CPU, dealt in same way as i486 with additional MMX instruction
8446 IDT Winchip2 CPU, dealt in same way as i486 with additional MMX and 3dNOW!
8447 instruction set support.
8449 Via C3 CPU with MMX and 3dNOW! instruction set support. (No scheduling is
8450 implemented for this chip.)
8452 Via C3-2 CPU with MMX and SSE instruction set support. (No scheduling is
8453 implemented for this chip.)
8456 While picking a specific @var{cpu-type} will schedule things appropriately
8457 for that particular chip, the compiler will not generate any code that
8458 does not run on the i386 without the @option{-march=@var{cpu-type}} option
8461 @item -march=@var{cpu-type}
8463 Generate instructions for the machine type @var{cpu-type}. The choices
8464 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
8465 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
8467 @item -mcpu=@var{cpu-type}
8469 A deprecated synonym for @option{-mtune}.
8478 @opindex mpentiumpro
8479 These options are synonyms for @option{-mtune=i386}, @option{-mtune=i486},
8480 @option{-mtune=pentium}, and @option{-mtune=pentiumpro} respectively.
8481 These synonyms are deprecated.
8483 @item -mfpmath=@var{unit}
8485 Generate floating point arithmetics for selected unit @var{unit}. The choices
8490 Use the standard 387 floating point coprocessor present majority of chips and
8491 emulated otherwise. Code compiled with this option will run almost everywhere.
8492 The temporary results are computed in 80bit precision instead of precision
8493 specified by the type resulting in slightly different results compared to most
8494 of other chips. See @option{-ffloat-store} for more detailed description.
8496 This is the default choice for i386 compiler.
8499 Use scalar floating point instructions present in the SSE instruction set.
8500 This instruction set is supported by Pentium3 and newer chips, in the AMD line
8501 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
8502 instruction set supports only single precision arithmetics, thus the double and
8503 extended precision arithmetics is still done using 387. Later version, present
8504 only in Pentium4 and the future AMD x86-64 chips supports double precision
8507 For the i386 compiler, you need to use @option{-march=@var{cpu-type}}, @option{-msse}
8508 or @option{-msse2} switches to enable SSE extensions and make this option
8509 effective. For the x86-64 compiler, these extensions are enabled by default.
8511 The resulting code should be considerably faster in the majority of cases and avoid
8512 the numerical instability problems of 387 code, but may break some existing
8513 code that expects temporaries to be 80bit.
8515 This is the default choice for the x86-64 compiler.
8518 Attempt to utilize both instruction sets at once. This effectively double the
8519 amount of available registers and on chips with separate execution units for
8520 387 and SSE the execution resources too. Use this option with care, as it is
8521 still experimental, because the GCC register allocator does not model separate
8522 functional units well resulting in instable performance.
8525 @item -masm=@var{dialect}
8526 @opindex masm=@var{dialect}
8527 Output asm instructions using selected @var{dialect}. Supported choices are
8528 @samp{intel} or @samp{att} (the default one).
8533 @opindex mno-ieee-fp
8534 Control whether or not the compiler uses IEEE floating point
8535 comparisons. These handle correctly the case where the result of a
8536 comparison is unordered.
8539 @opindex msoft-float
8540 Generate output containing library calls for floating point.
8541 @strong{Warning:} the requisite libraries are not part of GCC@.
8542 Normally the facilities of the machine's usual C compiler are used, but
8543 this can't be done directly in cross-compilation. You must make your
8544 own arrangements to provide suitable library functions for
8547 On machines where a function returns floating point results in the 80387
8548 register stack, some floating point opcodes may be emitted even if
8549 @option{-msoft-float} is used.
8551 @item -mno-fp-ret-in-387
8552 @opindex mno-fp-ret-in-387
8553 Do not use the FPU registers for return values of functions.
8555 The usual calling convention has functions return values of types
8556 @code{float} and @code{double} in an FPU register, even if there
8557 is no FPU@. The idea is that the operating system should emulate
8560 The option @option{-mno-fp-ret-in-387} causes such values to be returned
8561 in ordinary CPU registers instead.
8563 @item -mno-fancy-math-387
8564 @opindex mno-fancy-math-387
8565 Some 387 emulators do not support the @code{sin}, @code{cos} and
8566 @code{sqrt} instructions for the 387. Specify this option to avoid
8567 generating those instructions. This option is the default on FreeBSD,
8568 OpenBSD and NetBSD@. This option is overridden when @option{-march}
8569 indicates that the target cpu will always have an FPU and so the
8570 instruction will not need emulation. As of revision 2.6.1, these
8571 instructions are not generated unless you also use the
8572 @option{-funsafe-math-optimizations} switch.
8574 @item -malign-double
8575 @itemx -mno-align-double
8576 @opindex malign-double
8577 @opindex mno-align-double
8578 Control whether GCC aligns @code{double}, @code{long double}, and
8579 @code{long long} variables on a two word boundary or a one word
8580 boundary. Aligning @code{double} variables on a two word boundary will
8581 produce code that runs somewhat faster on a @samp{Pentium} at the
8582 expense of more memory.
8584 @strong{Warning:} if you use the @option{-malign-double} switch,
8585 structures containing the above types will be aligned differently than
8586 the published application binary interface specifications for the 386
8587 and will not be binary compatible with structures in code compiled
8588 without that switch.
8590 @item -m96bit-long-double
8591 @itemx -m128bit-long-double
8592 @opindex m96bit-long-double
8593 @opindex m128bit-long-double
8594 These switches control the size of @code{long double} type. The i386
8595 application binary interface specifies the size to be 96 bits,
8596 so @option{-m96bit-long-double} is the default in 32 bit mode.
8598 Modern architectures (Pentium and newer) would prefer @code{long double}
8599 to be aligned to an 8 or 16 byte boundary. In arrays or structures
8600 conforming to the ABI, this would not be possible. So specifying a
8601 @option{-m128bit-long-double} will align @code{long double}
8602 to a 16 byte boundary by padding the @code{long double} with an additional
8605 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
8606 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
8608 Notice that neither of these options enable any extra precision over the x87
8609 standard of 80 bits for a @code{long double}.
8611 @strong{Warning:} if you override the default value for your target ABI, the
8612 structures and arrays containing @code{long double} variables will change
8613 their size as well as function calling convention for function taking
8614 @code{long double} will be modified. Hence they will not be binary
8615 compatible with arrays or structures in code compiled without that switch.
8619 @itemx -mno-svr3-shlib
8620 @opindex msvr3-shlib
8621 @opindex mno-svr3-shlib
8622 Control whether GCC places uninitialized local variables into the
8623 @code{bss} or @code{data} segments. @option{-msvr3-shlib} places them
8624 into @code{bss}. These options are meaningful only on System V Release 3.
8628 Use a different function-calling convention, in which functions that
8629 take a fixed number of arguments return with the @code{ret} @var{num}
8630 instruction, which pops their arguments while returning. This saves one
8631 instruction in the caller since there is no need to pop the arguments
8634 You can specify that an individual function is called with this calling
8635 sequence with the function attribute @samp{stdcall}. You can also
8636 override the @option{-mrtd} option by using the function attribute
8637 @samp{cdecl}. @xref{Function Attributes}.
8639 @strong{Warning:} this calling convention is incompatible with the one
8640 normally used on Unix, so you cannot use it if you need to call
8641 libraries compiled with the Unix compiler.
8643 Also, you must provide function prototypes for all functions that
8644 take variable numbers of arguments (including @code{printf});
8645 otherwise incorrect code will be generated for calls to those
8648 In addition, seriously incorrect code will result if you call a
8649 function with too many arguments. (Normally, extra arguments are
8650 harmlessly ignored.)
8652 @item -mregparm=@var{num}
8654 Control how many registers are used to pass integer arguments. By
8655 default, no registers are used to pass arguments, and at most 3
8656 registers can be used. You can control this behavior for a specific
8657 function by using the function attribute @samp{regparm}.
8658 @xref{Function Attributes}.
8660 @strong{Warning:} if you use this switch, and
8661 @var{num} is nonzero, then you must build all modules with the same
8662 value, including any libraries. This includes the system libraries and
8665 @item -mpreferred-stack-boundary=@var{num}
8666 @opindex mpreferred-stack-boundary
8667 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
8668 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
8669 the default is 4 (16 bytes or 128 bits), except when optimizing for code
8670 size (@option{-Os}), in which case the default is the minimum correct
8671 alignment (4 bytes for x86, and 8 bytes for x86-64).
8673 On Pentium and PentiumPro, @code{double} and @code{long double} values
8674 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
8675 suffer significant run time performance penalties. On Pentium III, the
8676 Streaming SIMD Extension (SSE) data type @code{__m128} suffers similar
8677 penalties if it is not 16 byte aligned.
8679 To ensure proper alignment of this values on the stack, the stack boundary
8680 must be as aligned as that required by any value stored on the stack.
8681 Further, every function must be generated such that it keeps the stack
8682 aligned. Thus calling a function compiled with a higher preferred
8683 stack boundary from a function compiled with a lower preferred stack
8684 boundary will most likely misalign the stack. It is recommended that
8685 libraries that use callbacks always use the default setting.
8687 This extra alignment does consume extra stack space, and generally
8688 increases code size. Code that is sensitive to stack space usage, such
8689 as embedded systems and operating system kernels, may want to reduce the
8690 preferred alignment to @option{-mpreferred-stack-boundary=2}.
8708 These switches enable or disable the use of built-in functions that allow
8709 direct access to the MMX, SSE, SSE2, SSE3 and 3Dnow extensions of the
8712 @xref{X86 Built-in Functions}, for details of the functions enabled
8713 and disabled by these switches.
8715 To have SSE/SSE2 instructions generated automatically from floating-point
8716 code, see @option{-mfpmath=sse}.
8719 @itemx -mno-push-args
8721 @opindex mno-push-args
8722 Use PUSH operations to store outgoing parameters. This method is shorter
8723 and usually equally fast as method using SUB/MOV operations and is enabled
8724 by default. In some cases disabling it may improve performance because of
8725 improved scheduling and reduced dependencies.
8727 @item -maccumulate-outgoing-args
8728 @opindex maccumulate-outgoing-args
8729 If enabled, the maximum amount of space required for outgoing arguments will be
8730 computed in the function prologue. This is faster on most modern CPUs
8731 because of reduced dependencies, improved scheduling and reduced stack usage
8732 when preferred stack boundary is not equal to 2. The drawback is a notable
8733 increase in code size. This switch implies @option{-mno-push-args}.
8737 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
8738 on thread-safe exception handling must compile and link all code with the
8739 @option{-mthreads} option. When compiling, @option{-mthreads} defines
8740 @option{-D_MT}; when linking, it links in a special thread helper library
8741 @option{-lmingwthrd} which cleans up per thread exception handling data.
8743 @item -mno-align-stringops
8744 @opindex mno-align-stringops
8745 Do not align destination of inlined string operations. This switch reduces
8746 code size and improves performance in case the destination is already aligned,
8747 but GCC doesn't know about it.
8749 @item -minline-all-stringops
8750 @opindex minline-all-stringops
8751 By default GCC inlines string operations only when destination is known to be
8752 aligned at least to 4 byte boundary. This enables more inlining, increase code
8753 size, but may improve performance of code that depends on fast memcpy, strlen
8754 and memset for short lengths.
8756 @item -momit-leaf-frame-pointer
8757 @opindex momit-leaf-frame-pointer
8758 Don't keep the frame pointer in a register for leaf functions. This
8759 avoids the instructions to save, set up and restore frame pointers and
8760 makes an extra register available in leaf functions. The option
8761 @option{-fomit-frame-pointer} removes the frame pointer for all functions
8762 which might make debugging harder.
8764 @item -mtls-direct-seg-refs
8765 @itemx -mno-tls-direct-seg-refs
8766 @opindex mtls-direct-seg-refs
8767 Controls whether TLS variables may be accessed with offsets from the
8768 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
8769 or whether the thread base pointer must be added. Whether or not this
8770 is legal depends on the operating system, and whether it maps the
8771 segment to cover the entire TLS area.
8773 For systems that use GNU libc, the default is on.
8776 These @samp{-m} switches are supported in addition to the above
8777 on AMD x86-64 processors in 64-bit environments.
8784 Generate code for a 32-bit or 64-bit environment.
8785 The 32-bit environment sets int, long and pointer to 32 bits and
8786 generates code that runs on any i386 system.
8787 The 64-bit environment sets int to 32 bits and long and pointer
8788 to 64 bits and generates code for AMD's x86-64 architecture.
8791 @opindex no-red-zone
8792 Do not use a so called red zone for x86-64 code. The red zone is mandated
8793 by the x86-64 ABI, it is a 128-byte area beyond the location of the
8794 stack pointer that will not be modified by signal or interrupt handlers
8795 and therefore can be used for temporary data without adjusting the stack
8796 pointer. The flag @option{-mno-red-zone} disables this red zone.
8798 @item -mcmodel=small
8799 @opindex mcmodel=small
8800 Generate code for the small code model: the program and its symbols must
8801 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
8802 Programs can be statically or dynamically linked. This is the default
8805 @item -mcmodel=kernel
8806 @opindex mcmodel=kernel
8807 Generate code for the kernel code model. The kernel runs in the
8808 negative 2 GB of the address space.
8809 This model has to be used for Linux kernel code.
8811 @item -mcmodel=medium
8812 @opindex mcmodel=medium
8813 Generate code for the medium model: The program is linked in the lower 2
8814 GB of the address space but symbols can be located anywhere in the
8815 address space. Programs can be statically or dynamically linked, but
8816 building of shared libraries are not supported with the medium model.
8818 @item -mcmodel=large
8819 @opindex mcmodel=large
8820 Generate code for the large model: This model makes no assumptions
8821 about addresses and sizes of sections. Currently GCC does not implement
8826 @subsection IA-64 Options
8827 @cindex IA-64 Options
8829 These are the @samp{-m} options defined for the Intel IA-64 architecture.
8833 @opindex mbig-endian
8834 Generate code for a big endian target. This is the default for HP-UX@.
8836 @item -mlittle-endian
8837 @opindex mlittle-endian
8838 Generate code for a little endian target. This is the default for AIX5
8845 Generate (or don't) code for the GNU assembler. This is the default.
8846 @c Also, this is the default if the configure option @option{--with-gnu-as}
8853 Generate (or don't) code for the GNU linker. This is the default.
8854 @c Also, this is the default if the configure option @option{--with-gnu-ld}
8859 Generate code that does not use a global pointer register. The result
8860 is not position independent code, and violates the IA-64 ABI@.
8862 @item -mvolatile-asm-stop
8863 @itemx -mno-volatile-asm-stop
8864 @opindex mvolatile-asm-stop
8865 @opindex mno-volatile-asm-stop
8866 Generate (or don't) a stop bit immediately before and after volatile asm
8869 @item -mregister-names
8870 @itemx -mno-register-names
8871 @opindex mregister-names
8872 @opindex mno-register-names
8873 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
8874 the stacked registers. This may make assembler output more readable.
8880 Disable (or enable) optimizations that use the small data section. This may
8881 be useful for working around optimizer bugs.
8884 @opindex mconstant-gp
8885 Generate code that uses a single constant global pointer value. This is
8886 useful when compiling kernel code.
8890 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
8891 This is useful when compiling firmware code.
8893 @item -minline-float-divide-min-latency
8894 @opindex minline-float-divide-min-latency
8895 Generate code for inline divides of floating point values
8896 using the minimum latency algorithm.
8898 @item -minline-float-divide-max-throughput
8899 @opindex minline-float-divide-max-throughput
8900 Generate code for inline divides of floating point values
8901 using the maximum throughput algorithm.
8903 @item -minline-int-divide-min-latency
8904 @opindex minline-int-divide-min-latency
8905 Generate code for inline divides of integer values
8906 using the minimum latency algorithm.
8908 @item -minline-int-divide-max-throughput
8909 @opindex minline-int-divide-max-throughput
8910 Generate code for inline divides of integer values
8911 using the maximum throughput algorithm.
8913 @item -minline-sqrt-min-latency
8914 @opindex minline-sqrt-min-latency
8915 Generate code for inline square roots
8916 using the minimum latency algorithm.
8918 @item -minline-sqrt-max-throughput
8919 @opindex minline-sqrt-max-throughput
8920 Generate code for inline square roots
8921 using the maximum throughput algorithm.
8923 @item -mno-dwarf2-asm
8925 @opindex mno-dwarf2-asm
8926 @opindex mdwarf2-asm
8927 Don't (or do) generate assembler code for the DWARF2 line number debugging
8928 info. This may be useful when not using the GNU assembler.
8930 @item -mearly-stop-bits
8931 @itemx -mno-early-stop-bits
8932 @opindex mearly-stop-bits
8933 @opindex mno-early-stop-bits
8934 Allow stop bits to be placed earlier than immediately preceding the
8935 instruction that triggered the stop bit. This can improve instruction
8936 scheduling, but does not always do so.
8938 @item -mfixed-range=@var{register-range}
8939 @opindex mfixed-range
8940 Generate code treating the given register range as fixed registers.
8941 A fixed register is one that the register allocator can not use. This is
8942 useful when compiling kernel code. A register range is specified as
8943 two registers separated by a dash. Multiple register ranges can be
8944 specified separated by a comma.
8946 @item -mtls-size=@var{tls-size}
8948 Specify bit size of immediate TLS offsets. Valid values are 14, 22, and
8951 @item -mtune-arch=@var{cpu-type}
8953 Tune the instruction scheduling for a particular CPU, Valid values are
8954 itanium, itanium1, merced, itanium2, and mckinley.
8960 Add support for multithreading using the POSIX threads library. This
8961 option sets flags for both the preprocessor and linker. It does
8962 not affect the thread safety of object code produced by the compiler or
8963 that of libraries supplied with it. These are HP-UX specific flags.
8969 Generate code for a 32-bit or 64-bit environment.
8970 The 32-bit environment sets int, long and pointer to 32 bits.
8971 The 64-bit environment sets int to 32 bits and long and pointer
8972 to 64 bits. These are HP-UX specific flags.
8976 @node M32R/D Options
8977 @subsection M32R/D Options
8978 @cindex M32R/D options
8980 These @option{-m} options are defined for Renesas M32R/D architectures:
8985 Generate code for the M32R/2@.
8989 Generate code for the M32R/X@.
8993 Generate code for the M32R@. This is the default.
8996 @opindex mmodel=small
8997 Assume all objects live in the lower 16MB of memory (so that their addresses
8998 can be loaded with the @code{ld24} instruction), and assume all subroutines
8999 are reachable with the @code{bl} instruction.
9000 This is the default.
9002 The addressability of a particular object can be set with the
9003 @code{model} attribute.
9005 @item -mmodel=medium
9006 @opindex mmodel=medium
9007 Assume objects may be anywhere in the 32-bit address space (the compiler
9008 will generate @code{seth/add3} instructions to load their addresses), and
9009 assume all subroutines are reachable with the @code{bl} instruction.
9012 @opindex mmodel=large
9013 Assume objects may be anywhere in the 32-bit address space (the compiler
9014 will generate @code{seth/add3} instructions to load their addresses), and
9015 assume subroutines may not be reachable with the @code{bl} instruction
9016 (the compiler will generate the much slower @code{seth/add3/jl}
9017 instruction sequence).
9020 @opindex msdata=none
9021 Disable use of the small data area. Variables will be put into
9022 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
9023 @code{section} attribute has been specified).
9024 This is the default.
9026 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
9027 Objects may be explicitly put in the small data area with the
9028 @code{section} attribute using one of these sections.
9031 @opindex msdata=sdata
9032 Put small global and static data in the small data area, but do not
9033 generate special code to reference them.
9037 Put small global and static data in the small data area, and generate
9038 special instructions to reference them.
9042 @cindex smaller data references
9043 Put global and static objects less than or equal to @var{num} bytes
9044 into the small data or bss sections instead of the normal data or bss
9045 sections. The default value of @var{num} is 8.
9046 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
9047 for this option to have any effect.
9049 All modules should be compiled with the same @option{-G @var{num}} value.
9050 Compiling with different values of @var{num} may or may not work; if it
9051 doesn't the linker will give an error message---incorrect code will not be
9056 Makes the M32R specific code in the compiler display some statistics
9057 that might help in debugging programs.
9060 @opindex malign-loops
9061 Align all loops to a 32-byte boundary.
9063 @item -mno-align-loops
9064 @opindex mno-align-loops
9065 Do not enforce a 32-byte alignment for loops. This is the default.
9067 @item -missue-rate=@var{number}
9068 @opindex missue-rate=@var{number}
9069 Issue @var{number} instructions per cycle. @var{number} can only be 1
9072 @item -mbranch-cost=@var{number}
9073 @opindex mbranch-cost=@var{number}
9074 @var{number} can only be 1 or 2. If it is 1 then branches will be
9075 preferred over conditional code, if it is 2, then the opposite will
9078 @item -mflush-trap=@var{number}
9079 @opindex mflush-trap=@var{number}
9080 Specifies the trap number to use to flush the cache. The default is
9081 12. Valid numbers are between 0 and 15 inclusive.
9083 @item -mno-flush-trap
9084 @opindex mno-flush-trap
9085 Specifies that the cache cannot be flushed by using a trap.
9087 @item -mflush-func=@var{name}
9088 @opindex mflush-func=@var{name}
9089 Specifies the name of the operating system function to call to flush
9090 the cache. The default is @emph{_flush_cache}, but a function call
9091 will only be used if a trap is not available.
9093 @item -mno-flush-func
9094 @opindex mno-flush-func
9095 Indicates that there is no OS function for flushing the cache.
9099 @node M680x0 Options
9100 @subsection M680x0 Options
9101 @cindex M680x0 options
9103 These are the @samp{-m} options defined for the 68000 series. The default
9104 values for these options depends on which style of 68000 was selected when
9105 the compiler was configured; the defaults for the most common choices are
9113 Generate output for a 68000. This is the default
9114 when the compiler is configured for 68000-based systems.
9116 Use this option for microcontrollers with a 68000 or EC000 core,
9117 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
9123 Generate output for a 68020. This is the default
9124 when the compiler is configured for 68020-based systems.
9128 Generate output containing 68881 instructions for floating point.
9129 This is the default for most 68020 systems unless @option{--nfp} was
9130 specified when the compiler was configured.
9134 Generate output for a 68030. This is the default when the compiler is
9135 configured for 68030-based systems.
9139 Generate output for a 68040. This is the default when the compiler is
9140 configured for 68040-based systems.
9142 This option inhibits the use of 68881/68882 instructions that have to be
9143 emulated by software on the 68040. Use this option if your 68040 does not
9144 have code to emulate those instructions.
9148 Generate output for a 68060. This is the default when the compiler is
9149 configured for 68060-based systems.
9151 This option inhibits the use of 68020 and 68881/68882 instructions that
9152 have to be emulated by software on the 68060. Use this option if your 68060
9153 does not have code to emulate those instructions.
9157 Generate output for a CPU32. This is the default
9158 when the compiler is configured for CPU32-based systems.
9160 Use this option for microcontrollers with a
9161 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
9162 68336, 68340, 68341, 68349 and 68360.
9166 Generate output for a 520X ``coldfire'' family cpu. This is the default
9167 when the compiler is configured for 520X-based systems.
9169 Use this option for microcontroller with a 5200 core, including
9170 the MCF5202, MCF5203, MCF5204 and MCF5202.
9175 Generate output for a 68040, without using any of the new instructions.
9176 This results in code which can run relatively efficiently on either a
9177 68020/68881 or a 68030 or a 68040. The generated code does use the
9178 68881 instructions that are emulated on the 68040.
9182 Generate output for a 68060, without using any of the new instructions.
9183 This results in code which can run relatively efficiently on either a
9184 68020/68881 or a 68030 or a 68040. The generated code does use the
9185 68881 instructions that are emulated on the 68060.
9188 @opindex msoft-float
9189 Generate output containing library calls for floating point.
9190 @strong{Warning:} the requisite libraries are not available for all m68k
9191 targets. Normally the facilities of the machine's usual C compiler are
9192 used, but this can't be done directly in cross-compilation. You must
9193 make your own arrangements to provide suitable library functions for
9194 cross-compilation. The embedded targets @samp{m68k-*-aout} and
9195 @samp{m68k-*-coff} do provide software floating point support.
9199 Consider type @code{int} to be 16 bits wide, like @code{short int}.
9200 Additionally, parameters passed on the stack are also aligned to a
9201 16-bit boundary even on targets whose API mandates promotion to 32-bit.
9204 @opindex mnobitfield
9205 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
9206 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
9210 Do use the bit-field instructions. The @option{-m68020} option implies
9211 @option{-mbitfield}. This is the default if you use a configuration
9212 designed for a 68020.
9216 Use a different function-calling convention, in which functions
9217 that take a fixed number of arguments return with the @code{rtd}
9218 instruction, which pops their arguments while returning. This
9219 saves one instruction in the caller since there is no need to pop
9220 the arguments there.
9222 This calling convention is incompatible with the one normally
9223 used on Unix, so you cannot use it if you need to call libraries
9224 compiled with the Unix compiler.
9226 Also, you must provide function prototypes for all functions that
9227 take variable numbers of arguments (including @code{printf});
9228 otherwise incorrect code will be generated for calls to those
9231 In addition, seriously incorrect code will result if you call a
9232 function with too many arguments. (Normally, extra arguments are
9233 harmlessly ignored.)
9235 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
9236 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
9239 @itemx -mno-align-int
9241 @opindex mno-align-int
9242 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
9243 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
9244 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
9245 Aligning variables on 32-bit boundaries produces code that runs somewhat
9246 faster on processors with 32-bit busses at the expense of more memory.
9248 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
9249 align structures containing the above types differently than
9250 most published application binary interface specifications for the m68k.
9254 Use the pc-relative addressing mode of the 68000 directly, instead of
9255 using a global offset table. At present, this option implies @option{-fpic},
9256 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
9257 not presently supported with @option{-mpcrel}, though this could be supported for
9258 68020 and higher processors.
9260 @item -mno-strict-align
9261 @itemx -mstrict-align
9262 @opindex mno-strict-align
9263 @opindex mstrict-align
9264 Do not (do) assume that unaligned memory references will be handled by
9268 Generate code that allows the data segment to be located in a different
9269 area of memory from the text segment. This allows for execute in place in
9270 an environment without virtual memory management. This option implies
9274 Generate code that assumes that the data segment follows the text segment.
9275 This is the default.
9277 @item -mid-shared-library
9278 Generate code that supports shared libraries via the library ID method.
9279 This allows for execute in place and shared libraries in an environment
9280 without virtual memory management. This option implies @option{-fPIC}.
9282 @item -mno-id-shared-library
9283 Generate code that doesn't assume ID based shared libraries are being used.
9284 This is the default.
9286 @item -mshared-library-id=n
9287 Specified the identification number of the ID based shared library being
9288 compiled. Specifying a value of 0 will generate more compact code, specifying
9289 other values will force the allocation of that number to the current
9290 library but is no more space or time efficient than omitting this option.
9294 @node M68hc1x Options
9295 @subsection M68hc1x Options
9296 @cindex M68hc1x options
9298 These are the @samp{-m} options defined for the 68hc11 and 68hc12
9299 microcontrollers. The default values for these options depends on
9300 which style of microcontroller was selected when the compiler was configured;
9301 the defaults for the most common choices are given below.
9308 Generate output for a 68HC11. This is the default
9309 when the compiler is configured for 68HC11-based systems.
9315 Generate output for a 68HC12. This is the default
9316 when the compiler is configured for 68HC12-based systems.
9322 Generate output for a 68HCS12.
9325 @opindex mauto-incdec
9326 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
9333 Enable the use of 68HC12 min and max instructions.
9336 @itemx -mno-long-calls
9337 @opindex mlong-calls
9338 @opindex mno-long-calls
9339 Treat all calls as being far away (near). If calls are assumed to be
9340 far away, the compiler will use the @code{call} instruction to
9341 call a function and the @code{rtc} instruction for returning.
9345 Consider type @code{int} to be 16 bits wide, like @code{short int}.
9347 @item -msoft-reg-count=@var{count}
9348 @opindex msoft-reg-count
9349 Specify the number of pseudo-soft registers which are used for the
9350 code generation. The maximum number is 32. Using more pseudo-soft
9351 register may or may not result in better code depending on the program.
9352 The default is 4 for 68HC11 and 2 for 68HC12.
9357 @subsection MCore Options
9358 @cindex MCore options
9360 These are the @samp{-m} options defined for the Motorola M*Core
9368 @opindex mno-hardlit
9369 Inline constants into the code stream if it can be done in two
9370 instructions or less.
9376 Use the divide instruction. (Enabled by default).
9378 @item -mrelax-immediate
9379 @itemx -mno-relax-immediate
9380 @opindex mrelax-immediate
9381 @opindex mno-relax-immediate
9382 Allow arbitrary sized immediates in bit operations.
9384 @item -mwide-bitfields
9385 @itemx -mno-wide-bitfields
9386 @opindex mwide-bitfields
9387 @opindex mno-wide-bitfields
9388 Always treat bit-fields as int-sized.
9390 @item -m4byte-functions
9391 @itemx -mno-4byte-functions
9392 @opindex m4byte-functions
9393 @opindex mno-4byte-functions
9394 Force all functions to be aligned to a four byte boundary.
9396 @item -mcallgraph-data
9397 @itemx -mno-callgraph-data
9398 @opindex mcallgraph-data
9399 @opindex mno-callgraph-data
9400 Emit callgraph information.
9403 @itemx -mno-slow-bytes
9404 @opindex mslow-bytes
9405 @opindex mno-slow-bytes
9406 Prefer word access when reading byte quantities.
9408 @item -mlittle-endian
9410 @opindex mlittle-endian
9411 @opindex mbig-endian
9412 Generate code for a little endian target.
9418 Generate code for the 210 processor.
9422 @subsection MIPS Options
9423 @cindex MIPS options
9429 Generate big-endian code.
9433 Generate little-endian code. This is the default for @samp{mips*el-*-*}
9436 @item -march=@var{arch}
9438 Generate code that will run on @var{arch}, which can be the name of a
9439 generic MIPS ISA, or the name of a particular processor.
9441 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
9442 @samp{mips32}, @samp{mips32r2}, and @samp{mips64}.
9443 The processor names are:
9444 @samp{4kc}, @samp{4kp}, @samp{5kc}, @samp{20kc},
9446 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
9447 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000}, @samp{rm7000},
9451 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4130}, @samp{vr4300},
9452 @samp{vr5000}, @samp{vr5400} and @samp{vr5500}.
9453 The special value @samp{from-abi} selects the
9454 most compatible architecture for the selected ABI (that is,
9455 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
9457 In processor names, a final @samp{000} can be abbreviated as @samp{k}
9458 (for example, @samp{-march=r2k}). Prefixes are optional, and
9459 @samp{vr} may be written @samp{r}.
9461 GCC defines two macros based on the value of this option. The first
9462 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
9463 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
9464 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
9465 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
9466 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
9468 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
9469 above. In other words, it will have the full prefix and will not
9470 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
9471 the macro names the resolved architecture (either @samp{"mips1"} or
9472 @samp{"mips3"}). It names the default architecture when no
9473 @option{-march} option is given.
9475 @item -mtune=@var{arch}
9477 Optimize for @var{arch}. Among other things, this option controls
9478 the way instructions are scheduled, and the perceived cost of arithmetic
9479 operations. The list of @var{arch} values is the same as for
9482 When this option is not used, GCC will optimize for the processor
9483 specified by @option{-march}. By using @option{-march} and
9484 @option{-mtune} together, it is possible to generate code that will
9485 run on a family of processors, but optimize the code for one
9486 particular member of that family.
9488 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
9489 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
9490 @samp{-march} ones described above.
9494 Equivalent to @samp{-march=mips1}.
9498 Equivalent to @samp{-march=mips2}.
9502 Equivalent to @samp{-march=mips3}.
9506 Equivalent to @samp{-march=mips4}.
9510 Equivalent to @samp{-march=mips32}.
9514 Equivalent to @samp{-march=mips32r2}.
9518 Equivalent to @samp{-march=mips64}.
9524 Use (do not use) the MIPS16 ISA@.
9536 Generate code for the given ABI@.
9538 Note that the EABI has a 32-bit and a 64-bit variant. GCC normally
9539 generates 64-bit code when you select a 64-bit architecture, but you
9540 can use @option{-mgp32} to get 32-bit code instead.
9542 For information about the O64 ABI, see
9543 @w{@uref{http://gcc.gnu.org/projects/mipso64-abi.html}}.
9546 @itemx -mno-abicalls
9548 @opindex mno-abicalls
9549 Generate (do not generate) SVR4-style position-independent code.
9550 @option{-mabicalls} is the default for SVR4-based systems.
9556 Lift (do not lift) the usual restrictions on the size of the global
9559 GCC normally uses a single instruction to load values from the GOT@.
9560 While this is relatively efficient, it will only work if the GOT
9561 is smaller than about 64k. Anything larger will cause the linker
9562 to report an error such as:
9564 @cindex relocation truncated to fit (MIPS)
9566 relocation truncated to fit: R_MIPS_GOT16 foobar
9569 If this happens, you should recompile your code with @option{-mxgot}.
9570 It should then work with very large GOTs, although it will also be
9571 less efficient, since it will take three instructions to fetch the
9572 value of a global symbol.
9574 Note that some linkers can create multiple GOTs. If you have such a
9575 linker, you should only need to use @option{-mxgot} when a single object
9576 file accesses more than 64k's worth of GOT entries. Very few do.
9578 These options have no effect unless GCC is generating position
9583 Assume that general-purpose registers are 32 bits wide.
9587 Assume that general-purpose registers are 64 bits wide.
9591 Assume that floating-point registers are 32 bits wide.
9595 Assume that floating-point registers are 64 bits wide.
9598 @opindex mhard-float
9599 Use floating-point coprocessor instructions.
9602 @opindex msoft-float
9603 Do not use floating-point coprocessor instructions. Implement
9604 floating-point calculations using library calls instead.
9606 @item -msingle-float
9607 @opindex msingle-float
9608 Assume that the floating-point coprocessor only supports single-precision
9611 @itemx -mdouble-float
9612 @opindex mdouble-float
9613 Assume that the floating-point coprocessor supports double-precision
9614 operations. This is the default.
9616 @itemx -mpaired-single
9617 @itemx -mno-paired-single
9618 @opindex mpaired-single
9619 @opindex mno-paired-single
9620 Use (do not use) paired-single floating-point instructions.
9621 @xref{MIPS Paired-Single Support}. This option can only be used
9622 when generating 64-bit code and requires hardware floating-point
9623 support to be enabled.
9629 Use (do not use) the MIPS-3D ASE@. @xref{MIPS-3D Built-in Functions}.
9630 The option @option{-mips3d} implies @option{-mpaired-single}.
9634 Force @code{int} and @code{long} types to be 64 bits wide. See
9635 @option{-mlong32} for an explanation of the default and the way
9636 that the pointer size is determined.
9638 This option has been deprecated and will be removed in a future release.
9642 Force @code{long} types to be 64 bits wide. See @option{-mlong32} for
9643 an explanation of the default and the way that the pointer size is
9648 Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
9650 The default size of @code{int}s, @code{long}s and pointers depends on
9651 the ABI@. All the supported ABIs use 32-bit @code{int}s. The n64 ABI
9652 uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
9653 32-bit @code{long}s. Pointers are the same size as @code{long}s,
9654 or the same size as integer registers, whichever is smaller.
9658 @cindex smaller data references (MIPS)
9659 @cindex gp-relative references (MIPS)
9660 Put global and static items less than or equal to @var{num} bytes into
9661 the small data or bss section instead of the normal data or bss section.
9662 This allows the data to be accessed using a single instruction.
9664 All modules should be compiled with the same @option{-G @var{num}}
9667 @item -membedded-data
9668 @itemx -mno-embedded-data
9669 @opindex membedded-data
9670 @opindex mno-embedded-data
9671 Allocate variables to the read-only data section first if possible, then
9672 next in the small data section if possible, otherwise in data. This gives
9673 slightly slower code than the default, but reduces the amount of RAM required
9674 when executing, and thus may be preferred for some embedded systems.
9676 @item -muninit-const-in-rodata
9677 @itemx -mno-uninit-const-in-rodata
9678 @opindex muninit-const-in-rodata
9679 @opindex mno-uninit-const-in-rodata
9680 Put uninitialized @code{const} variables in the read-only data section.
9681 This option is only meaningful in conjunction with @option{-membedded-data}.
9683 @item -msplit-addresses
9684 @itemx -mno-split-addresses
9685 @opindex msplit-addresses
9686 @opindex mno-split-addresses
9687 Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
9688 relocation operators. This option has been superceded by
9689 @option{-mexplicit-relocs} but is retained for backwards compatibility.
9691 @item -mexplicit-relocs
9692 @itemx -mno-explicit-relocs
9693 @opindex mexplicit-relocs
9694 @opindex mno-explicit-relocs
9695 Use (do not use) assembler relocation operators when dealing with symbolic
9696 addresses. The alternative, selected by @option{-mno-explicit-relocs},
9697 is to use assembler macros instead.
9699 @option{-mexplicit-relocs} is the default if GCC was configured
9700 to use an assembler that supports relocation operators.
9702 @item -mcheck-zero-division
9703 @itemx -mno-check-zero-division
9704 @opindex mcheck-zero-division
9705 @opindex mno-check-zero-division
9706 Trap (do not trap) on integer division by zero. The default is
9707 @option{-mcheck-zero-division}.
9709 @item -mdivide-traps
9710 @itemx -mdivide-breaks
9711 @opindex mdivide-traps
9712 @opindex mdivide-breaks
9713 MIPS systems check for division by zero by generating either a
9714 conditional trap or a break instruction. Using traps results in
9715 smaller code, but is only supported on MIPS II and later. Also, some
9716 versions of the Linux kernel have a bug that prevents trap from
9717 generating the proper signal (@code{SIGFPE}). Use @option{-mdivide-traps} to
9718 allow conditional traps on architectures that support them and
9719 @option{-mdivide-breaks} to force the use of breaks.
9721 The default is usually @option{-mdivide-traps}, but this can be
9722 overridden at configure time using @option{--with-divide=breaks}.
9723 Divide-by-zero checks can be completely disabled using
9724 @option{-mno-check-zero-division}.
9730 Force (do not force) the use of @code{memcpy()} for non-trivial block
9731 moves. The default is @option{-mno-memcpy}, which allows GCC to inline
9732 most constant-sized copies.
9735 @itemx -mno-long-calls
9736 @opindex mlong-calls
9737 @opindex mno-long-calls
9738 Disable (do not disable) use of the @code{jal} instruction. Calling
9739 functions using @code{jal} is more efficient but requires the caller
9740 and callee to be in the same 256 megabyte segment.
9742 This option has no effect on abicalls code. The default is
9743 @option{-mno-long-calls}.
9749 Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
9750 instructions, as provided by the R4650 ISA@.
9753 @itemx -mno-fused-madd
9754 @opindex mfused-madd
9755 @opindex mno-fused-madd
9756 Enable (disable) use of the floating point multiply-accumulate
9757 instructions, when they are available. The default is
9758 @option{-mfused-madd}.
9760 When multiply-accumulate instructions are used, the intermediate
9761 product is calculated to infinite precision and is not subject to
9762 the FCSR Flush to Zero bit. This may be undesirable in some
9767 Tell the MIPS assembler to not run its preprocessor over user
9768 assembler files (with a @samp{.s} suffix) when assembling them.
9771 @itemx -mno-fix-r4000
9773 @opindex mno-fix-r4000
9774 Work around certain R4000 CPU errata:
9777 A double-word or a variable shift may give an incorrect result if executed
9778 immediately after starting an integer division.
9780 A double-word or a variable shift may give an incorrect result if executed
9781 while an integer multiplication is in progress.
9783 An integer division may give an incorrect result if started in a delay slot
9784 of a taken branch or a jump.
9788 @itemx -mno-fix-r4400
9790 @opindex mno-fix-r4400
9791 Work around certain R4400 CPU errata:
9794 A double-word or a variable shift may give an incorrect result if executed
9795 immediately after starting an integer division.
9799 @itemx -mno-fix-vr4120
9800 @opindex mfix-vr4120
9801 Work around certain VR4120 errata:
9804 @code{dmultu} does not always produce the correct result.
9806 @code{div} and @code{ddiv} do not always produce the correct result if one
9807 of the operands is negative.
9809 The workarounds for the division errata rely on special functions in
9810 @file{libgcc.a}. At present, these functions are only provided by
9811 the @code{mips64vr*-elf} configurations.
9813 Other VR4120 errata require a nop to be inserted between certain pairs of
9814 instructions. These errata are handled by the assembler, not by GCC itself.
9819 Work around certain SB-1 CPU core errata.
9820 (This flag currently works around the SB-1 revision 2
9821 ``F1'' and ``F2'' floating point errata.)
9823 @item -mflush-func=@var{func}
9824 @itemx -mno-flush-func
9825 @opindex mflush-func
9826 Specifies the function to call to flush the I and D caches, or to not
9827 call any such function. If called, the function must take the same
9828 arguments as the common @code{_flush_func()}, that is, the address of the
9829 memory range for which the cache is being flushed, the size of the
9830 memory range, and the number 3 (to flush both caches). The default
9831 depends on the target GCC was configured for, but commonly is either
9832 @samp{_flush_func} or @samp{__cpu_flush}.
9834 @item -mbranch-likely
9835 @itemx -mno-branch-likely
9836 @opindex mbranch-likely
9837 @opindex mno-branch-likely
9838 Enable or disable use of Branch Likely instructions, regardless of the
9839 default for the selected architecture. By default, Branch Likely
9840 instructions may be generated if they are supported by the selected
9841 architecture. An exception is for the MIPS32 and MIPS64 architectures
9842 and processors which implement those architectures; for those, Branch
9843 Likely instructions will not be generated by default because the MIPS32
9844 and MIPS64 architectures specifically deprecate their use.
9846 @item -mfp-exceptions
9847 @itemx -mno-fp-exceptions
9848 @opindex mfp-exceptions
9849 Specifies whether FP exceptions are enabled. This affects how we schedule
9850 FP instructions for some processors. The default is that FP exceptions are
9853 For instance, on the SB-1, if FP exceptions are disabled, and we are emitting
9854 64-bit code, then we can use both FP pipes. Otherwise, we can only use one
9857 @item -mvr4130-align
9858 @itemx -mno-vr4130-align
9859 @opindex mvr4130-align
9860 The VR4130 pipeline is two-way superscalar, but can only issue two
9861 instructions together if the first one is 8-byte aligned. When this
9862 option is enabled, GCC will align pairs of instructions that it
9863 thinks should execute in parallel.
9865 This option only has an effect when optimizing for the VR4130.
9866 It normally makes code faster, but at the expense of making it bigger.
9867 It is enabled by default at optimization level @option{-O3}.
9871 @subsection MMIX Options
9872 @cindex MMIX Options
9874 These options are defined for the MMIX:
9878 @itemx -mno-libfuncs
9880 @opindex mno-libfuncs
9881 Specify that intrinsic library functions are being compiled, passing all
9882 values in registers, no matter the size.
9887 @opindex mno-epsilon
9888 Generate floating-point comparison instructions that compare with respect
9889 to the @code{rE} epsilon register.
9891 @item -mabi=mmixware
9893 @opindex mabi-mmixware
9895 Generate code that passes function parameters and return values that (in
9896 the called function) are seen as registers @code{$0} and up, as opposed to
9897 the GNU ABI which uses global registers @code{$231} and up.
9900 @itemx -mno-zero-extend
9901 @opindex mzero-extend
9902 @opindex mno-zero-extend
9903 When reading data from memory in sizes shorter than 64 bits, use (do not
9904 use) zero-extending load instructions by default, rather than
9905 sign-extending ones.
9908 @itemx -mno-knuthdiv
9910 @opindex mno-knuthdiv
9911 Make the result of a division yielding a remainder have the same sign as
9912 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
9913 remainder follows the sign of the dividend. Both methods are
9914 arithmetically valid, the latter being almost exclusively used.
9916 @item -mtoplevel-symbols
9917 @itemx -mno-toplevel-symbols
9918 @opindex mtoplevel-symbols
9919 @opindex mno-toplevel-symbols
9920 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
9921 code can be used with the @code{PREFIX} assembly directive.
9925 Generate an executable in the ELF format, rather than the default
9926 @samp{mmo} format used by the @command{mmix} simulator.
9928 @item -mbranch-predict
9929 @itemx -mno-branch-predict
9930 @opindex mbranch-predict
9931 @opindex mno-branch-predict
9932 Use (do not use) the probable-branch instructions, when static branch
9933 prediction indicates a probable branch.
9935 @item -mbase-addresses
9936 @itemx -mno-base-addresses
9937 @opindex mbase-addresses
9938 @opindex mno-base-addresses
9939 Generate (do not generate) code that uses @emph{base addresses}. Using a
9940 base address automatically generates a request (handled by the assembler
9941 and the linker) for a constant to be set up in a global register. The
9942 register is used for one or more base address requests within the range 0
9943 to 255 from the value held in the register. The generally leads to short
9944 and fast code, but the number of different data items that can be
9945 addressed is limited. This means that a program that uses lots of static
9946 data may require @option{-mno-base-addresses}.
9949 @itemx -mno-single-exit
9950 @opindex msingle-exit
9951 @opindex mno-single-exit
9952 Force (do not force) generated code to have a single exit point in each
9956 @node MN10300 Options
9957 @subsection MN10300 Options
9958 @cindex MN10300 options
9960 These @option{-m} options are defined for Matsushita MN10300 architectures:
9965 Generate code to avoid bugs in the multiply instructions for the MN10300
9966 processors. This is the default.
9969 @opindex mno-mult-bug
9970 Do not generate code to avoid bugs in the multiply instructions for the
9975 Generate code which uses features specific to the AM33 processor.
9979 Do not generate code which uses features specific to the AM33 processor. This
9984 Do not link in the C run-time initialization object file.
9988 Indicate to the linker that it should perform a relaxation optimization pass
9989 to shorten branches, calls and absolute memory addresses. This option only
9990 has an effect when used on the command line for the final link step.
9992 This option makes symbolic debugging impossible.
9996 @subsection NS32K Options
9997 @cindex NS32K options
9999 These are the @samp{-m} options defined for the 32000 series. The default
10000 values for these options depends on which style of 32000 was selected when
10001 the compiler was configured; the defaults for the most common choices are
10009 Generate output for a 32032. This is the default
10010 when the compiler is configured for 32032 and 32016 based systems.
10016 Generate output for a 32332. This is the default
10017 when the compiler is configured for 32332-based systems.
10023 Generate output for a 32532. This is the default
10024 when the compiler is configured for 32532-based systems.
10028 Generate output containing 32081 instructions for floating point.
10029 This is the default for all systems.
10033 Generate output containing 32381 instructions for floating point. This
10034 also implies @option{-m32081}. The 32381 is only compatible with the 32332
10035 and 32532 cpus. This is the default for the pc532-netbsd configuration.
10038 @opindex mmulti-add
10039 Try and generate multiply-add floating point instructions @code{polyF}
10040 and @code{dotF}. This option is only available if the @option{-m32381}
10041 option is in effect. Using these instructions requires changes to
10042 register allocation which generally has a negative impact on
10043 performance. This option should only be enabled when compiling code
10044 particularly likely to make heavy use of multiply-add instructions.
10046 @item -mnomulti-add
10047 @opindex mnomulti-add
10048 Do not try and generate multiply-add floating point instructions
10049 @code{polyF} and @code{dotF}. This is the default on all platforms.
10052 @opindex msoft-float
10053 Generate output containing library calls for floating point.
10054 @strong{Warning:} the requisite libraries may not be available.
10056 @item -mieee-compare
10057 @itemx -mno-ieee-compare
10058 @opindex mieee-compare
10059 @opindex mno-ieee-compare
10060 Control whether or not the compiler uses IEEE floating point
10061 comparisons. These handle correctly the case where the result of a
10062 comparison is unordered.
10063 @strong{Warning:} the requisite kernel support may not be available.
10066 @opindex mnobitfield
10067 Do not use the bit-field instructions. On some machines it is faster to
10068 use shifting and masking operations. This is the default for the pc532.
10072 Do use the bit-field instructions. This is the default for all platforms
10077 Use a different function-calling convention, in which functions
10078 that take a fixed number of arguments return pop their
10079 arguments on return with the @code{ret} instruction.
10081 This calling convention is incompatible with the one normally
10082 used on Unix, so you cannot use it if you need to call libraries
10083 compiled with the Unix compiler.
10085 Also, you must provide function prototypes for all functions that
10086 take variable numbers of arguments (including @code{printf});
10087 otherwise incorrect code will be generated for calls to those
10090 In addition, seriously incorrect code will result if you call a
10091 function with too many arguments. (Normally, extra arguments are
10092 harmlessly ignored.)
10094 This option takes its name from the 680x0 @code{rtd} instruction.
10099 Use a different function-calling convention where the first two arguments
10100 are passed in registers.
10102 This calling convention is incompatible with the one normally
10103 used on Unix, so you cannot use it if you need to call libraries
10104 compiled with the Unix compiler.
10107 @opindex mnoregparam
10108 Do not pass any arguments in registers. This is the default for all
10113 It is OK to use the sb as an index register which is always loaded with
10114 zero. This is the default for the pc532-netbsd target.
10118 The sb register is not available for use or has not been initialized to
10119 zero by the run time system. This is the default for all targets except
10120 the pc532-netbsd. It is also implied whenever @option{-mhimem} or
10121 @option{-fpic} is set.
10125 Many ns32000 series addressing modes use displacements of up to 512MB@.
10126 If an address is above 512MB then displacements from zero can not be used.
10127 This option causes code to be generated which can be loaded above 512MB@.
10128 This may be useful for operating systems or ROM code.
10132 Assume code will be loaded in the first 512MB of virtual address space.
10133 This is the default for all platforms.
10137 @node PDP-11 Options
10138 @subsection PDP-11 Options
10139 @cindex PDP-11 Options
10141 These options are defined for the PDP-11:
10146 Use hardware FPP floating point. This is the default. (FIS floating
10147 point on the PDP-11/40 is not supported.)
10150 @opindex msoft-float
10151 Do not use hardware floating point.
10155 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
10159 Return floating-point results in memory. This is the default.
10163 Generate code for a PDP-11/40.
10167 Generate code for a PDP-11/45. This is the default.
10171 Generate code for a PDP-11/10.
10173 @item -mbcopy-builtin
10174 @opindex bcopy-builtin
10175 Use inline @code{movmemhi} patterns for copying memory. This is the
10180 Do not use inline @code{movmemhi} patterns for copying memory.
10186 Use 16-bit @code{int}. This is the default.
10192 Use 32-bit @code{int}.
10195 @itemx -mno-float32
10197 @opindex mno-float32
10198 Use 64-bit @code{float}. This is the default.
10201 @itemx -mno-float64
10203 @opindex mno-float64
10204 Use 32-bit @code{float}.
10208 Use @code{abshi2} pattern. This is the default.
10212 Do not use @code{abshi2} pattern.
10214 @item -mbranch-expensive
10215 @opindex mbranch-expensive
10216 Pretend that branches are expensive. This is for experimenting with
10217 code generation only.
10219 @item -mbranch-cheap
10220 @opindex mbranch-cheap
10221 Do not pretend that branches are expensive. This is the default.
10225 Generate code for a system with split I&D@.
10229 Generate code for a system without split I&D@. This is the default.
10233 Use Unix assembler syntax. This is the default when configured for
10234 @samp{pdp11-*-bsd}.
10238 Use DEC assembler syntax. This is the default when configured for any
10239 PDP-11 target other than @samp{pdp11-*-bsd}.
10242 @node PowerPC Options
10243 @subsection PowerPC Options
10244 @cindex PowerPC options
10246 These are listed under @xref{RS/6000 and PowerPC Options}.
10248 @node RS/6000 and PowerPC Options
10249 @subsection IBM RS/6000 and PowerPC Options
10250 @cindex RS/6000 and PowerPC Options
10251 @cindex IBM RS/6000 and PowerPC Options
10253 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
10260 @itemx -mno-powerpc
10261 @itemx -mpowerpc-gpopt
10262 @itemx -mno-powerpc-gpopt
10263 @itemx -mpowerpc-gfxopt
10264 @itemx -mno-powerpc-gfxopt
10266 @itemx -mno-powerpc64
10270 @opindex mno-power2
10272 @opindex mno-powerpc
10273 @opindex mpowerpc-gpopt
10274 @opindex mno-powerpc-gpopt
10275 @opindex mpowerpc-gfxopt
10276 @opindex mno-powerpc-gfxopt
10277 @opindex mpowerpc64
10278 @opindex mno-powerpc64
10279 GCC supports two related instruction set architectures for the
10280 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
10281 instructions supported by the @samp{rios} chip set used in the original
10282 RS/6000 systems and the @dfn{PowerPC} instruction set is the
10283 architecture of the Motorola MPC5xx, MPC6xx, MPC8xx microprocessors, and
10284 the IBM 4xx microprocessors.
10286 Neither architecture is a subset of the other. However there is a
10287 large common subset of instructions supported by both. An MQ
10288 register is included in processors supporting the POWER architecture.
10290 You use these options to specify which instructions are available on the
10291 processor you are using. The default value of these options is
10292 determined when configuring GCC@. Specifying the
10293 @option{-mcpu=@var{cpu_type}} overrides the specification of these
10294 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
10295 rather than the options listed above.
10297 The @option{-mpower} option allows GCC to generate instructions that
10298 are found only in the POWER architecture and to use the MQ register.
10299 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
10300 to generate instructions that are present in the POWER2 architecture but
10301 not the original POWER architecture.
10303 The @option{-mpowerpc} option allows GCC to generate instructions that
10304 are found only in the 32-bit subset of the PowerPC architecture.
10305 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
10306 GCC to use the optional PowerPC architecture instructions in the
10307 General Purpose group, including floating-point square root. Specifying
10308 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
10309 use the optional PowerPC architecture instructions in the Graphics
10310 group, including floating-point select.
10312 The @option{-mpowerpc64} option allows GCC to generate the additional
10313 64-bit instructions that are found in the full PowerPC64 architecture
10314 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
10315 @option{-mno-powerpc64}.
10317 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
10318 will use only the instructions in the common subset of both
10319 architectures plus some special AIX common-mode calls, and will not use
10320 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
10321 permits GCC to use any instruction from either architecture and to
10322 allow use of the MQ register; specify this for the Motorola MPC601.
10324 @item -mnew-mnemonics
10325 @itemx -mold-mnemonics
10326 @opindex mnew-mnemonics
10327 @opindex mold-mnemonics
10328 Select which mnemonics to use in the generated assembler code. With
10329 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
10330 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
10331 assembler mnemonics defined for the POWER architecture. Instructions
10332 defined in only one architecture have only one mnemonic; GCC uses that
10333 mnemonic irrespective of which of these options is specified.
10335 GCC defaults to the mnemonics appropriate for the architecture in
10336 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
10337 value of these option. Unless you are building a cross-compiler, you
10338 should normally not specify either @option{-mnew-mnemonics} or
10339 @option{-mold-mnemonics}, but should instead accept the default.
10341 @item -mcpu=@var{cpu_type}
10343 Set architecture type, register usage, choice of mnemonics, and
10344 instruction scheduling parameters for machine type @var{cpu_type}.
10345 Supported values for @var{cpu_type} are @samp{401}, @samp{403},
10346 @samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{505},
10347 @samp{601}, @samp{602}, @samp{603}, @samp{603e}, @samp{604},
10348 @samp{604e}, @samp{620}, @samp{630}, @samp{740}, @samp{7400},
10349 @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
10350 @samp{860}, @samp{970}, @samp{common}, @samp{ec603e}, @samp{G3},
10351 @samp{G4}, @samp{G5}, @samp{power}, @samp{power2}, @samp{power3},
10352 @samp{power4}, @samp{power5}, @samp{powerpc}, @samp{powerpc64},
10353 @samp{rios}, @samp{rios1}, @samp{rios2}, @samp{rsc}, and @samp{rs64a}.
10355 @option{-mcpu=common} selects a completely generic processor. Code
10356 generated under this option will run on any POWER or PowerPC processor.
10357 GCC will use only the instructions in the common subset of both
10358 architectures, and will not use the MQ register. GCC assumes a generic
10359 processor model for scheduling purposes.
10361 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
10362 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
10363 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
10364 types, with an appropriate, generic processor model assumed for
10365 scheduling purposes.
10367 The other options specify a specific processor. Code generated under
10368 those options will run best on that processor, and may not run at all on
10371 The @option{-mcpu} options automatically enable or disable the
10372 following options: @option{-maltivec}, @option{-mhard-float},
10373 @option{-mmfcrf}, @option{-mmultiple}, @option{-mnew-mnemonics},
10374 @option{-mpower}, @option{-mpower2}, @option{-mpowerpc64},
10375 @option{-mpowerpc-gpopt}, @option{-mpowerpc-gfxopt},
10376 @option{-mstring}. The particular options set for any particular CPU
10377 will vary between compiler versions, depending on what setting seems
10378 to produce optimal code for that CPU; it doesn't necessarily reflect
10379 the actual hardware's capabilities. If you wish to set an individual
10380 option to a particular value, you may specify it after the
10381 @option{-mcpu} option, like @samp{-mcpu=970 -mno-altivec}.
10383 On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
10384 not enabled or disabled by the @option{-mcpu} option at present, since
10385 AIX does not have full support for these options. You may still
10386 enable or disable them individually if you're sure it'll work in your
10389 @item -mtune=@var{cpu_type}
10391 Set the instruction scheduling parameters for machine type
10392 @var{cpu_type}, but do not set the architecture type, register usage, or
10393 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
10394 values for @var{cpu_type} are used for @option{-mtune} as for
10395 @option{-mcpu}. If both are specified, the code generated will use the
10396 architecture, registers, and mnemonics set by @option{-mcpu}, but the
10397 scheduling parameters set by @option{-mtune}.
10400 @itemx -mno-altivec
10402 @opindex mno-altivec
10403 Generate code that uses (does not use) AltiVec instructions, and also
10404 enable the use of built-in functions that allow more direct access to
10405 the AltiVec instruction set. You may also need to set
10406 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
10411 Extend the current ABI with SPE ABI extensions. This does not change
10412 the default ABI, instead it adds the SPE ABI extensions to the current
10416 @opindex mabi=no-spe
10417 Disable Booke SPE ABI extensions for the current ABI@.
10419 @item -misel=@var{yes/no}
10422 This switch enables or disables the generation of ISEL instructions.
10424 @item -mspe=@var{yes/no}
10427 This switch enables or disables the generation of SPE simd
10430 @item -mfloat-gprs=@var{yes/single/double/no}
10431 @itemx -mfloat-gprs
10432 @opindex mfloat-gprs
10433 This switch enables or disables the generation of floating point
10434 operations on the general purpose registers for architectures that
10437 The argument @var{yes} or @var{single} enables the use of
10438 single-precision floating point operations.
10440 The argument @var{double} enables the use of single and
10441 double-precision floating point operations.
10443 The argument @var{no} disables floating point operations on the
10444 general purpose registers.
10446 This option is currently only available on the MPC854x.
10452 Generate code for 32-bit or 64-bit environments of Darwin and SVR4
10453 targets (including GNU/Linux). The 32-bit environment sets int, long
10454 and pointer to 32 bits and generates code that runs on any PowerPC
10455 variant. The 64-bit environment sets int to 32 bits and long and
10456 pointer to 64 bits, and generates code for PowerPC64, as for
10457 @option{-mpowerpc64}.
10460 @itemx -mno-fp-in-toc
10461 @itemx -mno-sum-in-toc
10462 @itemx -mminimal-toc
10464 @opindex mno-fp-in-toc
10465 @opindex mno-sum-in-toc
10466 @opindex mminimal-toc
10467 Modify generation of the TOC (Table Of Contents), which is created for
10468 every executable file. The @option{-mfull-toc} option is selected by
10469 default. In that case, GCC will allocate at least one TOC entry for
10470 each unique non-automatic variable reference in your program. GCC
10471 will also place floating-point constants in the TOC@. However, only
10472 16,384 entries are available in the TOC@.
10474 If you receive a linker error message that saying you have overflowed
10475 the available TOC space, you can reduce the amount of TOC space used
10476 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
10477 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
10478 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
10479 generate code to calculate the sum of an address and a constant at
10480 run-time instead of putting that sum into the TOC@. You may specify one
10481 or both of these options. Each causes GCC to produce very slightly
10482 slower and larger code at the expense of conserving TOC space.
10484 If you still run out of space in the TOC even when you specify both of
10485 these options, specify @option{-mminimal-toc} instead. This option causes
10486 GCC to make only one TOC entry for every file. When you specify this
10487 option, GCC will produce code that is slower and larger but which
10488 uses extremely little TOC space. You may wish to use this option
10489 only on files that contain less frequently executed code.
10495 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
10496 @code{long} type, and the infrastructure needed to support them.
10497 Specifying @option{-maix64} implies @option{-mpowerpc64} and
10498 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
10499 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
10502 @itemx -mno-xl-compat
10503 @opindex mxl-compat
10504 @opindex mno-xl-compat
10505 Produce code that conforms more closely to IBM XLC semantics when using
10506 AIX-compatible ABI. Pass floating-point arguments to prototyped
10507 functions beyond the register save area (RSA) on the stack in addition
10508 to argument FPRs. Do not assume that most significant double in 128
10509 bit long double value is properly rounded when comparing values.
10511 The AIX calling convention was extended but not initially documented to
10512 handle an obscure K&R C case of calling a function that takes the
10513 address of its arguments with fewer arguments than declared. AIX XL
10514 compilers access floating point arguments which do not fit in the
10515 RSA from the stack when a subroutine is compiled without
10516 optimization. Because always storing floating-point arguments on the
10517 stack is inefficient and rarely needed, this option is not enabled by
10518 default and only is necessary when calling subroutines compiled by AIX
10519 XL compilers without optimization.
10523 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
10524 application written to use message passing with special startup code to
10525 enable the application to run. The system must have PE installed in the
10526 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
10527 must be overridden with the @option{-specs=} option to specify the
10528 appropriate directory location. The Parallel Environment does not
10529 support threads, so the @option{-mpe} option and the @option{-pthread}
10530 option are incompatible.
10532 @item -malign-natural
10533 @itemx -malign-power
10534 @opindex malign-natural
10535 @opindex malign-power
10536 On AIX, Darwin, and 64-bit PowerPC GNU/Linux, the option
10537 @option{-malign-natural} overrides the ABI-defined alignment of larger
10538 types, such as floating-point doubles, on their natural size-based boundary.
10539 The option @option{-malign-power} instructs GCC to follow the ABI-specified
10540 alignment rules. GCC defaults to the standard alignment defined in the ABI@.
10543 @itemx -mhard-float
10544 @opindex msoft-float
10545 @opindex mhard-float
10546 Generate code that does not use (uses) the floating-point register set.
10547 Software floating point emulation is provided if you use the
10548 @option{-msoft-float} option, and pass the option to GCC when linking.
10551 @itemx -mno-multiple
10553 @opindex mno-multiple
10554 Generate code that uses (does not use) the load multiple word
10555 instructions and the store multiple word instructions. These
10556 instructions are generated by default on POWER systems, and not
10557 generated on PowerPC systems. Do not use @option{-mmultiple} on little
10558 endian PowerPC systems, since those instructions do not work when the
10559 processor is in little endian mode. The exceptions are PPC740 and
10560 PPC750 which permit the instructions usage in little endian mode.
10565 @opindex mno-string
10566 Generate code that uses (does not use) the load string instructions
10567 and the store string word instructions to save multiple registers and
10568 do small block moves. These instructions are generated by default on
10569 POWER systems, and not generated on PowerPC systems. Do not use
10570 @option{-mstring} on little endian PowerPC systems, since those
10571 instructions do not work when the processor is in little endian mode.
10572 The exceptions are PPC740 and PPC750 which permit the instructions
10573 usage in little endian mode.
10578 @opindex mno-update
10579 Generate code that uses (does not use) the load or store instructions
10580 that update the base register to the address of the calculated memory
10581 location. These instructions are generated by default. If you use
10582 @option{-mno-update}, there is a small window between the time that the
10583 stack pointer is updated and the address of the previous frame is
10584 stored, which means code that walks the stack frame across interrupts or
10585 signals may get corrupted data.
10588 @itemx -mno-fused-madd
10589 @opindex mfused-madd
10590 @opindex mno-fused-madd
10591 Generate code that uses (does not use) the floating point multiply and
10592 accumulate instructions. These instructions are generated by default if
10593 hardware floating is used.
10595 @item -mno-bit-align
10597 @opindex mno-bit-align
10598 @opindex mbit-align
10599 On System V.4 and embedded PowerPC systems do not (do) force structures
10600 and unions that contain bit-fields to be aligned to the base type of the
10603 For example, by default a structure containing nothing but 8
10604 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
10605 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
10606 the structure would be aligned to a 1 byte boundary and be one byte in
10609 @item -mno-strict-align
10610 @itemx -mstrict-align
10611 @opindex mno-strict-align
10612 @opindex mstrict-align
10613 On System V.4 and embedded PowerPC systems do not (do) assume that
10614 unaligned memory references will be handled by the system.
10616 @item -mrelocatable
10617 @itemx -mno-relocatable
10618 @opindex mrelocatable
10619 @opindex mno-relocatable
10620 On embedded PowerPC systems generate code that allows (does not allow)
10621 the program to be relocated to a different address at runtime. If you
10622 use @option{-mrelocatable} on any module, all objects linked together must
10623 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
10625 @item -mrelocatable-lib
10626 @itemx -mno-relocatable-lib
10627 @opindex mrelocatable-lib
10628 @opindex mno-relocatable-lib
10629 On embedded PowerPC systems generate code that allows (does not allow)
10630 the program to be relocated to a different address at runtime. Modules
10631 compiled with @option{-mrelocatable-lib} can be linked with either modules
10632 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
10633 with modules compiled with the @option{-mrelocatable} options.
10639 On System V.4 and embedded PowerPC systems do not (do) assume that
10640 register 2 contains a pointer to a global area pointing to the addresses
10641 used in the program.
10644 @itemx -mlittle-endian
10646 @opindex mlittle-endian
10647 On System V.4 and embedded PowerPC systems compile code for the
10648 processor in little endian mode. The @option{-mlittle-endian} option is
10649 the same as @option{-mlittle}.
10652 @itemx -mbig-endian
10654 @opindex mbig-endian
10655 On System V.4 and embedded PowerPC systems compile code for the
10656 processor in big endian mode. The @option{-mbig-endian} option is
10657 the same as @option{-mbig}.
10659 @item -mdynamic-no-pic
10660 @opindex mdynamic-no-pic
10661 On Darwin and Mac OS X systems, compile code so that it is not
10662 relocatable, but that its external references are relocatable. The
10663 resulting code is suitable for applications, but not shared
10666 @item -mprioritize-restricted-insns=@var{priority}
10667 @opindex mprioritize-restricted-insns
10668 This option controls the priority that is assigned to
10669 dispatch-slot restricted instructions during the second scheduling
10670 pass. The argument @var{priority} takes the value @var{0/1/2} to assign
10671 @var{no/highest/second-highest} priority to dispatch slot restricted
10674 @item -msched-costly-dep=@var{dependence_type}
10675 @opindex msched-costly-dep
10676 This option controls which dependences are considered costly
10677 by the target during instruction scheduling. The argument
10678 @var{dependence_type} takes one of the following values:
10679 @var{no}: no dependence is costly,
10680 @var{all}: all dependences are costly,
10681 @var{true_store_to_load}: a true dependence from store to load is costly,
10682 @var{store_to_load}: any dependence from store to load is costly,
10683 @var{number}: any dependence which latency >= @var{number} is costly.
10685 @item -minsert-sched-nops=@var{scheme}
10686 @opindex minsert-sched-nops
10687 This option controls which nop insertion scheme will be used during
10688 the second scheduling pass. The argument @var{scheme} takes one of the
10690 @var{no}: Don't insert nops.
10691 @var{pad}: Pad with nops any dispatch group which has vacant issue slots,
10692 according to the scheduler's grouping.
10693 @var{regroup_exact}: Insert nops to force costly dependent insns into
10694 separate groups. Insert exactly as many nops as needed to force an insn
10695 to a new group, according to the estimated processor grouping.
10696 @var{number}: Insert nops to force costly dependent insns into
10697 separate groups. Insert @var{number} nops to force an insn to a new group.
10700 @opindex mcall-sysv
10701 On System V.4 and embedded PowerPC systems compile code using calling
10702 conventions that adheres to the March 1995 draft of the System V
10703 Application Binary Interface, PowerPC processor supplement. This is the
10704 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
10706 @item -mcall-sysv-eabi
10707 @opindex mcall-sysv-eabi
10708 Specify both @option{-mcall-sysv} and @option{-meabi} options.
10710 @item -mcall-sysv-noeabi
10711 @opindex mcall-sysv-noeabi
10712 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
10714 @item -mcall-solaris
10715 @opindex mcall-solaris
10716 On System V.4 and embedded PowerPC systems compile code for the Solaris
10720 @opindex mcall-linux
10721 On System V.4 and embedded PowerPC systems compile code for the
10722 Linux-based GNU system.
10726 On System V.4 and embedded PowerPC systems compile code for the
10727 Hurd-based GNU system.
10729 @item -mcall-netbsd
10730 @opindex mcall-netbsd
10731 On System V.4 and embedded PowerPC systems compile code for the
10732 NetBSD operating system.
10734 @item -maix-struct-return
10735 @opindex maix-struct-return
10736 Return all structures in memory (as specified by the AIX ABI)@.
10738 @item -msvr4-struct-return
10739 @opindex msvr4-struct-return
10740 Return structures smaller than 8 bytes in registers (as specified by the
10743 @item -mabi=altivec
10744 @opindex mabi=altivec
10745 Extend the current ABI with AltiVec ABI extensions. This does not
10746 change the default ABI, instead it adds the AltiVec ABI extensions to
10749 @item -mabi=no-altivec
10750 @opindex mabi=no-altivec
10751 Disable AltiVec ABI extensions for the current ABI@.
10754 @itemx -mno-prototype
10755 @opindex mprototype
10756 @opindex mno-prototype
10757 On System V.4 and embedded PowerPC systems assume that all calls to
10758 variable argument functions are properly prototyped. Otherwise, the
10759 compiler must insert an instruction before every non prototyped call to
10760 set or clear bit 6 of the condition code register (@var{CR}) to
10761 indicate whether floating point values were passed in the floating point
10762 registers in case the function takes a variable arguments. With
10763 @option{-mprototype}, only calls to prototyped variable argument functions
10764 will set or clear the bit.
10768 On embedded PowerPC systems, assume that the startup module is called
10769 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
10770 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
10775 On embedded PowerPC systems, assume that the startup module is called
10776 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
10781 On embedded PowerPC systems, assume that the startup module is called
10782 @file{crt0.o} and the standard C libraries are @file{libads.a} and
10785 @item -myellowknife
10786 @opindex myellowknife
10787 On embedded PowerPC systems, assume that the startup module is called
10788 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
10793 On System V.4 and embedded PowerPC systems, specify that you are
10794 compiling for a VxWorks system.
10798 Specify that you are compiling for the WindISS simulation environment.
10802 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
10803 header to indicate that @samp{eabi} extended relocations are used.
10809 On System V.4 and embedded PowerPC systems do (do not) adhere to the
10810 Embedded Applications Binary Interface (eabi) which is a set of
10811 modifications to the System V.4 specifications. Selecting @option{-meabi}
10812 means that the stack is aligned to an 8 byte boundary, a function
10813 @code{__eabi} is called to from @code{main} to set up the eabi
10814 environment, and the @option{-msdata} option can use both @code{r2} and
10815 @code{r13} to point to two separate small data areas. Selecting
10816 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
10817 do not call an initialization function from @code{main}, and the
10818 @option{-msdata} option will only use @code{r13} to point to a single
10819 small data area. The @option{-meabi} option is on by default if you
10820 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
10823 @opindex msdata=eabi
10824 On System V.4 and embedded PowerPC systems, put small initialized
10825 @code{const} global and static data in the @samp{.sdata2} section, which
10826 is pointed to by register @code{r2}. Put small initialized
10827 non-@code{const} global and static data in the @samp{.sdata} section,
10828 which is pointed to by register @code{r13}. Put small uninitialized
10829 global and static data in the @samp{.sbss} section, which is adjacent to
10830 the @samp{.sdata} section. The @option{-msdata=eabi} option is
10831 incompatible with the @option{-mrelocatable} option. The
10832 @option{-msdata=eabi} option also sets the @option{-memb} option.
10835 @opindex msdata=sysv
10836 On System V.4 and embedded PowerPC systems, put small global and static
10837 data in the @samp{.sdata} section, which is pointed to by register
10838 @code{r13}. Put small uninitialized global and static data in the
10839 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
10840 The @option{-msdata=sysv} option is incompatible with the
10841 @option{-mrelocatable} option.
10843 @item -msdata=default
10845 @opindex msdata=default
10847 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
10848 compile code the same as @option{-msdata=eabi}, otherwise compile code the
10849 same as @option{-msdata=sysv}.
10852 @opindex msdata-data
10853 On System V.4 and embedded PowerPC systems, put small global and static
10854 data in the @samp{.sdata} section. Put small uninitialized global and
10855 static data in the @samp{.sbss} section. Do not use register @code{r13}
10856 to address small data however. This is the default behavior unless
10857 other @option{-msdata} options are used.
10861 @opindex msdata=none
10863 On embedded PowerPC systems, put all initialized global and static data
10864 in the @samp{.data} section, and all uninitialized data in the
10865 @samp{.bss} section.
10869 @cindex smaller data references (PowerPC)
10870 @cindex .sdata/.sdata2 references (PowerPC)
10871 On embedded PowerPC systems, put global and static items less than or
10872 equal to @var{num} bytes into the small data or bss sections instead of
10873 the normal data or bss section. By default, @var{num} is 8. The
10874 @option{-G @var{num}} switch is also passed to the linker.
10875 All modules should be compiled with the same @option{-G @var{num}} value.
10878 @itemx -mno-regnames
10880 @opindex mno-regnames
10881 On System V.4 and embedded PowerPC systems do (do not) emit register
10882 names in the assembly language output using symbolic forms.
10885 @itemx -mno-longcall
10887 @opindex mno-longcall
10888 Default to making all function calls indirectly, using a register, so
10889 that functions which reside further than 32 megabytes (33,554,432
10890 bytes) from the current location can be called. This setting can be
10891 overridden by the @code{shortcall} function attribute, or by
10892 @code{#pragma longcall(0)}.
10894 Some linkers are capable of detecting out-of-range calls and generating
10895 glue code on the fly. On these systems, long calls are unnecessary and
10896 generate slower code. As of this writing, the AIX linker can do this,
10897 as can the GNU linker for PowerPC/64. It is planned to add this feature
10898 to the GNU linker for 32-bit PowerPC systems as well.
10900 On Darwin/PPC systems, @code{#pragma longcall} will generate ``jbsr
10901 callee, L42'', plus a ``branch island'' (glue code). The two target
10902 addresses represent the callee and the ``branch island''. The
10903 Darwin/PPC linker will prefer the first address and generate a ``bl
10904 callee'' if the PPC ``bl'' instruction will reach the callee directly;
10905 otherwise, the linker will generate ``bl L42'' to call the ``branch
10906 island''. The ``branch island'' is appended to the body of the
10907 calling function; it computes the full 32-bit address of the callee
10910 On Mach-O (Darwin) systems, this option directs the compiler emit to
10911 the glue for every direct call, and the Darwin linker decides whether
10912 to use or discard it.
10914 In the future, we may cause GCC to ignore all longcall specifications
10915 when the linker is known to generate glue.
10919 Adds support for multithreading with the @dfn{pthreads} library.
10920 This option sets flags for both the preprocessor and linker.
10924 @node S/390 and zSeries Options
10925 @subsection S/390 and zSeries Options
10926 @cindex S/390 and zSeries Options
10928 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
10932 @itemx -msoft-float
10933 @opindex mhard-float
10934 @opindex msoft-float
10935 Use (do not use) the hardware floating-point instructions and registers
10936 for floating-point operations. When @option{-msoft-float} is specified,
10937 functions in @file{libgcc.a} will be used to perform floating-point
10938 operations. When @option{-mhard-float} is specified, the compiler
10939 generates IEEE floating-point instructions. This is the default.
10942 @itemx -mno-backchain
10943 @opindex mbackchain
10944 @opindex mno-backchain
10945 Store (do not store) the address of the caller's frame as backchain pointer
10946 into the callee's stack frame.
10947 A backchain may be needed to allow debugging using tools that do not understand
10948 DWARF-2 call frame information.
10949 When @option{-mno-packed-stack} is in effect, the backchain pointer is stored
10950 at the bottom of the stack frame; when @option{-mpacked-stack} is in effect,
10951 the backchain is placed into the topmost word of the 96/160 byte register
10954 In general, code compiled with @option{-mbackchain} is call-compatible with
10955 code compiled with @option{-mmo-backchain}; however, use of the backchain
10956 for debugging purposes usually requires that the whole binary is built with
10957 @option{-mbackchain}. Note that the combination of @option{-mbackchain},
10958 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
10959 to build a linux kernel use @option{-msoft-float}.
10961 The default is to not maintain the backchain.
10963 @item -mpacked-stack
10964 @item -mno-packed-stack
10965 @opindex mpacked-stack
10966 @opindex mno-packed-stack
10967 Use (do not use) the packed stack layout. When @option{-mno-packed-stack} is
10968 specified, the compiler uses the all fields of the 96/160 byte register save
10969 area only for their default purpose; unused fields still take up stack space.
10970 When @option{-mpacked-stack} is specified, register save slots are densely
10971 packed at the top of the register save area; unused space is reused for other
10972 purposes, allowing for more efficient use of the available stack space.
10973 However, when @option{-mbackchain} is also in effect, the topmost word of
10974 the save area is always used to store the backchain, and the return address
10975 register is always saved two words below the backchain.
10977 As long as the stack frame backchain is not used, code generated with
10978 @option{-mpacked-stack} is call-compatible with code generated with
10979 @option{-mno-packed-stack}. Note that some non-FSF releases of GCC 2.95 for
10980 S/390 or zSeries generated code that uses the stack frame backchain at run
10981 time, not just for debugging purposes. Such code is not call-compatible
10982 with code compiled with @option{-mpacked-stack}. Also, note that the
10983 combination of @option{-mbackchain},
10984 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
10985 to build a linux kernel use @option{-msoft-float}.
10987 The default is to not use the packed stack layout.
10990 @itemx -mno-small-exec
10991 @opindex msmall-exec
10992 @opindex mno-small-exec
10993 Generate (or do not generate) code using the @code{bras} instruction
10994 to do subroutine calls.
10995 This only works reliably if the total executable size does not
10996 exceed 64k. The default is to use the @code{basr} instruction instead,
10997 which does not have this limitation.
11003 When @option{-m31} is specified, generate code compliant to the
11004 GNU/Linux for S/390 ABI@. When @option{-m64} is specified, generate
11005 code compliant to the GNU/Linux for zSeries ABI@. This allows GCC in
11006 particular to generate 64-bit instructions. For the @samp{s390}
11007 targets, the default is @option{-m31}, while the @samp{s390x}
11008 targets default to @option{-m64}.
11014 When @option{-mzarch} is specified, generate code using the
11015 instructions available on z/Architecture.
11016 When @option{-mesa} is specified, generate code using the
11017 instructions available on ESA/390. Note that @option{-mesa} is
11018 not possible with @option{-m64}.
11019 When generating code compliant to the GNU/Linux for S/390 ABI,
11020 the default is @option{-mesa}. When generating code compliant
11021 to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
11027 Generate (or do not generate) code using the @code{mvcle} instruction
11028 to perform block moves. When @option{-mno-mvcle} is specified,
11029 use a @code{mvc} loop instead. This is the default.
11035 Print (or do not print) additional debug information when compiling.
11036 The default is to not print debug information.
11038 @item -march=@var{cpu-type}
11040 Generate code that will run on @var{cpu-type}, which is the name of a system
11041 representing a certain processor type. Possible values for
11042 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, and @samp{z990}.
11043 When generating code using the instructions available on z/Architecture,
11044 the default is @option{-march=z900}. Otherwise, the default is
11045 @option{-march=g5}.
11047 @item -mtune=@var{cpu-type}
11049 Tune to @var{cpu-type} everything applicable about the generated code,
11050 except for the ABI and the set of available instructions.
11051 The list of @var{cpu-type} values is the same as for @option{-march}.
11052 The default is the value used for @option{-march}.
11055 @itemx -mno-tpf-trace
11056 @opindex mtpf-trace
11057 @opindex mno-tpf-trace
11058 Generate code that adds (does not add) in TPF OS specific branches to trace
11059 routines in the operating system. This option is off by default, even
11060 when compiling for the TPF OS@.
11063 @itemx -mno-fused-madd
11064 @opindex mfused-madd
11065 @opindex mno-fused-madd
11066 Generate code that uses (does not use) the floating point multiply and
11067 accumulate instructions. These instructions are generated by default if
11068 hardware floating point is used.
11070 @item -mwarn-framesize=@var{framesize}
11071 @opindex mwarn-framesize
11072 Emit a warning if the current function exceeds the given frame size. Because
11073 this is a compile time check it doesn't need to be a real problem when the program
11074 runs. It is intended to identify functions which most probably cause
11075 a stack overflow. It is useful to be used in an environment with limited stack
11076 size e.g.@: the linux kernel.
11078 @item -mwarn-dynamicstack
11079 @opindex mwarn-dynamicstack
11080 Emit a warning if the function calls alloca or uses dynamically
11081 sized arrays. This is generally a bad idea with a limited stack size.
11083 @item -mstack-guard=@var{stack-guard}
11084 @item -mstack-size=@var{stack-size}
11085 @opindex mstack-guard
11086 @opindex mstack-size
11087 These arguments always have to be used in conjunction. If they are present the s390
11088 back end emits additional instructions in the function prologue which trigger a trap
11089 if the stack size is @var{stack-guard} bytes above the @var{stack-size}
11090 (remember that the stack on s390 grows downward). These options are intended to
11091 be used to help debugging stack overflow problems. The additionally emitted code
11092 cause only little overhead and hence can also be used in production like systems
11093 without greater performance degradation. The given values have to be exact
11094 powers of 2 and @var{stack-size} has to be greater than @var{stack-guard}.
11095 In order to be efficient the extra code makes the assumption that the stack starts
11096 at an address aligned to the value given by @var{stack-size}.
11100 @subsection SH Options
11102 These @samp{-m} options are defined for the SH implementations:
11107 Generate code for the SH1.
11111 Generate code for the SH2.
11114 Generate code for the SH2e.
11118 Generate code for the SH3.
11122 Generate code for the SH3e.
11126 Generate code for the SH4 without a floating-point unit.
11128 @item -m4-single-only
11129 @opindex m4-single-only
11130 Generate code for the SH4 with a floating-point unit that only
11131 supports single-precision arithmetic.
11135 Generate code for the SH4 assuming the floating-point unit is in
11136 single-precision mode by default.
11140 Generate code for the SH4.
11144 Generate code for the SH4al-dsp, or for a SH4a in such a way that the
11145 floating-point unit is not used.
11147 @item -m4a-single-only
11148 @opindex m4a-single-only
11149 Generate code for the SH4a, in such a way that no double-precision
11150 floating point operations are used.
11153 @opindex m4a-single
11154 Generate code for the SH4a assuming the floating-point unit is in
11155 single-precision mode by default.
11159 Generate code for the SH4a.
11163 Same as @option{-m4a-nofpu}, except that it implicitly passes
11164 @option{-dsp} to the assembler. GCC doesn't generate any DSP
11165 instructions at the moment.
11169 Compile code for the processor in big endian mode.
11173 Compile code for the processor in little endian mode.
11177 Align doubles at 64-bit boundaries. Note that this changes the calling
11178 conventions, and thus some functions from the standard C library will
11179 not work unless you recompile it first with @option{-mdalign}.
11183 Shorten some address references at link time, when possible; uses the
11184 linker option @option{-relax}.
11188 Use 32-bit offsets in @code{switch} tables. The default is to use
11193 Enable the use of the instruction @code{fmovd}.
11197 Comply with the calling conventions defined by Renesas.
11201 Comply with the calling conventions defined by Renesas.
11205 Comply with the calling conventions defined for GCC before the Renesas
11206 conventions were available. This option is the default for all
11207 targets of the SH toolchain except for @samp{sh-symbianelf}.
11210 @opindex mnomacsave
11211 Mark the @code{MAC} register as call-clobbered, even if
11212 @option{-mhitachi} is given.
11216 Increase IEEE-compliance of floating-point code.
11220 Dump instruction size and location in the assembly code.
11223 @opindex mpadstruct
11224 This option is deprecated. It pads structures to multiple of 4 bytes,
11225 which is incompatible with the SH ABI@.
11229 Optimize for space instead of speed. Implied by @option{-Os}.
11232 @opindex mprefergot
11233 When generating position-independent code, emit function calls using
11234 the Global Offset Table instead of the Procedure Linkage Table.
11238 Generate a library function call to invalidate instruction cache
11239 entries, after fixing up a trampoline. This library function call
11240 doesn't assume it can write to the whole memory address space. This
11241 is the default when the target is @code{sh-*-linux*}.
11244 @node SPARC Options
11245 @subsection SPARC Options
11246 @cindex SPARC options
11248 These @samp{-m} options are supported on the SPARC:
11251 @item -mno-app-regs
11253 @opindex mno-app-regs
11255 Specify @option{-mapp-regs} to generate output using the global registers
11256 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
11259 To be fully SVR4 ABI compliant at the cost of some performance loss,
11260 specify @option{-mno-app-regs}. You should compile libraries and system
11261 software with this option.
11264 @itemx -mhard-float
11266 @opindex mhard-float
11267 Generate output containing floating point instructions. This is the
11271 @itemx -msoft-float
11273 @opindex msoft-float
11274 Generate output containing library calls for floating point.
11275 @strong{Warning:} the requisite libraries are not available for all SPARC
11276 targets. Normally the facilities of the machine's usual C compiler are
11277 used, but this cannot be done directly in cross-compilation. You must make
11278 your own arrangements to provide suitable library functions for
11279 cross-compilation. The embedded targets @samp{sparc-*-aout} and
11280 @samp{sparclite-*-*} do provide software floating point support.
11282 @option{-msoft-float} changes the calling convention in the output file;
11283 therefore, it is only useful if you compile @emph{all} of a program with
11284 this option. In particular, you need to compile @file{libgcc.a}, the
11285 library that comes with GCC, with @option{-msoft-float} in order for
11288 @item -mhard-quad-float
11289 @opindex mhard-quad-float
11290 Generate output containing quad-word (long double) floating point
11293 @item -msoft-quad-float
11294 @opindex msoft-quad-float
11295 Generate output containing library calls for quad-word (long double)
11296 floating point instructions. The functions called are those specified
11297 in the SPARC ABI@. This is the default.
11299 As of this writing, there are no SPARC implementations that have hardware
11300 support for the quad-word floating point instructions. They all invoke
11301 a trap handler for one of these instructions, and then the trap handler
11302 emulates the effect of the instruction. Because of the trap handler overhead,
11303 this is much slower than calling the ABI library routines. Thus the
11304 @option{-msoft-quad-float} option is the default.
11306 @item -mno-unaligned-doubles
11307 @itemx -munaligned-doubles
11308 @opindex mno-unaligned-doubles
11309 @opindex munaligned-doubles
11310 Assume that doubles have 8 byte alignment. This is the default.
11312 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
11313 alignment only if they are contained in another type, or if they have an
11314 absolute address. Otherwise, it assumes they have 4 byte alignment.
11315 Specifying this option avoids some rare compatibility problems with code
11316 generated by other compilers. It is not the default because it results
11317 in a performance loss, especially for floating point code.
11319 @item -mno-faster-structs
11320 @itemx -mfaster-structs
11321 @opindex mno-faster-structs
11322 @opindex mfaster-structs
11323 With @option{-mfaster-structs}, the compiler assumes that structures
11324 should have 8 byte alignment. This enables the use of pairs of
11325 @code{ldd} and @code{std} instructions for copies in structure
11326 assignment, in place of twice as many @code{ld} and @code{st} pairs.
11327 However, the use of this changed alignment directly violates the SPARC
11328 ABI@. Thus, it's intended only for use on targets where the developer
11329 acknowledges that their resulting code will not be directly in line with
11330 the rules of the ABI@.
11332 @item -mimpure-text
11333 @opindex mimpure-text
11334 @option{-mimpure-text}, used in addition to @option{-shared}, tells
11335 the compiler to not pass @option{-z text} to the linker when linking a
11336 shared object. Using this option, you can link position-dependent
11337 code into a shared object.
11339 @option{-mimpure-text} suppresses the ``relocations remain against
11340 allocatable but non-writable sections'' linker error message.
11341 However, the necessary relocations will trigger copy-on-write, and the
11342 shared object is not actually shared across processes. Instead of
11343 using @option{-mimpure-text}, you should compile all source code with
11344 @option{-fpic} or @option{-fPIC}.
11346 This option is only available on SunOS and Solaris.
11348 @item -mcpu=@var{cpu_type}
11350 Set the instruction set, register set, and instruction scheduling parameters
11351 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
11352 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
11353 @samp{f930}, @samp{f934}, @samp{hypersparc}, @samp{sparclite86x},
11354 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc}, and
11355 @samp{ultrasparc3}.
11357 Default instruction scheduling parameters are used for values that select
11358 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
11359 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
11361 Here is a list of each supported architecture and their supported
11366 v8: supersparc, hypersparc
11367 sparclite: f930, f934, sparclite86x
11369 v9: ultrasparc, ultrasparc3
11372 By default (unless configured otherwise), GCC generates code for the V7
11373 variant of the SPARC architecture. With @option{-mcpu=cypress}, the compiler
11374 additionally optimizes it for the Cypress CY7C602 chip, as used in the
11375 SPARCStation/SPARCServer 3xx series. This is also appropriate for the older
11376 SPARCStation 1, 2, IPX etc.
11378 With @option{-mcpu=v8}, GCC generates code for the V8 variant of the SPARC
11379 architecture. The only difference from V7 code is that the compiler emits
11380 the integer multiply and integer divide instructions which exist in SPARC-V8
11381 but not in SPARC-V7. With @option{-mcpu=supersparc}, the compiler additionally
11382 optimizes it for the SuperSPARC chip, as used in the SPARCStation 10, 1000 and
11385 With @option{-mcpu=sparclite}, GCC generates code for the SPARClite variant of
11386 the SPARC architecture. This adds the integer multiply, integer divide step
11387 and scan (@code{ffs}) instructions which exist in SPARClite but not in SPARC-V7.
11388 With @option{-mcpu=f930}, the compiler additionally optimizes it for the
11389 Fujitsu MB86930 chip, which is the original SPARClite, with no FPU@. With
11390 @option{-mcpu=f934}, the compiler additionally optimizes it for the Fujitsu
11391 MB86934 chip, which is the more recent SPARClite with FPU@.
11393 With @option{-mcpu=sparclet}, GCC generates code for the SPARClet variant of
11394 the SPARC architecture. This adds the integer multiply, multiply/accumulate,
11395 integer divide step and scan (@code{ffs}) instructions which exist in SPARClet
11396 but not in SPARC-V7. With @option{-mcpu=tsc701}, the compiler additionally
11397 optimizes it for the TEMIC SPARClet chip.
11399 With @option{-mcpu=v9}, GCC generates code for the V9 variant of the SPARC
11400 architecture. This adds 64-bit integer and floating-point move instructions,
11401 3 additional floating-point condition code registers and conditional move
11402 instructions. With @option{-mcpu=ultrasparc}, the compiler additionally
11403 optimizes it for the Sun UltraSPARC I/II chips. With
11404 @option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
11405 Sun UltraSPARC III chip.
11407 @item -mtune=@var{cpu_type}
11409 Set the instruction scheduling parameters for machine type
11410 @var{cpu_type}, but do not set the instruction set or register set that the
11411 option @option{-mcpu=@var{cpu_type}} would.
11413 The same values for @option{-mcpu=@var{cpu_type}} can be used for
11414 @option{-mtune=@var{cpu_type}}, but the only useful values are those
11415 that select a particular cpu implementation. Those are @samp{cypress},
11416 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
11417 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc}, and
11418 @samp{ultrasparc3}.
11423 @opindex mno-v8plus
11424 With @option{-mv8plus}, GCC generates code for the SPARC-V8+ ABI@. The
11425 difference from the V8 ABI is that the global and out registers are
11426 considered 64-bit wide. This is enabled by default on Solaris in 32-bit
11427 mode for all SPARC-V9 processors.
11433 With @option{-mvis}, GCC generates code that takes advantage of the UltraSPARC
11434 Visual Instruction Set extensions. The default is @option{-mno-vis}.
11437 These @samp{-m} options are supported in addition to the above
11438 on SPARC-V9 processors in 64-bit environments:
11441 @item -mlittle-endian
11442 @opindex mlittle-endian
11443 Generate code for a processor running in little-endian mode. It is only
11444 available for a few configurations and most notably not on Solaris.
11450 Generate code for a 32-bit or 64-bit environment.
11451 The 32-bit environment sets int, long and pointer to 32 bits.
11452 The 64-bit environment sets int to 32 bits and long and pointer
11455 @item -mcmodel=medlow
11456 @opindex mcmodel=medlow
11457 Generate code for the Medium/Low code model: 64-bit addresses, programs
11458 must be linked in the low 32 bits of memory. Programs can be statically
11459 or dynamically linked.
11461 @item -mcmodel=medmid
11462 @opindex mcmodel=medmid
11463 Generate code for the Medium/Middle code model: 64-bit addresses, programs
11464 must be linked in the low 44 bits of memory, the text and data segments must
11465 be less than 2GB in size and the data segment must be located within 2GB of
11468 @item -mcmodel=medany
11469 @opindex mcmodel=medany
11470 Generate code for the Medium/Anywhere code model: 64-bit addresses, programs
11471 may be linked anywhere in memory, the text and data segments must be less
11472 than 2GB in size and the data segment must be located within 2GB of the
11475 @item -mcmodel=embmedany
11476 @opindex mcmodel=embmedany
11477 Generate code for the Medium/Anywhere code model for embedded systems:
11478 64-bit addresses, the text and data segments must be less than 2GB in
11479 size, both starting anywhere in memory (determined at link time). The
11480 global register %g4 points to the base of the data segment. Programs
11481 are statically linked and PIC is not supported.
11484 @itemx -mno-stack-bias
11485 @opindex mstack-bias
11486 @opindex mno-stack-bias
11487 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
11488 frame pointer if present, are offset by @minus{}2047 which must be added back
11489 when making stack frame references. This is the default in 64-bit mode.
11490 Otherwise, assume no such offset is present.
11493 These switches are supported in addition to the above on Solaris:
11498 Add support for multithreading using the Solaris threads library. This
11499 option sets flags for both the preprocessor and linker. This option does
11500 not affect the thread safety of object code produced by the compiler or
11501 that of libraries supplied with it.
11505 Add support for multithreading using the POSIX threads library. This
11506 option sets flags for both the preprocessor and linker. This option does
11507 not affect the thread safety of object code produced by the compiler or
11508 that of libraries supplied with it.
11511 @node System V Options
11512 @subsection Options for System V
11514 These additional options are available on System V Release 4 for
11515 compatibility with other compilers on those systems:
11520 Create a shared object.
11521 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
11525 Identify the versions of each tool used by the compiler, in a
11526 @code{.ident} assembler directive in the output.
11530 Refrain from adding @code{.ident} directives to the output file (this is
11533 @item -YP,@var{dirs}
11535 Search the directories @var{dirs}, and no others, for libraries
11536 specified with @option{-l}.
11538 @item -Ym,@var{dir}
11540 Look in the directory @var{dir} to find the M4 preprocessor.
11541 The assembler uses this option.
11542 @c This is supposed to go with a -Yd for predefined M4 macro files, but
11543 @c the generic assembler that comes with Solaris takes just -Ym.
11546 @node TMS320C3x/C4x Options
11547 @subsection TMS320C3x/C4x Options
11548 @cindex TMS320C3x/C4x Options
11550 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
11554 @item -mcpu=@var{cpu_type}
11556 Set the instruction set, register set, and instruction scheduling
11557 parameters for machine type @var{cpu_type}. Supported values for
11558 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
11559 @samp{c44}. The default is @samp{c40} to generate code for the
11564 @itemx -msmall-memory
11566 @opindex mbig-memory
11568 @opindex msmall-memory
11570 Generates code for the big or small memory model. The small memory
11571 model assumed that all data fits into one 64K word page. At run-time
11572 the data page (DP) register must be set to point to the 64K page
11573 containing the .bss and .data program sections. The big memory model is
11574 the default and requires reloading of the DP register for every direct
11581 Allow (disallow) allocation of general integer operands into the block
11582 count register BK@.
11588 Enable (disable) generation of code using decrement and branch,
11589 DBcond(D), instructions. This is enabled by default for the C4x. To be
11590 on the safe side, this is disabled for the C3x, since the maximum
11591 iteration count on the C3x is @math{2^{23} + 1} (but who iterates loops more than
11592 @math{2^{23}} times on the C3x?). Note that GCC will try to reverse a loop so
11593 that it can utilize the decrement and branch instruction, but will give
11594 up if there is more than one memory reference in the loop. Thus a loop
11595 where the loop counter is decremented can generate slightly more
11596 efficient code, in cases where the RPTB instruction cannot be utilized.
11598 @item -mdp-isr-reload
11600 @opindex mdp-isr-reload
11602 Force the DP register to be saved on entry to an interrupt service
11603 routine (ISR), reloaded to point to the data section, and restored on
11604 exit from the ISR@. This should not be required unless someone has
11605 violated the small memory model by modifying the DP register, say within
11612 For the C3x use the 24-bit MPYI instruction for integer multiplies
11613 instead of a library call to guarantee 32-bit results. Note that if one
11614 of the operands is a constant, then the multiplication will be performed
11615 using shifts and adds. If the @option{-mmpyi} option is not specified for the C3x,
11616 then squaring operations are performed inline instead of a library call.
11619 @itemx -mno-fast-fix
11621 @opindex mno-fast-fix
11622 The C3x/C4x FIX instruction to convert a floating point value to an
11623 integer value chooses the nearest integer less than or equal to the
11624 floating point value rather than to the nearest integer. Thus if the
11625 floating point number is negative, the result will be incorrectly
11626 truncated an additional code is necessary to detect and correct this
11627 case. This option can be used to disable generation of the additional
11628 code required to correct the result.
11634 Enable (disable) generation of repeat block sequences using the RPTB
11635 instruction for zero overhead looping. The RPTB construct is only used
11636 for innermost loops that do not call functions or jump across the loop
11637 boundaries. There is no advantage having nested RPTB loops due to the
11638 overhead required to save and restore the RC, RS, and RE registers.
11639 This is enabled by default with @option{-O2}.
11641 @item -mrpts=@var{count}
11645 Enable (disable) the use of the single instruction repeat instruction
11646 RPTS@. If a repeat block contains a single instruction, and the loop
11647 count can be guaranteed to be less than the value @var{count}, GCC will
11648 emit a RPTS instruction instead of a RPTB@. If no value is specified,
11649 then a RPTS will be emitted even if the loop count cannot be determined
11650 at compile time. Note that the repeated instruction following RPTS does
11651 not have to be reloaded from memory each iteration, thus freeing up the
11652 CPU buses for operands. However, since interrupts are blocked by this
11653 instruction, it is disabled by default.
11655 @item -mloop-unsigned
11656 @itemx -mno-loop-unsigned
11657 @opindex mloop-unsigned
11658 @opindex mno-loop-unsigned
11659 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
11660 is @math{2^{31} + 1} since these instructions test if the iteration count is
11661 negative to terminate the loop. If the iteration count is unsigned
11662 there is a possibility than the @math{2^{31} + 1} maximum iteration count may be
11663 exceeded. This switch allows an unsigned iteration count.
11667 Try to emit an assembler syntax that the TI assembler (asm30) is happy
11668 with. This also enforces compatibility with the API employed by the TI
11669 C3x C compiler. For example, long doubles are passed as structures
11670 rather than in floating point registers.
11676 Generate code that uses registers (stack) for passing arguments to functions.
11677 By default, arguments are passed in registers where possible rather
11678 than by pushing arguments on to the stack.
11680 @item -mparallel-insns
11681 @itemx -mno-parallel-insns
11682 @opindex mparallel-insns
11683 @opindex mno-parallel-insns
11684 Allow the generation of parallel instructions. This is enabled by
11685 default with @option{-O2}.
11687 @item -mparallel-mpy
11688 @itemx -mno-parallel-mpy
11689 @opindex mparallel-mpy
11690 @opindex mno-parallel-mpy
11691 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
11692 provided @option{-mparallel-insns} is also specified. These instructions have
11693 tight register constraints which can pessimize the code generation
11694 of large functions.
11699 @subsection V850 Options
11700 @cindex V850 Options
11702 These @samp{-m} options are defined for V850 implementations:
11706 @itemx -mno-long-calls
11707 @opindex mlong-calls
11708 @opindex mno-long-calls
11709 Treat all calls as being far away (near). If calls are assumed to be
11710 far away, the compiler will always load the functions address up into a
11711 register, and call indirect through the pointer.
11717 Do not optimize (do optimize) basic blocks that use the same index
11718 pointer 4 or more times to copy pointer into the @code{ep} register, and
11719 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
11720 option is on by default if you optimize.
11722 @item -mno-prolog-function
11723 @itemx -mprolog-function
11724 @opindex mno-prolog-function
11725 @opindex mprolog-function
11726 Do not use (do use) external functions to save and restore registers
11727 at the prologue and epilogue of a function. The external functions
11728 are slower, but use less code space if more than one function saves
11729 the same number of registers. The @option{-mprolog-function} option
11730 is on by default if you optimize.
11734 Try to make the code as small as possible. At present, this just turns
11735 on the @option{-mep} and @option{-mprolog-function} options.
11737 @item -mtda=@var{n}
11739 Put static or global variables whose size is @var{n} bytes or less into
11740 the tiny data area that register @code{ep} points to. The tiny data
11741 area can hold up to 256 bytes in total (128 bytes for byte references).
11743 @item -msda=@var{n}
11745 Put static or global variables whose size is @var{n} bytes or less into
11746 the small data area that register @code{gp} points to. The small data
11747 area can hold up to 64 kilobytes.
11749 @item -mzda=@var{n}
11751 Put static or global variables whose size is @var{n} bytes or less into
11752 the first 32 kilobytes of memory.
11756 Specify that the target processor is the V850.
11759 @opindex mbig-switch
11760 Generate code suitable for big switch tables. Use this option only if
11761 the assembler/linker complain about out of range branches within a switch
11766 This option will cause r2 and r5 to be used in the code generated by
11767 the compiler. This setting is the default.
11769 @item -mno-app-regs
11770 @opindex mno-app-regs
11771 This option will cause r2 and r5 to be treated as fixed registers.
11775 Specify that the target processor is the V850E1. The preprocessor
11776 constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
11777 this option is used.
11781 Specify that the target processor is the V850E@. The preprocessor
11782 constant @samp{__v850e__} will be defined if this option is used.
11784 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
11785 are defined then a default target processor will be chosen and the
11786 relevant @samp{__v850*__} preprocessor constant will be defined.
11788 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
11789 defined, regardless of which processor variant is the target.
11791 @item -mdisable-callt
11792 @opindex mdisable-callt
11793 This option will suppress generation of the CALLT instruction for the
11794 v850e and v850e1 flavors of the v850 architecture. The default is
11795 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
11800 @subsection VAX Options
11801 @cindex VAX options
11803 These @samp{-m} options are defined for the VAX:
11808 Do not output certain jump instructions (@code{aobleq} and so on)
11809 that the Unix assembler for the VAX cannot handle across long
11814 Do output those jump instructions, on the assumption that you
11815 will assemble with the GNU assembler.
11819 Output code for g-format floating point numbers instead of d-format.
11822 @node x86-64 Options
11823 @subsection x86-64 Options
11824 @cindex x86-64 options
11826 These are listed under @xref{i386 and x86-64 Options}.
11828 @node Xstormy16 Options
11829 @subsection Xstormy16 Options
11830 @cindex Xstormy16 Options
11832 These options are defined for Xstormy16:
11837 Choose startup files and linker script suitable for the simulator.
11840 @node Xtensa Options
11841 @subsection Xtensa Options
11842 @cindex Xtensa Options
11844 These options are supported for Xtensa targets:
11848 @itemx -mno-const16
11850 @opindex mno-const16
11851 Enable or disable use of @code{CONST16} instructions for loading
11852 constant values. The @code{CONST16} instruction is currently not a
11853 standard option from Tensilica. When enabled, @code{CONST16}
11854 instructions are always used in place of the standard @code{L32R}
11855 instructions. The use of @code{CONST16} is enabled by default only if
11856 the @code{L32R} instruction is not available.
11859 @itemx -mno-fused-madd
11860 @opindex mfused-madd
11861 @opindex mno-fused-madd
11862 Enable or disable use of fused multiply/add and multiply/subtract
11863 instructions in the floating-point option. This has no effect if the
11864 floating-point option is not also enabled. Disabling fused multiply/add
11865 and multiply/subtract instructions forces the compiler to use separate
11866 instructions for the multiply and add/subtract operations. This may be
11867 desirable in some cases where strict IEEE 754-compliant results are
11868 required: the fused multiply add/subtract instructions do not round the
11869 intermediate result, thereby producing results with @emph{more} bits of
11870 precision than specified by the IEEE standard. Disabling fused multiply
11871 add/subtract instructions also ensures that the program output is not
11872 sensitive to the compiler's ability to combine multiply and add/subtract
11875 @item -mtext-section-literals
11876 @itemx -mno-text-section-literals
11877 @opindex mtext-section-literals
11878 @opindex mno-text-section-literals
11879 Control the treatment of literal pools. The default is
11880 @option{-mno-text-section-literals}, which places literals in a separate
11881 section in the output file. This allows the literal pool to be placed
11882 in a data RAM/ROM, and it also allows the linker to combine literal
11883 pools from separate object files to remove redundant literals and
11884 improve code size. With @option{-mtext-section-literals}, the literals
11885 are interspersed in the text section in order to keep them as close as
11886 possible to their references. This may be necessary for large assembly
11889 @item -mtarget-align
11890 @itemx -mno-target-align
11891 @opindex mtarget-align
11892 @opindex mno-target-align
11893 When this option is enabled, GCC instructs the assembler to
11894 automatically align instructions to reduce branch penalties at the
11895 expense of some code density. The assembler attempts to widen density
11896 instructions to align branch targets and the instructions following call
11897 instructions. If there are not enough preceding safe density
11898 instructions to align a target, no widening will be performed. The
11899 default is @option{-mtarget-align}. These options do not affect the
11900 treatment of auto-aligned instructions like @code{LOOP}, which the
11901 assembler will always align, either by widening density instructions or
11902 by inserting no-op instructions.
11905 @itemx -mno-longcalls
11906 @opindex mlongcalls
11907 @opindex mno-longcalls
11908 When this option is enabled, GCC instructs the assembler to translate
11909 direct calls to indirect calls unless it can determine that the target
11910 of a direct call is in the range allowed by the call instruction. This
11911 translation typically occurs for calls to functions in other source
11912 files. Specifically, the assembler translates a direct @code{CALL}
11913 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
11914 The default is @option{-mno-longcalls}. This option should be used in
11915 programs where the call target can potentially be out of range. This
11916 option is implemented in the assembler, not the compiler, so the
11917 assembly code generated by GCC will still show direct call
11918 instructions---look at the disassembled object code to see the actual
11919 instructions. Note that the assembler will use an indirect call for
11920 every cross-file call, not just those that really will be out of range.
11923 @node zSeries Options
11924 @subsection zSeries Options
11925 @cindex zSeries options
11927 These are listed under @xref{S/390 and zSeries Options}.
11929 @node Code Gen Options
11930 @section Options for Code Generation Conventions
11931 @cindex code generation conventions
11932 @cindex options, code generation
11933 @cindex run-time options
11935 These machine-independent options control the interface conventions
11936 used in code generation.
11938 Most of them have both positive and negative forms; the negative form
11939 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
11940 one of the forms is listed---the one which is not the default. You
11941 can figure out the other form by either removing @samp{no-} or adding
11945 @item -fbounds-check
11946 @opindex fbounds-check
11947 For front-ends that support it, generate additional code to check that
11948 indices used to access arrays are within the declared range. This is
11949 currently only supported by the Java and Fortran 77 front-ends, where
11950 this option defaults to true and false respectively.
11954 This option generates traps for signed overflow on addition, subtraction,
11955 multiplication operations.
11959 This option instructs the compiler to assume that signed arithmetic
11960 overflow of addition, subtraction and multiplication wraps around
11961 using twos-complement representation. This flag enables some optimizations
11962 and disables other. This option is enabled by default for the Java
11963 front-end, as required by the Java language specification.
11966 @opindex fexceptions
11967 Enable exception handling. Generates extra code needed to propagate
11968 exceptions. For some targets, this implies GCC will generate frame
11969 unwind information for all functions, which can produce significant data
11970 size overhead, although it does not affect execution. If you do not
11971 specify this option, GCC will enable it by default for languages like
11972 C++ which normally require exception handling, and disable it for
11973 languages like C that do not normally require it. However, you may need
11974 to enable this option when compiling C code that needs to interoperate
11975 properly with exception handlers written in C++. You may also wish to
11976 disable this option if you are compiling older C++ programs that don't
11977 use exception handling.
11979 @item -fnon-call-exceptions
11980 @opindex fnon-call-exceptions
11981 Generate code that allows trapping instructions to throw exceptions.
11982 Note that this requires platform-specific runtime support that does
11983 not exist everywhere. Moreover, it only allows @emph{trapping}
11984 instructions to throw exceptions, i.e.@: memory references or floating
11985 point instructions. It does not allow exceptions to be thrown from
11986 arbitrary signal handlers such as @code{SIGALRM}.
11988 @item -funwind-tables
11989 @opindex funwind-tables
11990 Similar to @option{-fexceptions}, except that it will just generate any needed
11991 static data, but will not affect the generated code in any other way.
11992 You will normally not enable this option; instead, a language processor
11993 that needs this handling would enable it on your behalf.
11995 @item -fasynchronous-unwind-tables
11996 @opindex fasynchronous-unwind-tables
11997 Generate unwind table in dwarf2 format, if supported by target machine. The
11998 table is exact at each instruction boundary, so it can be used for stack
11999 unwinding from asynchronous events (such as debugger or garbage collector).
12001 @item -fpcc-struct-return
12002 @opindex fpcc-struct-return
12003 Return ``short'' @code{struct} and @code{union} values in memory like
12004 longer ones, rather than in registers. This convention is less
12005 efficient, but it has the advantage of allowing intercallability between
12006 GCC-compiled files and files compiled with other compilers, particularly
12007 the Portable C Compiler (pcc).
12009 The precise convention for returning structures in memory depends
12010 on the target configuration macros.
12012 Short structures and unions are those whose size and alignment match
12013 that of some integer type.
12015 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
12016 switch is not binary compatible with code compiled with the
12017 @option{-freg-struct-return} switch.
12018 Use it to conform to a non-default application binary interface.
12020 @item -freg-struct-return
12021 @opindex freg-struct-return
12022 Return @code{struct} and @code{union} values in registers when possible.
12023 This is more efficient for small structures than
12024 @option{-fpcc-struct-return}.
12026 If you specify neither @option{-fpcc-struct-return} nor
12027 @option{-freg-struct-return}, GCC defaults to whichever convention is
12028 standard for the target. If there is no standard convention, GCC
12029 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
12030 the principal compiler. In those cases, we can choose the standard, and
12031 we chose the more efficient register return alternative.
12033 @strong{Warning:} code compiled with the @option{-freg-struct-return}
12034 switch is not binary compatible with code compiled with the
12035 @option{-fpcc-struct-return} switch.
12036 Use it to conform to a non-default application binary interface.
12038 @item -fshort-enums
12039 @opindex fshort-enums
12040 Allocate to an @code{enum} type only as many bytes as it needs for the
12041 declared range of possible values. Specifically, the @code{enum} type
12042 will be equivalent to the smallest integer type which has enough room.
12044 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
12045 code that is not binary compatible with code generated without that switch.
12046 Use it to conform to a non-default application binary interface.
12048 @item -fshort-double
12049 @opindex fshort-double
12050 Use the same size for @code{double} as for @code{float}.
12052 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
12053 code that is not binary compatible with code generated without that switch.
12054 Use it to conform to a non-default application binary interface.
12056 @item -fshort-wchar
12057 @opindex fshort-wchar
12058 Override the underlying type for @samp{wchar_t} to be @samp{short
12059 unsigned int} instead of the default for the target. This option is
12060 useful for building programs to run under WINE@.
12062 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
12063 code that is not binary compatible with code generated without that switch.
12064 Use it to conform to a non-default application binary interface.
12066 @item -fshared-data
12067 @opindex fshared-data
12068 Requests that the data and non-@code{const} variables of this
12069 compilation be shared data rather than private data. The distinction
12070 makes sense only on certain operating systems, where shared data is
12071 shared between processes running the same program, while private data
12072 exists in one copy per process.
12075 @opindex fno-common
12076 In C, allocate even uninitialized global variables in the data section of the
12077 object file, rather than generating them as common blocks. This has the
12078 effect that if the same variable is declared (without @code{extern}) in
12079 two different compilations, you will get an error when you link them.
12080 The only reason this might be useful is if you wish to verify that the
12081 program will work on other systems which always work this way.
12085 Ignore the @samp{#ident} directive.
12087 @item -finhibit-size-directive
12088 @opindex finhibit-size-directive
12089 Don't output a @code{.size} assembler directive, or anything else that
12090 would cause trouble if the function is split in the middle, and the
12091 two halves are placed at locations far apart in memory. This option is
12092 used when compiling @file{crtstuff.c}; you should not need to use it
12095 @item -fverbose-asm
12096 @opindex fverbose-asm
12097 Put extra commentary information in the generated assembly code to
12098 make it more readable. This option is generally only of use to those
12099 who actually need to read the generated assembly code (perhaps while
12100 debugging the compiler itself).
12102 @option{-fno-verbose-asm}, the default, causes the
12103 extra information to be omitted and is useful when comparing two assembler
12108 @cindex global offset table
12110 Generate position-independent code (PIC) suitable for use in a shared
12111 library, if supported for the target machine. Such code accesses all
12112 constant addresses through a global offset table (GOT)@. The dynamic
12113 loader resolves the GOT entries when the program starts (the dynamic
12114 loader is not part of GCC; it is part of the operating system). If
12115 the GOT size for the linked executable exceeds a machine-specific
12116 maximum size, you get an error message from the linker indicating that
12117 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
12118 instead. (These maximums are 8k on the SPARC and 32k
12119 on the m68k and RS/6000. The 386 has no such limit.)
12121 Position-independent code requires special support, and therefore works
12122 only on certain machines. For the 386, GCC supports PIC for System V
12123 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
12124 position-independent.
12128 If supported for the target machine, emit position-independent code,
12129 suitable for dynamic linking and avoiding any limit on the size of the
12130 global offset table. This option makes a difference on the m68k,
12131 PowerPC and SPARC@.
12133 Position-independent code requires special support, and therefore works
12134 only on certain machines.
12140 These options are similar to @option{-fpic} and @option{-fPIC}, but
12141 generated position independent code can be only linked into executables.
12142 Usually these options are used when @option{-pie} GCC option will be
12143 used during linking.
12145 @item -ffixed-@var{reg}
12147 Treat the register named @var{reg} as a fixed register; generated code
12148 should never refer to it (except perhaps as a stack pointer, frame
12149 pointer or in some other fixed role).
12151 @var{reg} must be the name of a register. The register names accepted
12152 are machine-specific and are defined in the @code{REGISTER_NAMES}
12153 macro in the machine description macro file.
12155 This flag does not have a negative form, because it specifies a
12158 @item -fcall-used-@var{reg}
12159 @opindex fcall-used
12160 Treat the register named @var{reg} as an allocable register that is
12161 clobbered by function calls. It may be allocated for temporaries or
12162 variables that do not live across a call. Functions compiled this way
12163 will not save and restore the register @var{reg}.
12165 It is an error to used this flag with the frame pointer or stack pointer.
12166 Use of this flag for other registers that have fixed pervasive roles in
12167 the machine's execution model will produce disastrous results.
12169 This flag does not have a negative form, because it specifies a
12172 @item -fcall-saved-@var{reg}
12173 @opindex fcall-saved
12174 Treat the register named @var{reg} as an allocable register saved by
12175 functions. It may be allocated even for temporaries or variables that
12176 live across a call. Functions compiled this way will save and restore
12177 the register @var{reg} if they use it.
12179 It is an error to used this flag with the frame pointer or stack pointer.
12180 Use of this flag for other registers that have fixed pervasive roles in
12181 the machine's execution model will produce disastrous results.
12183 A different sort of disaster will result from the use of this flag for
12184 a register in which function values may be returned.
12186 This flag does not have a negative form, because it specifies a
12189 @item -fpack-struct[=@var{n}]
12190 @opindex fpack-struct
12191 Without a value specified, pack all structure members together without
12192 holes. When a value is specified (which must be a small power of two), pack
12193 structure members according to this value, representing the maximum
12194 alignment (that is, objects with default alignment requirements larger than
12195 this will be output potentially unaligned at the next fitting location.
12197 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
12198 code that is not binary compatible with code generated without that switch.
12199 Additionally, it makes the code suboptimal.
12200 Use it to conform to a non-default application binary interface.
12202 @item -finstrument-functions
12203 @opindex finstrument-functions
12204 Generate instrumentation calls for entry and exit to functions. Just
12205 after function entry and just before function exit, the following
12206 profiling functions will be called with the address of the current
12207 function and its call site. (On some platforms,
12208 @code{__builtin_return_address} does not work beyond the current
12209 function, so the call site information may not be available to the
12210 profiling functions otherwise.)
12213 void __cyg_profile_func_enter (void *this_fn,
12215 void __cyg_profile_func_exit (void *this_fn,
12219 The first argument is the address of the start of the current function,
12220 which may be looked up exactly in the symbol table.
12222 This instrumentation is also done for functions expanded inline in other
12223 functions. The profiling calls will indicate where, conceptually, the
12224 inline function is entered and exited. This means that addressable
12225 versions of such functions must be available. If all your uses of a
12226 function are expanded inline, this may mean an additional expansion of
12227 code size. If you use @samp{extern inline} in your C code, an
12228 addressable version of such functions must be provided. (This is
12229 normally the case anyways, but if you get lucky and the optimizer always
12230 expands the functions inline, you might have gotten away without
12231 providing static copies.)
12233 A function may be given the attribute @code{no_instrument_function}, in
12234 which case this instrumentation will not be done. This can be used, for
12235 example, for the profiling functions listed above, high-priority
12236 interrupt routines, and any functions from which the profiling functions
12237 cannot safely be called (perhaps signal handlers, if the profiling
12238 routines generate output or allocate memory).
12240 @item -fstack-check
12241 @opindex fstack-check
12242 Generate code to verify that you do not go beyond the boundary of the
12243 stack. You should specify this flag if you are running in an
12244 environment with multiple threads, but only rarely need to specify it in
12245 a single-threaded environment since stack overflow is automatically
12246 detected on nearly all systems if there is only one stack.
12248 Note that this switch does not actually cause checking to be done; the
12249 operating system must do that. The switch causes generation of code
12250 to ensure that the operating system sees the stack being extended.
12252 @item -fstack-limit-register=@var{reg}
12253 @itemx -fstack-limit-symbol=@var{sym}
12254 @itemx -fno-stack-limit
12255 @opindex fstack-limit-register
12256 @opindex fstack-limit-symbol
12257 @opindex fno-stack-limit
12258 Generate code to ensure that the stack does not grow beyond a certain value,
12259 either the value of a register or the address of a symbol. If the stack
12260 would grow beyond the value, a signal is raised. For most targets,
12261 the signal is raised before the stack overruns the boundary, so
12262 it is possible to catch the signal without taking special precautions.
12264 For instance, if the stack starts at absolute address @samp{0x80000000}
12265 and grows downwards, you can use the flags
12266 @option{-fstack-limit-symbol=__stack_limit} and
12267 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
12268 of 128KB@. Note that this may only work with the GNU linker.
12270 @cindex aliasing of parameters
12271 @cindex parameters, aliased
12272 @item -fargument-alias
12273 @itemx -fargument-noalias
12274 @itemx -fargument-noalias-global
12275 @opindex fargument-alias
12276 @opindex fargument-noalias
12277 @opindex fargument-noalias-global
12278 Specify the possible relationships among parameters and between
12279 parameters and global data.
12281 @option{-fargument-alias} specifies that arguments (parameters) may
12282 alias each other and may alias global storage.@*
12283 @option{-fargument-noalias} specifies that arguments do not alias
12284 each other, but may alias global storage.@*
12285 @option{-fargument-noalias-global} specifies that arguments do not
12286 alias each other and do not alias global storage.
12288 Each language will automatically use whatever option is required by
12289 the language standard. You should not need to use these options yourself.
12291 @item -fleading-underscore
12292 @opindex fleading-underscore
12293 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
12294 change the way C symbols are represented in the object file. One use
12295 is to help link with legacy assembly code.
12297 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
12298 generate code that is not binary compatible with code generated without that
12299 switch. Use it to conform to a non-default application binary interface.
12300 Not all targets provide complete support for this switch.
12302 @item -ftls-model=@var{model}
12303 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
12304 The @var{model} argument should be one of @code{global-dynamic},
12305 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
12307 The default without @option{-fpic} is @code{initial-exec}; with
12308 @option{-fpic} the default is @code{global-dynamic}.
12310 @item -fvisibility=@var{default|internal|hidden|protected}
12311 @opindex fvisibility
12312 Set the default ELF image symbol visibility to the specified option---all
12313 symbols will be marked with this unless overridden within the code.
12314 Using this feature can very substantially improve linking and
12315 load times of shared object libraries, produce more optimized
12316 code, provide near-perfect API export and prevent symbol clashes.
12317 It is @strong{strongly} recommended that you use this in any shared objects
12320 Despite the nomenclature, @code{default} always means public ie;
12321 available to be linked against from outside the shared object.
12322 @code{protected} and @code{internal} are pretty useless in real-world
12323 usage so the only other commonly used option will be @code{hidden}.
12324 The default if @option{-fvisibility} isn't specified is
12325 @code{default}, i.e., make every
12326 symbol public---this causes the same behavior as previous versions of
12329 A good explanation of the benefits offered by ensuring ELF
12330 symbols have the correct visibility is given by ``How To Write
12331 Shared Libraries'' by Ulrich Drepper (which can be found at
12332 @w{@uref{http://people.redhat.com/~drepper/}})---however a superior
12333 solution made possible by this option to marking things hidden when
12334 the default is public is to make the default hidden and mark things
12335 public. This is the norm with DLL's on Windows and with @option{-fvisibility=hidden}
12336 and @code{__attribute__ ((visibility("default")))} instead of
12337 @code{__declspec(dllexport)} you get almost identical semantics with
12338 identical syntax. This is a great boon to those working with
12339 cross-platform projects.
12341 For those adding visibility support to existing code, you may find
12342 @samp{#pragma GCC visibility} of use. This works by you enclosing
12343 the declarations you wish to set visibility for with (for example)
12344 @samp{#pragma GCC visibility push(hidden)} and
12345 @samp{#pragma GCC visibility pop}. These can be nested up to sixteen
12346 times. Bear in mind that symbol visibility should be viewed @strong{as
12347 part of the API interface contract} and thus all new code should
12348 always specify visibility when it is not the default ie; declarations
12349 only for use within the local DSO should @strong{always} be marked explicitly
12350 as hidden as so to avoid PLT indirection overheads---making this
12351 abundantly clear also aids readability and self-documentation of the code.
12352 Note that due to ISO C++ specification requirements, operator new and
12353 operator delete must always be of default visibility.
12355 An overview of these techniques, their benefits and how to use them
12356 is at @w{@uref{http://www.nedprod.com/programs/gccvisibility.html}}.
12362 @node Environment Variables
12363 @section Environment Variables Affecting GCC
12364 @cindex environment variables
12366 @c man begin ENVIRONMENT
12367 This section describes several environment variables that affect how GCC
12368 operates. Some of them work by specifying directories or prefixes to use
12369 when searching for various kinds of files. Some are used to specify other
12370 aspects of the compilation environment.
12372 Note that you can also specify places to search using options such as
12373 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
12374 take precedence over places specified using environment variables, which
12375 in turn take precedence over those specified by the configuration of GCC@.
12376 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
12377 GNU Compiler Collection (GCC) Internals}.
12382 @c @itemx LC_COLLATE
12384 @c @itemx LC_MONETARY
12385 @c @itemx LC_NUMERIC
12390 @c @findex LC_COLLATE
12391 @findex LC_MESSAGES
12392 @c @findex LC_MONETARY
12393 @c @findex LC_NUMERIC
12397 These environment variables control the way that GCC uses
12398 localization information that allow GCC to work with different
12399 national conventions. GCC inspects the locale categories
12400 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
12401 so. These locale categories can be set to any value supported by your
12402 installation. A typical value is @samp{en_GB.UTF-8} for English in the United
12403 Kingdom encoded in UTF-8.
12405 The @env{LC_CTYPE} environment variable specifies character
12406 classification. GCC uses it to determine the character boundaries in
12407 a string; this is needed for some multibyte encodings that contain quote
12408 and escape characters that would otherwise be interpreted as a string
12411 The @env{LC_MESSAGES} environment variable specifies the language to
12412 use in diagnostic messages.
12414 If the @env{LC_ALL} environment variable is set, it overrides the value
12415 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
12416 and @env{LC_MESSAGES} default to the value of the @env{LANG}
12417 environment variable. If none of these variables are set, GCC
12418 defaults to traditional C English behavior.
12422 If @env{TMPDIR} is set, it specifies the directory to use for temporary
12423 files. GCC uses temporary files to hold the output of one stage of
12424 compilation which is to be used as input to the next stage: for example,
12425 the output of the preprocessor, which is the input to the compiler
12428 @item GCC_EXEC_PREFIX
12429 @findex GCC_EXEC_PREFIX
12430 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
12431 names of the subprograms executed by the compiler. No slash is added
12432 when this prefix is combined with the name of a subprogram, but you can
12433 specify a prefix that ends with a slash if you wish.
12435 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
12436 an appropriate prefix to use based on the pathname it was invoked with.
12438 If GCC cannot find the subprogram using the specified prefix, it
12439 tries looking in the usual places for the subprogram.
12441 The default value of @env{GCC_EXEC_PREFIX} is
12442 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the value
12443 of @code{prefix} when you ran the @file{configure} script.
12445 Other prefixes specified with @option{-B} take precedence over this prefix.
12447 This prefix is also used for finding files such as @file{crt0.o} that are
12450 In addition, the prefix is used in an unusual way in finding the
12451 directories to search for header files. For each of the standard
12452 directories whose name normally begins with @samp{/usr/local/lib/gcc}
12453 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
12454 replacing that beginning with the specified prefix to produce an
12455 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
12456 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
12457 These alternate directories are searched first; the standard directories
12460 @item COMPILER_PATH
12461 @findex COMPILER_PATH
12462 The value of @env{COMPILER_PATH} is a colon-separated list of
12463 directories, much like @env{PATH}. GCC tries the directories thus
12464 specified when searching for subprograms, if it can't find the
12465 subprograms using @env{GCC_EXEC_PREFIX}.
12468 @findex LIBRARY_PATH
12469 The value of @env{LIBRARY_PATH} is a colon-separated list of
12470 directories, much like @env{PATH}. When configured as a native compiler,
12471 GCC tries the directories thus specified when searching for special
12472 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
12473 using GCC also uses these directories when searching for ordinary
12474 libraries for the @option{-l} option (but directories specified with
12475 @option{-L} come first).
12479 @cindex locale definition
12480 This variable is used to pass locale information to the compiler. One way in
12481 which this information is used is to determine the character set to be used
12482 when character literals, string literals and comments are parsed in C and C++.
12483 When the compiler is configured to allow multibyte characters,
12484 the following values for @env{LANG} are recognized:
12488 Recognize JIS characters.
12490 Recognize SJIS characters.
12492 Recognize EUCJP characters.
12495 If @env{LANG} is not defined, or if it has some other value, then the
12496 compiler will use mblen and mbtowc as defined by the default locale to
12497 recognize and translate multibyte characters.
12501 Some additional environments variables affect the behavior of the
12504 @include cppenv.texi
12508 @node Precompiled Headers
12509 @section Using Precompiled Headers
12510 @cindex precompiled headers
12511 @cindex speed of compilation
12513 Often large projects have many header files that are included in every
12514 source file. The time the compiler takes to process these header files
12515 over and over again can account for nearly all of the time required to
12516 build the project. To make builds faster, GCC allows users to
12517 `precompile' a header file; then, if builds can use the precompiled
12518 header file they will be much faster.
12520 @strong{Caution:} There are a few known situations where GCC will
12521 crash when trying to use a precompiled header. If you have trouble
12522 with a precompiled header, you should remove the precompiled header
12523 and compile without it. In addition, please use GCC's on-line
12524 defect-tracking system to report any problems you encounter with
12525 precompiled headers. @xref{Bugs}.
12527 To create a precompiled header file, simply compile it as you would any
12528 other file, if necessary using the @option{-x} option to make the driver
12529 treat it as a C or C++ header file. You will probably want to use a
12530 tool like @command{make} to keep the precompiled header up-to-date when
12531 the headers it contains change.
12533 A precompiled header file will be searched for when @code{#include} is
12534 seen in the compilation. As it searches for the included file
12535 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
12536 compiler looks for a precompiled header in each directory just before it
12537 looks for the include file in that directory. The name searched for is
12538 the name specified in the @code{#include} with @samp{.gch} appended. If
12539 the precompiled header file can't be used, it is ignored.
12541 For instance, if you have @code{#include "all.h"}, and you have
12542 @file{all.h.gch} in the same directory as @file{all.h}, then the
12543 precompiled header file will be used if possible, and the original
12544 header will be used otherwise.
12546 Alternatively, you might decide to put the precompiled header file in a
12547 directory and use @option{-I} to ensure that directory is searched
12548 before (or instead of) the directory containing the original header.
12549 Then, if you want to check that the precompiled header file is always
12550 used, you can put a file of the same name as the original header in this
12551 directory containing an @code{#error} command.
12553 This also works with @option{-include}. So yet another way to use
12554 precompiled headers, good for projects not designed with precompiled
12555 header files in mind, is to simply take most of the header files used by
12556 a project, include them from another header file, precompile that header
12557 file, and @option{-include} the precompiled header. If the header files
12558 have guards against multiple inclusion, they will be skipped because
12559 they've already been included (in the precompiled header).
12561 If you need to precompile the same header file for different
12562 languages, targets, or compiler options, you can instead make a
12563 @emph{directory} named like @file{all.h.gch}, and put each precompiled
12564 header in the directory, perhaps using @option{-o}. It doesn't matter
12565 what you call the files in the directory, every precompiled header in
12566 the directory will be considered. The first precompiled header
12567 encountered in the directory that is valid for this compilation will
12568 be used; they're searched in no particular order.
12570 There are many other possibilities, limited only by your imagination,
12571 good sense, and the constraints of your build system.
12573 A precompiled header file can be used only when these conditions apply:
12577 Only one precompiled header can be used in a particular compilation.
12580 A precompiled header can't be used once the first C token is seen. You
12581 can have preprocessor directives before a precompiled header; you can
12582 even include a precompiled header from inside another header, so long as
12583 there are no C tokens before the @code{#include}.
12586 The precompiled header file must be produced for the same language as
12587 the current compilation. You can't use a C precompiled header for a C++
12591 The precompiled header file must be produced by the same compiler
12592 version and configuration as the current compilation is using.
12593 The easiest way to guarantee this is to use the same compiler binary
12594 for creating and using precompiled headers.
12597 Any macros defined before the precompiled header is included must
12598 either be defined in the same way as when the precompiled header was
12599 generated, or must not affect the precompiled header, which usually
12600 means that the they don't appear in the precompiled header at all.
12602 The @option{-D} option is one way to define a macro before a
12603 precompiled header is included; using a @code{#define} can also do it.
12604 There are also some options that define macros implicitly, like
12605 @option{-O} and @option{-Wdeprecated}; the same rule applies to macros
12608 @item If debugging information is output when using the precompiled
12609 header, using @option{-g} or similar, the same kind of debugging information
12610 must have been output when building the precompiled header. However,
12611 a precompiled header built using @option{-g} can be used in a compilation
12612 when no debugging information is being output.
12614 @item The same @option{-m} options must generally be used when building
12615 and using the precompiled header. @xref{Submodel Options},
12616 for any cases where this rule is relaxed.
12618 @item Each of the following options must be the same when building and using
12619 the precompiled header:
12621 @gccoptlist{-fexceptions -funit-at-a-time}
12624 Some other command-line options starting with @option{-f},
12625 @option{-p}, or @option{-O} must be defined in the same way as when
12626 the precompiled header was generated. At present, it's not clear
12627 which options are safe to change and which are not; the safest choice
12628 is to use exactly the same options when generating and using the
12629 precompiled header. The following are known to be safe:
12631 @gccoptlist{-fpreprocessed -pedantic-errors}
12635 For all of these except the last, the compiler will automatically
12636 ignore the precompiled header if the conditions aren't met. If you
12637 find an option combination that doesn't work and doesn't cause the
12638 precompiled header to be ignored, please consider filing a bug report,
12641 If you do use differing options when generating and using the
12642 precompiled header, the actual behavior will be a mixture of the
12643 behavior for the options. For instance, if you use @option{-g} to
12644 generate the precompiled header but not when using it, you may or may
12645 not get debugging information for routines in the precompiled header.
12647 @node Running Protoize
12648 @section Running Protoize
12650 The program @code{protoize} is an optional part of GCC@. You can use
12651 it to add prototypes to a program, thus converting the program to ISO
12652 C in one respect. The companion program @code{unprotoize} does the
12653 reverse: it removes argument types from any prototypes that are found.
12655 When you run these programs, you must specify a set of source files as
12656 command line arguments. The conversion programs start out by compiling
12657 these files to see what functions they define. The information gathered
12658 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
12660 After scanning comes actual conversion. The specified files are all
12661 eligible to be converted; any files they include (whether sources or
12662 just headers) are eligible as well.
12664 But not all the eligible files are converted. By default,
12665 @code{protoize} and @code{unprotoize} convert only source and header
12666 files in the current directory. You can specify additional directories
12667 whose files should be converted with the @option{-d @var{directory}}
12668 option. You can also specify particular files to exclude with the
12669 @option{-x @var{file}} option. A file is converted if it is eligible, its
12670 directory name matches one of the specified directory names, and its
12671 name within the directory has not been excluded.
12673 Basic conversion with @code{protoize} consists of rewriting most
12674 function definitions and function declarations to specify the types of
12675 the arguments. The only ones not rewritten are those for varargs
12678 @code{protoize} optionally inserts prototype declarations at the
12679 beginning of the source file, to make them available for any calls that
12680 precede the function's definition. Or it can insert prototype
12681 declarations with block scope in the blocks where undeclared functions
12684 Basic conversion with @code{unprotoize} consists of rewriting most
12685 function declarations to remove any argument types, and rewriting
12686 function definitions to the old-style pre-ISO form.
12688 Both conversion programs print a warning for any function declaration or
12689 definition that they can't convert. You can suppress these warnings
12692 The output from @code{protoize} or @code{unprotoize} replaces the
12693 original source file. The original file is renamed to a name ending
12694 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
12695 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
12696 for DOS) file already exists, then the source file is simply discarded.
12698 @code{protoize} and @code{unprotoize} both depend on GCC itself to
12699 scan the program and collect information about the functions it uses.
12700 So neither of these programs will work until GCC is installed.
12702 Here is a table of the options you can use with @code{protoize} and
12703 @code{unprotoize}. Each option works with both programs unless
12707 @item -B @var{directory}
12708 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
12709 usual directory (normally @file{/usr/local/lib}). This file contains
12710 prototype information about standard system functions. This option
12711 applies only to @code{protoize}.
12713 @item -c @var{compilation-options}
12714 Use @var{compilation-options} as the options when running @command{gcc} to
12715 produce the @samp{.X} files. The special option @option{-aux-info} is
12716 always passed in addition, to tell @command{gcc} to write a @samp{.X} file.
12718 Note that the compilation options must be given as a single argument to
12719 @code{protoize} or @code{unprotoize}. If you want to specify several
12720 @command{gcc} options, you must quote the entire set of compilation options
12721 to make them a single word in the shell.
12723 There are certain @command{gcc} arguments that you cannot use, because they
12724 would produce the wrong kind of output. These include @option{-g},
12725 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
12726 the @var{compilation-options}, they are ignored.
12729 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
12730 systems) instead of @samp{.c}. This is convenient if you are converting
12731 a C program to C++. This option applies only to @code{protoize}.
12734 Add explicit global declarations. This means inserting explicit
12735 declarations at the beginning of each source file for each function
12736 that is called in the file and was not declared. These declarations
12737 precede the first function definition that contains a call to an
12738 undeclared function. This option applies only to @code{protoize}.
12740 @item -i @var{string}
12741 Indent old-style parameter declarations with the string @var{string}.
12742 This option applies only to @code{protoize}.
12744 @code{unprotoize} converts prototyped function definitions to old-style
12745 function definitions, where the arguments are declared between the
12746 argument list and the initial @samp{@{}. By default, @code{unprotoize}
12747 uses five spaces as the indentation. If you want to indent with just
12748 one space instead, use @option{-i " "}.
12751 Keep the @samp{.X} files. Normally, they are deleted after conversion
12755 Add explicit local declarations. @code{protoize} with @option{-l} inserts
12756 a prototype declaration for each function in each block which calls the
12757 function without any declaration. This option applies only to
12761 Make no real changes. This mode just prints information about the conversions
12762 that would have been done without @option{-n}.
12765 Make no @samp{.save} files. The original files are simply deleted.
12766 Use this option with caution.
12768 @item -p @var{program}
12769 Use the program @var{program} as the compiler. Normally, the name
12770 @file{gcc} is used.
12773 Work quietly. Most warnings are suppressed.
12776 Print the version number, just like @option{-v} for @command{gcc}.
12779 If you need special compiler options to compile one of your program's
12780 source files, then you should generate that file's @samp{.X} file
12781 specially, by running @command{gcc} on that source file with the
12782 appropriate options and the option @option{-aux-info}. Then run
12783 @code{protoize} on the entire set of files. @code{protoize} will use
12784 the existing @samp{.X} file because it is newer than the source file.
12788 gcc -Dfoo=bar file1.c -aux-info file1.X
12793 You need to include the special files along with the rest in the
12794 @code{protoize} command, even though their @samp{.X} files already
12795 exist, because otherwise they won't get converted.
12797 @xref{Protoize Caveats}, for more information on how to use
12798 @code{protoize} successfully.