1 @c Copyright (C) 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
2 @c 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009
3 @c Free Software Foundation, Inc.
4 @c This is part of the GCC manual.
5 @c For copying conditions, see the file gcc.texi.
12 @c man begin COPYRIGHT
13 Copyright @copyright{} 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998,
14 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009
15 Free Software Foundation, Inc.
17 Permission is granted to copy, distribute and/or modify this document
18 under the terms of the GNU Free Documentation License, Version 1.2 or
19 any later version published by the Free Software Foundation; with the
20 Invariant Sections being ``GNU General Public License'' and ``Funding
21 Free Software'', the Front-Cover texts being (a) (see below), and with
22 the Back-Cover Texts being (b) (see below). A copy of the license is
23 included in the gfdl(7) man page.
25 (a) The FSF's Front-Cover Text is:
29 (b) The FSF's Back-Cover Text is:
31 You have freedom to copy and modify this GNU Manual, like GNU
32 software. Copies published by the Free Software Foundation raise
33 funds for GNU development.
35 @c Set file name and title for the man page.
37 @settitle GNU project C and C++ compiler
39 gcc [@option{-c}|@option{-S}|@option{-E}] [@option{-std=}@var{standard}]
40 [@option{-g}] [@option{-pg}] [@option{-O}@var{level}]
41 [@option{-W}@var{warn}@dots{}] [@option{-pedantic}]
42 [@option{-I}@var{dir}@dots{}] [@option{-L}@var{dir}@dots{}]
43 [@option{-D}@var{macro}[=@var{defn}]@dots{}] [@option{-U}@var{macro}]
44 [@option{-f}@var{option}@dots{}] [@option{-m}@var{machine-option}@dots{}]
45 [@option{-o} @var{outfile}] [@@@var{file}] @var{infile}@dots{}
47 Only the most useful options are listed here; see below for the
48 remainder. @samp{g++} accepts mostly the same options as @samp{gcc}.
51 gpl(7), gfdl(7), fsf-funding(7),
52 cpp(1), gcov(1), as(1), ld(1), gdb(1), adb(1), dbx(1), sdb(1)
53 and the Info entries for @file{gcc}, @file{cpp}, @file{as},
54 @file{ld}, @file{binutils} and @file{gdb}.
57 For instructions on reporting bugs, see
61 See the Info entry for @command{gcc}, or
62 @w{@uref{http://gcc.gnu.org/onlinedocs/gcc/Contributors.html}},
63 for contributors to GCC@.
68 @chapter GCC Command Options
69 @cindex GCC command options
70 @cindex command options
71 @cindex options, GCC command
73 @c man begin DESCRIPTION
74 When you invoke GCC, it normally does preprocessing, compilation,
75 assembly and linking. The ``overall options'' allow you to stop this
76 process at an intermediate stage. For example, the @option{-c} option
77 says not to run the linker. Then the output consists of object files
78 output by the assembler.
80 Other options are passed on to one stage of processing. Some options
81 control the preprocessor and others the compiler itself. Yet other
82 options control the assembler and linker; most of these are not
83 documented here, since you rarely need to use any of them.
85 @cindex C compilation options
86 Most of the command line options that you can use with GCC are useful
87 for C programs; when an option is only useful with another language
88 (usually C++), the explanation says so explicitly. If the description
89 for a particular option does not mention a source language, you can use
90 that option with all supported languages.
92 @cindex C++ compilation options
93 @xref{Invoking G++,,Compiling C++ Programs}, for a summary of special
94 options for compiling C++ programs.
96 @cindex grouping options
97 @cindex options, grouping
98 The @command{gcc} program accepts options and file names as operands. Many
99 options have multi-letter names; therefore multiple single-letter options
100 may @emph{not} be grouped: @option{-dv} is very different from @w{@samp{-d
103 @cindex order of options
104 @cindex options, order
105 You can mix options and other arguments. For the most part, the order
106 you use doesn't matter. Order does matter when you use several
107 options of the same kind; for example, if you specify @option{-L} more
108 than once, the directories are searched in the order specified. Also,
109 the placement of the @option{-l} option is significant.
111 Many options have long names starting with @samp{-f} or with
112 @samp{-W}---for example,
113 @option{-fmove-loop-invariants}, @option{-Wformat} and so on. Most of
114 these have both positive and negative forms; the negative form of
115 @option{-ffoo} would be @option{-fno-foo}. This manual documents
116 only one of these two forms, whichever one is not the default.
120 @xref{Option Index}, for an index to GCC's options.
123 * Option Summary:: Brief list of all options, without explanations.
124 * Overall Options:: Controlling the kind of output:
125 an executable, object files, assembler files,
126 or preprocessed source.
127 * Invoking G++:: Compiling C++ programs.
128 * C Dialect Options:: Controlling the variant of C language compiled.
129 * C++ Dialect Options:: Variations on C++.
130 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
132 * Language Independent Options:: Controlling how diagnostics should be
134 * Warning Options:: How picky should the compiler be?
135 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
136 * Optimize Options:: How much optimization?
137 * Preprocessor Options:: Controlling header files and macro definitions.
138 Also, getting dependency information for Make.
139 * Assembler Options:: Passing options to the assembler.
140 * Link Options:: Specifying libraries and so on.
141 * Directory Options:: Where to find header files and libraries.
142 Where to find the compiler executable files.
143 * Spec Files:: How to pass switches to sub-processes.
144 * Target Options:: Running a cross-compiler, or an old version of GCC.
145 * Submodel Options:: Specifying minor hardware or convention variations,
146 such as 68010 vs 68020.
147 * Code Gen Options:: Specifying conventions for function calls, data layout
149 * Environment Variables:: Env vars that affect GCC.
150 * Precompiled Headers:: Compiling a header once, and using it many times.
156 @section Option Summary
158 Here is a summary of all the options, grouped by type. Explanations are
159 in the following sections.
162 @item Overall Options
163 @xref{Overall Options,,Options Controlling the Kind of Output}.
164 @gccoptlist{-c -S -E -o @var{file} -combine -no-canonical-prefixes @gol
165 -pipe -pass-exit-codes @gol
166 -x @var{language} -v -### --help@r{[}=@var{class}@r{[},@dots{}@r{]]} --target-help @gol
167 --version -wrapper@@@var{file} -fplugin=@var{file} -fplugin-arg-@var{name}=@var{arg}}
169 @item C Language Options
170 @xref{C Dialect Options,,Options Controlling C Dialect}.
171 @gccoptlist{-ansi -std=@var{standard} -fgnu89-inline @gol
172 -aux-info @var{filename} @gol
173 -fno-asm -fno-builtin -fno-builtin-@var{function} @gol
174 -fhosted -ffreestanding -fopenmp -fms-extensions @gol
175 -trigraphs -no-integrated-cpp -traditional -traditional-cpp @gol
176 -fallow-single-precision -fcond-mismatch -flax-vector-conversions @gol
177 -fsigned-bitfields -fsigned-char @gol
178 -funsigned-bitfields -funsigned-char}
180 @item C++ Language Options
181 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
182 @gccoptlist{-fabi-version=@var{n} -fno-access-control -fcheck-new @gol
183 -fconserve-space -ffriend-injection @gol
184 -fno-elide-constructors @gol
185 -fno-enforce-eh-specs @gol
186 -ffor-scope -fno-for-scope -fno-gnu-keywords @gol
187 -fno-implicit-templates @gol
188 -fno-implicit-inline-templates @gol
189 -fno-implement-inlines -fms-extensions @gol
190 -fno-nonansi-builtins -fno-operator-names @gol
191 -fno-optional-diags -fpermissive @gol
192 -fno-pretty-templates @gol
193 -frepo -fno-rtti -fstats -ftemplate-depth-@var{n} @gol
194 -fno-threadsafe-statics -fuse-cxa-atexit -fno-weak -nostdinc++ @gol
195 -fno-default-inline -fvisibility-inlines-hidden @gol
196 -fvisibility-ms-compat @gol
197 -Wabi -Wctor-dtor-privacy @gol
198 -Wnon-virtual-dtor -Wreorder @gol
199 -Weffc++ -Wstrict-null-sentinel @gol
200 -Wno-non-template-friend -Wold-style-cast @gol
201 -Woverloaded-virtual -Wno-pmf-conversions @gol
204 @item Objective-C and Objective-C++ Language Options
205 @xref{Objective-C and Objective-C++ Dialect Options,,Options Controlling
206 Objective-C and Objective-C++ Dialects}.
207 @gccoptlist{-fconstant-string-class=@var{class-name} @gol
208 -fgnu-runtime -fnext-runtime @gol
209 -fno-nil-receivers @gol
210 -fobjc-call-cxx-cdtors @gol
211 -fobjc-direct-dispatch @gol
212 -fobjc-exceptions @gol
214 -freplace-objc-classes @gol
217 -Wassign-intercept @gol
218 -Wno-protocol -Wselector @gol
219 -Wstrict-selector-match @gol
220 -Wundeclared-selector}
222 @item Language Independent Options
223 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
224 @gccoptlist{-fmessage-length=@var{n} @gol
225 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]} @gol
226 -fdiagnostics-show-option}
228 @item Warning Options
229 @xref{Warning Options,,Options to Request or Suppress Warnings}.
230 @gccoptlist{-fsyntax-only -pedantic -pedantic-errors @gol
231 -w -Wextra -Wall -Waddress -Waggregate-return -Warray-bounds @gol
232 -Wno-attributes -Wno-builtin-macro-redefined @gol
233 -Wc++-compat -Wc++0x-compat -Wcast-align -Wcast-qual @gol
234 -Wchar-subscripts -Wclobbered -Wcomment @gol
235 -Wconversion -Wcoverage-mismatch -Wno-deprecated @gol
236 -Wno-deprecated-declarations -Wdisabled-optimization @gol
237 -Wno-div-by-zero -Wempty-body -Wenum-compare -Wno-endif-labels @gol
238 -Werror -Werror=* @gol
239 -Wfatal-errors -Wfloat-equal -Wformat -Wformat=2 @gol
240 -Wno-format-contains-nul -Wno-format-extra-args -Wformat-nonliteral @gol
241 -Wformat-security -Wformat-y2k @gol
242 -Wframe-larger-than=@var{len} -Wjump-misses-init -Wignored-qualifiers @gol
243 -Wimplicit -Wimplicit-function-declaration -Wimplicit-int @gol
244 -Winit-self -Winline @gol
245 -Wno-int-to-pointer-cast -Wno-invalid-offsetof @gol
246 -Winvalid-pch -Wlarger-than=@var{len} -Wunsafe-loop-optimizations @gol
247 -Wlogical-op -Wlong-long @gol
248 -Wmain -Wmissing-braces -Wmissing-field-initializers @gol
249 -Wmissing-format-attribute -Wmissing-include-dirs @gol
250 -Wmissing-noreturn -Wno-mudflap @gol
251 -Wno-multichar -Wnonnull -Wno-overflow @gol
252 -Woverlength-strings -Wpacked -Wpacked-bitfield-compat -Wpadded @gol
253 -Wparentheses -Wpedantic-ms-format -Wno-pedantic-ms-format @gol
254 -Wpointer-arith -Wno-pointer-to-int-cast @gol
255 -Wredundant-decls @gol
256 -Wreturn-type -Wsequence-point -Wshadow @gol
257 -Wsign-compare -Wsign-conversion -Wstack-protector @gol
258 -Wstrict-aliasing -Wstrict-aliasing=n @gol
259 -Wstrict-overflow -Wstrict-overflow=@var{n} @gol
260 -Wswitch -Wswitch-default -Wswitch-enum -Wsync-nand @gol
261 -Wsystem-headers -Wtrigraphs -Wtype-limits -Wundef -Wuninitialized @gol
262 -Wunknown-pragmas -Wno-pragmas -Wunreachable-code @gol
263 -Wunsuffixed-float-constants -Wunused -Wunused-function @gol
264 -Wunused-label -Wunused-parameter -Wno-unused-result -Wunused-value -Wunused-variable @gol
265 -Wvariadic-macros -Wvla @gol
266 -Wvolatile-register-var -Wwrite-strings}
268 @item C and Objective-C-only Warning Options
269 @gccoptlist{-Wbad-function-cast -Wmissing-declarations @gol
270 -Wmissing-parameter-type -Wmissing-prototypes -Wnested-externs @gol
271 -Wold-style-declaration -Wold-style-definition @gol
272 -Wstrict-prototypes -Wtraditional -Wtraditional-conversion @gol
273 -Wdeclaration-after-statement -Wpointer-sign}
275 @item Debugging Options
276 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
277 @gccoptlist{-d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
278 -fdbg-cnt-list -fdbg-cnt=@var{counter-value-list} @gol
279 -fdump-noaddr -fdump-unnumbered -fdump-unnumbered-links @gol
280 -fdump-translation-unit@r{[}-@var{n}@r{]} @gol
281 -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
282 -fdump-ipa-all -fdump-ipa-cgraph -fdump-ipa-inline @gol
283 -fdump-statistics @gol
285 -fdump-tree-original@r{[}-@var{n}@r{]} @gol
286 -fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
287 -fdump-tree-cfg -fdump-tree-vcg -fdump-tree-alias @gol
289 -fdump-tree-ssa@r{[}-@var{n}@r{]} -fdump-tree-pre@r{[}-@var{n}@r{]} @gol
290 -fdump-tree-ccp@r{[}-@var{n}@r{]} -fdump-tree-dce@r{[}-@var{n}@r{]} @gol
291 -fdump-tree-gimple@r{[}-raw@r{]} -fdump-tree-mudflap@r{[}-@var{n}@r{]} @gol
292 -fdump-tree-dom@r{[}-@var{n}@r{]} @gol
293 -fdump-tree-dse@r{[}-@var{n}@r{]} @gol
294 -fdump-tree-phiprop@r{[}-@var{n}@r{]} @gol
295 -fdump-tree-phiopt@r{[}-@var{n}@r{]} @gol
296 -fdump-tree-forwprop@r{[}-@var{n}@r{]} @gol
297 -fdump-tree-copyrename@r{[}-@var{n}@r{]} @gol
298 -fdump-tree-nrv -fdump-tree-vect @gol
299 -fdump-tree-sink @gol
300 -fdump-tree-sra@r{[}-@var{n}@r{]} @gol
301 -fdump-tree-forwprop@r{[}-@var{n}@r{]} @gol
302 -fdump-tree-fre@r{[}-@var{n}@r{]} @gol
303 -fdump-tree-vrp@r{[}-@var{n}@r{]} @gol
304 -ftree-vectorizer-verbose=@var{n} @gol
305 -fdump-tree-storeccp@r{[}-@var{n}@r{]} @gol
306 -fdump-final-insns=@var{file} @gol
307 -fcompare-debug@r{[}=@var{opts}@r{]} -fcompare-debug-second @gol
308 -feliminate-dwarf2-dups -feliminate-unused-debug-types @gol
309 -feliminate-unused-debug-symbols -femit-class-debug-always @gol
310 -fmem-report -fpre-ipa-mem-report -fpost-ipa-mem-report -fprofile-arcs @gol
311 -frandom-seed=@var{string} -fsched-verbose=@var{n} @gol
312 -fsel-sched-verbose -fsel-sched-dump-cfg -fsel-sched-pipelining-verbose @gol
313 -ftest-coverage -ftime-report -fvar-tracking @gol
314 -fvar-tracking-assigments -fvar-tracking-assignments-toggle @gol
315 -g -g@var{level} -gtoggle -gcoff -gdwarf-@var{version} @gol
316 -ggdb -gstabs -gstabs+ -gvms -gxcoff -gxcoff+ @gol
317 -fno-merge-debug-strings -fno-dwarf2-cfi-asm @gol
318 -fdebug-prefix-map=@var{old}=@var{new} @gol
319 -femit-struct-debug-baseonly -femit-struct-debug-reduced @gol
320 -femit-struct-debug-detailed@r{[}=@var{spec-list}@r{]} @gol
321 -p -pg -print-file-name=@var{library} -print-libgcc-file-name @gol
322 -print-multi-directory -print-multi-lib @gol
323 -print-prog-name=@var{program} -print-search-dirs -Q @gol
324 -print-sysroot -print-sysroot-headers-suffix @gol
325 -save-temps -save-temps=cwd -save-temps=obj -time@r{[}=@var{file}@r{]}}
327 @item Optimization Options
328 @xref{Optimize Options,,Options that Control Optimization}.
330 -falign-functions[=@var{n}] -falign-jumps[=@var{n}] @gol
331 -falign-labels[=@var{n}] -falign-loops[=@var{n}] -fassociative-math @gol
332 -fauto-inc-dec -fbranch-probabilities -fbranch-target-load-optimize @gol
333 -fbranch-target-load-optimize2 -fbtr-bb-exclusive -fcaller-saves @gol
334 -fcheck-data-deps -fconserve-stack -fcprop-registers -fcrossjumping @gol
335 -fcse-follow-jumps -fcse-skip-blocks -fcx-fortran-rules -fcx-limited-range @gol
336 -fdata-sections -fdce -fdce @gol
337 -fdelayed-branch -fdelete-null-pointer-checks -fdse -fdse @gol
338 -fearly-inlining -fipa-sra -fexpensive-optimizations -ffast-math @gol
339 -ffinite-math-only -ffloat-store -fexcess-precision=@var{style} @gol
340 -fforward-propagate -ffunction-sections @gol
341 -fgcse -fgcse-after-reload -fgcse-las -fgcse-lm @gol
342 -fgcse-sm -fif-conversion -fif-conversion2 -findirect-inlining @gol
343 -finline-functions -finline-functions-called-once -finline-limit=@var{n} @gol
344 -finline-small-functions -fipa-cp -fipa-cp-clone -fipa-matrix-reorg -fipa-pta @gol
345 -fipa-pure-const -fipa-reference -fipa-struct-reorg @gol
346 -fipa-type-escape -fira-algorithm=@var{algorithm} @gol
347 -fira-region=@var{region} -fira-coalesce -fno-ira-share-save-slots @gol
348 -fno-ira-share-spill-slots -fira-verbose=@var{n} @gol
349 -fivopts -fkeep-inline-functions -fkeep-static-consts @gol
350 -floop-block -floop-interchange -floop-strip-mine -fgraphite-identity @gol
351 -floop-parallelize-all @gol
352 -fmerge-all-constants -fmerge-constants -fmodulo-sched @gol
353 -fmodulo-sched-allow-regmoves -fmove-loop-invariants -fmudflap @gol
354 -fmudflapir -fmudflapth -fno-branch-count-reg -fno-default-inline @gol
355 -fno-defer-pop -fno-function-cse -fno-guess-branch-probability @gol
356 -fno-inline -fno-math-errno -fno-peephole -fno-peephole2 @gol
357 -fno-sched-interblock -fno-sched-spec -fno-signed-zeros @gol
358 -fno-toplevel-reorder -fno-trapping-math -fno-zero-initialized-in-bss @gol
359 -fomit-frame-pointer -foptimize-register-move -foptimize-sibling-calls @gol
360 -fpeel-loops -fpredictive-commoning -fprefetch-loop-arrays @gol
361 -fprofile-correction -fprofile-dir=@var{path} -fprofile-generate @gol
362 -fprofile-generate=@var{path} @gol
363 -fprofile-use -fprofile-use=@var{path} -fprofile-values @gol
364 -freciprocal-math -fregmove -frename-registers -freorder-blocks @gol
365 -freorder-blocks-and-partition -freorder-functions @gol
366 -frerun-cse-after-loop -freschedule-modulo-scheduled-loops @gol
367 -frounding-math -fsched2-use-superblocks @gol
368 -fsched2-use-traces -fsched-pressure @gol
369 -fsched-spec-load -fsched-spec-load-dangerous @gol
370 -fsched-stalled-insns-dep[=@var{n}] -fsched-stalled-insns[=@var{n}] @gol
371 -fsched-group-heuristic -fsched-critical-path-heuristic @gol
372 -fsched-spec-insn-heuristic -fsched-rank-heuristic @gol
373 -fsched-last-insn-heuristic -fsched-dep-count-heuristic @gol
374 -fschedule-insns -fschedule-insns2 -fsection-anchors @gol
375 -fselective-scheduling -fselective-scheduling2 @gol
376 -fsel-sched-pipelining -fsel-sched-pipelining-outer-loops @gol
377 -fsignaling-nans -fsingle-precision-constant -fsplit-ivs-in-unroller @gol
378 -fsplit-wide-types -fstack-protector -fstack-protector-all @gol
379 -fstrict-aliasing -fstrict-overflow -fthread-jumps -ftracer @gol
380 -ftree-builtin-call-dce -ftree-ccp -ftree-ch -ftree-copy-prop @gol
381 -ftree-copyrename -ftree-dce @gol
382 -ftree-dominator-opts -ftree-dse -ftree-forwprop -ftree-fre -ftree-loop-im @gol
383 -ftree-phiprop -ftree-loop-distribution @gol
384 -ftree-loop-ivcanon -ftree-loop-linear -ftree-loop-optimize @gol
385 -ftree-parallelize-loops=@var{n} -ftree-pre -ftree-pta -ftree-reassoc @gol
386 -ftree-sink -ftree-sra -ftree-switch-conversion @gol
387 -ftree-ter -ftree-vect-loop-version -ftree-vectorize -ftree-vrp @gol
388 -funit-at-a-time -funroll-all-loops -funroll-loops @gol
389 -funsafe-loop-optimizations -funsafe-math-optimizations -funswitch-loops @gol
390 -fvariable-expansion-in-unroller -fvect-cost-model -fvpt -fweb @gol
392 --param @var{name}=@var{value}
393 -O -O0 -O1 -O2 -O3 -Os}
395 @item Preprocessor Options
396 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
397 @gccoptlist{-A@var{question}=@var{answer} @gol
398 -A-@var{question}@r{[}=@var{answer}@r{]} @gol
399 -C -dD -dI -dM -dN @gol
400 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
401 -idirafter @var{dir} @gol
402 -include @var{file} -imacros @var{file} @gol
403 -iprefix @var{file} -iwithprefix @var{dir} @gol
404 -iwithprefixbefore @var{dir} -isystem @var{dir} @gol
405 -imultilib @var{dir} -isysroot @var{dir} @gol
406 -M -MM -MF -MG -MP -MQ -MT -nostdinc @gol
407 -P -fworking-directory -remap @gol
408 -trigraphs -undef -U@var{macro} -Wp,@var{option} @gol
409 -Xpreprocessor @var{option}}
411 @item Assembler Option
412 @xref{Assembler Options,,Passing Options to the Assembler}.
413 @gccoptlist{-Wa,@var{option} -Xassembler @var{option}}
416 @xref{Link Options,,Options for Linking}.
417 @gccoptlist{@var{object-file-name} -l@var{library} @gol
418 -nostartfiles -nodefaultlibs -nostdlib -pie -rdynamic @gol
419 -s -static -static-libgcc -static-libstdc++ -shared @gol
420 -shared-libgcc -symbolic @gol
421 -T @var{script} -Wl,@var{option} -Xlinker @var{option} @gol
424 @item Directory Options
425 @xref{Directory Options,,Options for Directory Search}.
426 @gccoptlist{-B@var{prefix} -I@var{dir} -iquote@var{dir} -L@var{dir}
427 -specs=@var{file} -I- --sysroot=@var{dir}}
430 @c I wrote this xref this way to avoid overfull hbox. -- rms
431 @xref{Target Options}.
432 @gccoptlist{-V @var{version} -b @var{machine}}
434 @item Machine Dependent Options
435 @xref{Submodel Options,,Hardware Models and Configurations}.
436 @c This list is ordered alphanumerically by subsection name.
437 @c Try and put the significant identifier (CPU or system) first,
438 @c so users have a clue at guessing where the ones they want will be.
441 @gccoptlist{-EB -EL @gol
442 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
443 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
446 @gccoptlist{-mapcs-frame -mno-apcs-frame @gol
447 -mabi=@var{name} @gol
448 -mapcs-stack-check -mno-apcs-stack-check @gol
449 -mapcs-float -mno-apcs-float @gol
450 -mapcs-reentrant -mno-apcs-reentrant @gol
451 -msched-prolog -mno-sched-prolog @gol
452 -mlittle-endian -mbig-endian -mwords-little-endian @gol
453 -mfloat-abi=@var{name} -msoft-float -mhard-float -mfpe @gol
454 -mfp16-format=@var{name}
455 -mthumb-interwork -mno-thumb-interwork @gol
456 -mcpu=@var{name} -march=@var{name} -mfpu=@var{name} @gol
457 -mstructure-size-boundary=@var{n} @gol
458 -mabort-on-noreturn @gol
459 -mlong-calls -mno-long-calls @gol
460 -msingle-pic-base -mno-single-pic-base @gol
461 -mpic-register=@var{reg} @gol
462 -mnop-fun-dllimport @gol
463 -mcirrus-fix-invalid-insns -mno-cirrus-fix-invalid-insns @gol
464 -mpoke-function-name @gol
466 -mtpcs-frame -mtpcs-leaf-frame @gol
467 -mcaller-super-interworking -mcallee-super-interworking @gol
469 -mword-relocations @gol
470 -mfix-cortex-m3-ldrd}
473 @gccoptlist{-mmcu=@var{mcu} -msize -mno-interrupts @gol
474 -mcall-prologues -mtiny-stack -mint8}
476 @emph{Blackfin Options}
477 @gccoptlist{-mcpu=@var{cpu}@r{[}-@var{sirevision}@r{]} @gol
478 -msim -momit-leaf-frame-pointer -mno-omit-leaf-frame-pointer @gol
479 -mspecld-anomaly -mno-specld-anomaly -mcsync-anomaly -mno-csync-anomaly @gol
480 -mlow-64k -mno-low64k -mstack-check-l1 -mid-shared-library @gol
481 -mno-id-shared-library -mshared-library-id=@var{n} @gol
482 -mleaf-id-shared-library -mno-leaf-id-shared-library @gol
483 -msep-data -mno-sep-data -mlong-calls -mno-long-calls @gol
484 -mfast-fp -minline-plt -mmulticore -mcorea -mcoreb -msdram @gol
488 @gccoptlist{-mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
489 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
490 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
491 -mstack-align -mdata-align -mconst-align @gol
492 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
493 -melf -maout -melinux -mlinux -sim -sim2 @gol
494 -mmul-bug-workaround -mno-mul-bug-workaround}
497 @gccoptlist{-mmac -mpush-args}
499 @emph{Darwin Options}
500 @gccoptlist{-all_load -allowable_client -arch -arch_errors_fatal @gol
501 -arch_only -bind_at_load -bundle -bundle_loader @gol
502 -client_name -compatibility_version -current_version @gol
504 -dependency-file -dylib_file -dylinker_install_name @gol
505 -dynamic -dynamiclib -exported_symbols_list @gol
506 -filelist -flat_namespace -force_cpusubtype_ALL @gol
507 -force_flat_namespace -headerpad_max_install_names @gol
509 -image_base -init -install_name -keep_private_externs @gol
510 -multi_module -multiply_defined -multiply_defined_unused @gol
511 -noall_load -no_dead_strip_inits_and_terms @gol
512 -nofixprebinding -nomultidefs -noprebind -noseglinkedit @gol
513 -pagezero_size -prebind -prebind_all_twolevel_modules @gol
514 -private_bundle -read_only_relocs -sectalign @gol
515 -sectobjectsymbols -whyload -seg1addr @gol
516 -sectcreate -sectobjectsymbols -sectorder @gol
517 -segaddr -segs_read_only_addr -segs_read_write_addr @gol
518 -seg_addr_table -seg_addr_table_filename -seglinkedit @gol
519 -segprot -segs_read_only_addr -segs_read_write_addr @gol
520 -single_module -static -sub_library -sub_umbrella @gol
521 -twolevel_namespace -umbrella -undefined @gol
522 -unexported_symbols_list -weak_reference_mismatches @gol
523 -whatsloaded -F -gused -gfull -mmacosx-version-min=@var{version} @gol
524 -mkernel -mone-byte-bool}
526 @emph{DEC Alpha Options}
527 @gccoptlist{-mno-fp-regs -msoft-float -malpha-as -mgas @gol
528 -mieee -mieee-with-inexact -mieee-conformant @gol
529 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
530 -mtrap-precision=@var{mode} -mbuild-constants @gol
531 -mcpu=@var{cpu-type} -mtune=@var{cpu-type} @gol
532 -mbwx -mmax -mfix -mcix @gol
533 -mfloat-vax -mfloat-ieee @gol
534 -mexplicit-relocs -msmall-data -mlarge-data @gol
535 -msmall-text -mlarge-text @gol
536 -mmemory-latency=@var{time}}
538 @emph{DEC Alpha/VMS Options}
539 @gccoptlist{-mvms-return-codes -mdebug-main=@var{prefix} -mmalloc64}
542 @gccoptlist{-msmall-model -mno-lsim}
545 @gccoptlist{-mgpr-32 -mgpr-64 -mfpr-32 -mfpr-64 @gol
546 -mhard-float -msoft-float @gol
547 -malloc-cc -mfixed-cc -mdword -mno-dword @gol
548 -mdouble -mno-double @gol
549 -mmedia -mno-media -mmuladd -mno-muladd @gol
550 -mfdpic -minline-plt -mgprel-ro -multilib-library-pic @gol
551 -mlinked-fp -mlong-calls -malign-labels @gol
552 -mlibrary-pic -macc-4 -macc-8 @gol
553 -mpack -mno-pack -mno-eflags -mcond-move -mno-cond-move @gol
554 -moptimize-membar -mno-optimize-membar @gol
555 -mscc -mno-scc -mcond-exec -mno-cond-exec @gol
556 -mvliw-branch -mno-vliw-branch @gol
557 -mmulti-cond-exec -mno-multi-cond-exec -mnested-cond-exec @gol
558 -mno-nested-cond-exec -mtomcat-stats @gol
562 @emph{GNU/Linux Options}
563 @gccoptlist{-muclibc}
565 @emph{H8/300 Options}
566 @gccoptlist{-mrelax -mh -ms -mn -mint32 -malign-300}
569 @gccoptlist{-march=@var{architecture-type} @gol
570 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
571 -mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
572 -mfixed-range=@var{register-range} @gol
573 -mjump-in-delay -mlinker-opt -mlong-calls @gol
574 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
575 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
576 -mno-jump-in-delay -mno-long-load-store @gol
577 -mno-portable-runtime -mno-soft-float @gol
578 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
579 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
580 -mschedule=@var{cpu-type} -mspace-regs -msio -mwsio @gol
581 -munix=@var{unix-std} -nolibdld -static -threads}
583 @emph{i386 and x86-64 Options}
584 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
585 -mfpmath=@var{unit} @gol
586 -masm=@var{dialect} -mno-fancy-math-387 @gol
587 -mno-fp-ret-in-387 -msoft-float @gol
588 -mno-wide-multiply -mrtd -malign-double @gol
589 -mpreferred-stack-boundary=@var{num}
590 -mincoming-stack-boundary=@var{num}
591 -mcld -mcx16 -msahf -mmovbe -mcrc32 -mrecip @gol
592 -mmmx -msse -msse2 -msse3 -mssse3 -msse4.1 -msse4.2 -msse4 -mavx @gol
594 -msse4a -m3dnow -mpopcnt -mabm @gol
595 -mthreads -mno-align-stringops -minline-all-stringops @gol
596 -minline-stringops-dynamically -mstringop-strategy=@var{alg} @gol
597 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
598 -m96bit-long-double -mregparm=@var{num} -msseregparm @gol
599 -mveclibabi=@var{type} -mpc32 -mpc64 -mpc80 -mstackrealign @gol
600 -momit-leaf-frame-pointer -mno-red-zone -mno-tls-direct-seg-refs @gol
601 -mcmodel=@var{code-model} -mabi=@var{name} @gol
602 -m32 -m64 -mlarge-data-threshold=@var{num} @gol
606 @gccoptlist{-mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
607 -mvolatile-asm-stop -mregister-names -msdata -mno-sdata @gol
608 -mconstant-gp -mauto-pic -mfused-madd @gol
609 -minline-float-divide-min-latency @gol
610 -minline-float-divide-max-throughput @gol
611 -mno-inline-float-divide @gol
612 -minline-int-divide-min-latency @gol
613 -minline-int-divide-max-throughput @gol
614 -mno-inline-int-divide @gol
615 -minline-sqrt-min-latency -minline-sqrt-max-throughput @gol
616 -mno-inline-sqrt @gol
617 -mdwarf2-asm -mearly-stop-bits @gol
618 -mfixed-range=@var{register-range} -mtls-size=@var{tls-size} @gol
619 -mtune=@var{cpu-type} -milp32 -mlp64 @gol
620 -msched-br-data-spec -msched-ar-data-spec -msched-control-spec @gol
621 -msched-br-in-data-spec -msched-ar-in-data-spec -msched-in-control-spec @gol
622 -msched-spec-ldc -msched-spec-control-ldc @gol
623 -msched-prefer-non-data-spec-insns -msched-prefer-non-control-spec-insns @gol
624 -msched-stop-bits-after-every-cycle -msched-count-spec-in-critical-path @gol
625 -msel-sched-dont-check-control-spec -msched-fp-mem-deps-zero-cost @gol
626 -msched-max-memory-insns-hard-limit -msched-max-memory-insns=@var{max-insns}}
628 @emph{IA-64/VMS Options}
629 @gccoptlist{-mvms-return-codes -mdebug-main=@var{prefix} -mmalloc64}
631 @emph{M32R/D Options}
632 @gccoptlist{-m32r2 -m32rx -m32r @gol
634 -malign-loops -mno-align-loops @gol
635 -missue-rate=@var{number} @gol
636 -mbranch-cost=@var{number} @gol
637 -mmodel=@var{code-size-model-type} @gol
638 -msdata=@var{sdata-type} @gol
639 -mno-flush-func -mflush-func=@var{name} @gol
640 -mno-flush-trap -mflush-trap=@var{number} @gol
644 @gccoptlist{-mcpu=@var{cpu} -msim -memregs=@var{number}}
646 @emph{M680x0 Options}
647 @gccoptlist{-march=@var{arch} -mcpu=@var{cpu} -mtune=@var{tune}
648 -m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
649 -m68060 -mcpu32 -m5200 -m5206e -m528x -m5307 -m5407 @gol
650 -mcfv4e -mbitfield -mno-bitfield -mc68000 -mc68020 @gol
651 -mnobitfield -mrtd -mno-rtd -mdiv -mno-div -mshort @gol
652 -mno-short -mhard-float -m68881 -msoft-float -mpcrel @gol
653 -malign-int -mstrict-align -msep-data -mno-sep-data @gol
654 -mshared-library-id=n -mid-shared-library -mno-id-shared-library @gol
657 @emph{M68hc1x Options}
658 @gccoptlist{-m6811 -m6812 -m68hc11 -m68hc12 -m68hcs12 @gol
659 -mauto-incdec -minmax -mlong-calls -mshort @gol
660 -msoft-reg-count=@var{count}}
663 @gccoptlist{-mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
664 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
665 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
666 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
667 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
670 @gccoptlist{-mabsdiff -mall-opts -maverage -mbased=@var{n} -mbitops @gol
671 -mc=@var{n} -mclip -mconfig=@var{name} -mcop -mcop32 -mcop64 -mivc2 @gol
672 -mdc -mdiv -meb -mel -mio-volatile -ml -mleadz -mm -mminmax @gol
673 -mmult -mno-opts -mrepeat -ms -msatur -msdram -msim -msimnovec -mtf @gol
677 @gccoptlist{-EL -EB -march=@var{arch} -mtune=@var{arch} @gol
678 -mips1 -mips2 -mips3 -mips4 -mips32 -mips32r2 @gol
679 -mips64 -mips64r2 @gol
680 -mips16 -mno-mips16 -mflip-mips16 @gol
681 -minterlink-mips16 -mno-interlink-mips16 @gol
682 -mabi=@var{abi} -mabicalls -mno-abicalls @gol
683 -mshared -mno-shared -mplt -mno-plt -mxgot -mno-xgot @gol
684 -mgp32 -mgp64 -mfp32 -mfp64 -mhard-float -msoft-float @gol
685 -msingle-float -mdouble-float -mdsp -mno-dsp -mdspr2 -mno-dspr2 @gol
686 -mfpu=@var{fpu-type} @gol
687 -msmartmips -mno-smartmips @gol
688 -mpaired-single -mno-paired-single -mdmx -mno-mdmx @gol
689 -mips3d -mno-mips3d -mmt -mno-mt -mllsc -mno-llsc @gol
690 -mlong64 -mlong32 -msym32 -mno-sym32 @gol
691 -G@var{num} -mlocal-sdata -mno-local-sdata @gol
692 -mextern-sdata -mno-extern-sdata -mgpopt -mno-gopt @gol
693 -membedded-data -mno-embedded-data @gol
694 -muninit-const-in-rodata -mno-uninit-const-in-rodata @gol
695 -mcode-readable=@var{setting} @gol
696 -msplit-addresses -mno-split-addresses @gol
697 -mexplicit-relocs -mno-explicit-relocs @gol
698 -mcheck-zero-division -mno-check-zero-division @gol
699 -mdivide-traps -mdivide-breaks @gol
700 -mmemcpy -mno-memcpy -mlong-calls -mno-long-calls @gol
701 -mmad -mno-mad -mfused-madd -mno-fused-madd -nocpp @gol
702 -mfix-r4000 -mno-fix-r4000 -mfix-r4400 -mno-fix-r4400 @gol
703 -mfix-r10000 -mno-fix-r10000 -mfix-vr4120 -mno-fix-vr4120 @gol
704 -mfix-vr4130 -mno-fix-vr4130 -mfix-sb1 -mno-fix-sb1 @gol
705 -mflush-func=@var{func} -mno-flush-func @gol
706 -mbranch-cost=@var{num} -mbranch-likely -mno-branch-likely @gol
707 -mfp-exceptions -mno-fp-exceptions @gol
708 -mvr4130-align -mno-vr4130-align -msynci -mno-synci @gol
709 -mrelax-pic-calls -mno-relax-pic-calls}
712 @gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
713 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
714 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
715 -mno-base-addresses -msingle-exit -mno-single-exit}
717 @emph{MN10300 Options}
718 @gccoptlist{-mmult-bug -mno-mult-bug @gol
719 -mam33 -mno-am33 @gol
720 -mam33-2 -mno-am33-2 @gol
721 -mreturn-pointer-on-d0 @gol
724 @emph{PDP-11 Options}
725 @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
726 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
727 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
728 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
729 -mbranch-expensive -mbranch-cheap @gol
730 -msplit -mno-split -munix-asm -mdec-asm}
732 @emph{picoChip Options}
733 @gccoptlist{-mae=@var{ae_type} -mvliw-lookahead=@var{N}
734 -msymbol-as-address -mno-inefficient-warnings}
736 @emph{PowerPC Options}
737 See RS/6000 and PowerPC Options.
739 @emph{RS/6000 and PowerPC Options}
740 @gccoptlist{-mcpu=@var{cpu-type} @gol
741 -mtune=@var{cpu-type} @gol
742 -mpower -mno-power -mpower2 -mno-power2 @gol
743 -mpowerpc -mpowerpc64 -mno-powerpc @gol
744 -maltivec -mno-altivec @gol
745 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
746 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
747 -mmfcrf -mno-mfcrf -mpopcntb -mno-popcntb -mpopcntd -mno-popcntd @gol
748 -mfprnd -mno-fprnd @gol
749 -mcmpb -mno-cmpb -mmfpgpr -mno-mfpgpr -mhard-dfp -mno-hard-dfp @gol
750 -mnew-mnemonics -mold-mnemonics @gol
751 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
752 -m64 -m32 -mxl-compat -mno-xl-compat -mpe @gol
753 -malign-power -malign-natural @gol
754 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
755 -msingle-float -mdouble-float -msimple-fpu @gol
756 -mstring -mno-string -mupdate -mno-update @gol
757 -mavoid-indexed-addresses -mno-avoid-indexed-addresses @gol
758 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
759 -mstrict-align -mno-strict-align -mrelocatable @gol
760 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
761 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
762 -mdynamic-no-pic -maltivec -mswdiv @gol
763 -mprioritize-restricted-insns=@var{priority} @gol
764 -msched-costly-dep=@var{dependence_type} @gol
765 -minsert-sched-nops=@var{scheme} @gol
766 -mcall-sysv -mcall-netbsd @gol
767 -maix-struct-return -msvr4-struct-return @gol
768 -mabi=@var{abi-type} -msecure-plt -mbss-plt @gol
769 -misel -mno-isel @gol
770 -misel=yes -misel=no @gol
772 -mspe=yes -mspe=no @gol
774 -mgen-cell-microcode -mwarn-cell-microcode @gol
775 -mvrsave -mno-vrsave @gol
776 -mmulhw -mno-mulhw @gol
777 -mdlmzb -mno-dlmzb @gol
778 -mfloat-gprs=yes -mfloat-gprs=no -mfloat-gprs=single -mfloat-gprs=double @gol
779 -mprototype -mno-prototype @gol
780 -msim -mmvme -mads -myellowknife -memb -msdata @gol
781 -msdata=@var{opt} -mvxworks -G @var{num} -pthread}
783 @emph{S/390 and zSeries Options}
784 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
785 -mhard-float -msoft-float -mhard-dfp -mno-hard-dfp @gol
786 -mlong-double-64 -mlong-double-128 @gol
787 -mbackchain -mno-backchain -mpacked-stack -mno-packed-stack @gol
788 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
789 -m64 -m31 -mdebug -mno-debug -mesa -mzarch @gol
790 -mtpf-trace -mno-tpf-trace -mfused-madd -mno-fused-madd @gol
791 -mwarn-framesize -mwarn-dynamicstack -mstack-size -mstack-guard}
794 @gccoptlist{-meb -mel @gol
798 -mscore5 -mscore5u -mscore7 -mscore7d}
801 @gccoptlist{-m1 -m2 -m2e @gol
802 -m2a-nofpu -m2a-single-only -m2a-single -m2a @gol
804 -m4-nofpu -m4-single-only -m4-single -m4 @gol
805 -m4a-nofpu -m4a-single-only -m4a-single -m4a -m4al @gol
806 -m5-64media -m5-64media-nofpu @gol
807 -m5-32media -m5-32media-nofpu @gol
808 -m5-compact -m5-compact-nofpu @gol
809 -mb -ml -mdalign -mrelax @gol
810 -mbigtable -mfmovd -mhitachi -mrenesas -mno-renesas -mnomacsave @gol
811 -mieee -mbitops -misize -minline-ic_invalidate -mpadstruct -mspace @gol
812 -mprefergot -musermode -multcost=@var{number} -mdiv=@var{strategy} @gol
813 -mdivsi3_libfunc=@var{name} -mfixed-range=@var{register-range} @gol
814 -madjust-unroll -mindexed-addressing -mgettrcost=@var{number} -mpt-fixed @gol
818 @gccoptlist{-mcpu=@var{cpu-type} @gol
819 -mtune=@var{cpu-type} @gol
820 -mcmodel=@var{code-model} @gol
821 -m32 -m64 -mapp-regs -mno-app-regs @gol
822 -mfaster-structs -mno-faster-structs @gol
823 -mfpu -mno-fpu -mhard-float -msoft-float @gol
824 -mhard-quad-float -msoft-quad-float @gol
825 -mimpure-text -mno-impure-text -mlittle-endian @gol
826 -mstack-bias -mno-stack-bias @gol
827 -munaligned-doubles -mno-unaligned-doubles @gol
828 -mv8plus -mno-v8plus -mvis -mno-vis
829 -threads -pthreads -pthread}
832 @gccoptlist{-mwarn-reloc -merror-reloc @gol
833 -msafe-dma -munsafe-dma @gol
835 -msmall-mem -mlarge-mem -mstdmain @gol
836 -mfixed-range=@var{register-range}}
838 @emph{System V Options}
839 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
842 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
843 -mprolog-function -mno-prolog-function -mspace @gol
844 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
845 -mapp-regs -mno-app-regs @gol
846 -mdisable-callt -mno-disable-callt @gol
852 @gccoptlist{-mg -mgnu -munix}
854 @emph{VxWorks Options}
855 @gccoptlist{-mrtp -non-static -Bstatic -Bdynamic @gol
856 -Xbind-lazy -Xbind-now}
858 @emph{x86-64 Options}
859 See i386 and x86-64 Options.
861 @emph{i386 and x86-64 Windows Options}
862 @gccoptlist{-mconsole -mcygwin -mno-cygwin -mdll
863 -mnop-fun-dllimport -mthread -municode -mwin32 -mwindows}
865 @emph{Xstormy16 Options}
868 @emph{Xtensa Options}
869 @gccoptlist{-mconst16 -mno-const16 @gol
870 -mfused-madd -mno-fused-madd @gol
871 -mserialize-volatile -mno-serialize-volatile @gol
872 -mtext-section-literals -mno-text-section-literals @gol
873 -mtarget-align -mno-target-align @gol
874 -mlongcalls -mno-longcalls}
876 @emph{zSeries Options}
877 See S/390 and zSeries Options.
879 @item Code Generation Options
880 @xref{Code Gen Options,,Options for Code Generation Conventions}.
881 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
882 -ffixed-@var{reg} -fexceptions @gol
883 -fnon-call-exceptions -funwind-tables @gol
884 -fasynchronous-unwind-tables @gol
885 -finhibit-size-directive -finstrument-functions @gol
886 -finstrument-functions-exclude-function-list=@var{sym},@var{sym},@dots{} @gol
887 -finstrument-functions-exclude-file-list=@var{file},@var{file},@dots{} @gol
888 -fno-common -fno-ident @gol
889 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
890 -fno-jump-tables @gol
891 -frecord-gcc-switches @gol
892 -freg-struct-return -fshort-enums @gol
893 -fshort-double -fshort-wchar @gol
894 -fverbose-asm -fpack-struct[=@var{n}] -fstack-check @gol
895 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
896 -fno-stack-limit -fargument-alias -fargument-noalias @gol
897 -fargument-noalias-global -fargument-noalias-anything @gol
898 -fleading-underscore -ftls-model=@var{model} @gol
899 -ftrapv -fwrapv -fbounds-check @gol
904 * Overall Options:: Controlling the kind of output:
905 an executable, object files, assembler files,
906 or preprocessed source.
907 * C Dialect Options:: Controlling the variant of C language compiled.
908 * C++ Dialect Options:: Variations on C++.
909 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
911 * Language Independent Options:: Controlling how diagnostics should be
913 * Warning Options:: How picky should the compiler be?
914 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
915 * Optimize Options:: How much optimization?
916 * Preprocessor Options:: Controlling header files and macro definitions.
917 Also, getting dependency information for Make.
918 * Assembler Options:: Passing options to the assembler.
919 * Link Options:: Specifying libraries and so on.
920 * Directory Options:: Where to find header files and libraries.
921 Where to find the compiler executable files.
922 * Spec Files:: How to pass switches to sub-processes.
923 * Target Options:: Running a cross-compiler, or an old version of GCC.
926 @node Overall Options
927 @section Options Controlling the Kind of Output
929 Compilation can involve up to four stages: preprocessing, compilation
930 proper, assembly and linking, always in that order. GCC is capable of
931 preprocessing and compiling several files either into several
932 assembler input files, or into one assembler input file; then each
933 assembler input file produces an object file, and linking combines all
934 the object files (those newly compiled, and those specified as input)
935 into an executable file.
937 @cindex file name suffix
938 For any given input file, the file name suffix determines what kind of
943 C source code which must be preprocessed.
946 C source code which should not be preprocessed.
949 C++ source code which should not be preprocessed.
952 Objective-C source code. Note that you must link with the @file{libobjc}
953 library to make an Objective-C program work.
956 Objective-C source code which should not be preprocessed.
960 Objective-C++ source code. Note that you must link with the @file{libobjc}
961 library to make an Objective-C++ program work. Note that @samp{.M} refers
962 to a literal capital M@.
965 Objective-C++ source code which should not be preprocessed.
968 C, C++, Objective-C or Objective-C++ header file to be turned into a
973 @itemx @var{file}.cxx
974 @itemx @var{file}.cpp
975 @itemx @var{file}.CPP
976 @itemx @var{file}.c++
978 C++ source code which must be preprocessed. Note that in @samp{.cxx},
979 the last two letters must both be literally @samp{x}. Likewise,
980 @samp{.C} refers to a literal capital C@.
984 Objective-C++ source code which must be preprocessed.
987 Objective-C++ source code which should not be preprocessed.
992 @itemx @var{file}.hxx
993 @itemx @var{file}.hpp
994 @itemx @var{file}.HPP
995 @itemx @var{file}.h++
996 @itemx @var{file}.tcc
997 C++ header file to be turned into a precompiled header.
1000 @itemx @var{file}.for
1001 @itemx @var{file}.ftn
1002 Fixed form Fortran source code which should not be preprocessed.
1005 @itemx @var{file}.FOR
1006 @itemx @var{file}.fpp
1007 @itemx @var{file}.FPP
1008 @itemx @var{file}.FTN
1009 Fixed form Fortran source code which must be preprocessed (with the traditional
1012 @item @var{file}.f90
1013 @itemx @var{file}.f95
1014 @itemx @var{file}.f03
1015 @itemx @var{file}.f08
1016 Free form Fortran source code which should not be preprocessed.
1018 @item @var{file}.F90
1019 @itemx @var{file}.F95
1020 @itemx @var{file}.F03
1021 @itemx @var{file}.F08
1022 Free form Fortran source code which must be preprocessed (with the
1023 traditional preprocessor).
1025 @c FIXME: Descriptions of Java file types.
1031 @item @var{file}.ads
1032 Ada source code file which contains a library unit declaration (a
1033 declaration of a package, subprogram, or generic, or a generic
1034 instantiation), or a library unit renaming declaration (a package,
1035 generic, or subprogram renaming declaration). Such files are also
1038 @item @var{file}.adb
1039 Ada source code file containing a library unit body (a subprogram or
1040 package body). Such files are also called @dfn{bodies}.
1042 @c GCC also knows about some suffixes for languages not yet included:
1053 @itemx @var{file}.sx
1054 Assembler code which must be preprocessed.
1057 An object file to be fed straight into linking.
1058 Any file name with no recognized suffix is treated this way.
1062 You can specify the input language explicitly with the @option{-x} option:
1065 @item -x @var{language}
1066 Specify explicitly the @var{language} for the following input files
1067 (rather than letting the compiler choose a default based on the file
1068 name suffix). This option applies to all following input files until
1069 the next @option{-x} option. Possible values for @var{language} are:
1071 c c-header c-cpp-output
1072 c++ c++-header c++-cpp-output
1073 objective-c objective-c-header objective-c-cpp-output
1074 objective-c++ objective-c++-header objective-c++-cpp-output
1075 assembler assembler-with-cpp
1077 f77 f77-cpp-input f95 f95-cpp-input
1082 Turn off any specification of a language, so that subsequent files are
1083 handled according to their file name suffixes (as they are if @option{-x}
1084 has not been used at all).
1086 @item -pass-exit-codes
1087 @opindex pass-exit-codes
1088 Normally the @command{gcc} program will exit with the code of 1 if any
1089 phase of the compiler returns a non-success return code. If you specify
1090 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
1091 numerically highest error produced by any phase that returned an error
1092 indication. The C, C++, and Fortran frontends return 4, if an internal
1093 compiler error is encountered.
1096 If you only want some of the stages of compilation, you can use
1097 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
1098 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
1099 @command{gcc} is to stop. Note that some combinations (for example,
1100 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
1105 Compile or assemble the source files, but do not link. The linking
1106 stage simply is not done. The ultimate output is in the form of an
1107 object file for each source file.
1109 By default, the object file name for a source file is made by replacing
1110 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
1112 Unrecognized input files, not requiring compilation or assembly, are
1117 Stop after the stage of compilation proper; do not assemble. The output
1118 is in the form of an assembler code file for each non-assembler input
1121 By default, the assembler file name for a source file is made by
1122 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
1124 Input files that don't require compilation are ignored.
1128 Stop after the preprocessing stage; do not run the compiler proper. The
1129 output is in the form of preprocessed source code, which is sent to the
1132 Input files which don't require preprocessing are ignored.
1134 @cindex output file option
1137 Place output in file @var{file}. This applies regardless to whatever
1138 sort of output is being produced, whether it be an executable file,
1139 an object file, an assembler file or preprocessed C code.
1141 If @option{-o} is not specified, the default is to put an executable
1142 file in @file{a.out}, the object file for
1143 @file{@var{source}.@var{suffix}} in @file{@var{source}.o}, its
1144 assembler file in @file{@var{source}.s}, a precompiled header file in
1145 @file{@var{source}.@var{suffix}.gch}, and all preprocessed C source on
1150 Print (on standard error output) the commands executed to run the stages
1151 of compilation. Also print the version number of the compiler driver
1152 program and of the preprocessor and the compiler proper.
1156 Like @option{-v} except the commands are not executed and all command
1157 arguments are quoted. This is useful for shell scripts to capture the
1158 driver-generated command lines.
1162 Use pipes rather than temporary files for communication between the
1163 various stages of compilation. This fails to work on some systems where
1164 the assembler is unable to read from a pipe; but the GNU assembler has
1169 If you are compiling multiple source files, this option tells the driver
1170 to pass all the source files to the compiler at once (for those
1171 languages for which the compiler can handle this). This will allow
1172 intermodule analysis (IMA) to be performed by the compiler. Currently the only
1173 language for which this is supported is C@. If you pass source files for
1174 multiple languages to the driver, using this option, the driver will invoke
1175 the compiler(s) that support IMA once each, passing each compiler all the
1176 source files appropriate for it. For those languages that do not support
1177 IMA this option will be ignored, and the compiler will be invoked once for
1178 each source file in that language. If you use this option in conjunction
1179 with @option{-save-temps}, the compiler will generate multiple
1181 (one for each source file), but only one (combined) @file{.o} or
1186 Print (on the standard output) a description of the command line options
1187 understood by @command{gcc}. If the @option{-v} option is also specified
1188 then @option{--help} will also be passed on to the various processes
1189 invoked by @command{gcc}, so that they can display the command line options
1190 they accept. If the @option{-Wextra} option has also been specified
1191 (prior to the @option{--help} option), then command line options which
1192 have no documentation associated with them will also be displayed.
1195 @opindex target-help
1196 Print (on the standard output) a description of target-specific command
1197 line options for each tool. For some targets extra target-specific
1198 information may also be printed.
1200 @item --help=@{@var{class}@r{|[}^@r{]}@var{qualifier}@}@r{[},@dots{}@r{]}
1201 Print (on the standard output) a description of the command line
1202 options understood by the compiler that fit into all specified classes
1203 and qualifiers. These are the supported classes:
1206 @item @samp{optimizers}
1207 This will display all of the optimization options supported by the
1210 @item @samp{warnings}
1211 This will display all of the options controlling warning messages
1212 produced by the compiler.
1215 This will display target-specific options. Unlike the
1216 @option{--target-help} option however, target-specific options of the
1217 linker and assembler will not be displayed. This is because those
1218 tools do not currently support the extended @option{--help=} syntax.
1221 This will display the values recognized by the @option{--param}
1224 @item @var{language}
1225 This will display the options supported for @var{language}, where
1226 @var{language} is the name of one of the languages supported in this
1230 This will display the options that are common to all languages.
1233 These are the supported qualifiers:
1236 @item @samp{undocumented}
1237 Display only those options which are undocumented.
1240 Display options which take an argument that appears after an equal
1241 sign in the same continuous piece of text, such as:
1242 @samp{--help=target}.
1244 @item @samp{separate}
1245 Display options which take an argument that appears as a separate word
1246 following the original option, such as: @samp{-o output-file}.
1249 Thus for example to display all the undocumented target-specific
1250 switches supported by the compiler the following can be used:
1253 --help=target,undocumented
1256 The sense of a qualifier can be inverted by prefixing it with the
1257 @samp{^} character, so for example to display all binary warning
1258 options (i.e., ones that are either on or off and that do not take an
1259 argument), which have a description the following can be used:
1262 --help=warnings,^joined,^undocumented
1265 The argument to @option{--help=} should not consist solely of inverted
1268 Combining several classes is possible, although this usually
1269 restricts the output by so much that there is nothing to display. One
1270 case where it does work however is when one of the classes is
1271 @var{target}. So for example to display all the target-specific
1272 optimization options the following can be used:
1275 --help=target,optimizers
1278 The @option{--help=} option can be repeated on the command line. Each
1279 successive use will display its requested class of options, skipping
1280 those that have already been displayed.
1282 If the @option{-Q} option appears on the command line before the
1283 @option{--help=} option, then the descriptive text displayed by
1284 @option{--help=} is changed. Instead of describing the displayed
1285 options, an indication is given as to whether the option is enabled,
1286 disabled or set to a specific value (assuming that the compiler
1287 knows this at the point where the @option{--help=} option is used).
1289 Here is a truncated example from the ARM port of @command{gcc}:
1292 % gcc -Q -mabi=2 --help=target -c
1293 The following options are target specific:
1295 -mabort-on-noreturn [disabled]
1299 The output is sensitive to the effects of previous command line
1300 options, so for example it is possible to find out which optimizations
1301 are enabled at @option{-O2} by using:
1304 -Q -O2 --help=optimizers
1307 Alternatively you can discover which binary optimizations are enabled
1308 by @option{-O3} by using:
1311 gcc -c -Q -O3 --help=optimizers > /tmp/O3-opts
1312 gcc -c -Q -O2 --help=optimizers > /tmp/O2-opts
1313 diff /tmp/O2-opts /tmp/O3-opts | grep enabled
1316 @item -no-canonical-prefixes
1317 @opindex no-canonical-prefixes
1318 Do not expand any symbolic links, resolve references to @samp{/../}
1319 or @samp{/./}, or make the path absolute when generating a relative
1324 Display the version number and copyrights of the invoked GCC@.
1328 Invoke all subcommands under a wrapper program. It takes a single
1329 comma separated list as an argument, which will be used to invoke
1333 gcc -c t.c -wrapper gdb,--args
1336 This will invoke all subprograms of gcc under "gdb --args",
1337 thus cc1 invocation will be "gdb --args cc1 ...".
1339 @item -fplugin=@var{name}.so
1340 Load the plugin code in file @var{name}.so, assumed to be a
1341 shared object to be dlopen'd by the compiler. The base name of
1342 the shared object file is used to identify the plugin for the
1343 purposes of argument parsing (See
1344 @option{-fplugin-arg-@var{name}-@var{key}=@var{value}} below).
1345 Each plugin should define the callback functions specified in the
1348 @item -fplugin-arg-@var{name}-@var{key}=@var{value}
1349 Define an argument called @var{key} with a value of @var{value}
1350 for the plugin called @var{name}.
1352 @include @value{srcdir}/../libiberty/at-file.texi
1356 @section Compiling C++ Programs
1358 @cindex suffixes for C++ source
1359 @cindex C++ source file suffixes
1360 C++ source files conventionally use one of the suffixes @samp{.C},
1361 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
1362 @samp{.cxx}; C++ header files often use @samp{.hh}, @samp{.hpp},
1363 @samp{.H}, or (for shared template code) @samp{.tcc}; and
1364 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
1365 files with these names and compiles them as C++ programs even if you
1366 call the compiler the same way as for compiling C programs (usually
1367 with the name @command{gcc}).
1371 However, the use of @command{gcc} does not add the C++ library.
1372 @command{g++} is a program that calls GCC and treats @samp{.c},
1373 @samp{.h} and @samp{.i} files as C++ source files instead of C source
1374 files unless @option{-x} is used, and automatically specifies linking
1375 against the C++ library. This program is also useful when
1376 precompiling a C header file with a @samp{.h} extension for use in C++
1377 compilations. On many systems, @command{g++} is also installed with
1378 the name @command{c++}.
1380 @cindex invoking @command{g++}
1381 When you compile C++ programs, you may specify many of the same
1382 command-line options that you use for compiling programs in any
1383 language; or command-line options meaningful for C and related
1384 languages; or options that are meaningful only for C++ programs.
1385 @xref{C Dialect Options,,Options Controlling C Dialect}, for
1386 explanations of options for languages related to C@.
1387 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
1388 explanations of options that are meaningful only for C++ programs.
1390 @node C Dialect Options
1391 @section Options Controlling C Dialect
1392 @cindex dialect options
1393 @cindex language dialect options
1394 @cindex options, dialect
1396 The following options control the dialect of C (or languages derived
1397 from C, such as C++, Objective-C and Objective-C++) that the compiler
1401 @cindex ANSI support
1405 In C mode, this is equivalent to @samp{-std=c89}. In C++ mode, it is
1406 equivalent to @samp{-std=c++98}.
1408 This turns off certain features of GCC that are incompatible with ISO
1409 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
1410 such as the @code{asm} and @code{typeof} keywords, and
1411 predefined macros such as @code{unix} and @code{vax} that identify the
1412 type of system you are using. It also enables the undesirable and
1413 rarely used ISO trigraph feature. For the C compiler,
1414 it disables recognition of C++ style @samp{//} comments as well as
1415 the @code{inline} keyword.
1417 The alternate keywords @code{__asm__}, @code{__extension__},
1418 @code{__inline__} and @code{__typeof__} continue to work despite
1419 @option{-ansi}. You would not want to use them in an ISO C program, of
1420 course, but it is useful to put them in header files that might be included
1421 in compilations done with @option{-ansi}. Alternate predefined macros
1422 such as @code{__unix__} and @code{__vax__} are also available, with or
1423 without @option{-ansi}.
1425 The @option{-ansi} option does not cause non-ISO programs to be
1426 rejected gratuitously. For that, @option{-pedantic} is required in
1427 addition to @option{-ansi}. @xref{Warning Options}.
1429 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1430 option is used. Some header files may notice this macro and refrain
1431 from declaring certain functions or defining certain macros that the
1432 ISO standard doesn't call for; this is to avoid interfering with any
1433 programs that might use these names for other things.
1435 Functions that would normally be built in but do not have semantics
1436 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1437 functions when @option{-ansi} is used. @xref{Other Builtins,,Other
1438 built-in functions provided by GCC}, for details of the functions
1443 Determine the language standard. @xref{Standards,,Language Standards
1444 Supported by GCC}, for details of these standard versions. This option
1445 is currently only supported when compiling C or C++.
1447 The compiler can accept several base standards, such as @samp{c89} or
1448 @samp{c++98}, and GNU dialects of those standards, such as
1449 @samp{gnu89} or @samp{gnu++98}. By specifying a base standard, the
1450 compiler will accept all programs following that standard and those
1451 using GNU extensions that do not contradict it. For example,
1452 @samp{-std=c89} turns off certain features of GCC that are
1453 incompatible with ISO C90, such as the @code{asm} and @code{typeof}
1454 keywords, but not other GNU extensions that do not have a meaning in
1455 ISO C90, such as omitting the middle term of a @code{?:}
1456 expression. On the other hand, by specifying a GNU dialect of a
1457 standard, all features the compiler support are enabled, even when
1458 those features change the meaning of the base standard and some
1459 strict-conforming programs may be rejected. The particular standard
1460 is used by @option{-pedantic} to identify which features are GNU
1461 extensions given that version of the standard. For example
1462 @samp{-std=gnu89 -pedantic} would warn about C++ style @samp{//}
1463 comments, while @samp{-std=gnu99 -pedantic} would not.
1465 A value for this option must be provided; possible values are
1470 Support all ISO C90 programs (certain GNU extensions that conflict
1471 with ISO C90 are disabled). Same as @option{-ansi} for C code.
1473 @item iso9899:199409
1474 ISO C90 as modified in amendment 1.
1480 ISO C99. Note that this standard is not yet fully supported; see
1481 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1482 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1485 GNU dialect of ISO C90 (including some C99 features). This
1486 is the default for C code.
1490 GNU dialect of ISO C99. When ISO C99 is fully implemented in GCC,
1491 this will become the default. The name @samp{gnu9x} is deprecated.
1494 The 1998 ISO C++ standard plus amendments. Same as @option{-ansi} for
1498 GNU dialect of @option{-std=c++98}. This is the default for
1502 The working draft of the upcoming ISO C++0x standard. This option
1503 enables experimental features that are likely to be included in
1504 C++0x. The working draft is constantly changing, and any feature that is
1505 enabled by this flag may be removed from future versions of GCC if it is
1506 not part of the C++0x standard.
1509 GNU dialect of @option{-std=c++0x}. This option enables
1510 experimental features that may be removed in future versions of GCC.
1513 @item -fgnu89-inline
1514 @opindex fgnu89-inline
1515 The option @option{-fgnu89-inline} tells GCC to use the traditional
1516 GNU semantics for @code{inline} functions when in C99 mode.
1517 @xref{Inline,,An Inline Function is As Fast As a Macro}. This option
1518 is accepted and ignored by GCC versions 4.1.3 up to but not including
1519 4.3. In GCC versions 4.3 and later it changes the behavior of GCC in
1520 C99 mode. Using this option is roughly equivalent to adding the
1521 @code{gnu_inline} function attribute to all inline functions
1522 (@pxref{Function Attributes}).
1524 The option @option{-fno-gnu89-inline} explicitly tells GCC to use the
1525 C99 semantics for @code{inline} when in C99 or gnu99 mode (i.e., it
1526 specifies the default behavior). This option was first supported in
1527 GCC 4.3. This option is not supported in C89 or gnu89 mode.
1529 The preprocessor macros @code{__GNUC_GNU_INLINE__} and
1530 @code{__GNUC_STDC_INLINE__} may be used to check which semantics are
1531 in effect for @code{inline} functions. @xref{Common Predefined
1532 Macros,,,cpp,The C Preprocessor}.
1534 @item -aux-info @var{filename}
1536 Output to the given filename prototyped declarations for all functions
1537 declared and/or defined in a translation unit, including those in header
1538 files. This option is silently ignored in any language other than C@.
1540 Besides declarations, the file indicates, in comments, the origin of
1541 each declaration (source file and line), whether the declaration was
1542 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1543 @samp{O} for old, respectively, in the first character after the line
1544 number and the colon), and whether it came from a declaration or a
1545 definition (@samp{C} or @samp{F}, respectively, in the following
1546 character). In the case of function definitions, a K&R-style list of
1547 arguments followed by their declarations is also provided, inside
1548 comments, after the declaration.
1552 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1553 keyword, so that code can use these words as identifiers. You can use
1554 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1555 instead. @option{-ansi} implies @option{-fno-asm}.
1557 In C++, this switch only affects the @code{typeof} keyword, since
1558 @code{asm} and @code{inline} are standard keywords. You may want to
1559 use the @option{-fno-gnu-keywords} flag instead, which has the same
1560 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1561 switch only affects the @code{asm} and @code{typeof} keywords, since
1562 @code{inline} is a standard keyword in ISO C99.
1565 @itemx -fno-builtin-@var{function}
1566 @opindex fno-builtin
1567 @cindex built-in functions
1568 Don't recognize built-in functions that do not begin with
1569 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1570 functions provided by GCC}, for details of the functions affected,
1571 including those which are not built-in functions when @option{-ansi} or
1572 @option{-std} options for strict ISO C conformance are used because they
1573 do not have an ISO standard meaning.
1575 GCC normally generates special code to handle certain built-in functions
1576 more efficiently; for instance, calls to @code{alloca} may become single
1577 instructions that adjust the stack directly, and calls to @code{memcpy}
1578 may become inline copy loops. The resulting code is often both smaller
1579 and faster, but since the function calls no longer appear as such, you
1580 cannot set a breakpoint on those calls, nor can you change the behavior
1581 of the functions by linking with a different library. In addition,
1582 when a function is recognized as a built-in function, GCC may use
1583 information about that function to warn about problems with calls to
1584 that function, or to generate more efficient code, even if the
1585 resulting code still contains calls to that function. For example,
1586 warnings are given with @option{-Wformat} for bad calls to
1587 @code{printf}, when @code{printf} is built in, and @code{strlen} is
1588 known not to modify global memory.
1590 With the @option{-fno-builtin-@var{function}} option
1591 only the built-in function @var{function} is
1592 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1593 function is named that is not built-in in this version of GCC, this
1594 option is ignored. There is no corresponding
1595 @option{-fbuiltin-@var{function}} option; if you wish to enable
1596 built-in functions selectively when using @option{-fno-builtin} or
1597 @option{-ffreestanding}, you may define macros such as:
1600 #define abs(n) __builtin_abs ((n))
1601 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1606 @cindex hosted environment
1608 Assert that compilation takes place in a hosted environment. This implies
1609 @option{-fbuiltin}. A hosted environment is one in which the
1610 entire standard library is available, and in which @code{main} has a return
1611 type of @code{int}. Examples are nearly everything except a kernel.
1612 This is equivalent to @option{-fno-freestanding}.
1614 @item -ffreestanding
1615 @opindex ffreestanding
1616 @cindex hosted environment
1618 Assert that compilation takes place in a freestanding environment. This
1619 implies @option{-fno-builtin}. A freestanding environment
1620 is one in which the standard library may not exist, and program startup may
1621 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1622 This is equivalent to @option{-fno-hosted}.
1624 @xref{Standards,,Language Standards Supported by GCC}, for details of
1625 freestanding and hosted environments.
1629 @cindex openmp parallel
1630 Enable handling of OpenMP directives @code{#pragma omp} in C/C++ and
1631 @code{!$omp} in Fortran. When @option{-fopenmp} is specified, the
1632 compiler generates parallel code according to the OpenMP Application
1633 Program Interface v3.0 @w{@uref{http://www.openmp.org/}}. This option
1634 implies @option{-pthread}, and thus is only supported on targets that
1635 have support for @option{-pthread}.
1637 @item -fms-extensions
1638 @opindex fms-extensions
1639 Accept some non-standard constructs used in Microsoft header files.
1641 Some cases of unnamed fields in structures and unions are only
1642 accepted with this option. @xref{Unnamed Fields,,Unnamed struct/union
1643 fields within structs/unions}, for details.
1647 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1648 options for strict ISO C conformance) implies @option{-trigraphs}.
1650 @item -no-integrated-cpp
1651 @opindex no-integrated-cpp
1652 Performs a compilation in two passes: preprocessing and compiling. This
1653 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1654 @option{-B} option. The user supplied compilation step can then add in
1655 an additional preprocessing step after normal preprocessing but before
1656 compiling. The default is to use the integrated cpp (internal cpp)
1658 The semantics of this option will change if "cc1", "cc1plus", and
1659 "cc1obj" are merged.
1661 @cindex traditional C language
1662 @cindex C language, traditional
1664 @itemx -traditional-cpp
1665 @opindex traditional-cpp
1666 @opindex traditional
1667 Formerly, these options caused GCC to attempt to emulate a pre-standard
1668 C compiler. They are now only supported with the @option{-E} switch.
1669 The preprocessor continues to support a pre-standard mode. See the GNU
1670 CPP manual for details.
1672 @item -fcond-mismatch
1673 @opindex fcond-mismatch
1674 Allow conditional expressions with mismatched types in the second and
1675 third arguments. The value of such an expression is void. This option
1676 is not supported for C++.
1678 @item -flax-vector-conversions
1679 @opindex flax-vector-conversions
1680 Allow implicit conversions between vectors with differing numbers of
1681 elements and/or incompatible element types. This option should not be
1684 @item -funsigned-char
1685 @opindex funsigned-char
1686 Let the type @code{char} be unsigned, like @code{unsigned char}.
1688 Each kind of machine has a default for what @code{char} should
1689 be. It is either like @code{unsigned char} by default or like
1690 @code{signed char} by default.
1692 Ideally, a portable program should always use @code{signed char} or
1693 @code{unsigned char} when it depends on the signedness of an object.
1694 But many programs have been written to use plain @code{char} and
1695 expect it to be signed, or expect it to be unsigned, depending on the
1696 machines they were written for. This option, and its inverse, let you
1697 make such a program work with the opposite default.
1699 The type @code{char} is always a distinct type from each of
1700 @code{signed char} or @code{unsigned char}, even though its behavior
1701 is always just like one of those two.
1704 @opindex fsigned-char
1705 Let the type @code{char} be signed, like @code{signed char}.
1707 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1708 the negative form of @option{-funsigned-char}. Likewise, the option
1709 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1711 @item -fsigned-bitfields
1712 @itemx -funsigned-bitfields
1713 @itemx -fno-signed-bitfields
1714 @itemx -fno-unsigned-bitfields
1715 @opindex fsigned-bitfields
1716 @opindex funsigned-bitfields
1717 @opindex fno-signed-bitfields
1718 @opindex fno-unsigned-bitfields
1719 These options control whether a bit-field is signed or unsigned, when the
1720 declaration does not use either @code{signed} or @code{unsigned}. By
1721 default, such a bit-field is signed, because this is consistent: the
1722 basic integer types such as @code{int} are signed types.
1725 @node C++ Dialect Options
1726 @section Options Controlling C++ Dialect
1728 @cindex compiler options, C++
1729 @cindex C++ options, command line
1730 @cindex options, C++
1731 This section describes the command-line options that are only meaningful
1732 for C++ programs; but you can also use most of the GNU compiler options
1733 regardless of what language your program is in. For example, you
1734 might compile a file @code{firstClass.C} like this:
1737 g++ -g -frepo -O -c firstClass.C
1741 In this example, only @option{-frepo} is an option meant
1742 only for C++ programs; you can use the other options with any
1743 language supported by GCC@.
1745 Here is a list of options that are @emph{only} for compiling C++ programs:
1749 @item -fabi-version=@var{n}
1750 @opindex fabi-version
1751 Use version @var{n} of the C++ ABI@. Version 2 is the version of the
1752 C++ ABI that first appeared in G++ 3.4. Version 1 is the version of
1753 the C++ ABI that first appeared in G++ 3.2. Version 0 will always be
1754 the version that conforms most closely to the C++ ABI specification.
1755 Therefore, the ABI obtained using version 0 will change as ABI bugs
1758 The default is version 2.
1760 @item -fno-access-control
1761 @opindex fno-access-control
1762 Turn off all access checking. This switch is mainly useful for working
1763 around bugs in the access control code.
1767 Check that the pointer returned by @code{operator new} is non-null
1768 before attempting to modify the storage allocated. This check is
1769 normally unnecessary because the C++ standard specifies that
1770 @code{operator new} will only return @code{0} if it is declared
1771 @samp{throw()}, in which case the compiler will always check the
1772 return value even without this option. In all other cases, when
1773 @code{operator new} has a non-empty exception specification, memory
1774 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1775 @samp{new (nothrow)}.
1777 @item -fconserve-space
1778 @opindex fconserve-space
1779 Put uninitialized or runtime-initialized global variables into the
1780 common segment, as C does. This saves space in the executable at the
1781 cost of not diagnosing duplicate definitions. If you compile with this
1782 flag and your program mysteriously crashes after @code{main()} has
1783 completed, you may have an object that is being destroyed twice because
1784 two definitions were merged.
1786 This option is no longer useful on most targets, now that support has
1787 been added for putting variables into BSS without making them common.
1789 @item -fno-deduce-init-list
1790 @opindex fno-deduce-init-list
1791 Disable deduction of a template type parameter as
1792 std::initializer_list from a brace-enclosed initializer list, i.e.
1795 template <class T> auto forward(T t) -> decltype (realfn (t))
1802 forward(@{1,2@}); // call forward<std::initializer_list<int>>
1806 This option is present because this deduction is an extension to the
1807 current specification in the C++0x working draft, and there was
1808 some concern about potential overload resolution problems.
1810 @item -ffriend-injection
1811 @opindex ffriend-injection
1812 Inject friend functions into the enclosing namespace, so that they are
1813 visible outside the scope of the class in which they are declared.
1814 Friend functions were documented to work this way in the old Annotated
1815 C++ Reference Manual, and versions of G++ before 4.1 always worked
1816 that way. However, in ISO C++ a friend function which is not declared
1817 in an enclosing scope can only be found using argument dependent
1818 lookup. This option causes friends to be injected as they were in
1821 This option is for compatibility, and may be removed in a future
1824 @item -fno-elide-constructors
1825 @opindex fno-elide-constructors
1826 The C++ standard allows an implementation to omit creating a temporary
1827 which is only used to initialize another object of the same type.
1828 Specifying this option disables that optimization, and forces G++ to
1829 call the copy constructor in all cases.
1831 @item -fno-enforce-eh-specs
1832 @opindex fno-enforce-eh-specs
1833 Don't generate code to check for violation of exception specifications
1834 at runtime. This option violates the C++ standard, but may be useful
1835 for reducing code size in production builds, much like defining
1836 @samp{NDEBUG}. This does not give user code permission to throw
1837 exceptions in violation of the exception specifications; the compiler
1838 will still optimize based on the specifications, so throwing an
1839 unexpected exception will result in undefined behavior.
1842 @itemx -fno-for-scope
1844 @opindex fno-for-scope
1845 If @option{-ffor-scope} is specified, the scope of variables declared in
1846 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1847 as specified by the C++ standard.
1848 If @option{-fno-for-scope} is specified, the scope of variables declared in
1849 a @i{for-init-statement} extends to the end of the enclosing scope,
1850 as was the case in old versions of G++, and other (traditional)
1851 implementations of C++.
1853 The default if neither flag is given to follow the standard,
1854 but to allow and give a warning for old-style code that would
1855 otherwise be invalid, or have different behavior.
1857 @item -fno-gnu-keywords
1858 @opindex fno-gnu-keywords
1859 Do not recognize @code{typeof} as a keyword, so that code can use this
1860 word as an identifier. You can use the keyword @code{__typeof__} instead.
1861 @option{-ansi} implies @option{-fno-gnu-keywords}.
1863 @item -fno-implicit-templates
1864 @opindex fno-implicit-templates
1865 Never emit code for non-inline templates which are instantiated
1866 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1867 @xref{Template Instantiation}, for more information.
1869 @item -fno-implicit-inline-templates
1870 @opindex fno-implicit-inline-templates
1871 Don't emit code for implicit instantiations of inline templates, either.
1872 The default is to handle inlines differently so that compiles with and
1873 without optimization will need the same set of explicit instantiations.
1875 @item -fno-implement-inlines
1876 @opindex fno-implement-inlines
1877 To save space, do not emit out-of-line copies of inline functions
1878 controlled by @samp{#pragma implementation}. This will cause linker
1879 errors if these functions are not inlined everywhere they are called.
1881 @item -fms-extensions
1882 @opindex fms-extensions
1883 Disable pedantic warnings about constructs used in MFC, such as implicit
1884 int and getting a pointer to member function via non-standard syntax.
1886 @item -fno-nonansi-builtins
1887 @opindex fno-nonansi-builtins
1888 Disable built-in declarations of functions that are not mandated by
1889 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1890 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1892 @item -fno-operator-names
1893 @opindex fno-operator-names
1894 Do not treat the operator name keywords @code{and}, @code{bitand},
1895 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1896 synonyms as keywords.
1898 @item -fno-optional-diags
1899 @opindex fno-optional-diags
1900 Disable diagnostics that the standard says a compiler does not need to
1901 issue. Currently, the only such diagnostic issued by G++ is the one for
1902 a name having multiple meanings within a class.
1905 @opindex fpermissive
1906 Downgrade some diagnostics about nonconformant code from errors to
1907 warnings. Thus, using @option{-fpermissive} will allow some
1908 nonconforming code to compile.
1910 @item -fno-pretty-templates
1911 @opindex fno-pretty-templates
1912 When an error message refers to a specialization of a function
1913 template, the compiler will normally print the signature of the
1914 template followed by the template arguments and any typedefs or
1915 typenames in the signature (e.g. @code{void f(T) [with T = int]}
1916 rather than @code{void f(int)}) so that it's clear which template is
1917 involved. When an error message refers to a specialization of a class
1918 template, the compiler will omit any template arguments which match
1919 the default template arguments for that template. If either of these
1920 behaviors make it harder to understand the error message rather than
1921 easier, using @option{-fno-pretty-templates} will disable them.
1925 Enable automatic template instantiation at link time. This option also
1926 implies @option{-fno-implicit-templates}. @xref{Template
1927 Instantiation}, for more information.
1931 Disable generation of information about every class with virtual
1932 functions for use by the C++ runtime type identification features
1933 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1934 of the language, you can save some space by using this flag. Note that
1935 exception handling uses the same information, but it will generate it as
1936 needed. The @samp{dynamic_cast} operator can still be used for casts that
1937 do not require runtime type information, i.e.@: casts to @code{void *} or to
1938 unambiguous base classes.
1942 Emit statistics about front-end processing at the end of the compilation.
1943 This information is generally only useful to the G++ development team.
1945 @item -ftemplate-depth-@var{n}
1946 @opindex ftemplate-depth
1947 Set the maximum instantiation depth for template classes to @var{n}.
1948 A limit on the template instantiation depth is needed to detect
1949 endless recursions during template class instantiation. ANSI/ISO C++
1950 conforming programs must not rely on a maximum depth greater than 17
1951 (changed to 1024 in C++0x).
1953 @item -fno-threadsafe-statics
1954 @opindex fno-threadsafe-statics
1955 Do not emit the extra code to use the routines specified in the C++
1956 ABI for thread-safe initialization of local statics. You can use this
1957 option to reduce code size slightly in code that doesn't need to be
1960 @item -fuse-cxa-atexit
1961 @opindex fuse-cxa-atexit
1962 Register destructors for objects with static storage duration with the
1963 @code{__cxa_atexit} function rather than the @code{atexit} function.
1964 This option is required for fully standards-compliant handling of static
1965 destructors, but will only work if your C library supports
1966 @code{__cxa_atexit}.
1968 @item -fno-use-cxa-get-exception-ptr
1969 @opindex fno-use-cxa-get-exception-ptr
1970 Don't use the @code{__cxa_get_exception_ptr} runtime routine. This
1971 will cause @code{std::uncaught_exception} to be incorrect, but is necessary
1972 if the runtime routine is not available.
1974 @item -fvisibility-inlines-hidden
1975 @opindex fvisibility-inlines-hidden
1976 This switch declares that the user does not attempt to compare
1977 pointers to inline methods where the addresses of the two functions
1978 were taken in different shared objects.
1980 The effect of this is that GCC may, effectively, mark inline methods with
1981 @code{__attribute__ ((visibility ("hidden")))} so that they do not
1982 appear in the export table of a DSO and do not require a PLT indirection
1983 when used within the DSO@. Enabling this option can have a dramatic effect
1984 on load and link times of a DSO as it massively reduces the size of the
1985 dynamic export table when the library makes heavy use of templates.
1987 The behavior of this switch is not quite the same as marking the
1988 methods as hidden directly, because it does not affect static variables
1989 local to the function or cause the compiler to deduce that
1990 the function is defined in only one shared object.
1992 You may mark a method as having a visibility explicitly to negate the
1993 effect of the switch for that method. For example, if you do want to
1994 compare pointers to a particular inline method, you might mark it as
1995 having default visibility. Marking the enclosing class with explicit
1996 visibility will have no effect.
1998 Explicitly instantiated inline methods are unaffected by this option
1999 as their linkage might otherwise cross a shared library boundary.
2000 @xref{Template Instantiation}.
2002 @item -fvisibility-ms-compat
2003 @opindex fvisibility-ms-compat
2004 This flag attempts to use visibility settings to make GCC's C++
2005 linkage model compatible with that of Microsoft Visual Studio.
2007 The flag makes these changes to GCC's linkage model:
2011 It sets the default visibility to @code{hidden}, like
2012 @option{-fvisibility=hidden}.
2015 Types, but not their members, are not hidden by default.
2018 The One Definition Rule is relaxed for types without explicit
2019 visibility specifications which are defined in more than one different
2020 shared object: those declarations are permitted if they would have
2021 been permitted when this option was not used.
2024 In new code it is better to use @option{-fvisibility=hidden} and
2025 export those classes which are intended to be externally visible.
2026 Unfortunately it is possible for code to rely, perhaps accidentally,
2027 on the Visual Studio behavior.
2029 Among the consequences of these changes are that static data members
2030 of the same type with the same name but defined in different shared
2031 objects will be different, so changing one will not change the other;
2032 and that pointers to function members defined in different shared
2033 objects may not compare equal. When this flag is given, it is a
2034 violation of the ODR to define types with the same name differently.
2038 Do not use weak symbol support, even if it is provided by the linker.
2039 By default, G++ will use weak symbols if they are available. This
2040 option exists only for testing, and should not be used by end-users;
2041 it will result in inferior code and has no benefits. This option may
2042 be removed in a future release of G++.
2046 Do not search for header files in the standard directories specific to
2047 C++, but do still search the other standard directories. (This option
2048 is used when building the C++ library.)
2051 In addition, these optimization, warning, and code generation options
2052 have meanings only for C++ programs:
2055 @item -fno-default-inline
2056 @opindex fno-default-inline
2057 Do not assume @samp{inline} for functions defined inside a class scope.
2058 @xref{Optimize Options,,Options That Control Optimization}. Note that these
2059 functions will have linkage like inline functions; they just won't be
2062 @item -Wabi @r{(C, Objective-C, C++ and Objective-C++ only)}
2065 Warn when G++ generates code that is probably not compatible with the
2066 vendor-neutral C++ ABI@. Although an effort has been made to warn about
2067 all such cases, there are probably some cases that are not warned about,
2068 even though G++ is generating incompatible code. There may also be
2069 cases where warnings are emitted even though the code that is generated
2072 You should rewrite your code to avoid these warnings if you are
2073 concerned about the fact that code generated by G++ may not be binary
2074 compatible with code generated by other compilers.
2076 The known incompatibilities at this point include:
2081 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
2082 pack data into the same byte as a base class. For example:
2085 struct A @{ virtual void f(); int f1 : 1; @};
2086 struct B : public A @{ int f2 : 1; @};
2090 In this case, G++ will place @code{B::f2} into the same byte
2091 as@code{A::f1}; other compilers will not. You can avoid this problem
2092 by explicitly padding @code{A} so that its size is a multiple of the
2093 byte size on your platform; that will cause G++ and other compilers to
2094 layout @code{B} identically.
2097 Incorrect handling of tail-padding for virtual bases. G++ does not use
2098 tail padding when laying out virtual bases. For example:
2101 struct A @{ virtual void f(); char c1; @};
2102 struct B @{ B(); char c2; @};
2103 struct C : public A, public virtual B @{@};
2107 In this case, G++ will not place @code{B} into the tail-padding for
2108 @code{A}; other compilers will. You can avoid this problem by
2109 explicitly padding @code{A} so that its size is a multiple of its
2110 alignment (ignoring virtual base classes); that will cause G++ and other
2111 compilers to layout @code{C} identically.
2114 Incorrect handling of bit-fields with declared widths greater than that
2115 of their underlying types, when the bit-fields appear in a union. For
2119 union U @{ int i : 4096; @};
2123 Assuming that an @code{int} does not have 4096 bits, G++ will make the
2124 union too small by the number of bits in an @code{int}.
2127 Empty classes can be placed at incorrect offsets. For example:
2137 struct C : public B, public A @{@};
2141 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
2142 it should be placed at offset zero. G++ mistakenly believes that the
2143 @code{A} data member of @code{B} is already at offset zero.
2146 Names of template functions whose types involve @code{typename} or
2147 template template parameters can be mangled incorrectly.
2150 template <typename Q>
2151 void f(typename Q::X) @{@}
2153 template <template <typename> class Q>
2154 void f(typename Q<int>::X) @{@}
2158 Instantiations of these templates may be mangled incorrectly.
2162 It also warns psABI related changes. The known psABI changes at this
2168 For SYSV/x86-64, when passing union with long double, it is changed to
2169 pass in memory as specified in psABI. For example:
2179 @code{union U} will always be passed in memory.
2183 @item -Wctor-dtor-privacy @r{(C++ and Objective-C++ only)}
2184 @opindex Wctor-dtor-privacy
2185 @opindex Wno-ctor-dtor-privacy
2186 Warn when a class seems unusable because all the constructors or
2187 destructors in that class are private, and it has neither friends nor
2188 public static member functions.
2190 @item -Wnon-virtual-dtor @r{(C++ and Objective-C++ only)}
2191 @opindex Wnon-virtual-dtor
2192 @opindex Wno-non-virtual-dtor
2193 Warn when a class has virtual functions and accessible non-virtual
2194 destructor, in which case it would be possible but unsafe to delete
2195 an instance of a derived class through a pointer to the base class.
2196 This warning is also enabled if -Weffc++ is specified.
2198 @item -Wreorder @r{(C++ and Objective-C++ only)}
2200 @opindex Wno-reorder
2201 @cindex reordering, warning
2202 @cindex warning for reordering of member initializers
2203 Warn when the order of member initializers given in the code does not
2204 match the order in which they must be executed. For instance:
2210 A(): j (0), i (1) @{ @}
2214 The compiler will rearrange the member initializers for @samp{i}
2215 and @samp{j} to match the declaration order of the members, emitting
2216 a warning to that effect. This warning is enabled by @option{-Wall}.
2219 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
2222 @item -Weffc++ @r{(C++ and Objective-C++ only)}
2225 Warn about violations of the following style guidelines from Scott Meyers'
2226 @cite{Effective C++} book:
2230 Item 11: Define a copy constructor and an assignment operator for classes
2231 with dynamically allocated memory.
2234 Item 12: Prefer initialization to assignment in constructors.
2237 Item 14: Make destructors virtual in base classes.
2240 Item 15: Have @code{operator=} return a reference to @code{*this}.
2243 Item 23: Don't try to return a reference when you must return an object.
2247 Also warn about violations of the following style guidelines from
2248 Scott Meyers' @cite{More Effective C++} book:
2252 Item 6: Distinguish between prefix and postfix forms of increment and
2253 decrement operators.
2256 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
2260 When selecting this option, be aware that the standard library
2261 headers do not obey all of these guidelines; use @samp{grep -v}
2262 to filter out those warnings.
2264 @item -Wstrict-null-sentinel @r{(C++ and Objective-C++ only)}
2265 @opindex Wstrict-null-sentinel
2266 @opindex Wno-strict-null-sentinel
2267 Warn also about the use of an uncasted @code{NULL} as sentinel. When
2268 compiling only with GCC this is a valid sentinel, as @code{NULL} is defined
2269 to @code{__null}. Although it is a null pointer constant not a null pointer,
2270 it is guaranteed to be of the same size as a pointer. But this use is
2271 not portable across different compilers.
2273 @item -Wno-non-template-friend @r{(C++ and Objective-C++ only)}
2274 @opindex Wno-non-template-friend
2275 @opindex Wnon-template-friend
2276 Disable warnings when non-templatized friend functions are declared
2277 within a template. Since the advent of explicit template specification
2278 support in G++, if the name of the friend is an unqualified-id (i.e.,
2279 @samp{friend foo(int)}), the C++ language specification demands that the
2280 friend declare or define an ordinary, nontemplate function. (Section
2281 14.5.3). Before G++ implemented explicit specification, unqualified-ids
2282 could be interpreted as a particular specialization of a templatized
2283 function. Because this non-conforming behavior is no longer the default
2284 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
2285 check existing code for potential trouble spots and is on by default.
2286 This new compiler behavior can be turned off with
2287 @option{-Wno-non-template-friend} which keeps the conformant compiler code
2288 but disables the helpful warning.
2290 @item -Wold-style-cast @r{(C++ and Objective-C++ only)}
2291 @opindex Wold-style-cast
2292 @opindex Wno-old-style-cast
2293 Warn if an old-style (C-style) cast to a non-void type is used within
2294 a C++ program. The new-style casts (@samp{dynamic_cast},
2295 @samp{static_cast}, @samp{reinterpret_cast}, and @samp{const_cast}) are
2296 less vulnerable to unintended effects and much easier to search for.
2298 @item -Woverloaded-virtual @r{(C++ and Objective-C++ only)}
2299 @opindex Woverloaded-virtual
2300 @opindex Wno-overloaded-virtual
2301 @cindex overloaded virtual fn, warning
2302 @cindex warning for overloaded virtual fn
2303 Warn when a function declaration hides virtual functions from a
2304 base class. For example, in:
2311 struct B: public A @{
2316 the @code{A} class version of @code{f} is hidden in @code{B}, and code
2324 will fail to compile.
2326 @item -Wno-pmf-conversions @r{(C++ and Objective-C++ only)}
2327 @opindex Wno-pmf-conversions
2328 @opindex Wpmf-conversions
2329 Disable the diagnostic for converting a bound pointer to member function
2332 @item -Wsign-promo @r{(C++ and Objective-C++ only)}
2333 @opindex Wsign-promo
2334 @opindex Wno-sign-promo
2335 Warn when overload resolution chooses a promotion from unsigned or
2336 enumerated type to a signed type, over a conversion to an unsigned type of
2337 the same size. Previous versions of G++ would try to preserve
2338 unsignedness, but the standard mandates the current behavior.
2343 A& operator = (int);
2353 In this example, G++ will synthesize a default @samp{A& operator =
2354 (const A&);}, while cfront will use the user-defined @samp{operator =}.
2357 @node Objective-C and Objective-C++ Dialect Options
2358 @section Options Controlling Objective-C and Objective-C++ Dialects
2360 @cindex compiler options, Objective-C and Objective-C++
2361 @cindex Objective-C and Objective-C++ options, command line
2362 @cindex options, Objective-C and Objective-C++
2363 (NOTE: This manual does not describe the Objective-C and Objective-C++
2364 languages themselves. See @xref{Standards,,Language Standards
2365 Supported by GCC}, for references.)
2367 This section describes the command-line options that are only meaningful
2368 for Objective-C and Objective-C++ programs, but you can also use most of
2369 the language-independent GNU compiler options.
2370 For example, you might compile a file @code{some_class.m} like this:
2373 gcc -g -fgnu-runtime -O -c some_class.m
2377 In this example, @option{-fgnu-runtime} is an option meant only for
2378 Objective-C and Objective-C++ programs; you can use the other options with
2379 any language supported by GCC@.
2381 Note that since Objective-C is an extension of the C language, Objective-C
2382 compilations may also use options specific to the C front-end (e.g.,
2383 @option{-Wtraditional}). Similarly, Objective-C++ compilations may use
2384 C++-specific options (e.g., @option{-Wabi}).
2386 Here is a list of options that are @emph{only} for compiling Objective-C
2387 and Objective-C++ programs:
2390 @item -fconstant-string-class=@var{class-name}
2391 @opindex fconstant-string-class
2392 Use @var{class-name} as the name of the class to instantiate for each
2393 literal string specified with the syntax @code{@@"@dots{}"}. The default
2394 class name is @code{NXConstantString} if the GNU runtime is being used, and
2395 @code{NSConstantString} if the NeXT runtime is being used (see below). The
2396 @option{-fconstant-cfstrings} option, if also present, will override the
2397 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
2398 to be laid out as constant CoreFoundation strings.
2401 @opindex fgnu-runtime
2402 Generate object code compatible with the standard GNU Objective-C
2403 runtime. This is the default for most types of systems.
2405 @item -fnext-runtime
2406 @opindex fnext-runtime
2407 Generate output compatible with the NeXT runtime. This is the default
2408 for NeXT-based systems, including Darwin and Mac OS X@. The macro
2409 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
2412 @item -fno-nil-receivers
2413 @opindex fno-nil-receivers
2414 Assume that all Objective-C message dispatches (e.g.,
2415 @code{[receiver message:arg]}) in this translation unit ensure that the receiver
2416 is not @code{nil}. This allows for more efficient entry points in the runtime
2417 to be used. Currently, this option is only available in conjunction with
2418 the NeXT runtime on Mac OS X 10.3 and later.
2420 @item -fobjc-call-cxx-cdtors
2421 @opindex fobjc-call-cxx-cdtors
2422 For each Objective-C class, check if any of its instance variables is a
2423 C++ object with a non-trivial default constructor. If so, synthesize a
2424 special @code{- (id) .cxx_construct} instance method that will run
2425 non-trivial default constructors on any such instance variables, in order,
2426 and then return @code{self}. Similarly, check if any instance variable
2427 is a C++ object with a non-trivial destructor, and if so, synthesize a
2428 special @code{- (void) .cxx_destruct} method that will run
2429 all such default destructors, in reverse order.
2431 The @code{- (id) .cxx_construct} and/or @code{- (void) .cxx_destruct} methods
2432 thusly generated will only operate on instance variables declared in the
2433 current Objective-C class, and not those inherited from superclasses. It
2434 is the responsibility of the Objective-C runtime to invoke all such methods
2435 in an object's inheritance hierarchy. The @code{- (id) .cxx_construct} methods
2436 will be invoked by the runtime immediately after a new object
2437 instance is allocated; the @code{- (void) .cxx_destruct} methods will
2438 be invoked immediately before the runtime deallocates an object instance.
2440 As of this writing, only the NeXT runtime on Mac OS X 10.4 and later has
2441 support for invoking the @code{- (id) .cxx_construct} and
2442 @code{- (void) .cxx_destruct} methods.
2444 @item -fobjc-direct-dispatch
2445 @opindex fobjc-direct-dispatch
2446 Allow fast jumps to the message dispatcher. On Darwin this is
2447 accomplished via the comm page.
2449 @item -fobjc-exceptions
2450 @opindex fobjc-exceptions
2451 Enable syntactic support for structured exception handling in Objective-C,
2452 similar to what is offered by C++ and Java. This option is
2453 unavailable in conjunction with the NeXT runtime on Mac OS X 10.2 and
2462 @@catch (AnObjCClass *exc) @{
2469 @@catch (AnotherClass *exc) @{
2472 @@catch (id allOthers) @{
2482 The @code{@@throw} statement may appear anywhere in an Objective-C or
2483 Objective-C++ program; when used inside of a @code{@@catch} block, the
2484 @code{@@throw} may appear without an argument (as shown above), in which case
2485 the object caught by the @code{@@catch} will be rethrown.
2487 Note that only (pointers to) Objective-C objects may be thrown and
2488 caught using this scheme. When an object is thrown, it will be caught
2489 by the nearest @code{@@catch} clause capable of handling objects of that type,
2490 analogously to how @code{catch} blocks work in C++ and Java. A
2491 @code{@@catch(id @dots{})} clause (as shown above) may also be provided to catch
2492 any and all Objective-C exceptions not caught by previous @code{@@catch}
2495 The @code{@@finally} clause, if present, will be executed upon exit from the
2496 immediately preceding @code{@@try @dots{} @@catch} section. This will happen
2497 regardless of whether any exceptions are thrown, caught or rethrown
2498 inside the @code{@@try @dots{} @@catch} section, analogously to the behavior
2499 of the @code{finally} clause in Java.
2501 There are several caveats to using the new exception mechanism:
2505 Although currently designed to be binary compatible with @code{NS_HANDLER}-style
2506 idioms provided by the @code{NSException} class, the new
2507 exceptions can only be used on Mac OS X 10.3 (Panther) and later
2508 systems, due to additional functionality needed in the (NeXT) Objective-C
2512 As mentioned above, the new exceptions do not support handling
2513 types other than Objective-C objects. Furthermore, when used from
2514 Objective-C++, the Objective-C exception model does not interoperate with C++
2515 exceptions at this time. This means you cannot @code{@@throw} an exception
2516 from Objective-C and @code{catch} it in C++, or vice versa
2517 (i.e., @code{throw @dots{} @@catch}).
2520 The @option{-fobjc-exceptions} switch also enables the use of synchronization
2521 blocks for thread-safe execution:
2524 @@synchronized (ObjCClass *guard) @{
2529 Upon entering the @code{@@synchronized} block, a thread of execution shall
2530 first check whether a lock has been placed on the corresponding @code{guard}
2531 object by another thread. If it has, the current thread shall wait until
2532 the other thread relinquishes its lock. Once @code{guard} becomes available,
2533 the current thread will place its own lock on it, execute the code contained in
2534 the @code{@@synchronized} block, and finally relinquish the lock (thereby
2535 making @code{guard} available to other threads).
2537 Unlike Java, Objective-C does not allow for entire methods to be marked
2538 @code{@@synchronized}. Note that throwing exceptions out of
2539 @code{@@synchronized} blocks is allowed, and will cause the guarding object
2540 to be unlocked properly.
2544 Enable garbage collection (GC) in Objective-C and Objective-C++ programs.
2546 @item -freplace-objc-classes
2547 @opindex freplace-objc-classes
2548 Emit a special marker instructing @command{ld(1)} not to statically link in
2549 the resulting object file, and allow @command{dyld(1)} to load it in at
2550 run time instead. This is used in conjunction with the Fix-and-Continue
2551 debugging mode, where the object file in question may be recompiled and
2552 dynamically reloaded in the course of program execution, without the need
2553 to restart the program itself. Currently, Fix-and-Continue functionality
2554 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
2559 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
2560 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
2561 compile time) with static class references that get initialized at load time,
2562 which improves run-time performance. Specifying the @option{-fzero-link} flag
2563 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
2564 to be retained. This is useful in Zero-Link debugging mode, since it allows
2565 for individual class implementations to be modified during program execution.
2569 Dump interface declarations for all classes seen in the source file to a
2570 file named @file{@var{sourcename}.decl}.
2572 @item -Wassign-intercept @r{(Objective-C and Objective-C++ only)}
2573 @opindex Wassign-intercept
2574 @opindex Wno-assign-intercept
2575 Warn whenever an Objective-C assignment is being intercepted by the
2578 @item -Wno-protocol @r{(Objective-C and Objective-C++ only)}
2579 @opindex Wno-protocol
2581 If a class is declared to implement a protocol, a warning is issued for
2582 every method in the protocol that is not implemented by the class. The
2583 default behavior is to issue a warning for every method not explicitly
2584 implemented in the class, even if a method implementation is inherited
2585 from the superclass. If you use the @option{-Wno-protocol} option, then
2586 methods inherited from the superclass are considered to be implemented,
2587 and no warning is issued for them.
2589 @item -Wselector @r{(Objective-C and Objective-C++ only)}
2591 @opindex Wno-selector
2592 Warn if multiple methods of different types for the same selector are
2593 found during compilation. The check is performed on the list of methods
2594 in the final stage of compilation. Additionally, a check is performed
2595 for each selector appearing in a @code{@@selector(@dots{})}
2596 expression, and a corresponding method for that selector has been found
2597 during compilation. Because these checks scan the method table only at
2598 the end of compilation, these warnings are not produced if the final
2599 stage of compilation is not reached, for example because an error is
2600 found during compilation, or because the @option{-fsyntax-only} option is
2603 @item -Wstrict-selector-match @r{(Objective-C and Objective-C++ only)}
2604 @opindex Wstrict-selector-match
2605 @opindex Wno-strict-selector-match
2606 Warn if multiple methods with differing argument and/or return types are
2607 found for a given selector when attempting to send a message using this
2608 selector to a receiver of type @code{id} or @code{Class}. When this flag
2609 is off (which is the default behavior), the compiler will omit such warnings
2610 if any differences found are confined to types which share the same size
2613 @item -Wundeclared-selector @r{(Objective-C and Objective-C++ only)}
2614 @opindex Wundeclared-selector
2615 @opindex Wno-undeclared-selector
2616 Warn if a @code{@@selector(@dots{})} expression referring to an
2617 undeclared selector is found. A selector is considered undeclared if no
2618 method with that name has been declared before the
2619 @code{@@selector(@dots{})} expression, either explicitly in an
2620 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
2621 an @code{@@implementation} section. This option always performs its
2622 checks as soon as a @code{@@selector(@dots{})} expression is found,
2623 while @option{-Wselector} only performs its checks in the final stage of
2624 compilation. This also enforces the coding style convention
2625 that methods and selectors must be declared before being used.
2627 @item -print-objc-runtime-info
2628 @opindex print-objc-runtime-info
2629 Generate C header describing the largest structure that is passed by
2634 @node Language Independent Options
2635 @section Options to Control Diagnostic Messages Formatting
2636 @cindex options to control diagnostics formatting
2637 @cindex diagnostic messages
2638 @cindex message formatting
2640 Traditionally, diagnostic messages have been formatted irrespective of
2641 the output device's aspect (e.g.@: its width, @dots{}). The options described
2642 below can be used to control the diagnostic messages formatting
2643 algorithm, e.g.@: how many characters per line, how often source location
2644 information should be reported. Right now, only the C++ front end can
2645 honor these options. However it is expected, in the near future, that
2646 the remaining front ends would be able to digest them correctly.
2649 @item -fmessage-length=@var{n}
2650 @opindex fmessage-length
2651 Try to format error messages so that they fit on lines of about @var{n}
2652 characters. The default is 72 characters for @command{g++} and 0 for the rest of
2653 the front ends supported by GCC@. If @var{n} is zero, then no
2654 line-wrapping will be done; each error message will appear on a single
2657 @opindex fdiagnostics-show-location
2658 @item -fdiagnostics-show-location=once
2659 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
2660 reporter to emit @emph{once} source location information; that is, in
2661 case the message is too long to fit on a single physical line and has to
2662 be wrapped, the source location won't be emitted (as prefix) again,
2663 over and over, in subsequent continuation lines. This is the default
2666 @item -fdiagnostics-show-location=every-line
2667 Only meaningful in line-wrapping mode. Instructs the diagnostic
2668 messages reporter to emit the same source location information (as
2669 prefix) for physical lines that result from the process of breaking
2670 a message which is too long to fit on a single line.
2672 @item -fdiagnostics-show-option
2673 @opindex fdiagnostics-show-option
2674 This option instructs the diagnostic machinery to add text to each
2675 diagnostic emitted, which indicates which command line option directly
2676 controls that diagnostic, when such an option is known to the
2677 diagnostic machinery.
2679 @item -Wcoverage-mismatch
2680 @opindex Wcoverage-mismatch
2681 Warn if feedback profiles do not match when using the
2682 @option{-fprofile-use} option.
2683 If a source file was changed between @option{-fprofile-gen} and
2684 @option{-fprofile-use}, the files with the profile feedback can fail
2685 to match the source file and GCC can not use the profile feedback
2686 information. By default, GCC emits an error message in this case.
2687 The option @option{-Wcoverage-mismatch} emits a warning instead of an
2688 error. GCC does not use appropriate feedback profiles, so using this
2689 option can result in poorly optimized code. This option is useful
2690 only in the case of very minor changes such as bug fixes to an
2695 @node Warning Options
2696 @section Options to Request or Suppress Warnings
2697 @cindex options to control warnings
2698 @cindex warning messages
2699 @cindex messages, warning
2700 @cindex suppressing warnings
2702 Warnings are diagnostic messages that report constructions which
2703 are not inherently erroneous but which are risky or suggest there
2704 may have been an error.
2706 The following language-independent options do not enable specific
2707 warnings but control the kinds of diagnostics produced by GCC.
2710 @cindex syntax checking
2712 @opindex fsyntax-only
2713 Check the code for syntax errors, but don't do anything beyond that.
2717 Inhibit all warning messages.
2722 Make all warnings into errors.
2727 Make the specified warning into an error. The specifier for a warning
2728 is appended, for example @option{-Werror=switch} turns the warnings
2729 controlled by @option{-Wswitch} into errors. This switch takes a
2730 negative form, to be used to negate @option{-Werror} for specific
2731 warnings, for example @option{-Wno-error=switch} makes
2732 @option{-Wswitch} warnings not be errors, even when @option{-Werror}
2733 is in effect. You can use the @option{-fdiagnostics-show-option}
2734 option to have each controllable warning amended with the option which
2735 controls it, to determine what to use with this option.
2737 Note that specifying @option{-Werror=}@var{foo} automatically implies
2738 @option{-W}@var{foo}. However, @option{-Wno-error=}@var{foo} does not
2741 @item -Wfatal-errors
2742 @opindex Wfatal-errors
2743 @opindex Wno-fatal-errors
2744 This option causes the compiler to abort compilation on the first error
2745 occurred rather than trying to keep going and printing further error
2750 You can request many specific warnings with options beginning
2751 @samp{-W}, for example @option{-Wimplicit} to request warnings on
2752 implicit declarations. Each of these specific warning options also
2753 has a negative form beginning @samp{-Wno-} to turn off warnings; for
2754 example, @option{-Wno-implicit}. This manual lists only one of the
2755 two forms, whichever is not the default. For further,
2756 language-specific options also refer to @ref{C++ Dialect Options} and
2757 @ref{Objective-C and Objective-C++ Dialect Options}.
2762 Issue all the warnings demanded by strict ISO C and ISO C++;
2763 reject all programs that use forbidden extensions, and some other
2764 programs that do not follow ISO C and ISO C++. For ISO C, follows the
2765 version of the ISO C standard specified by any @option{-std} option used.
2767 Valid ISO C and ISO C++ programs should compile properly with or without
2768 this option (though a rare few will require @option{-ansi} or a
2769 @option{-std} option specifying the required version of ISO C)@. However,
2770 without this option, certain GNU extensions and traditional C and C++
2771 features are supported as well. With this option, they are rejected.
2773 @option{-pedantic} does not cause warning messages for use of the
2774 alternate keywords whose names begin and end with @samp{__}. Pedantic
2775 warnings are also disabled in the expression that follows
2776 @code{__extension__}. However, only system header files should use
2777 these escape routes; application programs should avoid them.
2778 @xref{Alternate Keywords}.
2780 Some users try to use @option{-pedantic} to check programs for strict ISO
2781 C conformance. They soon find that it does not do quite what they want:
2782 it finds some non-ISO practices, but not all---only those for which
2783 ISO C @emph{requires} a diagnostic, and some others for which
2784 diagnostics have been added.
2786 A feature to report any failure to conform to ISO C might be useful in
2787 some instances, but would require considerable additional work and would
2788 be quite different from @option{-pedantic}. We don't have plans to
2789 support such a feature in the near future.
2791 Where the standard specified with @option{-std} represents a GNU
2792 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
2793 corresponding @dfn{base standard}, the version of ISO C on which the GNU
2794 extended dialect is based. Warnings from @option{-pedantic} are given
2795 where they are required by the base standard. (It would not make sense
2796 for such warnings to be given only for features not in the specified GNU
2797 C dialect, since by definition the GNU dialects of C include all
2798 features the compiler supports with the given option, and there would be
2799 nothing to warn about.)
2801 @item -pedantic-errors
2802 @opindex pedantic-errors
2803 Like @option{-pedantic}, except that errors are produced rather than
2809 This enables all the warnings about constructions that some users
2810 consider questionable, and that are easy to avoid (or modify to
2811 prevent the warning), even in conjunction with macros. This also
2812 enables some language-specific warnings described in @ref{C++ Dialect
2813 Options} and @ref{Objective-C and Objective-C++ Dialect Options}.
2815 @option{-Wall} turns on the following warning flags:
2817 @gccoptlist{-Waddress @gol
2818 -Warray-bounds @r{(only with} @option{-O2}@r{)} @gol
2820 -Wchar-subscripts @gol
2821 -Wenum-compare @r{(in C/Objc; this is on by default in C++)} @gol
2823 -Wimplicit-function-declaration @gol
2826 -Wmain @r{(only for C/ObjC and unless} @option{-ffreestanding}@r{)} @gol
2827 -Wmissing-braces @gol
2833 -Wsequence-point @gol
2834 -Wsign-compare @r{(only in C++)} @gol
2835 -Wstrict-aliasing @gol
2836 -Wstrict-overflow=1 @gol
2839 -Wuninitialized @gol
2840 -Wunknown-pragmas @gol
2841 -Wunused-function @gol
2844 -Wunused-variable @gol
2845 -Wvolatile-register-var @gol
2848 Note that some warning flags are not implied by @option{-Wall}. Some of
2849 them warn about constructions that users generally do not consider
2850 questionable, but which occasionally you might wish to check for;
2851 others warn about constructions that are necessary or hard to avoid in
2852 some cases, and there is no simple way to modify the code to suppress
2853 the warning. Some of them are enabled by @option{-Wextra} but many of
2854 them must be enabled individually.
2860 This enables some extra warning flags that are not enabled by
2861 @option{-Wall}. (This option used to be called @option{-W}. The older
2862 name is still supported, but the newer name is more descriptive.)
2864 @gccoptlist{-Wclobbered @gol
2866 -Wignored-qualifiers @gol
2867 -Wmissing-field-initializers @gol
2868 -Wmissing-parameter-type @r{(C only)} @gol
2869 -Wold-style-declaration @r{(C only)} @gol
2870 -Woverride-init @gol
2873 -Wuninitialized @gol
2874 -Wunused-parameter @r{(only with} @option{-Wunused} @r{or} @option{-Wall}@r{)} @gol
2877 The option @option{-Wextra} also prints warning messages for the
2883 A pointer is compared against integer zero with @samp{<}, @samp{<=},
2884 @samp{>}, or @samp{>=}.
2887 (C++ only) An enumerator and a non-enumerator both appear in a
2888 conditional expression.
2891 (C++ only) Ambiguous virtual bases.
2894 (C++ only) Subscripting an array which has been declared @samp{register}.
2897 (C++ only) Taking the address of a variable which has been declared
2901 (C++ only) A base class is not initialized in a derived class' copy
2906 @item -Wchar-subscripts
2907 @opindex Wchar-subscripts
2908 @opindex Wno-char-subscripts
2909 Warn if an array subscript has type @code{char}. This is a common cause
2910 of error, as programmers often forget that this type is signed on some
2912 This warning is enabled by @option{-Wall}.
2916 @opindex Wno-comment
2917 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
2918 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
2919 This warning is enabled by @option{-Wall}.
2924 @opindex ffreestanding
2925 @opindex fno-builtin
2926 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
2927 the arguments supplied have types appropriate to the format string
2928 specified, and that the conversions specified in the format string make
2929 sense. This includes standard functions, and others specified by format
2930 attributes (@pxref{Function Attributes}), in the @code{printf},
2931 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
2932 not in the C standard) families (or other target-specific families).
2933 Which functions are checked without format attributes having been
2934 specified depends on the standard version selected, and such checks of
2935 functions without the attribute specified are disabled by
2936 @option{-ffreestanding} or @option{-fno-builtin}.
2938 The formats are checked against the format features supported by GNU
2939 libc version 2.2. These include all ISO C90 and C99 features, as well
2940 as features from the Single Unix Specification and some BSD and GNU
2941 extensions. Other library implementations may not support all these
2942 features; GCC does not support warning about features that go beyond a
2943 particular library's limitations. However, if @option{-pedantic} is used
2944 with @option{-Wformat}, warnings will be given about format features not
2945 in the selected standard version (but not for @code{strfmon} formats,
2946 since those are not in any version of the C standard). @xref{C Dialect
2947 Options,,Options Controlling C Dialect}.
2949 Since @option{-Wformat} also checks for null format arguments for
2950 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
2952 @option{-Wformat} is included in @option{-Wall}. For more control over some
2953 aspects of format checking, the options @option{-Wformat-y2k},
2954 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
2955 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
2956 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
2959 @opindex Wformat-y2k
2960 @opindex Wno-format-y2k
2961 If @option{-Wformat} is specified, also warn about @code{strftime}
2962 formats which may yield only a two-digit year.
2964 @item -Wno-format-contains-nul
2965 @opindex Wno-format-contains-nul
2966 @opindex Wformat-contains-nul
2967 If @option{-Wformat} is specified, do not warn about format strings that
2970 @item -Wno-format-extra-args
2971 @opindex Wno-format-extra-args
2972 @opindex Wformat-extra-args
2973 If @option{-Wformat} is specified, do not warn about excess arguments to a
2974 @code{printf} or @code{scanf} format function. The C standard specifies
2975 that such arguments are ignored.
2977 Where the unused arguments lie between used arguments that are
2978 specified with @samp{$} operand number specifications, normally
2979 warnings are still given, since the implementation could not know what
2980 type to pass to @code{va_arg} to skip the unused arguments. However,
2981 in the case of @code{scanf} formats, this option will suppress the
2982 warning if the unused arguments are all pointers, since the Single
2983 Unix Specification says that such unused arguments are allowed.
2985 @item -Wno-format-zero-length @r{(C and Objective-C only)}
2986 @opindex Wno-format-zero-length
2987 @opindex Wformat-zero-length
2988 If @option{-Wformat} is specified, do not warn about zero-length formats.
2989 The C standard specifies that zero-length formats are allowed.
2991 @item -Wformat-nonliteral
2992 @opindex Wformat-nonliteral
2993 @opindex Wno-format-nonliteral
2994 If @option{-Wformat} is specified, also warn if the format string is not a
2995 string literal and so cannot be checked, unless the format function
2996 takes its format arguments as a @code{va_list}.
2998 @item -Wformat-security
2999 @opindex Wformat-security
3000 @opindex Wno-format-security
3001 If @option{-Wformat} is specified, also warn about uses of format
3002 functions that represent possible security problems. At present, this
3003 warns about calls to @code{printf} and @code{scanf} functions where the
3004 format string is not a string literal and there are no format arguments,
3005 as in @code{printf (foo);}. This may be a security hole if the format
3006 string came from untrusted input and contains @samp{%n}. (This is
3007 currently a subset of what @option{-Wformat-nonliteral} warns about, but
3008 in future warnings may be added to @option{-Wformat-security} that are not
3009 included in @option{-Wformat-nonliteral}.)
3013 @opindex Wno-format=2
3014 Enable @option{-Wformat} plus format checks not included in
3015 @option{-Wformat}. Currently equivalent to @samp{-Wformat
3016 -Wformat-nonliteral -Wformat-security -Wformat-y2k}.
3018 @item -Wnonnull @r{(C and Objective-C only)}
3020 @opindex Wno-nonnull
3021 Warn about passing a null pointer for arguments marked as
3022 requiring a non-null value by the @code{nonnull} function attribute.
3024 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
3025 can be disabled with the @option{-Wno-nonnull} option.
3027 @item -Wjump-misses-init @r{(C, Objective-C only)}
3028 @opindex Wjump-misses-init
3029 @opindex Wno-jump-misses-init
3030 Warn if a @code{goto} statement or a @code{switch} statement jumps
3031 forward across the initialization of a variable, or jumps backward to a
3032 label after the variable has been initialized. This only warns about
3033 variables which are initialized when they are declared. This warning is
3034 only supported for C and Objective C; in C++ this sort of branch is an
3037 @option{-Wjump-misses-init} is included in @option{-Wall} and
3038 @option{-Wc++-compat}. It can be disabled with the
3039 @option{-Wno-jump-misses-init} option.
3041 @item -Winit-self @r{(C, C++, Objective-C and Objective-C++ only)}
3043 @opindex Wno-init-self
3044 Warn about uninitialized variables which are initialized with themselves.
3045 Note this option can only be used with the @option{-Wuninitialized} option.
3047 For example, GCC will warn about @code{i} being uninitialized in the
3048 following snippet only when @option{-Winit-self} has been specified:
3059 @item -Wimplicit-int @r{(C and Objective-C only)}
3060 @opindex Wimplicit-int
3061 @opindex Wno-implicit-int
3062 Warn when a declaration does not specify a type.
3063 This warning is enabled by @option{-Wall}.
3065 @item -Wimplicit-function-declaration @r{(C and Objective-C only)}
3066 @opindex Wimplicit-function-declaration
3067 @opindex Wno-implicit-function-declaration
3068 Give a warning whenever a function is used before being declared. In
3069 C99 mode (@option{-std=c99} or @option{-std=gnu99}), this warning is
3070 enabled by default and it is made into an error by
3071 @option{-pedantic-errors}. This warning is also enabled by
3076 @opindex Wno-implicit
3077 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
3078 This warning is enabled by @option{-Wall}.
3080 @item -Wignored-qualifiers @r{(C and C++ only)}
3081 @opindex Wignored-qualifiers
3082 @opindex Wno-ignored-qualifiers
3083 Warn if the return type of a function has a type qualifier
3084 such as @code{const}. For ISO C such a type qualifier has no effect,
3085 since the value returned by a function is not an lvalue.
3086 For C++, the warning is only emitted for scalar types or @code{void}.
3087 ISO C prohibits qualified @code{void} return types on function
3088 definitions, so such return types always receive a warning
3089 even without this option.
3091 This warning is also enabled by @option{-Wextra}.
3096 Warn if the type of @samp{main} is suspicious. @samp{main} should be
3097 a function with external linkage, returning int, taking either zero
3098 arguments, two, or three arguments of appropriate types. This warning
3099 is enabled by default in C++ and is enabled by either @option{-Wall}
3100 or @option{-pedantic}.
3102 @item -Wmissing-braces
3103 @opindex Wmissing-braces
3104 @opindex Wno-missing-braces
3105 Warn if an aggregate or union initializer is not fully bracketed. In
3106 the following example, the initializer for @samp{a} is not fully
3107 bracketed, but that for @samp{b} is fully bracketed.
3110 int a[2][2] = @{ 0, 1, 2, 3 @};
3111 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
3114 This warning is enabled by @option{-Wall}.
3116 @item -Wmissing-include-dirs @r{(C, C++, Objective-C and Objective-C++ only)}
3117 @opindex Wmissing-include-dirs
3118 @opindex Wno-missing-include-dirs
3119 Warn if a user-supplied include directory does not exist.
3122 @opindex Wparentheses
3123 @opindex Wno-parentheses
3124 Warn if parentheses are omitted in certain contexts, such
3125 as when there is an assignment in a context where a truth value
3126 is expected, or when operators are nested whose precedence people
3127 often get confused about.
3129 Also warn if a comparison like @samp{x<=y<=z} appears; this is
3130 equivalent to @samp{(x<=y ? 1 : 0) <= z}, which is a different
3131 interpretation from that of ordinary mathematical notation.
3133 Also warn about constructions where there may be confusion to which
3134 @code{if} statement an @code{else} branch belongs. Here is an example of
3149 In C/C++, every @code{else} branch belongs to the innermost possible
3150 @code{if} statement, which in this example is @code{if (b)}. This is
3151 often not what the programmer expected, as illustrated in the above
3152 example by indentation the programmer chose. When there is the
3153 potential for this confusion, GCC will issue a warning when this flag
3154 is specified. To eliminate the warning, add explicit braces around
3155 the innermost @code{if} statement so there is no way the @code{else}
3156 could belong to the enclosing @code{if}. The resulting code would
3173 This warning is enabled by @option{-Wall}.
3175 @item -Wsequence-point
3176 @opindex Wsequence-point
3177 @opindex Wno-sequence-point
3178 Warn about code that may have undefined semantics because of violations
3179 of sequence point rules in the C and C++ standards.
3181 The C and C++ standards defines the order in which expressions in a C/C++
3182 program are evaluated in terms of @dfn{sequence points}, which represent
3183 a partial ordering between the execution of parts of the program: those
3184 executed before the sequence point, and those executed after it. These
3185 occur after the evaluation of a full expression (one which is not part
3186 of a larger expression), after the evaluation of the first operand of a
3187 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
3188 function is called (but after the evaluation of its arguments and the
3189 expression denoting the called function), and in certain other places.
3190 Other than as expressed by the sequence point rules, the order of
3191 evaluation of subexpressions of an expression is not specified. All
3192 these rules describe only a partial order rather than a total order,
3193 since, for example, if two functions are called within one expression
3194 with no sequence point between them, the order in which the functions
3195 are called is not specified. However, the standards committee have
3196 ruled that function calls do not overlap.
3198 It is not specified when between sequence points modifications to the
3199 values of objects take effect. Programs whose behavior depends on this
3200 have undefined behavior; the C and C++ standards specify that ``Between
3201 the previous and next sequence point an object shall have its stored
3202 value modified at most once by the evaluation of an expression.
3203 Furthermore, the prior value shall be read only to determine the value
3204 to be stored.''. If a program breaks these rules, the results on any
3205 particular implementation are entirely unpredictable.
3207 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
3208 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
3209 diagnosed by this option, and it may give an occasional false positive
3210 result, but in general it has been found fairly effective at detecting
3211 this sort of problem in programs.
3213 The standard is worded confusingly, therefore there is some debate
3214 over the precise meaning of the sequence point rules in subtle cases.
3215 Links to discussions of the problem, including proposed formal
3216 definitions, may be found on the GCC readings page, at
3217 @w{@uref{http://gcc.gnu.org/readings.html}}.
3219 This warning is enabled by @option{-Wall} for C and C++.
3222 @opindex Wreturn-type
3223 @opindex Wno-return-type
3224 Warn whenever a function is defined with a return-type that defaults
3225 to @code{int}. Also warn about any @code{return} statement with no
3226 return-value in a function whose return-type is not @code{void}
3227 (falling off the end of the function body is considered returning
3228 without a value), and about a @code{return} statement with an
3229 expression in a function whose return-type is @code{void}.
3231 For C++, a function without return type always produces a diagnostic
3232 message, even when @option{-Wno-return-type} is specified. The only
3233 exceptions are @samp{main} and functions defined in system headers.
3235 This warning is enabled by @option{-Wall}.
3240 Warn whenever a @code{switch} statement has an index of enumerated type
3241 and lacks a @code{case} for one or more of the named codes of that
3242 enumeration. (The presence of a @code{default} label prevents this
3243 warning.) @code{case} labels outside the enumeration range also
3244 provoke warnings when this option is used (even if there is a
3245 @code{default} label).
3246 This warning is enabled by @option{-Wall}.
3248 @item -Wswitch-default
3249 @opindex Wswitch-default
3250 @opindex Wno-switch-default
3251 Warn whenever a @code{switch} statement does not have a @code{default}
3255 @opindex Wswitch-enum
3256 @opindex Wno-switch-enum
3257 Warn whenever a @code{switch} statement has an index of enumerated type
3258 and lacks a @code{case} for one or more of the named codes of that
3259 enumeration. @code{case} labels outside the enumeration range also
3260 provoke warnings when this option is used. The only difference
3261 between @option{-Wswitch} and this option is that this option gives a
3262 warning about an omitted enumeration code even if there is a
3263 @code{default} label.
3265 @item -Wsync-nand @r{(C and C++ only)}
3267 @opindex Wno-sync-nand
3268 Warn when @code{__sync_fetch_and_nand} and @code{__sync_nand_and_fetch}
3269 built-in functions are used. These functions changed semantics in GCC 4.4.
3273 @opindex Wno-trigraphs
3274 Warn if any trigraphs are encountered that might change the meaning of
3275 the program (trigraphs within comments are not warned about).
3276 This warning is enabled by @option{-Wall}.
3278 @item -Wunused-function
3279 @opindex Wunused-function
3280 @opindex Wno-unused-function
3281 Warn whenever a static function is declared but not defined or a
3282 non-inline static function is unused.
3283 This warning is enabled by @option{-Wall}.
3285 @item -Wunused-label
3286 @opindex Wunused-label
3287 @opindex Wno-unused-label
3288 Warn whenever a label is declared but not used.
3289 This warning is enabled by @option{-Wall}.
3291 To suppress this warning use the @samp{unused} attribute
3292 (@pxref{Variable Attributes}).
3294 @item -Wunused-parameter
3295 @opindex Wunused-parameter
3296 @opindex Wno-unused-parameter
3297 Warn whenever a function parameter is unused aside from its declaration.
3299 To suppress this warning use the @samp{unused} attribute
3300 (@pxref{Variable Attributes}).
3302 @item -Wno-unused-result
3303 @opindex Wunused-result
3304 @opindex Wno-unused-result
3305 Do not warn if a caller of a function marked with attribute
3306 @code{warn_unused_result} (@pxref{Variable Attributes}) does not use
3307 its return value. The default is @option{-Wunused-result}.
3309 @item -Wunused-variable
3310 @opindex Wunused-variable
3311 @opindex Wno-unused-variable
3312 Warn whenever a local variable or non-constant static variable is unused
3313 aside from its declaration.
3314 This warning is enabled by @option{-Wall}.
3316 To suppress this warning use the @samp{unused} attribute
3317 (@pxref{Variable Attributes}).
3319 @item -Wunused-value
3320 @opindex Wunused-value
3321 @opindex Wno-unused-value
3322 Warn whenever a statement computes a result that is explicitly not
3323 used. To suppress this warning cast the unused expression to
3324 @samp{void}. This includes an expression-statement or the left-hand
3325 side of a comma expression that contains no side effects. For example,
3326 an expression such as @samp{x[i,j]} will cause a warning, while
3327 @samp{x[(void)i,j]} will not.
3329 This warning is enabled by @option{-Wall}.
3334 All the above @option{-Wunused} options combined.
3336 In order to get a warning about an unused function parameter, you must
3337 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
3338 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
3340 @item -Wuninitialized
3341 @opindex Wuninitialized
3342 @opindex Wno-uninitialized
3343 Warn if an automatic variable is used without first being initialized
3344 or if a variable may be clobbered by a @code{setjmp} call. In C++,
3345 warn if a non-static reference or non-static @samp{const} member
3346 appears in a class without constructors.
3348 If you want to warn about code which uses the uninitialized value of the
3349 variable in its own initializer, use the @option{-Winit-self} option.
3351 These warnings occur for individual uninitialized or clobbered
3352 elements of structure, union or array variables as well as for
3353 variables which are uninitialized or clobbered as a whole. They do
3354 not occur for variables or elements declared @code{volatile}. Because
3355 these warnings depend on optimization, the exact variables or elements
3356 for which there are warnings will depend on the precise optimization
3357 options and version of GCC used.
3359 Note that there may be no warning about a variable that is used only
3360 to compute a value that itself is never used, because such
3361 computations may be deleted by data flow analysis before the warnings
3364 These warnings are made optional because GCC is not smart
3365 enough to see all the reasons why the code might be correct
3366 despite appearing to have an error. Here is one example of how
3387 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
3388 always initialized, but GCC doesn't know this. Here is
3389 another common case:
3394 if (change_y) save_y = y, y = new_y;
3396 if (change_y) y = save_y;
3401 This has no bug because @code{save_y} is used only if it is set.
3403 @cindex @code{longjmp} warnings
3404 This option also warns when a non-volatile automatic variable might be
3405 changed by a call to @code{longjmp}. These warnings as well are possible
3406 only in optimizing compilation.
3408 The compiler sees only the calls to @code{setjmp}. It cannot know
3409 where @code{longjmp} will be called; in fact, a signal handler could
3410 call it at any point in the code. As a result, you may get a warning
3411 even when there is in fact no problem because @code{longjmp} cannot
3412 in fact be called at the place which would cause a problem.
3414 Some spurious warnings can be avoided if you declare all the functions
3415 you use that never return as @code{noreturn}. @xref{Function
3418 This warning is enabled by @option{-Wall} or @option{-Wextra}.
3420 @item -Wunknown-pragmas
3421 @opindex Wunknown-pragmas
3422 @opindex Wno-unknown-pragmas
3423 @cindex warning for unknown pragmas
3424 @cindex unknown pragmas, warning
3425 @cindex pragmas, warning of unknown
3426 Warn when a #pragma directive is encountered which is not understood by
3427 GCC@. If this command line option is used, warnings will even be issued
3428 for unknown pragmas in system header files. This is not the case if
3429 the warnings were only enabled by the @option{-Wall} command line option.
3432 @opindex Wno-pragmas
3434 Do not warn about misuses of pragmas, such as incorrect parameters,
3435 invalid syntax, or conflicts between pragmas. See also
3436 @samp{-Wunknown-pragmas}.
3438 @item -Wstrict-aliasing
3439 @opindex Wstrict-aliasing
3440 @opindex Wno-strict-aliasing
3441 This option is only active when @option{-fstrict-aliasing} is active.
3442 It warns about code which might break the strict aliasing rules that the
3443 compiler is using for optimization. The warning does not catch all
3444 cases, but does attempt to catch the more common pitfalls. It is
3445 included in @option{-Wall}.
3446 It is equivalent to @option{-Wstrict-aliasing=3}
3448 @item -Wstrict-aliasing=n
3449 @opindex Wstrict-aliasing=n
3450 @opindex Wno-strict-aliasing=n
3451 This option is only active when @option{-fstrict-aliasing} is active.
3452 It warns about code which might break the strict aliasing rules that the
3453 compiler is using for optimization.
3454 Higher levels correspond to higher accuracy (fewer false positives).
3455 Higher levels also correspond to more effort, similar to the way -O works.
3456 @option{-Wstrict-aliasing} is equivalent to @option{-Wstrict-aliasing=n},
3459 Level 1: Most aggressive, quick, least accurate.
3460 Possibly useful when higher levels
3461 do not warn but -fstrict-aliasing still breaks the code, as it has very few
3462 false negatives. However, it has many false positives.
3463 Warns for all pointer conversions between possibly incompatible types,
3464 even if never dereferenced. Runs in the frontend only.
3466 Level 2: Aggressive, quick, not too precise.
3467 May still have many false positives (not as many as level 1 though),
3468 and few false negatives (but possibly more than level 1).
3469 Unlike level 1, it only warns when an address is taken. Warns about
3470 incomplete types. Runs in the frontend only.
3472 Level 3 (default for @option{-Wstrict-aliasing}):
3473 Should have very few false positives and few false
3474 negatives. Slightly slower than levels 1 or 2 when optimization is enabled.
3475 Takes care of the common punn+dereference pattern in the frontend:
3476 @code{*(int*)&some_float}.
3477 If optimization is enabled, it also runs in the backend, where it deals
3478 with multiple statement cases using flow-sensitive points-to information.
3479 Only warns when the converted pointer is dereferenced.
3480 Does not warn about incomplete types.
3482 @item -Wstrict-overflow
3483 @itemx -Wstrict-overflow=@var{n}
3484 @opindex Wstrict-overflow
3485 @opindex Wno-strict-overflow
3486 This option is only active when @option{-fstrict-overflow} is active.
3487 It warns about cases where the compiler optimizes based on the
3488 assumption that signed overflow does not occur. Note that it does not
3489 warn about all cases where the code might overflow: it only warns
3490 about cases where the compiler implements some optimization. Thus
3491 this warning depends on the optimization level.
3493 An optimization which assumes that signed overflow does not occur is
3494 perfectly safe if the values of the variables involved are such that
3495 overflow never does, in fact, occur. Therefore this warning can
3496 easily give a false positive: a warning about code which is not
3497 actually a problem. To help focus on important issues, several
3498 warning levels are defined. No warnings are issued for the use of
3499 undefined signed overflow when estimating how many iterations a loop
3500 will require, in particular when determining whether a loop will be
3504 @item -Wstrict-overflow=1
3505 Warn about cases which are both questionable and easy to avoid. For
3506 example: @code{x + 1 > x}; with @option{-fstrict-overflow}, the
3507 compiler will simplify this to @code{1}. This level of
3508 @option{-Wstrict-overflow} is enabled by @option{-Wall}; higher levels
3509 are not, and must be explicitly requested.
3511 @item -Wstrict-overflow=2
3512 Also warn about other cases where a comparison is simplified to a
3513 constant. For example: @code{abs (x) >= 0}. This can only be
3514 simplified when @option{-fstrict-overflow} is in effect, because
3515 @code{abs (INT_MIN)} overflows to @code{INT_MIN}, which is less than
3516 zero. @option{-Wstrict-overflow} (with no level) is the same as
3517 @option{-Wstrict-overflow=2}.
3519 @item -Wstrict-overflow=3
3520 Also warn about other cases where a comparison is simplified. For
3521 example: @code{x + 1 > 1} will be simplified to @code{x > 0}.
3523 @item -Wstrict-overflow=4
3524 Also warn about other simplifications not covered by the above cases.
3525 For example: @code{(x * 10) / 5} will be simplified to @code{x * 2}.
3527 @item -Wstrict-overflow=5
3528 Also warn about cases where the compiler reduces the magnitude of a
3529 constant involved in a comparison. For example: @code{x + 2 > y} will
3530 be simplified to @code{x + 1 >= y}. This is reported only at the
3531 highest warning level because this simplification applies to many
3532 comparisons, so this warning level will give a very large number of
3536 @item -Warray-bounds
3537 @opindex Wno-array-bounds
3538 @opindex Warray-bounds
3539 This option is only active when @option{-ftree-vrp} is active
3540 (default for -O2 and above). It warns about subscripts to arrays
3541 that are always out of bounds. This warning is enabled by @option{-Wall}.
3543 @item -Wno-div-by-zero
3544 @opindex Wno-div-by-zero
3545 @opindex Wdiv-by-zero
3546 Do not warn about compile-time integer division by zero. Floating point
3547 division by zero is not warned about, as it can be a legitimate way of
3548 obtaining infinities and NaNs.
3550 @item -Wsystem-headers
3551 @opindex Wsystem-headers
3552 @opindex Wno-system-headers
3553 @cindex warnings from system headers
3554 @cindex system headers, warnings from
3555 Print warning messages for constructs found in system header files.
3556 Warnings from system headers are normally suppressed, on the assumption
3557 that they usually do not indicate real problems and would only make the
3558 compiler output harder to read. Using this command line option tells
3559 GCC to emit warnings from system headers as if they occurred in user
3560 code. However, note that using @option{-Wall} in conjunction with this
3561 option will @emph{not} warn about unknown pragmas in system
3562 headers---for that, @option{-Wunknown-pragmas} must also be used.
3565 @opindex Wfloat-equal
3566 @opindex Wno-float-equal
3567 Warn if floating point values are used in equality comparisons.
3569 The idea behind this is that sometimes it is convenient (for the
3570 programmer) to consider floating-point values as approximations to
3571 infinitely precise real numbers. If you are doing this, then you need
3572 to compute (by analyzing the code, or in some other way) the maximum or
3573 likely maximum error that the computation introduces, and allow for it
3574 when performing comparisons (and when producing output, but that's a
3575 different problem). In particular, instead of testing for equality, you
3576 would check to see whether the two values have ranges that overlap; and
3577 this is done with the relational operators, so equality comparisons are
3580 @item -Wtraditional @r{(C and Objective-C only)}
3581 @opindex Wtraditional
3582 @opindex Wno-traditional
3583 Warn about certain constructs that behave differently in traditional and
3584 ISO C@. Also warn about ISO C constructs that have no traditional C
3585 equivalent, and/or problematic constructs which should be avoided.
3589 Macro parameters that appear within string literals in the macro body.
3590 In traditional C macro replacement takes place within string literals,
3591 but does not in ISO C@.
3594 In traditional C, some preprocessor directives did not exist.
3595 Traditional preprocessors would only consider a line to be a directive
3596 if the @samp{#} appeared in column 1 on the line. Therefore
3597 @option{-Wtraditional} warns about directives that traditional C
3598 understands but would ignore because the @samp{#} does not appear as the
3599 first character on the line. It also suggests you hide directives like
3600 @samp{#pragma} not understood by traditional C by indenting them. Some
3601 traditional implementations would not recognize @samp{#elif}, so it
3602 suggests avoiding it altogether.
3605 A function-like macro that appears without arguments.
3608 The unary plus operator.
3611 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
3612 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
3613 constants.) Note, these suffixes appear in macros defined in the system
3614 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
3615 Use of these macros in user code might normally lead to spurious
3616 warnings, however GCC's integrated preprocessor has enough context to
3617 avoid warning in these cases.
3620 A function declared external in one block and then used after the end of
3624 A @code{switch} statement has an operand of type @code{long}.
3627 A non-@code{static} function declaration follows a @code{static} one.
3628 This construct is not accepted by some traditional C compilers.
3631 The ISO type of an integer constant has a different width or
3632 signedness from its traditional type. This warning is only issued if
3633 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
3634 typically represent bit patterns, are not warned about.
3637 Usage of ISO string concatenation is detected.
3640 Initialization of automatic aggregates.
3643 Identifier conflicts with labels. Traditional C lacks a separate
3644 namespace for labels.
3647 Initialization of unions. If the initializer is zero, the warning is
3648 omitted. This is done under the assumption that the zero initializer in
3649 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
3650 initializer warnings and relies on default initialization to zero in the
3654 Conversions by prototypes between fixed/floating point values and vice
3655 versa. The absence of these prototypes when compiling with traditional
3656 C would cause serious problems. This is a subset of the possible
3657 conversion warnings, for the full set use @option{-Wtraditional-conversion}.
3660 Use of ISO C style function definitions. This warning intentionally is
3661 @emph{not} issued for prototype declarations or variadic functions
3662 because these ISO C features will appear in your code when using
3663 libiberty's traditional C compatibility macros, @code{PARAMS} and
3664 @code{VPARAMS}. This warning is also bypassed for nested functions
3665 because that feature is already a GCC extension and thus not relevant to
3666 traditional C compatibility.
3669 @item -Wtraditional-conversion @r{(C and Objective-C only)}
3670 @opindex Wtraditional-conversion
3671 @opindex Wno-traditional-conversion
3672 Warn if a prototype causes a type conversion that is different from what
3673 would happen to the same argument in the absence of a prototype. This
3674 includes conversions of fixed point to floating and vice versa, and
3675 conversions changing the width or signedness of a fixed point argument
3676 except when the same as the default promotion.
3678 @item -Wdeclaration-after-statement @r{(C and Objective-C only)}
3679 @opindex Wdeclaration-after-statement
3680 @opindex Wno-declaration-after-statement
3681 Warn when a declaration is found after a statement in a block. This
3682 construct, known from C++, was introduced with ISO C99 and is by default
3683 allowed in GCC@. It is not supported by ISO C90 and was not supported by
3684 GCC versions before GCC 3.0. @xref{Mixed Declarations}.
3689 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
3691 @item -Wno-endif-labels
3692 @opindex Wno-endif-labels
3693 @opindex Wendif-labels
3694 Do not warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
3699 Warn whenever a local variable shadows another local variable, parameter or
3700 global variable or whenever a built-in function is shadowed.
3702 @item -Wlarger-than=@var{len}
3703 @opindex Wlarger-than=@var{len}
3704 @opindex Wlarger-than-@var{len}
3705 Warn whenever an object of larger than @var{len} bytes is defined.
3707 @item -Wframe-larger-than=@var{len}
3708 @opindex Wframe-larger-than
3709 Warn if the size of a function frame is larger than @var{len} bytes.
3710 The computation done to determine the stack frame size is approximate
3711 and not conservative.
3712 The actual requirements may be somewhat greater than @var{len}
3713 even if you do not get a warning. In addition, any space allocated
3714 via @code{alloca}, variable-length arrays, or related constructs
3715 is not included by the compiler when determining
3716 whether or not to issue a warning.
3718 @item -Wunsafe-loop-optimizations
3719 @opindex Wunsafe-loop-optimizations
3720 @opindex Wno-unsafe-loop-optimizations
3721 Warn if the loop cannot be optimized because the compiler could not
3722 assume anything on the bounds of the loop indices. With
3723 @option{-funsafe-loop-optimizations} warn if the compiler made
3726 @item -Wno-pedantic-ms-format @r{(MinGW targets only)}
3727 @opindex Wno-pedantic-ms-format
3728 @opindex Wpedantic-ms-format
3729 Disables the warnings about non-ISO @code{printf} / @code{scanf} format
3730 width specifiers @code{I32}, @code{I64}, and @code{I} used on Windows targets
3731 depending on the MS runtime, when you are using the options @option{-Wformat}
3732 and @option{-pedantic} without gnu-extensions.
3734 @item -Wpointer-arith
3735 @opindex Wpointer-arith
3736 @opindex Wno-pointer-arith
3737 Warn about anything that depends on the ``size of'' a function type or
3738 of @code{void}. GNU C assigns these types a size of 1, for
3739 convenience in calculations with @code{void *} pointers and pointers
3740 to functions. In C++, warn also when an arithmetic operation involves
3741 @code{NULL}. This warning is also enabled by @option{-pedantic}.
3744 @opindex Wtype-limits
3745 @opindex Wno-type-limits
3746 Warn if a comparison is always true or always false due to the limited
3747 range of the data type, but do not warn for constant expressions. For
3748 example, warn if an unsigned variable is compared against zero with
3749 @samp{<} or @samp{>=}. This warning is also enabled by
3752 @item -Wbad-function-cast @r{(C and Objective-C only)}
3753 @opindex Wbad-function-cast
3754 @opindex Wno-bad-function-cast
3755 Warn whenever a function call is cast to a non-matching type.
3756 For example, warn if @code{int malloc()} is cast to @code{anything *}.
3758 @item -Wc++-compat @r{(C and Objective-C only)}
3759 Warn about ISO C constructs that are outside of the common subset of
3760 ISO C and ISO C++, e.g.@: request for implicit conversion from
3761 @code{void *} to a pointer to non-@code{void} type.
3763 @item -Wc++0x-compat @r{(C++ and Objective-C++ only)}
3764 Warn about C++ constructs whose meaning differs between ISO C++ 1998 and
3765 ISO C++ 200x, e.g., identifiers in ISO C++ 1998 that will become keywords
3766 in ISO C++ 200x. This warning is enabled by @option{-Wall}.
3770 @opindex Wno-cast-qual
3771 Warn whenever a pointer is cast so as to remove a type qualifier from
3772 the target type. For example, warn if a @code{const char *} is cast
3773 to an ordinary @code{char *}.
3775 Also warn when making a cast which introduces a type qualifier in an
3776 unsafe way. For example, casting @code{char **} to @code{const char **}
3777 is unsafe, as in this example:
3780 /* p is char ** value. */
3781 const char **q = (const char **) p;
3782 /* Assignment of readonly string to const char * is OK. */
3784 /* Now char** pointer points to read-only memory. */
3789 @opindex Wcast-align
3790 @opindex Wno-cast-align
3791 Warn whenever a pointer is cast such that the required alignment of the
3792 target is increased. For example, warn if a @code{char *} is cast to
3793 an @code{int *} on machines where integers can only be accessed at
3794 two- or four-byte boundaries.
3796 @item -Wwrite-strings
3797 @opindex Wwrite-strings
3798 @opindex Wno-write-strings
3799 When compiling C, give string constants the type @code{const
3800 char[@var{length}]} so that copying the address of one into a
3801 non-@code{const} @code{char *} pointer will get a warning. These
3802 warnings will help you find at compile time code that can try to write
3803 into a string constant, but only if you have been very careful about
3804 using @code{const} in declarations and prototypes. Otherwise, it will
3805 just be a nuisance. This is why we did not make @option{-Wall} request
3808 When compiling C++, warn about the deprecated conversion from string
3809 literals to @code{char *}. This warning is enabled by default for C++
3814 @opindex Wno-clobbered
3815 Warn for variables that might be changed by @samp{longjmp} or
3816 @samp{vfork}. This warning is also enabled by @option{-Wextra}.
3819 @opindex Wconversion
3820 @opindex Wno-conversion
3821 Warn for implicit conversions that may alter a value. This includes
3822 conversions between real and integer, like @code{abs (x)} when
3823 @code{x} is @code{double}; conversions between signed and unsigned,
3824 like @code{unsigned ui = -1}; and conversions to smaller types, like
3825 @code{sqrtf (M_PI)}. Do not warn for explicit casts like @code{abs
3826 ((int) x)} and @code{ui = (unsigned) -1}, or if the value is not
3827 changed by the conversion like in @code{abs (2.0)}. Warnings about
3828 conversions between signed and unsigned integers can be disabled by
3829 using @option{-Wno-sign-conversion}.
3831 For C++, also warn for conversions between @code{NULL} and non-pointer
3832 types; confusing overload resolution for user-defined conversions; and
3833 conversions that will never use a type conversion operator:
3834 conversions to @code{void}, the same type, a base class or a reference
3835 to them. Warnings about conversions between signed and unsigned
3836 integers are disabled by default in C++ unless
3837 @option{-Wsign-conversion} is explicitly enabled.
3840 @opindex Wempty-body
3841 @opindex Wno-empty-body
3842 Warn if an empty body occurs in an @samp{if}, @samp{else} or @samp{do
3843 while} statement. This warning is also enabled by @option{-Wextra}.
3845 @item -Wenum-compare
3846 @opindex Wenum-compare
3847 @opindex Wno-enum-compare
3848 Warn about a comparison between values of different enum types. In C++
3849 this warning is enabled by default. In C this warning is enabled by
3852 @item -Wsign-compare
3853 @opindex Wsign-compare
3854 @opindex Wno-sign-compare
3855 @cindex warning for comparison of signed and unsigned values
3856 @cindex comparison of signed and unsigned values, warning
3857 @cindex signed and unsigned values, comparison warning
3858 Warn when a comparison between signed and unsigned values could produce
3859 an incorrect result when the signed value is converted to unsigned.
3860 This warning is also enabled by @option{-Wextra}; to get the other warnings
3861 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
3863 @item -Wsign-conversion
3864 @opindex Wsign-conversion
3865 @opindex Wno-sign-conversion
3866 Warn for implicit conversions that may change the sign of an integer
3867 value, like assigning a signed integer expression to an unsigned
3868 integer variable. An explicit cast silences the warning. In C, this
3869 option is enabled also by @option{-Wconversion}.
3873 @opindex Wno-address
3874 Warn about suspicious uses of memory addresses. These include using
3875 the address of a function in a conditional expression, such as
3876 @code{void func(void); if (func)}, and comparisons against the memory
3877 address of a string literal, such as @code{if (x == "abc")}. Such
3878 uses typically indicate a programmer error: the address of a function
3879 always evaluates to true, so their use in a conditional usually
3880 indicate that the programmer forgot the parentheses in a function
3881 call; and comparisons against string literals result in unspecified
3882 behavior and are not portable in C, so they usually indicate that the
3883 programmer intended to use @code{strcmp}. This warning is enabled by
3887 @opindex Wlogical-op
3888 @opindex Wno-logical-op
3889 Warn about suspicious uses of logical operators in expressions.
3890 This includes using logical operators in contexts where a
3891 bit-wise operator is likely to be expected.
3893 @item -Waggregate-return
3894 @opindex Waggregate-return
3895 @opindex Wno-aggregate-return
3896 Warn if any functions that return structures or unions are defined or
3897 called. (In languages where you can return an array, this also elicits
3900 @item -Wno-attributes
3901 @opindex Wno-attributes
3902 @opindex Wattributes
3903 Do not warn if an unexpected @code{__attribute__} is used, such as
3904 unrecognized attributes, function attributes applied to variables,
3905 etc. This will not stop errors for incorrect use of supported
3908 @item -Wno-builtin-macro-redefined
3909 @opindex Wno-builtin-macro-redefined
3910 @opindex Wbuiltin-macro-redefined
3911 Do not warn if certain built-in macros are redefined. This suppresses
3912 warnings for redefinition of @code{__TIMESTAMP__}, @code{__TIME__},
3913 @code{__DATE__}, @code{__FILE__}, and @code{__BASE_FILE__}.
3915 @item -Wstrict-prototypes @r{(C and Objective-C only)}
3916 @opindex Wstrict-prototypes
3917 @opindex Wno-strict-prototypes
3918 Warn if a function is declared or defined without specifying the
3919 argument types. (An old-style function definition is permitted without
3920 a warning if preceded by a declaration which specifies the argument
3923 @item -Wold-style-declaration @r{(C and Objective-C only)}
3924 @opindex Wold-style-declaration
3925 @opindex Wno-old-style-declaration
3926 Warn for obsolescent usages, according to the C Standard, in a
3927 declaration. For example, warn if storage-class specifiers like
3928 @code{static} are not the first things in a declaration. This warning
3929 is also enabled by @option{-Wextra}.
3931 @item -Wold-style-definition @r{(C and Objective-C only)}
3932 @opindex Wold-style-definition
3933 @opindex Wno-old-style-definition
3934 Warn if an old-style function definition is used. A warning is given
3935 even if there is a previous prototype.
3937 @item -Wmissing-parameter-type @r{(C and Objective-C only)}
3938 @opindex Wmissing-parameter-type
3939 @opindex Wno-missing-parameter-type
3940 A function parameter is declared without a type specifier in K&R-style
3947 This warning is also enabled by @option{-Wextra}.
3949 @item -Wmissing-prototypes @r{(C and Objective-C only)}
3950 @opindex Wmissing-prototypes
3951 @opindex Wno-missing-prototypes
3952 Warn if a global function is defined without a previous prototype
3953 declaration. This warning is issued even if the definition itself
3954 provides a prototype. The aim is to detect global functions that fail
3955 to be declared in header files.
3957 @item -Wmissing-declarations
3958 @opindex Wmissing-declarations
3959 @opindex Wno-missing-declarations
3960 Warn if a global function is defined without a previous declaration.
3961 Do so even if the definition itself provides a prototype.
3962 Use this option to detect global functions that are not declared in
3963 header files. In C++, no warnings are issued for function templates,
3964 or for inline functions, or for functions in anonymous namespaces.
3966 @item -Wmissing-field-initializers
3967 @opindex Wmissing-field-initializers
3968 @opindex Wno-missing-field-initializers
3972 Warn if a structure's initializer has some fields missing. For
3973 example, the following code would cause such a warning, because
3974 @code{x.h} is implicitly zero:
3977 struct s @{ int f, g, h; @};
3978 struct s x = @{ 3, 4 @};
3981 This option does not warn about designated initializers, so the following
3982 modification would not trigger a warning:
3985 struct s @{ int f, g, h; @};
3986 struct s x = @{ .f = 3, .g = 4 @};
3989 This warning is included in @option{-Wextra}. To get other @option{-Wextra}
3990 warnings without this one, use @samp{-Wextra -Wno-missing-field-initializers}.
3992 @item -Wmissing-noreturn
3993 @opindex Wmissing-noreturn
3994 @opindex Wno-missing-noreturn
3995 Warn about functions which might be candidates for attribute @code{noreturn}.
3996 Note these are only possible candidates, not absolute ones. Care should
3997 be taken to manually verify functions actually do not ever return before
3998 adding the @code{noreturn} attribute, otherwise subtle code generation
3999 bugs could be introduced. You will not get a warning for @code{main} in
4000 hosted C environments.
4002 @item -Wmissing-format-attribute
4003 @opindex Wmissing-format-attribute
4004 @opindex Wno-missing-format-attribute
4007 Warn about function pointers which might be candidates for @code{format}
4008 attributes. Note these are only possible candidates, not absolute ones.
4009 GCC will guess that function pointers with @code{format} attributes that
4010 are used in assignment, initialization, parameter passing or return
4011 statements should have a corresponding @code{format} attribute in the
4012 resulting type. I.e.@: the left-hand side of the assignment or
4013 initialization, the type of the parameter variable, or the return type
4014 of the containing function respectively should also have a @code{format}
4015 attribute to avoid the warning.
4017 GCC will also warn about function definitions which might be
4018 candidates for @code{format} attributes. Again, these are only
4019 possible candidates. GCC will guess that @code{format} attributes
4020 might be appropriate for any function that calls a function like
4021 @code{vprintf} or @code{vscanf}, but this might not always be the
4022 case, and some functions for which @code{format} attributes are
4023 appropriate may not be detected.
4025 @item -Wno-multichar
4026 @opindex Wno-multichar
4028 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
4029 Usually they indicate a typo in the user's code, as they have
4030 implementation-defined values, and should not be used in portable code.
4032 @item -Wnormalized=<none|id|nfc|nfkc>
4033 @opindex Wnormalized=
4036 @cindex character set, input normalization
4037 In ISO C and ISO C++, two identifiers are different if they are
4038 different sequences of characters. However, sometimes when characters
4039 outside the basic ASCII character set are used, you can have two
4040 different character sequences that look the same. To avoid confusion,
4041 the ISO 10646 standard sets out some @dfn{normalization rules} which
4042 when applied ensure that two sequences that look the same are turned into
4043 the same sequence. GCC can warn you if you are using identifiers which
4044 have not been normalized; this option controls that warning.
4046 There are four levels of warning that GCC supports. The default is
4047 @option{-Wnormalized=nfc}, which warns about any identifier which is
4048 not in the ISO 10646 ``C'' normalized form, @dfn{NFC}. NFC is the
4049 recommended form for most uses.
4051 Unfortunately, there are some characters which ISO C and ISO C++ allow
4052 in identifiers that when turned into NFC aren't allowable as
4053 identifiers. That is, there's no way to use these symbols in portable
4054 ISO C or C++ and have all your identifiers in NFC@.
4055 @option{-Wnormalized=id} suppresses the warning for these characters.
4056 It is hoped that future versions of the standards involved will correct
4057 this, which is why this option is not the default.
4059 You can switch the warning off for all characters by writing
4060 @option{-Wnormalized=none}. You would only want to do this if you
4061 were using some other normalization scheme (like ``D''), because
4062 otherwise you can easily create bugs that are literally impossible to see.
4064 Some characters in ISO 10646 have distinct meanings but look identical
4065 in some fonts or display methodologies, especially once formatting has
4066 been applied. For instance @code{\u207F}, ``SUPERSCRIPT LATIN SMALL
4067 LETTER N'', will display just like a regular @code{n} which has been
4068 placed in a superscript. ISO 10646 defines the @dfn{NFKC}
4069 normalization scheme to convert all these into a standard form as
4070 well, and GCC will warn if your code is not in NFKC if you use
4071 @option{-Wnormalized=nfkc}. This warning is comparable to warning
4072 about every identifier that contains the letter O because it might be
4073 confused with the digit 0, and so is not the default, but may be
4074 useful as a local coding convention if the programming environment is
4075 unable to be fixed to display these characters distinctly.
4077 @item -Wno-deprecated
4078 @opindex Wno-deprecated
4079 @opindex Wdeprecated
4080 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
4082 @item -Wno-deprecated-declarations
4083 @opindex Wno-deprecated-declarations
4084 @opindex Wdeprecated-declarations
4085 Do not warn about uses of functions (@pxref{Function Attributes}),
4086 variables (@pxref{Variable Attributes}), and types (@pxref{Type
4087 Attributes}) marked as deprecated by using the @code{deprecated}
4091 @opindex Wno-overflow
4093 Do not warn about compile-time overflow in constant expressions.
4095 @item -Woverride-init @r{(C and Objective-C only)}
4096 @opindex Woverride-init
4097 @opindex Wno-override-init
4101 Warn if an initialized field without side effects is overridden when
4102 using designated initializers (@pxref{Designated Inits, , Designated
4105 This warning is included in @option{-Wextra}. To get other
4106 @option{-Wextra} warnings without this one, use @samp{-Wextra
4107 -Wno-override-init}.
4112 Warn if a structure is given the packed attribute, but the packed
4113 attribute has no effect on the layout or size of the structure.
4114 Such structures may be mis-aligned for little benefit. For
4115 instance, in this code, the variable @code{f.x} in @code{struct bar}
4116 will be misaligned even though @code{struct bar} does not itself
4117 have the packed attribute:
4124 @} __attribute__((packed));
4132 @item -Wpacked-bitfield-compat
4133 @opindex Wpacked-bitfield-compat
4134 @opindex Wno-packed-bitfield-compat
4135 The 4.1, 4.2 and 4.3 series of GCC ignore the @code{packed} attribute
4136 on bit-fields of type @code{char}. This has been fixed in GCC 4.4 but
4137 the change can lead to differences in the structure layout. GCC
4138 informs you when the offset of such a field has changed in GCC 4.4.
4139 For example there is no longer a 4-bit padding between field @code{a}
4140 and @code{b} in this structure:
4147 @} __attribute__ ((packed));
4150 This warning is enabled by default. Use
4151 @option{-Wno-packed-bitfield-compat} to disable this warning.
4156 Warn if padding is included in a structure, either to align an element
4157 of the structure or to align the whole structure. Sometimes when this
4158 happens it is possible to rearrange the fields of the structure to
4159 reduce the padding and so make the structure smaller.
4161 @item -Wredundant-decls
4162 @opindex Wredundant-decls
4163 @opindex Wno-redundant-decls
4164 Warn if anything is declared more than once in the same scope, even in
4165 cases where multiple declaration is valid and changes nothing.
4167 @item -Wnested-externs @r{(C and Objective-C only)}
4168 @opindex Wnested-externs
4169 @opindex Wno-nested-externs
4170 Warn if an @code{extern} declaration is encountered within a function.
4172 @item -Wunreachable-code
4173 @opindex Wunreachable-code
4174 @opindex Wno-unreachable-code
4175 Warn if the compiler detects that code will never be executed.
4177 This option is intended to warn when the compiler detects that at
4178 least a whole line of source code will never be executed, because
4179 some condition is never satisfied or because it is after a
4180 procedure that never returns.
4182 It is possible for this option to produce a warning even though there
4183 are circumstances under which part of the affected line can be executed,
4184 so care should be taken when removing apparently-unreachable code.
4186 For instance, when a function is inlined, a warning may mean that the
4187 line is unreachable in only one inlined copy of the function.
4189 This option is not made part of @option{-Wall} because in a debugging
4190 version of a program there is often substantial code which checks
4191 correct functioning of the program and is, hopefully, unreachable
4192 because the program does work. Another common use of unreachable
4193 code is to provide behavior which is selectable at compile-time.
4198 Warn if a function can not be inlined and it was declared as inline.
4199 Even with this option, the compiler will not warn about failures to
4200 inline functions declared in system headers.
4202 The compiler uses a variety of heuristics to determine whether or not
4203 to inline a function. For example, the compiler takes into account
4204 the size of the function being inlined and the amount of inlining
4205 that has already been done in the current function. Therefore,
4206 seemingly insignificant changes in the source program can cause the
4207 warnings produced by @option{-Winline} to appear or disappear.
4209 @item -Wno-invalid-offsetof @r{(C++ and Objective-C++ only)}
4210 @opindex Wno-invalid-offsetof
4211 @opindex Winvalid-offsetof
4212 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
4213 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
4214 to a non-POD type is undefined. In existing C++ implementations,
4215 however, @samp{offsetof} typically gives meaningful results even when
4216 applied to certain kinds of non-POD types. (Such as a simple
4217 @samp{struct} that fails to be a POD type only by virtue of having a
4218 constructor.) This flag is for users who are aware that they are
4219 writing nonportable code and who have deliberately chosen to ignore the
4222 The restrictions on @samp{offsetof} may be relaxed in a future version
4223 of the C++ standard.
4225 @item -Wno-int-to-pointer-cast @r{(C and Objective-C only)}
4226 @opindex Wno-int-to-pointer-cast
4227 @opindex Wint-to-pointer-cast
4228 Suppress warnings from casts to pointer type of an integer of a
4231 @item -Wno-pointer-to-int-cast @r{(C and Objective-C only)}
4232 @opindex Wno-pointer-to-int-cast
4233 @opindex Wpointer-to-int-cast
4234 Suppress warnings from casts from a pointer to an integer type of a
4238 @opindex Winvalid-pch
4239 @opindex Wno-invalid-pch
4240 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
4241 the search path but can't be used.
4245 @opindex Wno-long-long
4246 Warn if @samp{long long} type is used. This is enabled by either
4247 @option{-pedantic} or @option{-Wtraditional} in ISO C90 and C++98
4248 modes. To inhibit the warning messages, use @option{-Wno-long-long}.
4250 @item -Wvariadic-macros
4251 @opindex Wvariadic-macros
4252 @opindex Wno-variadic-macros
4253 Warn if variadic macros are used in pedantic ISO C90 mode, or the GNU
4254 alternate syntax when in pedantic ISO C99 mode. This is default.
4255 To inhibit the warning messages, use @option{-Wno-variadic-macros}.
4260 Warn if variable length array is used in the code.
4261 @option{-Wno-vla} will prevent the @option{-pedantic} warning of
4262 the variable length array.
4264 @item -Wvolatile-register-var
4265 @opindex Wvolatile-register-var
4266 @opindex Wno-volatile-register-var
4267 Warn if a register variable is declared volatile. The volatile
4268 modifier does not inhibit all optimizations that may eliminate reads
4269 and/or writes to register variables. This warning is enabled by
4272 @item -Wdisabled-optimization
4273 @opindex Wdisabled-optimization
4274 @opindex Wno-disabled-optimization
4275 Warn if a requested optimization pass is disabled. This warning does
4276 not generally indicate that there is anything wrong with your code; it
4277 merely indicates that GCC's optimizers were unable to handle the code
4278 effectively. Often, the problem is that your code is too big or too
4279 complex; GCC will refuse to optimize programs when the optimization
4280 itself is likely to take inordinate amounts of time.
4282 @item -Wpointer-sign @r{(C and Objective-C only)}
4283 @opindex Wpointer-sign
4284 @opindex Wno-pointer-sign
4285 Warn for pointer argument passing or assignment with different signedness.
4286 This option is only supported for C and Objective-C@. It is implied by
4287 @option{-Wall} and by @option{-pedantic}, which can be disabled with
4288 @option{-Wno-pointer-sign}.
4290 @item -Wstack-protector
4291 @opindex Wstack-protector
4292 @opindex Wno-stack-protector
4293 This option is only active when @option{-fstack-protector} is active. It
4294 warns about functions that will not be protected against stack smashing.
4297 @opindex Wno-mudflap
4298 Suppress warnings about constructs that cannot be instrumented by
4301 @item -Woverlength-strings
4302 @opindex Woverlength-strings
4303 @opindex Wno-overlength-strings
4304 Warn about string constants which are longer than the ``minimum
4305 maximum'' length specified in the C standard. Modern compilers
4306 generally allow string constants which are much longer than the
4307 standard's minimum limit, but very portable programs should avoid
4308 using longer strings.
4310 The limit applies @emph{after} string constant concatenation, and does
4311 not count the trailing NUL@. In C89, the limit was 509 characters; in
4312 C99, it was raised to 4095. C++98 does not specify a normative
4313 minimum maximum, so we do not diagnose overlength strings in C++@.
4315 This option is implied by @option{-pedantic}, and can be disabled with
4316 @option{-Wno-overlength-strings}.
4318 @item -Wunsuffixed-float-constants
4319 @opindex Wunsuffixed-float-constants
4321 GCC will issue a warning for any floating constant that does not have
4322 a suffix. When used together with @option{-Wsystem-headers} it will
4323 warn about such constants in system header files. This can be useful
4324 when preparing code to use with the @code{FLOAT_CONST_DECIMAL64} pragma
4325 from the decimal floating-point extension to C99.
4328 @node Debugging Options
4329 @section Options for Debugging Your Program or GCC
4330 @cindex options, debugging
4331 @cindex debugging information options
4333 GCC has various special options that are used for debugging
4334 either your program or GCC:
4339 Produce debugging information in the operating system's native format
4340 (stabs, COFF, XCOFF, or DWARF 2)@. GDB can work with this debugging
4343 On most systems that use stabs format, @option{-g} enables use of extra
4344 debugging information that only GDB can use; this extra information
4345 makes debugging work better in GDB but will probably make other debuggers
4347 refuse to read the program. If you want to control for certain whether
4348 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
4349 @option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
4351 GCC allows you to use @option{-g} with
4352 @option{-O}. The shortcuts taken by optimized code may occasionally
4353 produce surprising results: some variables you declared may not exist
4354 at all; flow of control may briefly move where you did not expect it;
4355 some statements may not be executed because they compute constant
4356 results or their values were already at hand; some statements may
4357 execute in different places because they were moved out of loops.
4359 Nevertheless it proves possible to debug optimized output. This makes
4360 it reasonable to use the optimizer for programs that might have bugs.
4362 The following options are useful when GCC is generated with the
4363 capability for more than one debugging format.
4367 Produce debugging information for use by GDB@. This means to use the
4368 most expressive format available (DWARF 2, stabs, or the native format
4369 if neither of those are supported), including GDB extensions if at all
4374 Produce debugging information in stabs format (if that is supported),
4375 without GDB extensions. This is the format used by DBX on most BSD
4376 systems. On MIPS, Alpha and System V Release 4 systems this option
4377 produces stabs debugging output which is not understood by DBX or SDB@.
4378 On System V Release 4 systems this option requires the GNU assembler.
4380 @item -feliminate-unused-debug-symbols
4381 @opindex feliminate-unused-debug-symbols
4382 Produce debugging information in stabs format (if that is supported),
4383 for only symbols that are actually used.
4385 @item -femit-class-debug-always
4386 Instead of emitting debugging information for a C++ class in only one
4387 object file, emit it in all object files using the class. This option
4388 should be used only with debuggers that are unable to handle the way GCC
4389 normally emits debugging information for classes because using this
4390 option will increase the size of debugging information by as much as a
4395 Produce debugging information in stabs format (if that is supported),
4396 using GNU extensions understood only by the GNU debugger (GDB)@. The
4397 use of these extensions is likely to make other debuggers crash or
4398 refuse to read the program.
4402 Produce debugging information in COFF format (if that is supported).
4403 This is the format used by SDB on most System V systems prior to
4408 Produce debugging information in XCOFF format (if that is supported).
4409 This is the format used by the DBX debugger on IBM RS/6000 systems.
4413 Produce debugging information in XCOFF format (if that is supported),
4414 using GNU extensions understood only by the GNU debugger (GDB)@. The
4415 use of these extensions is likely to make other debuggers crash or
4416 refuse to read the program, and may cause assemblers other than the GNU
4417 assembler (GAS) to fail with an error.
4419 @item -gdwarf-@var{version}
4420 @opindex gdwarf-@var{version}
4421 Produce debugging information in DWARF format (if that is
4422 supported). This is the format used by DBX on IRIX 6. The value
4423 of @var{version} may be either 2, 3 or 4; the default version is 2.
4425 Note that with DWARF version 2 some ports require, and will always
4426 use, some non-conflicting DWARF 3 extensions in the unwind tables.
4428 Version 4 may require GDB 7.0 and @option{-fvar-tracking-assignments}
4429 for maximum benefit.
4433 Produce debugging information in VMS debug format (if that is
4434 supported). This is the format used by DEBUG on VMS systems.
4437 @itemx -ggdb@var{level}
4438 @itemx -gstabs@var{level}
4439 @itemx -gcoff@var{level}
4440 @itemx -gxcoff@var{level}
4441 @itemx -gvms@var{level}
4442 Request debugging information and also use @var{level} to specify how
4443 much information. The default level is 2.
4445 Level 0 produces no debug information at all. Thus, @option{-g0} negates
4448 Level 1 produces minimal information, enough for making backtraces in
4449 parts of the program that you don't plan to debug. This includes
4450 descriptions of functions and external variables, but no information
4451 about local variables and no line numbers.
4453 Level 3 includes extra information, such as all the macro definitions
4454 present in the program. Some debuggers support macro expansion when
4455 you use @option{-g3}.
4457 @option{-gdwarf-2} does not accept a concatenated debug level, because
4458 GCC used to support an option @option{-gdwarf} that meant to generate
4459 debug information in version 1 of the DWARF format (which is very
4460 different from version 2), and it would have been too confusing. That
4461 debug format is long obsolete, but the option cannot be changed now.
4462 Instead use an additional @option{-g@var{level}} option to change the
4463 debug level for DWARF.
4467 Turn off generation of debug info, if leaving out this option would have
4468 generated it, or turn it on at level 2 otherwise. The position of this
4469 argument in the command line does not matter, it takes effect after all
4470 other options are processed, and it does so only once, no matter how
4471 many times it is given. This is mainly intended to be used with
4472 @option{-fcompare-debug}.
4474 @item -fdump-final-insns@r{[}=@var{file}@r{]}
4475 @opindex fdump-final-insns
4476 Dump the final internal representation (RTL) to @var{file}. If the
4477 optional argument is omitted (or if @var{file} is @code{.}), the name
4478 of the dump file will be determined by appending @code{.gkd} to the
4479 compilation output file name.
4481 @item -fcompare-debug@r{[}=@var{opts}@r{]}
4482 @opindex fcompare-debug
4483 @opindex fno-compare-debug
4484 If no error occurs during compilation, run the compiler a second time,
4485 adding @var{opts} and @option{-fcompare-debug-second} to the arguments
4486 passed to the second compilation. Dump the final internal
4487 representation in both compilations, and print an error if they differ.
4489 If the equal sign is omitted, the default @option{-gtoggle} is used.
4491 The environment variable @env{GCC_COMPARE_DEBUG}, if defined, non-empty
4492 and nonzero, implicitly enables @option{-fcompare-debug}. If
4493 @env{GCC_COMPARE_DEBUG} is defined to a string starting with a dash,
4494 then it is used for @var{opts}, otherwise the default @option{-gtoggle}
4497 @option{-fcompare-debug=}, with the equal sign but without @var{opts},
4498 is equivalent to @option{-fno-compare-debug}, which disables the dumping
4499 of the final representation and the second compilation, preventing even
4500 @env{GCC_COMPARE_DEBUG} from taking effect.
4502 To verify full coverage during @option{-fcompare-debug} testing, set
4503 @env{GCC_COMPARE_DEBUG} to say @samp{-fcompare-debug-not-overridden},
4504 which GCC will reject as an invalid option in any actual compilation
4505 (rather than preprocessing, assembly or linking). To get just a
4506 warning, setting @env{GCC_COMPARE_DEBUG} to @samp{-w%n-fcompare-debug
4507 not overridden} will do.
4509 @item -fcompare-debug-second
4510 @opindex fcompare-debug-second
4511 This option is implicitly passed to the compiler for the second
4512 compilation requested by @option{-fcompare-debug}, along with options to
4513 silence warnings, and omitting other options that would cause
4514 side-effect compiler outputs to files or to the standard output. Dump
4515 files and preserved temporary files are renamed so as to contain the
4516 @code{.gk} additional extension during the second compilation, to avoid
4517 overwriting those generated by the first.
4519 When this option is passed to the compiler driver, it causes the
4520 @emph{first} compilation to be skipped, which makes it useful for little
4521 other than debugging the compiler proper.
4523 @item -feliminate-dwarf2-dups
4524 @opindex feliminate-dwarf2-dups
4525 Compress DWARF2 debugging information by eliminating duplicated
4526 information about each symbol. This option only makes sense when
4527 generating DWARF2 debugging information with @option{-gdwarf-2}.
4529 @item -femit-struct-debug-baseonly
4530 Emit debug information for struct-like types
4531 only when the base name of the compilation source file
4532 matches the base name of file in which the struct was defined.
4534 This option substantially reduces the size of debugging information,
4535 but at significant potential loss in type information to the debugger.
4536 See @option{-femit-struct-debug-reduced} for a less aggressive option.
4537 See @option{-femit-struct-debug-detailed} for more detailed control.
4539 This option works only with DWARF 2.
4541 @item -femit-struct-debug-reduced
4542 Emit debug information for struct-like types
4543 only when the base name of the compilation source file
4544 matches the base name of file in which the type was defined,
4545 unless the struct is a template or defined in a system header.
4547 This option significantly reduces the size of debugging information,
4548 with some potential loss in type information to the debugger.
4549 See @option{-femit-struct-debug-baseonly} for a more aggressive option.
4550 See @option{-femit-struct-debug-detailed} for more detailed control.
4552 This option works only with DWARF 2.
4554 @item -femit-struct-debug-detailed@r{[}=@var{spec-list}@r{]}
4555 Specify the struct-like types
4556 for which the compiler will generate debug information.
4557 The intent is to reduce duplicate struct debug information
4558 between different object files within the same program.
4560 This option is a detailed version of
4561 @option{-femit-struct-debug-reduced} and @option{-femit-struct-debug-baseonly},
4562 which will serve for most needs.
4564 A specification has the syntax
4565 [@samp{dir:}|@samp{ind:}][@samp{ord:}|@samp{gen:}](@samp{any}|@samp{sys}|@samp{base}|@samp{none})
4567 The optional first word limits the specification to
4568 structs that are used directly (@samp{dir:}) or used indirectly (@samp{ind:}).
4569 A struct type is used directly when it is the type of a variable, member.
4570 Indirect uses arise through pointers to structs.
4571 That is, when use of an incomplete struct would be legal, the use is indirect.
4573 @samp{struct one direct; struct two * indirect;}.
4575 The optional second word limits the specification to
4576 ordinary structs (@samp{ord:}) or generic structs (@samp{gen:}).
4577 Generic structs are a bit complicated to explain.
4578 For C++, these are non-explicit specializations of template classes,
4579 or non-template classes within the above.
4580 Other programming languages have generics,
4581 but @samp{-femit-struct-debug-detailed} does not yet implement them.
4583 The third word specifies the source files for those
4584 structs for which the compiler will emit debug information.
4585 The values @samp{none} and @samp{any} have the normal meaning.
4586 The value @samp{base} means that
4587 the base of name of the file in which the type declaration appears
4588 must match the base of the name of the main compilation file.
4589 In practice, this means that
4590 types declared in @file{foo.c} and @file{foo.h} will have debug information,
4591 but types declared in other header will not.
4592 The value @samp{sys} means those types satisfying @samp{base}
4593 or declared in system or compiler headers.
4595 You may need to experiment to determine the best settings for your application.
4597 The default is @samp{-femit-struct-debug-detailed=all}.
4599 This option works only with DWARF 2.
4601 @item -fno-merge-debug-strings
4602 @opindex fmerge-debug-strings
4603 @opindex fno-merge-debug-strings
4604 Direct the linker to not merge together strings in the debugging
4605 information which are identical in different object files. Merging is
4606 not supported by all assemblers or linkers. Merging decreases the size
4607 of the debug information in the output file at the cost of increasing
4608 link processing time. Merging is enabled by default.
4610 @item -fdebug-prefix-map=@var{old}=@var{new}
4611 @opindex fdebug-prefix-map
4612 When compiling files in directory @file{@var{old}}, record debugging
4613 information describing them as in @file{@var{new}} instead.
4615 @item -fno-dwarf2-cfi-asm
4616 @opindex fdwarf2-cfi-asm
4617 @opindex fno-dwarf2-cfi-asm
4618 Emit DWARF 2 unwind info as compiler generated @code{.eh_frame} section
4619 instead of using GAS @code{.cfi_*} directives.
4621 @cindex @command{prof}
4624 Generate extra code to write profile information suitable for the
4625 analysis program @command{prof}. You must use this option when compiling
4626 the source files you want data about, and you must also use it when
4629 @cindex @command{gprof}
4632 Generate extra code to write profile information suitable for the
4633 analysis program @command{gprof}. You must use this option when compiling
4634 the source files you want data about, and you must also use it when
4639 Makes the compiler print out each function name as it is compiled, and
4640 print some statistics about each pass when it finishes.
4643 @opindex ftime-report
4644 Makes the compiler print some statistics about the time consumed by each
4645 pass when it finishes.
4648 @opindex fmem-report
4649 Makes the compiler print some statistics about permanent memory
4650 allocation when it finishes.
4652 @item -fpre-ipa-mem-report
4653 @opindex fpre-ipa-mem-report
4654 @item -fpost-ipa-mem-report
4655 @opindex fpost-ipa-mem-report
4656 Makes the compiler print some statistics about permanent memory
4657 allocation before or after interprocedural optimization.
4659 @item -fprofile-arcs
4660 @opindex fprofile-arcs
4661 Add code so that program flow @dfn{arcs} are instrumented. During
4662 execution the program records how many times each branch and call is
4663 executed and how many times it is taken or returns. When the compiled
4664 program exits it saves this data to a file called
4665 @file{@var{auxname}.gcda} for each source file. The data may be used for
4666 profile-directed optimizations (@option{-fbranch-probabilities}), or for
4667 test coverage analysis (@option{-ftest-coverage}). Each object file's
4668 @var{auxname} is generated from the name of the output file, if
4669 explicitly specified and it is not the final executable, otherwise it is
4670 the basename of the source file. In both cases any suffix is removed
4671 (e.g.@: @file{foo.gcda} for input file @file{dir/foo.c}, or
4672 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
4673 @xref{Cross-profiling}.
4675 @cindex @command{gcov}
4679 This option is used to compile and link code instrumented for coverage
4680 analysis. The option is a synonym for @option{-fprofile-arcs}
4681 @option{-ftest-coverage} (when compiling) and @option{-lgcov} (when
4682 linking). See the documentation for those options for more details.
4687 Compile the source files with @option{-fprofile-arcs} plus optimization
4688 and code generation options. For test coverage analysis, use the
4689 additional @option{-ftest-coverage} option. You do not need to profile
4690 every source file in a program.
4693 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
4694 (the latter implies the former).
4697 Run the program on a representative workload to generate the arc profile
4698 information. This may be repeated any number of times. You can run
4699 concurrent instances of your program, and provided that the file system
4700 supports locking, the data files will be correctly updated. Also
4701 @code{fork} calls are detected and correctly handled (double counting
4705 For profile-directed optimizations, compile the source files again with
4706 the same optimization and code generation options plus
4707 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
4708 Control Optimization}).
4711 For test coverage analysis, use @command{gcov} to produce human readable
4712 information from the @file{.gcno} and @file{.gcda} files. Refer to the
4713 @command{gcov} documentation for further information.
4717 With @option{-fprofile-arcs}, for each function of your program GCC
4718 creates a program flow graph, then finds a spanning tree for the graph.
4719 Only arcs that are not on the spanning tree have to be instrumented: the
4720 compiler adds code to count the number of times that these arcs are
4721 executed. When an arc is the only exit or only entrance to a block, the
4722 instrumentation code can be added to the block; otherwise, a new basic
4723 block must be created to hold the instrumentation code.
4726 @item -ftest-coverage
4727 @opindex ftest-coverage
4728 Produce a notes file that the @command{gcov} code-coverage utility
4729 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
4730 show program coverage. Each source file's note file is called
4731 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
4732 above for a description of @var{auxname} and instructions on how to
4733 generate test coverage data. Coverage data will match the source files
4734 more closely, if you do not optimize.
4736 @item -fdbg-cnt-list
4737 @opindex fdbg-cnt-list
4738 Print the name and the counter upperbound for all debug counters.
4740 @item -fdbg-cnt=@var{counter-value-list}
4742 Set the internal debug counter upperbound. @var{counter-value-list}
4743 is a comma-separated list of @var{name}:@var{value} pairs
4744 which sets the upperbound of each debug counter @var{name} to @var{value}.
4745 All debug counters have the initial upperbound of @var{UINT_MAX},
4746 thus dbg_cnt() returns true always unless the upperbound is set by this option.
4747 e.g. With -fdbg-cnt=dce:10,tail_call:0
4748 dbg_cnt(dce) will return true only for first 10 invocations
4749 and dbg_cnt(tail_call) will return false always.
4751 @item -d@var{letters}
4752 @itemx -fdump-rtl-@var{pass}
4754 Says to make debugging dumps during compilation at times specified by
4755 @var{letters}. This is used for debugging the RTL-based passes of the
4756 compiler. The file names for most of the dumps are made by appending
4757 a pass number and a word to the @var{dumpname}, and the files are
4758 created in the directory of the output file. @var{dumpname} is
4759 generated from the name of the output file, if explicitly specified
4760 and it is not an executable, otherwise it is the basename of the
4761 source file. These switches may have different effects when
4762 @option{-E} is used for preprocessing.
4764 Debug dumps can be enabled with a @option{-fdump-rtl} switch or some
4765 @option{-d} option @var{letters}. Here are the possible
4766 letters for use in @var{pass} and @var{letters}, and their meanings:
4770 @item -fdump-rtl-alignments
4771 @opindex fdump-rtl-alignments
4772 Dump after branch alignments have been computed.
4774 @item -fdump-rtl-asmcons
4775 @opindex fdump-rtl-asmcons
4776 Dump after fixing rtl statements that have unsatisfied in/out constraints.
4778 @item -fdump-rtl-auto_inc_dec
4779 @opindex fdump-rtl-auto_inc_dec
4780 Dump after auto-inc-dec discovery. This pass is only run on
4781 architectures that have auto inc or auto dec instructions.
4783 @item -fdump-rtl-barriers
4784 @opindex fdump-rtl-barriers
4785 Dump after cleaning up the barrier instructions.
4787 @item -fdump-rtl-bbpart
4788 @opindex fdump-rtl-bbpart
4789 Dump after partitioning hot and cold basic blocks.
4791 @item -fdump-rtl-bbro
4792 @opindex fdump-rtl-bbro
4793 Dump after block reordering.
4795 @item -fdump-rtl-btl1
4796 @itemx -fdump-rtl-btl2
4797 @opindex fdump-rtl-btl2
4798 @opindex fdump-rtl-btl2
4799 @option{-fdump-rtl-btl1} and @option{-fdump-rtl-btl2} enable dumping
4800 after the two branch
4801 target load optimization passes.
4803 @item -fdump-rtl-bypass
4804 @opindex fdump-rtl-bypass
4805 Dump after jump bypassing and control flow optimizations.
4807 @item -fdump-rtl-combine
4808 @opindex fdump-rtl-combine
4809 Dump after the RTL instruction combination pass.
4811 @item -fdump-rtl-compgotos
4812 @opindex fdump-rtl-compgotos
4813 Dump after duplicating the computed gotos.
4815 @item -fdump-rtl-ce1
4816 @itemx -fdump-rtl-ce2
4817 @itemx -fdump-rtl-ce3
4818 @opindex fdump-rtl-ce1
4819 @opindex fdump-rtl-ce2
4820 @opindex fdump-rtl-ce3
4821 @option{-fdump-rtl-ce1}, @option{-fdump-rtl-ce2}, and
4822 @option{-fdump-rtl-ce3} enable dumping after the three
4823 if conversion passes.
4825 @itemx -fdump-rtl-cprop_hardreg
4826 @opindex fdump-rtl-cprop_hardreg
4827 Dump after hard register copy propagation.
4829 @itemx -fdump-rtl-csa
4830 @opindex fdump-rtl-csa
4831 Dump after combining stack adjustments.
4833 @item -fdump-rtl-cse1
4834 @itemx -fdump-rtl-cse2
4835 @opindex fdump-rtl-cse1
4836 @opindex fdump-rtl-cse2
4837 @option{-fdump-rtl-cse1} and @option{-fdump-rtl-cse2} enable dumping after
4838 the two common sub-expression elimination passes.
4840 @itemx -fdump-rtl-dce
4841 @opindex fdump-rtl-dce
4842 Dump after the standalone dead code elimination passes.
4844 @itemx -fdump-rtl-dbr
4845 @opindex fdump-rtl-dbr
4846 Dump after delayed branch scheduling.
4848 @item -fdump-rtl-dce1
4849 @itemx -fdump-rtl-dce2
4850 @opindex fdump-rtl-dce1
4851 @opindex fdump-rtl-dce2
4852 @option{-fdump-rtl-dce1} and @option{-fdump-rtl-dce2} enable dumping after
4853 the two dead store elimination passes.
4856 @opindex fdump-rtl-eh
4857 Dump after finalization of EH handling code.
4859 @item -fdump-rtl-eh_ranges
4860 @opindex fdump-rtl-eh_ranges
4861 Dump after conversion of EH handling range regions.
4863 @item -fdump-rtl-expand
4864 @opindex fdump-rtl-expand
4865 Dump after RTL generation.
4867 @item -fdump-rtl-fwprop1
4868 @itemx -fdump-rtl-fwprop2
4869 @opindex fdump-rtl-fwprop1
4870 @opindex fdump-rtl-fwprop2
4871 @option{-fdump-rtl-fwprop1} and @option{-fdump-rtl-fwprop2} enable
4872 dumping after the two forward propagation passes.
4874 @item -fdump-rtl-gcse1
4875 @itemx -fdump-rtl-gcse2
4876 @opindex fdump-rtl-gcse1
4877 @opindex fdump-rtl-gcse2
4878 @option{-fdump-rtl-gcse1} and @option{-fdump-rtl-gcse2} enable dumping
4879 after global common subexpression elimination.
4881 @item -fdump-rtl-init-regs
4882 @opindex fdump-rtl-init-regs
4883 Dump after the initialization of the registers.
4885 @item -fdump-rtl-initvals
4886 @opindex fdump-rtl-initvals
4887 Dump after the computation of the initial value sets.
4889 @itemx -fdump-rtl-into_cfglayout
4890 @opindex fdump-rtl-into_cfglayout
4891 Dump after converting to cfglayout mode.
4893 @item -fdump-rtl-ira
4894 @opindex fdump-rtl-ira
4895 Dump after iterated register allocation.
4897 @item -fdump-rtl-jump
4898 @opindex fdump-rtl-jump
4899 Dump after the second jump optimization.
4901 @item -fdump-rtl-loop2
4902 @opindex fdump-rtl-loop2
4903 @option{-fdump-rtl-loop2} enables dumping after the rtl
4904 loop optimization passes.
4906 @item -fdump-rtl-mach
4907 @opindex fdump-rtl-mach
4908 Dump after performing the machine dependent reorganization pass, if that
4911 @item -fdump-rtl-mode_sw
4912 @opindex fdump-rtl-mode_sw
4913 Dump after removing redundant mode switches.
4915 @item -fdump-rtl-rnreg
4916 @opindex fdump-rtl-rnreg
4917 Dump after register renumbering.
4919 @itemx -fdump-rtl-outof_cfglayout
4920 @opindex fdump-rtl-outof_cfglayout
4921 Dump after converting from cfglayout mode.
4923 @item -fdump-rtl-peephole2
4924 @opindex fdump-rtl-peephole2
4925 Dump after the peephole pass.
4927 @item -fdump-rtl-postreload
4928 @opindex fdump-rtl-postreload
4929 Dump after post-reload optimizations.
4931 @itemx -fdump-rtl-pro_and_epilogue
4932 @opindex fdump-rtl-pro_and_epilogue
4933 Dump after generating the function pro and epilogues.
4935 @item -fdump-rtl-regmove
4936 @opindex fdump-rtl-regmove
4937 Dump after the register move pass.
4939 @item -fdump-rtl-sched1
4940 @itemx -fdump-rtl-sched2
4941 @opindex fdump-rtl-sched1
4942 @opindex fdump-rtl-sched2
4943 @option{-fdump-rtl-sched1} and @option{-fdump-rtl-sched2} enable dumping
4944 after the basic block scheduling passes.
4946 @item -fdump-rtl-see
4947 @opindex fdump-rtl-see
4948 Dump after sign extension elimination.
4950 @item -fdump-rtl-seqabstr
4951 @opindex fdump-rtl-seqabstr
4952 Dump after common sequence discovery.
4954 @item -fdump-rtl-shorten
4955 @opindex fdump-rtl-shorten
4956 Dump after shortening branches.
4958 @item -fdump-rtl-sibling
4959 @opindex fdump-rtl-sibling
4960 Dump after sibling call optimizations.
4962 @item -fdump-rtl-split1
4963 @itemx -fdump-rtl-split2
4964 @itemx -fdump-rtl-split3
4965 @itemx -fdump-rtl-split4
4966 @itemx -fdump-rtl-split5
4967 @opindex fdump-rtl-split1
4968 @opindex fdump-rtl-split2
4969 @opindex fdump-rtl-split3
4970 @opindex fdump-rtl-split4
4971 @opindex fdump-rtl-split5
4972 @option{-fdump-rtl-split1}, @option{-fdump-rtl-split2},
4973 @option{-fdump-rtl-split3}, @option{-fdump-rtl-split4} and
4974 @option{-fdump-rtl-split5} enable dumping after five rounds of
4975 instruction splitting.
4977 @item -fdump-rtl-sms
4978 @opindex fdump-rtl-sms
4979 Dump after modulo scheduling. This pass is only run on some
4982 @item -fdump-rtl-stack
4983 @opindex fdump-rtl-stack
4984 Dump after conversion from GCC's "flat register file" registers to the
4985 x87's stack-like registers. This pass is only run on x86 variants.
4987 @item -fdump-rtl-subreg1
4988 @itemx -fdump-rtl-subreg2
4989 @opindex fdump-rtl-subreg1
4990 @opindex fdump-rtl-subreg2
4991 @option{-fdump-rtl-subreg1} and @option{-fdump-rtl-subreg2} enable dumping after
4992 the two subreg expansion passes.
4994 @item -fdump-rtl-unshare
4995 @opindex fdump-rtl-unshare
4996 Dump after all rtl has been unshared.
4998 @item -fdump-rtl-vartrack
4999 @opindex fdump-rtl-vartrack
5000 Dump after variable tracking.
5002 @item -fdump-rtl-vregs
5003 @opindex fdump-rtl-vregs
5004 Dump after converting virtual registers to hard registers.
5006 @item -fdump-rtl-web
5007 @opindex fdump-rtl-web
5008 Dump after live range splitting.
5010 @item -fdump-rtl-regclass
5011 @itemx -fdump-rtl-subregs_of_mode_init
5012 @itemx -fdump-rtl-subregs_of_mode_finish
5013 @itemx -fdump-rtl-dfinit
5014 @itemx -fdump-rtl-dfinish
5015 @opindex fdump-rtl-regclass
5016 @opindex fdump-rtl-subregs_of_mode_init
5017 @opindex fdump-rtl-subregs_of_mode_finish
5018 @opindex fdump-rtl-dfinit
5019 @opindex fdump-rtl-dfinish
5020 These dumps are defined but always produce empty files.
5022 @item -fdump-rtl-all
5023 @opindex fdump-rtl-all
5024 Produce all the dumps listed above.
5028 Annotate the assembler output with miscellaneous debugging information.
5032 Dump all macro definitions, at the end of preprocessing, in addition to
5037 Produce a core dump whenever an error occurs.
5041 Print statistics on memory usage, at the end of the run, to
5046 Annotate the assembler output with a comment indicating which
5047 pattern and alternative was used. The length of each instruction is
5052 Dump the RTL in the assembler output as a comment before each instruction.
5053 Also turns on @option{-dp} annotation.
5057 For each of the other indicated dump files (@option{-fdump-rtl-@var{pass}}),
5058 dump a representation of the control flow graph suitable for viewing with VCG
5059 to @file{@var{file}.@var{pass}.vcg}.
5063 Just generate RTL for a function instead of compiling it. Usually used
5064 with @option{-fdump-rtl-expand}.
5068 Dump debugging information during parsing, to standard error.
5072 @opindex fdump-noaddr
5073 When doing debugging dumps, suppress address output. This makes it more
5074 feasible to use diff on debugging dumps for compiler invocations with
5075 different compiler binaries and/or different
5076 text / bss / data / heap / stack / dso start locations.
5078 @item -fdump-unnumbered
5079 @opindex fdump-unnumbered
5080 When doing debugging dumps, suppress instruction numbers and address output.
5081 This makes it more feasible to use diff on debugging dumps for compiler
5082 invocations with different options, in particular with and without
5085 @item -fdump-unnumbered-links
5086 @opindex fdump-unnumbered-links
5087 When doing debugging dumps (see @option{-d} option above), suppress
5088 instruction numbers for the links to the previous and next instructions
5091 @item -fdump-translation-unit @r{(C++ only)}
5092 @itemx -fdump-translation-unit-@var{options} @r{(C++ only)}
5093 @opindex fdump-translation-unit
5094 Dump a representation of the tree structure for the entire translation
5095 unit to a file. The file name is made by appending @file{.tu} to the
5096 source file name, and the file is created in the same directory as the
5097 output file. If the @samp{-@var{options}} form is used, @var{options}
5098 controls the details of the dump as described for the
5099 @option{-fdump-tree} options.
5101 @item -fdump-class-hierarchy @r{(C++ only)}
5102 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
5103 @opindex fdump-class-hierarchy
5104 Dump a representation of each class's hierarchy and virtual function
5105 table layout to a file. The file name is made by appending
5106 @file{.class} to the source file name, and the file is created in the
5107 same directory as the output file. If the @samp{-@var{options}} form
5108 is used, @var{options} controls the details of the dump as described
5109 for the @option{-fdump-tree} options.
5111 @item -fdump-ipa-@var{switch}
5113 Control the dumping at various stages of inter-procedural analysis
5114 language tree to a file. The file name is generated by appending a
5115 switch specific suffix to the source file name, and the file is created
5116 in the same directory as the output file. The following dumps are
5121 Enables all inter-procedural analysis dumps.
5124 Dumps information about call-graph optimization, unused function removal,
5125 and inlining decisions.
5128 Dump after function inlining.
5132 @item -fdump-statistics-@var{option}
5133 @opindex fdump-statistics
5134 Enable and control dumping of pass statistics in a separate file. The
5135 file name is generated by appending a suffix ending in
5136 @samp{.statistics} to the source file name, and the file is created in
5137 the same directory as the output file. If the @samp{-@var{option}}
5138 form is used, @samp{-stats} will cause counters to be summed over the
5139 whole compilation unit while @samp{-details} will dump every event as
5140 the passes generate them. The default with no option is to sum
5141 counters for each function compiled.
5143 @item -fdump-tree-@var{switch}
5144 @itemx -fdump-tree-@var{switch}-@var{options}
5146 Control the dumping at various stages of processing the intermediate
5147 language tree to a file. The file name is generated by appending a
5148 switch specific suffix to the source file name, and the file is
5149 created in the same directory as the output file. If the
5150 @samp{-@var{options}} form is used, @var{options} is a list of
5151 @samp{-} separated options that control the details of the dump. Not
5152 all options are applicable to all dumps, those which are not
5153 meaningful will be ignored. The following options are available
5157 Print the address of each node. Usually this is not meaningful as it
5158 changes according to the environment and source file. Its primary use
5159 is for tying up a dump file with a debug environment.
5161 If @code{DECL_ASSEMBLER_NAME} has been set for a given decl, use that
5162 in the dump instead of @code{DECL_NAME}. Its primary use is ease of
5163 use working backward from mangled names in the assembly file.
5165 Inhibit dumping of members of a scope or body of a function merely
5166 because that scope has been reached. Only dump such items when they
5167 are directly reachable by some other path. When dumping pretty-printed
5168 trees, this option inhibits dumping the bodies of control structures.
5170 Print a raw representation of the tree. By default, trees are
5171 pretty-printed into a C-like representation.
5173 Enable more detailed dumps (not honored by every dump option).
5175 Enable dumping various statistics about the pass (not honored by every dump
5178 Enable showing basic block boundaries (disabled in raw dumps).
5180 Enable showing virtual operands for every statement.
5182 Enable showing line numbers for statements.
5184 Enable showing the unique ID (@code{DECL_UID}) for each variable.
5186 Enable showing the tree dump for each statement.
5188 Enable showing the EH region number holding each statement.
5190 Turn on all options, except @option{raw}, @option{slim}, @option{verbose}
5191 and @option{lineno}.
5194 The following tree dumps are possible:
5198 @opindex fdump-tree-original
5199 Dump before any tree based optimization, to @file{@var{file}.original}.
5202 @opindex fdump-tree-optimized
5203 Dump after all tree based optimization, to @file{@var{file}.optimized}.
5206 @opindex fdump-tree-gimple
5207 Dump each function before and after the gimplification pass to a file. The
5208 file name is made by appending @file{.gimple} to the source file name.
5211 @opindex fdump-tree-cfg
5212 Dump the control flow graph of each function to a file. The file name is
5213 made by appending @file{.cfg} to the source file name.
5216 @opindex fdump-tree-vcg
5217 Dump the control flow graph of each function to a file in VCG format. The
5218 file name is made by appending @file{.vcg} to the source file name. Note
5219 that if the file contains more than one function, the generated file cannot
5220 be used directly by VCG@. You will need to cut and paste each function's
5221 graph into its own separate file first.
5224 @opindex fdump-tree-ch
5225 Dump each function after copying loop headers. The file name is made by
5226 appending @file{.ch} to the source file name.
5229 @opindex fdump-tree-ssa
5230 Dump SSA related information to a file. The file name is made by appending
5231 @file{.ssa} to the source file name.
5234 @opindex fdump-tree-alias
5235 Dump aliasing information for each function. The file name is made by
5236 appending @file{.alias} to the source file name.
5239 @opindex fdump-tree-ccp
5240 Dump each function after CCP@. The file name is made by appending
5241 @file{.ccp} to the source file name.
5244 @opindex fdump-tree-storeccp
5245 Dump each function after STORE-CCP@. The file name is made by appending
5246 @file{.storeccp} to the source file name.
5249 @opindex fdump-tree-pre
5250 Dump trees after partial redundancy elimination. The file name is made
5251 by appending @file{.pre} to the source file name.
5254 @opindex fdump-tree-fre
5255 Dump trees after full redundancy elimination. The file name is made
5256 by appending @file{.fre} to the source file name.
5259 @opindex fdump-tree-copyprop
5260 Dump trees after copy propagation. The file name is made
5261 by appending @file{.copyprop} to the source file name.
5263 @item store_copyprop
5264 @opindex fdump-tree-store_copyprop
5265 Dump trees after store copy-propagation. The file name is made
5266 by appending @file{.store_copyprop} to the source file name.
5269 @opindex fdump-tree-dce
5270 Dump each function after dead code elimination. The file name is made by
5271 appending @file{.dce} to the source file name.
5274 @opindex fdump-tree-mudflap
5275 Dump each function after adding mudflap instrumentation. The file name is
5276 made by appending @file{.mudflap} to the source file name.
5279 @opindex fdump-tree-sra
5280 Dump each function after performing scalar replacement of aggregates. The
5281 file name is made by appending @file{.sra} to the source file name.
5284 @opindex fdump-tree-sink
5285 Dump each function after performing code sinking. The file name is made
5286 by appending @file{.sink} to the source file name.
5289 @opindex fdump-tree-dom
5290 Dump each function after applying dominator tree optimizations. The file
5291 name is made by appending @file{.dom} to the source file name.
5294 @opindex fdump-tree-dse
5295 Dump each function after applying dead store elimination. The file
5296 name is made by appending @file{.dse} to the source file name.
5299 @opindex fdump-tree-phiopt
5300 Dump each function after optimizing PHI nodes into straightline code. The file
5301 name is made by appending @file{.phiopt} to the source file name.
5304 @opindex fdump-tree-forwprop
5305 Dump each function after forward propagating single use variables. The file
5306 name is made by appending @file{.forwprop} to the source file name.
5309 @opindex fdump-tree-copyrename
5310 Dump each function after applying the copy rename optimization. The file
5311 name is made by appending @file{.copyrename} to the source file name.
5314 @opindex fdump-tree-nrv
5315 Dump each function after applying the named return value optimization on
5316 generic trees. The file name is made by appending @file{.nrv} to the source
5320 @opindex fdump-tree-vect
5321 Dump each function after applying vectorization of loops. The file name is
5322 made by appending @file{.vect} to the source file name.
5325 @opindex fdump-tree-vrp
5326 Dump each function after Value Range Propagation (VRP). The file name
5327 is made by appending @file{.vrp} to the source file name.
5330 @opindex fdump-tree-all
5331 Enable all the available tree dumps with the flags provided in this option.
5334 @item -ftree-vectorizer-verbose=@var{n}
5335 @opindex ftree-vectorizer-verbose
5336 This option controls the amount of debugging output the vectorizer prints.
5337 This information is written to standard error, unless
5338 @option{-fdump-tree-all} or @option{-fdump-tree-vect} is specified,
5339 in which case it is output to the usual dump listing file, @file{.vect}.
5340 For @var{n}=0 no diagnostic information is reported.
5341 If @var{n}=1 the vectorizer reports each loop that got vectorized,
5342 and the total number of loops that got vectorized.
5343 If @var{n}=2 the vectorizer also reports non-vectorized loops that passed
5344 the first analysis phase (vect_analyze_loop_form) - i.e.@: countable,
5345 inner-most, single-bb, single-entry/exit loops. This is the same verbosity
5346 level that @option{-fdump-tree-vect-stats} uses.
5347 Higher verbosity levels mean either more information dumped for each
5348 reported loop, or same amount of information reported for more loops:
5349 If @var{n}=3, alignment related information is added to the reports.
5350 If @var{n}=4, data-references related information (e.g.@: memory dependences,
5351 memory access-patterns) is added to the reports.
5352 If @var{n}=5, the vectorizer reports also non-vectorized inner-most loops
5353 that did not pass the first analysis phase (i.e., may not be countable, or
5354 may have complicated control-flow).
5355 If @var{n}=6, the vectorizer reports also non-vectorized nested loops.
5356 For @var{n}=7, all the information the vectorizer generates during its
5357 analysis and transformation is reported. This is the same verbosity level
5358 that @option{-fdump-tree-vect-details} uses.
5360 @item -frandom-seed=@var{string}
5361 @opindex frandom-seed
5362 This option provides a seed that GCC uses when it would otherwise use
5363 random numbers. It is used to generate certain symbol names
5364 that have to be different in every compiled file. It is also used to
5365 place unique stamps in coverage data files and the object files that
5366 produce them. You can use the @option{-frandom-seed} option to produce
5367 reproducibly identical object files.
5369 The @var{string} should be different for every file you compile.
5371 @item -fsched-verbose=@var{n}
5372 @opindex fsched-verbose
5373 On targets that use instruction scheduling, this option controls the
5374 amount of debugging output the scheduler prints. This information is
5375 written to standard error, unless @option{-fdump-rtl-sched1} or
5376 @option{-fdump-rtl-sched2} is specified, in which case it is output
5377 to the usual dump listing file, @file{.sched} or @file{.sched2}
5378 respectively. However for @var{n} greater than nine, the output is
5379 always printed to standard error.
5381 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
5382 same information as @option{-fdump-rtl-sched1} and @option{-fdump-rtl-sched2}.
5383 For @var{n} greater than one, it also output basic block probabilities,
5384 detailed ready list information and unit/insn info. For @var{n} greater
5385 than two, it includes RTL at abort point, control-flow and regions info.
5386 And for @var{n} over four, @option{-fsched-verbose} also includes
5390 @itemx -save-temps=cwd
5392 Store the usual ``temporary'' intermediate files permanently; place them
5393 in the current directory and name them based on the source file. Thus,
5394 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
5395 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
5396 preprocessed @file{foo.i} output file even though the compiler now
5397 normally uses an integrated preprocessor.
5399 When used in combination with the @option{-x} command line option,
5400 @option{-save-temps} is sensible enough to avoid over writing an
5401 input source file with the same extension as an intermediate file.
5402 The corresponding intermediate file may be obtained by renaming the
5403 source file before using @option{-save-temps}.
5405 If you invoke GCC in parallel, compiling several different source
5406 files that share a common base name in different subdirectories or the
5407 same source file compiled for multiple output destinations, it is
5408 likely that the different parallel compilers will interfere with each
5409 other, and overwrite the temporary files. For instance:
5412 gcc -save-temps -o outdir1/foo.o indir1/foo.c&
5413 gcc -save-temps -o outdir2/foo.o indir2/foo.c&
5416 may result in @file{foo.i} and @file{foo.o} being written to
5417 simultaneously by both compilers.
5419 @item -save-temps=obj
5420 @opindex save-temps=obj
5421 Store the usual ``temporary'' intermediate files permanently. If the
5422 @option{-o} option is used, the temporary files are based on the
5423 object file. If the @option{-o} option is not used, the
5424 @option{-save-temps=obj} switch behaves like @option{-save-temps}.
5429 gcc -save-temps=obj -c foo.c
5430 gcc -save-temps=obj -c bar.c -o dir/xbar.o
5431 gcc -save-temps=obj foobar.c -o dir2/yfoobar
5434 would create @file{foo.i}, @file{foo.s}, @file{dir/xbar.i},
5435 @file{dir/xbar.s}, @file{dir2/yfoobar.i}, @file{dir2/yfoobar.s}, and
5436 @file{dir2/yfoobar.o}.
5438 @item -time@r{[}=@var{file}@r{]}
5440 Report the CPU time taken by each subprocess in the compilation
5441 sequence. For C source files, this is the compiler proper and assembler
5442 (plus the linker if linking is done).
5444 Without the specification of an output file, the output looks like this:
5451 The first number on each line is the ``user time'', that is time spent
5452 executing the program itself. The second number is ``system time'',
5453 time spent executing operating system routines on behalf of the program.
5454 Both numbers are in seconds.
5456 With the specification of an output file, the output is appended to the
5457 named file, and it looks like this:
5460 0.12 0.01 cc1 @var{options}
5461 0.00 0.01 as @var{options}
5464 The ``user time'' and the ``system time'' are moved before the program
5465 name, and the options passed to the program are displayed, so that one
5466 can later tell what file was being compiled, and with which options.
5468 @item -fvar-tracking
5469 @opindex fvar-tracking
5470 Run variable tracking pass. It computes where variables are stored at each
5471 position in code. Better debugging information is then generated
5472 (if the debugging information format supports this information).
5474 It is enabled by default when compiling with optimization (@option{-Os},
5475 @option{-O}, @option{-O2}, @dots{}), debugging information (@option{-g}) and
5476 the debug info format supports it.
5478 @item -fvar-tracking-assignments
5479 @opindex fvar-tracking-assignments
5480 @opindex fno-var-tracking-assignments
5481 Annotate assignments to user variables early in the compilation and
5482 attempt to carry the annotations over throughout the compilation all the
5483 way to the end, in an attempt to improve debug information while
5484 optimizing. Use of @option{-gdwarf-4} is recommended along with it.
5486 It can be enabled even if var-tracking is disabled, in which case
5487 annotations will be created and maintained, but discarded at the end.
5489 @item -fvar-tracking-assignments-toggle
5490 @opindex fvar-tracking-assignments-toggle
5491 @opindex fno-var-tracking-assignments-toggle
5492 Toggle @option{-fvar-tracking-assignments}, in the same way that
5493 @option{-gtoggle} toggles @option{-g}.
5495 @item -print-file-name=@var{library}
5496 @opindex print-file-name
5497 Print the full absolute name of the library file @var{library} that
5498 would be used when linking---and don't do anything else. With this
5499 option, GCC does not compile or link anything; it just prints the
5502 @item -print-multi-directory
5503 @opindex print-multi-directory
5504 Print the directory name corresponding to the multilib selected by any
5505 other switches present in the command line. This directory is supposed
5506 to exist in @env{GCC_EXEC_PREFIX}.
5508 @item -print-multi-lib
5509 @opindex print-multi-lib
5510 Print the mapping from multilib directory names to compiler switches
5511 that enable them. The directory name is separated from the switches by
5512 @samp{;}, and each switch starts with an @samp{@@} instead of the
5513 @samp{-}, without spaces between multiple switches. This is supposed to
5514 ease shell-processing.
5516 @item -print-prog-name=@var{program}
5517 @opindex print-prog-name
5518 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
5520 @item -print-libgcc-file-name
5521 @opindex print-libgcc-file-name
5522 Same as @option{-print-file-name=libgcc.a}.
5524 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
5525 but you do want to link with @file{libgcc.a}. You can do
5528 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
5531 @item -print-search-dirs
5532 @opindex print-search-dirs
5533 Print the name of the configured installation directory and a list of
5534 program and library directories @command{gcc} will search---and don't do anything else.
5536 This is useful when @command{gcc} prints the error message
5537 @samp{installation problem, cannot exec cpp0: No such file or directory}.
5538 To resolve this you either need to put @file{cpp0} and the other compiler
5539 components where @command{gcc} expects to find them, or you can set the environment
5540 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
5541 Don't forget the trailing @samp{/}.
5542 @xref{Environment Variables}.
5544 @item -print-sysroot
5545 @opindex print-sysroot
5546 Print the target sysroot directory that will be used during
5547 compilation. This is the target sysroot specified either at configure
5548 time or using the @option{--sysroot} option, possibly with an extra
5549 suffix that depends on compilation options. If no target sysroot is
5550 specified, the option prints nothing.
5552 @item -print-sysroot-headers-suffix
5553 @opindex print-sysroot-headers-suffix
5554 Print the suffix added to the target sysroot when searching for
5555 headers, or give an error if the compiler is not configured with such
5556 a suffix---and don't do anything else.
5559 @opindex dumpmachine
5560 Print the compiler's target machine (for example,
5561 @samp{i686-pc-linux-gnu})---and don't do anything else.
5564 @opindex dumpversion
5565 Print the compiler version (for example, @samp{3.0})---and don't do
5570 Print the compiler's built-in specs---and don't do anything else. (This
5571 is used when GCC itself is being built.) @xref{Spec Files}.
5573 @item -feliminate-unused-debug-types
5574 @opindex feliminate-unused-debug-types
5575 Normally, when producing DWARF2 output, GCC will emit debugging
5576 information for all types declared in a compilation
5577 unit, regardless of whether or not they are actually used
5578 in that compilation unit. Sometimes this is useful, such as
5579 if, in the debugger, you want to cast a value to a type that is
5580 not actually used in your program (but is declared). More often,
5581 however, this results in a significant amount of wasted space.
5582 With this option, GCC will avoid producing debug symbol output
5583 for types that are nowhere used in the source file being compiled.
5586 @node Optimize Options
5587 @section Options That Control Optimization
5588 @cindex optimize options
5589 @cindex options, optimization
5591 These options control various sorts of optimizations.
5593 Without any optimization option, the compiler's goal is to reduce the
5594 cost of compilation and to make debugging produce the expected
5595 results. Statements are independent: if you stop the program with a
5596 breakpoint between statements, you can then assign a new value to any
5597 variable or change the program counter to any other statement in the
5598 function and get exactly the results you would expect from the source
5601 Turning on optimization flags makes the compiler attempt to improve
5602 the performance and/or code size at the expense of compilation time
5603 and possibly the ability to debug the program.
5605 The compiler performs optimization based on the knowledge it has of the
5606 program. Compiling multiple files at once to a single output file mode allows
5607 the compiler to use information gained from all of the files when compiling
5610 Not all optimizations are controlled directly by a flag. Only
5611 optimizations that have a flag are listed in this section.
5613 Depending on the target and how GCC was configured, a slightly different
5614 set of optimizations may be enabled at each @option{-O} level than
5615 those listed here. You can invoke GCC with @samp{-Q --help=optimizers}
5616 to find out the exact set of optimizations that are enabled at each level.
5617 @xref{Overall Options}, for examples.
5624 Optimize. Optimizing compilation takes somewhat more time, and a lot
5625 more memory for a large function.
5627 With @option{-O}, the compiler tries to reduce code size and execution
5628 time, without performing any optimizations that take a great deal of
5631 @option{-O} turns on the following optimization flags:
5634 -fcprop-registers @gol
5637 -fdelayed-branch @gol
5639 -fguess-branch-probability @gol
5640 -fif-conversion2 @gol
5641 -fif-conversion @gol
5642 -fipa-pure-const @gol
5643 -fipa-reference @gol
5645 -fsplit-wide-types @gol
5646 -ftree-builtin-call-dce @gol
5649 -ftree-copyrename @gol
5651 -ftree-dominator-opts @gol
5653 -ftree-forwprop @gol
5661 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
5662 where doing so does not interfere with debugging.
5666 Optimize even more. GCC performs nearly all supported optimizations
5667 that do not involve a space-speed tradeoff.
5668 As compared to @option{-O}, this option increases both compilation time
5669 and the performance of the generated code.
5671 @option{-O2} turns on all optimization flags specified by @option{-O}. It
5672 also turns on the following optimization flags:
5673 @gccoptlist{-fthread-jumps @gol
5674 -falign-functions -falign-jumps @gol
5675 -falign-loops -falign-labels @gol
5678 -fcse-follow-jumps -fcse-skip-blocks @gol
5679 -fdelete-null-pointer-checks @gol
5680 -fexpensive-optimizations @gol
5681 -fgcse -fgcse-lm @gol
5682 -finline-small-functions @gol
5683 -findirect-inlining @gol
5685 -foptimize-sibling-calls @gol
5688 -freorder-blocks -freorder-functions @gol
5689 -frerun-cse-after-loop @gol
5690 -fsched-interblock -fsched-spec @gol
5691 -fschedule-insns -fschedule-insns2 @gol
5692 -fstrict-aliasing -fstrict-overflow @gol
5693 -ftree-switch-conversion @gol
5697 Please note the warning under @option{-fgcse} about
5698 invoking @option{-O2} on programs that use computed gotos.
5702 Optimize yet more. @option{-O3} turns on all optimizations specified
5703 by @option{-O2} and also turns on the @option{-finline-functions},
5704 @option{-funswitch-loops}, @option{-fpredictive-commoning},
5705 @option{-fgcse-after-reload} and @option{-ftree-vectorize} options.
5709 Reduce compilation time and make debugging produce the expected
5710 results. This is the default.
5714 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
5715 do not typically increase code size. It also performs further
5716 optimizations designed to reduce code size.
5718 @option{-Os} disables the following optimization flags:
5719 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
5720 -falign-labels -freorder-blocks -freorder-blocks-and-partition @gol
5721 -fprefetch-loop-arrays -ftree-vect-loop-version}
5723 If you use multiple @option{-O} options, with or without level numbers,
5724 the last such option is the one that is effective.
5727 Options of the form @option{-f@var{flag}} specify machine-independent
5728 flags. Most flags have both positive and negative forms; the negative
5729 form of @option{-ffoo} would be @option{-fno-foo}. In the table
5730 below, only one of the forms is listed---the one you typically will
5731 use. You can figure out the other form by either removing @samp{no-}
5734 The following options control specific optimizations. They are either
5735 activated by @option{-O} options or are related to ones that are. You
5736 can use the following flags in the rare cases when ``fine-tuning'' of
5737 optimizations to be performed is desired.
5740 @item -fno-default-inline
5741 @opindex fno-default-inline
5742 Do not make member functions inline by default merely because they are
5743 defined inside the class scope (C++ only). Otherwise, when you specify
5744 @w{@option{-O}}, member functions defined inside class scope are compiled
5745 inline by default; i.e., you don't need to add @samp{inline} in front of
5746 the member function name.
5748 @item -fno-defer-pop
5749 @opindex fno-defer-pop
5750 Always pop the arguments to each function call as soon as that function
5751 returns. For machines which must pop arguments after a function call,
5752 the compiler normally lets arguments accumulate on the stack for several
5753 function calls and pops them all at once.
5755 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5757 @item -fforward-propagate
5758 @opindex fforward-propagate
5759 Perform a forward propagation pass on RTL@. The pass tries to combine two
5760 instructions and checks if the result can be simplified. If loop unrolling
5761 is active, two passes are performed and the second is scheduled after
5764 This option is enabled by default at optimization levels @option{-O},
5765 @option{-O2}, @option{-O3}, @option{-Os}.
5767 @item -fomit-frame-pointer
5768 @opindex fomit-frame-pointer
5769 Don't keep the frame pointer in a register for functions that
5770 don't need one. This avoids the instructions to save, set up and
5771 restore frame pointers; it also makes an extra register available
5772 in many functions. @strong{It also makes debugging impossible on
5775 On some machines, such as the VAX, this flag has no effect, because
5776 the standard calling sequence automatically handles the frame pointer
5777 and nothing is saved by pretending it doesn't exist. The
5778 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
5779 whether a target machine supports this flag. @xref{Registers,,Register
5780 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
5782 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5784 @item -foptimize-sibling-calls
5785 @opindex foptimize-sibling-calls
5786 Optimize sibling and tail recursive calls.
5788 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
5792 Don't pay attention to the @code{inline} keyword. Normally this option
5793 is used to keep the compiler from expanding any functions inline.
5794 Note that if you are not optimizing, no functions can be expanded inline.
5796 @item -finline-small-functions
5797 @opindex finline-small-functions
5798 Integrate functions into their callers when their body is smaller than expected
5799 function call code (so overall size of program gets smaller). The compiler
5800 heuristically decides which functions are simple enough to be worth integrating
5803 Enabled at level @option{-O2}.
5805 @item -findirect-inlining
5806 @opindex findirect-inlining
5807 Inline also indirect calls that are discovered to be known at compile
5808 time thanks to previous inlining. This option has any effect only
5809 when inlining itself is turned on by the @option{-finline-functions}
5810 or @option{-finline-small-functions} options.
5812 Enabled at level @option{-O2}.
5814 @item -finline-functions
5815 @opindex finline-functions
5816 Integrate all simple functions into their callers. The compiler
5817 heuristically decides which functions are simple enough to be worth
5818 integrating in this way.
5820 If all calls to a given function are integrated, and the function is
5821 declared @code{static}, then the function is normally not output as
5822 assembler code in its own right.
5824 Enabled at level @option{-O3}.
5826 @item -finline-functions-called-once
5827 @opindex finline-functions-called-once
5828 Consider all @code{static} functions called once for inlining into their
5829 caller even if they are not marked @code{inline}. If a call to a given
5830 function is integrated, then the function is not output as assembler code
5833 Enabled at levels @option{-O1}, @option{-O2}, @option{-O3} and @option{-Os}.
5835 @item -fearly-inlining
5836 @opindex fearly-inlining
5837 Inline functions marked by @code{always_inline} and functions whose body seems
5838 smaller than the function call overhead early before doing
5839 @option{-fprofile-generate} instrumentation and real inlining pass. Doing so
5840 makes profiling significantly cheaper and usually inlining faster on programs
5841 having large chains of nested wrapper functions.
5847 Perform interprocedural scalar replacement of aggregates, removal of
5848 unused parameters and replacement of parameters passed by reference
5849 by parameters passed by value.
5851 Enabled at levels @option{-O2}, @option{-O3} and @option{-Os}.
5853 @item -finline-limit=@var{n}
5854 @opindex finline-limit
5855 By default, GCC limits the size of functions that can be inlined. This flag
5856 allows coarse control of this limit. @var{n} is the size of functions that
5857 can be inlined in number of pseudo instructions.
5859 Inlining is actually controlled by a number of parameters, which may be
5860 specified individually by using @option{--param @var{name}=@var{value}}.
5861 The @option{-finline-limit=@var{n}} option sets some of these parameters
5865 @item max-inline-insns-single
5866 is set to @var{n}/2.
5867 @item max-inline-insns-auto
5868 is set to @var{n}/2.
5871 See below for a documentation of the individual
5872 parameters controlling inlining and for the defaults of these parameters.
5874 @emph{Note:} there may be no value to @option{-finline-limit} that results
5875 in default behavior.
5877 @emph{Note:} pseudo instruction represents, in this particular context, an
5878 abstract measurement of function's size. In no way does it represent a count
5879 of assembly instructions and as such its exact meaning might change from one
5880 release to an another.
5882 @item -fkeep-inline-functions
5883 @opindex fkeep-inline-functions
5884 In C, emit @code{static} functions that are declared @code{inline}
5885 into the object file, even if the function has been inlined into all
5886 of its callers. This switch does not affect functions using the
5887 @code{extern inline} extension in GNU C89@. In C++, emit any and all
5888 inline functions into the object file.
5890 @item -fkeep-static-consts
5891 @opindex fkeep-static-consts
5892 Emit variables declared @code{static const} when optimization isn't turned
5893 on, even if the variables aren't referenced.
5895 GCC enables this option by default. If you want to force the compiler to
5896 check if the variable was referenced, regardless of whether or not
5897 optimization is turned on, use the @option{-fno-keep-static-consts} option.
5899 @item -fmerge-constants
5900 @opindex fmerge-constants
5901 Attempt to merge identical constants (string constants and floating point
5902 constants) across compilation units.
5904 This option is the default for optimized compilation if the assembler and
5905 linker support it. Use @option{-fno-merge-constants} to inhibit this
5908 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5910 @item -fmerge-all-constants
5911 @opindex fmerge-all-constants
5912 Attempt to merge identical constants and identical variables.
5914 This option implies @option{-fmerge-constants}. In addition to
5915 @option{-fmerge-constants} this considers e.g.@: even constant initialized
5916 arrays or initialized constant variables with integral or floating point
5917 types. Languages like C or C++ require each variable, including multiple
5918 instances of the same variable in recursive calls, to have distinct locations,
5919 so using this option will result in non-conforming
5922 @item -fmodulo-sched
5923 @opindex fmodulo-sched
5924 Perform swing modulo scheduling immediately before the first scheduling
5925 pass. This pass looks at innermost loops and reorders their
5926 instructions by overlapping different iterations.
5928 @item -fmodulo-sched-allow-regmoves
5929 @opindex fmodulo-sched-allow-regmoves
5930 Perform more aggressive SMS based modulo scheduling with register moves
5931 allowed. By setting this flag certain anti-dependences edges will be
5932 deleted which will trigger the generation of reg-moves based on the
5933 life-range analysis. This option is effective only with
5934 @option{-fmodulo-sched} enabled.
5936 @item -fno-branch-count-reg
5937 @opindex fno-branch-count-reg
5938 Do not use ``decrement and branch'' instructions on a count register,
5939 but instead generate a sequence of instructions that decrement a
5940 register, compare it against zero, then branch based upon the result.
5941 This option is only meaningful on architectures that support such
5942 instructions, which include x86, PowerPC, IA-64 and S/390.
5944 The default is @option{-fbranch-count-reg}.
5946 @item -fno-function-cse
5947 @opindex fno-function-cse
5948 Do not put function addresses in registers; make each instruction that
5949 calls a constant function contain the function's address explicitly.
5951 This option results in less efficient code, but some strange hacks
5952 that alter the assembler output may be confused by the optimizations
5953 performed when this option is not used.
5955 The default is @option{-ffunction-cse}
5957 @item -fno-zero-initialized-in-bss
5958 @opindex fno-zero-initialized-in-bss
5959 If the target supports a BSS section, GCC by default puts variables that
5960 are initialized to zero into BSS@. This can save space in the resulting
5963 This option turns off this behavior because some programs explicitly
5964 rely on variables going to the data section. E.g., so that the
5965 resulting executable can find the beginning of that section and/or make
5966 assumptions based on that.
5968 The default is @option{-fzero-initialized-in-bss}.
5970 @item -fmudflap -fmudflapth -fmudflapir
5974 @cindex bounds checking
5976 For front-ends that support it (C and C++), instrument all risky
5977 pointer/array dereferencing operations, some standard library
5978 string/heap functions, and some other associated constructs with
5979 range/validity tests. Modules so instrumented should be immune to
5980 buffer overflows, invalid heap use, and some other classes of C/C++
5981 programming errors. The instrumentation relies on a separate runtime
5982 library (@file{libmudflap}), which will be linked into a program if
5983 @option{-fmudflap} is given at link time. Run-time behavior of the
5984 instrumented program is controlled by the @env{MUDFLAP_OPTIONS}
5985 environment variable. See @code{env MUDFLAP_OPTIONS=-help a.out}
5988 Use @option{-fmudflapth} instead of @option{-fmudflap} to compile and to
5989 link if your program is multi-threaded. Use @option{-fmudflapir}, in
5990 addition to @option{-fmudflap} or @option{-fmudflapth}, if
5991 instrumentation should ignore pointer reads. This produces less
5992 instrumentation (and therefore faster execution) and still provides
5993 some protection against outright memory corrupting writes, but allows
5994 erroneously read data to propagate within a program.
5996 @item -fthread-jumps
5997 @opindex fthread-jumps
5998 Perform optimizations where we check to see if a jump branches to a
5999 location where another comparison subsumed by the first is found. If
6000 so, the first branch is redirected to either the destination of the
6001 second branch or a point immediately following it, depending on whether
6002 the condition is known to be true or false.
6004 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6006 @item -fsplit-wide-types
6007 @opindex fsplit-wide-types
6008 When using a type that occupies multiple registers, such as @code{long
6009 long} on a 32-bit system, split the registers apart and allocate them
6010 independently. This normally generates better code for those types,
6011 but may make debugging more difficult.
6013 Enabled at levels @option{-O}, @option{-O2}, @option{-O3},
6016 @item -fcse-follow-jumps
6017 @opindex fcse-follow-jumps
6018 In common subexpression elimination (CSE), scan through jump instructions
6019 when the target of the jump is not reached by any other path. For
6020 example, when CSE encounters an @code{if} statement with an
6021 @code{else} clause, CSE will follow the jump when the condition
6024 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6026 @item -fcse-skip-blocks
6027 @opindex fcse-skip-blocks
6028 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
6029 follow jumps which conditionally skip over blocks. When CSE
6030 encounters a simple @code{if} statement with no else clause,
6031 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
6032 body of the @code{if}.
6034 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6036 @item -frerun-cse-after-loop
6037 @opindex frerun-cse-after-loop
6038 Re-run common subexpression elimination after loop optimizations has been
6041 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6045 Perform a global common subexpression elimination pass.
6046 This pass also performs global constant and copy propagation.
6048 @emph{Note:} When compiling a program using computed gotos, a GCC
6049 extension, you may get better runtime performance if you disable
6050 the global common subexpression elimination pass by adding
6051 @option{-fno-gcse} to the command line.
6053 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6057 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
6058 attempt to move loads which are only killed by stores into themselves. This
6059 allows a loop containing a load/store sequence to be changed to a load outside
6060 the loop, and a copy/store within the loop.
6062 Enabled by default when gcse is enabled.
6066 When @option{-fgcse-sm} is enabled, a store motion pass is run after
6067 global common subexpression elimination. This pass will attempt to move
6068 stores out of loops. When used in conjunction with @option{-fgcse-lm},
6069 loops containing a load/store sequence can be changed to a load before
6070 the loop and a store after the loop.
6072 Not enabled at any optimization level.
6076 When @option{-fgcse-las} is enabled, the global common subexpression
6077 elimination pass eliminates redundant loads that come after stores to the
6078 same memory location (both partial and full redundancies).
6080 Not enabled at any optimization level.
6082 @item -fgcse-after-reload
6083 @opindex fgcse-after-reload
6084 When @option{-fgcse-after-reload} is enabled, a redundant load elimination
6085 pass is performed after reload. The purpose of this pass is to cleanup
6088 @item -funsafe-loop-optimizations
6089 @opindex funsafe-loop-optimizations
6090 If given, the loop optimizer will assume that loop indices do not
6091 overflow, and that the loops with nontrivial exit condition are not
6092 infinite. This enables a wider range of loop optimizations even if
6093 the loop optimizer itself cannot prove that these assumptions are valid.
6094 Using @option{-Wunsafe-loop-optimizations}, the compiler will warn you
6095 if it finds this kind of loop.
6097 @item -fcrossjumping
6098 @opindex fcrossjumping
6099 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
6100 resulting code may or may not perform better than without cross-jumping.
6102 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6104 @item -fauto-inc-dec
6105 @opindex fauto-inc-dec
6106 Combine increments or decrements of addresses with memory accesses.
6107 This pass is always skipped on architectures that do not have
6108 instructions to support this. Enabled by default at @option{-O} and
6109 higher on architectures that support this.
6113 Perform dead code elimination (DCE) on RTL@.
6114 Enabled by default at @option{-O} and higher.
6118 Perform dead store elimination (DSE) on RTL@.
6119 Enabled by default at @option{-O} and higher.
6121 @item -fif-conversion
6122 @opindex fif-conversion
6123 Attempt to transform conditional jumps into branch-less equivalents. This
6124 include use of conditional moves, min, max, set flags and abs instructions, and
6125 some tricks doable by standard arithmetics. The use of conditional execution
6126 on chips where it is available is controlled by @code{if-conversion2}.
6128 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6130 @item -fif-conversion2
6131 @opindex fif-conversion2
6132 Use conditional execution (where available) to transform conditional jumps into
6133 branch-less equivalents.
6135 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6137 @item -fdelete-null-pointer-checks
6138 @opindex fdelete-null-pointer-checks
6139 Assume that programs cannot safely dereference null pointers, and that
6140 no code or data element resides there. This enables simple constant
6141 folding optimizations at all optimization levels. In addition, other
6142 optimization passes in GCC use this flag to control global dataflow
6143 analyses that eliminate useless checks for null pointers; these assume
6144 that if a pointer is checked after it has already been dereferenced,
6147 Note however that in some environments this assumption is not true.
6148 Use @option{-fno-delete-null-pointer-checks} to disable this optimization
6149 for programs which depend on that behavior.
6151 Some targets, especially embedded ones, disable this option at all levels.
6152 Otherwise it is enabled at all levels: @option{-O0}, @option{-O1},
6153 @option{-O2}, @option{-O3}, @option{-Os}. Passes that use the information
6154 are enabled independently at different optimization levels.
6156 @item -fexpensive-optimizations
6157 @opindex fexpensive-optimizations
6158 Perform a number of minor optimizations that are relatively expensive.
6160 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6162 @item -foptimize-register-move
6164 @opindex foptimize-register-move
6166 Attempt to reassign register numbers in move instructions and as
6167 operands of other simple instructions in order to maximize the amount of
6168 register tying. This is especially helpful on machines with two-operand
6171 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
6174 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6176 @item -fira-algorithm=@var{algorithm}
6177 Use specified coloring algorithm for the integrated register
6178 allocator. The @var{algorithm} argument should be @code{priority} or
6179 @code{CB}. The first algorithm specifies Chow's priority coloring,
6180 the second one specifies Chaitin-Briggs coloring. The second
6181 algorithm can be unimplemented for some architectures. If it is
6182 implemented, it is the default because Chaitin-Briggs coloring as a
6183 rule generates a better code.
6185 @item -fira-region=@var{region}
6186 Use specified regions for the integrated register allocator. The
6187 @var{region} argument should be one of @code{all}, @code{mixed}, or
6188 @code{one}. The first value means using all loops as register
6189 allocation regions, the second value which is the default means using
6190 all loops except for loops with small register pressure as the
6191 regions, and third one means using all function as a single region.
6192 The first value can give best result for machines with small size and
6193 irregular register set, the third one results in faster and generates
6194 decent code and the smallest size code, and the default value usually
6195 give the best results in most cases and for most architectures.
6197 @item -fira-coalesce
6198 @opindex fira-coalesce
6199 Do optimistic register coalescing. This option might be profitable for
6200 architectures with big regular register files.
6202 @item -fno-ira-share-save-slots
6203 @opindex fno-ira-share-save-slots
6204 Switch off sharing stack slots used for saving call used hard
6205 registers living through a call. Each hard register will get a
6206 separate stack slot and as a result function stack frame will be
6209 @item -fno-ira-share-spill-slots
6210 @opindex fno-ira-share-spill-slots
6211 Switch off sharing stack slots allocated for pseudo-registers. Each
6212 pseudo-register which did not get a hard register will get a separate
6213 stack slot and as a result function stack frame will be bigger.
6215 @item -fira-verbose=@var{n}
6216 @opindex fira-verbose
6217 Set up how verbose dump file for the integrated register allocator
6218 will be. Default value is 5. If the value is greater or equal to 10,
6219 the dump file will be stderr as if the value were @var{n} minus 10.
6221 @item -fdelayed-branch
6222 @opindex fdelayed-branch
6223 If supported for the target machine, attempt to reorder instructions
6224 to exploit instruction slots available after delayed branch
6227 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6229 @item -fschedule-insns
6230 @opindex fschedule-insns
6231 If supported for the target machine, attempt to reorder instructions to
6232 eliminate execution stalls due to required data being unavailable. This
6233 helps machines that have slow floating point or memory load instructions
6234 by allowing other instructions to be issued until the result of the load
6235 or floating point instruction is required.
6237 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6239 @item -fschedule-insns2
6240 @opindex fschedule-insns2
6241 Similar to @option{-fschedule-insns}, but requests an additional pass of
6242 instruction scheduling after register allocation has been done. This is
6243 especially useful on machines with a relatively small number of
6244 registers and where memory load instructions take more than one cycle.
6246 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6248 @item -fno-sched-interblock
6249 @opindex fno-sched-interblock
6250 Don't schedule instructions across basic blocks. This is normally
6251 enabled by default when scheduling before register allocation, i.e.@:
6252 with @option{-fschedule-insns} or at @option{-O2} or higher.
6254 @item -fno-sched-spec
6255 @opindex fno-sched-spec
6256 Don't allow speculative motion of non-load instructions. This is normally
6257 enabled by default when scheduling before register allocation, i.e.@:
6258 with @option{-fschedule-insns} or at @option{-O2} or higher.
6260 @item -fsched-pressure
6261 @opindex fsched-pressure
6262 Enable register pressure sensitive insn scheduling before the register
6263 allocation. This only makes sense when scheduling before register
6264 allocation is enabled, i.e.@: with @option{-fschedule-insns} or at
6265 @option{-O2} or higher. Usage of this option can improve the
6266 generated code and decrease its size by preventing register pressure
6267 increase above the number of available hard registers and as a
6268 consequence register spills in the register allocation.
6270 @item -fsched-spec-load
6271 @opindex fsched-spec-load
6272 Allow speculative motion of some load instructions. This only makes
6273 sense when scheduling before register allocation, i.e.@: with
6274 @option{-fschedule-insns} or at @option{-O2} or higher.
6276 @item -fsched-spec-load-dangerous
6277 @opindex fsched-spec-load-dangerous
6278 Allow speculative motion of more load instructions. This only makes
6279 sense when scheduling before register allocation, i.e.@: with
6280 @option{-fschedule-insns} or at @option{-O2} or higher.
6282 @item -fsched-stalled-insns
6283 @itemx -fsched-stalled-insns=@var{n}
6284 @opindex fsched-stalled-insns
6285 Define how many insns (if any) can be moved prematurely from the queue
6286 of stalled insns into the ready list, during the second scheduling pass.
6287 @option{-fno-sched-stalled-insns} means that no insns will be moved
6288 prematurely, @option{-fsched-stalled-insns=0} means there is no limit
6289 on how many queued insns can be moved prematurely.
6290 @option{-fsched-stalled-insns} without a value is equivalent to
6291 @option{-fsched-stalled-insns=1}.
6293 @item -fsched-stalled-insns-dep
6294 @itemx -fsched-stalled-insns-dep=@var{n}
6295 @opindex fsched-stalled-insns-dep
6296 Define how many insn groups (cycles) will be examined for a dependency
6297 on a stalled insn that is candidate for premature removal from the queue
6298 of stalled insns. This has an effect only during the second scheduling pass,
6299 and only if @option{-fsched-stalled-insns} is used.
6300 @option{-fno-sched-stalled-insns-dep} is equivalent to
6301 @option{-fsched-stalled-insns-dep=0}.
6302 @option{-fsched-stalled-insns-dep} without a value is equivalent to
6303 @option{-fsched-stalled-insns-dep=1}.
6305 @item -fsched2-use-superblocks
6306 @opindex fsched2-use-superblocks
6307 When scheduling after register allocation, do use superblock scheduling
6308 algorithm. Superblock scheduling allows motion across basic block boundaries
6309 resulting on faster schedules. This option is experimental, as not all machine
6310 descriptions used by GCC model the CPU closely enough to avoid unreliable
6311 results from the algorithm.
6313 This only makes sense when scheduling after register allocation, i.e.@: with
6314 @option{-fschedule-insns2} or at @option{-O2} or higher.
6316 @item -fsched-group-heuristic
6317 @opindex fsched-group-heuristic
6318 Enable the group heuristic in the scheduler. This heuristic favors
6319 the instruction that belongs to a schedule group. This is enabled
6320 by default when scheduling is enabled, i.e.@: with @option{-fschedule-insns}
6321 or @option{-fschedule-insns2} or at @option{-O2} or higher.
6323 @item -fsched-critical-path-heuristic
6324 @opindex fsched-critical-path-heuristic
6325 Enable the critical-path heuristic in the scheduler. This heuristic favors
6326 instructions on the critical path. This is enabled by default when
6327 scheduling is enabled, i.e.@: with @option{-fschedule-insns}
6328 or @option{-fschedule-insns2} or at @option{-O2} or higher.
6330 @item -fsched-spec-insn-heuristic
6331 @opindex fsched-spec-insn-heuristic
6332 Enable the speculative instruction heuristic in the scheduler. This
6333 heuristic favors speculative instructions with greater dependency weakness.
6334 This is enabled by default when scheduling is enabled, i.e.@:
6335 with @option{-fschedule-insns} or @option{-fschedule-insns2}
6336 or at @option{-O2} or higher.
6338 @item -fsched-rank-heuristic
6339 @opindex fsched-rank-heuristic
6340 Enable the rank heuristic in the scheduler. This heuristic favors
6341 the instruction belonging to a basic block with greater size or frequency.
6342 This is enabled by default when scheduling is enabled, i.e.@:
6343 with @option{-fschedule-insns} or @option{-fschedule-insns2} or
6344 at @option{-O2} or higher.
6346 @item -fsched-last-insn-heuristic
6347 @opindex fsched-last-insn-heuristic
6348 Enable the last-instruction heuristic in the scheduler. This heuristic
6349 favors the instruction that is less dependent on the last instruction
6350 scheduled. This is enabled by default when scheduling is enabled,
6351 i.e.@: with @option{-fschedule-insns} or @option{-fschedule-insns2} or
6352 at @option{-O2} or higher.
6354 @item -fsched-dep-count-heuristic
6355 @opindex fsched-dep-count-heuristic
6356 Enable the dependent-count heuristic in the scheduler. This heuristic
6357 favors the instruction that has more instructions depending on it.
6358 This is enabled by default when scheduling is enabled, i.e.@:
6359 with @option{-fschedule-insns} or @option{-fschedule-insns2} or
6360 at @option{-O2} or higher.
6362 @item -fsched2-use-traces
6363 @opindex fsched2-use-traces
6364 Use @option{-fsched2-use-superblocks} algorithm when scheduling after register
6365 allocation and additionally perform code duplication in order to increase the
6366 size of superblocks using tracer pass. See @option{-ftracer} for details on
6369 This mode should produce faster but significantly longer programs. Also
6370 without @option{-fbranch-probabilities} the traces constructed may not
6371 match the reality and hurt the performance. This only makes
6372 sense when scheduling after register allocation, i.e.@: with
6373 @option{-fschedule-insns2} or at @option{-O2} or higher.
6375 @item -freschedule-modulo-scheduled-loops
6376 @opindex freschedule-modulo-scheduled-loops
6377 The modulo scheduling comes before the traditional scheduling, if a loop
6378 was modulo scheduled we may want to prevent the later scheduling passes
6379 from changing its schedule, we use this option to control that.
6381 @item -fselective-scheduling
6382 @opindex fselective-scheduling
6383 Schedule instructions using selective scheduling algorithm. Selective
6384 scheduling runs instead of the first scheduler pass.
6386 @item -fselective-scheduling2
6387 @opindex fselective-scheduling2
6388 Schedule instructions using selective scheduling algorithm. Selective
6389 scheduling runs instead of the second scheduler pass.
6391 @item -fsel-sched-pipelining
6392 @opindex fsel-sched-pipelining
6393 Enable software pipelining of innermost loops during selective scheduling.
6394 This option has no effect until one of @option{-fselective-scheduling} or
6395 @option{-fselective-scheduling2} is turned on.
6397 @item -fsel-sched-pipelining-outer-loops
6398 @opindex fsel-sched-pipelining-outer-loops
6399 When pipelining loops during selective scheduling, also pipeline outer loops.
6400 This option has no effect until @option{-fsel-sched-pipelining} is turned on.
6402 @item -fcaller-saves
6403 @opindex fcaller-saves
6404 Enable values to be allocated in registers that will be clobbered by
6405 function calls, by emitting extra instructions to save and restore the
6406 registers around such calls. Such allocation is done only when it
6407 seems to result in better code than would otherwise be produced.
6409 This option is always enabled by default on certain machines, usually
6410 those which have no call-preserved registers to use instead.
6412 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6414 @item -fconserve-stack
6415 @opindex fconserve-stack
6416 Attempt to minimize stack usage. The compiler will attempt to use less
6417 stack space, even if that makes the program slower. This option
6418 implies setting the @option{large-stack-frame} parameter to 100
6419 and the @option{large-stack-frame-growth} parameter to 400.
6421 @item -ftree-reassoc
6422 @opindex ftree-reassoc
6423 Perform reassociation on trees. This flag is enabled by default
6424 at @option{-O} and higher.
6428 Perform partial redundancy elimination (PRE) on trees. This flag is
6429 enabled by default at @option{-O2} and @option{-O3}.
6431 @item -ftree-forwprop
6432 @opindex ftree-forwprop
6433 Perform forward propagation on trees. This flag is enabled by default
6434 at @option{-O} and higher.
6438 Perform full redundancy elimination (FRE) on trees. The difference
6439 between FRE and PRE is that FRE only considers expressions
6440 that are computed on all paths leading to the redundant computation.
6441 This analysis is faster than PRE, though it exposes fewer redundancies.
6442 This flag is enabled by default at @option{-O} and higher.
6444 @item -ftree-phiprop
6445 @opindex ftree-phiprop
6446 Perform hoisting of loads from conditional pointers on trees. This
6447 pass is enabled by default at @option{-O} and higher.
6449 @item -ftree-copy-prop
6450 @opindex ftree-copy-prop
6451 Perform copy propagation on trees. This pass eliminates unnecessary
6452 copy operations. This flag is enabled by default at @option{-O} and
6455 @item -fipa-pure-const
6456 @opindex fipa-pure-const
6457 Discover which functions are pure or constant.
6458 Enabled by default at @option{-O} and higher.
6460 @item -fipa-reference
6461 @opindex fipa-reference
6462 Discover which static variables do not escape cannot escape the
6464 Enabled by default at @option{-O} and higher.
6466 @item -fipa-struct-reorg
6467 @opindex fipa-struct-reorg
6468 Perform structure reorganization optimization, that change C-like structures
6469 layout in order to better utilize spatial locality. This transformation is
6470 affective for programs containing arrays of structures. Available in two
6471 compilation modes: profile-based (enabled with @option{-fprofile-generate})
6472 or static (which uses built-in heuristics). Require @option{-fipa-type-escape}
6473 to provide the safety of this transformation. It works only in whole program
6474 mode, so it requires @option{-fwhole-program} and @option{-combine} to be
6475 enabled. Structures considered @samp{cold} by this transformation are not
6476 affected (see @option{--param struct-reorg-cold-struct-ratio=@var{value}}).
6478 With this flag, the program debug info reflects a new structure layout.
6482 Perform interprocedural pointer analysis. This option is experimental
6483 and does not affect generated code.
6487 Perform interprocedural constant propagation.
6488 This optimization analyzes the program to determine when values passed
6489 to functions are constants and then optimizes accordingly.
6490 This optimization can substantially increase performance
6491 if the application has constants passed to functions.
6492 This flag is enabled by default at @option{-O2}, @option{-Os} and @option{-O3}.
6494 @item -fipa-cp-clone
6495 @opindex fipa-cp-clone
6496 Perform function cloning to make interprocedural constant propagation stronger.
6497 When enabled, interprocedural constant propagation will perform function cloning
6498 when externally visible function can be called with constant arguments.
6499 Because this optimization can create multiple copies of functions,
6500 it may significantly increase code size
6501 (see @option{--param ipcp-unit-growth=@var{value}}).
6502 This flag is enabled by default at @option{-O3}.
6504 @item -fipa-matrix-reorg
6505 @opindex fipa-matrix-reorg
6506 Perform matrix flattening and transposing.
6507 Matrix flattening tries to replace an @math{m}-dimensional matrix
6508 with its equivalent @math{n}-dimensional matrix, where @math{n < m}.
6509 This reduces the level of indirection needed for accessing the elements
6510 of the matrix. The second optimization is matrix transposing that
6511 attempts to change the order of the matrix's dimensions in order to
6512 improve cache locality.
6513 Both optimizations need the @option{-fwhole-program} flag.
6514 Transposing is enabled only if profiling information is available.
6518 Perform forward store motion on trees. This flag is
6519 enabled by default at @option{-O} and higher.
6523 Perform sparse conditional constant propagation (CCP) on trees. This
6524 pass only operates on local scalar variables and is enabled by default
6525 at @option{-O} and higher.
6527 @item -ftree-switch-conversion
6528 Perform conversion of simple initializations in a switch to
6529 initializations from a scalar array. This flag is enabled by default
6530 at @option{-O2} and higher.
6534 Perform dead code elimination (DCE) on trees. This flag is enabled by
6535 default at @option{-O} and higher.
6537 @item -ftree-builtin-call-dce
6538 @opindex ftree-builtin-call-dce
6539 Perform conditional dead code elimination (DCE) for calls to builtin functions
6540 that may set @code{errno} but are otherwise side-effect free. This flag is
6541 enabled by default at @option{-O2} and higher if @option{-Os} is not also
6544 @item -ftree-dominator-opts
6545 @opindex ftree-dominator-opts
6546 Perform a variety of simple scalar cleanups (constant/copy
6547 propagation, redundancy elimination, range propagation and expression
6548 simplification) based on a dominator tree traversal. This also
6549 performs jump threading (to reduce jumps to jumps). This flag is
6550 enabled by default at @option{-O} and higher.
6554 Perform dead store elimination (DSE) on trees. A dead store is a store into
6555 a memory location which will later be overwritten by another store without
6556 any intervening loads. In this case the earlier store can be deleted. This
6557 flag is enabled by default at @option{-O} and higher.
6561 Perform loop header copying on trees. This is beneficial since it increases
6562 effectiveness of code motion optimizations. It also saves one jump. This flag
6563 is enabled by default at @option{-O} and higher. It is not enabled
6564 for @option{-Os}, since it usually increases code size.
6566 @item -ftree-loop-optimize
6567 @opindex ftree-loop-optimize
6568 Perform loop optimizations on trees. This flag is enabled by default
6569 at @option{-O} and higher.
6571 @item -ftree-loop-linear
6572 @opindex ftree-loop-linear
6573 Perform linear loop transformations on tree. This flag can improve cache
6574 performance and allow further loop optimizations to take place.
6576 @item -floop-interchange
6577 Perform loop interchange transformations on loops. Interchanging two
6578 nested loops switches the inner and outer loops. For example, given a
6583 A(J, I) = A(J, I) * C
6587 loop interchange will transform the loop as if the user had written:
6591 A(J, I) = A(J, I) * C
6595 which can be beneficial when @code{N} is larger than the caches,
6596 because in Fortran, the elements of an array are stored in memory
6597 contiguously by column, and the original loop iterates over rows,
6598 potentially creating at each access a cache miss. This optimization
6599 applies to all the languages supported by GCC and is not limited to
6600 Fortran. To use this code transformation, GCC has to be configured
6601 with @option{--with-ppl} and @option{--with-cloog} to enable the
6602 Graphite loop transformation infrastructure.
6604 @item -floop-strip-mine
6605 Perform loop strip mining transformations on loops. Strip mining
6606 splits a loop into two nested loops. The outer loop has strides
6607 equal to the strip size and the inner loop has strides of the
6608 original loop within a strip. For example, given a loop like:
6614 loop strip mining will transform the loop as if the user had written:
6617 DO I = II, min (II + 3, N)
6622 This optimization applies to all the languages supported by GCC and is
6623 not limited to Fortran. To use this code transformation, GCC has to
6624 be configured with @option{--with-ppl} and @option{--with-cloog} to
6625 enable the Graphite loop transformation infrastructure.
6628 Perform loop blocking transformations on loops. Blocking strip mines
6629 each loop in the loop nest such that the memory accesses of the
6630 element loops fit inside caches. For example, given a loop like:
6634 A(J, I) = B(I) + C(J)
6638 loop blocking will transform the loop as if the user had written:
6642 DO I = II, min (II + 63, N)
6643 DO J = JJ, min (JJ + 63, M)
6644 A(J, I) = B(I) + C(J)
6650 which can be beneficial when @code{M} is larger than the caches,
6651 because the innermost loop will iterate over a smaller amount of data
6652 that can be kept in the caches. This optimization applies to all the
6653 languages supported by GCC and is not limited to Fortran. To use this
6654 code transformation, GCC has to be configured with @option{--with-ppl}
6655 and @option{--with-cloog} to enable the Graphite loop transformation
6658 @item -fgraphite-identity
6659 @opindex fgraphite-identity
6660 Enable the identity transformation for graphite. For every SCoP we generate
6661 the polyhedral representation and transform it back to gimple. Using
6662 @option{-fgraphite-identity} we can check the costs or benefits of the
6663 GIMPLE -> GRAPHITE -> GIMPLE transformation. Some minimal optimizations
6664 are also performed by the code generator CLooG, like index splitting and
6665 dead code elimination in loops.
6667 @item -floop-parallelize-all
6668 Use the Graphite data dependence analysis to identify loops that can
6669 be parallelized. Parallelize all the loops that can be analyzed to
6670 not contain loop carried dependences without checking that it is
6671 profitable to parallelize the loops.
6673 @item -fcheck-data-deps
6674 @opindex fcheck-data-deps
6675 Compare the results of several data dependence analyzers. This option
6676 is used for debugging the data dependence analyzers.
6678 @item -ftree-loop-distribution
6679 Perform loop distribution. This flag can improve cache performance on
6680 big loop bodies and allow further loop optimizations, like
6681 parallelization or vectorization, to take place. For example, the loop
6698 @item -ftree-loop-im
6699 @opindex ftree-loop-im
6700 Perform loop invariant motion on trees. This pass moves only invariants that
6701 would be hard to handle at RTL level (function calls, operations that expand to
6702 nontrivial sequences of insns). With @option{-funswitch-loops} it also moves
6703 operands of conditions that are invariant out of the loop, so that we can use
6704 just trivial invariantness analysis in loop unswitching. The pass also includes
6707 @item -ftree-loop-ivcanon
6708 @opindex ftree-loop-ivcanon
6709 Create a canonical counter for number of iterations in the loop for that
6710 determining number of iterations requires complicated analysis. Later
6711 optimizations then may determine the number easily. Useful especially
6712 in connection with unrolling.
6716 Perform induction variable optimizations (strength reduction, induction
6717 variable merging and induction variable elimination) on trees.
6719 @item -ftree-parallelize-loops=n
6720 @opindex ftree-parallelize-loops
6721 Parallelize loops, i.e., split their iteration space to run in n threads.
6722 This is only possible for loops whose iterations are independent
6723 and can be arbitrarily reordered. The optimization is only
6724 profitable on multiprocessor machines, for loops that are CPU-intensive,
6725 rather than constrained e.g.@: by memory bandwidth. This option
6726 implies @option{-pthread}, and thus is only supported on targets
6727 that have support for @option{-pthread}.
6731 Perform function-local points-to analysis on trees. This flag is
6732 enabled by default at @option{-O} and higher.
6736 Perform scalar replacement of aggregates. This pass replaces structure
6737 references with scalars to prevent committing structures to memory too
6738 early. This flag is enabled by default at @option{-O} and higher.
6740 @item -ftree-copyrename
6741 @opindex ftree-copyrename
6742 Perform copy renaming on trees. This pass attempts to rename compiler
6743 temporaries to other variables at copy locations, usually resulting in
6744 variable names which more closely resemble the original variables. This flag
6745 is enabled by default at @option{-O} and higher.
6749 Perform temporary expression replacement during the SSA->normal phase. Single
6750 use/single def temporaries are replaced at their use location with their
6751 defining expression. This results in non-GIMPLE code, but gives the expanders
6752 much more complex trees to work on resulting in better RTL generation. This is
6753 enabled by default at @option{-O} and higher.
6755 @item -ftree-vectorize
6756 @opindex ftree-vectorize
6757 Perform loop vectorization on trees. This flag is enabled by default at
6760 @item -ftree-vect-loop-version
6761 @opindex ftree-vect-loop-version
6762 Perform loop versioning when doing loop vectorization on trees. When a loop
6763 appears to be vectorizable except that data alignment or data dependence cannot
6764 be determined at compile time then vectorized and non-vectorized versions of
6765 the loop are generated along with runtime checks for alignment or dependence
6766 to control which version is executed. This option is enabled by default
6767 except at level @option{-Os} where it is disabled.
6769 @item -fvect-cost-model
6770 @opindex fvect-cost-model
6771 Enable cost model for vectorization.
6775 Perform Value Range Propagation on trees. This is similar to the
6776 constant propagation pass, but instead of values, ranges of values are
6777 propagated. This allows the optimizers to remove unnecessary range
6778 checks like array bound checks and null pointer checks. This is
6779 enabled by default at @option{-O2} and higher. Null pointer check
6780 elimination is only done if @option{-fdelete-null-pointer-checks} is
6785 Perform tail duplication to enlarge superblock size. This transformation
6786 simplifies the control flow of the function allowing other optimizations to do
6789 @item -funroll-loops
6790 @opindex funroll-loops
6791 Unroll loops whose number of iterations can be determined at compile
6792 time or upon entry to the loop. @option{-funroll-loops} implies
6793 @option{-frerun-cse-after-loop}. This option makes code larger,
6794 and may or may not make it run faster.
6796 @item -funroll-all-loops
6797 @opindex funroll-all-loops
6798 Unroll all loops, even if their number of iterations is uncertain when
6799 the loop is entered. This usually makes programs run more slowly.
6800 @option{-funroll-all-loops} implies the same options as
6801 @option{-funroll-loops},
6803 @item -fsplit-ivs-in-unroller
6804 @opindex fsplit-ivs-in-unroller
6805 Enables expressing of values of induction variables in later iterations
6806 of the unrolled loop using the value in the first iteration. This breaks
6807 long dependency chains, thus improving efficiency of the scheduling passes.
6809 Combination of @option{-fweb} and CSE is often sufficient to obtain the
6810 same effect. However in cases the loop body is more complicated than
6811 a single basic block, this is not reliable. It also does not work at all
6812 on some of the architectures due to restrictions in the CSE pass.
6814 This optimization is enabled by default.
6816 @item -fvariable-expansion-in-unroller
6817 @opindex fvariable-expansion-in-unroller
6818 With this option, the compiler will create multiple copies of some
6819 local variables when unrolling a loop which can result in superior code.
6821 @item -fpredictive-commoning
6822 @opindex fpredictive-commoning
6823 Perform predictive commoning optimization, i.e., reusing computations
6824 (especially memory loads and stores) performed in previous
6825 iterations of loops.
6827 This option is enabled at level @option{-O3}.
6829 @item -fprefetch-loop-arrays
6830 @opindex fprefetch-loop-arrays
6831 If supported by the target machine, generate instructions to prefetch
6832 memory to improve the performance of loops that access large arrays.
6834 This option may generate better or worse code; results are highly
6835 dependent on the structure of loops within the source code.
6837 Disabled at level @option{-Os}.
6840 @itemx -fno-peephole2
6841 @opindex fno-peephole
6842 @opindex fno-peephole2
6843 Disable any machine-specific peephole optimizations. The difference
6844 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
6845 are implemented in the compiler; some targets use one, some use the
6846 other, a few use both.
6848 @option{-fpeephole} is enabled by default.
6849 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6851 @item -fno-guess-branch-probability
6852 @opindex fno-guess-branch-probability
6853 Do not guess branch probabilities using heuristics.
6855 GCC will use heuristics to guess branch probabilities if they are
6856 not provided by profiling feedback (@option{-fprofile-arcs}). These
6857 heuristics are based on the control flow graph. If some branch probabilities
6858 are specified by @samp{__builtin_expect}, then the heuristics will be
6859 used to guess branch probabilities for the rest of the control flow graph,
6860 taking the @samp{__builtin_expect} info into account. The interactions
6861 between the heuristics and @samp{__builtin_expect} can be complex, and in
6862 some cases, it may be useful to disable the heuristics so that the effects
6863 of @samp{__builtin_expect} are easier to understand.
6865 The default is @option{-fguess-branch-probability} at levels
6866 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6868 @item -freorder-blocks
6869 @opindex freorder-blocks
6870 Reorder basic blocks in the compiled function in order to reduce number of
6871 taken branches and improve code locality.
6873 Enabled at levels @option{-O2}, @option{-O3}.
6875 @item -freorder-blocks-and-partition
6876 @opindex freorder-blocks-and-partition
6877 In addition to reordering basic blocks in the compiled function, in order
6878 to reduce number of taken branches, partitions hot and cold basic blocks
6879 into separate sections of the assembly and .o files, to improve
6880 paging and cache locality performance.
6882 This optimization is automatically turned off in the presence of
6883 exception handling, for linkonce sections, for functions with a user-defined
6884 section attribute and on any architecture that does not support named
6887 @item -freorder-functions
6888 @opindex freorder-functions
6889 Reorder functions in the object file in order to
6890 improve code locality. This is implemented by using special
6891 subsections @code{.text.hot} for most frequently executed functions and
6892 @code{.text.unlikely} for unlikely executed functions. Reordering is done by
6893 the linker so object file format must support named sections and linker must
6894 place them in a reasonable way.
6896 Also profile feedback must be available in to make this option effective. See
6897 @option{-fprofile-arcs} for details.
6899 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6901 @item -fstrict-aliasing
6902 @opindex fstrict-aliasing
6903 Allow the compiler to assume the strictest aliasing rules applicable to
6904 the language being compiled. For C (and C++), this activates
6905 optimizations based on the type of expressions. In particular, an
6906 object of one type is assumed never to reside at the same address as an
6907 object of a different type, unless the types are almost the same. For
6908 example, an @code{unsigned int} can alias an @code{int}, but not a
6909 @code{void*} or a @code{double}. A character type may alias any other
6912 @anchor{Type-punning}Pay special attention to code like this:
6925 The practice of reading from a different union member than the one most
6926 recently written to (called ``type-punning'') is common. Even with
6927 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
6928 is accessed through the union type. So, the code above will work as
6929 expected. @xref{Structures unions enumerations and bit-fields
6930 implementation}. However, this code might not:
6941 Similarly, access by taking the address, casting the resulting pointer
6942 and dereferencing the result has undefined behavior, even if the cast
6943 uses a union type, e.g.:
6947 return ((union a_union *) &d)->i;
6951 The @option{-fstrict-aliasing} option is enabled at levels
6952 @option{-O2}, @option{-O3}, @option{-Os}.
6954 @item -fstrict-overflow
6955 @opindex fstrict-overflow
6956 Allow the compiler to assume strict signed overflow rules, depending
6957 on the language being compiled. For C (and C++) this means that
6958 overflow when doing arithmetic with signed numbers is undefined, which
6959 means that the compiler may assume that it will not happen. This
6960 permits various optimizations. For example, the compiler will assume
6961 that an expression like @code{i + 10 > i} will always be true for
6962 signed @code{i}. This assumption is only valid if signed overflow is
6963 undefined, as the expression is false if @code{i + 10} overflows when
6964 using twos complement arithmetic. When this option is in effect any
6965 attempt to determine whether an operation on signed numbers will
6966 overflow must be written carefully to not actually involve overflow.
6968 This option also allows the compiler to assume strict pointer
6969 semantics: given a pointer to an object, if adding an offset to that
6970 pointer does not produce a pointer to the same object, the addition is
6971 undefined. This permits the compiler to conclude that @code{p + u >
6972 p} is always true for a pointer @code{p} and unsigned integer
6973 @code{u}. This assumption is only valid because pointer wraparound is
6974 undefined, as the expression is false if @code{p + u} overflows using
6975 twos complement arithmetic.
6977 See also the @option{-fwrapv} option. Using @option{-fwrapv} means
6978 that integer signed overflow is fully defined: it wraps. When
6979 @option{-fwrapv} is used, there is no difference between
6980 @option{-fstrict-overflow} and @option{-fno-strict-overflow} for
6981 integers. With @option{-fwrapv} certain types of overflow are
6982 permitted. For example, if the compiler gets an overflow when doing
6983 arithmetic on constants, the overflowed value can still be used with
6984 @option{-fwrapv}, but not otherwise.
6986 The @option{-fstrict-overflow} option is enabled at levels
6987 @option{-O2}, @option{-O3}, @option{-Os}.
6989 @item -falign-functions
6990 @itemx -falign-functions=@var{n}
6991 @opindex falign-functions
6992 Align the start of functions to the next power-of-two greater than
6993 @var{n}, skipping up to @var{n} bytes. For instance,
6994 @option{-falign-functions=32} aligns functions to the next 32-byte
6995 boundary, but @option{-falign-functions=24} would align to the next
6996 32-byte boundary only if this can be done by skipping 23 bytes or less.
6998 @option{-fno-align-functions} and @option{-falign-functions=1} are
6999 equivalent and mean that functions will not be aligned.
7001 Some assemblers only support this flag when @var{n} is a power of two;
7002 in that case, it is rounded up.
7004 If @var{n} is not specified or is zero, use a machine-dependent default.
7006 Enabled at levels @option{-O2}, @option{-O3}.
7008 @item -falign-labels
7009 @itemx -falign-labels=@var{n}
7010 @opindex falign-labels
7011 Align all branch targets to a power-of-two boundary, skipping up to
7012 @var{n} bytes like @option{-falign-functions}. This option can easily
7013 make code slower, because it must insert dummy operations for when the
7014 branch target is reached in the usual flow of the code.
7016 @option{-fno-align-labels} and @option{-falign-labels=1} are
7017 equivalent and mean that labels will not be aligned.
7019 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
7020 are greater than this value, then their values are used instead.
7022 If @var{n} is not specified or is zero, use a machine-dependent default
7023 which is very likely to be @samp{1}, meaning no alignment.
7025 Enabled at levels @option{-O2}, @option{-O3}.
7028 @itemx -falign-loops=@var{n}
7029 @opindex falign-loops
7030 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
7031 like @option{-falign-functions}. The hope is that the loop will be
7032 executed many times, which will make up for any execution of the dummy
7035 @option{-fno-align-loops} and @option{-falign-loops=1} are
7036 equivalent and mean that loops will not be aligned.
7038 If @var{n} is not specified or is zero, use a machine-dependent default.
7040 Enabled at levels @option{-O2}, @option{-O3}.
7043 @itemx -falign-jumps=@var{n}
7044 @opindex falign-jumps
7045 Align branch targets to a power-of-two boundary, for branch targets
7046 where the targets can only be reached by jumping, skipping up to @var{n}
7047 bytes like @option{-falign-functions}. In this case, no dummy operations
7050 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
7051 equivalent and mean that loops will not be aligned.
7053 If @var{n} is not specified or is zero, use a machine-dependent default.
7055 Enabled at levels @option{-O2}, @option{-O3}.
7057 @item -funit-at-a-time
7058 @opindex funit-at-a-time
7059 This option is left for compatibility reasons. @option{-funit-at-a-time}
7060 has no effect, while @option{-fno-unit-at-a-time} implies
7061 @option{-fno-toplevel-reorder} and @option{-fno-section-anchors}.
7065 @item -fno-toplevel-reorder
7066 @opindex fno-toplevel-reorder
7067 Do not reorder top-level functions, variables, and @code{asm}
7068 statements. Output them in the same order that they appear in the
7069 input file. When this option is used, unreferenced static variables
7070 will not be removed. This option is intended to support existing code
7071 which relies on a particular ordering. For new code, it is better to
7074 Enabled at level @option{-O0}. When disabled explicitly, it also imply
7075 @option{-fno-section-anchors} that is otherwise enabled at @option{-O0} on some
7080 Constructs webs as commonly used for register allocation purposes and assign
7081 each web individual pseudo register. This allows the register allocation pass
7082 to operate on pseudos directly, but also strengthens several other optimization
7083 passes, such as CSE, loop optimizer and trivial dead code remover. It can,
7084 however, make debugging impossible, since variables will no longer stay in a
7087 Enabled by default with @option{-funroll-loops}.
7089 @item -fwhole-program
7090 @opindex fwhole-program
7091 Assume that the current compilation unit represents the whole program being
7092 compiled. All public functions and variables with the exception of @code{main}
7093 and those merged by attribute @code{externally_visible} become static functions
7094 and in effect are optimized more aggressively by interprocedural optimizers.
7095 While this option is equivalent to proper use of the @code{static} keyword for
7096 programs consisting of a single file, in combination with option
7097 @option{--combine} this flag can be used to compile many smaller scale C
7098 programs since the functions and variables become local for the whole combined
7099 compilation unit, not for the single source file itself.
7101 This option implies @option{-fwhole-file} for Fortran programs.
7103 @item -fcprop-registers
7104 @opindex fcprop-registers
7105 After register allocation and post-register allocation instruction splitting,
7106 we perform a copy-propagation pass to try to reduce scheduling dependencies
7107 and occasionally eliminate the copy.
7109 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
7111 @item -fprofile-correction
7112 @opindex fprofile-correction
7113 Profiles collected using an instrumented binary for multi-threaded programs may
7114 be inconsistent due to missed counter updates. When this option is specified,
7115 GCC will use heuristics to correct or smooth out such inconsistencies. By
7116 default, GCC will emit an error message when an inconsistent profile is detected.
7118 @item -fprofile-dir=@var{path}
7119 @opindex fprofile-dir
7121 Set the directory to search the profile data files in to @var{path}.
7122 This option affects only the profile data generated by
7123 @option{-fprofile-generate}, @option{-ftest-coverage}, @option{-fprofile-arcs}
7124 and used by @option{-fprofile-use} and @option{-fbranch-probabilities}
7125 and its related options.
7126 By default, GCC will use the current directory as @var{path}
7127 thus the profile data file will appear in the same directory as the object file.
7129 @item -fprofile-generate
7130 @itemx -fprofile-generate=@var{path}
7131 @opindex fprofile-generate
7133 Enable options usually used for instrumenting application to produce
7134 profile useful for later recompilation with profile feedback based
7135 optimization. You must use @option{-fprofile-generate} both when
7136 compiling and when linking your program.
7138 The following options are enabled: @code{-fprofile-arcs}, @code{-fprofile-values}, @code{-fvpt}.
7140 If @var{path} is specified, GCC will look at the @var{path} to find
7141 the profile feedback data files. See @option{-fprofile-dir}.
7144 @itemx -fprofile-use=@var{path}
7145 @opindex fprofile-use
7146 Enable profile feedback directed optimizations, and optimizations
7147 generally profitable only with profile feedback available.
7149 The following options are enabled: @code{-fbranch-probabilities}, @code{-fvpt},
7150 @code{-funroll-loops}, @code{-fpeel-loops}, @code{-ftracer}
7152 By default, GCC emits an error message if the feedback profiles do not
7153 match the source code. This error can be turned into a warning by using
7154 @option{-Wcoverage-mismatch}. Note this may result in poorly optimized
7157 If @var{path} is specified, GCC will look at the @var{path} to find
7158 the profile feedback data files. See @option{-fprofile-dir}.
7161 The following options control compiler behavior regarding floating
7162 point arithmetic. These options trade off between speed and
7163 correctness. All must be specifically enabled.
7167 @opindex ffloat-store
7168 Do not store floating point variables in registers, and inhibit other
7169 options that might change whether a floating point value is taken from a
7172 @cindex floating point precision
7173 This option prevents undesirable excess precision on machines such as
7174 the 68000 where the floating registers (of the 68881) keep more
7175 precision than a @code{double} is supposed to have. Similarly for the
7176 x86 architecture. For most programs, the excess precision does only
7177 good, but a few programs rely on the precise definition of IEEE floating
7178 point. Use @option{-ffloat-store} for such programs, after modifying
7179 them to store all pertinent intermediate computations into variables.
7181 @item -fexcess-precision=@var{style}
7182 @opindex fexcess-precision
7183 This option allows further control over excess precision on machines
7184 where floating-point registers have more precision than the IEEE
7185 @code{float} and @code{double} types and the processor does not
7186 support operations rounding to those types. By default,
7187 @option{-fexcess-precision=fast} is in effect; this means that
7188 operations are carried out in the precision of the registers and that
7189 it is unpredictable when rounding to the types specified in the source
7190 code takes place. When compiling C, if
7191 @option{-fexcess-precision=standard} is specified then excess
7192 precision will follow the rules specified in ISO C99; in particular,
7193 both casts and assignments cause values to be rounded to their
7194 semantic types (whereas @option{-ffloat-store} only affects
7195 assignments). This option is enabled by default for C if a strict
7196 conformance option such as @option{-std=c99} is used.
7199 @option{-fexcess-precision=standard} is not implemented for languages
7200 other than C, and has no effect if
7201 @option{-funsafe-math-optimizations} or @option{-ffast-math} is
7202 specified. On the x86, it also has no effect if @option{-mfpmath=sse}
7203 or @option{-mfpmath=sse+387} is specified; in the former case, IEEE
7204 semantics apply without excess precision, and in the latter, rounding
7209 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations},
7210 @option{-ffinite-math-only}, @option{-fno-rounding-math},
7211 @option{-fno-signaling-nans} and @option{-fcx-limited-range}.
7213 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
7215 This option is not turned on by any @option{-O} option since
7216 it can result in incorrect output for programs which depend on
7217 an exact implementation of IEEE or ISO rules/specifications for
7218 math functions. It may, however, yield faster code for programs
7219 that do not require the guarantees of these specifications.
7221 @item -fno-math-errno
7222 @opindex fno-math-errno
7223 Do not set ERRNO after calling math functions that are executed
7224 with a single instruction, e.g., sqrt. A program that relies on
7225 IEEE exceptions for math error handling may want to use this flag
7226 for speed while maintaining IEEE arithmetic compatibility.
7228 This option is not turned on by any @option{-O} option since
7229 it can result in incorrect output for programs which depend on
7230 an exact implementation of IEEE or ISO rules/specifications for
7231 math functions. It may, however, yield faster code for programs
7232 that do not require the guarantees of these specifications.
7234 The default is @option{-fmath-errno}.
7236 On Darwin systems, the math library never sets @code{errno}. There is
7237 therefore no reason for the compiler to consider the possibility that
7238 it might, and @option{-fno-math-errno} is the default.
7240 @item -funsafe-math-optimizations
7241 @opindex funsafe-math-optimizations
7243 Allow optimizations for floating-point arithmetic that (a) assume
7244 that arguments and results are valid and (b) may violate IEEE or
7245 ANSI standards. When used at link-time, it may include libraries
7246 or startup files that change the default FPU control word or other
7247 similar optimizations.
7249 This option is not turned on by any @option{-O} option since
7250 it can result in incorrect output for programs which depend on
7251 an exact implementation of IEEE or ISO rules/specifications for
7252 math functions. It may, however, yield faster code for programs
7253 that do not require the guarantees of these specifications.
7254 Enables @option{-fno-signed-zeros}, @option{-fno-trapping-math},
7255 @option{-fassociative-math} and @option{-freciprocal-math}.
7257 The default is @option{-fno-unsafe-math-optimizations}.
7259 @item -fassociative-math
7260 @opindex fassociative-math
7262 Allow re-association of operands in series of floating-point operations.
7263 This violates the ISO C and C++ language standard by possibly changing
7264 computation result. NOTE: re-ordering may change the sign of zero as
7265 well as ignore NaNs and inhibit or create underflow or overflow (and
7266 thus cannot be used on a code which relies on rounding behavior like
7267 @code{(x + 2**52) - 2**52)}. May also reorder floating-point comparisons
7268 and thus may not be used when ordered comparisons are required.
7269 This option requires that both @option{-fno-signed-zeros} and
7270 @option{-fno-trapping-math} be in effect. Moreover, it doesn't make
7271 much sense with @option{-frounding-math}.
7273 The default is @option{-fno-associative-math}.
7275 @item -freciprocal-math
7276 @opindex freciprocal-math
7278 Allow the reciprocal of a value to be used instead of dividing by
7279 the value if this enables optimizations. For example @code{x / y}
7280 can be replaced with @code{x * (1/y)} which is useful if @code{(1/y)}
7281 is subject to common subexpression elimination. Note that this loses
7282 precision and increases the number of flops operating on the value.
7284 The default is @option{-fno-reciprocal-math}.
7286 @item -ffinite-math-only
7287 @opindex ffinite-math-only
7288 Allow optimizations for floating-point arithmetic that assume
7289 that arguments and results are not NaNs or +-Infs.
7291 This option is not turned on by any @option{-O} option since
7292 it can result in incorrect output for programs which depend on
7293 an exact implementation of IEEE or ISO rules/specifications for
7294 math functions. It may, however, yield faster code for programs
7295 that do not require the guarantees of these specifications.
7297 The default is @option{-fno-finite-math-only}.
7299 @item -fno-signed-zeros
7300 @opindex fno-signed-zeros
7301 Allow optimizations for floating point arithmetic that ignore the
7302 signedness of zero. IEEE arithmetic specifies the behavior of
7303 distinct +0.0 and @minus{}0.0 values, which then prohibits simplification
7304 of expressions such as x+0.0 or 0.0*x (even with @option{-ffinite-math-only}).
7305 This option implies that the sign of a zero result isn't significant.
7307 The default is @option{-fsigned-zeros}.
7309 @item -fno-trapping-math
7310 @opindex fno-trapping-math
7311 Compile code assuming that floating-point operations cannot generate
7312 user-visible traps. These traps include division by zero, overflow,
7313 underflow, inexact result and invalid operation. This option requires
7314 that @option{-fno-signaling-nans} be in effect. Setting this option may
7315 allow faster code if one relies on ``non-stop'' IEEE arithmetic, for example.
7317 This option should never be turned on by any @option{-O} option since
7318 it can result in incorrect output for programs which depend on
7319 an exact implementation of IEEE or ISO rules/specifications for
7322 The default is @option{-ftrapping-math}.
7324 @item -frounding-math
7325 @opindex frounding-math
7326 Disable transformations and optimizations that assume default floating
7327 point rounding behavior. This is round-to-zero for all floating point
7328 to integer conversions, and round-to-nearest for all other arithmetic
7329 truncations. This option should be specified for programs that change
7330 the FP rounding mode dynamically, or that may be executed with a
7331 non-default rounding mode. This option disables constant folding of
7332 floating point expressions at compile-time (which may be affected by
7333 rounding mode) and arithmetic transformations that are unsafe in the
7334 presence of sign-dependent rounding modes.
7336 The default is @option{-fno-rounding-math}.
7338 This option is experimental and does not currently guarantee to
7339 disable all GCC optimizations that are affected by rounding mode.
7340 Future versions of GCC may provide finer control of this setting
7341 using C99's @code{FENV_ACCESS} pragma. This command line option
7342 will be used to specify the default state for @code{FENV_ACCESS}.
7344 @item -fsignaling-nans
7345 @opindex fsignaling-nans
7346 Compile code assuming that IEEE signaling NaNs may generate user-visible
7347 traps during floating-point operations. Setting this option disables
7348 optimizations that may change the number of exceptions visible with
7349 signaling NaNs. This option implies @option{-ftrapping-math}.
7351 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
7354 The default is @option{-fno-signaling-nans}.
7356 This option is experimental and does not currently guarantee to
7357 disable all GCC optimizations that affect signaling NaN behavior.
7359 @item -fsingle-precision-constant
7360 @opindex fsingle-precision-constant
7361 Treat floating point constant as single precision constant instead of
7362 implicitly converting it to double precision constant.
7364 @item -fcx-limited-range
7365 @opindex fcx-limited-range
7366 When enabled, this option states that a range reduction step is not
7367 needed when performing complex division. Also, there is no checking
7368 whether the result of a complex multiplication or division is @code{NaN
7369 + I*NaN}, with an attempt to rescue the situation in that case. The
7370 default is @option{-fno-cx-limited-range}, but is enabled by
7371 @option{-ffast-math}.
7373 This option controls the default setting of the ISO C99
7374 @code{CX_LIMITED_RANGE} pragma. Nevertheless, the option applies to
7377 @item -fcx-fortran-rules
7378 @opindex fcx-fortran-rules
7379 Complex multiplication and division follow Fortran rules. Range
7380 reduction is done as part of complex division, but there is no checking
7381 whether the result of a complex multiplication or division is @code{NaN
7382 + I*NaN}, with an attempt to rescue the situation in that case.
7384 The default is @option{-fno-cx-fortran-rules}.
7388 The following options control optimizations that may improve
7389 performance, but are not enabled by any @option{-O} options. This
7390 section includes experimental options that may produce broken code.
7393 @item -fbranch-probabilities
7394 @opindex fbranch-probabilities
7395 After running a program compiled with @option{-fprofile-arcs}
7396 (@pxref{Debugging Options,, Options for Debugging Your Program or
7397 @command{gcc}}), you can compile it a second time using
7398 @option{-fbranch-probabilities}, to improve optimizations based on
7399 the number of times each branch was taken. When the program
7400 compiled with @option{-fprofile-arcs} exits it saves arc execution
7401 counts to a file called @file{@var{sourcename}.gcda} for each source
7402 file. The information in this data file is very dependent on the
7403 structure of the generated code, so you must use the same source code
7404 and the same optimization options for both compilations.
7406 With @option{-fbranch-probabilities}, GCC puts a
7407 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
7408 These can be used to improve optimization. Currently, they are only
7409 used in one place: in @file{reorg.c}, instead of guessing which path a
7410 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
7411 exactly determine which path is taken more often.
7413 @item -fprofile-values
7414 @opindex fprofile-values
7415 If combined with @option{-fprofile-arcs}, it adds code so that some
7416 data about values of expressions in the program is gathered.
7418 With @option{-fbranch-probabilities}, it reads back the data gathered
7419 from profiling values of expressions and adds @samp{REG_VALUE_PROFILE}
7420 notes to instructions for their later usage in optimizations.
7422 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
7426 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
7427 a code to gather information about values of expressions.
7429 With @option{-fbranch-probabilities}, it reads back the data gathered
7430 and actually performs the optimizations based on them.
7431 Currently the optimizations include specialization of division operation
7432 using the knowledge about the value of the denominator.
7434 @item -frename-registers
7435 @opindex frename-registers
7436 Attempt to avoid false dependencies in scheduled code by making use
7437 of registers left over after register allocation. This optimization
7438 will most benefit processors with lots of registers. Depending on the
7439 debug information format adopted by the target, however, it can
7440 make debugging impossible, since variables will no longer stay in
7441 a ``home register''.
7443 Enabled by default with @option{-funroll-loops}.
7447 Perform tail duplication to enlarge superblock size. This transformation
7448 simplifies the control flow of the function allowing other optimizations to do
7451 Enabled with @option{-fprofile-use}.
7453 @item -funroll-loops
7454 @opindex funroll-loops
7455 Unroll loops whose number of iterations can be determined at compile time or
7456 upon entry to the loop. @option{-funroll-loops} implies
7457 @option{-frerun-cse-after-loop}, @option{-fweb} and @option{-frename-registers}.
7458 It also turns on complete loop peeling (i.e.@: complete removal of loops with
7459 small constant number of iterations). This option makes code larger, and may
7460 or may not make it run faster.
7462 Enabled with @option{-fprofile-use}.
7464 @item -funroll-all-loops
7465 @opindex funroll-all-loops
7466 Unroll all loops, even if their number of iterations is uncertain when
7467 the loop is entered. This usually makes programs run more slowly.
7468 @option{-funroll-all-loops} implies the same options as
7469 @option{-funroll-loops}.
7472 @opindex fpeel-loops
7473 Peels the loops for that there is enough information that they do not
7474 roll much (from profile feedback). It also turns on complete loop peeling
7475 (i.e.@: complete removal of loops with small constant number of iterations).
7477 Enabled with @option{-fprofile-use}.
7479 @item -fmove-loop-invariants
7480 @opindex fmove-loop-invariants
7481 Enables the loop invariant motion pass in the RTL loop optimizer. Enabled
7482 at level @option{-O1}
7484 @item -funswitch-loops
7485 @opindex funswitch-loops
7486 Move branches with loop invariant conditions out of the loop, with duplicates
7487 of the loop on both branches (modified according to result of the condition).
7489 @item -ffunction-sections
7490 @itemx -fdata-sections
7491 @opindex ffunction-sections
7492 @opindex fdata-sections
7493 Place each function or data item into its own section in the output
7494 file if the target supports arbitrary sections. The name of the
7495 function or the name of the data item determines the section's name
7498 Use these options on systems where the linker can perform optimizations
7499 to improve locality of reference in the instruction space. Most systems
7500 using the ELF object format and SPARC processors running Solaris 2 have
7501 linkers with such optimizations. AIX may have these optimizations in
7504 Only use these options when there are significant benefits from doing
7505 so. When you specify these options, the assembler and linker will
7506 create larger object and executable files and will also be slower.
7507 You will not be able to use @code{gprof} on all systems if you
7508 specify this option and you may have problems with debugging if
7509 you specify both this option and @option{-g}.
7511 @item -fbranch-target-load-optimize
7512 @opindex fbranch-target-load-optimize
7513 Perform branch target register load optimization before prologue / epilogue
7515 The use of target registers can typically be exposed only during reload,
7516 thus hoisting loads out of loops and doing inter-block scheduling needs
7517 a separate optimization pass.
7519 @item -fbranch-target-load-optimize2
7520 @opindex fbranch-target-load-optimize2
7521 Perform branch target register load optimization after prologue / epilogue
7524 @item -fbtr-bb-exclusive
7525 @opindex fbtr-bb-exclusive
7526 When performing branch target register load optimization, don't reuse
7527 branch target registers in within any basic block.
7529 @item -fstack-protector
7530 @opindex fstack-protector
7531 Emit extra code to check for buffer overflows, such as stack smashing
7532 attacks. This is done by adding a guard variable to functions with
7533 vulnerable objects. This includes functions that call alloca, and
7534 functions with buffers larger than 8 bytes. The guards are initialized
7535 when a function is entered and then checked when the function exits.
7536 If a guard check fails, an error message is printed and the program exits.
7538 @item -fstack-protector-all
7539 @opindex fstack-protector-all
7540 Like @option{-fstack-protector} except that all functions are protected.
7542 @item -fsection-anchors
7543 @opindex fsection-anchors
7544 Try to reduce the number of symbolic address calculations by using
7545 shared ``anchor'' symbols to address nearby objects. This transformation
7546 can help to reduce the number of GOT entries and GOT accesses on some
7549 For example, the implementation of the following function @code{foo}:
7553 int foo (void) @{ return a + b + c; @}
7556 would usually calculate the addresses of all three variables, but if you
7557 compile it with @option{-fsection-anchors}, it will access the variables
7558 from a common anchor point instead. The effect is similar to the
7559 following pseudocode (which isn't valid C):
7564 register int *xr = &x;
7565 return xr[&a - &x] + xr[&b - &x] + xr[&c - &x];
7569 Not all targets support this option.
7571 @item --param @var{name}=@var{value}
7573 In some places, GCC uses various constants to control the amount of
7574 optimization that is done. For example, GCC will not inline functions
7575 that contain more that a certain number of instructions. You can
7576 control some of these constants on the command-line using the
7577 @option{--param} option.
7579 The names of specific parameters, and the meaning of the values, are
7580 tied to the internals of the compiler, and are subject to change
7581 without notice in future releases.
7583 In each case, the @var{value} is an integer. The allowable choices for
7584 @var{name} are given in the following table:
7587 @item struct-reorg-cold-struct-ratio
7588 The threshold ratio (as a percentage) between a structure frequency
7589 and the frequency of the hottest structure in the program. This parameter
7590 is used by struct-reorg optimization enabled by @option{-fipa-struct-reorg}.
7591 We say that if the ratio of a structure frequency, calculated by profiling,
7592 to the hottest structure frequency in the program is less than this
7593 parameter, then structure reorganization is not applied to this structure.
7596 @item predictable-branch-cost-outcome
7597 When branch is predicted to be taken with probability lower than this threshold
7598 (in percent), then it is considered well predictable. The default is 10.
7600 @item max-crossjump-edges
7601 The maximum number of incoming edges to consider for crossjumping.
7602 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
7603 the number of edges incoming to each block. Increasing values mean
7604 more aggressive optimization, making the compile time increase with
7605 probably small improvement in executable size.
7607 @item min-crossjump-insns
7608 The minimum number of instructions which must be matched at the end
7609 of two blocks before crossjumping will be performed on them. This
7610 value is ignored in the case where all instructions in the block being
7611 crossjumped from are matched. The default value is 5.
7613 @item max-grow-copy-bb-insns
7614 The maximum code size expansion factor when copying basic blocks
7615 instead of jumping. The expansion is relative to a jump instruction.
7616 The default value is 8.
7618 @item max-goto-duplication-insns
7619 The maximum number of instructions to duplicate to a block that jumps
7620 to a computed goto. To avoid @math{O(N^2)} behavior in a number of
7621 passes, GCC factors computed gotos early in the compilation process,
7622 and unfactors them as late as possible. Only computed jumps at the
7623 end of a basic blocks with no more than max-goto-duplication-insns are
7624 unfactored. The default value is 8.
7626 @item max-delay-slot-insn-search
7627 The maximum number of instructions to consider when looking for an
7628 instruction to fill a delay slot. If more than this arbitrary number of
7629 instructions is searched, the time savings from filling the delay slot
7630 will be minimal so stop searching. Increasing values mean more
7631 aggressive optimization, making the compile time increase with probably
7632 small improvement in executable run time.
7634 @item max-delay-slot-live-search
7635 When trying to fill delay slots, the maximum number of instructions to
7636 consider when searching for a block with valid live register
7637 information. Increasing this arbitrarily chosen value means more
7638 aggressive optimization, increasing the compile time. This parameter
7639 should be removed when the delay slot code is rewritten to maintain the
7642 @item max-gcse-memory
7643 The approximate maximum amount of memory that will be allocated in
7644 order to perform the global common subexpression elimination
7645 optimization. If more memory than specified is required, the
7646 optimization will not be done.
7648 @item max-pending-list-length
7649 The maximum number of pending dependencies scheduling will allow
7650 before flushing the current state and starting over. Large functions
7651 with few branches or calls can create excessively large lists which
7652 needlessly consume memory and resources.
7654 @item max-inline-insns-single
7655 Several parameters control the tree inliner used in gcc.
7656 This number sets the maximum number of instructions (counted in GCC's
7657 internal representation) in a single function that the tree inliner
7658 will consider for inlining. This only affects functions declared
7659 inline and methods implemented in a class declaration (C++).
7660 The default value is 300.
7662 @item max-inline-insns-auto
7663 When you use @option{-finline-functions} (included in @option{-O3}),
7664 a lot of functions that would otherwise not be considered for inlining
7665 by the compiler will be investigated. To those functions, a different
7666 (more restrictive) limit compared to functions declared inline can
7668 The default value is 50.
7670 @item large-function-insns
7671 The limit specifying really large functions. For functions larger than this
7672 limit after inlining, inlining is constrained by
7673 @option{--param large-function-growth}. This parameter is useful primarily
7674 to avoid extreme compilation time caused by non-linear algorithms used by the
7676 The default value is 2700.
7678 @item large-function-growth
7679 Specifies maximal growth of large function caused by inlining in percents.
7680 The default value is 100 which limits large function growth to 2.0 times
7683 @item large-unit-insns
7684 The limit specifying large translation unit. Growth caused by inlining of
7685 units larger than this limit is limited by @option{--param inline-unit-growth}.
7686 For small units this might be too tight (consider unit consisting of function A
7687 that is inline and B that just calls A three time. If B is small relative to
7688 A, the growth of unit is 300\% and yet such inlining is very sane. For very
7689 large units consisting of small inlineable functions however the overall unit
7690 growth limit is needed to avoid exponential explosion of code size. Thus for
7691 smaller units, the size is increased to @option{--param large-unit-insns}
7692 before applying @option{--param inline-unit-growth}. The default is 10000
7694 @item inline-unit-growth
7695 Specifies maximal overall growth of the compilation unit caused by inlining.
7696 The default value is 30 which limits unit growth to 1.3 times the original
7699 @item ipcp-unit-growth
7700 Specifies maximal overall growth of the compilation unit caused by
7701 interprocedural constant propagation. The default value is 10 which limits
7702 unit growth to 1.1 times the original size.
7704 @item large-stack-frame
7705 The limit specifying large stack frames. While inlining the algorithm is trying
7706 to not grow past this limit too much. Default value is 256 bytes.
7708 @item large-stack-frame-growth
7709 Specifies maximal growth of large stack frames caused by inlining in percents.
7710 The default value is 1000 which limits large stack frame growth to 11 times
7713 @item max-inline-insns-recursive
7714 @itemx max-inline-insns-recursive-auto
7715 Specifies maximum number of instructions out-of-line copy of self recursive inline
7716 function can grow into by performing recursive inlining.
7718 For functions declared inline @option{--param max-inline-insns-recursive} is
7719 taken into account. For function not declared inline, recursive inlining
7720 happens only when @option{-finline-functions} (included in @option{-O3}) is
7721 enabled and @option{--param max-inline-insns-recursive-auto} is used. The
7722 default value is 450.
7724 @item max-inline-recursive-depth
7725 @itemx max-inline-recursive-depth-auto
7726 Specifies maximum recursion depth used by the recursive inlining.
7728 For functions declared inline @option{--param max-inline-recursive-depth} is
7729 taken into account. For function not declared inline, recursive inlining
7730 happens only when @option{-finline-functions} (included in @option{-O3}) is
7731 enabled and @option{--param max-inline-recursive-depth-auto} is used. The
7734 @item min-inline-recursive-probability
7735 Recursive inlining is profitable only for function having deep recursion
7736 in average and can hurt for function having little recursion depth by
7737 increasing the prologue size or complexity of function body to other
7740 When profile feedback is available (see @option{-fprofile-generate}) the actual
7741 recursion depth can be guessed from probability that function will recurse via
7742 given call expression. This parameter limits inlining only to call expression
7743 whose probability exceeds given threshold (in percents). The default value is
7746 @item early-inlining-insns
7747 Specify growth that early inliner can make. In effect it increases amount of
7748 inlining for code having large abstraction penalty. The default value is 8.
7750 @item max-early-inliner-iterations
7751 @itemx max-early-inliner-iterations
7752 Limit of iterations of early inliner. This basically bounds number of nested
7753 indirect calls early inliner can resolve. Deeper chains are still handled by
7756 @item min-vect-loop-bound
7757 The minimum number of iterations under which a loop will not get vectorized
7758 when @option{-ftree-vectorize} is used. The number of iterations after
7759 vectorization needs to be greater than the value specified by this option
7760 to allow vectorization. The default value is 0.
7762 @item max-unrolled-insns
7763 The maximum number of instructions that a loop should have if that loop
7764 is unrolled, and if the loop is unrolled, it determines how many times
7765 the loop code is unrolled.
7767 @item max-average-unrolled-insns
7768 The maximum number of instructions biased by probabilities of their execution
7769 that a loop should have if that loop is unrolled, and if the loop is unrolled,
7770 it determines how many times the loop code is unrolled.
7772 @item max-unroll-times
7773 The maximum number of unrollings of a single loop.
7775 @item max-peeled-insns
7776 The maximum number of instructions that a loop should have if that loop
7777 is peeled, and if the loop is peeled, it determines how many times
7778 the loop code is peeled.
7780 @item max-peel-times
7781 The maximum number of peelings of a single loop.
7783 @item max-completely-peeled-insns
7784 The maximum number of insns of a completely peeled loop.
7786 @item max-completely-peel-times
7787 The maximum number of iterations of a loop to be suitable for complete peeling.
7789 @item max-unswitch-insns
7790 The maximum number of insns of an unswitched loop.
7792 @item max-unswitch-level
7793 The maximum number of branches unswitched in a single loop.
7796 The minimum cost of an expensive expression in the loop invariant motion.
7798 @item iv-consider-all-candidates-bound
7799 Bound on number of candidates for induction variables below that
7800 all candidates are considered for each use in induction variable
7801 optimizations. Only the most relevant candidates are considered
7802 if there are more candidates, to avoid quadratic time complexity.
7804 @item iv-max-considered-uses
7805 The induction variable optimizations give up on loops that contain more
7806 induction variable uses.
7808 @item iv-always-prune-cand-set-bound
7809 If number of candidates in the set is smaller than this value,
7810 we always try to remove unnecessary ivs from the set during its
7811 optimization when a new iv is added to the set.
7813 @item scev-max-expr-size
7814 Bound on size of expressions used in the scalar evolutions analyzer.
7815 Large expressions slow the analyzer.
7817 @item omega-max-vars
7818 The maximum number of variables in an Omega constraint system.
7819 The default value is 128.
7821 @item omega-max-geqs
7822 The maximum number of inequalities in an Omega constraint system.
7823 The default value is 256.
7826 The maximum number of equalities in an Omega constraint system.
7827 The default value is 128.
7829 @item omega-max-wild-cards
7830 The maximum number of wildcard variables that the Omega solver will
7831 be able to insert. The default value is 18.
7833 @item omega-hash-table-size
7834 The size of the hash table in the Omega solver. The default value is
7837 @item omega-max-keys
7838 The maximal number of keys used by the Omega solver. The default
7841 @item omega-eliminate-redundant-constraints
7842 When set to 1, use expensive methods to eliminate all redundant
7843 constraints. The default value is 0.
7845 @item vect-max-version-for-alignment-checks
7846 The maximum number of runtime checks that can be performed when
7847 doing loop versioning for alignment in the vectorizer. See option
7848 ftree-vect-loop-version for more information.
7850 @item vect-max-version-for-alias-checks
7851 The maximum number of runtime checks that can be performed when
7852 doing loop versioning for alias in the vectorizer. See option
7853 ftree-vect-loop-version for more information.
7855 @item max-iterations-to-track
7857 The maximum number of iterations of a loop the brute force algorithm
7858 for analysis of # of iterations of the loop tries to evaluate.
7860 @item hot-bb-count-fraction
7861 Select fraction of the maximal count of repetitions of basic block in program
7862 given basic block needs to have to be considered hot.
7864 @item hot-bb-frequency-fraction
7865 Select fraction of the maximal frequency of executions of basic block in
7866 function given basic block needs to have to be considered hot
7868 @item max-predicted-iterations
7869 The maximum number of loop iterations we predict statically. This is useful
7870 in cases where function contain single loop with known bound and other loop
7871 with unknown. We predict the known number of iterations correctly, while
7872 the unknown number of iterations average to roughly 10. This means that the
7873 loop without bounds would appear artificially cold relative to the other one.
7875 @item align-threshold
7877 Select fraction of the maximal frequency of executions of basic block in
7878 function given basic block will get aligned.
7880 @item align-loop-iterations
7882 A loop expected to iterate at lest the selected number of iterations will get
7885 @item tracer-dynamic-coverage
7886 @itemx tracer-dynamic-coverage-feedback
7888 This value is used to limit superblock formation once the given percentage of
7889 executed instructions is covered. This limits unnecessary code size
7892 The @option{tracer-dynamic-coverage-feedback} is used only when profile
7893 feedback is available. The real profiles (as opposed to statically estimated
7894 ones) are much less balanced allowing the threshold to be larger value.
7896 @item tracer-max-code-growth
7897 Stop tail duplication once code growth has reached given percentage. This is
7898 rather hokey argument, as most of the duplicates will be eliminated later in
7899 cross jumping, so it may be set to much higher values than is the desired code
7902 @item tracer-min-branch-ratio
7904 Stop reverse growth when the reverse probability of best edge is less than this
7905 threshold (in percent).
7907 @item tracer-min-branch-ratio
7908 @itemx tracer-min-branch-ratio-feedback
7910 Stop forward growth if the best edge do have probability lower than this
7913 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
7914 compilation for profile feedback and one for compilation without. The value
7915 for compilation with profile feedback needs to be more conservative (higher) in
7916 order to make tracer effective.
7918 @item max-cse-path-length
7920 Maximum number of basic blocks on path that cse considers. The default is 10.
7923 The maximum instructions CSE process before flushing. The default is 1000.
7925 @item ggc-min-expand
7927 GCC uses a garbage collector to manage its own memory allocation. This
7928 parameter specifies the minimum percentage by which the garbage
7929 collector's heap should be allowed to expand between collections.
7930 Tuning this may improve compilation speed; it has no effect on code
7933 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
7934 RAM >= 1GB@. If @code{getrlimit} is available, the notion of "RAM" is
7935 the smallest of actual RAM and @code{RLIMIT_DATA} or @code{RLIMIT_AS}. If
7936 GCC is not able to calculate RAM on a particular platform, the lower
7937 bound of 30% is used. Setting this parameter and
7938 @option{ggc-min-heapsize} to zero causes a full collection to occur at
7939 every opportunity. This is extremely slow, but can be useful for
7942 @item ggc-min-heapsize
7944 Minimum size of the garbage collector's heap before it begins bothering
7945 to collect garbage. The first collection occurs after the heap expands
7946 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
7947 tuning this may improve compilation speed, and has no effect on code
7950 The default is the smaller of RAM/8, RLIMIT_RSS, or a limit which
7951 tries to ensure that RLIMIT_DATA or RLIMIT_AS are not exceeded, but
7952 with a lower bound of 4096 (four megabytes) and an upper bound of
7953 131072 (128 megabytes). If GCC is not able to calculate RAM on a
7954 particular platform, the lower bound is used. Setting this parameter
7955 very large effectively disables garbage collection. Setting this
7956 parameter and @option{ggc-min-expand} to zero causes a full collection
7957 to occur at every opportunity.
7959 @item max-reload-search-insns
7960 The maximum number of instruction reload should look backward for equivalent
7961 register. Increasing values mean more aggressive optimization, making the
7962 compile time increase with probably slightly better performance. The default
7965 @item max-cselib-memory-locations
7966 The maximum number of memory locations cselib should take into account.
7967 Increasing values mean more aggressive optimization, making the compile time
7968 increase with probably slightly better performance. The default value is 500.
7970 @item reorder-blocks-duplicate
7971 @itemx reorder-blocks-duplicate-feedback
7973 Used by basic block reordering pass to decide whether to use unconditional
7974 branch or duplicate the code on its destination. Code is duplicated when its
7975 estimated size is smaller than this value multiplied by the estimated size of
7976 unconditional jump in the hot spots of the program.
7978 The @option{reorder-block-duplicate-feedback} is used only when profile
7979 feedback is available and may be set to higher values than
7980 @option{reorder-block-duplicate} since information about the hot spots is more
7983 @item max-sched-ready-insns
7984 The maximum number of instructions ready to be issued the scheduler should
7985 consider at any given time during the first scheduling pass. Increasing
7986 values mean more thorough searches, making the compilation time increase
7987 with probably little benefit. The default value is 100.
7989 @item max-sched-region-blocks
7990 The maximum number of blocks in a region to be considered for
7991 interblock scheduling. The default value is 10.
7993 @item max-pipeline-region-blocks
7994 The maximum number of blocks in a region to be considered for
7995 pipelining in the selective scheduler. The default value is 15.
7997 @item max-sched-region-insns
7998 The maximum number of insns in a region to be considered for
7999 interblock scheduling. The default value is 100.
8001 @item max-pipeline-region-insns
8002 The maximum number of insns in a region to be considered for
8003 pipelining in the selective scheduler. The default value is 200.
8006 The minimum probability (in percents) of reaching a source block
8007 for interblock speculative scheduling. The default value is 40.
8009 @item max-sched-extend-regions-iters
8010 The maximum number of iterations through CFG to extend regions.
8011 0 - disable region extension,
8012 N - do at most N iterations.
8013 The default value is 0.
8015 @item max-sched-insn-conflict-delay
8016 The maximum conflict delay for an insn to be considered for speculative motion.
8017 The default value is 3.
8019 @item sched-spec-prob-cutoff
8020 The minimal probability of speculation success (in percents), so that
8021 speculative insn will be scheduled.
8022 The default value is 40.
8024 @item sched-mem-true-dep-cost
8025 Minimal distance (in CPU cycles) between store and load targeting same
8026 memory locations. The default value is 1.
8028 @item selsched-max-lookahead
8029 The maximum size of the lookahead window of selective scheduling. It is a
8030 depth of search for available instructions.
8031 The default value is 50.
8033 @item selsched-max-sched-times
8034 The maximum number of times that an instruction will be scheduled during
8035 selective scheduling. This is the limit on the number of iterations
8036 through which the instruction may be pipelined. The default value is 2.
8038 @item selsched-max-insns-to-rename
8039 The maximum number of best instructions in the ready list that are considered
8040 for renaming in the selective scheduler. The default value is 2.
8042 @item max-last-value-rtl
8043 The maximum size measured as number of RTLs that can be recorded in an expression
8044 in combiner for a pseudo register as last known value of that register. The default
8047 @item integer-share-limit
8048 Small integer constants can use a shared data structure, reducing the
8049 compiler's memory usage and increasing its speed. This sets the maximum
8050 value of a shared integer constant. The default value is 256.
8052 @item min-virtual-mappings
8053 Specifies the minimum number of virtual mappings in the incremental
8054 SSA updater that should be registered to trigger the virtual mappings
8055 heuristic defined by virtual-mappings-ratio. The default value is
8058 @item virtual-mappings-ratio
8059 If the number of virtual mappings is virtual-mappings-ratio bigger
8060 than the number of virtual symbols to be updated, then the incremental
8061 SSA updater switches to a full update for those symbols. The default
8064 @item ssp-buffer-size
8065 The minimum size of buffers (i.e.@: arrays) that will receive stack smashing
8066 protection when @option{-fstack-protection} is used.
8068 @item max-jump-thread-duplication-stmts
8069 Maximum number of statements allowed in a block that needs to be
8070 duplicated when threading jumps.
8072 @item max-fields-for-field-sensitive
8073 Maximum number of fields in a structure we will treat in
8074 a field sensitive manner during pointer analysis. The default is zero
8075 for -O0, and -O1 and 100 for -Os, -O2, and -O3.
8077 @item prefetch-latency
8078 Estimate on average number of instructions that are executed before
8079 prefetch finishes. The distance we prefetch ahead is proportional
8080 to this constant. Increasing this number may also lead to less
8081 streams being prefetched (see @option{simultaneous-prefetches}).
8083 @item simultaneous-prefetches
8084 Maximum number of prefetches that can run at the same time.
8086 @item l1-cache-line-size
8087 The size of cache line in L1 cache, in bytes.
8090 The size of L1 cache, in kilobytes.
8093 The size of L2 cache, in kilobytes.
8095 @item min-insn-to-prefetch-ratio
8096 The minimum ratio between the number of instructions and the
8097 number of prefetches to enable prefetching in a loop with an
8100 @item prefetch-min-insn-to-mem-ratio
8101 The minimum ratio between the number of instructions and the
8102 number of memory references to enable prefetching in a loop.
8104 @item use-canonical-types
8105 Whether the compiler should use the ``canonical'' type system. By
8106 default, this should always be 1, which uses a more efficient internal
8107 mechanism for comparing types in C++ and Objective-C++. However, if
8108 bugs in the canonical type system are causing compilation failures,
8109 set this value to 0 to disable canonical types.
8111 @item switch-conversion-max-branch-ratio
8112 Switch initialization conversion will refuse to create arrays that are
8113 bigger than @option{switch-conversion-max-branch-ratio} times the number of
8114 branches in the switch.
8116 @item max-partial-antic-length
8117 Maximum length of the partial antic set computed during the tree
8118 partial redundancy elimination optimization (@option{-ftree-pre}) when
8119 optimizing at @option{-O3} and above. For some sorts of source code
8120 the enhanced partial redundancy elimination optimization can run away,
8121 consuming all of the memory available on the host machine. This
8122 parameter sets a limit on the length of the sets that are computed,
8123 which prevents the runaway behavior. Setting a value of 0 for
8124 this parameter will allow an unlimited set length.
8126 @item sccvn-max-scc-size
8127 Maximum size of a strongly connected component (SCC) during SCCVN
8128 processing. If this limit is hit, SCCVN processing for the whole
8129 function will not be done and optimizations depending on it will
8130 be disabled. The default maximum SCC size is 10000.
8132 @item ira-max-loops-num
8133 IRA uses a regional register allocation by default. If a function
8134 contains loops more than number given by the parameter, only at most
8135 given number of the most frequently executed loops will form regions
8136 for the regional register allocation. The default value of the
8139 @item ira-max-conflict-table-size
8140 Although IRA uses a sophisticated algorithm of compression conflict
8141 table, the table can be still big for huge functions. If the conflict
8142 table for a function could be more than size in MB given by the
8143 parameter, the conflict table is not built and faster, simpler, and
8144 lower quality register allocation algorithm will be used. The
8145 algorithm do not use pseudo-register conflicts. The default value of
8146 the parameter is 2000.
8148 @item loop-invariant-max-bbs-in-loop
8149 Loop invariant motion can be very expensive, both in compile time and
8150 in amount of needed compile time memory, with very large loops. Loops
8151 with more basic blocks than this parameter won't have loop invariant
8152 motion optimization performed on them. The default value of the
8153 parameter is 1000 for -O1 and 10000 for -O2 and above.
8155 @item min-nondebug-insn-uid
8156 Use uids starting at this parameter for nondebug insns. The range below
8157 the parameter is reserved exclusively for debug insns created by
8158 @option{-fvar-tracking-assignments}, but debug insns may get
8159 (non-overlapping) uids above it if the reserved range is exhausted.
8161 @item ipa-sra-ptr-growth-factor
8162 IPA-SRA will replace a pointer to an aggregate with one or more new
8163 parameters only when their cumulative size is less or equal to
8164 @option{ipa-sra-ptr-growth-factor} times the size of the original
8170 @node Preprocessor Options
8171 @section Options Controlling the Preprocessor
8172 @cindex preprocessor options
8173 @cindex options, preprocessor
8175 These options control the C preprocessor, which is run on each C source
8176 file before actual compilation.
8178 If you use the @option{-E} option, nothing is done except preprocessing.
8179 Some of these options make sense only together with @option{-E} because
8180 they cause the preprocessor output to be unsuitable for actual
8184 @item -Wp,@var{option}
8186 You can use @option{-Wp,@var{option}} to bypass the compiler driver
8187 and pass @var{option} directly through to the preprocessor. If
8188 @var{option} contains commas, it is split into multiple options at the
8189 commas. However, many options are modified, translated or interpreted
8190 by the compiler driver before being passed to the preprocessor, and
8191 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
8192 interface is undocumented and subject to change, so whenever possible
8193 you should avoid using @option{-Wp} and let the driver handle the
8196 @item -Xpreprocessor @var{option}
8197 @opindex Xpreprocessor
8198 Pass @var{option} as an option to the preprocessor. You can use this to
8199 supply system-specific preprocessor options which GCC does not know how to
8202 If you want to pass an option that takes an argument, you must use
8203 @option{-Xpreprocessor} twice, once for the option and once for the argument.
8206 @include cppopts.texi
8208 @node Assembler Options
8209 @section Passing Options to the Assembler
8211 @c prevent bad page break with this line
8212 You can pass options to the assembler.
8215 @item -Wa,@var{option}
8217 Pass @var{option} as an option to the assembler. If @var{option}
8218 contains commas, it is split into multiple options at the commas.
8220 @item -Xassembler @var{option}
8222 Pass @var{option} as an option to the assembler. You can use this to
8223 supply system-specific assembler options which GCC does not know how to
8226 If you want to pass an option that takes an argument, you must use
8227 @option{-Xassembler} twice, once for the option and once for the argument.
8232 @section Options for Linking
8233 @cindex link options
8234 @cindex options, linking
8236 These options come into play when the compiler links object files into
8237 an executable output file. They are meaningless if the compiler is
8238 not doing a link step.
8242 @item @var{object-file-name}
8243 A file name that does not end in a special recognized suffix is
8244 considered to name an object file or library. (Object files are
8245 distinguished from libraries by the linker according to the file
8246 contents.) If linking is done, these object files are used as input
8255 If any of these options is used, then the linker is not run, and
8256 object file names should not be used as arguments. @xref{Overall
8260 @item -l@var{library}
8261 @itemx -l @var{library}
8263 Search the library named @var{library} when linking. (The second
8264 alternative with the library as a separate argument is only for
8265 POSIX compliance and is not recommended.)
8267 It makes a difference where in the command you write this option; the
8268 linker searches and processes libraries and object files in the order they
8269 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
8270 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
8271 to functions in @samp{z}, those functions may not be loaded.
8273 The linker searches a standard list of directories for the library,
8274 which is actually a file named @file{lib@var{library}.a}. The linker
8275 then uses this file as if it had been specified precisely by name.
8277 The directories searched include several standard system directories
8278 plus any that you specify with @option{-L}.
8280 Normally the files found this way are library files---archive files
8281 whose members are object files. The linker handles an archive file by
8282 scanning through it for members which define symbols that have so far
8283 been referenced but not defined. But if the file that is found is an
8284 ordinary object file, it is linked in the usual fashion. The only
8285 difference between using an @option{-l} option and specifying a file name
8286 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
8287 and searches several directories.
8291 You need this special case of the @option{-l} option in order to
8292 link an Objective-C or Objective-C++ program.
8295 @opindex nostartfiles
8296 Do not use the standard system startup files when linking.
8297 The standard system libraries are used normally, unless @option{-nostdlib}
8298 or @option{-nodefaultlibs} is used.
8300 @item -nodefaultlibs
8301 @opindex nodefaultlibs
8302 Do not use the standard system libraries when linking.
8303 Only the libraries you specify will be passed to the linker, options
8304 specifying linkage of the system libraries, such as @code{-static-libgcc}
8305 or @code{-shared-libgcc}, will be ignored.
8306 The standard startup files are used normally, unless @option{-nostartfiles}
8307 is used. The compiler may generate calls to @code{memcmp},
8308 @code{memset}, @code{memcpy} and @code{memmove}.
8309 These entries are usually resolved by entries in
8310 libc. These entry points should be supplied through some other
8311 mechanism when this option is specified.
8315 Do not use the standard system startup files or libraries when linking.
8316 No startup files and only the libraries you specify will be passed to
8317 the linker, options specifying linkage of the system libraries, such as
8318 @code{-static-libgcc} or @code{-shared-libgcc}, will be ignored.
8319 The compiler may generate calls to @code{memcmp}, @code{memset},
8320 @code{memcpy} and @code{memmove}.
8321 These entries are usually resolved by entries in
8322 libc. These entry points should be supplied through some other
8323 mechanism when this option is specified.
8325 @cindex @option{-lgcc}, use with @option{-nostdlib}
8326 @cindex @option{-nostdlib} and unresolved references
8327 @cindex unresolved references and @option{-nostdlib}
8328 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
8329 @cindex @option{-nodefaultlibs} and unresolved references
8330 @cindex unresolved references and @option{-nodefaultlibs}
8331 One of the standard libraries bypassed by @option{-nostdlib} and
8332 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
8333 that GCC uses to overcome shortcomings of particular machines, or special
8334 needs for some languages.
8335 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
8336 Collection (GCC) Internals},
8337 for more discussion of @file{libgcc.a}.)
8338 In most cases, you need @file{libgcc.a} even when you want to avoid
8339 other standard libraries. In other words, when you specify @option{-nostdlib}
8340 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
8341 This ensures that you have no unresolved references to internal GCC
8342 library subroutines. (For example, @samp{__main}, used to ensure C++
8343 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
8344 GNU Compiler Collection (GCC) Internals}.)
8348 Produce a position independent executable on targets which support it.
8349 For predictable results, you must also specify the same set of options
8350 that were used to generate code (@option{-fpie}, @option{-fPIE},
8351 or model suboptions) when you specify this option.
8355 Pass the flag @option{-export-dynamic} to the ELF linker, on targets
8356 that support it. This instructs the linker to add all symbols, not
8357 only used ones, to the dynamic symbol table. This option is needed
8358 for some uses of @code{dlopen} or to allow obtaining backtraces
8359 from within a program.
8363 Remove all symbol table and relocation information from the executable.
8367 On systems that support dynamic linking, this prevents linking with the shared
8368 libraries. On other systems, this option has no effect.
8372 Produce a shared object which can then be linked with other objects to
8373 form an executable. Not all systems support this option. For predictable
8374 results, you must also specify the same set of options that were used to
8375 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
8376 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
8377 needs to build supplementary stub code for constructors to work. On
8378 multi-libbed systems, @samp{gcc -shared} must select the correct support
8379 libraries to link against. Failing to supply the correct flags may lead
8380 to subtle defects. Supplying them in cases where they are not necessary
8383 @item -shared-libgcc
8384 @itemx -static-libgcc
8385 @opindex shared-libgcc
8386 @opindex static-libgcc
8387 On systems that provide @file{libgcc} as a shared library, these options
8388 force the use of either the shared or static version respectively.
8389 If no shared version of @file{libgcc} was built when the compiler was
8390 configured, these options have no effect.
8392 There are several situations in which an application should use the
8393 shared @file{libgcc} instead of the static version. The most common
8394 of these is when the application wishes to throw and catch exceptions
8395 across different shared libraries. In that case, each of the libraries
8396 as well as the application itself should use the shared @file{libgcc}.
8398 Therefore, the G++ and GCJ drivers automatically add
8399 @option{-shared-libgcc} whenever you build a shared library or a main
8400 executable, because C++ and Java programs typically use exceptions, so
8401 this is the right thing to do.
8403 If, instead, you use the GCC driver to create shared libraries, you may
8404 find that they will not always be linked with the shared @file{libgcc}.
8405 If GCC finds, at its configuration time, that you have a non-GNU linker
8406 or a GNU linker that does not support option @option{--eh-frame-hdr},
8407 it will link the shared version of @file{libgcc} into shared libraries
8408 by default. Otherwise, it will take advantage of the linker and optimize
8409 away the linking with the shared version of @file{libgcc}, linking with
8410 the static version of libgcc by default. This allows exceptions to
8411 propagate through such shared libraries, without incurring relocation
8412 costs at library load time.
8414 However, if a library or main executable is supposed to throw or catch
8415 exceptions, you must link it using the G++ or GCJ driver, as appropriate
8416 for the languages used in the program, or using the option
8417 @option{-shared-libgcc}, such that it is linked with the shared
8420 @item -static-libstdc++
8421 When the @command{g++} program is used to link a C++ program, it will
8422 normally automatically link against @option{libstdc++}. If
8423 @file{libstdc++} is available as a shared library, and the
8424 @option{-static} option is not used, then this will link against the
8425 shared version of @file{libstdc++}. That is normally fine. However, it
8426 is sometimes useful to freeze the version of @file{libstdc++} used by
8427 the program without going all the way to a fully static link. The
8428 @option{-static-libstdc++} option directs the @command{g++} driver to
8429 link @file{libstdc++} statically, without necessarily linking other
8430 libraries statically.
8434 Bind references to global symbols when building a shared object. Warn
8435 about any unresolved references (unless overridden by the link editor
8436 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
8439 @item -T @var{script}
8441 @cindex linker script
8442 Use @var{script} as the linker script. This option is supported by most
8443 systems using the GNU linker. On some targets, such as bare-board
8444 targets without an operating system, the @option{-T} option may be required
8445 when linking to avoid references to undefined symbols.
8447 @item -Xlinker @var{option}
8449 Pass @var{option} as an option to the linker. You can use this to
8450 supply system-specific linker options which GCC does not know how to
8453 If you want to pass an option that takes a separate argument, you must use
8454 @option{-Xlinker} twice, once for the option and once for the argument.
8455 For example, to pass @option{-assert definitions}, you must write
8456 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
8457 @option{-Xlinker "-assert definitions"}, because this passes the entire
8458 string as a single argument, which is not what the linker expects.
8460 When using the GNU linker, it is usually more convenient to pass
8461 arguments to linker options using the @option{@var{option}=@var{value}}
8462 syntax than as separate arguments. For example, you can specify
8463 @samp{-Xlinker -Map=output.map} rather than
8464 @samp{-Xlinker -Map -Xlinker output.map}. Other linkers may not support
8465 this syntax for command-line options.
8467 @item -Wl,@var{option}
8469 Pass @var{option} as an option to the linker. If @var{option} contains
8470 commas, it is split into multiple options at the commas. You can use this
8471 syntax to pass an argument to the option.
8472 For example, @samp{-Wl,-Map,output.map} passes @samp{-Map output.map} to the
8473 linker. When using the GNU linker, you can also get the same effect with
8474 @samp{-Wl,-Map=output.map}.
8476 @item -u @var{symbol}
8478 Pretend the symbol @var{symbol} is undefined, to force linking of
8479 library modules to define it. You can use @option{-u} multiple times with
8480 different symbols to force loading of additional library modules.
8483 @node Directory Options
8484 @section Options for Directory Search
8485 @cindex directory options
8486 @cindex options, directory search
8489 These options specify directories to search for header files, for
8490 libraries and for parts of the compiler:
8495 Add the directory @var{dir} to the head of the list of directories to be
8496 searched for header files. This can be used to override a system header
8497 file, substituting your own version, since these directories are
8498 searched before the system header file directories. However, you should
8499 not use this option to add directories that contain vendor-supplied
8500 system header files (use @option{-isystem} for that). If you use more than
8501 one @option{-I} option, the directories are scanned in left-to-right
8502 order; the standard system directories come after.
8504 If a standard system include directory, or a directory specified with
8505 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
8506 option will be ignored. The directory will still be searched but as a
8507 system directory at its normal position in the system include chain.
8508 This is to ensure that GCC's procedure to fix buggy system headers and
8509 the ordering for the include_next directive are not inadvertently changed.
8510 If you really need to change the search order for system directories,
8511 use the @option{-nostdinc} and/or @option{-isystem} options.
8513 @item -iquote@var{dir}
8515 Add the directory @var{dir} to the head of the list of directories to
8516 be searched for header files only for the case of @samp{#include
8517 "@var{file}"}; they are not searched for @samp{#include <@var{file}>},
8518 otherwise just like @option{-I}.
8522 Add directory @var{dir} to the list of directories to be searched
8525 @item -B@var{prefix}
8527 This option specifies where to find the executables, libraries,
8528 include files, and data files of the compiler itself.
8530 The compiler driver program runs one or more of the subprograms
8531 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
8532 @var{prefix} as a prefix for each program it tries to run, both with and
8533 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
8535 For each subprogram to be run, the compiler driver first tries the
8536 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
8537 was not specified, the driver tries two standard prefixes, which are
8538 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
8539 those results in a file name that is found, the unmodified program
8540 name is searched for using the directories specified in your
8541 @env{PATH} environment variable.
8543 The compiler will check to see if the path provided by the @option{-B}
8544 refers to a directory, and if necessary it will add a directory
8545 separator character at the end of the path.
8547 @option{-B} prefixes that effectively specify directory names also apply
8548 to libraries in the linker, because the compiler translates these
8549 options into @option{-L} options for the linker. They also apply to
8550 includes files in the preprocessor, because the compiler translates these
8551 options into @option{-isystem} options for the preprocessor. In this case,
8552 the compiler appends @samp{include} to the prefix.
8554 The run-time support file @file{libgcc.a} can also be searched for using
8555 the @option{-B} prefix, if needed. If it is not found there, the two
8556 standard prefixes above are tried, and that is all. The file is left
8557 out of the link if it is not found by those means.
8559 Another way to specify a prefix much like the @option{-B} prefix is to use
8560 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
8563 As a special kludge, if the path provided by @option{-B} is
8564 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
8565 9, then it will be replaced by @file{[dir/]include}. This is to help
8566 with boot-strapping the compiler.
8568 @item -specs=@var{file}
8570 Process @var{file} after the compiler reads in the standard @file{specs}
8571 file, in order to override the defaults that the @file{gcc} driver
8572 program uses when determining what switches to pass to @file{cc1},
8573 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
8574 @option{-specs=@var{file}} can be specified on the command line, and they
8575 are processed in order, from left to right.
8577 @item --sysroot=@var{dir}
8579 Use @var{dir} as the logical root directory for headers and libraries.
8580 For example, if the compiler would normally search for headers in
8581 @file{/usr/include} and libraries in @file{/usr/lib}, it will instead
8582 search @file{@var{dir}/usr/include} and @file{@var{dir}/usr/lib}.
8584 If you use both this option and the @option{-isysroot} option, then
8585 the @option{--sysroot} option will apply to libraries, but the
8586 @option{-isysroot} option will apply to header files.
8588 The GNU linker (beginning with version 2.16) has the necessary support
8589 for this option. If your linker does not support this option, the
8590 header file aspect of @option{--sysroot} will still work, but the
8591 library aspect will not.
8595 This option has been deprecated. Please use @option{-iquote} instead for
8596 @option{-I} directories before the @option{-I-} and remove the @option{-I-}.
8597 Any directories you specify with @option{-I} options before the @option{-I-}
8598 option are searched only for the case of @samp{#include "@var{file}"};
8599 they are not searched for @samp{#include <@var{file}>}.
8601 If additional directories are specified with @option{-I} options after
8602 the @option{-I-}, these directories are searched for all @samp{#include}
8603 directives. (Ordinarily @emph{all} @option{-I} directories are used
8606 In addition, the @option{-I-} option inhibits the use of the current
8607 directory (where the current input file came from) as the first search
8608 directory for @samp{#include "@var{file}"}. There is no way to
8609 override this effect of @option{-I-}. With @option{-I.} you can specify
8610 searching the directory which was current when the compiler was
8611 invoked. That is not exactly the same as what the preprocessor does
8612 by default, but it is often satisfactory.
8614 @option{-I-} does not inhibit the use of the standard system directories
8615 for header files. Thus, @option{-I-} and @option{-nostdinc} are
8622 @section Specifying subprocesses and the switches to pass to them
8625 @command{gcc} is a driver program. It performs its job by invoking a
8626 sequence of other programs to do the work of compiling, assembling and
8627 linking. GCC interprets its command-line parameters and uses these to
8628 deduce which programs it should invoke, and which command-line options
8629 it ought to place on their command lines. This behavior is controlled
8630 by @dfn{spec strings}. In most cases there is one spec string for each
8631 program that GCC can invoke, but a few programs have multiple spec
8632 strings to control their behavior. The spec strings built into GCC can
8633 be overridden by using the @option{-specs=} command-line switch to specify
8636 @dfn{Spec files} are plaintext files that are used to construct spec
8637 strings. They consist of a sequence of directives separated by blank
8638 lines. The type of directive is determined by the first non-whitespace
8639 character on the line and it can be one of the following:
8642 @item %@var{command}
8643 Issues a @var{command} to the spec file processor. The commands that can
8647 @item %include <@var{file}>
8649 Search for @var{file} and insert its text at the current point in the
8652 @item %include_noerr <@var{file}>
8653 @cindex %include_noerr
8654 Just like @samp{%include}, but do not generate an error message if the include
8655 file cannot be found.
8657 @item %rename @var{old_name} @var{new_name}
8659 Rename the spec string @var{old_name} to @var{new_name}.
8663 @item *[@var{spec_name}]:
8664 This tells the compiler to create, override or delete the named spec
8665 string. All lines after this directive up to the next directive or
8666 blank line are considered to be the text for the spec string. If this
8667 results in an empty string then the spec will be deleted. (Or, if the
8668 spec did not exist, then nothing will happened.) Otherwise, if the spec
8669 does not currently exist a new spec will be created. If the spec does
8670 exist then its contents will be overridden by the text of this
8671 directive, unless the first character of that text is the @samp{+}
8672 character, in which case the text will be appended to the spec.
8674 @item [@var{suffix}]:
8675 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
8676 and up to the next directive or blank line are considered to make up the
8677 spec string for the indicated suffix. When the compiler encounters an
8678 input file with the named suffix, it will processes the spec string in
8679 order to work out how to compile that file. For example:
8686 This says that any input file whose name ends in @samp{.ZZ} should be
8687 passed to the program @samp{z-compile}, which should be invoked with the
8688 command-line switch @option{-input} and with the result of performing the
8689 @samp{%i} substitution. (See below.)
8691 As an alternative to providing a spec string, the text that follows a
8692 suffix directive can be one of the following:
8695 @item @@@var{language}
8696 This says that the suffix is an alias for a known @var{language}. This is
8697 similar to using the @option{-x} command-line switch to GCC to specify a
8698 language explicitly. For example:
8705 Says that .ZZ files are, in fact, C++ source files.
8708 This causes an error messages saying:
8711 @var{name} compiler not installed on this system.
8715 GCC already has an extensive list of suffixes built into it.
8716 This directive will add an entry to the end of the list of suffixes, but
8717 since the list is searched from the end backwards, it is effectively
8718 possible to override earlier entries using this technique.
8722 GCC has the following spec strings built into it. Spec files can
8723 override these strings or create their own. Note that individual
8724 targets can also add their own spec strings to this list.
8727 asm Options to pass to the assembler
8728 asm_final Options to pass to the assembler post-processor
8729 cpp Options to pass to the C preprocessor
8730 cc1 Options to pass to the C compiler
8731 cc1plus Options to pass to the C++ compiler
8732 endfile Object files to include at the end of the link
8733 link Options to pass to the linker
8734 lib Libraries to include on the command line to the linker
8735 libgcc Decides which GCC support library to pass to the linker
8736 linker Sets the name of the linker
8737 predefines Defines to be passed to the C preprocessor
8738 signed_char Defines to pass to CPP to say whether @code{char} is signed
8740 startfile Object files to include at the start of the link
8743 Here is a small example of a spec file:
8749 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
8752 This example renames the spec called @samp{lib} to @samp{old_lib} and
8753 then overrides the previous definition of @samp{lib} with a new one.
8754 The new definition adds in some extra command-line options before
8755 including the text of the old definition.
8757 @dfn{Spec strings} are a list of command-line options to be passed to their
8758 corresponding program. In addition, the spec strings can contain
8759 @samp{%}-prefixed sequences to substitute variable text or to
8760 conditionally insert text into the command line. Using these constructs
8761 it is possible to generate quite complex command lines.
8763 Here is a table of all defined @samp{%}-sequences for spec
8764 strings. Note that spaces are not generated automatically around the
8765 results of expanding these sequences. Therefore you can concatenate them
8766 together or combine them with constant text in a single argument.
8770 Substitute one @samp{%} into the program name or argument.
8773 Substitute the name of the input file being processed.
8776 Substitute the basename of the input file being processed.
8777 This is the substring up to (and not including) the last period
8778 and not including the directory.
8781 This is the same as @samp{%b}, but include the file suffix (text after
8785 Marks the argument containing or following the @samp{%d} as a
8786 temporary file name, so that that file will be deleted if GCC exits
8787 successfully. Unlike @samp{%g}, this contributes no text to the
8790 @item %g@var{suffix}
8791 Substitute a file name that has suffix @var{suffix} and is chosen
8792 once per compilation, and mark the argument in the same way as
8793 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
8794 name is now chosen in a way that is hard to predict even when previously
8795 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
8796 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
8797 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
8798 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
8799 was simply substituted with a file name chosen once per compilation,
8800 without regard to any appended suffix (which was therefore treated
8801 just like ordinary text), making such attacks more likely to succeed.
8803 @item %u@var{suffix}
8804 Like @samp{%g}, but generates a new temporary file name even if
8805 @samp{%u@var{suffix}} was already seen.
8807 @item %U@var{suffix}
8808 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
8809 new one if there is no such last file name. In the absence of any
8810 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
8811 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
8812 would involve the generation of two distinct file names, one
8813 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
8814 simply substituted with a file name chosen for the previous @samp{%u},
8815 without regard to any appended suffix.
8817 @item %j@var{suffix}
8818 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
8819 writable, and if save-temps is off; otherwise, substitute the name
8820 of a temporary file, just like @samp{%u}. This temporary file is not
8821 meant for communication between processes, but rather as a junk
8824 @item %|@var{suffix}
8825 @itemx %m@var{suffix}
8826 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
8827 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
8828 all. These are the two most common ways to instruct a program that it
8829 should read from standard input or write to standard output. If you
8830 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
8831 construct: see for example @file{f/lang-specs.h}.
8833 @item %.@var{SUFFIX}
8834 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
8835 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
8836 terminated by the next space or %.
8839 Marks the argument containing or following the @samp{%w} as the
8840 designated output file of this compilation. This puts the argument
8841 into the sequence of arguments that @samp{%o} will substitute later.
8844 Substitutes the names of all the output files, with spaces
8845 automatically placed around them. You should write spaces
8846 around the @samp{%o} as well or the results are undefined.
8847 @samp{%o} is for use in the specs for running the linker.
8848 Input files whose names have no recognized suffix are not compiled
8849 at all, but they are included among the output files, so they will
8853 Substitutes the suffix for object files. Note that this is
8854 handled specially when it immediately follows @samp{%g, %u, or %U},
8855 because of the need for those to form complete file names. The
8856 handling is such that @samp{%O} is treated exactly as if it had already
8857 been substituted, except that @samp{%g, %u, and %U} do not currently
8858 support additional @var{suffix} characters following @samp{%O} as they would
8859 following, for example, @samp{.o}.
8862 Substitutes the standard macro predefinitions for the
8863 current target machine. Use this when running @code{cpp}.
8866 Like @samp{%p}, but puts @samp{__} before and after the name of each
8867 predefined macro, except for macros that start with @samp{__} or with
8868 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
8872 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
8873 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}),
8874 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
8875 and @option{-imultilib} as necessary.
8878 Current argument is the name of a library or startup file of some sort.
8879 Search for that file in a standard list of directories and substitute
8880 the full name found. The current working directory is included in the
8881 list of directories scanned.
8884 Current argument is the name of a linker script. Search for that file
8885 in the current list of directories to scan for libraries. If the file
8886 is located insert a @option{--script} option into the command line
8887 followed by the full path name found. If the file is not found then
8888 generate an error message. Note: the current working directory is not
8892 Print @var{str} as an error message. @var{str} is terminated by a newline.
8893 Use this when inconsistent options are detected.
8896 Substitute the contents of spec string @var{name} at this point.
8899 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
8901 @item %x@{@var{option}@}
8902 Accumulate an option for @samp{%X}.
8905 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
8909 Output the accumulated assembler options specified by @option{-Wa}.
8912 Output the accumulated preprocessor options specified by @option{-Wp}.
8915 Process the @code{asm} spec. This is used to compute the
8916 switches to be passed to the assembler.
8919 Process the @code{asm_final} spec. This is a spec string for
8920 passing switches to an assembler post-processor, if such a program is
8924 Process the @code{link} spec. This is the spec for computing the
8925 command line passed to the linker. Typically it will make use of the
8926 @samp{%L %G %S %D and %E} sequences.
8929 Dump out a @option{-L} option for each directory that GCC believes might
8930 contain startup files. If the target supports multilibs then the
8931 current multilib directory will be prepended to each of these paths.
8934 Process the @code{lib} spec. This is a spec string for deciding which
8935 libraries should be included on the command line to the linker.
8938 Process the @code{libgcc} spec. This is a spec string for deciding
8939 which GCC support library should be included on the command line to the linker.
8942 Process the @code{startfile} spec. This is a spec for deciding which
8943 object files should be the first ones passed to the linker. Typically
8944 this might be a file named @file{crt0.o}.
8947 Process the @code{endfile} spec. This is a spec string that specifies
8948 the last object files that will be passed to the linker.
8951 Process the @code{cpp} spec. This is used to construct the arguments
8952 to be passed to the C preprocessor.
8955 Process the @code{cc1} spec. This is used to construct the options to be
8956 passed to the actual C compiler (@samp{cc1}).
8959 Process the @code{cc1plus} spec. This is used to construct the options to be
8960 passed to the actual C++ compiler (@samp{cc1plus}).
8963 Substitute the variable part of a matched option. See below.
8964 Note that each comma in the substituted string is replaced by
8968 Remove all occurrences of @code{-S} from the command line. Note---this
8969 command is position dependent. @samp{%} commands in the spec string
8970 before this one will see @code{-S}, @samp{%} commands in the spec string
8971 after this one will not.
8973 @item %:@var{function}(@var{args})
8974 Call the named function @var{function}, passing it @var{args}.
8975 @var{args} is first processed as a nested spec string, then split
8976 into an argument vector in the usual fashion. The function returns
8977 a string which is processed as if it had appeared literally as part
8978 of the current spec.
8980 The following built-in spec functions are provided:
8984 The @code{getenv} spec function takes two arguments: an environment
8985 variable name and a string. If the environment variable is not
8986 defined, a fatal error is issued. Otherwise, the return value is the
8987 value of the environment variable concatenated with the string. For
8988 example, if @env{TOPDIR} is defined as @file{/path/to/top}, then:
8991 %:getenv(TOPDIR /include)
8994 expands to @file{/path/to/top/include}.
8996 @item @code{if-exists}
8997 The @code{if-exists} spec function takes one argument, an absolute
8998 pathname to a file. If the file exists, @code{if-exists} returns the
8999 pathname. Here is a small example of its usage:
9003 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
9006 @item @code{if-exists-else}
9007 The @code{if-exists-else} spec function is similar to the @code{if-exists}
9008 spec function, except that it takes two arguments. The first argument is
9009 an absolute pathname to a file. If the file exists, @code{if-exists-else}
9010 returns the pathname. If it does not exist, it returns the second argument.
9011 This way, @code{if-exists-else} can be used to select one file or another,
9012 based on the existence of the first. Here is a small example of its usage:
9016 crt0%O%s %:if-exists(crti%O%s) \
9017 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
9020 @item @code{replace-outfile}
9021 The @code{replace-outfile} spec function takes two arguments. It looks for the
9022 first argument in the outfiles array and replaces it with the second argument. Here
9023 is a small example of its usage:
9026 %@{fgnu-runtime:%:replace-outfile(-lobjc -lobjc-gnu)@}
9029 @item @code{print-asm-header}
9030 The @code{print-asm-header} function takes no arguments and simply
9031 prints a banner like:
9037 Use "-Wa,OPTION" to pass "OPTION" to the assembler.
9040 It is used to separate compiler options from assembler options
9041 in the @option{--target-help} output.
9045 Substitutes the @code{-S} switch, if that switch was given to GCC@.
9046 If that switch was not specified, this substitutes nothing. Note that
9047 the leading dash is omitted when specifying this option, and it is
9048 automatically inserted if the substitution is performed. Thus the spec
9049 string @samp{%@{foo@}} would match the command-line option @option{-foo}
9050 and would output the command line option @option{-foo}.
9052 @item %W@{@code{S}@}
9053 Like %@{@code{S}@} but mark last argument supplied within as a file to be
9056 @item %@{@code{S}*@}
9057 Substitutes all the switches specified to GCC whose names start
9058 with @code{-S}, but which also take an argument. This is used for
9059 switches like @option{-o}, @option{-D}, @option{-I}, etc.
9060 GCC considers @option{-o foo} as being
9061 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
9062 text, including the space. Thus two arguments would be generated.
9064 @item %@{@code{S}*&@code{T}*@}
9065 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
9066 (the order of @code{S} and @code{T} in the spec is not significant).
9067 There can be any number of ampersand-separated variables; for each the
9068 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
9070 @item %@{@code{S}:@code{X}@}
9071 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
9073 @item %@{!@code{S}:@code{X}@}
9074 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
9076 @item %@{@code{S}*:@code{X}@}
9077 Substitutes @code{X} if one or more switches whose names start with
9078 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
9079 once, no matter how many such switches appeared. However, if @code{%*}
9080 appears somewhere in @code{X}, then @code{X} will be substituted once
9081 for each matching switch, with the @code{%*} replaced by the part of
9082 that switch that matched the @code{*}.
9084 @item %@{.@code{S}:@code{X}@}
9085 Substitutes @code{X}, if processing a file with suffix @code{S}.
9087 @item %@{!.@code{S}:@code{X}@}
9088 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
9090 @item %@{,@code{S}:@code{X}@}
9091 Substitutes @code{X}, if processing a file for language @code{S}.
9093 @item %@{!,@code{S}:@code{X}@}
9094 Substitutes @code{X}, if not processing a file for language @code{S}.
9096 @item %@{@code{S}|@code{P}:@code{X}@}
9097 Substitutes @code{X} if either @code{-S} or @code{-P} was given to
9098 GCC@. This may be combined with @samp{!}, @samp{.}, @samp{,}, and
9099 @code{*} sequences as well, although they have a stronger binding than
9100 the @samp{|}. If @code{%*} appears in @code{X}, all of the
9101 alternatives must be starred, and only the first matching alternative
9104 For example, a spec string like this:
9107 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
9110 will output the following command-line options from the following input
9111 command-line options:
9116 -d fred.c -foo -baz -boggle
9117 -d jim.d -bar -baz -boggle
9120 @item %@{S:X; T:Y; :D@}
9122 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
9123 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
9124 be as many clauses as you need. This may be combined with @code{.},
9125 @code{,}, @code{!}, @code{|}, and @code{*} as needed.
9130 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
9131 construct may contain other nested @samp{%} constructs or spaces, or
9132 even newlines. They are processed as usual, as described above.
9133 Trailing white space in @code{X} is ignored. White space may also
9134 appear anywhere on the left side of the colon in these constructs,
9135 except between @code{.} or @code{*} and the corresponding word.
9137 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
9138 handled specifically in these constructs. If another value of
9139 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
9140 @option{-W} switch is found later in the command line, the earlier
9141 switch value is ignored, except with @{@code{S}*@} where @code{S} is
9142 just one letter, which passes all matching options.
9144 The character @samp{|} at the beginning of the predicate text is used to
9145 indicate that a command should be piped to the following command, but
9146 only if @option{-pipe} is specified.
9148 It is built into GCC which switches take arguments and which do not.
9149 (You might think it would be useful to generalize this to allow each
9150 compiler's spec to say which switches take arguments. But this cannot
9151 be done in a consistent fashion. GCC cannot even decide which input
9152 files have been specified without knowing which switches take arguments,
9153 and it must know which input files to compile in order to tell which
9156 GCC also knows implicitly that arguments starting in @option{-l} are to be
9157 treated as compiler output files, and passed to the linker in their
9158 proper position among the other output files.
9160 @c man begin OPTIONS
9162 @node Target Options
9163 @section Specifying Target Machine and Compiler Version
9164 @cindex target options
9165 @cindex cross compiling
9166 @cindex specifying machine version
9167 @cindex specifying compiler version and target machine
9168 @cindex compiler version, specifying
9169 @cindex target machine, specifying
9171 The usual way to run GCC is to run the executable called @file{gcc}, or
9172 @file{<machine>-gcc} when cross-compiling, or
9173 @file{<machine>-gcc-<version>} to run a version other than the one that
9174 was installed last. Sometimes this is inconvenient, so GCC provides
9175 options that will switch to another cross-compiler or version.
9178 @item -b @var{machine}
9180 The argument @var{machine} specifies the target machine for compilation.
9182 The value to use for @var{machine} is the same as was specified as the
9183 machine type when configuring GCC as a cross-compiler. For
9184 example, if a cross-compiler was configured with @samp{configure
9185 arm-elf}, meaning to compile for an arm processor with elf binaries,
9186 then you would specify @option{-b arm-elf} to run that cross compiler.
9187 Because there are other options beginning with @option{-b}, the
9188 configuration must contain a hyphen, or @option{-b} alone should be one
9189 argument followed by the configuration in the next argument.
9191 @item -V @var{version}
9193 The argument @var{version} specifies which version of GCC to run.
9194 This is useful when multiple versions are installed. For example,
9195 @var{version} might be @samp{4.0}, meaning to run GCC version 4.0.
9198 The @option{-V} and @option{-b} options work by running the
9199 @file{<machine>-gcc-<version>} executable, so there's no real reason to
9200 use them if you can just run that directly.
9202 @node Submodel Options
9203 @section Hardware Models and Configurations
9204 @cindex submodel options
9205 @cindex specifying hardware config
9206 @cindex hardware models and configurations, specifying
9207 @cindex machine dependent options
9209 Earlier we discussed the standard option @option{-b} which chooses among
9210 different installed compilers for completely different target
9211 machines, such as VAX vs.@: 68000 vs.@: 80386.
9213 In addition, each of these target machine types can have its own
9214 special options, starting with @samp{-m}, to choose among various
9215 hardware models or configurations---for example, 68010 vs 68020,
9216 floating coprocessor or none. A single installed version of the
9217 compiler can compile for any model or configuration, according to the
9220 Some configurations of the compiler also support additional special
9221 options, usually for compatibility with other compilers on the same
9224 @c This list is ordered alphanumerically by subsection name.
9225 @c It should be the same order and spelling as these options are listed
9226 @c in Machine Dependent Options
9232 * Blackfin Options::
9236 * DEC Alpha Options::
9237 * DEC Alpha/VMS Options::
9240 * GNU/Linux Options::
9243 * i386 and x86-64 Options::
9244 * i386 and x86-64 Windows Options::
9246 * IA-64/VMS Options::
9257 * picoChip Options::
9259 * RS/6000 and PowerPC Options::
9260 * S/390 and zSeries Options::
9265 * System V Options::
9270 * Xstormy16 Options::
9276 @subsection ARC Options
9279 These options are defined for ARC implementations:
9284 Compile code for little endian mode. This is the default.
9288 Compile code for big endian mode.
9291 @opindex mmangle-cpu
9292 Prepend the name of the cpu to all public symbol names.
9293 In multiple-processor systems, there are many ARC variants with different
9294 instruction and register set characteristics. This flag prevents code
9295 compiled for one cpu to be linked with code compiled for another.
9296 No facility exists for handling variants that are ``almost identical''.
9297 This is an all or nothing option.
9299 @item -mcpu=@var{cpu}
9301 Compile code for ARC variant @var{cpu}.
9302 Which variants are supported depend on the configuration.
9303 All variants support @option{-mcpu=base}, this is the default.
9305 @item -mtext=@var{text-section}
9306 @itemx -mdata=@var{data-section}
9307 @itemx -mrodata=@var{readonly-data-section}
9311 Put functions, data, and readonly data in @var{text-section},
9312 @var{data-section}, and @var{readonly-data-section} respectively
9313 by default. This can be overridden with the @code{section} attribute.
9314 @xref{Variable Attributes}.
9316 @item -mfix-cortex-m3-ldrd
9317 @opindex mfix-cortex-m3-ldrd
9318 Some Cortex-M3 cores can cause data corruption when @code{ldrd} instructions
9319 with overlapping destination and base registers are used. This option avoids
9320 generating these instructions. This option is enabled by default when
9321 @option{-mcpu=cortex-m3} is specified.
9326 @subsection ARM Options
9329 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
9333 @item -mabi=@var{name}
9335 Generate code for the specified ABI@. Permissible values are: @samp{apcs-gnu},
9336 @samp{atpcs}, @samp{aapcs}, @samp{aapcs-linux} and @samp{iwmmxt}.
9339 @opindex mapcs-frame
9340 Generate a stack frame that is compliant with the ARM Procedure Call
9341 Standard for all functions, even if this is not strictly necessary for
9342 correct execution of the code. Specifying @option{-fomit-frame-pointer}
9343 with this option will cause the stack frames not to be generated for
9344 leaf functions. The default is @option{-mno-apcs-frame}.
9348 This is a synonym for @option{-mapcs-frame}.
9351 @c not currently implemented
9352 @item -mapcs-stack-check
9353 @opindex mapcs-stack-check
9354 Generate code to check the amount of stack space available upon entry to
9355 every function (that actually uses some stack space). If there is
9356 insufficient space available then either the function
9357 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
9358 called, depending upon the amount of stack space required. The run time
9359 system is required to provide these functions. The default is
9360 @option{-mno-apcs-stack-check}, since this produces smaller code.
9362 @c not currently implemented
9364 @opindex mapcs-float
9365 Pass floating point arguments using the float point registers. This is
9366 one of the variants of the APCS@. This option is recommended if the
9367 target hardware has a floating point unit or if a lot of floating point
9368 arithmetic is going to be performed by the code. The default is
9369 @option{-mno-apcs-float}, since integer only code is slightly increased in
9370 size if @option{-mapcs-float} is used.
9372 @c not currently implemented
9373 @item -mapcs-reentrant
9374 @opindex mapcs-reentrant
9375 Generate reentrant, position independent code. The default is
9376 @option{-mno-apcs-reentrant}.
9379 @item -mthumb-interwork
9380 @opindex mthumb-interwork
9381 Generate code which supports calling between the ARM and Thumb
9382 instruction sets. Without this option the two instruction sets cannot
9383 be reliably used inside one program. The default is
9384 @option{-mno-thumb-interwork}, since slightly larger code is generated
9385 when @option{-mthumb-interwork} is specified.
9387 @item -mno-sched-prolog
9388 @opindex mno-sched-prolog
9389 Prevent the reordering of instructions in the function prolog, or the
9390 merging of those instruction with the instructions in the function's
9391 body. This means that all functions will start with a recognizable set
9392 of instructions (or in fact one of a choice from a small set of
9393 different function prologues), and this information can be used to
9394 locate the start if functions inside an executable piece of code. The
9395 default is @option{-msched-prolog}.
9397 @item -mfloat-abi=@var{name}
9399 Specifies which floating-point ABI to use. Permissible values
9400 are: @samp{soft}, @samp{softfp} and @samp{hard}.
9402 Specifying @samp{soft} causes GCC to generate output containing
9403 library calls for floating-point operations.
9404 @samp{softfp} allows the generation of code using hardware floating-point
9405 instructions, but still uses the soft-float calling conventions.
9406 @samp{hard} allows generation of floating-point instructions
9407 and uses FPU-specific calling conventions.
9409 The default depends on the specific target configuration. Note that
9410 the hard-float and soft-float ABIs are not link-compatible; you must
9411 compile your entire program with the same ABI, and link with a
9412 compatible set of libraries.
9415 @opindex mhard-float
9416 Equivalent to @option{-mfloat-abi=hard}.
9419 @opindex msoft-float
9420 Equivalent to @option{-mfloat-abi=soft}.
9422 @item -mlittle-endian
9423 @opindex mlittle-endian
9424 Generate code for a processor running in little-endian mode. This is
9425 the default for all standard configurations.
9428 @opindex mbig-endian
9429 Generate code for a processor running in big-endian mode; the default is
9430 to compile code for a little-endian processor.
9432 @item -mwords-little-endian
9433 @opindex mwords-little-endian
9434 This option only applies when generating code for big-endian processors.
9435 Generate code for a little-endian word order but a big-endian byte
9436 order. That is, a byte order of the form @samp{32107654}. Note: this
9437 option should only be used if you require compatibility with code for
9438 big-endian ARM processors generated by versions of the compiler prior to
9441 @item -mcpu=@var{name}
9443 This specifies the name of the target ARM processor. GCC uses this name
9444 to determine what kind of instructions it can emit when generating
9445 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
9446 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
9447 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
9448 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
9449 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
9451 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm7tdmi-s},
9452 @samp{arm710t}, @samp{arm720t}, @samp{arm740t},
9453 @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
9454 @samp{strongarm1110},
9455 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
9456 @samp{arm920t}, @samp{arm922t}, @samp{arm946e-s}, @samp{arm966e-s},
9457 @samp{arm968e-s}, @samp{arm926ej-s}, @samp{arm940t}, @samp{arm9tdmi},
9458 @samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ej-s},
9459 @samp{arm10e}, @samp{arm1020e}, @samp{arm1022e},
9460 @samp{arm1136j-s}, @samp{arm1136jf-s}, @samp{mpcore}, @samp{mpcorenovfp},
9461 @samp{arm1156t2-s}, @samp{arm1156t2f-s}, @samp{arm1176jz-s}, @samp{arm1176jzf-s},
9462 @samp{cortex-a8}, @samp{cortex-a9},
9463 @samp{cortex-r4}, @samp{cortex-r4f}, @samp{cortex-m3},
9466 @samp{xscale}, @samp{iwmmxt}, @samp{iwmmxt2}, @samp{ep9312}.
9468 @item -mtune=@var{name}
9470 This option is very similar to the @option{-mcpu=} option, except that
9471 instead of specifying the actual target processor type, and hence
9472 restricting which instructions can be used, it specifies that GCC should
9473 tune the performance of the code as if the target were of the type
9474 specified in this option, but still choosing the instructions that it
9475 will generate based on the cpu specified by a @option{-mcpu=} option.
9476 For some ARM implementations better performance can be obtained by using
9479 @item -march=@var{name}
9481 This specifies the name of the target ARM architecture. GCC uses this
9482 name to determine what kind of instructions it can emit when generating
9483 assembly code. This option can be used in conjunction with or instead
9484 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
9485 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
9486 @samp{armv5}, @samp{armv5t}, @samp{armv5e}, @samp{armv5te},
9487 @samp{armv6}, @samp{armv6j},
9488 @samp{armv6t2}, @samp{armv6z}, @samp{armv6zk}, @samp{armv6-m},
9489 @samp{armv7}, @samp{armv7-a}, @samp{armv7-r}, @samp{armv7-m},
9490 @samp{iwmmxt}, @samp{iwmmxt2}, @samp{ep9312}.
9492 @item -mfpu=@var{name}
9493 @itemx -mfpe=@var{number}
9494 @itemx -mfp=@var{number}
9498 This specifies what floating point hardware (or hardware emulation) is
9499 available on the target. Permissible names are: @samp{fpa}, @samp{fpe2},
9500 @samp{fpe3}, @samp{maverick}, @samp{vfp}, @samp{vfpv3}, @samp{vfpv3-d16},
9501 @samp{neon}, and @samp{neon-fp16}. @option{-mfp} and @option{-mfpe}
9502 are synonyms for @option{-mfpu}=@samp{fpe}@var{number}, for compatibility
9503 with older versions of GCC@.
9505 If @option{-msoft-float} is specified this specifies the format of
9506 floating point values.
9508 @item -mfp16-format=@var{name}
9509 @opindex mfp16-format
9510 Specify the format of the @code{__fp16} half-precision floating-point type.
9511 Permissible names are @samp{none}, @samp{ieee}, and @samp{alternative};
9512 the default is @samp{none}, in which case the @code{__fp16} type is not
9513 defined. @xref{Half-Precision}, for more information.
9515 @item -mstructure-size-boundary=@var{n}
9516 @opindex mstructure-size-boundary
9517 The size of all structures and unions will be rounded up to a multiple
9518 of the number of bits set by this option. Permissible values are 8, 32
9519 and 64. The default value varies for different toolchains. For the COFF
9520 targeted toolchain the default value is 8. A value of 64 is only allowed
9521 if the underlying ABI supports it.
9523 Specifying the larger number can produce faster, more efficient code, but
9524 can also increase the size of the program. Different values are potentially
9525 incompatible. Code compiled with one value cannot necessarily expect to
9526 work with code or libraries compiled with another value, if they exchange
9527 information using structures or unions.
9529 @item -mabort-on-noreturn
9530 @opindex mabort-on-noreturn
9531 Generate a call to the function @code{abort} at the end of a
9532 @code{noreturn} function. It will be executed if the function tries to
9536 @itemx -mno-long-calls
9537 @opindex mlong-calls
9538 @opindex mno-long-calls
9539 Tells the compiler to perform function calls by first loading the
9540 address of the function into a register and then performing a subroutine
9541 call on this register. This switch is needed if the target function
9542 will lie outside of the 64 megabyte addressing range of the offset based
9543 version of subroutine call instruction.
9545 Even if this switch is enabled, not all function calls will be turned
9546 into long calls. The heuristic is that static functions, functions
9547 which have the @samp{short-call} attribute, functions that are inside
9548 the scope of a @samp{#pragma no_long_calls} directive and functions whose
9549 definitions have already been compiled within the current compilation
9550 unit, will not be turned into long calls. The exception to this rule is
9551 that weak function definitions, functions with the @samp{long-call}
9552 attribute or the @samp{section} attribute, and functions that are within
9553 the scope of a @samp{#pragma long_calls} directive, will always be
9554 turned into long calls.
9556 This feature is not enabled by default. Specifying
9557 @option{-mno-long-calls} will restore the default behavior, as will
9558 placing the function calls within the scope of a @samp{#pragma
9559 long_calls_off} directive. Note these switches have no effect on how
9560 the compiler generates code to handle function calls via function
9563 @item -msingle-pic-base
9564 @opindex msingle-pic-base
9565 Treat the register used for PIC addressing as read-only, rather than
9566 loading it in the prologue for each function. The run-time system is
9567 responsible for initializing this register with an appropriate value
9568 before execution begins.
9570 @item -mpic-register=@var{reg}
9571 @opindex mpic-register
9572 Specify the register to be used for PIC addressing. The default is R10
9573 unless stack-checking is enabled, when R9 is used.
9575 @item -mcirrus-fix-invalid-insns
9576 @opindex mcirrus-fix-invalid-insns
9577 @opindex mno-cirrus-fix-invalid-insns
9578 Insert NOPs into the instruction stream to in order to work around
9579 problems with invalid Maverick instruction combinations. This option
9580 is only valid if the @option{-mcpu=ep9312} option has been used to
9581 enable generation of instructions for the Cirrus Maverick floating
9582 point co-processor. This option is not enabled by default, since the
9583 problem is only present in older Maverick implementations. The default
9584 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
9587 @item -mpoke-function-name
9588 @opindex mpoke-function-name
9589 Write the name of each function into the text section, directly
9590 preceding the function prologue. The generated code is similar to this:
9594 .ascii "arm_poke_function_name", 0
9597 .word 0xff000000 + (t1 - t0)
9598 arm_poke_function_name
9600 stmfd sp!, @{fp, ip, lr, pc@}
9604 When performing a stack backtrace, code can inspect the value of
9605 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
9606 location @code{pc - 12} and the top 8 bits are set, then we know that
9607 there is a function name embedded immediately preceding this location
9608 and has length @code{((pc[-3]) & 0xff000000)}.
9612 Generate code for the Thumb instruction set. The default is to
9613 use the 32-bit ARM instruction set.
9614 This option automatically enables either 16-bit Thumb-1 or
9615 mixed 16/32-bit Thumb-2 instructions based on the @option{-mcpu=@var{name}}
9616 and @option{-march=@var{name}} options. This option is not passed to the
9617 assembler. If you want to force assembler files to be interpreted as Thumb code,
9618 either add a @samp{.thumb} directive to the source or pass the @option{-mthumb}
9619 option directly to the assembler by prefixing it with @option{-Wa}.
9622 @opindex mtpcs-frame
9623 Generate a stack frame that is compliant with the Thumb Procedure Call
9624 Standard for all non-leaf functions. (A leaf function is one that does
9625 not call any other functions.) The default is @option{-mno-tpcs-frame}.
9627 @item -mtpcs-leaf-frame
9628 @opindex mtpcs-leaf-frame
9629 Generate a stack frame that is compliant with the Thumb Procedure Call
9630 Standard for all leaf functions. (A leaf function is one that does
9631 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
9633 @item -mcallee-super-interworking
9634 @opindex mcallee-super-interworking
9635 Gives all externally visible functions in the file being compiled an ARM
9636 instruction set header which switches to Thumb mode before executing the
9637 rest of the function. This allows these functions to be called from
9638 non-interworking code. This option is not valid in AAPCS configurations
9639 because interworking is enabled by default.
9641 @item -mcaller-super-interworking
9642 @opindex mcaller-super-interworking
9643 Allows calls via function pointers (including virtual functions) to
9644 execute correctly regardless of whether the target code has been
9645 compiled for interworking or not. There is a small overhead in the cost
9646 of executing a function pointer if this option is enabled. This option
9647 is not valid in AAPCS configurations because interworking is enabled
9650 @item -mtp=@var{name}
9652 Specify the access model for the thread local storage pointer. The valid
9653 models are @option{soft}, which generates calls to @code{__aeabi_read_tp},
9654 @option{cp15}, which fetches the thread pointer from @code{cp15} directly
9655 (supported in the arm6k architecture), and @option{auto}, which uses the
9656 best available method for the selected processor. The default setting is
9659 @item -mword-relocations
9660 @opindex mword-relocations
9661 Only generate absolute relocations on word sized values (i.e. R_ARM_ABS32).
9662 This is enabled by default on targets (uClinux, SymbianOS) where the runtime
9663 loader imposes this restriction, and when @option{-fpic} or @option{-fPIC}
9669 @subsection AVR Options
9672 These options are defined for AVR implementations:
9675 @item -mmcu=@var{mcu}
9677 Specify ATMEL AVR instruction set or MCU type.
9679 Instruction set avr1 is for the minimal AVR core, not supported by the C
9680 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
9681 attiny11, attiny12, attiny15, attiny28).
9683 Instruction set avr2 (default) is for the classic AVR core with up to
9684 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
9685 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
9686 at90c8534, at90s8535).
9688 Instruction set avr3 is for the classic AVR core with up to 128K program
9689 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
9691 Instruction set avr4 is for the enhanced AVR core with up to 8K program
9692 memory space (MCU types: atmega8, atmega83, atmega85).
9694 Instruction set avr5 is for the enhanced AVR core with up to 128K program
9695 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
9696 atmega64, atmega128, at43usb355, at94k).
9700 Output instruction sizes to the asm file.
9702 @item -mno-interrupts
9703 @opindex mno-interrupts
9704 Generated code is not compatible with hardware interrupts.
9705 Code size will be smaller.
9707 @item -mcall-prologues
9708 @opindex mcall-prologues
9709 Functions prologues/epilogues expanded as call to appropriate
9710 subroutines. Code size will be smaller.
9713 @opindex mtiny-stack
9714 Change only the low 8 bits of the stack pointer.
9718 Assume int to be 8 bit integer. This affects the sizes of all types: A
9719 char will be 1 byte, an int will be 1 byte, a long will be 2 bytes
9720 and long long will be 4 bytes. Please note that this option does not
9721 comply to the C standards, but it will provide you with smaller code
9725 @node Blackfin Options
9726 @subsection Blackfin Options
9727 @cindex Blackfin Options
9730 @item -mcpu=@var{cpu}@r{[}-@var{sirevision}@r{]}
9732 Specifies the name of the target Blackfin processor. Currently, @var{cpu}
9733 can be one of @samp{bf512}, @samp{bf514}, @samp{bf516}, @samp{bf518},
9734 @samp{bf522}, @samp{bf523}, @samp{bf524}, @samp{bf525}, @samp{bf526},
9735 @samp{bf527}, @samp{bf531}, @samp{bf532}, @samp{bf533},
9736 @samp{bf534}, @samp{bf536}, @samp{bf537}, @samp{bf538}, @samp{bf539},
9737 @samp{bf542}, @samp{bf544}, @samp{bf547}, @samp{bf548}, @samp{bf549},
9738 @samp{bf542m}, @samp{bf544m}, @samp{bf547m}, @samp{bf548m}, @samp{bf549m},
9740 The optional @var{sirevision} specifies the silicon revision of the target
9741 Blackfin processor. Any workarounds available for the targeted silicon revision
9742 will be enabled. If @var{sirevision} is @samp{none}, no workarounds are enabled.
9743 If @var{sirevision} is @samp{any}, all workarounds for the targeted processor
9744 will be enabled. The @code{__SILICON_REVISION__} macro is defined to two
9745 hexadecimal digits representing the major and minor numbers in the silicon
9746 revision. If @var{sirevision} is @samp{none}, the @code{__SILICON_REVISION__}
9747 is not defined. If @var{sirevision} is @samp{any}, the
9748 @code{__SILICON_REVISION__} is defined to be @code{0xffff}.
9749 If this optional @var{sirevision} is not used, GCC assumes the latest known
9750 silicon revision of the targeted Blackfin processor.
9752 Support for @samp{bf561} is incomplete. For @samp{bf561},
9753 Only the processor macro is defined.
9754 Without this option, @samp{bf532} is used as the processor by default.
9755 The corresponding predefined processor macros for @var{cpu} is to
9756 be defined. And for @samp{bfin-elf} toolchain, this causes the hardware BSP
9757 provided by libgloss to be linked in if @option{-msim} is not given.
9761 Specifies that the program will be run on the simulator. This causes
9762 the simulator BSP provided by libgloss to be linked in. This option
9763 has effect only for @samp{bfin-elf} toolchain.
9764 Certain other options, such as @option{-mid-shared-library} and
9765 @option{-mfdpic}, imply @option{-msim}.
9767 @item -momit-leaf-frame-pointer
9768 @opindex momit-leaf-frame-pointer
9769 Don't keep the frame pointer in a register for leaf functions. This
9770 avoids the instructions to save, set up and restore frame pointers and
9771 makes an extra register available in leaf functions. The option
9772 @option{-fomit-frame-pointer} removes the frame pointer for all functions
9773 which might make debugging harder.
9775 @item -mspecld-anomaly
9776 @opindex mspecld-anomaly
9777 When enabled, the compiler will ensure that the generated code does not
9778 contain speculative loads after jump instructions. If this option is used,
9779 @code{__WORKAROUND_SPECULATIVE_LOADS} is defined.
9781 @item -mno-specld-anomaly
9782 @opindex mno-specld-anomaly
9783 Don't generate extra code to prevent speculative loads from occurring.
9785 @item -mcsync-anomaly
9786 @opindex mcsync-anomaly
9787 When enabled, the compiler will ensure that the generated code does not
9788 contain CSYNC or SSYNC instructions too soon after conditional branches.
9789 If this option is used, @code{__WORKAROUND_SPECULATIVE_SYNCS} is defined.
9791 @item -mno-csync-anomaly
9792 @opindex mno-csync-anomaly
9793 Don't generate extra code to prevent CSYNC or SSYNC instructions from
9794 occurring too soon after a conditional branch.
9798 When enabled, the compiler is free to take advantage of the knowledge that
9799 the entire program fits into the low 64k of memory.
9802 @opindex mno-low-64k
9803 Assume that the program is arbitrarily large. This is the default.
9805 @item -mstack-check-l1
9806 @opindex mstack-check-l1
9807 Do stack checking using information placed into L1 scratchpad memory by the
9810 @item -mid-shared-library
9811 @opindex mid-shared-library
9812 Generate code that supports shared libraries via the library ID method.
9813 This allows for execute in place and shared libraries in an environment
9814 without virtual memory management. This option implies @option{-fPIC}.
9815 With a @samp{bfin-elf} target, this option implies @option{-msim}.
9817 @item -mno-id-shared-library
9818 @opindex mno-id-shared-library
9819 Generate code that doesn't assume ID based shared libraries are being used.
9820 This is the default.
9822 @item -mleaf-id-shared-library
9823 @opindex mleaf-id-shared-library
9824 Generate code that supports shared libraries via the library ID method,
9825 but assumes that this library or executable won't link against any other
9826 ID shared libraries. That allows the compiler to use faster code for jumps
9829 @item -mno-leaf-id-shared-library
9830 @opindex mno-leaf-id-shared-library
9831 Do not assume that the code being compiled won't link against any ID shared
9832 libraries. Slower code will be generated for jump and call insns.
9834 @item -mshared-library-id=n
9835 @opindex mshared-library-id
9836 Specified the identification number of the ID based shared library being
9837 compiled. Specifying a value of 0 will generate more compact code, specifying
9838 other values will force the allocation of that number to the current
9839 library but is no more space or time efficient than omitting this option.
9843 Generate code that allows the data segment to be located in a different
9844 area of memory from the text segment. This allows for execute in place in
9845 an environment without virtual memory management by eliminating relocations
9846 against the text section.
9849 @opindex mno-sep-data
9850 Generate code that assumes that the data segment follows the text segment.
9851 This is the default.
9854 @itemx -mno-long-calls
9855 @opindex mlong-calls
9856 @opindex mno-long-calls
9857 Tells the compiler to perform function calls by first loading the
9858 address of the function into a register and then performing a subroutine
9859 call on this register. This switch is needed if the target function
9860 will lie outside of the 24 bit addressing range of the offset based
9861 version of subroutine call instruction.
9863 This feature is not enabled by default. Specifying
9864 @option{-mno-long-calls} will restore the default behavior. Note these
9865 switches have no effect on how the compiler generates code to handle
9866 function calls via function pointers.
9870 Link with the fast floating-point library. This library relaxes some of
9871 the IEEE floating-point standard's rules for checking inputs against
9872 Not-a-Number (NAN), in the interest of performance.
9875 @opindex minline-plt
9876 Enable inlining of PLT entries in function calls to functions that are
9877 not known to bind locally. It has no effect without @option{-mfdpic}.
9881 Build standalone application for multicore Blackfin processor. Proper
9882 start files and link scripts will be used to support multicore.
9883 This option defines @code{__BFIN_MULTICORE}. It can only be used with
9884 @option{-mcpu=bf561@r{[}-@var{sirevision}@r{]}}. It can be used with
9885 @option{-mcorea} or @option{-mcoreb}. If it's used without
9886 @option{-mcorea} or @option{-mcoreb}, single application/dual core
9887 programming model is used. In this model, the main function of Core B
9888 should be named as coreb_main. If it's used with @option{-mcorea} or
9889 @option{-mcoreb}, one application per core programming model is used.
9890 If this option is not used, single core application programming
9895 Build standalone application for Core A of BF561 when using
9896 one application per core programming model. Proper start files
9897 and link scripts will be used to support Core A. This option
9898 defines @code{__BFIN_COREA}. It must be used with @option{-mmulticore}.
9902 Build standalone application for Core B of BF561 when using
9903 one application per core programming model. Proper start files
9904 and link scripts will be used to support Core B. This option
9905 defines @code{__BFIN_COREB}. When this option is used, coreb_main
9906 should be used instead of main. It must be used with
9907 @option{-mmulticore}.
9911 Build standalone application for SDRAM. Proper start files and
9912 link scripts will be used to put the application into SDRAM.
9913 Loader should initialize SDRAM before loading the application
9914 into SDRAM. This option defines @code{__BFIN_SDRAM}.
9918 Assume that ICPLBs are enabled at runtime. This has an effect on certain
9919 anomaly workarounds. For Linux targets, the default is to assume ICPLBs
9920 are enabled; for standalone applications the default is off.
9924 @subsection CRIS Options
9925 @cindex CRIS Options
9927 These options are defined specifically for the CRIS ports.
9930 @item -march=@var{architecture-type}
9931 @itemx -mcpu=@var{architecture-type}
9934 Generate code for the specified architecture. The choices for
9935 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
9936 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX@.
9937 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
9940 @item -mtune=@var{architecture-type}
9942 Tune to @var{architecture-type} everything applicable about the generated
9943 code, except for the ABI and the set of available instructions. The
9944 choices for @var{architecture-type} are the same as for
9945 @option{-march=@var{architecture-type}}.
9947 @item -mmax-stack-frame=@var{n}
9948 @opindex mmax-stack-frame
9949 Warn when the stack frame of a function exceeds @var{n} bytes.
9955 The options @option{-metrax4} and @option{-metrax100} are synonyms for
9956 @option{-march=v3} and @option{-march=v8} respectively.
9958 @item -mmul-bug-workaround
9959 @itemx -mno-mul-bug-workaround
9960 @opindex mmul-bug-workaround
9961 @opindex mno-mul-bug-workaround
9962 Work around a bug in the @code{muls} and @code{mulu} instructions for CPU
9963 models where it applies. This option is active by default.
9967 Enable CRIS-specific verbose debug-related information in the assembly
9968 code. This option also has the effect to turn off the @samp{#NO_APP}
9969 formatted-code indicator to the assembler at the beginning of the
9974 Do not use condition-code results from previous instruction; always emit
9975 compare and test instructions before use of condition codes.
9977 @item -mno-side-effects
9978 @opindex mno-side-effects
9979 Do not emit instructions with side-effects in addressing modes other than
9983 @itemx -mno-stack-align
9985 @itemx -mno-data-align
9986 @itemx -mconst-align
9987 @itemx -mno-const-align
9988 @opindex mstack-align
9989 @opindex mno-stack-align
9990 @opindex mdata-align
9991 @opindex mno-data-align
9992 @opindex mconst-align
9993 @opindex mno-const-align
9994 These options (no-options) arranges (eliminate arrangements) for the
9995 stack-frame, individual data and constants to be aligned for the maximum
9996 single data access size for the chosen CPU model. The default is to
9997 arrange for 32-bit alignment. ABI details such as structure layout are
9998 not affected by these options.
10006 Similar to the stack- data- and const-align options above, these options
10007 arrange for stack-frame, writable data and constants to all be 32-bit,
10008 16-bit or 8-bit aligned. The default is 32-bit alignment.
10010 @item -mno-prologue-epilogue
10011 @itemx -mprologue-epilogue
10012 @opindex mno-prologue-epilogue
10013 @opindex mprologue-epilogue
10014 With @option{-mno-prologue-epilogue}, the normal function prologue and
10015 epilogue that sets up the stack-frame are omitted and no return
10016 instructions or return sequences are generated in the code. Use this
10017 option only together with visual inspection of the compiled code: no
10018 warnings or errors are generated when call-saved registers must be saved,
10019 or storage for local variable needs to be allocated.
10023 @opindex mno-gotplt
10025 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
10026 instruction sequences that load addresses for functions from the PLT part
10027 of the GOT rather than (traditional on other architectures) calls to the
10028 PLT@. The default is @option{-mgotplt}.
10032 Legacy no-op option only recognized with the cris-axis-elf and
10033 cris-axis-linux-gnu targets.
10037 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
10041 This option, recognized for the cris-axis-elf arranges
10042 to link with input-output functions from a simulator library. Code,
10043 initialized data and zero-initialized data are allocated consecutively.
10047 Like @option{-sim}, but pass linker options to locate initialized data at
10048 0x40000000 and zero-initialized data at 0x80000000.
10052 @subsection CRX Options
10053 @cindex CRX Options
10055 These options are defined specifically for the CRX ports.
10061 Enable the use of multiply-accumulate instructions. Disabled by default.
10064 @opindex mpush-args
10065 Push instructions will be used to pass outgoing arguments when functions
10066 are called. Enabled by default.
10069 @node Darwin Options
10070 @subsection Darwin Options
10071 @cindex Darwin options
10073 These options are defined for all architectures running the Darwin operating
10076 FSF GCC on Darwin does not create ``fat'' object files; it will create
10077 an object file for the single architecture that it was built to
10078 target. Apple's GCC on Darwin does create ``fat'' files if multiple
10079 @option{-arch} options are used; it does so by running the compiler or
10080 linker multiple times and joining the results together with
10083 The subtype of the file created (like @samp{ppc7400} or @samp{ppc970} or
10084 @samp{i686}) is determined by the flags that specify the ISA
10085 that GCC is targetting, like @option{-mcpu} or @option{-march}. The
10086 @option{-force_cpusubtype_ALL} option can be used to override this.
10088 The Darwin tools vary in their behavior when presented with an ISA
10089 mismatch. The assembler, @file{as}, will only permit instructions to
10090 be used that are valid for the subtype of the file it is generating,
10091 so you cannot put 64-bit instructions in a @samp{ppc750} object file.
10092 The linker for shared libraries, @file{/usr/bin/libtool}, will fail
10093 and print an error if asked to create a shared library with a less
10094 restrictive subtype than its input files (for instance, trying to put
10095 a @samp{ppc970} object file in a @samp{ppc7400} library). The linker
10096 for executables, @file{ld}, will quietly give the executable the most
10097 restrictive subtype of any of its input files.
10102 Add the framework directory @var{dir} to the head of the list of
10103 directories to be searched for header files. These directories are
10104 interleaved with those specified by @option{-I} options and are
10105 scanned in a left-to-right order.
10107 A framework directory is a directory with frameworks in it. A
10108 framework is a directory with a @samp{"Headers"} and/or
10109 @samp{"PrivateHeaders"} directory contained directly in it that ends
10110 in @samp{".framework"}. The name of a framework is the name of this
10111 directory excluding the @samp{".framework"}. Headers associated with
10112 the framework are found in one of those two directories, with
10113 @samp{"Headers"} being searched first. A subframework is a framework
10114 directory that is in a framework's @samp{"Frameworks"} directory.
10115 Includes of subframework headers can only appear in a header of a
10116 framework that contains the subframework, or in a sibling subframework
10117 header. Two subframeworks are siblings if they occur in the same
10118 framework. A subframework should not have the same name as a
10119 framework, a warning will be issued if this is violated. Currently a
10120 subframework cannot have subframeworks, in the future, the mechanism
10121 may be extended to support this. The standard frameworks can be found
10122 in @samp{"/System/Library/Frameworks"} and
10123 @samp{"/Library/Frameworks"}. An example include looks like
10124 @code{#include <Framework/header.h>}, where @samp{Framework} denotes
10125 the name of the framework and header.h is found in the
10126 @samp{"PrivateHeaders"} or @samp{"Headers"} directory.
10128 @item -iframework@var{dir}
10129 @opindex iframework
10130 Like @option{-F} except the directory is a treated as a system
10131 directory. The main difference between this @option{-iframework} and
10132 @option{-F} is that with @option{-iframework} the compiler does not
10133 warn about constructs contained within header files found via
10134 @var{dir}. This option is valid only for the C family of languages.
10138 Emit debugging information for symbols that are used. For STABS
10139 debugging format, this enables @option{-feliminate-unused-debug-symbols}.
10140 This is by default ON@.
10144 Emit debugging information for all symbols and types.
10146 @item -mmacosx-version-min=@var{version}
10147 The earliest version of MacOS X that this executable will run on
10148 is @var{version}. Typical values of @var{version} include @code{10.1},
10149 @code{10.2}, and @code{10.3.9}.
10151 If the compiler was built to use the system's headers by default,
10152 then the default for this option is the system version on which the
10153 compiler is running, otherwise the default is to make choices which
10154 are compatible with as many systems and code bases as possible.
10158 Enable kernel development mode. The @option{-mkernel} option sets
10159 @option{-static}, @option{-fno-common}, @option{-fno-cxa-atexit},
10160 @option{-fno-exceptions}, @option{-fno-non-call-exceptions},
10161 @option{-fapple-kext}, @option{-fno-weak} and @option{-fno-rtti} where
10162 applicable. This mode also sets @option{-mno-altivec},
10163 @option{-msoft-float}, @option{-fno-builtin} and
10164 @option{-mlong-branch} for PowerPC targets.
10166 @item -mone-byte-bool
10167 @opindex mone-byte-bool
10168 Override the defaults for @samp{bool} so that @samp{sizeof(bool)==1}.
10169 By default @samp{sizeof(bool)} is @samp{4} when compiling for
10170 Darwin/PowerPC and @samp{1} when compiling for Darwin/x86, so this
10171 option has no effect on x86.
10173 @strong{Warning:} The @option{-mone-byte-bool} switch causes GCC
10174 to generate code that is not binary compatible with code generated
10175 without that switch. Using this switch may require recompiling all
10176 other modules in a program, including system libraries. Use this
10177 switch to conform to a non-default data model.
10179 @item -mfix-and-continue
10180 @itemx -ffix-and-continue
10181 @itemx -findirect-data
10182 @opindex mfix-and-continue
10183 @opindex ffix-and-continue
10184 @opindex findirect-data
10185 Generate code suitable for fast turn around development. Needed to
10186 enable gdb to dynamically load @code{.o} files into already running
10187 programs. @option{-findirect-data} and @option{-ffix-and-continue}
10188 are provided for backwards compatibility.
10192 Loads all members of static archive libraries.
10193 See man ld(1) for more information.
10195 @item -arch_errors_fatal
10196 @opindex arch_errors_fatal
10197 Cause the errors having to do with files that have the wrong architecture
10200 @item -bind_at_load
10201 @opindex bind_at_load
10202 Causes the output file to be marked such that the dynamic linker will
10203 bind all undefined references when the file is loaded or launched.
10207 Produce a Mach-o bundle format file.
10208 See man ld(1) for more information.
10210 @item -bundle_loader @var{executable}
10211 @opindex bundle_loader
10212 This option specifies the @var{executable} that will be loading the build
10213 output file being linked. See man ld(1) for more information.
10216 @opindex dynamiclib
10217 When passed this option, GCC will produce a dynamic library instead of
10218 an executable when linking, using the Darwin @file{libtool} command.
10220 @item -force_cpusubtype_ALL
10221 @opindex force_cpusubtype_ALL
10222 This causes GCC's output file to have the @var{ALL} subtype, instead of
10223 one controlled by the @option{-mcpu} or @option{-march} option.
10225 @item -allowable_client @var{client_name}
10226 @itemx -client_name
10227 @itemx -compatibility_version
10228 @itemx -current_version
10230 @itemx -dependency-file
10232 @itemx -dylinker_install_name
10234 @itemx -exported_symbols_list
10236 @itemx -flat_namespace
10237 @itemx -force_flat_namespace
10238 @itemx -headerpad_max_install_names
10241 @itemx -install_name
10242 @itemx -keep_private_externs
10243 @itemx -multi_module
10244 @itemx -multiply_defined
10245 @itemx -multiply_defined_unused
10247 @itemx -no_dead_strip_inits_and_terms
10248 @itemx -nofixprebinding
10249 @itemx -nomultidefs
10251 @itemx -noseglinkedit
10252 @itemx -pagezero_size
10254 @itemx -prebind_all_twolevel_modules
10255 @itemx -private_bundle
10256 @itemx -read_only_relocs
10258 @itemx -sectobjectsymbols
10262 @itemx -sectobjectsymbols
10265 @itemx -segs_read_only_addr
10266 @itemx -segs_read_write_addr
10267 @itemx -seg_addr_table
10268 @itemx -seg_addr_table_filename
10269 @itemx -seglinkedit
10271 @itemx -segs_read_only_addr
10272 @itemx -segs_read_write_addr
10273 @itemx -single_module
10275 @itemx -sub_library
10276 @itemx -sub_umbrella
10277 @itemx -twolevel_namespace
10280 @itemx -unexported_symbols_list
10281 @itemx -weak_reference_mismatches
10282 @itemx -whatsloaded
10283 @opindex allowable_client
10284 @opindex client_name
10285 @opindex compatibility_version
10286 @opindex current_version
10287 @opindex dead_strip
10288 @opindex dependency-file
10289 @opindex dylib_file
10290 @opindex dylinker_install_name
10292 @opindex exported_symbols_list
10294 @opindex flat_namespace
10295 @opindex force_flat_namespace
10296 @opindex headerpad_max_install_names
10297 @opindex image_base
10299 @opindex install_name
10300 @opindex keep_private_externs
10301 @opindex multi_module
10302 @opindex multiply_defined
10303 @opindex multiply_defined_unused
10304 @opindex noall_load
10305 @opindex no_dead_strip_inits_and_terms
10306 @opindex nofixprebinding
10307 @opindex nomultidefs
10309 @opindex noseglinkedit
10310 @opindex pagezero_size
10312 @opindex prebind_all_twolevel_modules
10313 @opindex private_bundle
10314 @opindex read_only_relocs
10316 @opindex sectobjectsymbols
10319 @opindex sectcreate
10320 @opindex sectobjectsymbols
10323 @opindex segs_read_only_addr
10324 @opindex segs_read_write_addr
10325 @opindex seg_addr_table
10326 @opindex seg_addr_table_filename
10327 @opindex seglinkedit
10329 @opindex segs_read_only_addr
10330 @opindex segs_read_write_addr
10331 @opindex single_module
10333 @opindex sub_library
10334 @opindex sub_umbrella
10335 @opindex twolevel_namespace
10338 @opindex unexported_symbols_list
10339 @opindex weak_reference_mismatches
10340 @opindex whatsloaded
10341 These options are passed to the Darwin linker. The Darwin linker man page
10342 describes them in detail.
10345 @node DEC Alpha Options
10346 @subsection DEC Alpha Options
10348 These @samp{-m} options are defined for the DEC Alpha implementations:
10351 @item -mno-soft-float
10352 @itemx -msoft-float
10353 @opindex mno-soft-float
10354 @opindex msoft-float
10355 Use (do not use) the hardware floating-point instructions for
10356 floating-point operations. When @option{-msoft-float} is specified,
10357 functions in @file{libgcc.a} will be used to perform floating-point
10358 operations. Unless they are replaced by routines that emulate the
10359 floating-point operations, or compiled in such a way as to call such
10360 emulations routines, these routines will issue floating-point
10361 operations. If you are compiling for an Alpha without floating-point
10362 operations, you must ensure that the library is built so as not to call
10365 Note that Alpha implementations without floating-point operations are
10366 required to have floating-point registers.
10369 @itemx -mno-fp-regs
10371 @opindex mno-fp-regs
10372 Generate code that uses (does not use) the floating-point register set.
10373 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
10374 register set is not used, floating point operands are passed in integer
10375 registers as if they were integers and floating-point results are passed
10376 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
10377 so any function with a floating-point argument or return value called by code
10378 compiled with @option{-mno-fp-regs} must also be compiled with that
10381 A typical use of this option is building a kernel that does not use,
10382 and hence need not save and restore, any floating-point registers.
10386 The Alpha architecture implements floating-point hardware optimized for
10387 maximum performance. It is mostly compliant with the IEEE floating
10388 point standard. However, for full compliance, software assistance is
10389 required. This option generates code fully IEEE compliant code
10390 @emph{except} that the @var{inexact-flag} is not maintained (see below).
10391 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
10392 defined during compilation. The resulting code is less efficient but is
10393 able to correctly support denormalized numbers and exceptional IEEE
10394 values such as not-a-number and plus/minus infinity. Other Alpha
10395 compilers call this option @option{-ieee_with_no_inexact}.
10397 @item -mieee-with-inexact
10398 @opindex mieee-with-inexact
10399 This is like @option{-mieee} except the generated code also maintains
10400 the IEEE @var{inexact-flag}. Turning on this option causes the
10401 generated code to implement fully-compliant IEEE math. In addition to
10402 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
10403 macro. On some Alpha implementations the resulting code may execute
10404 significantly slower than the code generated by default. Since there is
10405 very little code that depends on the @var{inexact-flag}, you should
10406 normally not specify this option. Other Alpha compilers call this
10407 option @option{-ieee_with_inexact}.
10409 @item -mfp-trap-mode=@var{trap-mode}
10410 @opindex mfp-trap-mode
10411 This option controls what floating-point related traps are enabled.
10412 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
10413 The trap mode can be set to one of four values:
10417 This is the default (normal) setting. The only traps that are enabled
10418 are the ones that cannot be disabled in software (e.g., division by zero
10422 In addition to the traps enabled by @samp{n}, underflow traps are enabled
10426 Like @samp{u}, but the instructions are marked to be safe for software
10427 completion (see Alpha architecture manual for details).
10430 Like @samp{su}, but inexact traps are enabled as well.
10433 @item -mfp-rounding-mode=@var{rounding-mode}
10434 @opindex mfp-rounding-mode
10435 Selects the IEEE rounding mode. Other Alpha compilers call this option
10436 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
10441 Normal IEEE rounding mode. Floating point numbers are rounded towards
10442 the nearest machine number or towards the even machine number in case
10446 Round towards minus infinity.
10449 Chopped rounding mode. Floating point numbers are rounded towards zero.
10452 Dynamic rounding mode. A field in the floating point control register
10453 (@var{fpcr}, see Alpha architecture reference manual) controls the
10454 rounding mode in effect. The C library initializes this register for
10455 rounding towards plus infinity. Thus, unless your program modifies the
10456 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
10459 @item -mtrap-precision=@var{trap-precision}
10460 @opindex mtrap-precision
10461 In the Alpha architecture, floating point traps are imprecise. This
10462 means without software assistance it is impossible to recover from a
10463 floating trap and program execution normally needs to be terminated.
10464 GCC can generate code that can assist operating system trap handlers
10465 in determining the exact location that caused a floating point trap.
10466 Depending on the requirements of an application, different levels of
10467 precisions can be selected:
10471 Program precision. This option is the default and means a trap handler
10472 can only identify which program caused a floating point exception.
10475 Function precision. The trap handler can determine the function that
10476 caused a floating point exception.
10479 Instruction precision. The trap handler can determine the exact
10480 instruction that caused a floating point exception.
10483 Other Alpha compilers provide the equivalent options called
10484 @option{-scope_safe} and @option{-resumption_safe}.
10486 @item -mieee-conformant
10487 @opindex mieee-conformant
10488 This option marks the generated code as IEEE conformant. You must not
10489 use this option unless you also specify @option{-mtrap-precision=i} and either
10490 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
10491 is to emit the line @samp{.eflag 48} in the function prologue of the
10492 generated assembly file. Under DEC Unix, this has the effect that
10493 IEEE-conformant math library routines will be linked in.
10495 @item -mbuild-constants
10496 @opindex mbuild-constants
10497 Normally GCC examines a 32- or 64-bit integer constant to
10498 see if it can construct it from smaller constants in two or three
10499 instructions. If it cannot, it will output the constant as a literal and
10500 generate code to load it from the data segment at runtime.
10502 Use this option to require GCC to construct @emph{all} integer constants
10503 using code, even if it takes more instructions (the maximum is six).
10505 You would typically use this option to build a shared library dynamic
10506 loader. Itself a shared library, it must relocate itself in memory
10507 before it can find the variables and constants in its own data segment.
10513 Select whether to generate code to be assembled by the vendor-supplied
10514 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
10532 Indicate whether GCC should generate code to use the optional BWX,
10533 CIX, FIX and MAX instruction sets. The default is to use the instruction
10534 sets supported by the CPU type specified via @option{-mcpu=} option or that
10535 of the CPU on which GCC was built if none was specified.
10538 @itemx -mfloat-ieee
10539 @opindex mfloat-vax
10540 @opindex mfloat-ieee
10541 Generate code that uses (does not use) VAX F and G floating point
10542 arithmetic instead of IEEE single and double precision.
10544 @item -mexplicit-relocs
10545 @itemx -mno-explicit-relocs
10546 @opindex mexplicit-relocs
10547 @opindex mno-explicit-relocs
10548 Older Alpha assemblers provided no way to generate symbol relocations
10549 except via assembler macros. Use of these macros does not allow
10550 optimal instruction scheduling. GNU binutils as of version 2.12
10551 supports a new syntax that allows the compiler to explicitly mark
10552 which relocations should apply to which instructions. This option
10553 is mostly useful for debugging, as GCC detects the capabilities of
10554 the assembler when it is built and sets the default accordingly.
10557 @itemx -mlarge-data
10558 @opindex msmall-data
10559 @opindex mlarge-data
10560 When @option{-mexplicit-relocs} is in effect, static data is
10561 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
10562 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
10563 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
10564 16-bit relocations off of the @code{$gp} register. This limits the
10565 size of the small data area to 64KB, but allows the variables to be
10566 directly accessed via a single instruction.
10568 The default is @option{-mlarge-data}. With this option the data area
10569 is limited to just below 2GB@. Programs that require more than 2GB of
10570 data must use @code{malloc} or @code{mmap} to allocate the data in the
10571 heap instead of in the program's data segment.
10573 When generating code for shared libraries, @option{-fpic} implies
10574 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
10577 @itemx -mlarge-text
10578 @opindex msmall-text
10579 @opindex mlarge-text
10580 When @option{-msmall-text} is used, the compiler assumes that the
10581 code of the entire program (or shared library) fits in 4MB, and is
10582 thus reachable with a branch instruction. When @option{-msmall-data}
10583 is used, the compiler can assume that all local symbols share the
10584 same @code{$gp} value, and thus reduce the number of instructions
10585 required for a function call from 4 to 1.
10587 The default is @option{-mlarge-text}.
10589 @item -mcpu=@var{cpu_type}
10591 Set the instruction set and instruction scheduling parameters for
10592 machine type @var{cpu_type}. You can specify either the @samp{EV}
10593 style name or the corresponding chip number. GCC supports scheduling
10594 parameters for the EV4, EV5 and EV6 family of processors and will
10595 choose the default values for the instruction set from the processor
10596 you specify. If you do not specify a processor type, GCC will default
10597 to the processor on which the compiler was built.
10599 Supported values for @var{cpu_type} are
10605 Schedules as an EV4 and has no instruction set extensions.
10609 Schedules as an EV5 and has no instruction set extensions.
10613 Schedules as an EV5 and supports the BWX extension.
10618 Schedules as an EV5 and supports the BWX and MAX extensions.
10622 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
10626 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
10629 Native Linux/GNU toolchains also support the value @samp{native},
10630 which selects the best architecture option for the host processor.
10631 @option{-mcpu=native} has no effect if GCC does not recognize
10634 @item -mtune=@var{cpu_type}
10636 Set only the instruction scheduling parameters for machine type
10637 @var{cpu_type}. The instruction set is not changed.
10639 Native Linux/GNU toolchains also support the value @samp{native},
10640 which selects the best architecture option for the host processor.
10641 @option{-mtune=native} has no effect if GCC does not recognize
10644 @item -mmemory-latency=@var{time}
10645 @opindex mmemory-latency
10646 Sets the latency the scheduler should assume for typical memory
10647 references as seen by the application. This number is highly
10648 dependent on the memory access patterns used by the application
10649 and the size of the external cache on the machine.
10651 Valid options for @var{time} are
10655 A decimal number representing clock cycles.
10661 The compiler contains estimates of the number of clock cycles for
10662 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
10663 (also called Dcache, Scache, and Bcache), as well as to main memory.
10664 Note that L3 is only valid for EV5.
10669 @node DEC Alpha/VMS Options
10670 @subsection DEC Alpha/VMS Options
10672 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
10675 @item -mvms-return-codes
10676 @opindex mvms-return-codes
10677 Return VMS condition codes from main. The default is to return POSIX
10678 style condition (e.g.@: error) codes.
10680 @item -mdebug-main=@var{prefix}
10681 @opindex mdebug-main=@var{prefix}
10682 Flag the first routine whose name starts with @var{prefix} as the main
10683 routine for the debugger.
10687 Default to 64bit memory allocation routines.
10691 @subsection FR30 Options
10692 @cindex FR30 Options
10694 These options are defined specifically for the FR30 port.
10698 @item -msmall-model
10699 @opindex msmall-model
10700 Use the small address space model. This can produce smaller code, but
10701 it does assume that all symbolic values and addresses will fit into a
10706 Assume that run-time support has been provided and so there is no need
10707 to include the simulator library (@file{libsim.a}) on the linker
10713 @subsection FRV Options
10714 @cindex FRV Options
10720 Only use the first 32 general purpose registers.
10725 Use all 64 general purpose registers.
10730 Use only the first 32 floating point registers.
10735 Use all 64 floating point registers
10738 @opindex mhard-float
10740 Use hardware instructions for floating point operations.
10743 @opindex msoft-float
10745 Use library routines for floating point operations.
10750 Dynamically allocate condition code registers.
10755 Do not try to dynamically allocate condition code registers, only
10756 use @code{icc0} and @code{fcc0}.
10761 Change ABI to use double word insns.
10766 Do not use double word instructions.
10771 Use floating point double instructions.
10774 @opindex mno-double
10776 Do not use floating point double instructions.
10781 Use media instructions.
10786 Do not use media instructions.
10791 Use multiply and add/subtract instructions.
10794 @opindex mno-muladd
10796 Do not use multiply and add/subtract instructions.
10801 Select the FDPIC ABI, that uses function descriptors to represent
10802 pointers to functions. Without any PIC/PIE-related options, it
10803 implies @option{-fPIE}. With @option{-fpic} or @option{-fpie}, it
10804 assumes GOT entries and small data are within a 12-bit range from the
10805 GOT base address; with @option{-fPIC} or @option{-fPIE}, GOT offsets
10806 are computed with 32 bits.
10807 With a @samp{bfin-elf} target, this option implies @option{-msim}.
10810 @opindex minline-plt
10812 Enable inlining of PLT entries in function calls to functions that are
10813 not known to bind locally. It has no effect without @option{-mfdpic}.
10814 It's enabled by default if optimizing for speed and compiling for
10815 shared libraries (i.e., @option{-fPIC} or @option{-fpic}), or when an
10816 optimization option such as @option{-O3} or above is present in the
10822 Assume a large TLS segment when generating thread-local code.
10827 Do not assume a large TLS segment when generating thread-local code.
10832 Enable the use of @code{GPREL} relocations in the FDPIC ABI for data
10833 that is known to be in read-only sections. It's enabled by default,
10834 except for @option{-fpic} or @option{-fpie}: even though it may help
10835 make the global offset table smaller, it trades 1 instruction for 4.
10836 With @option{-fPIC} or @option{-fPIE}, it trades 3 instructions for 4,
10837 one of which may be shared by multiple symbols, and it avoids the need
10838 for a GOT entry for the referenced symbol, so it's more likely to be a
10839 win. If it is not, @option{-mno-gprel-ro} can be used to disable it.
10841 @item -multilib-library-pic
10842 @opindex multilib-library-pic
10844 Link with the (library, not FD) pic libraries. It's implied by
10845 @option{-mlibrary-pic}, as well as by @option{-fPIC} and
10846 @option{-fpic} without @option{-mfdpic}. You should never have to use
10850 @opindex mlinked-fp
10852 Follow the EABI requirement of always creating a frame pointer whenever
10853 a stack frame is allocated. This option is enabled by default and can
10854 be disabled with @option{-mno-linked-fp}.
10857 @opindex mlong-calls
10859 Use indirect addressing to call functions outside the current
10860 compilation unit. This allows the functions to be placed anywhere
10861 within the 32-bit address space.
10863 @item -malign-labels
10864 @opindex malign-labels
10866 Try to align labels to an 8-byte boundary by inserting nops into the
10867 previous packet. This option only has an effect when VLIW packing
10868 is enabled. It doesn't create new packets; it merely adds nops to
10871 @item -mlibrary-pic
10872 @opindex mlibrary-pic
10874 Generate position-independent EABI code.
10879 Use only the first four media accumulator registers.
10884 Use all eight media accumulator registers.
10889 Pack VLIW instructions.
10894 Do not pack VLIW instructions.
10897 @opindex mno-eflags
10899 Do not mark ABI switches in e_flags.
10902 @opindex mcond-move
10904 Enable the use of conditional-move instructions (default).
10906 This switch is mainly for debugging the compiler and will likely be removed
10907 in a future version.
10909 @item -mno-cond-move
10910 @opindex mno-cond-move
10912 Disable the use of conditional-move instructions.
10914 This switch is mainly for debugging the compiler and will likely be removed
10915 in a future version.
10920 Enable the use of conditional set instructions (default).
10922 This switch is mainly for debugging the compiler and will likely be removed
10923 in a future version.
10928 Disable the use of conditional set instructions.
10930 This switch is mainly for debugging the compiler and will likely be removed
10931 in a future version.
10934 @opindex mcond-exec
10936 Enable the use of conditional execution (default).
10938 This switch is mainly for debugging the compiler and will likely be removed
10939 in a future version.
10941 @item -mno-cond-exec
10942 @opindex mno-cond-exec
10944 Disable the use of conditional execution.
10946 This switch is mainly for debugging the compiler and will likely be removed
10947 in a future version.
10949 @item -mvliw-branch
10950 @opindex mvliw-branch
10952 Run a pass to pack branches into VLIW instructions (default).
10954 This switch is mainly for debugging the compiler and will likely be removed
10955 in a future version.
10957 @item -mno-vliw-branch
10958 @opindex mno-vliw-branch
10960 Do not run a pass to pack branches into VLIW instructions.
10962 This switch is mainly for debugging the compiler and will likely be removed
10963 in a future version.
10965 @item -mmulti-cond-exec
10966 @opindex mmulti-cond-exec
10968 Enable optimization of @code{&&} and @code{||} in conditional execution
10971 This switch is mainly for debugging the compiler and will likely be removed
10972 in a future version.
10974 @item -mno-multi-cond-exec
10975 @opindex mno-multi-cond-exec
10977 Disable optimization of @code{&&} and @code{||} in conditional execution.
10979 This switch is mainly for debugging the compiler and will likely be removed
10980 in a future version.
10982 @item -mnested-cond-exec
10983 @opindex mnested-cond-exec
10985 Enable nested conditional execution optimizations (default).
10987 This switch is mainly for debugging the compiler and will likely be removed
10988 in a future version.
10990 @item -mno-nested-cond-exec
10991 @opindex mno-nested-cond-exec
10993 Disable nested conditional execution optimizations.
10995 This switch is mainly for debugging the compiler and will likely be removed
10996 in a future version.
10998 @item -moptimize-membar
10999 @opindex moptimize-membar
11001 This switch removes redundant @code{membar} instructions from the
11002 compiler generated code. It is enabled by default.
11004 @item -mno-optimize-membar
11005 @opindex mno-optimize-membar
11007 This switch disables the automatic removal of redundant @code{membar}
11008 instructions from the generated code.
11010 @item -mtomcat-stats
11011 @opindex mtomcat-stats
11013 Cause gas to print out tomcat statistics.
11015 @item -mcpu=@var{cpu}
11018 Select the processor type for which to generate code. Possible values are
11019 @samp{frv}, @samp{fr550}, @samp{tomcat}, @samp{fr500}, @samp{fr450},
11020 @samp{fr405}, @samp{fr400}, @samp{fr300} and @samp{simple}.
11024 @node GNU/Linux Options
11025 @subsection GNU/Linux Options
11027 These @samp{-m} options are defined for GNU/Linux targets:
11032 Use the GNU C library instead of uClibc. This is the default except
11033 on @samp{*-*-linux-*uclibc*} targets.
11037 Use uClibc instead of the GNU C library. This is the default on
11038 @samp{*-*-linux-*uclibc*} targets.
11041 @node H8/300 Options
11042 @subsection H8/300 Options
11044 These @samp{-m} options are defined for the H8/300 implementations:
11049 Shorten some address references at link time, when possible; uses the
11050 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
11051 ld, Using ld}, for a fuller description.
11055 Generate code for the H8/300H@.
11059 Generate code for the H8S@.
11063 Generate code for the H8S and H8/300H in the normal mode. This switch
11064 must be used either with @option{-mh} or @option{-ms}.
11068 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
11072 Make @code{int} data 32 bits by default.
11075 @opindex malign-300
11076 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
11077 The default for the H8/300H and H8S is to align longs and floats on 4
11079 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
11080 This option has no effect on the H8/300.
11084 @subsection HPPA Options
11085 @cindex HPPA Options
11087 These @samp{-m} options are defined for the HPPA family of computers:
11090 @item -march=@var{architecture-type}
11092 Generate code for the specified architecture. The choices for
11093 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
11094 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
11095 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
11096 architecture option for your machine. Code compiled for lower numbered
11097 architectures will run on higher numbered architectures, but not the
11100 @item -mpa-risc-1-0
11101 @itemx -mpa-risc-1-1
11102 @itemx -mpa-risc-2-0
11103 @opindex mpa-risc-1-0
11104 @opindex mpa-risc-1-1
11105 @opindex mpa-risc-2-0
11106 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
11109 @opindex mbig-switch
11110 Generate code suitable for big switch tables. Use this option only if
11111 the assembler/linker complain about out of range branches within a switch
11114 @item -mjump-in-delay
11115 @opindex mjump-in-delay
11116 Fill delay slots of function calls with unconditional jump instructions
11117 by modifying the return pointer for the function call to be the target
11118 of the conditional jump.
11120 @item -mdisable-fpregs
11121 @opindex mdisable-fpregs
11122 Prevent floating point registers from being used in any manner. This is
11123 necessary for compiling kernels which perform lazy context switching of
11124 floating point registers. If you use this option and attempt to perform
11125 floating point operations, the compiler will abort.
11127 @item -mdisable-indexing
11128 @opindex mdisable-indexing
11129 Prevent the compiler from using indexing address modes. This avoids some
11130 rather obscure problems when compiling MIG generated code under MACH@.
11132 @item -mno-space-regs
11133 @opindex mno-space-regs
11134 Generate code that assumes the target has no space registers. This allows
11135 GCC to generate faster indirect calls and use unscaled index address modes.
11137 Such code is suitable for level 0 PA systems and kernels.
11139 @item -mfast-indirect-calls
11140 @opindex mfast-indirect-calls
11141 Generate code that assumes calls never cross space boundaries. This
11142 allows GCC to emit code which performs faster indirect calls.
11144 This option will not work in the presence of shared libraries or nested
11147 @item -mfixed-range=@var{register-range}
11148 @opindex mfixed-range
11149 Generate code treating the given register range as fixed registers.
11150 A fixed register is one that the register allocator can not use. This is
11151 useful when compiling kernel code. A register range is specified as
11152 two registers separated by a dash. Multiple register ranges can be
11153 specified separated by a comma.
11155 @item -mlong-load-store
11156 @opindex mlong-load-store
11157 Generate 3-instruction load and store sequences as sometimes required by
11158 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
11161 @item -mportable-runtime
11162 @opindex mportable-runtime
11163 Use the portable calling conventions proposed by HP for ELF systems.
11167 Enable the use of assembler directives only GAS understands.
11169 @item -mschedule=@var{cpu-type}
11171 Schedule code according to the constraints for the machine type
11172 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
11173 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
11174 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
11175 proper scheduling option for your machine. The default scheduling is
11179 @opindex mlinker-opt
11180 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
11181 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
11182 linkers in which they give bogus error messages when linking some programs.
11185 @opindex msoft-float
11186 Generate output containing library calls for floating point.
11187 @strong{Warning:} the requisite libraries are not available for all HPPA
11188 targets. Normally the facilities of the machine's usual C compiler are
11189 used, but this cannot be done directly in cross-compilation. You must make
11190 your own arrangements to provide suitable library functions for
11193 @option{-msoft-float} changes the calling convention in the output file;
11194 therefore, it is only useful if you compile @emph{all} of a program with
11195 this option. In particular, you need to compile @file{libgcc.a}, the
11196 library that comes with GCC, with @option{-msoft-float} in order for
11201 Generate the predefine, @code{_SIO}, for server IO@. The default is
11202 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
11203 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO@. These
11204 options are available under HP-UX and HI-UX@.
11208 Use GNU ld specific options. This passes @option{-shared} to ld when
11209 building a shared library. It is the default when GCC is configured,
11210 explicitly or implicitly, with the GNU linker. This option does not
11211 have any affect on which ld is called, it only changes what parameters
11212 are passed to that ld. The ld that is called is determined by the
11213 @option{--with-ld} configure option, GCC's program search path, and
11214 finally by the user's @env{PATH}. The linker used by GCC can be printed
11215 using @samp{which `gcc -print-prog-name=ld`}. This option is only available
11216 on the 64 bit HP-UX GCC, i.e.@: configured with @samp{hppa*64*-*-hpux*}.
11220 Use HP ld specific options. This passes @option{-b} to ld when building
11221 a shared library and passes @option{+Accept TypeMismatch} to ld on all
11222 links. It is the default when GCC is configured, explicitly or
11223 implicitly, with the HP linker. This option does not have any affect on
11224 which ld is called, it only changes what parameters are passed to that
11225 ld. The ld that is called is determined by the @option{--with-ld}
11226 configure option, GCC's program search path, and finally by the user's
11227 @env{PATH}. The linker used by GCC can be printed using @samp{which
11228 `gcc -print-prog-name=ld`}. This option is only available on the 64 bit
11229 HP-UX GCC, i.e.@: configured with @samp{hppa*64*-*-hpux*}.
11232 @opindex mno-long-calls
11233 Generate code that uses long call sequences. This ensures that a call
11234 is always able to reach linker generated stubs. The default is to generate
11235 long calls only when the distance from the call site to the beginning
11236 of the function or translation unit, as the case may be, exceeds a
11237 predefined limit set by the branch type being used. The limits for
11238 normal calls are 7,600,000 and 240,000 bytes, respectively for the
11239 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
11242 Distances are measured from the beginning of functions when using the
11243 @option{-ffunction-sections} option, or when using the @option{-mgas}
11244 and @option{-mno-portable-runtime} options together under HP-UX with
11247 It is normally not desirable to use this option as it will degrade
11248 performance. However, it may be useful in large applications,
11249 particularly when partial linking is used to build the application.
11251 The types of long calls used depends on the capabilities of the
11252 assembler and linker, and the type of code being generated. The
11253 impact on systems that support long absolute calls, and long pic
11254 symbol-difference or pc-relative calls should be relatively small.
11255 However, an indirect call is used on 32-bit ELF systems in pic code
11256 and it is quite long.
11258 @item -munix=@var{unix-std}
11260 Generate compiler predefines and select a startfile for the specified
11261 UNIX standard. The choices for @var{unix-std} are @samp{93}, @samp{95}
11262 and @samp{98}. @samp{93} is supported on all HP-UX versions. @samp{95}
11263 is available on HP-UX 10.10 and later. @samp{98} is available on HP-UX
11264 11.11 and later. The default values are @samp{93} for HP-UX 10.00,
11265 @samp{95} for HP-UX 10.10 though to 11.00, and @samp{98} for HP-UX 11.11
11268 @option{-munix=93} provides the same predefines as GCC 3.3 and 3.4.
11269 @option{-munix=95} provides additional predefines for @code{XOPEN_UNIX}
11270 and @code{_XOPEN_SOURCE_EXTENDED}, and the startfile @file{unix95.o}.
11271 @option{-munix=98} provides additional predefines for @code{_XOPEN_UNIX},
11272 @code{_XOPEN_SOURCE_EXTENDED}, @code{_INCLUDE__STDC_A1_SOURCE} and
11273 @code{_INCLUDE_XOPEN_SOURCE_500}, and the startfile @file{unix98.o}.
11275 It is @emph{important} to note that this option changes the interfaces
11276 for various library routines. It also affects the operational behavior
11277 of the C library. Thus, @emph{extreme} care is needed in using this
11280 Library code that is intended to operate with more than one UNIX
11281 standard must test, set and restore the variable @var{__xpg4_extended_mask}
11282 as appropriate. Most GNU software doesn't provide this capability.
11286 Suppress the generation of link options to search libdld.sl when the
11287 @option{-static} option is specified on HP-UX 10 and later.
11291 The HP-UX implementation of setlocale in libc has a dependency on
11292 libdld.sl. There isn't an archive version of libdld.sl. Thus,
11293 when the @option{-static} option is specified, special link options
11294 are needed to resolve this dependency.
11296 On HP-UX 10 and later, the GCC driver adds the necessary options to
11297 link with libdld.sl when the @option{-static} option is specified.
11298 This causes the resulting binary to be dynamic. On the 64-bit port,
11299 the linkers generate dynamic binaries by default in any case. The
11300 @option{-nolibdld} option can be used to prevent the GCC driver from
11301 adding these link options.
11305 Add support for multithreading with the @dfn{dce thread} library
11306 under HP-UX@. This option sets flags for both the preprocessor and
11310 @node i386 and x86-64 Options
11311 @subsection Intel 386 and AMD x86-64 Options
11312 @cindex i386 Options
11313 @cindex x86-64 Options
11314 @cindex Intel 386 Options
11315 @cindex AMD x86-64 Options
11317 These @samp{-m} options are defined for the i386 and x86-64 family of
11321 @item -mtune=@var{cpu-type}
11323 Tune to @var{cpu-type} everything applicable about the generated code, except
11324 for the ABI and the set of available instructions. The choices for
11325 @var{cpu-type} are:
11328 Produce code optimized for the most common IA32/AMD64/EM64T processors.
11329 If you know the CPU on which your code will run, then you should use
11330 the corresponding @option{-mtune} option instead of
11331 @option{-mtune=generic}. But, if you do not know exactly what CPU users
11332 of your application will have, then you should use this option.
11334 As new processors are deployed in the marketplace, the behavior of this
11335 option will change. Therefore, if you upgrade to a newer version of
11336 GCC, the code generated option will change to reflect the processors
11337 that were most common when that version of GCC was released.
11339 There is no @option{-march=generic} option because @option{-march}
11340 indicates the instruction set the compiler can use, and there is no
11341 generic instruction set applicable to all processors. In contrast,
11342 @option{-mtune} indicates the processor (or, in this case, collection of
11343 processors) for which the code is optimized.
11345 This selects the CPU to tune for at compilation time by determining
11346 the processor type of the compiling machine. Using @option{-mtune=native}
11347 will produce code optimized for the local machine under the constraints
11348 of the selected instruction set. Using @option{-march=native} will
11349 enable all instruction subsets supported by the local machine (hence
11350 the result might not run on different machines).
11352 Original Intel's i386 CPU@.
11354 Intel's i486 CPU@. (No scheduling is implemented for this chip.)
11355 @item i586, pentium
11356 Intel Pentium CPU with no MMX support.
11358 Intel PentiumMMX CPU based on Pentium core with MMX instruction set support.
11360 Intel PentiumPro CPU@.
11362 Same as @code{generic}, but when used as @code{march} option, PentiumPro
11363 instruction set will be used, so the code will run on all i686 family chips.
11365 Intel Pentium2 CPU based on PentiumPro core with MMX instruction set support.
11366 @item pentium3, pentium3m
11367 Intel Pentium3 CPU based on PentiumPro core with MMX and SSE instruction set
11370 Low power version of Intel Pentium3 CPU with MMX, SSE and SSE2 instruction set
11371 support. Used by Centrino notebooks.
11372 @item pentium4, pentium4m
11373 Intel Pentium4 CPU with MMX, SSE and SSE2 instruction set support.
11375 Improved version of Intel Pentium4 CPU with MMX, SSE, SSE2 and SSE3 instruction
11378 Improved version of Intel Pentium4 CPU with 64-bit extensions, MMX, SSE,
11379 SSE2 and SSE3 instruction set support.
11381 Intel Core2 CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3 and SSSE3
11382 instruction set support.
11384 Intel Atom CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3 and SSSE3
11385 instruction set support.
11387 AMD K6 CPU with MMX instruction set support.
11389 Improved versions of AMD K6 CPU with MMX and 3dNOW!@: instruction set support.
11390 @item athlon, athlon-tbird
11391 AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW!@: and SSE prefetch instructions
11393 @item athlon-4, athlon-xp, athlon-mp
11394 Improved AMD Athlon CPU with MMX, 3dNOW!, enhanced 3dNOW!@: and full SSE
11395 instruction set support.
11396 @item k8, opteron, athlon64, athlon-fx
11397 AMD K8 core based CPUs with x86-64 instruction set support. (This supersets
11398 MMX, SSE, SSE2, 3dNOW!, enhanced 3dNOW!@: and 64-bit instruction set extensions.)
11399 @item k8-sse3, opteron-sse3, athlon64-sse3
11400 Improved versions of k8, opteron and athlon64 with SSE3 instruction set support.
11401 @item amdfam10, barcelona
11402 AMD Family 10h core based CPUs with x86-64 instruction set support. (This
11403 supersets MMX, SSE, SSE2, SSE3, SSE4A, 3dNOW!, enhanced 3dNOW!, ABM and 64-bit
11404 instruction set extensions.)
11406 IDT Winchip C6 CPU, dealt in same way as i486 with additional MMX instruction
11409 IDT Winchip2 CPU, dealt in same way as i486 with additional MMX and 3dNOW!@:
11410 instruction set support.
11412 Via C3 CPU with MMX and 3dNOW!@: instruction set support. (No scheduling is
11413 implemented for this chip.)
11415 Via C3-2 CPU with MMX and SSE instruction set support. (No scheduling is
11416 implemented for this chip.)
11418 Embedded AMD CPU with MMX and 3dNOW! instruction set support.
11421 While picking a specific @var{cpu-type} will schedule things appropriately
11422 for that particular chip, the compiler will not generate any code that
11423 does not run on the i386 without the @option{-march=@var{cpu-type}} option
11426 @item -march=@var{cpu-type}
11428 Generate instructions for the machine type @var{cpu-type}. The choices
11429 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
11430 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
11432 @item -mcpu=@var{cpu-type}
11434 A deprecated synonym for @option{-mtune}.
11436 @item -mfpmath=@var{unit}
11438 Generate floating point arithmetics for selected unit @var{unit}. The choices
11439 for @var{unit} are:
11443 Use the standard 387 floating point coprocessor present majority of chips and
11444 emulated otherwise. Code compiled with this option will run almost everywhere.
11445 The temporary results are computed in 80bit precision instead of precision
11446 specified by the type resulting in slightly different results compared to most
11447 of other chips. See @option{-ffloat-store} for more detailed description.
11449 This is the default choice for i386 compiler.
11452 Use scalar floating point instructions present in the SSE instruction set.
11453 This instruction set is supported by Pentium3 and newer chips, in the AMD line
11454 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
11455 instruction set supports only single precision arithmetics, thus the double and
11456 extended precision arithmetics is still done using 387. Later version, present
11457 only in Pentium4 and the future AMD x86-64 chips supports double precision
11460 For the i386 compiler, you need to use @option{-march=@var{cpu-type}}, @option{-msse}
11461 or @option{-msse2} switches to enable SSE extensions and make this option
11462 effective. For the x86-64 compiler, these extensions are enabled by default.
11464 The resulting code should be considerably faster in the majority of cases and avoid
11465 the numerical instability problems of 387 code, but may break some existing
11466 code that expects temporaries to be 80bit.
11468 This is the default choice for the x86-64 compiler.
11473 Attempt to utilize both instruction sets at once. This effectively double the
11474 amount of available registers and on chips with separate execution units for
11475 387 and SSE the execution resources too. Use this option with care, as it is
11476 still experimental, because the GCC register allocator does not model separate
11477 functional units well resulting in instable performance.
11480 @item -masm=@var{dialect}
11481 @opindex masm=@var{dialect}
11482 Output asm instructions using selected @var{dialect}. Supported
11483 choices are @samp{intel} or @samp{att} (the default one). Darwin does
11484 not support @samp{intel}.
11487 @itemx -mno-ieee-fp
11489 @opindex mno-ieee-fp
11490 Control whether or not the compiler uses IEEE floating point
11491 comparisons. These handle correctly the case where the result of a
11492 comparison is unordered.
11495 @opindex msoft-float
11496 Generate output containing library calls for floating point.
11497 @strong{Warning:} the requisite libraries are not part of GCC@.
11498 Normally the facilities of the machine's usual C compiler are used, but
11499 this can't be done directly in cross-compilation. You must make your
11500 own arrangements to provide suitable library functions for
11503 On machines where a function returns floating point results in the 80387
11504 register stack, some floating point opcodes may be emitted even if
11505 @option{-msoft-float} is used.
11507 @item -mno-fp-ret-in-387
11508 @opindex mno-fp-ret-in-387
11509 Do not use the FPU registers for return values of functions.
11511 The usual calling convention has functions return values of types
11512 @code{float} and @code{double} in an FPU register, even if there
11513 is no FPU@. The idea is that the operating system should emulate
11516 The option @option{-mno-fp-ret-in-387} causes such values to be returned
11517 in ordinary CPU registers instead.
11519 @item -mno-fancy-math-387
11520 @opindex mno-fancy-math-387
11521 Some 387 emulators do not support the @code{sin}, @code{cos} and
11522 @code{sqrt} instructions for the 387. Specify this option to avoid
11523 generating those instructions. This option is the default on FreeBSD,
11524 OpenBSD and NetBSD@. This option is overridden when @option{-march}
11525 indicates that the target cpu will always have an FPU and so the
11526 instruction will not need emulation. As of revision 2.6.1, these
11527 instructions are not generated unless you also use the
11528 @option{-funsafe-math-optimizations} switch.
11530 @item -malign-double
11531 @itemx -mno-align-double
11532 @opindex malign-double
11533 @opindex mno-align-double
11534 Control whether GCC aligns @code{double}, @code{long double}, and
11535 @code{long long} variables on a two word boundary or a one word
11536 boundary. Aligning @code{double} variables on a two word boundary will
11537 produce code that runs somewhat faster on a @samp{Pentium} at the
11538 expense of more memory.
11540 On x86-64, @option{-malign-double} is enabled by default.
11542 @strong{Warning:} if you use the @option{-malign-double} switch,
11543 structures containing the above types will be aligned differently than
11544 the published application binary interface specifications for the 386
11545 and will not be binary compatible with structures in code compiled
11546 without that switch.
11548 @item -m96bit-long-double
11549 @itemx -m128bit-long-double
11550 @opindex m96bit-long-double
11551 @opindex m128bit-long-double
11552 These switches control the size of @code{long double} type. The i386
11553 application binary interface specifies the size to be 96 bits,
11554 so @option{-m96bit-long-double} is the default in 32 bit mode.
11556 Modern architectures (Pentium and newer) would prefer @code{long double}
11557 to be aligned to an 8 or 16 byte boundary. In arrays or structures
11558 conforming to the ABI, this would not be possible. So specifying a
11559 @option{-m128bit-long-double} will align @code{long double}
11560 to a 16 byte boundary by padding the @code{long double} with an additional
11563 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
11564 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
11566 Notice that neither of these options enable any extra precision over the x87
11567 standard of 80 bits for a @code{long double}.
11569 @strong{Warning:} if you override the default value for your target ABI, the
11570 structures and arrays containing @code{long double} variables will change
11571 their size as well as function calling convention for function taking
11572 @code{long double} will be modified. Hence they will not be binary
11573 compatible with arrays or structures in code compiled without that switch.
11575 @item -mlarge-data-threshold=@var{number}
11576 @opindex mlarge-data-threshold=@var{number}
11577 When @option{-mcmodel=medium} is specified, the data greater than
11578 @var{threshold} are placed in large data section. This value must be the
11579 same across all object linked into the binary and defaults to 65535.
11583 Use a different function-calling convention, in which functions that
11584 take a fixed number of arguments return with the @code{ret} @var{num}
11585 instruction, which pops their arguments while returning. This saves one
11586 instruction in the caller since there is no need to pop the arguments
11589 You can specify that an individual function is called with this calling
11590 sequence with the function attribute @samp{stdcall}. You can also
11591 override the @option{-mrtd} option by using the function attribute
11592 @samp{cdecl}. @xref{Function Attributes}.
11594 @strong{Warning:} this calling convention is incompatible with the one
11595 normally used on Unix, so you cannot use it if you need to call
11596 libraries compiled with the Unix compiler.
11598 Also, you must provide function prototypes for all functions that
11599 take variable numbers of arguments (including @code{printf});
11600 otherwise incorrect code will be generated for calls to those
11603 In addition, seriously incorrect code will result if you call a
11604 function with too many arguments. (Normally, extra arguments are
11605 harmlessly ignored.)
11607 @item -mregparm=@var{num}
11609 Control how many registers are used to pass integer arguments. By
11610 default, no registers are used to pass arguments, and at most 3
11611 registers can be used. You can control this behavior for a specific
11612 function by using the function attribute @samp{regparm}.
11613 @xref{Function Attributes}.
11615 @strong{Warning:} if you use this switch, and
11616 @var{num} is nonzero, then you must build all modules with the same
11617 value, including any libraries. This includes the system libraries and
11621 @opindex msseregparm
11622 Use SSE register passing conventions for float and double arguments
11623 and return values. You can control this behavior for a specific
11624 function by using the function attribute @samp{sseregparm}.
11625 @xref{Function Attributes}.
11627 @strong{Warning:} if you use this switch then you must build all
11628 modules with the same value, including any libraries. This includes
11629 the system libraries and startup modules.
11638 Set 80387 floating-point precision to 32, 64 or 80 bits. When @option{-mpc32}
11639 is specified, the significands of results of floating-point operations are
11640 rounded to 24 bits (single precision); @option{-mpc64} rounds the
11641 significands of results of floating-point operations to 53 bits (double
11642 precision) and @option{-mpc80} rounds the significands of results of
11643 floating-point operations to 64 bits (extended double precision), which is
11644 the default. When this option is used, floating-point operations in higher
11645 precisions are not available to the programmer without setting the FPU
11646 control word explicitly.
11648 Setting the rounding of floating-point operations to less than the default
11649 80 bits can speed some programs by 2% or more. Note that some mathematical
11650 libraries assume that extended precision (80 bit) floating-point operations
11651 are enabled by default; routines in such libraries could suffer significant
11652 loss of accuracy, typically through so-called "catastrophic cancellation",
11653 when this option is used to set the precision to less than extended precision.
11655 @item -mstackrealign
11656 @opindex mstackrealign
11657 Realign the stack at entry. On the Intel x86, the @option{-mstackrealign}
11658 option will generate an alternate prologue and epilogue that realigns the
11659 runtime stack if necessary. This supports mixing legacy codes that keep
11660 a 4-byte aligned stack with modern codes that keep a 16-byte stack for
11661 SSE compatibility. See also the attribute @code{force_align_arg_pointer},
11662 applicable to individual functions.
11664 @item -mpreferred-stack-boundary=@var{num}
11665 @opindex mpreferred-stack-boundary
11666 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
11667 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
11668 the default is 4 (16 bytes or 128 bits).
11670 @item -mincoming-stack-boundary=@var{num}
11671 @opindex mincoming-stack-boundary
11672 Assume the incoming stack is aligned to a 2 raised to @var{num} byte
11673 boundary. If @option{-mincoming-stack-boundary} is not specified,
11674 the one specified by @option{-mpreferred-stack-boundary} will be used.
11676 On Pentium and PentiumPro, @code{double} and @code{long double} values
11677 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
11678 suffer significant run time performance penalties. On Pentium III, the
11679 Streaming SIMD Extension (SSE) data type @code{__m128} may not work
11680 properly if it is not 16 byte aligned.
11682 To ensure proper alignment of this values on the stack, the stack boundary
11683 must be as aligned as that required by any value stored on the stack.
11684 Further, every function must be generated such that it keeps the stack
11685 aligned. Thus calling a function compiled with a higher preferred
11686 stack boundary from a function compiled with a lower preferred stack
11687 boundary will most likely misalign the stack. It is recommended that
11688 libraries that use callbacks always use the default setting.
11690 This extra alignment does consume extra stack space, and generally
11691 increases code size. Code that is sensitive to stack space usage, such
11692 as embedded systems and operating system kernels, may want to reduce the
11693 preferred alignment to @option{-mpreferred-stack-boundary=2}.
11731 These switches enable or disable the use of instructions in the MMX,
11732 SSE, SSE2, SSE3, SSSE3, SSE4.1, AVX, AES, PCLMUL, SSE4A, ABM or
11733 3DNow!@: extended instruction sets.
11734 These extensions are also available as built-in functions: see
11735 @ref{X86 Built-in Functions}, for details of the functions enabled and
11736 disabled by these switches.
11738 To have SSE/SSE2 instructions generated automatically from floating-point
11739 code (as opposed to 387 instructions), see @option{-mfpmath=sse}.
11741 GCC depresses SSEx instructions when @option{-mavx} is used. Instead, it
11742 generates new AVX instructions or AVX equivalence for all SSEx instructions
11745 These options will enable GCC to use these extended instructions in
11746 generated code, even without @option{-mfpmath=sse}. Applications which
11747 perform runtime CPU detection must compile separate files for each
11748 supported architecture, using the appropriate flags. In particular,
11749 the file containing the CPU detection code should be compiled without
11754 This option instructs GCC to emit a @code{cld} instruction in the prologue
11755 of functions that use string instructions. String instructions depend on
11756 the DF flag to select between autoincrement or autodecrement mode. While the
11757 ABI specifies the DF flag to be cleared on function entry, some operating
11758 systems violate this specification by not clearing the DF flag in their
11759 exception dispatchers. The exception handler can be invoked with the DF flag
11760 set which leads to wrong direction mode, when string instructions are used.
11761 This option can be enabled by default on 32-bit x86 targets by configuring
11762 GCC with the @option{--enable-cld} configure option. Generation of @code{cld}
11763 instructions can be suppressed with the @option{-mno-cld} compiler option
11768 This option will enable GCC to use CMPXCHG16B instruction in generated code.
11769 CMPXCHG16B allows for atomic operations on 128-bit double quadword (or oword)
11770 data types. This is useful for high resolution counters that could be updated
11771 by multiple processors (or cores). This instruction is generated as part of
11772 atomic built-in functions: see @ref{Atomic Builtins} for details.
11776 This option will enable GCC to use SAHF instruction in generated 64-bit code.
11777 Early Intel CPUs with Intel 64 lacked LAHF and SAHF instructions supported
11778 by AMD64 until introduction of Pentium 4 G1 step in December 2005. LAHF and
11779 SAHF are load and store instructions, respectively, for certain status flags.
11780 In 64-bit mode, SAHF instruction is used to optimize @code{fmod}, @code{drem}
11781 or @code{remainder} built-in functions: see @ref{Other Builtins} for details.
11785 This option will enable GCC to use movbe instruction to implement
11786 @code{__builtin_bswap32} and @code{__builtin_bswap64}.
11790 This option will enable built-in functions, @code{__builtin_ia32_crc32qi},
11791 @code{__builtin_ia32_crc32hi}. @code{__builtin_ia32_crc32si} and
11792 @code{__builtin_ia32_crc32di} to generate the crc32 machine instruction.
11796 This option will enable GCC to use RCPSS and RSQRTSS instructions (and their
11797 vectorized variants RCPPS and RSQRTPS) with an additional Newton-Raphson step
11798 to increase precision instead of DIVSS and SQRTSS (and their vectorized
11799 variants) for single precision floating point arguments. These instructions
11800 are generated only when @option{-funsafe-math-optimizations} is enabled
11801 together with @option{-finite-math-only} and @option{-fno-trapping-math}.
11802 Note that while the throughput of the sequence is higher than the throughput
11803 of the non-reciprocal instruction, the precision of the sequence can be
11804 decreased by up to 2 ulp (i.e. the inverse of 1.0 equals 0.99999994).
11806 @item -mveclibabi=@var{type}
11807 @opindex mveclibabi
11808 Specifies the ABI type to use for vectorizing intrinsics using an
11809 external library. Supported types are @code{svml} for the Intel short
11810 vector math library and @code{acml} for the AMD math core library style
11811 of interfacing. GCC will currently emit calls to @code{vmldExp2},
11812 @code{vmldLn2}, @code{vmldLog102}, @code{vmldLog102}, @code{vmldPow2},
11813 @code{vmldTanh2}, @code{vmldTan2}, @code{vmldAtan2}, @code{vmldAtanh2},
11814 @code{vmldCbrt2}, @code{vmldSinh2}, @code{vmldSin2}, @code{vmldAsinh2},
11815 @code{vmldAsin2}, @code{vmldCosh2}, @code{vmldCos2}, @code{vmldAcosh2},
11816 @code{vmldAcos2}, @code{vmlsExp4}, @code{vmlsLn4}, @code{vmlsLog104},
11817 @code{vmlsLog104}, @code{vmlsPow4}, @code{vmlsTanh4}, @code{vmlsTan4},
11818 @code{vmlsAtan4}, @code{vmlsAtanh4}, @code{vmlsCbrt4}, @code{vmlsSinh4},
11819 @code{vmlsSin4}, @code{vmlsAsinh4}, @code{vmlsAsin4}, @code{vmlsCosh4},
11820 @code{vmlsCos4}, @code{vmlsAcosh4} and @code{vmlsAcos4} for corresponding
11821 function type when @option{-mveclibabi=svml} is used and @code{__vrd2_sin},
11822 @code{__vrd2_cos}, @code{__vrd2_exp}, @code{__vrd2_log}, @code{__vrd2_log2},
11823 @code{__vrd2_log10}, @code{__vrs4_sinf}, @code{__vrs4_cosf},
11824 @code{__vrs4_expf}, @code{__vrs4_logf}, @code{__vrs4_log2f},
11825 @code{__vrs4_log10f} and @code{__vrs4_powf} for corresponding function type
11826 when @option{-mveclibabi=acml} is used. Both @option{-ftree-vectorize} and
11827 @option{-funsafe-math-optimizations} have to be enabled. A SVML or ACML ABI
11828 compatible library will have to be specified at link time.
11830 @item -mabi=@var{name}
11832 Generate code for the specified calling convention. Permissible values
11833 are: @samp{sysv} for the ABI used on GNU/Linux and other systems and
11834 @samp{ms} for the Microsoft ABI. The default is to use the Microsoft
11835 ABI when targeting Windows. On all other systems, the default is the
11836 SYSV ABI. You can control this behavior for a specific function by
11837 using the function attribute @samp{ms_abi}/@samp{sysv_abi}.
11838 @xref{Function Attributes}.
11841 @itemx -mno-push-args
11842 @opindex mpush-args
11843 @opindex mno-push-args
11844 Use PUSH operations to store outgoing parameters. This method is shorter
11845 and usually equally fast as method using SUB/MOV operations and is enabled
11846 by default. In some cases disabling it may improve performance because of
11847 improved scheduling and reduced dependencies.
11849 @item -maccumulate-outgoing-args
11850 @opindex maccumulate-outgoing-args
11851 If enabled, the maximum amount of space required for outgoing arguments will be
11852 computed in the function prologue. This is faster on most modern CPUs
11853 because of reduced dependencies, improved scheduling and reduced stack usage
11854 when preferred stack boundary is not equal to 2. The drawback is a notable
11855 increase in code size. This switch implies @option{-mno-push-args}.
11859 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
11860 on thread-safe exception handling must compile and link all code with the
11861 @option{-mthreads} option. When compiling, @option{-mthreads} defines
11862 @option{-D_MT}; when linking, it links in a special thread helper library
11863 @option{-lmingwthrd} which cleans up per thread exception handling data.
11865 @item -mno-align-stringops
11866 @opindex mno-align-stringops
11867 Do not align destination of inlined string operations. This switch reduces
11868 code size and improves performance in case the destination is already aligned,
11869 but GCC doesn't know about it.
11871 @item -minline-all-stringops
11872 @opindex minline-all-stringops
11873 By default GCC inlines string operations only when destination is known to be
11874 aligned at least to 4 byte boundary. This enables more inlining, increase code
11875 size, but may improve performance of code that depends on fast memcpy, strlen
11876 and memset for short lengths.
11878 @item -minline-stringops-dynamically
11879 @opindex minline-stringops-dynamically
11880 For string operation of unknown size, inline runtime checks so for small
11881 blocks inline code is used, while for large blocks library call is used.
11883 @item -mstringop-strategy=@var{alg}
11884 @opindex mstringop-strategy=@var{alg}
11885 Overwrite internal decision heuristic about particular algorithm to inline
11886 string operation with. The allowed values are @code{rep_byte},
11887 @code{rep_4byte}, @code{rep_8byte} for expanding using i386 @code{rep} prefix
11888 of specified size, @code{byte_loop}, @code{loop}, @code{unrolled_loop} for
11889 expanding inline loop, @code{libcall} for always expanding library call.
11891 @item -momit-leaf-frame-pointer
11892 @opindex momit-leaf-frame-pointer
11893 Don't keep the frame pointer in a register for leaf functions. This
11894 avoids the instructions to save, set up and restore frame pointers and
11895 makes an extra register available in leaf functions. The option
11896 @option{-fomit-frame-pointer} removes the frame pointer for all functions
11897 which might make debugging harder.
11899 @item -mtls-direct-seg-refs
11900 @itemx -mno-tls-direct-seg-refs
11901 @opindex mtls-direct-seg-refs
11902 Controls whether TLS variables may be accessed with offsets from the
11903 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
11904 or whether the thread base pointer must be added. Whether or not this
11905 is legal depends on the operating system, and whether it maps the
11906 segment to cover the entire TLS area.
11908 For systems that use GNU libc, the default is on.
11911 @itemx -mno-sse2avx
11913 Specify that the assembler should encode SSE instructions with VEX
11914 prefix. The option @option{-mavx} turns this on by default.
11917 These @samp{-m} switches are supported in addition to the above
11918 on AMD x86-64 processors in 64-bit environments.
11925 Generate code for a 32-bit or 64-bit environment.
11926 The 32-bit environment sets int, long and pointer to 32 bits and
11927 generates code that runs on any i386 system.
11928 The 64-bit environment sets int to 32 bits and long and pointer
11929 to 64 bits and generates code for AMD's x86-64 architecture. For
11930 darwin only the -m64 option turns off the @option{-fno-pic} and
11931 @option{-mdynamic-no-pic} options.
11933 @item -mno-red-zone
11934 @opindex mno-red-zone
11935 Do not use a so called red zone for x86-64 code. The red zone is mandated
11936 by the x86-64 ABI, it is a 128-byte area beyond the location of the
11937 stack pointer that will not be modified by signal or interrupt handlers
11938 and therefore can be used for temporary data without adjusting the stack
11939 pointer. The flag @option{-mno-red-zone} disables this red zone.
11941 @item -mcmodel=small
11942 @opindex mcmodel=small
11943 Generate code for the small code model: the program and its symbols must
11944 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
11945 Programs can be statically or dynamically linked. This is the default
11948 @item -mcmodel=kernel
11949 @opindex mcmodel=kernel
11950 Generate code for the kernel code model. The kernel runs in the
11951 negative 2 GB of the address space.
11952 This model has to be used for Linux kernel code.
11954 @item -mcmodel=medium
11955 @opindex mcmodel=medium
11956 Generate code for the medium model: The program is linked in the lower 2
11957 GB of the address space. Small symbols are also placed there. Symbols
11958 with sizes larger than @option{-mlarge-data-threshold} are put into
11959 large data or bss sections and can be located above 2GB. Programs can
11960 be statically or dynamically linked.
11962 @item -mcmodel=large
11963 @opindex mcmodel=large
11964 Generate code for the large model: This model makes no assumptions
11965 about addresses and sizes of sections.
11968 @node IA-64 Options
11969 @subsection IA-64 Options
11970 @cindex IA-64 Options
11972 These are the @samp{-m} options defined for the Intel IA-64 architecture.
11976 @opindex mbig-endian
11977 Generate code for a big endian target. This is the default for HP-UX@.
11979 @item -mlittle-endian
11980 @opindex mlittle-endian
11981 Generate code for a little endian target. This is the default for AIX5
11987 @opindex mno-gnu-as
11988 Generate (or don't) code for the GNU assembler. This is the default.
11989 @c Also, this is the default if the configure option @option{--with-gnu-as}
11995 @opindex mno-gnu-ld
11996 Generate (or don't) code for the GNU linker. This is the default.
11997 @c Also, this is the default if the configure option @option{--with-gnu-ld}
12002 Generate code that does not use a global pointer register. The result
12003 is not position independent code, and violates the IA-64 ABI@.
12005 @item -mvolatile-asm-stop
12006 @itemx -mno-volatile-asm-stop
12007 @opindex mvolatile-asm-stop
12008 @opindex mno-volatile-asm-stop
12009 Generate (or don't) a stop bit immediately before and after volatile asm
12012 @item -mregister-names
12013 @itemx -mno-register-names
12014 @opindex mregister-names
12015 @opindex mno-register-names
12016 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
12017 the stacked registers. This may make assembler output more readable.
12023 Disable (or enable) optimizations that use the small data section. This may
12024 be useful for working around optimizer bugs.
12026 @item -mconstant-gp
12027 @opindex mconstant-gp
12028 Generate code that uses a single constant global pointer value. This is
12029 useful when compiling kernel code.
12033 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
12034 This is useful when compiling firmware code.
12036 @item -minline-float-divide-min-latency
12037 @opindex minline-float-divide-min-latency
12038 Generate code for inline divides of floating point values
12039 using the minimum latency algorithm.
12041 @item -minline-float-divide-max-throughput
12042 @opindex minline-float-divide-max-throughput
12043 Generate code for inline divides of floating point values
12044 using the maximum throughput algorithm.
12046 @item -mno-inline-float-divide
12047 @opindex mno-inline-float-divide
12048 Do not generate inline code for divides of floating point values.
12050 @item -minline-int-divide-min-latency
12051 @opindex minline-int-divide-min-latency
12052 Generate code for inline divides of integer values
12053 using the minimum latency algorithm.
12055 @item -minline-int-divide-max-throughput
12056 @opindex minline-int-divide-max-throughput
12057 Generate code for inline divides of integer values
12058 using the maximum throughput algorithm.
12060 @item -mno-inline-int-divide
12061 @opindex mno-inline-int-divide
12062 Do not generate inline code for divides of integer values.
12064 @item -minline-sqrt-min-latency
12065 @opindex minline-sqrt-min-latency
12066 Generate code for inline square roots
12067 using the minimum latency algorithm.
12069 @item -minline-sqrt-max-throughput
12070 @opindex minline-sqrt-max-throughput
12071 Generate code for inline square roots
12072 using the maximum throughput algorithm.
12074 @item -mno-inline-sqrt
12075 @opindex mno-inline-sqrt
12076 Do not generate inline code for sqrt.
12079 @itemx -mno-fused-madd
12080 @opindex mfused-madd
12081 @opindex mno-fused-madd
12082 Do (don't) generate code that uses the fused multiply/add or multiply/subtract
12083 instructions. The default is to use these instructions.
12085 @item -mno-dwarf2-asm
12086 @itemx -mdwarf2-asm
12087 @opindex mno-dwarf2-asm
12088 @opindex mdwarf2-asm
12089 Don't (or do) generate assembler code for the DWARF2 line number debugging
12090 info. This may be useful when not using the GNU assembler.
12092 @item -mearly-stop-bits
12093 @itemx -mno-early-stop-bits
12094 @opindex mearly-stop-bits
12095 @opindex mno-early-stop-bits
12096 Allow stop bits to be placed earlier than immediately preceding the
12097 instruction that triggered the stop bit. This can improve instruction
12098 scheduling, but does not always do so.
12100 @item -mfixed-range=@var{register-range}
12101 @opindex mfixed-range
12102 Generate code treating the given register range as fixed registers.
12103 A fixed register is one that the register allocator can not use. This is
12104 useful when compiling kernel code. A register range is specified as
12105 two registers separated by a dash. Multiple register ranges can be
12106 specified separated by a comma.
12108 @item -mtls-size=@var{tls-size}
12110 Specify bit size of immediate TLS offsets. Valid values are 14, 22, and
12113 @item -mtune=@var{cpu-type}
12115 Tune the instruction scheduling for a particular CPU, Valid values are
12116 itanium, itanium1, merced, itanium2, and mckinley.
12122 Generate code for a 32-bit or 64-bit environment.
12123 The 32-bit environment sets int, long and pointer to 32 bits.
12124 The 64-bit environment sets int to 32 bits and long and pointer
12125 to 64 bits. These are HP-UX specific flags.
12127 @item -mno-sched-br-data-spec
12128 @itemx -msched-br-data-spec
12129 @opindex mno-sched-br-data-spec
12130 @opindex msched-br-data-spec
12131 (Dis/En)able data speculative scheduling before reload.
12132 This will result in generation of the ld.a instructions and
12133 the corresponding check instructions (ld.c / chk.a).
12134 The default is 'disable'.
12136 @item -msched-ar-data-spec
12137 @itemx -mno-sched-ar-data-spec
12138 @opindex msched-ar-data-spec
12139 @opindex mno-sched-ar-data-spec
12140 (En/Dis)able data speculative scheduling after reload.
12141 This will result in generation of the ld.a instructions and
12142 the corresponding check instructions (ld.c / chk.a).
12143 The default is 'enable'.
12145 @item -mno-sched-control-spec
12146 @itemx -msched-control-spec
12147 @opindex mno-sched-control-spec
12148 @opindex msched-control-spec
12149 (Dis/En)able control speculative scheduling. This feature is
12150 available only during region scheduling (i.e.@: before reload).
12151 This will result in generation of the ld.s instructions and
12152 the corresponding check instructions chk.s .
12153 The default is 'disable'.
12155 @item -msched-br-in-data-spec
12156 @itemx -mno-sched-br-in-data-spec
12157 @opindex msched-br-in-data-spec
12158 @opindex mno-sched-br-in-data-spec
12159 (En/Dis)able speculative scheduling of the instructions that
12160 are dependent on the data speculative loads before reload.
12161 This is effective only with @option{-msched-br-data-spec} enabled.
12162 The default is 'enable'.
12164 @item -msched-ar-in-data-spec
12165 @itemx -mno-sched-ar-in-data-spec
12166 @opindex msched-ar-in-data-spec
12167 @opindex mno-sched-ar-in-data-spec
12168 (En/Dis)able speculative scheduling of the instructions that
12169 are dependent on the data speculative loads after reload.
12170 This is effective only with @option{-msched-ar-data-spec} enabled.
12171 The default is 'enable'.
12173 @item -msched-in-control-spec
12174 @itemx -mno-sched-in-control-spec
12175 @opindex msched-in-control-spec
12176 @opindex mno-sched-in-control-spec
12177 (En/Dis)able speculative scheduling of the instructions that
12178 are dependent on the control speculative loads.
12179 This is effective only with @option{-msched-control-spec} enabled.
12180 The default is 'enable'.
12182 @item -mno-sched-prefer-non-data-spec-insns
12183 @itemx -msched-prefer-non-data-spec-insns
12184 @opindex mno-sched-prefer-non-data-spec-insns
12185 @opindex msched-prefer-non-data-spec-insns
12186 If enabled, data speculative instructions will be chosen for schedule
12187 only if there are no other choices at the moment. This will make
12188 the use of the data speculation much more conservative.
12189 The default is 'disable'.
12191 @item -mno-sched-prefer-non-control-spec-insns
12192 @itemx -msched-prefer-non-control-spec-insns
12193 @opindex mno-sched-prefer-non-control-spec-insns
12194 @opindex msched-prefer-non-control-spec-insns
12195 If enabled, control speculative instructions will be chosen for schedule
12196 only if there are no other choices at the moment. This will make
12197 the use of the control speculation much more conservative.
12198 The default is 'disable'.
12200 @item -mno-sched-count-spec-in-critical-path
12201 @itemx -msched-count-spec-in-critical-path
12202 @opindex mno-sched-count-spec-in-critical-path
12203 @opindex msched-count-spec-in-critical-path
12204 If enabled, speculative dependencies will be considered during
12205 computation of the instructions priorities. This will make the use of the
12206 speculation a bit more conservative.
12207 The default is 'disable'.
12209 @item -msched-spec-ldc
12210 @opindex msched-spec-ldc
12211 Use a simple data speculation check. This option is on by default.
12213 @item -msched-control-spec-ldc
12214 @opindex msched-spec-ldc
12215 Use a simple check for control speculation. This option is on by default.
12217 @item -msched-stop-bits-after-every-cycle
12218 @opindex msched-stop-bits-after-every-cycle
12219 Place a stop bit after every cycle when scheduling. This option is on
12222 @item -msched-fp-mem-deps-zero-cost
12223 @opindex msched-fp-mem-deps-zero-cost
12224 Assume that floating-point stores and loads are not likely to cause a conflict
12225 when placed into the same instruction group. This option is disabled by
12228 @item -msel-sched-dont-check-control-spec
12229 @opindex msel-sched-dont-check-control-spec
12230 Generate checks for control speculation in selective scheduling.
12231 This flag is disabled by default.
12233 @item -msched-max-memory-insns=@var{max-insns}
12234 @opindex msched-max-memory-insns
12235 Limit on the number of memory insns per instruction group, giving lower
12236 priority to subsequent memory insns attempting to schedule in the same
12237 instruction group. Frequently useful to prevent cache bank conflicts.
12238 The default value is 1.
12240 @item -msched-max-memory-insns-hard-limit
12241 @opindex msched-max-memory-insns-hard-limit
12242 Disallow more than `msched-max-memory-insns' in instruction group.
12243 Otherwise, limit is `soft' meaning that we would prefer non-memory operations
12244 when limit is reached but may still schedule memory operations.
12248 @node IA-64/VMS Options
12249 @subsection IA-64/VMS Options
12251 These @samp{-m} options are defined for the IA-64/VMS implementations:
12254 @item -mvms-return-codes
12255 @opindex mvms-return-codes
12256 Return VMS condition codes from main. The default is to return POSIX
12257 style condition (e.g.@ error) codes.
12259 @item -mdebug-main=@var{prefix}
12260 @opindex mdebug-main=@var{prefix}
12261 Flag the first routine whose name starts with @var{prefix} as the main
12262 routine for the debugger.
12266 Default to 64bit memory allocation routines.
12270 @subsection M32C Options
12271 @cindex M32C options
12274 @item -mcpu=@var{name}
12276 Select the CPU for which code is generated. @var{name} may be one of
12277 @samp{r8c} for the R8C/Tiny series, @samp{m16c} for the M16C (up to
12278 /60) series, @samp{m32cm} for the M16C/80 series, or @samp{m32c} for
12279 the M32C/80 series.
12283 Specifies that the program will be run on the simulator. This causes
12284 an alternate runtime library to be linked in which supports, for
12285 example, file I/O@. You must not use this option when generating
12286 programs that will run on real hardware; you must provide your own
12287 runtime library for whatever I/O functions are needed.
12289 @item -memregs=@var{number}
12291 Specifies the number of memory-based pseudo-registers GCC will use
12292 during code generation. These pseudo-registers will be used like real
12293 registers, so there is a tradeoff between GCC's ability to fit the
12294 code into available registers, and the performance penalty of using
12295 memory instead of registers. Note that all modules in a program must
12296 be compiled with the same value for this option. Because of that, you
12297 must not use this option with the default runtime libraries gcc
12302 @node M32R/D Options
12303 @subsection M32R/D Options
12304 @cindex M32R/D options
12306 These @option{-m} options are defined for Renesas M32R/D architectures:
12311 Generate code for the M32R/2@.
12315 Generate code for the M32R/X@.
12319 Generate code for the M32R@. This is the default.
12321 @item -mmodel=small
12322 @opindex mmodel=small
12323 Assume all objects live in the lower 16MB of memory (so that their addresses
12324 can be loaded with the @code{ld24} instruction), and assume all subroutines
12325 are reachable with the @code{bl} instruction.
12326 This is the default.
12328 The addressability of a particular object can be set with the
12329 @code{model} attribute.
12331 @item -mmodel=medium
12332 @opindex mmodel=medium
12333 Assume objects may be anywhere in the 32-bit address space (the compiler
12334 will generate @code{seth/add3} instructions to load their addresses), and
12335 assume all subroutines are reachable with the @code{bl} instruction.
12337 @item -mmodel=large
12338 @opindex mmodel=large
12339 Assume objects may be anywhere in the 32-bit address space (the compiler
12340 will generate @code{seth/add3} instructions to load their addresses), and
12341 assume subroutines may not be reachable with the @code{bl} instruction
12342 (the compiler will generate the much slower @code{seth/add3/jl}
12343 instruction sequence).
12346 @opindex msdata=none
12347 Disable use of the small data area. Variables will be put into
12348 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
12349 @code{section} attribute has been specified).
12350 This is the default.
12352 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
12353 Objects may be explicitly put in the small data area with the
12354 @code{section} attribute using one of these sections.
12356 @item -msdata=sdata
12357 @opindex msdata=sdata
12358 Put small global and static data in the small data area, but do not
12359 generate special code to reference them.
12362 @opindex msdata=use
12363 Put small global and static data in the small data area, and generate
12364 special instructions to reference them.
12368 @cindex smaller data references
12369 Put global and static objects less than or equal to @var{num} bytes
12370 into the small data or bss sections instead of the normal data or bss
12371 sections. The default value of @var{num} is 8.
12372 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
12373 for this option to have any effect.
12375 All modules should be compiled with the same @option{-G @var{num}} value.
12376 Compiling with different values of @var{num} may or may not work; if it
12377 doesn't the linker will give an error message---incorrect code will not be
12382 Makes the M32R specific code in the compiler display some statistics
12383 that might help in debugging programs.
12385 @item -malign-loops
12386 @opindex malign-loops
12387 Align all loops to a 32-byte boundary.
12389 @item -mno-align-loops
12390 @opindex mno-align-loops
12391 Do not enforce a 32-byte alignment for loops. This is the default.
12393 @item -missue-rate=@var{number}
12394 @opindex missue-rate=@var{number}
12395 Issue @var{number} instructions per cycle. @var{number} can only be 1
12398 @item -mbranch-cost=@var{number}
12399 @opindex mbranch-cost=@var{number}
12400 @var{number} can only be 1 or 2. If it is 1 then branches will be
12401 preferred over conditional code, if it is 2, then the opposite will
12404 @item -mflush-trap=@var{number}
12405 @opindex mflush-trap=@var{number}
12406 Specifies the trap number to use to flush the cache. The default is
12407 12. Valid numbers are between 0 and 15 inclusive.
12409 @item -mno-flush-trap
12410 @opindex mno-flush-trap
12411 Specifies that the cache cannot be flushed by using a trap.
12413 @item -mflush-func=@var{name}
12414 @opindex mflush-func=@var{name}
12415 Specifies the name of the operating system function to call to flush
12416 the cache. The default is @emph{_flush_cache}, but a function call
12417 will only be used if a trap is not available.
12419 @item -mno-flush-func
12420 @opindex mno-flush-func
12421 Indicates that there is no OS function for flushing the cache.
12425 @node M680x0 Options
12426 @subsection M680x0 Options
12427 @cindex M680x0 options
12429 These are the @samp{-m} options defined for M680x0 and ColdFire processors.
12430 The default settings depend on which architecture was selected when
12431 the compiler was configured; the defaults for the most common choices
12435 @item -march=@var{arch}
12437 Generate code for a specific M680x0 or ColdFire instruction set
12438 architecture. Permissible values of @var{arch} for M680x0
12439 architectures are: @samp{68000}, @samp{68010}, @samp{68020},
12440 @samp{68030}, @samp{68040}, @samp{68060} and @samp{cpu32}. ColdFire
12441 architectures are selected according to Freescale's ISA classification
12442 and the permissible values are: @samp{isaa}, @samp{isaaplus},
12443 @samp{isab} and @samp{isac}.
12445 gcc defines a macro @samp{__mcf@var{arch}__} whenever it is generating
12446 code for a ColdFire target. The @var{arch} in this macro is one of the
12447 @option{-march} arguments given above.
12449 When used together, @option{-march} and @option{-mtune} select code
12450 that runs on a family of similar processors but that is optimized
12451 for a particular microarchitecture.
12453 @item -mcpu=@var{cpu}
12455 Generate code for a specific M680x0 or ColdFire processor.
12456 The M680x0 @var{cpu}s are: @samp{68000}, @samp{68010}, @samp{68020},
12457 @samp{68030}, @samp{68040}, @samp{68060}, @samp{68302}, @samp{68332}
12458 and @samp{cpu32}. The ColdFire @var{cpu}s are given by the table
12459 below, which also classifies the CPUs into families:
12461 @multitable @columnfractions 0.20 0.80
12462 @item @strong{Family} @tab @strong{@samp{-mcpu} arguments}
12463 @item @samp{51} @tab @samp{51} @samp{51ac} @samp{51cn} @samp{51em} @samp{51qe}
12464 @item @samp{5206} @tab @samp{5202} @samp{5204} @samp{5206}
12465 @item @samp{5206e} @tab @samp{5206e}
12466 @item @samp{5208} @tab @samp{5207} @samp{5208}
12467 @item @samp{5211a} @tab @samp{5210a} @samp{5211a}
12468 @item @samp{5213} @tab @samp{5211} @samp{5212} @samp{5213}
12469 @item @samp{5216} @tab @samp{5214} @samp{5216}
12470 @item @samp{52235} @tab @samp{52230} @samp{52231} @samp{52232} @samp{52233} @samp{52234} @samp{52235}
12471 @item @samp{5225} @tab @samp{5224} @samp{5225}
12472 @item @samp{52259} @tab @samp{52252} @samp{52254} @samp{52255} @samp{52256} @samp{52258} @samp{52259}
12473 @item @samp{5235} @tab @samp{5232} @samp{5233} @samp{5234} @samp{5235} @samp{523x}
12474 @item @samp{5249} @tab @samp{5249}
12475 @item @samp{5250} @tab @samp{5250}
12476 @item @samp{5271} @tab @samp{5270} @samp{5271}
12477 @item @samp{5272} @tab @samp{5272}
12478 @item @samp{5275} @tab @samp{5274} @samp{5275}
12479 @item @samp{5282} @tab @samp{5280} @samp{5281} @samp{5282} @samp{528x}
12480 @item @samp{53017} @tab @samp{53011} @samp{53012} @samp{53013} @samp{53014} @samp{53015} @samp{53016} @samp{53017}
12481 @item @samp{5307} @tab @samp{5307}
12482 @item @samp{5329} @tab @samp{5327} @samp{5328} @samp{5329} @samp{532x}
12483 @item @samp{5373} @tab @samp{5372} @samp{5373} @samp{537x}
12484 @item @samp{5407} @tab @samp{5407}
12485 @item @samp{5475} @tab @samp{5470} @samp{5471} @samp{5472} @samp{5473} @samp{5474} @samp{5475} @samp{547x} @samp{5480} @samp{5481} @samp{5482} @samp{5483} @samp{5484} @samp{5485}
12488 @option{-mcpu=@var{cpu}} overrides @option{-march=@var{arch}} if
12489 @var{arch} is compatible with @var{cpu}. Other combinations of
12490 @option{-mcpu} and @option{-march} are rejected.
12492 gcc defines the macro @samp{__mcf_cpu_@var{cpu}} when ColdFire target
12493 @var{cpu} is selected. It also defines @samp{__mcf_family_@var{family}},
12494 where the value of @var{family} is given by the table above.
12496 @item -mtune=@var{tune}
12498 Tune the code for a particular microarchitecture, within the
12499 constraints set by @option{-march} and @option{-mcpu}.
12500 The M680x0 microarchitectures are: @samp{68000}, @samp{68010},
12501 @samp{68020}, @samp{68030}, @samp{68040}, @samp{68060}
12502 and @samp{cpu32}. The ColdFire microarchitectures
12503 are: @samp{cfv1}, @samp{cfv2}, @samp{cfv3}, @samp{cfv4} and @samp{cfv4e}.
12505 You can also use @option{-mtune=68020-40} for code that needs
12506 to run relatively well on 68020, 68030 and 68040 targets.
12507 @option{-mtune=68020-60} is similar but includes 68060 targets
12508 as well. These two options select the same tuning decisions as
12509 @option{-m68020-40} and @option{-m68020-60} respectively.
12511 gcc defines the macros @samp{__mc@var{arch}} and @samp{__mc@var{arch}__}
12512 when tuning for 680x0 architecture @var{arch}. It also defines
12513 @samp{mc@var{arch}} unless either @option{-ansi} or a non-GNU @option{-std}
12514 option is used. If gcc is tuning for a range of architectures,
12515 as selected by @option{-mtune=68020-40} or @option{-mtune=68020-60},
12516 it defines the macros for every architecture in the range.
12518 gcc also defines the macro @samp{__m@var{uarch}__} when tuning for
12519 ColdFire microarchitecture @var{uarch}, where @var{uarch} is one
12520 of the arguments given above.
12526 Generate output for a 68000. This is the default
12527 when the compiler is configured for 68000-based systems.
12528 It is equivalent to @option{-march=68000}.
12530 Use this option for microcontrollers with a 68000 or EC000 core,
12531 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
12535 Generate output for a 68010. This is the default
12536 when the compiler is configured for 68010-based systems.
12537 It is equivalent to @option{-march=68010}.
12543 Generate output for a 68020. This is the default
12544 when the compiler is configured for 68020-based systems.
12545 It is equivalent to @option{-march=68020}.
12549 Generate output for a 68030. This is the default when the compiler is
12550 configured for 68030-based systems. It is equivalent to
12551 @option{-march=68030}.
12555 Generate output for a 68040. This is the default when the compiler is
12556 configured for 68040-based systems. It is equivalent to
12557 @option{-march=68040}.
12559 This option inhibits the use of 68881/68882 instructions that have to be
12560 emulated by software on the 68040. Use this option if your 68040 does not
12561 have code to emulate those instructions.
12565 Generate output for a 68060. This is the default when the compiler is
12566 configured for 68060-based systems. It is equivalent to
12567 @option{-march=68060}.
12569 This option inhibits the use of 68020 and 68881/68882 instructions that
12570 have to be emulated by software on the 68060. Use this option if your 68060
12571 does not have code to emulate those instructions.
12575 Generate output for a CPU32. This is the default
12576 when the compiler is configured for CPU32-based systems.
12577 It is equivalent to @option{-march=cpu32}.
12579 Use this option for microcontrollers with a
12580 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
12581 68336, 68340, 68341, 68349 and 68360.
12585 Generate output for a 520X ColdFire CPU@. This is the default
12586 when the compiler is configured for 520X-based systems.
12587 It is equivalent to @option{-mcpu=5206}, and is now deprecated
12588 in favor of that option.
12590 Use this option for microcontroller with a 5200 core, including
12591 the MCF5202, MCF5203, MCF5204 and MCF5206.
12595 Generate output for a 5206e ColdFire CPU@. The option is now
12596 deprecated in favor of the equivalent @option{-mcpu=5206e}.
12600 Generate output for a member of the ColdFire 528X family.
12601 The option is now deprecated in favor of the equivalent
12602 @option{-mcpu=528x}.
12606 Generate output for a ColdFire 5307 CPU@. The option is now deprecated
12607 in favor of the equivalent @option{-mcpu=5307}.
12611 Generate output for a ColdFire 5407 CPU@. The option is now deprecated
12612 in favor of the equivalent @option{-mcpu=5407}.
12616 Generate output for a ColdFire V4e family CPU (e.g.@: 547x/548x).
12617 This includes use of hardware floating point instructions.
12618 The option is equivalent to @option{-mcpu=547x}, and is now
12619 deprecated in favor of that option.
12623 Generate output for a 68040, without using any of the new instructions.
12624 This results in code which can run relatively efficiently on either a
12625 68020/68881 or a 68030 or a 68040. The generated code does use the
12626 68881 instructions that are emulated on the 68040.
12628 The option is equivalent to @option{-march=68020} @option{-mtune=68020-40}.
12632 Generate output for a 68060, without using any of the new instructions.
12633 This results in code which can run relatively efficiently on either a
12634 68020/68881 or a 68030 or a 68040. The generated code does use the
12635 68881 instructions that are emulated on the 68060.
12637 The option is equivalent to @option{-march=68020} @option{-mtune=68020-60}.
12641 @opindex mhard-float
12643 Generate floating-point instructions. This is the default for 68020
12644 and above, and for ColdFire devices that have an FPU@. It defines the
12645 macro @samp{__HAVE_68881__} on M680x0 targets and @samp{__mcffpu__}
12646 on ColdFire targets.
12649 @opindex msoft-float
12650 Do not generate floating-point instructions; use library calls instead.
12651 This is the default for 68000, 68010, and 68832 targets. It is also
12652 the default for ColdFire devices that have no FPU.
12658 Generate (do not generate) ColdFire hardware divide and remainder
12659 instructions. If @option{-march} is used without @option{-mcpu},
12660 the default is ``on'' for ColdFire architectures and ``off'' for M680x0
12661 architectures. Otherwise, the default is taken from the target CPU
12662 (either the default CPU, or the one specified by @option{-mcpu}). For
12663 example, the default is ``off'' for @option{-mcpu=5206} and ``on'' for
12664 @option{-mcpu=5206e}.
12666 gcc defines the macro @samp{__mcfhwdiv__} when this option is enabled.
12670 Consider type @code{int} to be 16 bits wide, like @code{short int}.
12671 Additionally, parameters passed on the stack are also aligned to a
12672 16-bit boundary even on targets whose API mandates promotion to 32-bit.
12676 Do not consider type @code{int} to be 16 bits wide. This is the default.
12679 @itemx -mno-bitfield
12680 @opindex mnobitfield
12681 @opindex mno-bitfield
12682 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
12683 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
12687 Do use the bit-field instructions. The @option{-m68020} option implies
12688 @option{-mbitfield}. This is the default if you use a configuration
12689 designed for a 68020.
12693 Use a different function-calling convention, in which functions
12694 that take a fixed number of arguments return with the @code{rtd}
12695 instruction, which pops their arguments while returning. This
12696 saves one instruction in the caller since there is no need to pop
12697 the arguments there.
12699 This calling convention is incompatible with the one normally
12700 used on Unix, so you cannot use it if you need to call libraries
12701 compiled with the Unix compiler.
12703 Also, you must provide function prototypes for all functions that
12704 take variable numbers of arguments (including @code{printf});
12705 otherwise incorrect code will be generated for calls to those
12708 In addition, seriously incorrect code will result if you call a
12709 function with too many arguments. (Normally, extra arguments are
12710 harmlessly ignored.)
12712 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
12713 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
12717 Do not use the calling conventions selected by @option{-mrtd}.
12718 This is the default.
12721 @itemx -mno-align-int
12722 @opindex malign-int
12723 @opindex mno-align-int
12724 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
12725 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
12726 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
12727 Aligning variables on 32-bit boundaries produces code that runs somewhat
12728 faster on processors with 32-bit busses at the expense of more memory.
12730 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
12731 align structures containing the above types differently than
12732 most published application binary interface specifications for the m68k.
12736 Use the pc-relative addressing mode of the 68000 directly, instead of
12737 using a global offset table. At present, this option implies @option{-fpic},
12738 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
12739 not presently supported with @option{-mpcrel}, though this could be supported for
12740 68020 and higher processors.
12742 @item -mno-strict-align
12743 @itemx -mstrict-align
12744 @opindex mno-strict-align
12745 @opindex mstrict-align
12746 Do not (do) assume that unaligned memory references will be handled by
12750 Generate code that allows the data segment to be located in a different
12751 area of memory from the text segment. This allows for execute in place in
12752 an environment without virtual memory management. This option implies
12755 @item -mno-sep-data
12756 Generate code that assumes that the data segment follows the text segment.
12757 This is the default.
12759 @item -mid-shared-library
12760 Generate code that supports shared libraries via the library ID method.
12761 This allows for execute in place and shared libraries in an environment
12762 without virtual memory management. This option implies @option{-fPIC}.
12764 @item -mno-id-shared-library
12765 Generate code that doesn't assume ID based shared libraries are being used.
12766 This is the default.
12768 @item -mshared-library-id=n
12769 Specified the identification number of the ID based shared library being
12770 compiled. Specifying a value of 0 will generate more compact code, specifying
12771 other values will force the allocation of that number to the current
12772 library but is no more space or time efficient than omitting this option.
12778 When generating position-independent code for ColdFire, generate code
12779 that works if the GOT has more than 8192 entries. This code is
12780 larger and slower than code generated without this option. On M680x0
12781 processors, this option is not needed; @option{-fPIC} suffices.
12783 GCC normally uses a single instruction to load values from the GOT@.
12784 While this is relatively efficient, it only works if the GOT
12785 is smaller than about 64k. Anything larger causes the linker
12786 to report an error such as:
12788 @cindex relocation truncated to fit (ColdFire)
12790 relocation truncated to fit: R_68K_GOT16O foobar
12793 If this happens, you should recompile your code with @option{-mxgot}.
12794 It should then work with very large GOTs. However, code generated with
12795 @option{-mxgot} is less efficient, since it takes 4 instructions to fetch
12796 the value of a global symbol.
12798 Note that some linkers, including newer versions of the GNU linker,
12799 can create multiple GOTs and sort GOT entries. If you have such a linker,
12800 you should only need to use @option{-mxgot} when compiling a single
12801 object file that accesses more than 8192 GOT entries. Very few do.
12803 These options have no effect unless GCC is generating
12804 position-independent code.
12808 @node M68hc1x Options
12809 @subsection M68hc1x Options
12810 @cindex M68hc1x options
12812 These are the @samp{-m} options defined for the 68hc11 and 68hc12
12813 microcontrollers. The default values for these options depends on
12814 which style of microcontroller was selected when the compiler was configured;
12815 the defaults for the most common choices are given below.
12822 Generate output for a 68HC11. This is the default
12823 when the compiler is configured for 68HC11-based systems.
12829 Generate output for a 68HC12. This is the default
12830 when the compiler is configured for 68HC12-based systems.
12836 Generate output for a 68HCS12.
12838 @item -mauto-incdec
12839 @opindex mauto-incdec
12840 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
12847 Enable the use of 68HC12 min and max instructions.
12850 @itemx -mno-long-calls
12851 @opindex mlong-calls
12852 @opindex mno-long-calls
12853 Treat all calls as being far away (near). If calls are assumed to be
12854 far away, the compiler will use the @code{call} instruction to
12855 call a function and the @code{rtc} instruction for returning.
12859 Consider type @code{int} to be 16 bits wide, like @code{short int}.
12861 @item -msoft-reg-count=@var{count}
12862 @opindex msoft-reg-count
12863 Specify the number of pseudo-soft registers which are used for the
12864 code generation. The maximum number is 32. Using more pseudo-soft
12865 register may or may not result in better code depending on the program.
12866 The default is 4 for 68HC11 and 2 for 68HC12.
12870 @node MCore Options
12871 @subsection MCore Options
12872 @cindex MCore options
12874 These are the @samp{-m} options defined for the Motorola M*Core
12880 @itemx -mno-hardlit
12882 @opindex mno-hardlit
12883 Inline constants into the code stream if it can be done in two
12884 instructions or less.
12890 Use the divide instruction. (Enabled by default).
12892 @item -mrelax-immediate
12893 @itemx -mno-relax-immediate
12894 @opindex mrelax-immediate
12895 @opindex mno-relax-immediate
12896 Allow arbitrary sized immediates in bit operations.
12898 @item -mwide-bitfields
12899 @itemx -mno-wide-bitfields
12900 @opindex mwide-bitfields
12901 @opindex mno-wide-bitfields
12902 Always treat bit-fields as int-sized.
12904 @item -m4byte-functions
12905 @itemx -mno-4byte-functions
12906 @opindex m4byte-functions
12907 @opindex mno-4byte-functions
12908 Force all functions to be aligned to a four byte boundary.
12910 @item -mcallgraph-data
12911 @itemx -mno-callgraph-data
12912 @opindex mcallgraph-data
12913 @opindex mno-callgraph-data
12914 Emit callgraph information.
12917 @itemx -mno-slow-bytes
12918 @opindex mslow-bytes
12919 @opindex mno-slow-bytes
12920 Prefer word access when reading byte quantities.
12922 @item -mlittle-endian
12923 @itemx -mbig-endian
12924 @opindex mlittle-endian
12925 @opindex mbig-endian
12926 Generate code for a little endian target.
12932 Generate code for the 210 processor.
12936 Assume that run-time support has been provided and so omit the
12937 simulator library (@file{libsim.a)} from the linker command line.
12939 @item -mstack-increment=@var{size}
12940 @opindex mstack-increment
12941 Set the maximum amount for a single stack increment operation. Large
12942 values can increase the speed of programs which contain functions
12943 that need a large amount of stack space, but they can also trigger a
12944 segmentation fault if the stack is extended too much. The default
12950 @subsection MeP Options
12951 @cindex MeP options
12957 Enables the @code{abs} instruction, which is the absolute difference
12958 between two registers.
12962 Enables all the optional instructions - average, multiply, divide, bit
12963 operations, leading zero, absolute difference, min/max, clip, and
12969 Enables the @code{ave} instruction, which computes the average of two
12972 @item -mbased=@var{n}
12974 Variables of size @var{n} bytes or smaller will be placed in the
12975 @code{.based} section by default. Based variables use the @code{$tp}
12976 register as a base register, and there is a 128 byte limit to the
12977 @code{.based} section.
12981 Enables the bit operation instructions - bit test (@code{btstm}), set
12982 (@code{bsetm}), clear (@code{bclrm}), invert (@code{bnotm}), and
12983 test-and-set (@code{tas}).
12985 @item -mc=@var{name}
12987 Selects which section constant data will be placed in. @var{name} may
12988 be @code{tiny}, @code{near}, or @code{far}.
12992 Enables the @code{clip} instruction. Note that @code{-mclip} is not
12993 useful unless you also provide @code{-mminmax}.
12995 @item -mconfig=@var{name}
12997 Selects one of the build-in core configurations. Each MeP chip has
12998 one or more modules in it; each module has a core CPU and a variety of
12999 coprocessors, optional instructions, and peripherals. The
13000 @code{MeP-Integrator} tool, not part of GCC, provides these
13001 configurations through this option; using this option is the same as
13002 using all the corresponding command line options. The default
13003 configuration is @code{default}.
13007 Enables the coprocessor instructions. By default, this is a 32-bit
13008 coprocessor. Note that the coprocessor is normally enabled via the
13009 @code{-mconfig=} option.
13013 Enables the 32-bit coprocessor's instructions.
13017 Enables the 64-bit coprocessor's instructions.
13021 Enables IVC2 scheduling. IVC2 is a 64-bit VLIW coprocessor.
13025 Causes constant variables to be placed in the @code{.near} section.
13029 Enables the @code{div} and @code{divu} instructions.
13033 Generate big-endian code.
13037 Generate little-endian code.
13039 @item -mio-volatile
13040 @opindex mio-volatile
13041 Tells the compiler that any variable marked with the @code{io}
13042 attribute is to be considered volatile.
13046 Causes variables to be assigned to the @code{.far} section by default.
13050 Enables the @code{leadz} (leading zero) instruction.
13054 Causes variables to be assigned to the @code{.near} section by default.
13058 Enables the @code{min} and @code{max} instructions.
13062 Enables the multiplication and multiply-accumulate instructions.
13066 Disables all the optional instructions enabled by @code{-mall-opts}.
13070 Enables the @code{repeat} and @code{erepeat} instructions, used for
13071 low-overhead looping.
13075 Causes all variables to default to the @code{.tiny} section. Note
13076 that there is a 65536 byte limit to this section. Accesses to these
13077 variables use the @code{%gp} base register.
13081 Enables the saturation instructions. Note that the compiler does not
13082 currently generate these itself, but this option is included for
13083 compatibility with other tools, like @code{as}.
13087 Link the SDRAM-based runtime instead of the default ROM-based runtime.
13091 Link the simulator runtime libraries.
13095 Link the simulator runtime libraries, excluding built-in support
13096 for reset and exception vectors and tables.
13100 Causes all functions to default to the @code{.far} section. Without
13101 this option, functions default to the @code{.near} section.
13103 @item -mtiny=@var{n}
13105 Variables that are @var{n} bytes or smaller will be allocated to the
13106 @code{.tiny} section. These variables use the @code{$gp} base
13107 register. The default for this option is 4, but note that there's a
13108 65536 byte limit to the @code{.tiny} section.
13113 @subsection MIPS Options
13114 @cindex MIPS options
13120 Generate big-endian code.
13124 Generate little-endian code. This is the default for @samp{mips*el-*-*}
13127 @item -march=@var{arch}
13129 Generate code that will run on @var{arch}, which can be the name of a
13130 generic MIPS ISA, or the name of a particular processor.
13132 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
13133 @samp{mips32}, @samp{mips32r2}, @samp{mips64} and @samp{mips64r2}.
13134 The processor names are:
13135 @samp{4kc}, @samp{4km}, @samp{4kp}, @samp{4ksc},
13136 @samp{4kec}, @samp{4kem}, @samp{4kep}, @samp{4ksd},
13137 @samp{5kc}, @samp{5kf},
13139 @samp{24kc}, @samp{24kf2_1}, @samp{24kf1_1},
13140 @samp{24kec}, @samp{24kef2_1}, @samp{24kef1_1},
13141 @samp{34kc}, @samp{34kf2_1}, @samp{34kf1_1},
13142 @samp{74kc}, @samp{74kf2_1}, @samp{74kf1_1}, @samp{74kf3_2},
13143 @samp{1004kc}, @samp{1004kf2_1}, @samp{1004kf1_1},
13144 @samp{loongson2e}, @samp{loongson2f},
13148 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
13149 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000},
13150 @samp{rm7000}, @samp{rm9000},
13151 @samp{r10000}, @samp{r12000}, @samp{r14000}, @samp{r16000},
13154 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4130}, @samp{vr4300},
13155 @samp{vr5000}, @samp{vr5400}, @samp{vr5500}
13157 The special value @samp{from-abi} selects the
13158 most compatible architecture for the selected ABI (that is,
13159 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
13161 Native Linux/GNU toolchains also support the value @samp{native},
13162 which selects the best architecture option for the host processor.
13163 @option{-march=native} has no effect if GCC does not recognize
13166 In processor names, a final @samp{000} can be abbreviated as @samp{k}
13167 (for example, @samp{-march=r2k}). Prefixes are optional, and
13168 @samp{vr} may be written @samp{r}.
13170 Names of the form @samp{@var{n}f2_1} refer to processors with
13171 FPUs clocked at half the rate of the core, names of the form
13172 @samp{@var{n}f1_1} refer to processors with FPUs clocked at the same
13173 rate as the core, and names of the form @samp{@var{n}f3_2} refer to
13174 processors with FPUs clocked a ratio of 3:2 with respect to the core.
13175 For compatibility reasons, @samp{@var{n}f} is accepted as a synonym
13176 for @samp{@var{n}f2_1} while @samp{@var{n}x} and @samp{@var{b}fx} are
13177 accepted as synonyms for @samp{@var{n}f1_1}.
13179 GCC defines two macros based on the value of this option. The first
13180 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
13181 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
13182 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
13183 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
13184 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
13186 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
13187 above. In other words, it will have the full prefix and will not
13188 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
13189 the macro names the resolved architecture (either @samp{"mips1"} or
13190 @samp{"mips3"}). It names the default architecture when no
13191 @option{-march} option is given.
13193 @item -mtune=@var{arch}
13195 Optimize for @var{arch}. Among other things, this option controls
13196 the way instructions are scheduled, and the perceived cost of arithmetic
13197 operations. The list of @var{arch} values is the same as for
13200 When this option is not used, GCC will optimize for the processor
13201 specified by @option{-march}. By using @option{-march} and
13202 @option{-mtune} together, it is possible to generate code that will
13203 run on a family of processors, but optimize the code for one
13204 particular member of that family.
13206 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
13207 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
13208 @samp{-march} ones described above.
13212 Equivalent to @samp{-march=mips1}.
13216 Equivalent to @samp{-march=mips2}.
13220 Equivalent to @samp{-march=mips3}.
13224 Equivalent to @samp{-march=mips4}.
13228 Equivalent to @samp{-march=mips32}.
13232 Equivalent to @samp{-march=mips32r2}.
13236 Equivalent to @samp{-march=mips64}.
13240 Equivalent to @samp{-march=mips64r2}.
13245 @opindex mno-mips16
13246 Generate (do not generate) MIPS16 code. If GCC is targetting a
13247 MIPS32 or MIPS64 architecture, it will make use of the MIPS16e ASE@.
13249 MIPS16 code generation can also be controlled on a per-function basis
13250 by means of @code{mips16} and @code{nomips16} attributes.
13251 @xref{Function Attributes}, for more information.
13253 @item -mflip-mips16
13254 @opindex mflip-mips16
13255 Generate MIPS16 code on alternating functions. This option is provided
13256 for regression testing of mixed MIPS16/non-MIPS16 code generation, and is
13257 not intended for ordinary use in compiling user code.
13259 @item -minterlink-mips16
13260 @itemx -mno-interlink-mips16
13261 @opindex minterlink-mips16
13262 @opindex mno-interlink-mips16
13263 Require (do not require) that non-MIPS16 code be link-compatible with
13266 For example, non-MIPS16 code cannot jump directly to MIPS16 code;
13267 it must either use a call or an indirect jump. @option{-minterlink-mips16}
13268 therefore disables direct jumps unless GCC knows that the target of the
13269 jump is not MIPS16.
13281 Generate code for the given ABI@.
13283 Note that the EABI has a 32-bit and a 64-bit variant. GCC normally
13284 generates 64-bit code when you select a 64-bit architecture, but you
13285 can use @option{-mgp32} to get 32-bit code instead.
13287 For information about the O64 ABI, see
13288 @w{@uref{http://gcc.gnu.org/projects/mipso64-abi.html}}.
13290 GCC supports a variant of the o32 ABI in which floating-point registers
13291 are 64 rather than 32 bits wide. You can select this combination with
13292 @option{-mabi=32} @option{-mfp64}. This ABI relies on the @samp{mthc1}
13293 and @samp{mfhc1} instructions and is therefore only supported for
13294 MIPS32R2 processors.
13296 The register assignments for arguments and return values remain the
13297 same, but each scalar value is passed in a single 64-bit register
13298 rather than a pair of 32-bit registers. For example, scalar
13299 floating-point values are returned in @samp{$f0} only, not a
13300 @samp{$f0}/@samp{$f1} pair. The set of call-saved registers also
13301 remains the same, but all 64 bits are saved.
13304 @itemx -mno-abicalls
13306 @opindex mno-abicalls
13307 Generate (do not generate) code that is suitable for SVR4-style
13308 dynamic objects. @option{-mabicalls} is the default for SVR4-based
13313 Generate (do not generate) code that is fully position-independent,
13314 and that can therefore be linked into shared libraries. This option
13315 only affects @option{-mabicalls}.
13317 All @option{-mabicalls} code has traditionally been position-independent,
13318 regardless of options like @option{-fPIC} and @option{-fpic}. However,
13319 as an extension, the GNU toolchain allows executables to use absolute
13320 accesses for locally-binding symbols. It can also use shorter GP
13321 initialization sequences and generate direct calls to locally-defined
13322 functions. This mode is selected by @option{-mno-shared}.
13324 @option{-mno-shared} depends on binutils 2.16 or higher and generates
13325 objects that can only be linked by the GNU linker. However, the option
13326 does not affect the ABI of the final executable; it only affects the ABI
13327 of relocatable objects. Using @option{-mno-shared} will generally make
13328 executables both smaller and quicker.
13330 @option{-mshared} is the default.
13336 Assume (do not assume) that the static and dynamic linkers
13337 support PLTs and copy relocations. This option only affects
13338 @samp{-mno-shared -mabicalls}. For the n64 ABI, this option
13339 has no effect without @samp{-msym32}.
13341 You can make @option{-mplt} the default by configuring
13342 GCC with @option{--with-mips-plt}. The default is
13343 @option{-mno-plt} otherwise.
13349 Lift (do not lift) the usual restrictions on the size of the global
13352 GCC normally uses a single instruction to load values from the GOT@.
13353 While this is relatively efficient, it will only work if the GOT
13354 is smaller than about 64k. Anything larger will cause the linker
13355 to report an error such as:
13357 @cindex relocation truncated to fit (MIPS)
13359 relocation truncated to fit: R_MIPS_GOT16 foobar
13362 If this happens, you should recompile your code with @option{-mxgot}.
13363 It should then work with very large GOTs, although it will also be
13364 less efficient, since it will take three instructions to fetch the
13365 value of a global symbol.
13367 Note that some linkers can create multiple GOTs. If you have such a
13368 linker, you should only need to use @option{-mxgot} when a single object
13369 file accesses more than 64k's worth of GOT entries. Very few do.
13371 These options have no effect unless GCC is generating position
13376 Assume that general-purpose registers are 32 bits wide.
13380 Assume that general-purpose registers are 64 bits wide.
13384 Assume that floating-point registers are 32 bits wide.
13388 Assume that floating-point registers are 64 bits wide.
13391 @opindex mhard-float
13392 Use floating-point coprocessor instructions.
13395 @opindex msoft-float
13396 Do not use floating-point coprocessor instructions. Implement
13397 floating-point calculations using library calls instead.
13399 @item -msingle-float
13400 @opindex msingle-float
13401 Assume that the floating-point coprocessor only supports single-precision
13404 @item -mdouble-float
13405 @opindex mdouble-float
13406 Assume that the floating-point coprocessor supports double-precision
13407 operations. This is the default.
13413 Use (do not use) @samp{ll}, @samp{sc}, and @samp{sync} instructions to
13414 implement atomic memory built-in functions. When neither option is
13415 specified, GCC will use the instructions if the target architecture
13418 @option{-mllsc} is useful if the runtime environment can emulate the
13419 instructions and @option{-mno-llsc} can be useful when compiling for
13420 nonstandard ISAs. You can make either option the default by
13421 configuring GCC with @option{--with-llsc} and @option{--without-llsc}
13422 respectively. @option{--with-llsc} is the default for some
13423 configurations; see the installation documentation for details.
13429 Use (do not use) revision 1 of the MIPS DSP ASE@.
13430 @xref{MIPS DSP Built-in Functions}. This option defines the
13431 preprocessor macro @samp{__mips_dsp}. It also defines
13432 @samp{__mips_dsp_rev} to 1.
13438 Use (do not use) revision 2 of the MIPS DSP ASE@.
13439 @xref{MIPS DSP Built-in Functions}. This option defines the
13440 preprocessor macros @samp{__mips_dsp} and @samp{__mips_dspr2}.
13441 It also defines @samp{__mips_dsp_rev} to 2.
13444 @itemx -mno-smartmips
13445 @opindex msmartmips
13446 @opindex mno-smartmips
13447 Use (do not use) the MIPS SmartMIPS ASE.
13449 @item -mpaired-single
13450 @itemx -mno-paired-single
13451 @opindex mpaired-single
13452 @opindex mno-paired-single
13453 Use (do not use) paired-single floating-point instructions.
13454 @xref{MIPS Paired-Single Support}. This option requires
13455 hardware floating-point support to be enabled.
13461 Use (do not use) MIPS Digital Media Extension instructions.
13462 This option can only be used when generating 64-bit code and requires
13463 hardware floating-point support to be enabled.
13468 @opindex mno-mips3d
13469 Use (do not use) the MIPS-3D ASE@. @xref{MIPS-3D Built-in Functions}.
13470 The option @option{-mips3d} implies @option{-mpaired-single}.
13476 Use (do not use) MT Multithreading instructions.
13480 Force @code{long} types to be 64 bits wide. See @option{-mlong32} for
13481 an explanation of the default and the way that the pointer size is
13486 Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
13488 The default size of @code{int}s, @code{long}s and pointers depends on
13489 the ABI@. All the supported ABIs use 32-bit @code{int}s. The n64 ABI
13490 uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
13491 32-bit @code{long}s. Pointers are the same size as @code{long}s,
13492 or the same size as integer registers, whichever is smaller.
13498 Assume (do not assume) that all symbols have 32-bit values, regardless
13499 of the selected ABI@. This option is useful in combination with
13500 @option{-mabi=64} and @option{-mno-abicalls} because it allows GCC
13501 to generate shorter and faster references to symbolic addresses.
13505 Put definitions of externally-visible data in a small data section
13506 if that data is no bigger than @var{num} bytes. GCC can then access
13507 the data more efficiently; see @option{-mgpopt} for details.
13509 The default @option{-G} option depends on the configuration.
13511 @item -mlocal-sdata
13512 @itemx -mno-local-sdata
13513 @opindex mlocal-sdata
13514 @opindex mno-local-sdata
13515 Extend (do not extend) the @option{-G} behavior to local data too,
13516 such as to static variables in C@. @option{-mlocal-sdata} is the
13517 default for all configurations.
13519 If the linker complains that an application is using too much small data,
13520 you might want to try rebuilding the less performance-critical parts with
13521 @option{-mno-local-sdata}. You might also want to build large
13522 libraries with @option{-mno-local-sdata}, so that the libraries leave
13523 more room for the main program.
13525 @item -mextern-sdata
13526 @itemx -mno-extern-sdata
13527 @opindex mextern-sdata
13528 @opindex mno-extern-sdata
13529 Assume (do not assume) that externally-defined data will be in
13530 a small data section if that data is within the @option{-G} limit.
13531 @option{-mextern-sdata} is the default for all configurations.
13533 If you compile a module @var{Mod} with @option{-mextern-sdata} @option{-G
13534 @var{num}} @option{-mgpopt}, and @var{Mod} references a variable @var{Var}
13535 that is no bigger than @var{num} bytes, you must make sure that @var{Var}
13536 is placed in a small data section. If @var{Var} is defined by another
13537 module, you must either compile that module with a high-enough
13538 @option{-G} setting or attach a @code{section} attribute to @var{Var}'s
13539 definition. If @var{Var} is common, you must link the application
13540 with a high-enough @option{-G} setting.
13542 The easiest way of satisfying these restrictions is to compile
13543 and link every module with the same @option{-G} option. However,
13544 you may wish to build a library that supports several different
13545 small data limits. You can do this by compiling the library with
13546 the highest supported @option{-G} setting and additionally using
13547 @option{-mno-extern-sdata} to stop the library from making assumptions
13548 about externally-defined data.
13554 Use (do not use) GP-relative accesses for symbols that are known to be
13555 in a small data section; see @option{-G}, @option{-mlocal-sdata} and
13556 @option{-mextern-sdata}. @option{-mgpopt} is the default for all
13559 @option{-mno-gpopt} is useful for cases where the @code{$gp} register
13560 might not hold the value of @code{_gp}. For example, if the code is
13561 part of a library that might be used in a boot monitor, programs that
13562 call boot monitor routines will pass an unknown value in @code{$gp}.
13563 (In such situations, the boot monitor itself would usually be compiled
13564 with @option{-G0}.)
13566 @option{-mno-gpopt} implies @option{-mno-local-sdata} and
13567 @option{-mno-extern-sdata}.
13569 @item -membedded-data
13570 @itemx -mno-embedded-data
13571 @opindex membedded-data
13572 @opindex mno-embedded-data
13573 Allocate variables to the read-only data section first if possible, then
13574 next in the small data section if possible, otherwise in data. This gives
13575 slightly slower code than the default, but reduces the amount of RAM required
13576 when executing, and thus may be preferred for some embedded systems.
13578 @item -muninit-const-in-rodata
13579 @itemx -mno-uninit-const-in-rodata
13580 @opindex muninit-const-in-rodata
13581 @opindex mno-uninit-const-in-rodata
13582 Put uninitialized @code{const} variables in the read-only data section.
13583 This option is only meaningful in conjunction with @option{-membedded-data}.
13585 @item -mcode-readable=@var{setting}
13586 @opindex mcode-readable
13587 Specify whether GCC may generate code that reads from executable sections.
13588 There are three possible settings:
13591 @item -mcode-readable=yes
13592 Instructions may freely access executable sections. This is the
13595 @item -mcode-readable=pcrel
13596 MIPS16 PC-relative load instructions can access executable sections,
13597 but other instructions must not do so. This option is useful on 4KSc
13598 and 4KSd processors when the code TLBs have the Read Inhibit bit set.
13599 It is also useful on processors that can be configured to have a dual
13600 instruction/data SRAM interface and that, like the M4K, automatically
13601 redirect PC-relative loads to the instruction RAM.
13603 @item -mcode-readable=no
13604 Instructions must not access executable sections. This option can be
13605 useful on targets that are configured to have a dual instruction/data
13606 SRAM interface but that (unlike the M4K) do not automatically redirect
13607 PC-relative loads to the instruction RAM.
13610 @item -msplit-addresses
13611 @itemx -mno-split-addresses
13612 @opindex msplit-addresses
13613 @opindex mno-split-addresses
13614 Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
13615 relocation operators. This option has been superseded by
13616 @option{-mexplicit-relocs} but is retained for backwards compatibility.
13618 @item -mexplicit-relocs
13619 @itemx -mno-explicit-relocs
13620 @opindex mexplicit-relocs
13621 @opindex mno-explicit-relocs
13622 Use (do not use) assembler relocation operators when dealing with symbolic
13623 addresses. The alternative, selected by @option{-mno-explicit-relocs},
13624 is to use assembler macros instead.
13626 @option{-mexplicit-relocs} is the default if GCC was configured
13627 to use an assembler that supports relocation operators.
13629 @item -mcheck-zero-division
13630 @itemx -mno-check-zero-division
13631 @opindex mcheck-zero-division
13632 @opindex mno-check-zero-division
13633 Trap (do not trap) on integer division by zero.
13635 The default is @option{-mcheck-zero-division}.
13637 @item -mdivide-traps
13638 @itemx -mdivide-breaks
13639 @opindex mdivide-traps
13640 @opindex mdivide-breaks
13641 MIPS systems check for division by zero by generating either a
13642 conditional trap or a break instruction. Using traps results in
13643 smaller code, but is only supported on MIPS II and later. Also, some
13644 versions of the Linux kernel have a bug that prevents trap from
13645 generating the proper signal (@code{SIGFPE}). Use @option{-mdivide-traps} to
13646 allow conditional traps on architectures that support them and
13647 @option{-mdivide-breaks} to force the use of breaks.
13649 The default is usually @option{-mdivide-traps}, but this can be
13650 overridden at configure time using @option{--with-divide=breaks}.
13651 Divide-by-zero checks can be completely disabled using
13652 @option{-mno-check-zero-division}.
13657 @opindex mno-memcpy
13658 Force (do not force) the use of @code{memcpy()} for non-trivial block
13659 moves. The default is @option{-mno-memcpy}, which allows GCC to inline
13660 most constant-sized copies.
13663 @itemx -mno-long-calls
13664 @opindex mlong-calls
13665 @opindex mno-long-calls
13666 Disable (do not disable) use of the @code{jal} instruction. Calling
13667 functions using @code{jal} is more efficient but requires the caller
13668 and callee to be in the same 256 megabyte segment.
13670 This option has no effect on abicalls code. The default is
13671 @option{-mno-long-calls}.
13677 Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
13678 instructions, as provided by the R4650 ISA@.
13681 @itemx -mno-fused-madd
13682 @opindex mfused-madd
13683 @opindex mno-fused-madd
13684 Enable (disable) use of the floating point multiply-accumulate
13685 instructions, when they are available. The default is
13686 @option{-mfused-madd}.
13688 When multiply-accumulate instructions are used, the intermediate
13689 product is calculated to infinite precision and is not subject to
13690 the FCSR Flush to Zero bit. This may be undesirable in some
13695 Tell the MIPS assembler to not run its preprocessor over user
13696 assembler files (with a @samp{.s} suffix) when assembling them.
13699 @itemx -mno-fix-r4000
13700 @opindex mfix-r4000
13701 @opindex mno-fix-r4000
13702 Work around certain R4000 CPU errata:
13705 A double-word or a variable shift may give an incorrect result if executed
13706 immediately after starting an integer division.
13708 A double-word or a variable shift may give an incorrect result if executed
13709 while an integer multiplication is in progress.
13711 An integer division may give an incorrect result if started in a delay slot
13712 of a taken branch or a jump.
13716 @itemx -mno-fix-r4400
13717 @opindex mfix-r4400
13718 @opindex mno-fix-r4400
13719 Work around certain R4400 CPU errata:
13722 A double-word or a variable shift may give an incorrect result if executed
13723 immediately after starting an integer division.
13727 @itemx -mno-fix-r10000
13728 @opindex mfix-r10000
13729 @opindex mno-fix-r10000
13730 Work around certain R10000 errata:
13733 @code{ll}/@code{sc} sequences may not behave atomically on revisions
13734 prior to 3.0. They may deadlock on revisions 2.6 and earlier.
13737 This option can only be used if the target architecture supports
13738 branch-likely instructions. @option{-mfix-r10000} is the default when
13739 @option{-march=r10000} is used; @option{-mno-fix-r10000} is the default
13743 @itemx -mno-fix-vr4120
13744 @opindex mfix-vr4120
13745 Work around certain VR4120 errata:
13748 @code{dmultu} does not always produce the correct result.
13750 @code{div} and @code{ddiv} do not always produce the correct result if one
13751 of the operands is negative.
13753 The workarounds for the division errata rely on special functions in
13754 @file{libgcc.a}. At present, these functions are only provided by
13755 the @code{mips64vr*-elf} configurations.
13757 Other VR4120 errata require a nop to be inserted between certain pairs of
13758 instructions. These errata are handled by the assembler, not by GCC itself.
13761 @opindex mfix-vr4130
13762 Work around the VR4130 @code{mflo}/@code{mfhi} errata. The
13763 workarounds are implemented by the assembler rather than by GCC,
13764 although GCC will avoid using @code{mflo} and @code{mfhi} if the
13765 VR4130 @code{macc}, @code{macchi}, @code{dmacc} and @code{dmacchi}
13766 instructions are available instead.
13769 @itemx -mno-fix-sb1
13771 Work around certain SB-1 CPU core errata.
13772 (This flag currently works around the SB-1 revision 2
13773 ``F1'' and ``F2'' floating point errata.)
13775 @item -mr10k-cache-barrier=@var{setting}
13776 @opindex mr10k-cache-barrier
13777 Specify whether GCC should insert cache barriers to avoid the
13778 side-effects of speculation on R10K processors.
13780 In common with many processors, the R10K tries to predict the outcome
13781 of a conditional branch and speculatively executes instructions from
13782 the ``taken'' branch. It later aborts these instructions if the
13783 predicted outcome was wrong. However, on the R10K, even aborted
13784 instructions can have side effects.
13786 This problem only affects kernel stores and, depending on the system,
13787 kernel loads. As an example, a speculatively-executed store may load
13788 the target memory into cache and mark the cache line as dirty, even if
13789 the store itself is later aborted. If a DMA operation writes to the
13790 same area of memory before the ``dirty'' line is flushed, the cached
13791 data will overwrite the DMA-ed data. See the R10K processor manual
13792 for a full description, including other potential problems.
13794 One workaround is to insert cache barrier instructions before every memory
13795 access that might be speculatively executed and that might have side
13796 effects even if aborted. @option{-mr10k-cache-barrier=@var{setting}}
13797 controls GCC's implementation of this workaround. It assumes that
13798 aborted accesses to any byte in the following regions will not have
13803 the memory occupied by the current function's stack frame;
13806 the memory occupied by an incoming stack argument;
13809 the memory occupied by an object with a link-time-constant address.
13812 It is the kernel's responsibility to ensure that speculative
13813 accesses to these regions are indeed safe.
13815 If the input program contains a function declaration such as:
13821 then the implementation of @code{foo} must allow @code{j foo} and
13822 @code{jal foo} to be executed speculatively. GCC honors this
13823 restriction for functions it compiles itself. It expects non-GCC
13824 functions (such as hand-written assembly code) to do the same.
13826 The option has three forms:
13829 @item -mr10k-cache-barrier=load-store
13830 Insert a cache barrier before a load or store that might be
13831 speculatively executed and that might have side effects even
13834 @item -mr10k-cache-barrier=store
13835 Insert a cache barrier before a store that might be speculatively
13836 executed and that might have side effects even if aborted.
13838 @item -mr10k-cache-barrier=none
13839 Disable the insertion of cache barriers. This is the default setting.
13842 @item -mflush-func=@var{func}
13843 @itemx -mno-flush-func
13844 @opindex mflush-func
13845 Specifies the function to call to flush the I and D caches, or to not
13846 call any such function. If called, the function must take the same
13847 arguments as the common @code{_flush_func()}, that is, the address of the
13848 memory range for which the cache is being flushed, the size of the
13849 memory range, and the number 3 (to flush both caches). The default
13850 depends on the target GCC was configured for, but commonly is either
13851 @samp{_flush_func} or @samp{__cpu_flush}.
13853 @item mbranch-cost=@var{num}
13854 @opindex mbranch-cost
13855 Set the cost of branches to roughly @var{num} ``simple'' instructions.
13856 This cost is only a heuristic and is not guaranteed to produce
13857 consistent results across releases. A zero cost redundantly selects
13858 the default, which is based on the @option{-mtune} setting.
13860 @item -mbranch-likely
13861 @itemx -mno-branch-likely
13862 @opindex mbranch-likely
13863 @opindex mno-branch-likely
13864 Enable or disable use of Branch Likely instructions, regardless of the
13865 default for the selected architecture. By default, Branch Likely
13866 instructions may be generated if they are supported by the selected
13867 architecture. An exception is for the MIPS32 and MIPS64 architectures
13868 and processors which implement those architectures; for those, Branch
13869 Likely instructions will not be generated by default because the MIPS32
13870 and MIPS64 architectures specifically deprecate their use.
13872 @item -mfp-exceptions
13873 @itemx -mno-fp-exceptions
13874 @opindex mfp-exceptions
13875 Specifies whether FP exceptions are enabled. This affects how we schedule
13876 FP instructions for some processors. The default is that FP exceptions are
13879 For instance, on the SB-1, if FP exceptions are disabled, and we are emitting
13880 64-bit code, then we can use both FP pipes. Otherwise, we can only use one
13883 @item -mvr4130-align
13884 @itemx -mno-vr4130-align
13885 @opindex mvr4130-align
13886 The VR4130 pipeline is two-way superscalar, but can only issue two
13887 instructions together if the first one is 8-byte aligned. When this
13888 option is enabled, GCC will align pairs of instructions that it
13889 thinks should execute in parallel.
13891 This option only has an effect when optimizing for the VR4130.
13892 It normally makes code faster, but at the expense of making it bigger.
13893 It is enabled by default at optimization level @option{-O3}.
13898 Enable (disable) generation of @code{synci} instructions on
13899 architectures that support it. The @code{synci} instructions (if
13900 enabled) will be generated when @code{__builtin___clear_cache()} is
13903 This option defaults to @code{-mno-synci}, but the default can be
13904 overridden by configuring with @code{--with-synci}.
13906 When compiling code for single processor systems, it is generally safe
13907 to use @code{synci}. However, on many multi-core (SMP) systems, it
13908 will not invalidate the instruction caches on all cores and may lead
13909 to undefined behavior.
13911 @item -mrelax-pic-calls
13912 @itemx -mno-relax-pic-calls
13913 @opindex mrelax-pic-calls
13914 Try to turn PIC calls that are normally dispatched via register
13915 @code{$25} into direct calls. This is only possible if the linker can
13916 resolve the destination at link-time and if the destination is within
13917 range for a direct call.
13919 @option{-mrelax-pic-calls} is the default if GCC was configured to use
13920 an assembler and a linker that supports the @code{.reloc} assembly
13921 directive and @code{-mexplicit-relocs} is in effect. With
13922 @code{-mno-explicit-relocs}, this optimization can be performed by the
13923 assembler and the linker alone without help from the compiler.
13927 @subsection MMIX Options
13928 @cindex MMIX Options
13930 These options are defined for the MMIX:
13934 @itemx -mno-libfuncs
13936 @opindex mno-libfuncs
13937 Specify that intrinsic library functions are being compiled, passing all
13938 values in registers, no matter the size.
13941 @itemx -mno-epsilon
13943 @opindex mno-epsilon
13944 Generate floating-point comparison instructions that compare with respect
13945 to the @code{rE} epsilon register.
13947 @item -mabi=mmixware
13949 @opindex mabi=mmixware
13951 Generate code that passes function parameters and return values that (in
13952 the called function) are seen as registers @code{$0} and up, as opposed to
13953 the GNU ABI which uses global registers @code{$231} and up.
13955 @item -mzero-extend
13956 @itemx -mno-zero-extend
13957 @opindex mzero-extend
13958 @opindex mno-zero-extend
13959 When reading data from memory in sizes shorter than 64 bits, use (do not
13960 use) zero-extending load instructions by default, rather than
13961 sign-extending ones.
13964 @itemx -mno-knuthdiv
13966 @opindex mno-knuthdiv
13967 Make the result of a division yielding a remainder have the same sign as
13968 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
13969 remainder follows the sign of the dividend. Both methods are
13970 arithmetically valid, the latter being almost exclusively used.
13972 @item -mtoplevel-symbols
13973 @itemx -mno-toplevel-symbols
13974 @opindex mtoplevel-symbols
13975 @opindex mno-toplevel-symbols
13976 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
13977 code can be used with the @code{PREFIX} assembly directive.
13981 Generate an executable in the ELF format, rather than the default
13982 @samp{mmo} format used by the @command{mmix} simulator.
13984 @item -mbranch-predict
13985 @itemx -mno-branch-predict
13986 @opindex mbranch-predict
13987 @opindex mno-branch-predict
13988 Use (do not use) the probable-branch instructions, when static branch
13989 prediction indicates a probable branch.
13991 @item -mbase-addresses
13992 @itemx -mno-base-addresses
13993 @opindex mbase-addresses
13994 @opindex mno-base-addresses
13995 Generate (do not generate) code that uses @emph{base addresses}. Using a
13996 base address automatically generates a request (handled by the assembler
13997 and the linker) for a constant to be set up in a global register. The
13998 register is used for one or more base address requests within the range 0
13999 to 255 from the value held in the register. The generally leads to short
14000 and fast code, but the number of different data items that can be
14001 addressed is limited. This means that a program that uses lots of static
14002 data may require @option{-mno-base-addresses}.
14004 @item -msingle-exit
14005 @itemx -mno-single-exit
14006 @opindex msingle-exit
14007 @opindex mno-single-exit
14008 Force (do not force) generated code to have a single exit point in each
14012 @node MN10300 Options
14013 @subsection MN10300 Options
14014 @cindex MN10300 options
14016 These @option{-m} options are defined for Matsushita MN10300 architectures:
14021 Generate code to avoid bugs in the multiply instructions for the MN10300
14022 processors. This is the default.
14024 @item -mno-mult-bug
14025 @opindex mno-mult-bug
14026 Do not generate code to avoid bugs in the multiply instructions for the
14027 MN10300 processors.
14031 Generate code which uses features specific to the AM33 processor.
14035 Do not generate code which uses features specific to the AM33 processor. This
14038 @item -mreturn-pointer-on-d0
14039 @opindex mreturn-pointer-on-d0
14040 When generating a function which returns a pointer, return the pointer
14041 in both @code{a0} and @code{d0}. Otherwise, the pointer is returned
14042 only in a0, and attempts to call such functions without a prototype
14043 would result in errors. Note that this option is on by default; use
14044 @option{-mno-return-pointer-on-d0} to disable it.
14048 Do not link in the C run-time initialization object file.
14052 Indicate to the linker that it should perform a relaxation optimization pass
14053 to shorten branches, calls and absolute memory addresses. This option only
14054 has an effect when used on the command line for the final link step.
14056 This option makes symbolic debugging impossible.
14059 @node PDP-11 Options
14060 @subsection PDP-11 Options
14061 @cindex PDP-11 Options
14063 These options are defined for the PDP-11:
14068 Use hardware FPP floating point. This is the default. (FIS floating
14069 point on the PDP-11/40 is not supported.)
14072 @opindex msoft-float
14073 Do not use hardware floating point.
14077 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
14081 Return floating-point results in memory. This is the default.
14085 Generate code for a PDP-11/40.
14089 Generate code for a PDP-11/45. This is the default.
14093 Generate code for a PDP-11/10.
14095 @item -mbcopy-builtin
14096 @opindex mbcopy-builtin
14097 Use inline @code{movmemhi} patterns for copying memory. This is the
14102 Do not use inline @code{movmemhi} patterns for copying memory.
14108 Use 16-bit @code{int}. This is the default.
14114 Use 32-bit @code{int}.
14117 @itemx -mno-float32
14119 @opindex mno-float32
14120 Use 64-bit @code{float}. This is the default.
14123 @itemx -mno-float64
14125 @opindex mno-float64
14126 Use 32-bit @code{float}.
14130 Use @code{abshi2} pattern. This is the default.
14134 Do not use @code{abshi2} pattern.
14136 @item -mbranch-expensive
14137 @opindex mbranch-expensive
14138 Pretend that branches are expensive. This is for experimenting with
14139 code generation only.
14141 @item -mbranch-cheap
14142 @opindex mbranch-cheap
14143 Do not pretend that branches are expensive. This is the default.
14147 Generate code for a system with split I&D@.
14151 Generate code for a system without split I&D@. This is the default.
14155 Use Unix assembler syntax. This is the default when configured for
14156 @samp{pdp11-*-bsd}.
14160 Use DEC assembler syntax. This is the default when configured for any
14161 PDP-11 target other than @samp{pdp11-*-bsd}.
14164 @node picoChip Options
14165 @subsection picoChip Options
14166 @cindex picoChip options
14168 These @samp{-m} options are defined for picoChip implementations:
14172 @item -mae=@var{ae_type}
14174 Set the instruction set, register set, and instruction scheduling
14175 parameters for array element type @var{ae_type}. Supported values
14176 for @var{ae_type} are @samp{ANY}, @samp{MUL}, and @samp{MAC}.
14178 @option{-mae=ANY} selects a completely generic AE type. Code
14179 generated with this option will run on any of the other AE types. The
14180 code will not be as efficient as it would be if compiled for a specific
14181 AE type, and some types of operation (e.g., multiplication) will not
14182 work properly on all types of AE.
14184 @option{-mae=MUL} selects a MUL AE type. This is the most useful AE type
14185 for compiled code, and is the default.
14187 @option{-mae=MAC} selects a DSP-style MAC AE. Code compiled with this
14188 option may suffer from poor performance of byte (char) manipulation,
14189 since the DSP AE does not provide hardware support for byte load/stores.
14191 @item -msymbol-as-address
14192 Enable the compiler to directly use a symbol name as an address in a
14193 load/store instruction, without first loading it into a
14194 register. Typically, the use of this option will generate larger
14195 programs, which run faster than when the option isn't used. However, the
14196 results vary from program to program, so it is left as a user option,
14197 rather than being permanently enabled.
14199 @item -mno-inefficient-warnings
14200 Disables warnings about the generation of inefficient code. These
14201 warnings can be generated, for example, when compiling code which
14202 performs byte-level memory operations on the MAC AE type. The MAC AE has
14203 no hardware support for byte-level memory operations, so all byte
14204 load/stores must be synthesized from word load/store operations. This is
14205 inefficient and a warning will be generated indicating to the programmer
14206 that they should rewrite the code to avoid byte operations, or to target
14207 an AE type which has the necessary hardware support. This option enables
14208 the warning to be turned off.
14212 @node PowerPC Options
14213 @subsection PowerPC Options
14214 @cindex PowerPC options
14216 These are listed under @xref{RS/6000 and PowerPC Options}.
14218 @node RS/6000 and PowerPC Options
14219 @subsection IBM RS/6000 and PowerPC Options
14220 @cindex RS/6000 and PowerPC Options
14221 @cindex IBM RS/6000 and PowerPC Options
14223 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
14230 @itemx -mno-powerpc
14231 @itemx -mpowerpc-gpopt
14232 @itemx -mno-powerpc-gpopt
14233 @itemx -mpowerpc-gfxopt
14234 @itemx -mno-powerpc-gfxopt
14236 @itemx -mno-powerpc64
14240 @itemx -mno-popcntb
14242 @itemx -mno-popcntd
14250 @itemx -mno-hard-dfp
14254 @opindex mno-power2
14256 @opindex mno-powerpc
14257 @opindex mpowerpc-gpopt
14258 @opindex mno-powerpc-gpopt
14259 @opindex mpowerpc-gfxopt
14260 @opindex mno-powerpc-gfxopt
14261 @opindex mpowerpc64
14262 @opindex mno-powerpc64
14266 @opindex mno-popcntb
14268 @opindex mno-popcntd
14274 @opindex mno-mfpgpr
14276 @opindex mno-hard-dfp
14277 GCC supports two related instruction set architectures for the
14278 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
14279 instructions supported by the @samp{rios} chip set used in the original
14280 RS/6000 systems and the @dfn{PowerPC} instruction set is the
14281 architecture of the Freescale MPC5xx, MPC6xx, MPC8xx microprocessors, and
14282 the IBM 4xx, 6xx, and follow-on microprocessors.
14284 Neither architecture is a subset of the other. However there is a
14285 large common subset of instructions supported by both. An MQ
14286 register is included in processors supporting the POWER architecture.
14288 You use these options to specify which instructions are available on the
14289 processor you are using. The default value of these options is
14290 determined when configuring GCC@. Specifying the
14291 @option{-mcpu=@var{cpu_type}} overrides the specification of these
14292 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
14293 rather than the options listed above.
14295 The @option{-mpower} option allows GCC to generate instructions that
14296 are found only in the POWER architecture and to use the MQ register.
14297 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
14298 to generate instructions that are present in the POWER2 architecture but
14299 not the original POWER architecture.
14301 The @option{-mpowerpc} option allows GCC to generate instructions that
14302 are found only in the 32-bit subset of the PowerPC architecture.
14303 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
14304 GCC to use the optional PowerPC architecture instructions in the
14305 General Purpose group, including floating-point square root. Specifying
14306 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
14307 use the optional PowerPC architecture instructions in the Graphics
14308 group, including floating-point select.
14310 The @option{-mmfcrf} option allows GCC to generate the move from
14311 condition register field instruction implemented on the POWER4
14312 processor and other processors that support the PowerPC V2.01
14314 The @option{-mpopcntb} option allows GCC to generate the popcount and
14315 double precision FP reciprocal estimate instruction implemented on the
14316 POWER5 processor and other processors that support the PowerPC V2.02
14318 The @option{-mpopcntd} option allows GCC to generate the popcount
14319 instruction implemented on the POWER7 processor and other processors
14320 that support the PowerPC V2.06 architecture.
14321 The @option{-mfprnd} option allows GCC to generate the FP round to
14322 integer instructions implemented on the POWER5+ processor and other
14323 processors that support the PowerPC V2.03 architecture.
14324 The @option{-mcmpb} option allows GCC to generate the compare bytes
14325 instruction implemented on the POWER6 processor and other processors
14326 that support the PowerPC V2.05 architecture.
14327 The @option{-mmfpgpr} option allows GCC to generate the FP move to/from
14328 general purpose register instructions implemented on the POWER6X
14329 processor and other processors that support the extended PowerPC V2.05
14331 The @option{-mhard-dfp} option allows GCC to generate the decimal floating
14332 point instructions implemented on some POWER processors.
14334 The @option{-mpowerpc64} option allows GCC to generate the additional
14335 64-bit instructions that are found in the full PowerPC64 architecture
14336 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
14337 @option{-mno-powerpc64}.
14339 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
14340 will use only the instructions in the common subset of both
14341 architectures plus some special AIX common-mode calls, and will not use
14342 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
14343 permits GCC to use any instruction from either architecture and to
14344 allow use of the MQ register; specify this for the Motorola MPC601.
14346 @item -mnew-mnemonics
14347 @itemx -mold-mnemonics
14348 @opindex mnew-mnemonics
14349 @opindex mold-mnemonics
14350 Select which mnemonics to use in the generated assembler code. With
14351 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
14352 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
14353 assembler mnemonics defined for the POWER architecture. Instructions
14354 defined in only one architecture have only one mnemonic; GCC uses that
14355 mnemonic irrespective of which of these options is specified.
14357 GCC defaults to the mnemonics appropriate for the architecture in
14358 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
14359 value of these option. Unless you are building a cross-compiler, you
14360 should normally not specify either @option{-mnew-mnemonics} or
14361 @option{-mold-mnemonics}, but should instead accept the default.
14363 @item -mcpu=@var{cpu_type}
14365 Set architecture type, register usage, choice of mnemonics, and
14366 instruction scheduling parameters for machine type @var{cpu_type}.
14367 Supported values for @var{cpu_type} are @samp{401}, @samp{403},
14368 @samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{464}, @samp{464fp},
14369 @samp{505}, @samp{601}, @samp{602}, @samp{603}, @samp{603e}, @samp{604},
14370 @samp{604e}, @samp{620}, @samp{630}, @samp{740}, @samp{7400},
14371 @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
14372 @samp{860}, @samp{970}, @samp{8540}, @samp{e300c2}, @samp{e300c3},
14373 @samp{e500mc}, @samp{ec603e}, @samp{G3}, @samp{G4}, @samp{G5},
14374 @samp{power}, @samp{power2}, @samp{power3}, @samp{power4},
14375 @samp{power5}, @samp{power5+}, @samp{power6}, @samp{power6x}, @samp{power7}
14376 @samp{common}, @samp{powerpc}, @samp{powerpc64}, @samp{rios},
14377 @samp{rios1}, @samp{rios2}, @samp{rsc}, and @samp{rs64}.
14379 @option{-mcpu=common} selects a completely generic processor. Code
14380 generated under this option will run on any POWER or PowerPC processor.
14381 GCC will use only the instructions in the common subset of both
14382 architectures, and will not use the MQ register. GCC assumes a generic
14383 processor model for scheduling purposes.
14385 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
14386 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
14387 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
14388 types, with an appropriate, generic processor model assumed for
14389 scheduling purposes.
14391 The other options specify a specific processor. Code generated under
14392 those options will run best on that processor, and may not run at all on
14395 The @option{-mcpu} options automatically enable or disable the
14398 @gccoptlist{-maltivec -mfprnd -mhard-float -mmfcrf -mmultiple @gol
14399 -mnew-mnemonics -mpopcntb -mpopcntd -mpower -mpower2 -mpowerpc64 @gol
14400 -mpowerpc-gpopt -mpowerpc-gfxopt -msingle-float -mdouble-float @gol
14401 -msimple-fpu -mstring -mmulhw -mdlmzb -mmfpgpr -mvsx}
14403 The particular options set for any particular CPU will vary between
14404 compiler versions, depending on what setting seems to produce optimal
14405 code for that CPU; it doesn't necessarily reflect the actual hardware's
14406 capabilities. If you wish to set an individual option to a particular
14407 value, you may specify it after the @option{-mcpu} option, like
14408 @samp{-mcpu=970 -mno-altivec}.
14410 On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
14411 not enabled or disabled by the @option{-mcpu} option at present because
14412 AIX does not have full support for these options. You may still
14413 enable or disable them individually if you're sure it'll work in your
14416 @item -mtune=@var{cpu_type}
14418 Set the instruction scheduling parameters for machine type
14419 @var{cpu_type}, but do not set the architecture type, register usage, or
14420 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
14421 values for @var{cpu_type} are used for @option{-mtune} as for
14422 @option{-mcpu}. If both are specified, the code generated will use the
14423 architecture, registers, and mnemonics set by @option{-mcpu}, but the
14424 scheduling parameters set by @option{-mtune}.
14430 Generate code to compute division as reciprocal estimate and iterative
14431 refinement, creating opportunities for increased throughput. This
14432 feature requires: optional PowerPC Graphics instruction set for single
14433 precision and FRE instruction for double precision, assuming divides
14434 cannot generate user-visible traps, and the domain values not include
14435 Infinities, denormals or zero denominator.
14438 @itemx -mno-altivec
14440 @opindex mno-altivec
14441 Generate code that uses (does not use) AltiVec instructions, and also
14442 enable the use of built-in functions that allow more direct access to
14443 the AltiVec instruction set. You may also need to set
14444 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
14450 @opindex mno-vrsave
14451 Generate VRSAVE instructions when generating AltiVec code.
14453 @item -mgen-cell-microcode
14454 @opindex mgen-cell-microcode
14455 Generate Cell microcode instructions
14457 @item -mwarn-cell-microcode
14458 @opindex mwarn-cell-microcode
14459 Warning when a Cell microcode instruction is going to emitted. An example
14460 of a Cell microcode instruction is a variable shift.
14463 @opindex msecure-plt
14464 Generate code that allows ld and ld.so to build executables and shared
14465 libraries with non-exec .plt and .got sections. This is a PowerPC
14466 32-bit SYSV ABI option.
14470 Generate code that uses a BSS .plt section that ld.so fills in, and
14471 requires .plt and .got sections that are both writable and executable.
14472 This is a PowerPC 32-bit SYSV ABI option.
14478 This switch enables or disables the generation of ISEL instructions.
14480 @item -misel=@var{yes/no}
14481 This switch has been deprecated. Use @option{-misel} and
14482 @option{-mno-isel} instead.
14488 This switch enables or disables the generation of SPE simd
14494 @opindex mno-paired
14495 This switch enables or disables the generation of PAIRED simd
14498 @item -mspe=@var{yes/no}
14499 This option has been deprecated. Use @option{-mspe} and
14500 @option{-mno-spe} instead.
14506 Generate code that uses (does not use) vector/scalar (VSX)
14507 instructions, and also enable the use of built-in functions that allow
14508 more direct access to the VSX instruction set.
14510 @item -mfloat-gprs=@var{yes/single/double/no}
14511 @itemx -mfloat-gprs
14512 @opindex mfloat-gprs
14513 This switch enables or disables the generation of floating point
14514 operations on the general purpose registers for architectures that
14517 The argument @var{yes} or @var{single} enables the use of
14518 single-precision floating point operations.
14520 The argument @var{double} enables the use of single and
14521 double-precision floating point operations.
14523 The argument @var{no} disables floating point operations on the
14524 general purpose registers.
14526 This option is currently only available on the MPC854x.
14532 Generate code for 32-bit or 64-bit environments of Darwin and SVR4
14533 targets (including GNU/Linux). The 32-bit environment sets int, long
14534 and pointer to 32 bits and generates code that runs on any PowerPC
14535 variant. The 64-bit environment sets int to 32 bits and long and
14536 pointer to 64 bits, and generates code for PowerPC64, as for
14537 @option{-mpowerpc64}.
14540 @itemx -mno-fp-in-toc
14541 @itemx -mno-sum-in-toc
14542 @itemx -mminimal-toc
14544 @opindex mno-fp-in-toc
14545 @opindex mno-sum-in-toc
14546 @opindex mminimal-toc
14547 Modify generation of the TOC (Table Of Contents), which is created for
14548 every executable file. The @option{-mfull-toc} option is selected by
14549 default. In that case, GCC will allocate at least one TOC entry for
14550 each unique non-automatic variable reference in your program. GCC
14551 will also place floating-point constants in the TOC@. However, only
14552 16,384 entries are available in the TOC@.
14554 If you receive a linker error message that saying you have overflowed
14555 the available TOC space, you can reduce the amount of TOC space used
14556 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
14557 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
14558 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
14559 generate code to calculate the sum of an address and a constant at
14560 run-time instead of putting that sum into the TOC@. You may specify one
14561 or both of these options. Each causes GCC to produce very slightly
14562 slower and larger code at the expense of conserving TOC space.
14564 If you still run out of space in the TOC even when you specify both of
14565 these options, specify @option{-mminimal-toc} instead. This option causes
14566 GCC to make only one TOC entry for every file. When you specify this
14567 option, GCC will produce code that is slower and larger but which
14568 uses extremely little TOC space. You may wish to use this option
14569 only on files that contain less frequently executed code.
14575 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
14576 @code{long} type, and the infrastructure needed to support them.
14577 Specifying @option{-maix64} implies @option{-mpowerpc64} and
14578 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
14579 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
14582 @itemx -mno-xl-compat
14583 @opindex mxl-compat
14584 @opindex mno-xl-compat
14585 Produce code that conforms more closely to IBM XL compiler semantics
14586 when using AIX-compatible ABI@. Pass floating-point arguments to
14587 prototyped functions beyond the register save area (RSA) on the stack
14588 in addition to argument FPRs. Do not assume that most significant
14589 double in 128-bit long double value is properly rounded when comparing
14590 values and converting to double. Use XL symbol names for long double
14593 The AIX calling convention was extended but not initially documented to
14594 handle an obscure K&R C case of calling a function that takes the
14595 address of its arguments with fewer arguments than declared. IBM XL
14596 compilers access floating point arguments which do not fit in the
14597 RSA from the stack when a subroutine is compiled without
14598 optimization. Because always storing floating-point arguments on the
14599 stack is inefficient and rarely needed, this option is not enabled by
14600 default and only is necessary when calling subroutines compiled by IBM
14601 XL compilers without optimization.
14605 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
14606 application written to use message passing with special startup code to
14607 enable the application to run. The system must have PE installed in the
14608 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
14609 must be overridden with the @option{-specs=} option to specify the
14610 appropriate directory location. The Parallel Environment does not
14611 support threads, so the @option{-mpe} option and the @option{-pthread}
14612 option are incompatible.
14614 @item -malign-natural
14615 @itemx -malign-power
14616 @opindex malign-natural
14617 @opindex malign-power
14618 On AIX, 32-bit Darwin, and 64-bit PowerPC GNU/Linux, the option
14619 @option{-malign-natural} overrides the ABI-defined alignment of larger
14620 types, such as floating-point doubles, on their natural size-based boundary.
14621 The option @option{-malign-power} instructs GCC to follow the ABI-specified
14622 alignment rules. GCC defaults to the standard alignment defined in the ABI@.
14624 On 64-bit Darwin, natural alignment is the default, and @option{-malign-power}
14628 @itemx -mhard-float
14629 @opindex msoft-float
14630 @opindex mhard-float
14631 Generate code that does not use (uses) the floating-point register set.
14632 Software floating point emulation is provided if you use the
14633 @option{-msoft-float} option, and pass the option to GCC when linking.
14635 @item -msingle-float
14636 @itemx -mdouble-float
14637 @opindex msingle-float
14638 @opindex mdouble-float
14639 Generate code for single or double-precision floating point operations.
14640 @option{-mdouble-float} implies @option{-msingle-float}.
14643 @opindex msimple-fpu
14644 Do not generate sqrt and div instructions for hardware floating point unit.
14648 Specify type of floating point unit. Valid values are @var{sp_lite}
14649 (equivalent to -msingle-float -msimple-fpu), @var{dp_lite} (equivalent
14650 to -mdouble-float -msimple-fpu), @var{sp_full} (equivalent to -msingle-float),
14651 and @var{dp_full} (equivalent to -mdouble-float).
14654 @opindex mxilinx-fpu
14655 Perform optimizations for floating point unit on Xilinx PPC 405/440.
14658 @itemx -mno-multiple
14660 @opindex mno-multiple
14661 Generate code that uses (does not use) the load multiple word
14662 instructions and the store multiple word instructions. These
14663 instructions are generated by default on POWER systems, and not
14664 generated on PowerPC systems. Do not use @option{-mmultiple} on little
14665 endian PowerPC systems, since those instructions do not work when the
14666 processor is in little endian mode. The exceptions are PPC740 and
14667 PPC750 which permit the instructions usage in little endian mode.
14672 @opindex mno-string
14673 Generate code that uses (does not use) the load string instructions
14674 and the store string word instructions to save multiple registers and
14675 do small block moves. These instructions are generated by default on
14676 POWER systems, and not generated on PowerPC systems. Do not use
14677 @option{-mstring} on little endian PowerPC systems, since those
14678 instructions do not work when the processor is in little endian mode.
14679 The exceptions are PPC740 and PPC750 which permit the instructions
14680 usage in little endian mode.
14685 @opindex mno-update
14686 Generate code that uses (does not use) the load or store instructions
14687 that update the base register to the address of the calculated memory
14688 location. These instructions are generated by default. If you use
14689 @option{-mno-update}, there is a small window between the time that the
14690 stack pointer is updated and the address of the previous frame is
14691 stored, which means code that walks the stack frame across interrupts or
14692 signals may get corrupted data.
14694 @item -mavoid-indexed-addresses
14695 @item -mno-avoid-indexed-addresses
14696 @opindex mavoid-indexed-addresses
14697 @opindex mno-avoid-indexed-addresses
14698 Generate code that tries to avoid (not avoid) the use of indexed load
14699 or store instructions. These instructions can incur a performance
14700 penalty on Power6 processors in certain situations, such as when
14701 stepping through large arrays that cross a 16M boundary. This option
14702 is enabled by default when targetting Power6 and disabled otherwise.
14705 @itemx -mno-fused-madd
14706 @opindex mfused-madd
14707 @opindex mno-fused-madd
14708 Generate code that uses (does not use) the floating point multiply and
14709 accumulate instructions. These instructions are generated by default if
14710 hardware floating is used.
14716 Generate code that uses (does not use) the half-word multiply and
14717 multiply-accumulate instructions on the IBM 405, 440 and 464 processors.
14718 These instructions are generated by default when targetting those
14725 Generate code that uses (does not use) the string-search @samp{dlmzb}
14726 instruction on the IBM 405, 440 and 464 processors. This instruction is
14727 generated by default when targetting those processors.
14729 @item -mno-bit-align
14731 @opindex mno-bit-align
14732 @opindex mbit-align
14733 On System V.4 and embedded PowerPC systems do not (do) force structures
14734 and unions that contain bit-fields to be aligned to the base type of the
14737 For example, by default a structure containing nothing but 8
14738 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
14739 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
14740 the structure would be aligned to a 1 byte boundary and be one byte in
14743 @item -mno-strict-align
14744 @itemx -mstrict-align
14745 @opindex mno-strict-align
14746 @opindex mstrict-align
14747 On System V.4 and embedded PowerPC systems do not (do) assume that
14748 unaligned memory references will be handled by the system.
14750 @item -mrelocatable
14751 @itemx -mno-relocatable
14752 @opindex mrelocatable
14753 @opindex mno-relocatable
14754 On embedded PowerPC systems generate code that allows (does not allow)
14755 the program to be relocated to a different address at runtime. If you
14756 use @option{-mrelocatable} on any module, all objects linked together must
14757 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
14759 @item -mrelocatable-lib
14760 @itemx -mno-relocatable-lib
14761 @opindex mrelocatable-lib
14762 @opindex mno-relocatable-lib
14763 On embedded PowerPC systems generate code that allows (does not allow)
14764 the program to be relocated to a different address at runtime. Modules
14765 compiled with @option{-mrelocatable-lib} can be linked with either modules
14766 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
14767 with modules compiled with the @option{-mrelocatable} options.
14773 On System V.4 and embedded PowerPC systems do not (do) assume that
14774 register 2 contains a pointer to a global area pointing to the addresses
14775 used in the program.
14778 @itemx -mlittle-endian
14780 @opindex mlittle-endian
14781 On System V.4 and embedded PowerPC systems compile code for the
14782 processor in little endian mode. The @option{-mlittle-endian} option is
14783 the same as @option{-mlittle}.
14786 @itemx -mbig-endian
14788 @opindex mbig-endian
14789 On System V.4 and embedded PowerPC systems compile code for the
14790 processor in big endian mode. The @option{-mbig-endian} option is
14791 the same as @option{-mbig}.
14793 @item -mdynamic-no-pic
14794 @opindex mdynamic-no-pic
14795 On Darwin and Mac OS X systems, compile code so that it is not
14796 relocatable, but that its external references are relocatable. The
14797 resulting code is suitable for applications, but not shared
14800 @item -mprioritize-restricted-insns=@var{priority}
14801 @opindex mprioritize-restricted-insns
14802 This option controls the priority that is assigned to
14803 dispatch-slot restricted instructions during the second scheduling
14804 pass. The argument @var{priority} takes the value @var{0/1/2} to assign
14805 @var{no/highest/second-highest} priority to dispatch slot restricted
14808 @item -msched-costly-dep=@var{dependence_type}
14809 @opindex msched-costly-dep
14810 This option controls which dependences are considered costly
14811 by the target during instruction scheduling. The argument
14812 @var{dependence_type} takes one of the following values:
14813 @var{no}: no dependence is costly,
14814 @var{all}: all dependences are costly,
14815 @var{true_store_to_load}: a true dependence from store to load is costly,
14816 @var{store_to_load}: any dependence from store to load is costly,
14817 @var{number}: any dependence which latency >= @var{number} is costly.
14819 @item -minsert-sched-nops=@var{scheme}
14820 @opindex minsert-sched-nops
14821 This option controls which nop insertion scheme will be used during
14822 the second scheduling pass. The argument @var{scheme} takes one of the
14824 @var{no}: Don't insert nops.
14825 @var{pad}: Pad with nops any dispatch group which has vacant issue slots,
14826 according to the scheduler's grouping.
14827 @var{regroup_exact}: Insert nops to force costly dependent insns into
14828 separate groups. Insert exactly as many nops as needed to force an insn
14829 to a new group, according to the estimated processor grouping.
14830 @var{number}: Insert nops to force costly dependent insns into
14831 separate groups. Insert @var{number} nops to force an insn to a new group.
14834 @opindex mcall-sysv
14835 On System V.4 and embedded PowerPC systems compile code using calling
14836 conventions that adheres to the March 1995 draft of the System V
14837 Application Binary Interface, PowerPC processor supplement. This is the
14838 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
14840 @item -mcall-sysv-eabi
14842 @opindex mcall-sysv-eabi
14843 @opindex mcall-eabi
14844 Specify both @option{-mcall-sysv} and @option{-meabi} options.
14846 @item -mcall-sysv-noeabi
14847 @opindex mcall-sysv-noeabi
14848 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
14850 @item -mcall-aixdesc
14852 On System V.4 and embedded PowerPC systems compile code for the AIX
14856 @opindex mcall-linux
14857 On System V.4 and embedded PowerPC systems compile code for the
14858 Linux-based GNU system.
14862 On System V.4 and embedded PowerPC systems compile code for the
14863 Hurd-based GNU system.
14865 @item -mcall-freebsd
14866 @opindex mcall-freebsd
14867 On System V.4 and embedded PowerPC systems compile code for the
14868 FreeBSD operating system.
14870 @item -mcall-netbsd
14871 @opindex mcall-netbsd
14872 On System V.4 and embedded PowerPC systems compile code for the
14873 NetBSD operating system.
14875 @item -mcall-openbsd
14876 @opindex mcall-netbsd
14877 On System V.4 and embedded PowerPC systems compile code for the
14878 OpenBSD operating system.
14880 @item -maix-struct-return
14881 @opindex maix-struct-return
14882 Return all structures in memory (as specified by the AIX ABI)@.
14884 @item -msvr4-struct-return
14885 @opindex msvr4-struct-return
14886 Return structures smaller than 8 bytes in registers (as specified by the
14889 @item -mabi=@var{abi-type}
14891 Extend the current ABI with a particular extension, or remove such extension.
14892 Valid values are @var{altivec}, @var{no-altivec}, @var{spe},
14893 @var{no-spe}, @var{ibmlongdouble}, @var{ieeelongdouble}@.
14897 Extend the current ABI with SPE ABI extensions. This does not change
14898 the default ABI, instead it adds the SPE ABI extensions to the current
14902 @opindex mabi=no-spe
14903 Disable Booke SPE ABI extensions for the current ABI@.
14905 @item -mabi=ibmlongdouble
14906 @opindex mabi=ibmlongdouble
14907 Change the current ABI to use IBM extended precision long double.
14908 This is a PowerPC 32-bit SYSV ABI option.
14910 @item -mabi=ieeelongdouble
14911 @opindex mabi=ieeelongdouble
14912 Change the current ABI to use IEEE extended precision long double.
14913 This is a PowerPC 32-bit Linux ABI option.
14916 @itemx -mno-prototype
14917 @opindex mprototype
14918 @opindex mno-prototype
14919 On System V.4 and embedded PowerPC systems assume that all calls to
14920 variable argument functions are properly prototyped. Otherwise, the
14921 compiler must insert an instruction before every non prototyped call to
14922 set or clear bit 6 of the condition code register (@var{CR}) to
14923 indicate whether floating point values were passed in the floating point
14924 registers in case the function takes a variable arguments. With
14925 @option{-mprototype}, only calls to prototyped variable argument functions
14926 will set or clear the bit.
14930 On embedded PowerPC systems, assume that the startup module is called
14931 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
14932 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}
14937 On embedded PowerPC systems, assume that the startup module is called
14938 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
14943 On embedded PowerPC systems, assume that the startup module is called
14944 @file{crt0.o} and the standard C libraries are @file{libads.a} and
14947 @item -myellowknife
14948 @opindex myellowknife
14949 On embedded PowerPC systems, assume that the startup module is called
14950 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
14955 On System V.4 and embedded PowerPC systems, specify that you are
14956 compiling for a VxWorks system.
14960 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
14961 header to indicate that @samp{eabi} extended relocations are used.
14967 On System V.4 and embedded PowerPC systems do (do not) adhere to the
14968 Embedded Applications Binary Interface (eabi) which is a set of
14969 modifications to the System V.4 specifications. Selecting @option{-meabi}
14970 means that the stack is aligned to an 8 byte boundary, a function
14971 @code{__eabi} is called to from @code{main} to set up the eabi
14972 environment, and the @option{-msdata} option can use both @code{r2} and
14973 @code{r13} to point to two separate small data areas. Selecting
14974 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
14975 do not call an initialization function from @code{main}, and the
14976 @option{-msdata} option will only use @code{r13} to point to a single
14977 small data area. The @option{-meabi} option is on by default if you
14978 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
14981 @opindex msdata=eabi
14982 On System V.4 and embedded PowerPC systems, put small initialized
14983 @code{const} global and static data in the @samp{.sdata2} section, which
14984 is pointed to by register @code{r2}. Put small initialized
14985 non-@code{const} global and static data in the @samp{.sdata} section,
14986 which is pointed to by register @code{r13}. Put small uninitialized
14987 global and static data in the @samp{.sbss} section, which is adjacent to
14988 the @samp{.sdata} section. The @option{-msdata=eabi} option is
14989 incompatible with the @option{-mrelocatable} option. The
14990 @option{-msdata=eabi} option also sets the @option{-memb} option.
14993 @opindex msdata=sysv
14994 On System V.4 and embedded PowerPC systems, put small global and static
14995 data in the @samp{.sdata} section, which is pointed to by register
14996 @code{r13}. Put small uninitialized global and static data in the
14997 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
14998 The @option{-msdata=sysv} option is incompatible with the
14999 @option{-mrelocatable} option.
15001 @item -msdata=default
15003 @opindex msdata=default
15005 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
15006 compile code the same as @option{-msdata=eabi}, otherwise compile code the
15007 same as @option{-msdata=sysv}.
15010 @opindex msdata=data
15011 On System V.4 and embedded PowerPC systems, put small global
15012 data in the @samp{.sdata} section. Put small uninitialized global
15013 data in the @samp{.sbss} section. Do not use register @code{r13}
15014 to address small data however. This is the default behavior unless
15015 other @option{-msdata} options are used.
15019 @opindex msdata=none
15021 On embedded PowerPC systems, put all initialized global and static data
15022 in the @samp{.data} section, and all uninitialized data in the
15023 @samp{.bss} section.
15027 @cindex smaller data references (PowerPC)
15028 @cindex .sdata/.sdata2 references (PowerPC)
15029 On embedded PowerPC systems, put global and static items less than or
15030 equal to @var{num} bytes into the small data or bss sections instead of
15031 the normal data or bss section. By default, @var{num} is 8. The
15032 @option{-G @var{num}} switch is also passed to the linker.
15033 All modules should be compiled with the same @option{-G @var{num}} value.
15036 @itemx -mno-regnames
15038 @opindex mno-regnames
15039 On System V.4 and embedded PowerPC systems do (do not) emit register
15040 names in the assembly language output using symbolic forms.
15043 @itemx -mno-longcall
15045 @opindex mno-longcall
15046 By default assume that all calls are far away so that a longer more
15047 expensive calling sequence is required. This is required for calls
15048 further than 32 megabytes (33,554,432 bytes) from the current location.
15049 A short call will be generated if the compiler knows
15050 the call cannot be that far away. This setting can be overridden by
15051 the @code{shortcall} function attribute, or by @code{#pragma
15054 Some linkers are capable of detecting out-of-range calls and generating
15055 glue code on the fly. On these systems, long calls are unnecessary and
15056 generate slower code. As of this writing, the AIX linker can do this,
15057 as can the GNU linker for PowerPC/64. It is planned to add this feature
15058 to the GNU linker for 32-bit PowerPC systems as well.
15060 On Darwin/PPC systems, @code{#pragma longcall} will generate ``jbsr
15061 callee, L42'', plus a ``branch island'' (glue code). The two target
15062 addresses represent the callee and the ``branch island''. The
15063 Darwin/PPC linker will prefer the first address and generate a ``bl
15064 callee'' if the PPC ``bl'' instruction will reach the callee directly;
15065 otherwise, the linker will generate ``bl L42'' to call the ``branch
15066 island''. The ``branch island'' is appended to the body of the
15067 calling function; it computes the full 32-bit address of the callee
15070 On Mach-O (Darwin) systems, this option directs the compiler emit to
15071 the glue for every direct call, and the Darwin linker decides whether
15072 to use or discard it.
15074 In the future, we may cause GCC to ignore all longcall specifications
15075 when the linker is known to generate glue.
15077 @item -mtls-markers
15078 @itemx -mno-tls-markers
15079 @opindex mtls-markers
15080 @opindex mno-tls-markers
15081 Mark (do not mark) calls to @code{__tls_get_addr} with a relocation
15082 specifying the function argument. The relocation allows ld to
15083 reliably associate function call with argument setup instructions for
15084 TLS optimization, which in turn allows gcc to better schedule the
15089 Adds support for multithreading with the @dfn{pthreads} library.
15090 This option sets flags for both the preprocessor and linker.
15094 @node S/390 and zSeries Options
15095 @subsection S/390 and zSeries Options
15096 @cindex S/390 and zSeries Options
15098 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
15102 @itemx -msoft-float
15103 @opindex mhard-float
15104 @opindex msoft-float
15105 Use (do not use) the hardware floating-point instructions and registers
15106 for floating-point operations. When @option{-msoft-float} is specified,
15107 functions in @file{libgcc.a} will be used to perform floating-point
15108 operations. When @option{-mhard-float} is specified, the compiler
15109 generates IEEE floating-point instructions. This is the default.
15112 @itemx -mno-hard-dfp
15114 @opindex mno-hard-dfp
15115 Use (do not use) the hardware decimal-floating-point instructions for
15116 decimal-floating-point operations. When @option{-mno-hard-dfp} is
15117 specified, functions in @file{libgcc.a} will be used to perform
15118 decimal-floating-point operations. When @option{-mhard-dfp} is
15119 specified, the compiler generates decimal-floating-point hardware
15120 instructions. This is the default for @option{-march=z9-ec} or higher.
15122 @item -mlong-double-64
15123 @itemx -mlong-double-128
15124 @opindex mlong-double-64
15125 @opindex mlong-double-128
15126 These switches control the size of @code{long double} type. A size
15127 of 64bit makes the @code{long double} type equivalent to the @code{double}
15128 type. This is the default.
15131 @itemx -mno-backchain
15132 @opindex mbackchain
15133 @opindex mno-backchain
15134 Store (do not store) the address of the caller's frame as backchain pointer
15135 into the callee's stack frame.
15136 A backchain may be needed to allow debugging using tools that do not understand
15137 DWARF-2 call frame information.
15138 When @option{-mno-packed-stack} is in effect, the backchain pointer is stored
15139 at the bottom of the stack frame; when @option{-mpacked-stack} is in effect,
15140 the backchain is placed into the topmost word of the 96/160 byte register
15143 In general, code compiled with @option{-mbackchain} is call-compatible with
15144 code compiled with @option{-mmo-backchain}; however, use of the backchain
15145 for debugging purposes usually requires that the whole binary is built with
15146 @option{-mbackchain}. Note that the combination of @option{-mbackchain},
15147 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
15148 to build a linux kernel use @option{-msoft-float}.
15150 The default is to not maintain the backchain.
15152 @item -mpacked-stack
15153 @itemx -mno-packed-stack
15154 @opindex mpacked-stack
15155 @opindex mno-packed-stack
15156 Use (do not use) the packed stack layout. When @option{-mno-packed-stack} is
15157 specified, the compiler uses the all fields of the 96/160 byte register save
15158 area only for their default purpose; unused fields still take up stack space.
15159 When @option{-mpacked-stack} is specified, register save slots are densely
15160 packed at the top of the register save area; unused space is reused for other
15161 purposes, allowing for more efficient use of the available stack space.
15162 However, when @option{-mbackchain} is also in effect, the topmost word of
15163 the save area is always used to store the backchain, and the return address
15164 register is always saved two words below the backchain.
15166 As long as the stack frame backchain is not used, code generated with
15167 @option{-mpacked-stack} is call-compatible with code generated with
15168 @option{-mno-packed-stack}. Note that some non-FSF releases of GCC 2.95 for
15169 S/390 or zSeries generated code that uses the stack frame backchain at run
15170 time, not just for debugging purposes. Such code is not call-compatible
15171 with code compiled with @option{-mpacked-stack}. Also, note that the
15172 combination of @option{-mbackchain},
15173 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
15174 to build a linux kernel use @option{-msoft-float}.
15176 The default is to not use the packed stack layout.
15179 @itemx -mno-small-exec
15180 @opindex msmall-exec
15181 @opindex mno-small-exec
15182 Generate (or do not generate) code using the @code{bras} instruction
15183 to do subroutine calls.
15184 This only works reliably if the total executable size does not
15185 exceed 64k. The default is to use the @code{basr} instruction instead,
15186 which does not have this limitation.
15192 When @option{-m31} is specified, generate code compliant to the
15193 GNU/Linux for S/390 ABI@. When @option{-m64} is specified, generate
15194 code compliant to the GNU/Linux for zSeries ABI@. This allows GCC in
15195 particular to generate 64-bit instructions. For the @samp{s390}
15196 targets, the default is @option{-m31}, while the @samp{s390x}
15197 targets default to @option{-m64}.
15203 When @option{-mzarch} is specified, generate code using the
15204 instructions available on z/Architecture.
15205 When @option{-mesa} is specified, generate code using the
15206 instructions available on ESA/390. Note that @option{-mesa} is
15207 not possible with @option{-m64}.
15208 When generating code compliant to the GNU/Linux for S/390 ABI,
15209 the default is @option{-mesa}. When generating code compliant
15210 to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
15216 Generate (or do not generate) code using the @code{mvcle} instruction
15217 to perform block moves. When @option{-mno-mvcle} is specified,
15218 use a @code{mvc} loop instead. This is the default unless optimizing for
15225 Print (or do not print) additional debug information when compiling.
15226 The default is to not print debug information.
15228 @item -march=@var{cpu-type}
15230 Generate code that will run on @var{cpu-type}, which is the name of a system
15231 representing a certain processor type. Possible values for
15232 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, @samp{z990},
15233 @samp{z9-109}, @samp{z9-ec} and @samp{z10}.
15234 When generating code using the instructions available on z/Architecture,
15235 the default is @option{-march=z900}. Otherwise, the default is
15236 @option{-march=g5}.
15238 @item -mtune=@var{cpu-type}
15240 Tune to @var{cpu-type} everything applicable about the generated code,
15241 except for the ABI and the set of available instructions.
15242 The list of @var{cpu-type} values is the same as for @option{-march}.
15243 The default is the value used for @option{-march}.
15246 @itemx -mno-tpf-trace
15247 @opindex mtpf-trace
15248 @opindex mno-tpf-trace
15249 Generate code that adds (does not add) in TPF OS specific branches to trace
15250 routines in the operating system. This option is off by default, even
15251 when compiling for the TPF OS@.
15254 @itemx -mno-fused-madd
15255 @opindex mfused-madd
15256 @opindex mno-fused-madd
15257 Generate code that uses (does not use) the floating point multiply and
15258 accumulate instructions. These instructions are generated by default if
15259 hardware floating point is used.
15261 @item -mwarn-framesize=@var{framesize}
15262 @opindex mwarn-framesize
15263 Emit a warning if the current function exceeds the given frame size. Because
15264 this is a compile time check it doesn't need to be a real problem when the program
15265 runs. It is intended to identify functions which most probably cause
15266 a stack overflow. It is useful to be used in an environment with limited stack
15267 size e.g.@: the linux kernel.
15269 @item -mwarn-dynamicstack
15270 @opindex mwarn-dynamicstack
15271 Emit a warning if the function calls alloca or uses dynamically
15272 sized arrays. This is generally a bad idea with a limited stack size.
15274 @item -mstack-guard=@var{stack-guard}
15275 @itemx -mstack-size=@var{stack-size}
15276 @opindex mstack-guard
15277 @opindex mstack-size
15278 If these options are provided the s390 back end emits additional instructions in
15279 the function prologue which trigger a trap if the stack size is @var{stack-guard}
15280 bytes above the @var{stack-size} (remember that the stack on s390 grows downward).
15281 If the @var{stack-guard} option is omitted the smallest power of 2 larger than
15282 the frame size of the compiled function is chosen.
15283 These options are intended to be used to help debugging stack overflow problems.
15284 The additionally emitted code causes only little overhead and hence can also be
15285 used in production like systems without greater performance degradation. The given
15286 values have to be exact powers of 2 and @var{stack-size} has to be greater than
15287 @var{stack-guard} without exceeding 64k.
15288 In order to be efficient the extra code makes the assumption that the stack starts
15289 at an address aligned to the value given by @var{stack-size}.
15290 The @var{stack-guard} option can only be used in conjunction with @var{stack-size}.
15293 @node Score Options
15294 @subsection Score Options
15295 @cindex Score Options
15297 These options are defined for Score implementations:
15302 Compile code for big endian mode. This is the default.
15306 Compile code for little endian mode.
15310 Disable generate bcnz instruction.
15314 Enable generate unaligned load and store instruction.
15318 Enable the use of multiply-accumulate instructions. Disabled by default.
15322 Specify the SCORE5 as the target architecture.
15326 Specify the SCORE5U of the target architecture.
15330 Specify the SCORE7 as the target architecture. This is the default.
15334 Specify the SCORE7D as the target architecture.
15338 @subsection SH Options
15340 These @samp{-m} options are defined for the SH implementations:
15345 Generate code for the SH1.
15349 Generate code for the SH2.
15352 Generate code for the SH2e.
15356 Generate code for the SH2a without FPU, or for a SH2a-FPU in such a way
15357 that the floating-point unit is not used.
15359 @item -m2a-single-only
15360 @opindex m2a-single-only
15361 Generate code for the SH2a-FPU, in such a way that no double-precision
15362 floating point operations are used.
15365 @opindex m2a-single
15366 Generate code for the SH2a-FPU assuming the floating-point unit is in
15367 single-precision mode by default.
15371 Generate code for the SH2a-FPU assuming the floating-point unit is in
15372 double-precision mode by default.
15376 Generate code for the SH3.
15380 Generate code for the SH3e.
15384 Generate code for the SH4 without a floating-point unit.
15386 @item -m4-single-only
15387 @opindex m4-single-only
15388 Generate code for the SH4 with a floating-point unit that only
15389 supports single-precision arithmetic.
15393 Generate code for the SH4 assuming the floating-point unit is in
15394 single-precision mode by default.
15398 Generate code for the SH4.
15402 Generate code for the SH4al-dsp, or for a SH4a in such a way that the
15403 floating-point unit is not used.
15405 @item -m4a-single-only
15406 @opindex m4a-single-only
15407 Generate code for the SH4a, in such a way that no double-precision
15408 floating point operations are used.
15411 @opindex m4a-single
15412 Generate code for the SH4a assuming the floating-point unit is in
15413 single-precision mode by default.
15417 Generate code for the SH4a.
15421 Same as @option{-m4a-nofpu}, except that it implicitly passes
15422 @option{-dsp} to the assembler. GCC doesn't generate any DSP
15423 instructions at the moment.
15427 Compile code for the processor in big endian mode.
15431 Compile code for the processor in little endian mode.
15435 Align doubles at 64-bit boundaries. Note that this changes the calling
15436 conventions, and thus some functions from the standard C library will
15437 not work unless you recompile it first with @option{-mdalign}.
15441 Shorten some address references at link time, when possible; uses the
15442 linker option @option{-relax}.
15446 Use 32-bit offsets in @code{switch} tables. The default is to use
15451 Enable the use of bit manipulation instructions on SH2A.
15455 Enable the use of the instruction @code{fmovd}. Check @option{-mdalign} for
15456 alignment constraints.
15460 Comply with the calling conventions defined by Renesas.
15464 Comply with the calling conventions defined by Renesas.
15468 Comply with the calling conventions defined for GCC before the Renesas
15469 conventions were available. This option is the default for all
15470 targets of the SH toolchain except for @samp{sh-symbianelf}.
15473 @opindex mnomacsave
15474 Mark the @code{MAC} register as call-clobbered, even if
15475 @option{-mhitachi} is given.
15479 Increase IEEE-compliance of floating-point code.
15480 At the moment, this is equivalent to @option{-fno-finite-math-only}.
15481 When generating 16 bit SH opcodes, getting IEEE-conforming results for
15482 comparisons of NANs / infinities incurs extra overhead in every
15483 floating point comparison, therefore the default is set to
15484 @option{-ffinite-math-only}.
15486 @item -minline-ic_invalidate
15487 @opindex minline-ic_invalidate
15488 Inline code to invalidate instruction cache entries after setting up
15489 nested function trampolines.
15490 This option has no effect if -musermode is in effect and the selected
15491 code generation option (e.g. -m4) does not allow the use of the icbi
15493 If the selected code generation option does not allow the use of the icbi
15494 instruction, and -musermode is not in effect, the inlined code will
15495 manipulate the instruction cache address array directly with an associative
15496 write. This not only requires privileged mode, but it will also
15497 fail if the cache line had been mapped via the TLB and has become unmapped.
15501 Dump instruction size and location in the assembly code.
15504 @opindex mpadstruct
15505 This option is deprecated. It pads structures to multiple of 4 bytes,
15506 which is incompatible with the SH ABI@.
15510 Optimize for space instead of speed. Implied by @option{-Os}.
15513 @opindex mprefergot
15514 When generating position-independent code, emit function calls using
15515 the Global Offset Table instead of the Procedure Linkage Table.
15519 Don't generate privileged mode only code; implies -mno-inline-ic_invalidate
15520 if the inlined code would not work in user mode.
15521 This is the default when the target is @code{sh-*-linux*}.
15523 @item -multcost=@var{number}
15524 @opindex multcost=@var{number}
15525 Set the cost to assume for a multiply insn.
15527 @item -mdiv=@var{strategy}
15528 @opindex mdiv=@var{strategy}
15529 Set the division strategy to use for SHmedia code. @var{strategy} must be
15530 one of: call, call2, fp, inv, inv:minlat, inv20u, inv20l, inv:call,
15531 inv:call2, inv:fp .
15532 "fp" performs the operation in floating point. This has a very high latency,
15533 but needs only a few instructions, so it might be a good choice if
15534 your code has enough easily exploitable ILP to allow the compiler to
15535 schedule the floating point instructions together with other instructions.
15536 Division by zero causes a floating point exception.
15537 "inv" uses integer operations to calculate the inverse of the divisor,
15538 and then multiplies the dividend with the inverse. This strategy allows
15539 cse and hoisting of the inverse calculation. Division by zero calculates
15540 an unspecified result, but does not trap.
15541 "inv:minlat" is a variant of "inv" where if no cse / hoisting opportunities
15542 have been found, or if the entire operation has been hoisted to the same
15543 place, the last stages of the inverse calculation are intertwined with the
15544 final multiply to reduce the overall latency, at the expense of using a few
15545 more instructions, and thus offering fewer scheduling opportunities with
15547 "call" calls a library function that usually implements the inv:minlat
15549 This gives high code density for m5-*media-nofpu compilations.
15550 "call2" uses a different entry point of the same library function, where it
15551 assumes that a pointer to a lookup table has already been set up, which
15552 exposes the pointer load to cse / code hoisting optimizations.
15553 "inv:call", "inv:call2" and "inv:fp" all use the "inv" algorithm for initial
15554 code generation, but if the code stays unoptimized, revert to the "call",
15555 "call2", or "fp" strategies, respectively. Note that the
15556 potentially-trapping side effect of division by zero is carried by a
15557 separate instruction, so it is possible that all the integer instructions
15558 are hoisted out, but the marker for the side effect stays where it is.
15559 A recombination to fp operations or a call is not possible in that case.
15560 "inv20u" and "inv20l" are variants of the "inv:minlat" strategy. In the case
15561 that the inverse calculation was nor separated from the multiply, they speed
15562 up division where the dividend fits into 20 bits (plus sign where applicable),
15563 by inserting a test to skip a number of operations in this case; this test
15564 slows down the case of larger dividends. inv20u assumes the case of a such
15565 a small dividend to be unlikely, and inv20l assumes it to be likely.
15567 @item -mdivsi3_libfunc=@var{name}
15568 @opindex mdivsi3_libfunc=@var{name}
15569 Set the name of the library function used for 32 bit signed division to
15570 @var{name}. This only affect the name used in the call and inv:call
15571 division strategies, and the compiler will still expect the same
15572 sets of input/output/clobbered registers as if this option was not present.
15574 @item -mfixed-range=@var{register-range}
15575 @opindex mfixed-range
15576 Generate code treating the given register range as fixed registers.
15577 A fixed register is one that the register allocator can not use. This is
15578 useful when compiling kernel code. A register range is specified as
15579 two registers separated by a dash. Multiple register ranges can be
15580 specified separated by a comma.
15582 @item -madjust-unroll
15583 @opindex madjust-unroll
15584 Throttle unrolling to avoid thrashing target registers.
15585 This option only has an effect if the gcc code base supports the
15586 TARGET_ADJUST_UNROLL_MAX target hook.
15588 @item -mindexed-addressing
15589 @opindex mindexed-addressing
15590 Enable the use of the indexed addressing mode for SHmedia32/SHcompact.
15591 This is only safe if the hardware and/or OS implement 32 bit wrap-around
15592 semantics for the indexed addressing mode. The architecture allows the
15593 implementation of processors with 64 bit MMU, which the OS could use to
15594 get 32 bit addressing, but since no current hardware implementation supports
15595 this or any other way to make the indexed addressing mode safe to use in
15596 the 32 bit ABI, the default is -mno-indexed-addressing.
15598 @item -mgettrcost=@var{number}
15599 @opindex mgettrcost=@var{number}
15600 Set the cost assumed for the gettr instruction to @var{number}.
15601 The default is 2 if @option{-mpt-fixed} is in effect, 100 otherwise.
15605 Assume pt* instructions won't trap. This will generally generate better
15606 scheduled code, but is unsafe on current hardware. The current architecture
15607 definition says that ptabs and ptrel trap when the target anded with 3 is 3.
15608 This has the unintentional effect of making it unsafe to schedule ptabs /
15609 ptrel before a branch, or hoist it out of a loop. For example,
15610 __do_global_ctors, a part of libgcc that runs constructors at program
15611 startup, calls functions in a list which is delimited by @minus{}1. With the
15612 -mpt-fixed option, the ptabs will be done before testing against @minus{}1.
15613 That means that all the constructors will be run a bit quicker, but when
15614 the loop comes to the end of the list, the program crashes because ptabs
15615 loads @minus{}1 into a target register. Since this option is unsafe for any
15616 hardware implementing the current architecture specification, the default
15617 is -mno-pt-fixed. Unless the user specifies a specific cost with
15618 @option{-mgettrcost}, -mno-pt-fixed also implies @option{-mgettrcost=100};
15619 this deters register allocation using target registers for storing
15622 @item -minvalid-symbols
15623 @opindex minvalid-symbols
15624 Assume symbols might be invalid. Ordinary function symbols generated by
15625 the compiler will always be valid to load with movi/shori/ptabs or
15626 movi/shori/ptrel, but with assembler and/or linker tricks it is possible
15627 to generate symbols that will cause ptabs / ptrel to trap.
15628 This option is only meaningful when @option{-mno-pt-fixed} is in effect.
15629 It will then prevent cross-basic-block cse, hoisting and most scheduling
15630 of symbol loads. The default is @option{-mno-invalid-symbols}.
15633 @node SPARC Options
15634 @subsection SPARC Options
15635 @cindex SPARC options
15637 These @samp{-m} options are supported on the SPARC:
15640 @item -mno-app-regs
15642 @opindex mno-app-regs
15644 Specify @option{-mapp-regs} to generate output using the global registers
15645 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
15648 To be fully SVR4 ABI compliant at the cost of some performance loss,
15649 specify @option{-mno-app-regs}. You should compile libraries and system
15650 software with this option.
15653 @itemx -mhard-float
15655 @opindex mhard-float
15656 Generate output containing floating point instructions. This is the
15660 @itemx -msoft-float
15662 @opindex msoft-float
15663 Generate output containing library calls for floating point.
15664 @strong{Warning:} the requisite libraries are not available for all SPARC
15665 targets. Normally the facilities of the machine's usual C compiler are
15666 used, but this cannot be done directly in cross-compilation. You must make
15667 your own arrangements to provide suitable library functions for
15668 cross-compilation. The embedded targets @samp{sparc-*-aout} and
15669 @samp{sparclite-*-*} do provide software floating point support.
15671 @option{-msoft-float} changes the calling convention in the output file;
15672 therefore, it is only useful if you compile @emph{all} of a program with
15673 this option. In particular, you need to compile @file{libgcc.a}, the
15674 library that comes with GCC, with @option{-msoft-float} in order for
15677 @item -mhard-quad-float
15678 @opindex mhard-quad-float
15679 Generate output containing quad-word (long double) floating point
15682 @item -msoft-quad-float
15683 @opindex msoft-quad-float
15684 Generate output containing library calls for quad-word (long double)
15685 floating point instructions. The functions called are those specified
15686 in the SPARC ABI@. This is the default.
15688 As of this writing, there are no SPARC implementations that have hardware
15689 support for the quad-word floating point instructions. They all invoke
15690 a trap handler for one of these instructions, and then the trap handler
15691 emulates the effect of the instruction. Because of the trap handler overhead,
15692 this is much slower than calling the ABI library routines. Thus the
15693 @option{-msoft-quad-float} option is the default.
15695 @item -mno-unaligned-doubles
15696 @itemx -munaligned-doubles
15697 @opindex mno-unaligned-doubles
15698 @opindex munaligned-doubles
15699 Assume that doubles have 8 byte alignment. This is the default.
15701 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
15702 alignment only if they are contained in another type, or if they have an
15703 absolute address. Otherwise, it assumes they have 4 byte alignment.
15704 Specifying this option avoids some rare compatibility problems with code
15705 generated by other compilers. It is not the default because it results
15706 in a performance loss, especially for floating point code.
15708 @item -mno-faster-structs
15709 @itemx -mfaster-structs
15710 @opindex mno-faster-structs
15711 @opindex mfaster-structs
15712 With @option{-mfaster-structs}, the compiler assumes that structures
15713 should have 8 byte alignment. This enables the use of pairs of
15714 @code{ldd} and @code{std} instructions for copies in structure
15715 assignment, in place of twice as many @code{ld} and @code{st} pairs.
15716 However, the use of this changed alignment directly violates the SPARC
15717 ABI@. Thus, it's intended only for use on targets where the developer
15718 acknowledges that their resulting code will not be directly in line with
15719 the rules of the ABI@.
15721 @item -mimpure-text
15722 @opindex mimpure-text
15723 @option{-mimpure-text}, used in addition to @option{-shared}, tells
15724 the compiler to not pass @option{-z text} to the linker when linking a
15725 shared object. Using this option, you can link position-dependent
15726 code into a shared object.
15728 @option{-mimpure-text} suppresses the ``relocations remain against
15729 allocatable but non-writable sections'' linker error message.
15730 However, the necessary relocations will trigger copy-on-write, and the
15731 shared object is not actually shared across processes. Instead of
15732 using @option{-mimpure-text}, you should compile all source code with
15733 @option{-fpic} or @option{-fPIC}.
15735 This option is only available on SunOS and Solaris.
15737 @item -mcpu=@var{cpu_type}
15739 Set the instruction set, register set, and instruction scheduling parameters
15740 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
15741 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
15742 @samp{f930}, @samp{f934}, @samp{hypersparc}, @samp{sparclite86x},
15743 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc},
15744 @samp{ultrasparc3}, @samp{niagara} and @samp{niagara2}.
15746 Default instruction scheduling parameters are used for values that select
15747 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
15748 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
15750 Here is a list of each supported architecture and their supported
15755 v8: supersparc, hypersparc
15756 sparclite: f930, f934, sparclite86x
15758 v9: ultrasparc, ultrasparc3, niagara, niagara2
15761 By default (unless configured otherwise), GCC generates code for the V7
15762 variant of the SPARC architecture. With @option{-mcpu=cypress}, the compiler
15763 additionally optimizes it for the Cypress CY7C602 chip, as used in the
15764 SPARCStation/SPARCServer 3xx series. This is also appropriate for the older
15765 SPARCStation 1, 2, IPX etc.
15767 With @option{-mcpu=v8}, GCC generates code for the V8 variant of the SPARC
15768 architecture. The only difference from V7 code is that the compiler emits
15769 the integer multiply and integer divide instructions which exist in SPARC-V8
15770 but not in SPARC-V7. With @option{-mcpu=supersparc}, the compiler additionally
15771 optimizes it for the SuperSPARC chip, as used in the SPARCStation 10, 1000 and
15774 With @option{-mcpu=sparclite}, GCC generates code for the SPARClite variant of
15775 the SPARC architecture. This adds the integer multiply, integer divide step
15776 and scan (@code{ffs}) instructions which exist in SPARClite but not in SPARC-V7.
15777 With @option{-mcpu=f930}, the compiler additionally optimizes it for the
15778 Fujitsu MB86930 chip, which is the original SPARClite, with no FPU@. With
15779 @option{-mcpu=f934}, the compiler additionally optimizes it for the Fujitsu
15780 MB86934 chip, which is the more recent SPARClite with FPU@.
15782 With @option{-mcpu=sparclet}, GCC generates code for the SPARClet variant of
15783 the SPARC architecture. This adds the integer multiply, multiply/accumulate,
15784 integer divide step and scan (@code{ffs}) instructions which exist in SPARClet
15785 but not in SPARC-V7. With @option{-mcpu=tsc701}, the compiler additionally
15786 optimizes it for the TEMIC SPARClet chip.
15788 With @option{-mcpu=v9}, GCC generates code for the V9 variant of the SPARC
15789 architecture. This adds 64-bit integer and floating-point move instructions,
15790 3 additional floating-point condition code registers and conditional move
15791 instructions. With @option{-mcpu=ultrasparc}, the compiler additionally
15792 optimizes it for the Sun UltraSPARC I/II/IIi chips. With
15793 @option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
15794 Sun UltraSPARC III/III+/IIIi/IIIi+/IV/IV+ chips. With
15795 @option{-mcpu=niagara}, the compiler additionally optimizes it for
15796 Sun UltraSPARC T1 chips. With @option{-mcpu=niagara2}, the compiler
15797 additionally optimizes it for Sun UltraSPARC T2 chips.
15799 @item -mtune=@var{cpu_type}
15801 Set the instruction scheduling parameters for machine type
15802 @var{cpu_type}, but do not set the instruction set or register set that the
15803 option @option{-mcpu=@var{cpu_type}} would.
15805 The same values for @option{-mcpu=@var{cpu_type}} can be used for
15806 @option{-mtune=@var{cpu_type}}, but the only useful values are those
15807 that select a particular cpu implementation. Those are @samp{cypress},
15808 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
15809 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc},
15810 @samp{ultrasparc3}, @samp{niagara}, and @samp{niagara2}.
15815 @opindex mno-v8plus
15816 With @option{-mv8plus}, GCC generates code for the SPARC-V8+ ABI@. The
15817 difference from the V8 ABI is that the global and out registers are
15818 considered 64-bit wide. This is enabled by default on Solaris in 32-bit
15819 mode for all SPARC-V9 processors.
15825 With @option{-mvis}, GCC generates code that takes advantage of the UltraSPARC
15826 Visual Instruction Set extensions. The default is @option{-mno-vis}.
15829 These @samp{-m} options are supported in addition to the above
15830 on SPARC-V9 processors in 64-bit environments:
15833 @item -mlittle-endian
15834 @opindex mlittle-endian
15835 Generate code for a processor running in little-endian mode. It is only
15836 available for a few configurations and most notably not on Solaris and Linux.
15842 Generate code for a 32-bit or 64-bit environment.
15843 The 32-bit environment sets int, long and pointer to 32 bits.
15844 The 64-bit environment sets int to 32 bits and long and pointer
15847 @item -mcmodel=medlow
15848 @opindex mcmodel=medlow
15849 Generate code for the Medium/Low code model: 64-bit addresses, programs
15850 must be linked in the low 32 bits of memory. Programs can be statically
15851 or dynamically linked.
15853 @item -mcmodel=medmid
15854 @opindex mcmodel=medmid
15855 Generate code for the Medium/Middle code model: 64-bit addresses, programs
15856 must be linked in the low 44 bits of memory, the text and data segments must
15857 be less than 2GB in size and the data segment must be located within 2GB of
15860 @item -mcmodel=medany
15861 @opindex mcmodel=medany
15862 Generate code for the Medium/Anywhere code model: 64-bit addresses, programs
15863 may be linked anywhere in memory, the text and data segments must be less
15864 than 2GB in size and the data segment must be located within 2GB of the
15867 @item -mcmodel=embmedany
15868 @opindex mcmodel=embmedany
15869 Generate code for the Medium/Anywhere code model for embedded systems:
15870 64-bit addresses, the text and data segments must be less than 2GB in
15871 size, both starting anywhere in memory (determined at link time). The
15872 global register %g4 points to the base of the data segment. Programs
15873 are statically linked and PIC is not supported.
15876 @itemx -mno-stack-bias
15877 @opindex mstack-bias
15878 @opindex mno-stack-bias
15879 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
15880 frame pointer if present, are offset by @minus{}2047 which must be added back
15881 when making stack frame references. This is the default in 64-bit mode.
15882 Otherwise, assume no such offset is present.
15885 These switches are supported in addition to the above on Solaris:
15890 Add support for multithreading using the Solaris threads library. This
15891 option sets flags for both the preprocessor and linker. This option does
15892 not affect the thread safety of object code produced by the compiler or
15893 that of libraries supplied with it.
15897 Add support for multithreading using the POSIX threads library. This
15898 option sets flags for both the preprocessor and linker. This option does
15899 not affect the thread safety of object code produced by the compiler or
15900 that of libraries supplied with it.
15904 This is a synonym for @option{-pthreads}.
15908 @subsection SPU Options
15909 @cindex SPU options
15911 These @samp{-m} options are supported on the SPU:
15915 @itemx -merror-reloc
15916 @opindex mwarn-reloc
15917 @opindex merror-reloc
15919 The loader for SPU does not handle dynamic relocations. By default, GCC
15920 will give an error when it generates code that requires a dynamic
15921 relocation. @option{-mno-error-reloc} disables the error,
15922 @option{-mwarn-reloc} will generate a warning instead.
15925 @itemx -munsafe-dma
15927 @opindex munsafe-dma
15929 Instructions which initiate or test completion of DMA must not be
15930 reordered with respect to loads and stores of the memory which is being
15931 accessed. Users typically address this problem using the volatile
15932 keyword, but that can lead to inefficient code in places where the
15933 memory is known to not change. Rather than mark the memory as volatile
15934 we treat the DMA instructions as potentially effecting all memory. With
15935 @option{-munsafe-dma} users must use the volatile keyword to protect
15938 @item -mbranch-hints
15939 @opindex mbranch-hints
15941 By default, GCC will generate a branch hint instruction to avoid
15942 pipeline stalls for always taken or probably taken branches. A hint
15943 will not be generated closer than 8 instructions away from its branch.
15944 There is little reason to disable them, except for debugging purposes,
15945 or to make an object a little bit smaller.
15949 @opindex msmall-mem
15950 @opindex mlarge-mem
15952 By default, GCC generates code assuming that addresses are never larger
15953 than 18 bits. With @option{-mlarge-mem} code is generated that assumes
15954 a full 32 bit address.
15959 By default, GCC links against startup code that assumes the SPU-style
15960 main function interface (which has an unconventional parameter list).
15961 With @option{-mstdmain}, GCC will link your program against startup
15962 code that assumes a C99-style interface to @code{main}, including a
15963 local copy of @code{argv} strings.
15965 @item -mfixed-range=@var{register-range}
15966 @opindex mfixed-range
15967 Generate code treating the given register range as fixed registers.
15968 A fixed register is one that the register allocator can not use. This is
15969 useful when compiling kernel code. A register range is specified as
15970 two registers separated by a dash. Multiple register ranges can be
15971 specified separated by a comma.
15974 @itemx -mdual-nops=@var{n}
15975 @opindex mdual-nops
15976 By default, GCC will insert nops to increase dual issue when it expects
15977 it to increase performance. @var{n} can be a value from 0 to 10. A
15978 smaller @var{n} will insert fewer nops. 10 is the default, 0 is the
15979 same as @option{-mno-dual-nops}. Disabled with @option{-Os}.
15981 @item -mhint-max-nops=@var{n}
15982 @opindex mhint-max-nops
15983 Maximum number of nops to insert for a branch hint. A branch hint must
15984 be at least 8 instructions away from the branch it is effecting. GCC
15985 will insert up to @var{n} nops to enforce this, otherwise it will not
15986 generate the branch hint.
15988 @item -mhint-max-distance=@var{n}
15989 @opindex mhint-max-distance
15990 The encoding of the branch hint instruction limits the hint to be within
15991 256 instructions of the branch it is effecting. By default, GCC makes
15992 sure it is within 125.
15995 @opindex msafe-hints
15996 Work around a hardware bug which causes the SPU to stall indefinitely.
15997 By default, GCC will insert the @code{hbrp} instruction to make sure
15998 this stall won't happen.
16002 @node System V Options
16003 @subsection Options for System V
16005 These additional options are available on System V Release 4 for
16006 compatibility with other compilers on those systems:
16011 Create a shared object.
16012 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
16016 Identify the versions of each tool used by the compiler, in a
16017 @code{.ident} assembler directive in the output.
16021 Refrain from adding @code{.ident} directives to the output file (this is
16024 @item -YP,@var{dirs}
16026 Search the directories @var{dirs}, and no others, for libraries
16027 specified with @option{-l}.
16029 @item -Ym,@var{dir}
16031 Look in the directory @var{dir} to find the M4 preprocessor.
16032 The assembler uses this option.
16033 @c This is supposed to go with a -Yd for predefined M4 macro files, but
16034 @c the generic assembler that comes with Solaris takes just -Ym.
16038 @subsection V850 Options
16039 @cindex V850 Options
16041 These @samp{-m} options are defined for V850 implementations:
16045 @itemx -mno-long-calls
16046 @opindex mlong-calls
16047 @opindex mno-long-calls
16048 Treat all calls as being far away (near). If calls are assumed to be
16049 far away, the compiler will always load the functions address up into a
16050 register, and call indirect through the pointer.
16056 Do not optimize (do optimize) basic blocks that use the same index
16057 pointer 4 or more times to copy pointer into the @code{ep} register, and
16058 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
16059 option is on by default if you optimize.
16061 @item -mno-prolog-function
16062 @itemx -mprolog-function
16063 @opindex mno-prolog-function
16064 @opindex mprolog-function
16065 Do not use (do use) external functions to save and restore registers
16066 at the prologue and epilogue of a function. The external functions
16067 are slower, but use less code space if more than one function saves
16068 the same number of registers. The @option{-mprolog-function} option
16069 is on by default if you optimize.
16073 Try to make the code as small as possible. At present, this just turns
16074 on the @option{-mep} and @option{-mprolog-function} options.
16076 @item -mtda=@var{n}
16078 Put static or global variables whose size is @var{n} bytes or less into
16079 the tiny data area that register @code{ep} points to. The tiny data
16080 area can hold up to 256 bytes in total (128 bytes for byte references).
16082 @item -msda=@var{n}
16084 Put static or global variables whose size is @var{n} bytes or less into
16085 the small data area that register @code{gp} points to. The small data
16086 area can hold up to 64 kilobytes.
16088 @item -mzda=@var{n}
16090 Put static or global variables whose size is @var{n} bytes or less into
16091 the first 32 kilobytes of memory.
16095 Specify that the target processor is the V850.
16098 @opindex mbig-switch
16099 Generate code suitable for big switch tables. Use this option only if
16100 the assembler/linker complain about out of range branches within a switch
16105 This option will cause r2 and r5 to be used in the code generated by
16106 the compiler. This setting is the default.
16108 @item -mno-app-regs
16109 @opindex mno-app-regs
16110 This option will cause r2 and r5 to be treated as fixed registers.
16114 Specify that the target processor is the V850E1. The preprocessor
16115 constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
16116 this option is used.
16120 Specify that the target processor is the V850E@. The preprocessor
16121 constant @samp{__v850e__} will be defined if this option is used.
16123 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
16124 are defined then a default target processor will be chosen and the
16125 relevant @samp{__v850*__} preprocessor constant will be defined.
16127 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
16128 defined, regardless of which processor variant is the target.
16130 @item -mdisable-callt
16131 @opindex mdisable-callt
16132 This option will suppress generation of the CALLT instruction for the
16133 v850e and v850e1 flavors of the v850 architecture. The default is
16134 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
16139 @subsection VAX Options
16140 @cindex VAX options
16142 These @samp{-m} options are defined for the VAX:
16147 Do not output certain jump instructions (@code{aobleq} and so on)
16148 that the Unix assembler for the VAX cannot handle across long
16153 Do output those jump instructions, on the assumption that you
16154 will assemble with the GNU assembler.
16158 Output code for g-format floating point numbers instead of d-format.
16161 @node VxWorks Options
16162 @subsection VxWorks Options
16163 @cindex VxWorks Options
16165 The options in this section are defined for all VxWorks targets.
16166 Options specific to the target hardware are listed with the other
16167 options for that target.
16172 GCC can generate code for both VxWorks kernels and real time processes
16173 (RTPs). This option switches from the former to the latter. It also
16174 defines the preprocessor macro @code{__RTP__}.
16177 @opindex non-static
16178 Link an RTP executable against shared libraries rather than static
16179 libraries. The options @option{-static} and @option{-shared} can
16180 also be used for RTPs (@pxref{Link Options}); @option{-static}
16187 These options are passed down to the linker. They are defined for
16188 compatibility with Diab.
16191 @opindex Xbind-lazy
16192 Enable lazy binding of function calls. This option is equivalent to
16193 @option{-Wl,-z,now} and is defined for compatibility with Diab.
16197 Disable lazy binding of function calls. This option is the default and
16198 is defined for compatibility with Diab.
16201 @node x86-64 Options
16202 @subsection x86-64 Options
16203 @cindex x86-64 options
16205 These are listed under @xref{i386 and x86-64 Options}.
16207 @node i386 and x86-64 Windows Options
16208 @subsection i386 and x86-64 Windows Options
16209 @cindex i386 and x86-64 Windows Options
16211 These additional options are available for Windows targets:
16216 This option is available for Cygwin and MinGW targets. It
16217 specifies that a console application is to be generated, by
16218 instructing the linker to set the PE header subsystem type
16219 required for console applications.
16220 This is the default behavior for Cygwin and MinGW targets.
16224 This option is available for Cygwin targets. It specifies that
16225 the Cygwin internal interface is to be used for predefined
16226 preprocessor macros, C runtime libraries and related linker
16227 paths and options. For Cygwin targets this is the default behavior.
16228 This option is deprecated and will be removed in a future release.
16231 @opindex mno-cygwin
16232 This option is available for Cygwin targets. It specifies that
16233 the MinGW internal interface is to be used instead of Cygwin's, by
16234 setting MinGW-related predefined macros and linker paths and default
16236 This option is deprecated and will be removed in a future release.
16240 This option is available for Cygwin and MinGW targets. It
16241 specifies that a DLL - a dynamic link library - is to be
16242 generated, enabling the selection of the required runtime
16243 startup object and entry point.
16245 @item -mnop-fun-dllimport
16246 @opindex mnop-fun-dllimport
16247 This option is available for Cygwin and MinGW targets. It
16248 specifies that the dllimport attribute should be ignored.
16252 This option is available for MinGW targets. It specifies
16253 that MinGW-specific thread support is to be used.
16257 This option is available for mingw-w64 targets. It specifies
16258 that the UNICODE macro is getting pre-defined and that the
16259 unicode capable runtime startup code is choosen.
16263 This option is available for Cygwin and MinGW targets. It
16264 specifies that the typical Windows pre-defined macros are to
16265 be set in the pre-processor, but does not influence the choice
16266 of runtime library/startup code.
16270 This option is available for Cygwin and MinGW targets. It
16271 specifies that a GUI application is to be generated by
16272 instructing the linker to set the PE header subsystem type
16275 @item -mpe-aligned-commons
16276 @opindex mpe-aligned-commons
16277 This option is available for Cygwin and MinGW targets. It
16278 specifies that the GNU extension to the PE file format that
16279 permits the correct alignment of COMMON variables should be
16280 used when generating code. It will be enabled by default if
16281 GCC detects that the target assembler found during configuration
16282 supports the feature.
16285 See also under @ref{i386 and x86-64 Options} for standard options.
16287 @node Xstormy16 Options
16288 @subsection Xstormy16 Options
16289 @cindex Xstormy16 Options
16291 These options are defined for Xstormy16:
16296 Choose startup files and linker script suitable for the simulator.
16299 @node Xtensa Options
16300 @subsection Xtensa Options
16301 @cindex Xtensa Options
16303 These options are supported for Xtensa targets:
16307 @itemx -mno-const16
16309 @opindex mno-const16
16310 Enable or disable use of @code{CONST16} instructions for loading
16311 constant values. The @code{CONST16} instruction is currently not a
16312 standard option from Tensilica. When enabled, @code{CONST16}
16313 instructions are always used in place of the standard @code{L32R}
16314 instructions. The use of @code{CONST16} is enabled by default only if
16315 the @code{L32R} instruction is not available.
16318 @itemx -mno-fused-madd
16319 @opindex mfused-madd
16320 @opindex mno-fused-madd
16321 Enable or disable use of fused multiply/add and multiply/subtract
16322 instructions in the floating-point option. This has no effect if the
16323 floating-point option is not also enabled. Disabling fused multiply/add
16324 and multiply/subtract instructions forces the compiler to use separate
16325 instructions for the multiply and add/subtract operations. This may be
16326 desirable in some cases where strict IEEE 754-compliant results are
16327 required: the fused multiply add/subtract instructions do not round the
16328 intermediate result, thereby producing results with @emph{more} bits of
16329 precision than specified by the IEEE standard. Disabling fused multiply
16330 add/subtract instructions also ensures that the program output is not
16331 sensitive to the compiler's ability to combine multiply and add/subtract
16334 @item -mserialize-volatile
16335 @itemx -mno-serialize-volatile
16336 @opindex mserialize-volatile
16337 @opindex mno-serialize-volatile
16338 When this option is enabled, GCC inserts @code{MEMW} instructions before
16339 @code{volatile} memory references to guarantee sequential consistency.
16340 The default is @option{-mserialize-volatile}. Use
16341 @option{-mno-serialize-volatile} to omit the @code{MEMW} instructions.
16343 @item -mtext-section-literals
16344 @itemx -mno-text-section-literals
16345 @opindex mtext-section-literals
16346 @opindex mno-text-section-literals
16347 Control the treatment of literal pools. The default is
16348 @option{-mno-text-section-literals}, which places literals in a separate
16349 section in the output file. This allows the literal pool to be placed
16350 in a data RAM/ROM, and it also allows the linker to combine literal
16351 pools from separate object files to remove redundant literals and
16352 improve code size. With @option{-mtext-section-literals}, the literals
16353 are interspersed in the text section in order to keep them as close as
16354 possible to their references. This may be necessary for large assembly
16357 @item -mtarget-align
16358 @itemx -mno-target-align
16359 @opindex mtarget-align
16360 @opindex mno-target-align
16361 When this option is enabled, GCC instructs the assembler to
16362 automatically align instructions to reduce branch penalties at the
16363 expense of some code density. The assembler attempts to widen density
16364 instructions to align branch targets and the instructions following call
16365 instructions. If there are not enough preceding safe density
16366 instructions to align a target, no widening will be performed. The
16367 default is @option{-mtarget-align}. These options do not affect the
16368 treatment of auto-aligned instructions like @code{LOOP}, which the
16369 assembler will always align, either by widening density instructions or
16370 by inserting no-op instructions.
16373 @itemx -mno-longcalls
16374 @opindex mlongcalls
16375 @opindex mno-longcalls
16376 When this option is enabled, GCC instructs the assembler to translate
16377 direct calls to indirect calls unless it can determine that the target
16378 of a direct call is in the range allowed by the call instruction. This
16379 translation typically occurs for calls to functions in other source
16380 files. Specifically, the assembler translates a direct @code{CALL}
16381 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
16382 The default is @option{-mno-longcalls}. This option should be used in
16383 programs where the call target can potentially be out of range. This
16384 option is implemented in the assembler, not the compiler, so the
16385 assembly code generated by GCC will still show direct call
16386 instructions---look at the disassembled object code to see the actual
16387 instructions. Note that the assembler will use an indirect call for
16388 every cross-file call, not just those that really will be out of range.
16391 @node zSeries Options
16392 @subsection zSeries Options
16393 @cindex zSeries options
16395 These are listed under @xref{S/390 and zSeries Options}.
16397 @node Code Gen Options
16398 @section Options for Code Generation Conventions
16399 @cindex code generation conventions
16400 @cindex options, code generation
16401 @cindex run-time options
16403 These machine-independent options control the interface conventions
16404 used in code generation.
16406 Most of them have both positive and negative forms; the negative form
16407 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
16408 one of the forms is listed---the one which is not the default. You
16409 can figure out the other form by either removing @samp{no-} or adding
16413 @item -fbounds-check
16414 @opindex fbounds-check
16415 For front-ends that support it, generate additional code to check that
16416 indices used to access arrays are within the declared range. This is
16417 currently only supported by the Java and Fortran front-ends, where
16418 this option defaults to true and false respectively.
16422 This option generates traps for signed overflow on addition, subtraction,
16423 multiplication operations.
16427 This option instructs the compiler to assume that signed arithmetic
16428 overflow of addition, subtraction and multiplication wraps around
16429 using twos-complement representation. This flag enables some optimizations
16430 and disables others. This option is enabled by default for the Java
16431 front-end, as required by the Java language specification.
16434 @opindex fexceptions
16435 Enable exception handling. Generates extra code needed to propagate
16436 exceptions. For some targets, this implies GCC will generate frame
16437 unwind information for all functions, which can produce significant data
16438 size overhead, although it does not affect execution. If you do not
16439 specify this option, GCC will enable it by default for languages like
16440 C++ which normally require exception handling, and disable it for
16441 languages like C that do not normally require it. However, you may need
16442 to enable this option when compiling C code that needs to interoperate
16443 properly with exception handlers written in C++. You may also wish to
16444 disable this option if you are compiling older C++ programs that don't
16445 use exception handling.
16447 @item -fnon-call-exceptions
16448 @opindex fnon-call-exceptions
16449 Generate code that allows trapping instructions to throw exceptions.
16450 Note that this requires platform-specific runtime support that does
16451 not exist everywhere. Moreover, it only allows @emph{trapping}
16452 instructions to throw exceptions, i.e.@: memory references or floating
16453 point instructions. It does not allow exceptions to be thrown from
16454 arbitrary signal handlers such as @code{SIGALRM}.
16456 @item -funwind-tables
16457 @opindex funwind-tables
16458 Similar to @option{-fexceptions}, except that it will just generate any needed
16459 static data, but will not affect the generated code in any other way.
16460 You will normally not enable this option; instead, a language processor
16461 that needs this handling would enable it on your behalf.
16463 @item -fasynchronous-unwind-tables
16464 @opindex fasynchronous-unwind-tables
16465 Generate unwind table in dwarf2 format, if supported by target machine. The
16466 table is exact at each instruction boundary, so it can be used for stack
16467 unwinding from asynchronous events (such as debugger or garbage collector).
16469 @item -fpcc-struct-return
16470 @opindex fpcc-struct-return
16471 Return ``short'' @code{struct} and @code{union} values in memory like
16472 longer ones, rather than in registers. This convention is less
16473 efficient, but it has the advantage of allowing intercallability between
16474 GCC-compiled files and files compiled with other compilers, particularly
16475 the Portable C Compiler (pcc).
16477 The precise convention for returning structures in memory depends
16478 on the target configuration macros.
16480 Short structures and unions are those whose size and alignment match
16481 that of some integer type.
16483 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
16484 switch is not binary compatible with code compiled with the
16485 @option{-freg-struct-return} switch.
16486 Use it to conform to a non-default application binary interface.
16488 @item -freg-struct-return
16489 @opindex freg-struct-return
16490 Return @code{struct} and @code{union} values in registers when possible.
16491 This is more efficient for small structures than
16492 @option{-fpcc-struct-return}.
16494 If you specify neither @option{-fpcc-struct-return} nor
16495 @option{-freg-struct-return}, GCC defaults to whichever convention is
16496 standard for the target. If there is no standard convention, GCC
16497 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
16498 the principal compiler. In those cases, we can choose the standard, and
16499 we chose the more efficient register return alternative.
16501 @strong{Warning:} code compiled with the @option{-freg-struct-return}
16502 switch is not binary compatible with code compiled with the
16503 @option{-fpcc-struct-return} switch.
16504 Use it to conform to a non-default application binary interface.
16506 @item -fshort-enums
16507 @opindex fshort-enums
16508 Allocate to an @code{enum} type only as many bytes as it needs for the
16509 declared range of possible values. Specifically, the @code{enum} type
16510 will be equivalent to the smallest integer type which has enough room.
16512 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
16513 code that is not binary compatible with code generated without that switch.
16514 Use it to conform to a non-default application binary interface.
16516 @item -fshort-double
16517 @opindex fshort-double
16518 Use the same size for @code{double} as for @code{float}.
16520 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
16521 code that is not binary compatible with code generated without that switch.
16522 Use it to conform to a non-default application binary interface.
16524 @item -fshort-wchar
16525 @opindex fshort-wchar
16526 Override the underlying type for @samp{wchar_t} to be @samp{short
16527 unsigned int} instead of the default for the target. This option is
16528 useful for building programs to run under WINE@.
16530 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
16531 code that is not binary compatible with code generated without that switch.
16532 Use it to conform to a non-default application binary interface.
16535 @opindex fno-common
16536 In C code, controls the placement of uninitialized global variables.
16537 Unix C compilers have traditionally permitted multiple definitions of
16538 such variables in different compilation units by placing the variables
16540 This is the behavior specified by @option{-fcommon}, and is the default
16541 for GCC on most targets.
16542 On the other hand, this behavior is not required by ISO C, and on some
16543 targets may carry a speed or code size penalty on variable references.
16544 The @option{-fno-common} option specifies that the compiler should place
16545 uninitialized global variables in the data section of the object file,
16546 rather than generating them as common blocks.
16547 This has the effect that if the same variable is declared
16548 (without @code{extern}) in two different compilations,
16549 you will get a multiple-definition error when you link them.
16550 In this case, you must compile with @option{-fcommon} instead.
16551 Compiling with @option{-fno-common} is useful on targets for which
16552 it provides better performance, or if you wish to verify that the
16553 program will work on other systems which always treat uninitialized
16554 variable declarations this way.
16558 Ignore the @samp{#ident} directive.
16560 @item -finhibit-size-directive
16561 @opindex finhibit-size-directive
16562 Don't output a @code{.size} assembler directive, or anything else that
16563 would cause trouble if the function is split in the middle, and the
16564 two halves are placed at locations far apart in memory. This option is
16565 used when compiling @file{crtstuff.c}; you should not need to use it
16568 @item -fverbose-asm
16569 @opindex fverbose-asm
16570 Put extra commentary information in the generated assembly code to
16571 make it more readable. This option is generally only of use to those
16572 who actually need to read the generated assembly code (perhaps while
16573 debugging the compiler itself).
16575 @option{-fno-verbose-asm}, the default, causes the
16576 extra information to be omitted and is useful when comparing two assembler
16579 @item -frecord-gcc-switches
16580 @opindex frecord-gcc-switches
16581 This switch causes the command line that was used to invoke the
16582 compiler to be recorded into the object file that is being created.
16583 This switch is only implemented on some targets and the exact format
16584 of the recording is target and binary file format dependent, but it
16585 usually takes the form of a section containing ASCII text. This
16586 switch is related to the @option{-fverbose-asm} switch, but that
16587 switch only records information in the assembler output file as
16588 comments, so it never reaches the object file.
16592 @cindex global offset table
16594 Generate position-independent code (PIC) suitable for use in a shared
16595 library, if supported for the target machine. Such code accesses all
16596 constant addresses through a global offset table (GOT)@. The dynamic
16597 loader resolves the GOT entries when the program starts (the dynamic
16598 loader is not part of GCC; it is part of the operating system). If
16599 the GOT size for the linked executable exceeds a machine-specific
16600 maximum size, you get an error message from the linker indicating that
16601 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
16602 instead. (These maximums are 8k on the SPARC and 32k
16603 on the m68k and RS/6000. The 386 has no such limit.)
16605 Position-independent code requires special support, and therefore works
16606 only on certain machines. For the 386, GCC supports PIC for System V
16607 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
16608 position-independent.
16610 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
16615 If supported for the target machine, emit position-independent code,
16616 suitable for dynamic linking and avoiding any limit on the size of the
16617 global offset table. This option makes a difference on the m68k,
16618 PowerPC and SPARC@.
16620 Position-independent code requires special support, and therefore works
16621 only on certain machines.
16623 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
16630 These options are similar to @option{-fpic} and @option{-fPIC}, but
16631 generated position independent code can be only linked into executables.
16632 Usually these options are used when @option{-pie} GCC option will be
16633 used during linking.
16635 @option{-fpie} and @option{-fPIE} both define the macros
16636 @code{__pie__} and @code{__PIE__}. The macros have the value 1
16637 for @option{-fpie} and 2 for @option{-fPIE}.
16639 @item -fno-jump-tables
16640 @opindex fno-jump-tables
16641 Do not use jump tables for switch statements even where it would be
16642 more efficient than other code generation strategies. This option is
16643 of use in conjunction with @option{-fpic} or @option{-fPIC} for
16644 building code which forms part of a dynamic linker and cannot
16645 reference the address of a jump table. On some targets, jump tables
16646 do not require a GOT and this option is not needed.
16648 @item -ffixed-@var{reg}
16650 Treat the register named @var{reg} as a fixed register; generated code
16651 should never refer to it (except perhaps as a stack pointer, frame
16652 pointer or in some other fixed role).
16654 @var{reg} must be the name of a register. The register names accepted
16655 are machine-specific and are defined in the @code{REGISTER_NAMES}
16656 macro in the machine description macro file.
16658 This flag does not have a negative form, because it specifies a
16661 @item -fcall-used-@var{reg}
16662 @opindex fcall-used
16663 Treat the register named @var{reg} as an allocable register that is
16664 clobbered by function calls. It may be allocated for temporaries or
16665 variables that do not live across a call. Functions compiled this way
16666 will not save and restore the register @var{reg}.
16668 It is an error to used this flag with the frame pointer or stack pointer.
16669 Use of this flag for other registers that have fixed pervasive roles in
16670 the machine's execution model will produce disastrous results.
16672 This flag does not have a negative form, because it specifies a
16675 @item -fcall-saved-@var{reg}
16676 @opindex fcall-saved
16677 Treat the register named @var{reg} as an allocable register saved by
16678 functions. It may be allocated even for temporaries or variables that
16679 live across a call. Functions compiled this way will save and restore
16680 the register @var{reg} if they use it.
16682 It is an error to used this flag with the frame pointer or stack pointer.
16683 Use of this flag for other registers that have fixed pervasive roles in
16684 the machine's execution model will produce disastrous results.
16686 A different sort of disaster will result from the use of this flag for
16687 a register in which function values may be returned.
16689 This flag does not have a negative form, because it specifies a
16692 @item -fpack-struct[=@var{n}]
16693 @opindex fpack-struct
16694 Without a value specified, pack all structure members together without
16695 holes. When a value is specified (which must be a small power of two), pack
16696 structure members according to this value, representing the maximum
16697 alignment (that is, objects with default alignment requirements larger than
16698 this will be output potentially unaligned at the next fitting location.
16700 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
16701 code that is not binary compatible with code generated without that switch.
16702 Additionally, it makes the code suboptimal.
16703 Use it to conform to a non-default application binary interface.
16705 @item -finstrument-functions
16706 @opindex finstrument-functions
16707 Generate instrumentation calls for entry and exit to functions. Just
16708 after function entry and just before function exit, the following
16709 profiling functions will be called with the address of the current
16710 function and its call site. (On some platforms,
16711 @code{__builtin_return_address} does not work beyond the current
16712 function, so the call site information may not be available to the
16713 profiling functions otherwise.)
16716 void __cyg_profile_func_enter (void *this_fn,
16718 void __cyg_profile_func_exit (void *this_fn,
16722 The first argument is the address of the start of the current function,
16723 which may be looked up exactly in the symbol table.
16725 This instrumentation is also done for functions expanded inline in other
16726 functions. The profiling calls will indicate where, conceptually, the
16727 inline function is entered and exited. This means that addressable
16728 versions of such functions must be available. If all your uses of a
16729 function are expanded inline, this may mean an additional expansion of
16730 code size. If you use @samp{extern inline} in your C code, an
16731 addressable version of such functions must be provided. (This is
16732 normally the case anyways, but if you get lucky and the optimizer always
16733 expands the functions inline, you might have gotten away without
16734 providing static copies.)
16736 A function may be given the attribute @code{no_instrument_function}, in
16737 which case this instrumentation will not be done. This can be used, for
16738 example, for the profiling functions listed above, high-priority
16739 interrupt routines, and any functions from which the profiling functions
16740 cannot safely be called (perhaps signal handlers, if the profiling
16741 routines generate output or allocate memory).
16743 @item -finstrument-functions-exclude-file-list=@var{file},@var{file},@dots{}
16744 @opindex finstrument-functions-exclude-file-list
16746 Set the list of functions that are excluded from instrumentation (see
16747 the description of @code{-finstrument-functions}). If the file that
16748 contains a function definition matches with one of @var{file}, then
16749 that function is not instrumented. The match is done on substrings:
16750 if the @var{file} parameter is a substring of the file name, it is
16751 considered to be a match.
16754 @code{-finstrument-functions-exclude-file-list=/bits/stl,include/sys}
16755 will exclude any inline function defined in files whose pathnames
16756 contain @code{/bits/stl} or @code{include/sys}.
16758 If, for some reason, you want to include letter @code{','} in one of
16759 @var{sym}, write @code{'\,'}. For example,
16760 @code{-finstrument-functions-exclude-file-list='\,\,tmp'}
16761 (note the single quote surrounding the option).
16763 @item -finstrument-functions-exclude-function-list=@var{sym},@var{sym},@dots{}
16764 @opindex finstrument-functions-exclude-function-list
16766 This is similar to @code{-finstrument-functions-exclude-file-list},
16767 but this option sets the list of function names to be excluded from
16768 instrumentation. The function name to be matched is its user-visible
16769 name, such as @code{vector<int> blah(const vector<int> &)}, not the
16770 internal mangled name (e.g., @code{_Z4blahRSt6vectorIiSaIiEE}). The
16771 match is done on substrings: if the @var{sym} parameter is a substring
16772 of the function name, it is considered to be a match. For C99 and C++
16773 extended identifiers, the function name must be given in UTF-8, not
16774 using universal character names.
16776 @item -fstack-check
16777 @opindex fstack-check
16778 Generate code to verify that you do not go beyond the boundary of the
16779 stack. You should specify this flag if you are running in an
16780 environment with multiple threads, but only rarely need to specify it in
16781 a single-threaded environment since stack overflow is automatically
16782 detected on nearly all systems if there is only one stack.
16784 Note that this switch does not actually cause checking to be done; the
16785 operating system or the language runtime must do that. The switch causes
16786 generation of code to ensure that they see the stack being extended.
16788 You can additionally specify a string parameter: @code{no} means no
16789 checking, @code{generic} means force the use of old-style checking,
16790 @code{specific} means use the best checking method and is equivalent
16791 to bare @option{-fstack-check}.
16793 Old-style checking is a generic mechanism that requires no specific
16794 target support in the compiler but comes with the following drawbacks:
16798 Modified allocation strategy for large objects: they will always be
16799 allocated dynamically if their size exceeds a fixed threshold.
16802 Fixed limit on the size of the static frame of functions: when it is
16803 topped by a particular function, stack checking is not reliable and
16804 a warning is issued by the compiler.
16807 Inefficiency: because of both the modified allocation strategy and the
16808 generic implementation, the performances of the code are hampered.
16811 Note that old-style stack checking is also the fallback method for
16812 @code{specific} if no target support has been added in the compiler.
16814 @item -fstack-limit-register=@var{reg}
16815 @itemx -fstack-limit-symbol=@var{sym}
16816 @itemx -fno-stack-limit
16817 @opindex fstack-limit-register
16818 @opindex fstack-limit-symbol
16819 @opindex fno-stack-limit
16820 Generate code to ensure that the stack does not grow beyond a certain value,
16821 either the value of a register or the address of a symbol. If the stack
16822 would grow beyond the value, a signal is raised. For most targets,
16823 the signal is raised before the stack overruns the boundary, so
16824 it is possible to catch the signal without taking special precautions.
16826 For instance, if the stack starts at absolute address @samp{0x80000000}
16827 and grows downwards, you can use the flags
16828 @option{-fstack-limit-symbol=__stack_limit} and
16829 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
16830 of 128KB@. Note that this may only work with the GNU linker.
16832 @cindex aliasing of parameters
16833 @cindex parameters, aliased
16834 @item -fargument-alias
16835 @itemx -fargument-noalias
16836 @itemx -fargument-noalias-global
16837 @itemx -fargument-noalias-anything
16838 @opindex fargument-alias
16839 @opindex fargument-noalias
16840 @opindex fargument-noalias-global
16841 @opindex fargument-noalias-anything
16842 Specify the possible relationships among parameters and between
16843 parameters and global data.
16845 @option{-fargument-alias} specifies that arguments (parameters) may
16846 alias each other and may alias global storage.@*
16847 @option{-fargument-noalias} specifies that arguments do not alias
16848 each other, but may alias global storage.@*
16849 @option{-fargument-noalias-global} specifies that arguments do not
16850 alias each other and do not alias global storage.
16851 @option{-fargument-noalias-anything} specifies that arguments do not
16852 alias any other storage.
16854 Each language will automatically use whatever option is required by
16855 the language standard. You should not need to use these options yourself.
16857 @item -fleading-underscore
16858 @opindex fleading-underscore
16859 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
16860 change the way C symbols are represented in the object file. One use
16861 is to help link with legacy assembly code.
16863 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
16864 generate code that is not binary compatible with code generated without that
16865 switch. Use it to conform to a non-default application binary interface.
16866 Not all targets provide complete support for this switch.
16868 @item -ftls-model=@var{model}
16869 @opindex ftls-model
16870 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
16871 The @var{model} argument should be one of @code{global-dynamic},
16872 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
16874 The default without @option{-fpic} is @code{initial-exec}; with
16875 @option{-fpic} the default is @code{global-dynamic}.
16877 @item -fvisibility=@var{default|internal|hidden|protected}
16878 @opindex fvisibility
16879 Set the default ELF image symbol visibility to the specified option---all
16880 symbols will be marked with this unless overridden within the code.
16881 Using this feature can very substantially improve linking and
16882 load times of shared object libraries, produce more optimized
16883 code, provide near-perfect API export and prevent symbol clashes.
16884 It is @strong{strongly} recommended that you use this in any shared objects
16887 Despite the nomenclature, @code{default} always means public ie;
16888 available to be linked against from outside the shared object.
16889 @code{protected} and @code{internal} are pretty useless in real-world
16890 usage so the only other commonly used option will be @code{hidden}.
16891 The default if @option{-fvisibility} isn't specified is
16892 @code{default}, i.e., make every
16893 symbol public---this causes the same behavior as previous versions of
16896 A good explanation of the benefits offered by ensuring ELF
16897 symbols have the correct visibility is given by ``How To Write
16898 Shared Libraries'' by Ulrich Drepper (which can be found at
16899 @w{@uref{http://people.redhat.com/~drepper/}})---however a superior
16900 solution made possible by this option to marking things hidden when
16901 the default is public is to make the default hidden and mark things
16902 public. This is the norm with DLL's on Windows and with @option{-fvisibility=hidden}
16903 and @code{__attribute__ ((visibility("default")))} instead of
16904 @code{__declspec(dllexport)} you get almost identical semantics with
16905 identical syntax. This is a great boon to those working with
16906 cross-platform projects.
16908 For those adding visibility support to existing code, you may find
16909 @samp{#pragma GCC visibility} of use. This works by you enclosing
16910 the declarations you wish to set visibility for with (for example)
16911 @samp{#pragma GCC visibility push(hidden)} and
16912 @samp{#pragma GCC visibility pop}.
16913 Bear in mind that symbol visibility should be viewed @strong{as
16914 part of the API interface contract} and thus all new code should
16915 always specify visibility when it is not the default ie; declarations
16916 only for use within the local DSO should @strong{always} be marked explicitly
16917 as hidden as so to avoid PLT indirection overheads---making this
16918 abundantly clear also aids readability and self-documentation of the code.
16919 Note that due to ISO C++ specification requirements, operator new and
16920 operator delete must always be of default visibility.
16922 Be aware that headers from outside your project, in particular system
16923 headers and headers from any other library you use, may not be
16924 expecting to be compiled with visibility other than the default. You
16925 may need to explicitly say @samp{#pragma GCC visibility push(default)}
16926 before including any such headers.
16928 @samp{extern} declarations are not affected by @samp{-fvisibility}, so
16929 a lot of code can be recompiled with @samp{-fvisibility=hidden} with
16930 no modifications. However, this means that calls to @samp{extern}
16931 functions with no explicit visibility will use the PLT, so it is more
16932 effective to use @samp{__attribute ((visibility))} and/or
16933 @samp{#pragma GCC visibility} to tell the compiler which @samp{extern}
16934 declarations should be treated as hidden.
16936 Note that @samp{-fvisibility} does affect C++ vague linkage
16937 entities. This means that, for instance, an exception class that will
16938 be thrown between DSOs must be explicitly marked with default
16939 visibility so that the @samp{type_info} nodes will be unified between
16942 An overview of these techniques, their benefits and how to use them
16943 is at @w{@uref{http://gcc.gnu.org/wiki/Visibility}}.
16949 @node Environment Variables
16950 @section Environment Variables Affecting GCC
16951 @cindex environment variables
16953 @c man begin ENVIRONMENT
16954 This section describes several environment variables that affect how GCC
16955 operates. Some of them work by specifying directories or prefixes to use
16956 when searching for various kinds of files. Some are used to specify other
16957 aspects of the compilation environment.
16959 Note that you can also specify places to search using options such as
16960 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
16961 take precedence over places specified using environment variables, which
16962 in turn take precedence over those specified by the configuration of GCC@.
16963 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
16964 GNU Compiler Collection (GCC) Internals}.
16969 @c @itemx LC_COLLATE
16971 @c @itemx LC_MONETARY
16972 @c @itemx LC_NUMERIC
16977 @c @findex LC_COLLATE
16978 @findex LC_MESSAGES
16979 @c @findex LC_MONETARY
16980 @c @findex LC_NUMERIC
16984 These environment variables control the way that GCC uses
16985 localization information that allow GCC to work with different
16986 national conventions. GCC inspects the locale categories
16987 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
16988 so. These locale categories can be set to any value supported by your
16989 installation. A typical value is @samp{en_GB.UTF-8} for English in the United
16990 Kingdom encoded in UTF-8.
16992 The @env{LC_CTYPE} environment variable specifies character
16993 classification. GCC uses it to determine the character boundaries in
16994 a string; this is needed for some multibyte encodings that contain quote
16995 and escape characters that would otherwise be interpreted as a string
16998 The @env{LC_MESSAGES} environment variable specifies the language to
16999 use in diagnostic messages.
17001 If the @env{LC_ALL} environment variable is set, it overrides the value
17002 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
17003 and @env{LC_MESSAGES} default to the value of the @env{LANG}
17004 environment variable. If none of these variables are set, GCC
17005 defaults to traditional C English behavior.
17009 If @env{TMPDIR} is set, it specifies the directory to use for temporary
17010 files. GCC uses temporary files to hold the output of one stage of
17011 compilation which is to be used as input to the next stage: for example,
17012 the output of the preprocessor, which is the input to the compiler
17015 @item GCC_EXEC_PREFIX
17016 @findex GCC_EXEC_PREFIX
17017 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
17018 names of the subprograms executed by the compiler. No slash is added
17019 when this prefix is combined with the name of a subprogram, but you can
17020 specify a prefix that ends with a slash if you wish.
17022 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
17023 an appropriate prefix to use based on the pathname it was invoked with.
17025 If GCC cannot find the subprogram using the specified prefix, it
17026 tries looking in the usual places for the subprogram.
17028 The default value of @env{GCC_EXEC_PREFIX} is
17029 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the prefix to
17030 the installed compiler. In many cases @var{prefix} is the value
17031 of @code{prefix} when you ran the @file{configure} script.
17033 Other prefixes specified with @option{-B} take precedence over this prefix.
17035 This prefix is also used for finding files such as @file{crt0.o} that are
17038 In addition, the prefix is used in an unusual way in finding the
17039 directories to search for header files. For each of the standard
17040 directories whose name normally begins with @samp{/usr/local/lib/gcc}
17041 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
17042 replacing that beginning with the specified prefix to produce an
17043 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
17044 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
17045 These alternate directories are searched first; the standard directories
17046 come next. If a standard directory begins with the configured
17047 @var{prefix} then the value of @var{prefix} is replaced by
17048 @env{GCC_EXEC_PREFIX} when looking for header files.
17050 @item COMPILER_PATH
17051 @findex COMPILER_PATH
17052 The value of @env{COMPILER_PATH} is a colon-separated list of
17053 directories, much like @env{PATH}. GCC tries the directories thus
17054 specified when searching for subprograms, if it can't find the
17055 subprograms using @env{GCC_EXEC_PREFIX}.
17058 @findex LIBRARY_PATH
17059 The value of @env{LIBRARY_PATH} is a colon-separated list of
17060 directories, much like @env{PATH}. When configured as a native compiler,
17061 GCC tries the directories thus specified when searching for special
17062 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
17063 using GCC also uses these directories when searching for ordinary
17064 libraries for the @option{-l} option (but directories specified with
17065 @option{-L} come first).
17069 @cindex locale definition
17070 This variable is used to pass locale information to the compiler. One way in
17071 which this information is used is to determine the character set to be used
17072 when character literals, string literals and comments are parsed in C and C++.
17073 When the compiler is configured to allow multibyte characters,
17074 the following values for @env{LANG} are recognized:
17078 Recognize JIS characters.
17080 Recognize SJIS characters.
17082 Recognize EUCJP characters.
17085 If @env{LANG} is not defined, or if it has some other value, then the
17086 compiler will use mblen and mbtowc as defined by the default locale to
17087 recognize and translate multibyte characters.
17091 Some additional environments variables affect the behavior of the
17094 @include cppenv.texi
17098 @node Precompiled Headers
17099 @section Using Precompiled Headers
17100 @cindex precompiled headers
17101 @cindex speed of compilation
17103 Often large projects have many header files that are included in every
17104 source file. The time the compiler takes to process these header files
17105 over and over again can account for nearly all of the time required to
17106 build the project. To make builds faster, GCC allows users to
17107 `precompile' a header file; then, if builds can use the precompiled
17108 header file they will be much faster.
17110 To create a precompiled header file, simply compile it as you would any
17111 other file, if necessary using the @option{-x} option to make the driver
17112 treat it as a C or C++ header file. You will probably want to use a
17113 tool like @command{make} to keep the precompiled header up-to-date when
17114 the headers it contains change.
17116 A precompiled header file will be searched for when @code{#include} is
17117 seen in the compilation. As it searches for the included file
17118 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
17119 compiler looks for a precompiled header in each directory just before it
17120 looks for the include file in that directory. The name searched for is
17121 the name specified in the @code{#include} with @samp{.gch} appended. If
17122 the precompiled header file can't be used, it is ignored.
17124 For instance, if you have @code{#include "all.h"}, and you have
17125 @file{all.h.gch} in the same directory as @file{all.h}, then the
17126 precompiled header file will be used if possible, and the original
17127 header will be used otherwise.
17129 Alternatively, you might decide to put the precompiled header file in a
17130 directory and use @option{-I} to ensure that directory is searched
17131 before (or instead of) the directory containing the original header.
17132 Then, if you want to check that the precompiled header file is always
17133 used, you can put a file of the same name as the original header in this
17134 directory containing an @code{#error} command.
17136 This also works with @option{-include}. So yet another way to use
17137 precompiled headers, good for projects not designed with precompiled
17138 header files in mind, is to simply take most of the header files used by
17139 a project, include them from another header file, precompile that header
17140 file, and @option{-include} the precompiled header. If the header files
17141 have guards against multiple inclusion, they will be skipped because
17142 they've already been included (in the precompiled header).
17144 If you need to precompile the same header file for different
17145 languages, targets, or compiler options, you can instead make a
17146 @emph{directory} named like @file{all.h.gch}, and put each precompiled
17147 header in the directory, perhaps using @option{-o}. It doesn't matter
17148 what you call the files in the directory, every precompiled header in
17149 the directory will be considered. The first precompiled header
17150 encountered in the directory that is valid for this compilation will
17151 be used; they're searched in no particular order.
17153 There are many other possibilities, limited only by your imagination,
17154 good sense, and the constraints of your build system.
17156 A precompiled header file can be used only when these conditions apply:
17160 Only one precompiled header can be used in a particular compilation.
17163 A precompiled header can't be used once the first C token is seen. You
17164 can have preprocessor directives before a precompiled header; you can
17165 even include a precompiled header from inside another header, so long as
17166 there are no C tokens before the @code{#include}.
17169 The precompiled header file must be produced for the same language as
17170 the current compilation. You can't use a C precompiled header for a C++
17174 The precompiled header file must have been produced by the same compiler
17175 binary as the current compilation is using.
17178 Any macros defined before the precompiled header is included must
17179 either be defined in the same way as when the precompiled header was
17180 generated, or must not affect the precompiled header, which usually
17181 means that they don't appear in the precompiled header at all.
17183 The @option{-D} option is one way to define a macro before a
17184 precompiled header is included; using a @code{#define} can also do it.
17185 There are also some options that define macros implicitly, like
17186 @option{-O} and @option{-Wdeprecated}; the same rule applies to macros
17189 @item If debugging information is output when using the precompiled
17190 header, using @option{-g} or similar, the same kind of debugging information
17191 must have been output when building the precompiled header. However,
17192 a precompiled header built using @option{-g} can be used in a compilation
17193 when no debugging information is being output.
17195 @item The same @option{-m} options must generally be used when building
17196 and using the precompiled header. @xref{Submodel Options},
17197 for any cases where this rule is relaxed.
17199 @item Each of the following options must be the same when building and using
17200 the precompiled header:
17202 @gccoptlist{-fexceptions}
17205 Some other command-line options starting with @option{-f},
17206 @option{-p}, or @option{-O} must be defined in the same way as when
17207 the precompiled header was generated. At present, it's not clear
17208 which options are safe to change and which are not; the safest choice
17209 is to use exactly the same options when generating and using the
17210 precompiled header. The following are known to be safe:
17212 @gccoptlist{-fmessage-length= -fpreprocessed -fsched-interblock @gol
17213 -fsched-spec -fsched-spec-load -fsched-spec-load-dangerous @gol
17214 -fsched-verbose=<number> -fschedule-insns -fvisibility= @gol
17219 For all of these except the last, the compiler will automatically
17220 ignore the precompiled header if the conditions aren't met. If you
17221 find an option combination that doesn't work and doesn't cause the
17222 precompiled header to be ignored, please consider filing a bug report,
17225 If you do use differing options when generating and using the
17226 precompiled header, the actual behavior will be a mixture of the
17227 behavior for the options. For instance, if you use @option{-g} to
17228 generate the precompiled header but not when using it, you may or may
17229 not get debugging information for routines in the precompiled header.