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, 2010
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, 2010
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 1.3 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} @gol
168 -fdump-ada-spec@r{[}-slim@r{]}}
170 @item C Language Options
171 @xref{C Dialect Options,,Options Controlling C Dialect}.
172 @gccoptlist{-ansi -std=@var{standard} -fgnu89-inline @gol
173 -aux-info @var{filename} @gol
174 -fno-asm -fno-builtin -fno-builtin-@var{function} @gol
175 -fhosted -ffreestanding -fopenmp -fms-extensions @gol
176 -trigraphs -no-integrated-cpp -traditional -traditional-cpp @gol
177 -fallow-single-precision -fcond-mismatch -flax-vector-conversions @gol
178 -fsigned-bitfields -fsigned-char @gol
179 -funsigned-bitfields -funsigned-char}
181 @item C++ Language Options
182 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
183 @gccoptlist{-fabi-version=@var{n} -fno-access-control -fcheck-new @gol
184 -fconserve-space -ffriend-injection @gol
185 -fno-elide-constructors @gol
186 -fno-enforce-eh-specs @gol
187 -ffor-scope -fno-for-scope -fno-gnu-keywords @gol
188 -fno-implicit-templates @gol
189 -fno-implicit-inline-templates @gol
190 -fno-implement-inlines -fms-extensions @gol
191 -fno-nonansi-builtins -fnothrow-opt -fno-operator-names @gol
192 -fno-optional-diags -fpermissive @gol
193 -fno-pretty-templates @gol
194 -frepo -fno-rtti -fstats -ftemplate-depth=@var{n} @gol
195 -fno-threadsafe-statics -fuse-cxa-atexit -fno-weak -nostdinc++ @gol
196 -fno-default-inline -fvisibility-inlines-hidden @gol
197 -fvisibility-ms-compat @gol
198 -Wabi -Wconversion-null -Wctor-dtor-privacy @gol
199 -Wnoexcept -Wnon-virtual-dtor -Wreorder @gol
200 -Weffc++ -Wstrict-null-sentinel @gol
201 -Wno-non-template-friend -Wold-style-cast @gol
202 -Woverloaded-virtual -Wno-pmf-conversions @gol
205 @item Objective-C and Objective-C++ Language Options
206 @xref{Objective-C and Objective-C++ Dialect Options,,Options Controlling
207 Objective-C and Objective-C++ Dialects}.
208 @gccoptlist{-fconstant-string-class=@var{class-name} @gol
209 -fgnu-runtime -fnext-runtime @gol
210 -fno-nil-receivers @gol
211 -fobjc-call-cxx-cdtors @gol
212 -fobjc-direct-dispatch @gol
213 -fobjc-exceptions @gol
215 -freplace-objc-classes @gol
218 -Wassign-intercept @gol
219 -Wno-protocol -Wselector @gol
220 -Wstrict-selector-match @gol
221 -Wundeclared-selector}
223 @item Language Independent Options
224 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
225 @gccoptlist{-fmessage-length=@var{n} @gol
226 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]} @gol
227 -fdiagnostics-show-option}
229 @item Warning Options
230 @xref{Warning Options,,Options to Request or Suppress Warnings}.
231 @gccoptlist{-fsyntax-only -pedantic -pedantic-errors @gol
232 -w -Wextra -Wall -Waddress -Waggregate-return -Warray-bounds @gol
233 -Wno-attributes -Wno-builtin-macro-redefined @gol
234 -Wc++-compat -Wc++0x-compat -Wcast-align -Wcast-qual @gol
235 -Wchar-subscripts -Wclobbered -Wcomment @gol
236 -Wconversion -Wcoverage-mismatch -Wcpp -Wno-deprecated @gol
237 -Wno-deprecated-declarations -Wdisabled-optimization @gol
238 -Wno-div-by-zero -Wempty-body -Wenum-compare -Wno-endif-labels @gol
239 -Werror -Werror=* @gol
240 -Wfatal-errors -Wfloat-equal -Wformat -Wformat=2 @gol
241 -Wno-format-contains-nul -Wno-format-extra-args -Wformat-nonliteral @gol
242 -Wformat-security -Wformat-y2k @gol
243 -Wframe-larger-than=@var{len} -Wjump-misses-init -Wignored-qualifiers @gol
244 -Wimplicit -Wimplicit-function-declaration -Wimplicit-int @gol
245 -Winit-self -Winline @gol
246 -Wno-int-to-pointer-cast -Wno-invalid-offsetof @gol
247 -Winvalid-pch -Wlarger-than=@var{len} -Wunsafe-loop-optimizations @gol
248 -Wlogical-op -Wlong-long @gol
249 -Wmain -Wmissing-braces -Wmissing-field-initializers @gol
250 -Wmissing-format-attribute -Wmissing-include-dirs @gol
252 -Wno-multichar -Wnonnull -Wno-overflow @gol
253 -Woverlength-strings -Wpacked -Wpacked-bitfield-compat -Wpadded @gol
254 -Wparentheses -Wpedantic-ms-format -Wno-pedantic-ms-format @gol
255 -Wpointer-arith -Wno-pointer-to-int-cast @gol
256 -Wredundant-decls @gol
257 -Wreturn-type -Wsequence-point -Wshadow @gol
258 -Wsign-compare -Wsign-conversion -Wstack-protector @gol
259 -Wstrict-aliasing -Wstrict-aliasing=n @gol
260 -Wstrict-overflow -Wstrict-overflow=@var{n} @gol
261 -Wsuggest-attribute=@r{[}pure@r{|}const@r{|}noreturn@r{]} @gol
262 -Wswitch -Wswitch-default -Wswitch-enum -Wsync-nand @gol
263 -Wsystem-headers -Wtrampolines -Wtrigraphs -Wtype-limits -Wundef @gol
264 -Wuninitialized -Wunknown-pragmas -Wno-pragmas @gol
265 -Wunsuffixed-float-constants -Wunused -Wunused-function @gol
266 -Wunused-label -Wunused-parameter -Wno-unused-result -Wunused-value -Wunused-variable @gol
267 -Wunused-but-set-parameter -Wunused-but-set-variable -Wvariadic-macros -Wvla @gol
268 -Wvolatile-register-var -Wwrite-strings}
270 @item C and Objective-C-only Warning Options
271 @gccoptlist{-Wbad-function-cast -Wmissing-declarations @gol
272 -Wmissing-parameter-type -Wmissing-prototypes -Wnested-externs @gol
273 -Wold-style-declaration -Wold-style-definition @gol
274 -Wstrict-prototypes -Wtraditional -Wtraditional-conversion @gol
275 -Wdeclaration-after-statement -Wpointer-sign}
277 @item Debugging Options
278 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
279 @gccoptlist{-d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
280 -fdbg-cnt-list -fdbg-cnt=@var{counter-value-list} @gol
281 -fdump-noaddr -fdump-unnumbered -fdump-unnumbered-links @gol
282 -fdump-translation-unit@r{[}-@var{n}@r{]} @gol
283 -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
284 -fdump-ipa-all -fdump-ipa-cgraph -fdump-ipa-inline @gol
285 -fdump-statistics @gol
287 -fdump-tree-original@r{[}-@var{n}@r{]} @gol
288 -fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
289 -fdump-tree-cfg -fdump-tree-vcg -fdump-tree-alias @gol
291 -fdump-tree-ssa@r{[}-@var{n}@r{]} -fdump-tree-pre@r{[}-@var{n}@r{]} @gol
292 -fdump-tree-ccp@r{[}-@var{n}@r{]} -fdump-tree-dce@r{[}-@var{n}@r{]} @gol
293 -fdump-tree-gimple@r{[}-raw@r{]} -fdump-tree-mudflap@r{[}-@var{n}@r{]} @gol
294 -fdump-tree-dom@r{[}-@var{n}@r{]} @gol
295 -fdump-tree-dse@r{[}-@var{n}@r{]} @gol
296 -fdump-tree-phiprop@r{[}-@var{n}@r{]} @gol
297 -fdump-tree-phiopt@r{[}-@var{n}@r{]} @gol
298 -fdump-tree-forwprop@r{[}-@var{n}@r{]} @gol
299 -fdump-tree-copyrename@r{[}-@var{n}@r{]} @gol
300 -fdump-tree-nrv -fdump-tree-vect @gol
301 -fdump-tree-sink @gol
302 -fdump-tree-sra@r{[}-@var{n}@r{]} @gol
303 -fdump-tree-forwprop@r{[}-@var{n}@r{]} @gol
304 -fdump-tree-fre@r{[}-@var{n}@r{]} @gol
305 -fdump-tree-vrp@r{[}-@var{n}@r{]} @gol
306 -ftree-vectorizer-verbose=@var{n} @gol
307 -fdump-tree-storeccp@r{[}-@var{n}@r{]} @gol
308 -fdump-final-insns=@var{file} @gol
309 -fcompare-debug@r{[}=@var{opts}@r{]} -fcompare-debug-second @gol
310 -feliminate-dwarf2-dups -feliminate-unused-debug-types @gol
311 -feliminate-unused-debug-symbols -femit-class-debug-always @gol
312 -fenable-icf-debug @gol
313 -fmem-report -fpre-ipa-mem-report -fpost-ipa-mem-report -fprofile-arcs @gol
314 -frandom-seed=@var{string} -fsched-verbose=@var{n} @gol
315 -fsel-sched-verbose -fsel-sched-dump-cfg -fsel-sched-pipelining-verbose @gol
316 -ftest-coverage -ftime-report -fvar-tracking @gol
317 -fvar-tracking-assignments -fvar-tracking-assignments-toggle @gol
318 -g -g@var{level} -gtoggle -gcoff -gdwarf-@var{version} @gol
319 -ggdb -gstabs -gstabs+ -gstrict-dwarf -gno-strict-dwarf @gol
320 -gvms -gxcoff -gxcoff+ @gol
321 -fno-merge-debug-strings -fno-dwarf2-cfi-asm @gol
322 -fdebug-prefix-map=@var{old}=@var{new} @gol
323 -femit-struct-debug-baseonly -femit-struct-debug-reduced @gol
324 -femit-struct-debug-detailed@r{[}=@var{spec-list}@r{]} @gol
325 -p -pg -print-file-name=@var{library} -print-libgcc-file-name @gol
326 -print-multi-directory -print-multi-lib -print-multi-os-directory @gol
327 -print-prog-name=@var{program} -print-search-dirs -Q @gol
328 -print-sysroot -print-sysroot-headers-suffix @gol
329 -save-temps -save-temps=cwd -save-temps=obj -time@r{[}=@var{file}@r{]}}
331 @item Optimization Options
332 @xref{Optimize Options,,Options that Control Optimization}.
334 -falign-functions[=@var{n}] -falign-jumps[=@var{n}] @gol
335 -falign-labels[=@var{n}] -falign-loops[=@var{n}] -fassociative-math @gol
336 -fauto-inc-dec -fbranch-probabilities -fbranch-target-load-optimize @gol
337 -fbranch-target-load-optimize2 -fbtr-bb-exclusive -fcaller-saves @gol
338 -fcheck-data-deps -fconserve-stack -fcprop-registers -fcrossjumping @gol
339 -fcse-follow-jumps -fcse-skip-blocks -fcx-fortran-rules -fcx-limited-range @gol
340 -fdata-sections -fdce -fdce @gol
341 -fdelayed-branch -fdelete-null-pointer-checks -fdse -fdse @gol
342 -fearly-inlining -fipa-sra -fexpensive-optimizations -ffast-math @gol
343 -ffinite-math-only -ffloat-store -fexcess-precision=@var{style} @gol
344 -fforward-propagate -ffunction-sections @gol
345 -fgcse -fgcse-after-reload -fgcse-las -fgcse-lm -fgraphite-identity @gol
346 -fgcse-sm -fif-conversion -fif-conversion2 -findirect-inlining @gol
347 -finline-functions -finline-functions-called-once -finline-limit=@var{n} @gol
348 -finline-small-functions -fipa-cp -fipa-cp-clone -fipa-matrix-reorg -fipa-pta @gol
349 -fipa-profile -fipa-pure-const -fipa-reference -fipa-struct-reorg @gol
350 -fira-algorithm=@var{algorithm} @gol
351 -fira-region=@var{region} -fira-coalesce @gol
352 -fira-loop-pressure -fno-ira-share-save-slots @gol
353 -fno-ira-share-spill-slots -fira-verbose=@var{n} @gol
354 -fivopts -fkeep-inline-functions -fkeep-static-consts @gol
355 -floop-block -floop-interchange -floop-strip-mine @gol
356 -floop-parallelize-all -flto -flto-compression-level -flto-report -fltrans @gol
357 -fltrans-output-list -fmerge-all-constants -fmerge-constants -fmodulo-sched @gol
358 -fmodulo-sched-allow-regmoves -fmove-loop-invariants -fmudflap @gol
359 -fmudflapir -fmudflapth -fno-branch-count-reg -fno-default-inline @gol
360 -fno-defer-pop -fno-function-cse -fno-guess-branch-probability @gol
361 -fno-inline -fno-math-errno -fno-peephole -fno-peephole2 @gol
362 -fno-sched-interblock -fno-sched-spec -fno-signed-zeros @gol
363 -fno-toplevel-reorder -fno-trapping-math -fno-zero-initialized-in-bss @gol
364 -fomit-frame-pointer -foptimize-register-move -foptimize-sibling-calls @gol
365 -fpartial-inlining -fpeel-loops -fpredictive-commoning -fprefetch-loop-arrays @gol
366 -fprofile-correction -fprofile-dir=@var{path} -fprofile-generate @gol
367 -fprofile-generate=@var{path} @gol
368 -fprofile-use -fprofile-use=@var{path} -fprofile-values @gol
369 -freciprocal-math -fregmove -frename-registers -freorder-blocks @gol
370 -freorder-blocks-and-partition -freorder-functions @gol
371 -frerun-cse-after-loop -freschedule-modulo-scheduled-loops @gol
372 -frounding-math -fsched2-use-superblocks -fsched-pressure @gol
373 -fsched-spec-load -fsched-spec-load-dangerous @gol
374 -fsched-stalled-insns-dep[=@var{n}] -fsched-stalled-insns[=@var{n}] @gol
375 -fsched-group-heuristic -fsched-critical-path-heuristic @gol
376 -fsched-spec-insn-heuristic -fsched-rank-heuristic @gol
377 -fsched-last-insn-heuristic -fsched-dep-count-heuristic @gol
378 -fschedule-insns -fschedule-insns2 -fsection-anchors @gol
379 -fselective-scheduling -fselective-scheduling2 @gol
380 -fsel-sched-pipelining -fsel-sched-pipelining-outer-loops @gol
381 -fsignaling-nans -fsingle-precision-constant -fsplit-ivs-in-unroller @gol
382 -fsplit-wide-types -fstack-protector -fstack-protector-all @gol
383 -fstrict-aliasing -fstrict-overflow -fthread-jumps -ftracer @gol
384 -ftree-builtin-call-dce -ftree-ccp -ftree-ch -ftree-copy-prop @gol
385 -ftree-copyrename -ftree-dce -ftree-dominator-opts -ftree-dse @gol
386 -ftree-forwprop -ftree-fre -ftree-loop-if-convert -ftree-loop-im @gol
387 -ftree-phiprop -ftree-loop-distribution @gol
388 -ftree-loop-ivcanon -ftree-loop-linear -ftree-loop-optimize @gol
389 -ftree-parallelize-loops=@var{n} -ftree-pre -ftree-pta -ftree-reassoc @gol
390 -ftree-sink -ftree-sra -ftree-switch-conversion @gol
391 -ftree-ter -ftree-vect-loop-version -ftree-vectorize -ftree-vrp @gol
392 -funit-at-a-time -funroll-all-loops -funroll-loops @gol
393 -funsafe-loop-optimizations -funsafe-math-optimizations -funswitch-loops @gol
394 -fvariable-expansion-in-unroller -fvect-cost-model -fvpt -fweb @gol
395 -fwhole-program -fwhopr[=@var{n}] -fwpa -fuse-linker-plugin @gol
396 --param @var{name}=@var{value}
397 -O -O0 -O1 -O2 -O3 -Os -Ofast}
399 @item Preprocessor Options
400 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
401 @gccoptlist{-A@var{question}=@var{answer} @gol
402 -A-@var{question}@r{[}=@var{answer}@r{]} @gol
403 -C -dD -dI -dM -dN @gol
404 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
405 -idirafter @var{dir} @gol
406 -include @var{file} -imacros @var{file} @gol
407 -iprefix @var{file} -iwithprefix @var{dir} @gol
408 -iwithprefixbefore @var{dir} -isystem @var{dir} @gol
409 -imultilib @var{dir} -isysroot @var{dir} @gol
410 -M -MM -MF -MG -MP -MQ -MT -nostdinc @gol
411 -P -fworking-directory -remap @gol
412 -trigraphs -undef -U@var{macro} -Wp,@var{option} @gol
413 -Xpreprocessor @var{option}}
415 @item Assembler Option
416 @xref{Assembler Options,,Passing Options to the Assembler}.
417 @gccoptlist{-Wa,@var{option} -Xassembler @var{option}}
420 @xref{Link Options,,Options for Linking}.
421 @gccoptlist{@var{object-file-name} -l@var{library} @gol
422 -nostartfiles -nodefaultlibs -nostdlib -pie -rdynamic @gol
423 -s -static -static-libgcc -static-libstdc++ -shared @gol
424 -shared-libgcc -symbolic @gol
425 -T @var{script} -Wl,@var{option} -Xlinker @var{option} @gol
428 @item Directory Options
429 @xref{Directory Options,,Options for Directory Search}.
430 @gccoptlist{-B@var{prefix} -I@var{dir} -iplugindir=@var{dir}}
431 -iquote@var{dir} -L@var{dir} -specs=@var{file} -I-
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} -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{-mglibc -muclibc -mbionic -mandroid @gol
564 -tno-android-cc -tno-android-ld}
566 @emph{H8/300 Options}
567 @gccoptlist{-mrelax -mh -ms -mn -mint32 -malign-300}
570 @gccoptlist{-march=@var{architecture-type} @gol
571 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
572 -mfast-indirect-calls -mgas -mgnu-ld -mhp-ld @gol
573 -mfixed-range=@var{register-range} @gol
574 -mjump-in-delay -mlinker-opt -mlong-calls @gol
575 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
576 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
577 -mno-jump-in-delay -mno-long-load-store @gol
578 -mno-portable-runtime -mno-soft-float @gol
579 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
580 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
581 -mschedule=@var{cpu-type} -mspace-regs -msio -mwsio @gol
582 -munix=@var{unix-std} -nolibdld -static -threads}
584 @emph{i386 and x86-64 Options}
585 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
586 -mfpmath=@var{unit} @gol
587 -masm=@var{dialect} -mno-fancy-math-387 @gol
588 -mno-fp-ret-in-387 -msoft-float @gol
589 -mno-wide-multiply -mrtd -malign-double @gol
590 -mpreferred-stack-boundary=@var{num}
591 -mincoming-stack-boundary=@var{num}
592 -mcld -mcx16 -msahf -mmovbe -mcrc32 -mrecip @gol
593 -mmmx -msse -msse2 -msse3 -mssse3 -msse4.1 -msse4.2 -msse4 -mavx @gol
594 -maes -mpclmul -mfsgsbase -mrdrnd -mf16c -mfused-madd @gol
595 -msse4a -m3dnow -mpopcnt -mabm -mfma4 -mxop -mlwp @gol
596 -mthreads -mno-align-stringops -minline-all-stringops @gol
597 -minline-stringops-dynamically -mstringop-strategy=@var{alg} @gol
598 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
599 -m96bit-long-double -mregparm=@var{num} -msseregparm @gol
600 -mveclibabi=@var{type} -mpc32 -mpc64 -mpc80 -mstackrealign @gol
601 -momit-leaf-frame-pointer -mno-red-zone -mno-tls-direct-seg-refs @gol
602 -mcmodel=@var{code-model} -mabi=@var{name} @gol
603 -m32 -m64 -mlarge-data-threshold=@var{num} @gol
607 @gccoptlist{-mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
608 -mvolatile-asm-stop -mregister-names -msdata -mno-sdata @gol
609 -mconstant-gp -mauto-pic -mfused-madd @gol
610 -minline-float-divide-min-latency @gol
611 -minline-float-divide-max-throughput @gol
612 -mno-inline-float-divide @gol
613 -minline-int-divide-min-latency @gol
614 -minline-int-divide-max-throughput @gol
615 -mno-inline-int-divide @gol
616 -minline-sqrt-min-latency -minline-sqrt-max-throughput @gol
617 -mno-inline-sqrt @gol
618 -mdwarf2-asm -mearly-stop-bits @gol
619 -mfixed-range=@var{register-range} -mtls-size=@var{tls-size} @gol
620 -mtune=@var{cpu-type} -milp32 -mlp64 @gol
621 -msched-br-data-spec -msched-ar-data-spec -msched-control-spec @gol
622 -msched-br-in-data-spec -msched-ar-in-data-spec -msched-in-control-spec @gol
623 -msched-spec-ldc -msched-spec-control-ldc @gol
624 -msched-prefer-non-data-spec-insns -msched-prefer-non-control-spec-insns @gol
625 -msched-stop-bits-after-every-cycle -msched-count-spec-in-critical-path @gol
626 -msel-sched-dont-check-control-spec -msched-fp-mem-deps-zero-cost @gol
627 -msched-max-memory-insns-hard-limit -msched-max-memory-insns=@var{max-insns}}
629 @emph{IA-64/VMS Options}
630 @gccoptlist{-mvms-return-codes -mdebug-main=@var{prefix} -mmalloc64}
633 @gccoptlist{-mbarrel-shift-enabled -mdivide-enabled -mmultiply-enabled @gol
634 -msign-extend-enabled -muser-enabled}
636 @emph{M32R/D Options}
637 @gccoptlist{-m32r2 -m32rx -m32r @gol
639 -malign-loops -mno-align-loops @gol
640 -missue-rate=@var{number} @gol
641 -mbranch-cost=@var{number} @gol
642 -mmodel=@var{code-size-model-type} @gol
643 -msdata=@var{sdata-type} @gol
644 -mno-flush-func -mflush-func=@var{name} @gol
645 -mno-flush-trap -mflush-trap=@var{number} @gol
649 @gccoptlist{-mcpu=@var{cpu} -msim -memregs=@var{number}}
651 @emph{M680x0 Options}
652 @gccoptlist{-march=@var{arch} -mcpu=@var{cpu} -mtune=@var{tune}
653 -m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
654 -m68060 -mcpu32 -m5200 -m5206e -m528x -m5307 -m5407 @gol
655 -mcfv4e -mbitfield -mno-bitfield -mc68000 -mc68020 @gol
656 -mnobitfield -mrtd -mno-rtd -mdiv -mno-div -mshort @gol
657 -mno-short -mhard-float -m68881 -msoft-float -mpcrel @gol
658 -malign-int -mstrict-align -msep-data -mno-sep-data @gol
659 -mshared-library-id=n -mid-shared-library -mno-id-shared-library @gol
662 @emph{M68hc1x Options}
663 @gccoptlist{-m6811 -m6812 -m68hc11 -m68hc12 -m68hcs12 @gol
664 -mauto-incdec -minmax -mlong-calls -mshort @gol
665 -msoft-reg-count=@var{count}}
668 @gccoptlist{-mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
669 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
670 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
671 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
672 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
675 @gccoptlist{-mabsdiff -mall-opts -maverage -mbased=@var{n} -mbitops @gol
676 -mc=@var{n} -mclip -mconfig=@var{name} -mcop -mcop32 -mcop64 -mivc2 @gol
677 -mdc -mdiv -meb -mel -mio-volatile -ml -mleadz -mm -mminmax @gol
678 -mmult -mno-opts -mrepeat -ms -msatur -msdram -msim -msimnovec -mtf @gol
682 @gccoptlist{-EL -EB -march=@var{arch} -mtune=@var{arch} @gol
683 -mips1 -mips2 -mips3 -mips4 -mips32 -mips32r2 @gol
684 -mips64 -mips64r2 @gol
685 -mips16 -mno-mips16 -mflip-mips16 @gol
686 -minterlink-mips16 -mno-interlink-mips16 @gol
687 -mabi=@var{abi} -mabicalls -mno-abicalls @gol
688 -mshared -mno-shared -mplt -mno-plt -mxgot -mno-xgot @gol
689 -mgp32 -mgp64 -mfp32 -mfp64 -mhard-float -msoft-float @gol
690 -msingle-float -mdouble-float -mdsp -mno-dsp -mdspr2 -mno-dspr2 @gol
691 -mfpu=@var{fpu-type} @gol
692 -msmartmips -mno-smartmips @gol
693 -mpaired-single -mno-paired-single -mdmx -mno-mdmx @gol
694 -mips3d -mno-mips3d -mmt -mno-mt -mllsc -mno-llsc @gol
695 -mlong64 -mlong32 -msym32 -mno-sym32 @gol
696 -G@var{num} -mlocal-sdata -mno-local-sdata @gol
697 -mextern-sdata -mno-extern-sdata -mgpopt -mno-gopt @gol
698 -membedded-data -mno-embedded-data @gol
699 -muninit-const-in-rodata -mno-uninit-const-in-rodata @gol
700 -mcode-readable=@var{setting} @gol
701 -msplit-addresses -mno-split-addresses @gol
702 -mexplicit-relocs -mno-explicit-relocs @gol
703 -mcheck-zero-division -mno-check-zero-division @gol
704 -mdivide-traps -mdivide-breaks @gol
705 -mmemcpy -mno-memcpy -mlong-calls -mno-long-calls @gol
706 -mmad -mno-mad -mfused-madd -mno-fused-madd -nocpp @gol
707 -mfix-r4000 -mno-fix-r4000 -mfix-r4400 -mno-fix-r4400 @gol
708 -mfix-r10000 -mno-fix-r10000 -mfix-vr4120 -mno-fix-vr4120 @gol
709 -mfix-vr4130 -mno-fix-vr4130 -mfix-sb1 -mno-fix-sb1 @gol
710 -mflush-func=@var{func} -mno-flush-func @gol
711 -mbranch-cost=@var{num} -mbranch-likely -mno-branch-likely @gol
712 -mfp-exceptions -mno-fp-exceptions @gol
713 -mvr4130-align -mno-vr4130-align -msynci -mno-synci @gol
714 -mrelax-pic-calls -mno-relax-pic-calls -mmcount-ra-address}
717 @gccoptlist{-mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
718 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
719 -melf -mbranch-predict -mno-branch-predict -mbase-addresses @gol
720 -mno-base-addresses -msingle-exit -mno-single-exit}
722 @emph{MN10300 Options}
723 @gccoptlist{-mmult-bug -mno-mult-bug @gol
724 -mam33 -mno-am33 @gol
725 -mam33-2 -mno-am33-2 @gol
726 -mreturn-pointer-on-d0 @gol
729 @emph{PDP-11 Options}
730 @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
731 -mbcopy -mbcopy-builtin -mint32 -mno-int16 @gol
732 -mint16 -mno-int32 -mfloat32 -mno-float64 @gol
733 -mfloat64 -mno-float32 -mabshi -mno-abshi @gol
734 -mbranch-expensive -mbranch-cheap @gol
735 -msplit -mno-split -munix-asm -mdec-asm}
737 @emph{picoChip Options}
738 @gccoptlist{-mae=@var{ae_type} -mvliw-lookahead=@var{N}
739 -msymbol-as-address -mno-inefficient-warnings}
741 @emph{PowerPC Options}
742 See RS/6000 and PowerPC Options.
744 @emph{RS/6000 and PowerPC Options}
745 @gccoptlist{-mcpu=@var{cpu-type} @gol
746 -mtune=@var{cpu-type} @gol
747 -mcmodel=@var{code-model} @gol
748 -mpower -mno-power -mpower2 -mno-power2 @gol
749 -mpowerpc -mpowerpc64 -mno-powerpc @gol
750 -maltivec -mno-altivec @gol
751 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
752 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
753 -mmfcrf -mno-mfcrf -mpopcntb -mno-popcntb -mpopcntd -mno-popcntd @gol
754 -mfprnd -mno-fprnd @gol
755 -mcmpb -mno-cmpb -mmfpgpr -mno-mfpgpr -mhard-dfp -mno-hard-dfp @gol
756 -mnew-mnemonics -mold-mnemonics @gol
757 -mfull-toc -mminimal-toc -mno-fp-in-toc -mno-sum-in-toc @gol
758 -m64 -m32 -mxl-compat -mno-xl-compat -mpe @gol
759 -malign-power -malign-natural @gol
760 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
761 -msingle-float -mdouble-float -msimple-fpu @gol
762 -mstring -mno-string -mupdate -mno-update @gol
763 -mavoid-indexed-addresses -mno-avoid-indexed-addresses @gol
764 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
765 -mstrict-align -mno-strict-align -mrelocatable @gol
766 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
767 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
768 -mdynamic-no-pic -maltivec -mswdiv @gol
769 -mprioritize-restricted-insns=@var{priority} @gol
770 -msched-costly-dep=@var{dependence_type} @gol
771 -minsert-sched-nops=@var{scheme} @gol
772 -mcall-sysv -mcall-netbsd @gol
773 -maix-struct-return -msvr4-struct-return @gol
774 -mabi=@var{abi-type} -msecure-plt -mbss-plt @gol
775 -misel -mno-isel @gol
776 -misel=yes -misel=no @gol
778 -mspe=yes -mspe=no @gol
780 -mgen-cell-microcode -mwarn-cell-microcode @gol
781 -mvrsave -mno-vrsave @gol
782 -mmulhw -mno-mulhw @gol
783 -mdlmzb -mno-dlmzb @gol
784 -mfloat-gprs=yes -mfloat-gprs=no -mfloat-gprs=single -mfloat-gprs=double @gol
785 -mprototype -mno-prototype @gol
786 -msim -mmvme -mads -myellowknife -memb -msdata @gol
787 -msdata=@var{opt} -mvxworks -G @var{num} -pthread @gol
788 -mrecip -mrecip=@var{opt} -mno-recip -mrecip-precision -mno-recip-precision}
791 @gccoptlist{-m64bit-doubles -m32bit-doubles -fpu -nofpu@gol
793 -mbig-endian-data -mlittle-endian-data @gol
796 -mas100-syntax -mno-as100-syntax@gol
798 -mmax-constant-size=@gol
800 -msave-acc-in-interrupts}
802 @emph{S/390 and zSeries Options}
803 @gccoptlist{-mtune=@var{cpu-type} -march=@var{cpu-type} @gol
804 -mhard-float -msoft-float -mhard-dfp -mno-hard-dfp @gol
805 -mlong-double-64 -mlong-double-128 @gol
806 -mbackchain -mno-backchain -mpacked-stack -mno-packed-stack @gol
807 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
808 -m64 -m31 -mdebug -mno-debug -mesa -mzarch @gol
809 -mtpf-trace -mno-tpf-trace -mfused-madd -mno-fused-madd @gol
810 -mwarn-framesize -mwarn-dynamicstack -mstack-size -mstack-guard}
813 @gccoptlist{-meb -mel @gol
817 -mscore5 -mscore5u -mscore7 -mscore7d}
820 @gccoptlist{-m1 -m2 -m2e @gol
821 -m2a-nofpu -m2a-single-only -m2a-single -m2a @gol
823 -m4-nofpu -m4-single-only -m4-single -m4 @gol
824 -m4a-nofpu -m4a-single-only -m4a-single -m4a -m4al @gol
825 -m5-64media -m5-64media-nofpu @gol
826 -m5-32media -m5-32media-nofpu @gol
827 -m5-compact -m5-compact-nofpu @gol
828 -mb -ml -mdalign -mrelax @gol
829 -mbigtable -mfmovd -mhitachi -mrenesas -mno-renesas -mnomacsave @gol
830 -mieee -mbitops -misize -minline-ic_invalidate -mpadstruct -mspace @gol
831 -mprefergot -musermode -multcost=@var{number} -mdiv=@var{strategy} @gol
832 -mdivsi3_libfunc=@var{name} -mfixed-range=@var{register-range} @gol
833 -madjust-unroll -mindexed-addressing -mgettrcost=@var{number} -mpt-fixed @gol
834 -maccumulate-outgoing-args -minvalid-symbols}
837 @gccoptlist{-mcpu=@var{cpu-type} @gol
838 -mtune=@var{cpu-type} @gol
839 -mcmodel=@var{code-model} @gol
840 -m32 -m64 -mapp-regs -mno-app-regs @gol
841 -mfaster-structs -mno-faster-structs @gol
842 -mfpu -mno-fpu -mhard-float -msoft-float @gol
843 -mhard-quad-float -msoft-quad-float @gol
844 -mimpure-text -mno-impure-text -mlittle-endian @gol
845 -mstack-bias -mno-stack-bias @gol
846 -munaligned-doubles -mno-unaligned-doubles @gol
847 -mv8plus -mno-v8plus -mvis -mno-vis
848 -threads -pthreads -pthread}
851 @gccoptlist{-mwarn-reloc -merror-reloc @gol
852 -msafe-dma -munsafe-dma @gol
854 -msmall-mem -mlarge-mem -mstdmain @gol
855 -mfixed-range=@var{register-range} @gol
857 -maddress-space-conversion -mno-address-space-conversion @gol
858 -mcache-size=@var{cache-size} @gol
859 -matomic-updates -mno-atomic-updates}
861 @emph{System V Options}
862 @gccoptlist{-Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
865 @gccoptlist{-mlong-calls -mno-long-calls -mep -mno-ep @gol
866 -mprolog-function -mno-prolog-function -mspace @gol
867 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
868 -mapp-regs -mno-app-regs @gol
869 -mdisable-callt -mno-disable-callt @gol
875 @gccoptlist{-mg -mgnu -munix}
877 @emph{VxWorks Options}
878 @gccoptlist{-mrtp -non-static -Bstatic -Bdynamic @gol
879 -Xbind-lazy -Xbind-now}
881 @emph{x86-64 Options}
882 See i386 and x86-64 Options.
884 @emph{i386 and x86-64 Windows Options}
885 @gccoptlist{-mconsole -mcygwin -mno-cygwin -mdll
886 -mnop-fun-dllimport -mthread -municode -mwin32 -mwindows
887 -fno-set-stack-executable}
889 @emph{Xstormy16 Options}
892 @emph{Xtensa Options}
893 @gccoptlist{-mconst16 -mno-const16 @gol
894 -mfused-madd -mno-fused-madd @gol
896 -mserialize-volatile -mno-serialize-volatile @gol
897 -mtext-section-literals -mno-text-section-literals @gol
898 -mtarget-align -mno-target-align @gol
899 -mlongcalls -mno-longcalls}
901 @emph{zSeries Options}
902 See S/390 and zSeries Options.
904 @item Code Generation Options
905 @xref{Code Gen Options,,Options for Code Generation Conventions}.
906 @gccoptlist{-fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
907 -ffixed-@var{reg} -fexceptions @gol
908 -fnon-call-exceptions -funwind-tables @gol
909 -fasynchronous-unwind-tables @gol
910 -finhibit-size-directive -finstrument-functions @gol
911 -finstrument-functions-exclude-function-list=@var{sym},@var{sym},@dots{} @gol
912 -finstrument-functions-exclude-file-list=@var{file},@var{file},@dots{} @gol
913 -fno-common -fno-ident @gol
914 -fpcc-struct-return -fpic -fPIC -fpie -fPIE @gol
915 -fno-jump-tables @gol
916 -frecord-gcc-switches @gol
917 -freg-struct-return -fshort-enums @gol
918 -fshort-double -fshort-wchar @gol
919 -fverbose-asm -fpack-struct[=@var{n}] -fstack-check @gol
920 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
921 -fno-stack-limit @gol
922 -fleading-underscore -ftls-model=@var{model} @gol
923 -ftrapv -fwrapv -fbounds-check @gol
928 * Overall Options:: Controlling the kind of output:
929 an executable, object files, assembler files,
930 or preprocessed source.
931 * C Dialect Options:: Controlling the variant of C language compiled.
932 * C++ Dialect Options:: Variations on C++.
933 * Objective-C and Objective-C++ Dialect Options:: Variations on Objective-C
935 * Language Independent Options:: Controlling how diagnostics should be
937 * Warning Options:: How picky should the compiler be?
938 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
939 * Optimize Options:: How much optimization?
940 * Preprocessor Options:: Controlling header files and macro definitions.
941 Also, getting dependency information for Make.
942 * Assembler Options:: Passing options to the assembler.
943 * Link Options:: Specifying libraries and so on.
944 * Directory Options:: Where to find header files and libraries.
945 Where to find the compiler executable files.
946 * Spec Files:: How to pass switches to sub-processes.
947 * Target Options:: Running a cross-compiler, or an old version of GCC.
950 @node Overall Options
951 @section Options Controlling the Kind of Output
953 Compilation can involve up to four stages: preprocessing, compilation
954 proper, assembly and linking, always in that order. GCC is capable of
955 preprocessing and compiling several files either into several
956 assembler input files, or into one assembler input file; then each
957 assembler input file produces an object file, and linking combines all
958 the object files (those newly compiled, and those specified as input)
959 into an executable file.
961 @cindex file name suffix
962 For any given input file, the file name suffix determines what kind of
967 C source code which must be preprocessed.
970 C source code which should not be preprocessed.
973 C++ source code which should not be preprocessed.
976 Objective-C source code. Note that you must link with the @file{libobjc}
977 library to make an Objective-C program work.
980 Objective-C source code which should not be preprocessed.
984 Objective-C++ source code. Note that you must link with the @file{libobjc}
985 library to make an Objective-C++ program work. Note that @samp{.M} refers
986 to a literal capital M@.
989 Objective-C++ source code which should not be preprocessed.
992 C, C++, Objective-C or Objective-C++ header file to be turned into a
993 precompiled header (default), or C, C++ header file to be turned into an
994 Ada spec (via the @option{-fdump-ada-spec} switch).
998 @itemx @var{file}.cxx
999 @itemx @var{file}.cpp
1000 @itemx @var{file}.CPP
1001 @itemx @var{file}.c++
1003 C++ source code which must be preprocessed. Note that in @samp{.cxx},
1004 the last two letters must both be literally @samp{x}. Likewise,
1005 @samp{.C} refers to a literal capital C@.
1009 Objective-C++ source code which must be preprocessed.
1011 @item @var{file}.mii
1012 Objective-C++ source code which should not be preprocessed.
1016 @itemx @var{file}.hp
1017 @itemx @var{file}.hxx
1018 @itemx @var{file}.hpp
1019 @itemx @var{file}.HPP
1020 @itemx @var{file}.h++
1021 @itemx @var{file}.tcc
1022 C++ header file to be turned into a precompiled header or Ada spec.
1025 @itemx @var{file}.for
1026 @itemx @var{file}.ftn
1027 Fixed form Fortran source code which should not be preprocessed.
1030 @itemx @var{file}.FOR
1031 @itemx @var{file}.fpp
1032 @itemx @var{file}.FPP
1033 @itemx @var{file}.FTN
1034 Fixed form Fortran source code which must be preprocessed (with the traditional
1037 @item @var{file}.f90
1038 @itemx @var{file}.f95
1039 @itemx @var{file}.f03
1040 @itemx @var{file}.f08
1041 Free form Fortran source code which should not be preprocessed.
1043 @item @var{file}.F90
1044 @itemx @var{file}.F95
1045 @itemx @var{file}.F03
1046 @itemx @var{file}.F08
1047 Free form Fortran source code which must be preprocessed (with the
1048 traditional preprocessor).
1050 @c FIXME: Descriptions of Java file types.
1056 @item @var{file}.ads
1057 Ada source code file which contains a library unit declaration (a
1058 declaration of a package, subprogram, or generic, or a generic
1059 instantiation), or a library unit renaming declaration (a package,
1060 generic, or subprogram renaming declaration). Such files are also
1063 @item @var{file}.adb
1064 Ada source code file containing a library unit body (a subprogram or
1065 package body). Such files are also called @dfn{bodies}.
1067 @c GCC also knows about some suffixes for languages not yet included:
1078 @itemx @var{file}.sx
1079 Assembler code which must be preprocessed.
1082 An object file to be fed straight into linking.
1083 Any file name with no recognized suffix is treated this way.
1087 You can specify the input language explicitly with the @option{-x} option:
1090 @item -x @var{language}
1091 Specify explicitly the @var{language} for the following input files
1092 (rather than letting the compiler choose a default based on the file
1093 name suffix). This option applies to all following input files until
1094 the next @option{-x} option. Possible values for @var{language} are:
1096 c c-header c-cpp-output
1097 c++ c++-header c++-cpp-output
1098 objective-c objective-c-header objective-c-cpp-output
1099 objective-c++ objective-c++-header objective-c++-cpp-output
1100 assembler assembler-with-cpp
1102 f77 f77-cpp-input f95 f95-cpp-input
1107 Turn off any specification of a language, so that subsequent files are
1108 handled according to their file name suffixes (as they are if @option{-x}
1109 has not been used at all).
1111 @item -pass-exit-codes
1112 @opindex pass-exit-codes
1113 Normally the @command{gcc} program will exit with the code of 1 if any
1114 phase of the compiler returns a non-success return code. If you specify
1115 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
1116 numerically highest error produced by any phase that returned an error
1117 indication. The C, C++, and Fortran frontends return 4, if an internal
1118 compiler error is encountered.
1121 If you only want some of the stages of compilation, you can use
1122 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
1123 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
1124 @command{gcc} is to stop. Note that some combinations (for example,
1125 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
1130 Compile or assemble the source files, but do not link. The linking
1131 stage simply is not done. The ultimate output is in the form of an
1132 object file for each source file.
1134 By default, the object file name for a source file is made by replacing
1135 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
1137 Unrecognized input files, not requiring compilation or assembly, are
1142 Stop after the stage of compilation proper; do not assemble. The output
1143 is in the form of an assembler code file for each non-assembler input
1146 By default, the assembler file name for a source file is made by
1147 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
1149 Input files that don't require compilation are ignored.
1153 Stop after the preprocessing stage; do not run the compiler proper. The
1154 output is in the form of preprocessed source code, which is sent to the
1157 Input files which don't require preprocessing are ignored.
1159 @cindex output file option
1162 Place output in file @var{file}. This applies regardless to whatever
1163 sort of output is being produced, whether it be an executable file,
1164 an object file, an assembler file or preprocessed C code.
1166 If @option{-o} is not specified, the default is to put an executable
1167 file in @file{a.out}, the object file for
1168 @file{@var{source}.@var{suffix}} in @file{@var{source}.o}, its
1169 assembler file in @file{@var{source}.s}, a precompiled header file in
1170 @file{@var{source}.@var{suffix}.gch}, and all preprocessed C source on
1175 Print (on standard error output) the commands executed to run the stages
1176 of compilation. Also print the version number of the compiler driver
1177 program and of the preprocessor and the compiler proper.
1181 Like @option{-v} except the commands are not executed and arguments
1182 are quoted unless they contain only alphanumeric characters or @code{./-_}.
1183 This is useful for shell scripts to capture the driver-generated command lines.
1187 Use pipes rather than temporary files for communication between the
1188 various stages of compilation. This fails to work on some systems where
1189 the assembler is unable to read from a pipe; but the GNU assembler has
1194 If you are compiling multiple source files, this option tells the driver
1195 to pass all the source files to the compiler at once (for those
1196 languages for which the compiler can handle this). This will allow
1197 intermodule analysis (IMA) to be performed by the compiler. Currently the only
1198 language for which this is supported is C@. If you pass source files for
1199 multiple languages to the driver, using this option, the driver will invoke
1200 the compiler(s) that support IMA once each, passing each compiler all the
1201 source files appropriate for it. For those languages that do not support
1202 IMA this option will be ignored, and the compiler will be invoked once for
1203 each source file in that language. If you use this option in conjunction
1204 with @option{-save-temps}, the compiler will generate multiple
1206 (one for each source file), but only one (combined) @file{.o} or
1211 Print (on the standard output) a description of the command line options
1212 understood by @command{gcc}. If the @option{-v} option is also specified
1213 then @option{--help} will also be passed on to the various processes
1214 invoked by @command{gcc}, so that they can display the command line options
1215 they accept. If the @option{-Wextra} option has also been specified
1216 (prior to the @option{--help} option), then command line options which
1217 have no documentation associated with them will also be displayed.
1220 @opindex target-help
1221 Print (on the standard output) a description of target-specific command
1222 line options for each tool. For some targets extra target-specific
1223 information may also be printed.
1225 @item --help=@{@var{class}@r{|[}^@r{]}@var{qualifier}@}@r{[},@dots{}@r{]}
1226 Print (on the standard output) a description of the command line
1227 options understood by the compiler that fit into all specified classes
1228 and qualifiers. These are the supported classes:
1231 @item @samp{optimizers}
1232 This will display all of the optimization options supported by the
1235 @item @samp{warnings}
1236 This will display all of the options controlling warning messages
1237 produced by the compiler.
1240 This will display target-specific options. Unlike the
1241 @option{--target-help} option however, target-specific options of the
1242 linker and assembler will not be displayed. This is because those
1243 tools do not currently support the extended @option{--help=} syntax.
1246 This will display the values recognized by the @option{--param}
1249 @item @var{language}
1250 This will display the options supported for @var{language}, where
1251 @var{language} is the name of one of the languages supported in this
1255 This will display the options that are common to all languages.
1258 These are the supported qualifiers:
1261 @item @samp{undocumented}
1262 Display only those options which are undocumented.
1265 Display options which take an argument that appears after an equal
1266 sign in the same continuous piece of text, such as:
1267 @samp{--help=target}.
1269 @item @samp{separate}
1270 Display options which take an argument that appears as a separate word
1271 following the original option, such as: @samp{-o output-file}.
1274 Thus for example to display all the undocumented target-specific
1275 switches supported by the compiler the following can be used:
1278 --help=target,undocumented
1281 The sense of a qualifier can be inverted by prefixing it with the
1282 @samp{^} character, so for example to display all binary warning
1283 options (i.e., ones that are either on or off and that do not take an
1284 argument), which have a description the following can be used:
1287 --help=warnings,^joined,^undocumented
1290 The argument to @option{--help=} should not consist solely of inverted
1293 Combining several classes is possible, although this usually
1294 restricts the output by so much that there is nothing to display. One
1295 case where it does work however is when one of the classes is
1296 @var{target}. So for example to display all the target-specific
1297 optimization options the following can be used:
1300 --help=target,optimizers
1303 The @option{--help=} option can be repeated on the command line. Each
1304 successive use will display its requested class of options, skipping
1305 those that have already been displayed.
1307 If the @option{-Q} option appears on the command line before the
1308 @option{--help=} option, then the descriptive text displayed by
1309 @option{--help=} is changed. Instead of describing the displayed
1310 options, an indication is given as to whether the option is enabled,
1311 disabled or set to a specific value (assuming that the compiler
1312 knows this at the point where the @option{--help=} option is used).
1314 Here is a truncated example from the ARM port of @command{gcc}:
1317 % gcc -Q -mabi=2 --help=target -c
1318 The following options are target specific:
1320 -mabort-on-noreturn [disabled]
1324 The output is sensitive to the effects of previous command line
1325 options, so for example it is possible to find out which optimizations
1326 are enabled at @option{-O2} by using:
1329 -Q -O2 --help=optimizers
1332 Alternatively you can discover which binary optimizations are enabled
1333 by @option{-O3} by using:
1336 gcc -c -Q -O3 --help=optimizers > /tmp/O3-opts
1337 gcc -c -Q -O2 --help=optimizers > /tmp/O2-opts
1338 diff /tmp/O2-opts /tmp/O3-opts | grep enabled
1341 @item -no-canonical-prefixes
1342 @opindex no-canonical-prefixes
1343 Do not expand any symbolic links, resolve references to @samp{/../}
1344 or @samp{/./}, or make the path absolute when generating a relative
1349 Display the version number and copyrights of the invoked GCC@.
1353 Invoke all subcommands under a wrapper program. It takes a single
1354 comma separated list as an argument, which will be used to invoke
1358 gcc -c t.c -wrapper gdb,--args
1361 This will invoke all subprograms of gcc under "gdb --args",
1362 thus cc1 invocation will be "gdb --args cc1 ...".
1364 @item -fplugin=@var{name}.so
1365 Load the plugin code in file @var{name}.so, assumed to be a
1366 shared object to be dlopen'd by the compiler. The base name of
1367 the shared object file is used to identify the plugin for the
1368 purposes of argument parsing (See
1369 @option{-fplugin-arg-@var{name}-@var{key}=@var{value}} below).
1370 Each plugin should define the callback functions specified in the
1373 @item -fplugin-arg-@var{name}-@var{key}=@var{value}
1374 Define an argument called @var{key} with a value of @var{value}
1375 for the plugin called @var{name}.
1377 @item -fdump-ada-spec@r{[}-slim@r{]}
1378 For C and C++ source and include files, generate corresponding Ada
1379 specs. @xref{Generating Ada Bindings for C and C++ headers,,, gnat_ugn,
1380 GNAT User's Guide}, which provides detailed documentation on this feature.
1382 @include @value{srcdir}/../libiberty/at-file.texi
1386 @section Compiling C++ Programs
1388 @cindex suffixes for C++ source
1389 @cindex C++ source file suffixes
1390 C++ source files conventionally use one of the suffixes @samp{.C},
1391 @samp{.cc}, @samp{.cpp}, @samp{.CPP}, @samp{.c++}, @samp{.cp}, or
1392 @samp{.cxx}; C++ header files often use @samp{.hh}, @samp{.hpp},
1393 @samp{.H}, or (for shared template code) @samp{.tcc}; and
1394 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
1395 files with these names and compiles them as C++ programs even if you
1396 call the compiler the same way as for compiling C programs (usually
1397 with the name @command{gcc}).
1401 However, the use of @command{gcc} does not add the C++ library.
1402 @command{g++} is a program that calls GCC and treats @samp{.c},
1403 @samp{.h} and @samp{.i} files as C++ source files instead of C source
1404 files unless @option{-x} is used, and automatically specifies linking
1405 against the C++ library. This program is also useful when
1406 precompiling a C header file with a @samp{.h} extension for use in C++
1407 compilations. On many systems, @command{g++} is also installed with
1408 the name @command{c++}.
1410 @cindex invoking @command{g++}
1411 When you compile C++ programs, you may specify many of the same
1412 command-line options that you use for compiling programs in any
1413 language; or command-line options meaningful for C and related
1414 languages; or options that are meaningful only for C++ programs.
1415 @xref{C Dialect Options,,Options Controlling C Dialect}, for
1416 explanations of options for languages related to C@.
1417 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
1418 explanations of options that are meaningful only for C++ programs.
1420 @node C Dialect Options
1421 @section Options Controlling C Dialect
1422 @cindex dialect options
1423 @cindex language dialect options
1424 @cindex options, dialect
1426 The following options control the dialect of C (or languages derived
1427 from C, such as C++, Objective-C and Objective-C++) that the compiler
1431 @cindex ANSI support
1435 In C mode, this is equivalent to @samp{-std=c90}. In C++ mode, it is
1436 equivalent to @samp{-std=c++98}.
1438 This turns off certain features of GCC that are incompatible with ISO
1439 C90 (when compiling C code), or of standard C++ (when compiling C++ code),
1440 such as the @code{asm} and @code{typeof} keywords, and
1441 predefined macros such as @code{unix} and @code{vax} that identify the
1442 type of system you are using. It also enables the undesirable and
1443 rarely used ISO trigraph feature. For the C compiler,
1444 it disables recognition of C++ style @samp{//} comments as well as
1445 the @code{inline} keyword.
1447 The alternate keywords @code{__asm__}, @code{__extension__},
1448 @code{__inline__} and @code{__typeof__} continue to work despite
1449 @option{-ansi}. You would not want to use them in an ISO C program, of
1450 course, but it is useful to put them in header files that might be included
1451 in compilations done with @option{-ansi}. Alternate predefined macros
1452 such as @code{__unix__} and @code{__vax__} are also available, with or
1453 without @option{-ansi}.
1455 The @option{-ansi} option does not cause non-ISO programs to be
1456 rejected gratuitously. For that, @option{-pedantic} is required in
1457 addition to @option{-ansi}. @xref{Warning Options}.
1459 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
1460 option is used. Some header files may notice this macro and refrain
1461 from declaring certain functions or defining certain macros that the
1462 ISO standard doesn't call for; this is to avoid interfering with any
1463 programs that might use these names for other things.
1465 Functions that would normally be built in but do not have semantics
1466 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
1467 functions when @option{-ansi} is used. @xref{Other Builtins,,Other
1468 built-in functions provided by GCC}, for details of the functions
1473 Determine the language standard. @xref{Standards,,Language Standards
1474 Supported by GCC}, for details of these standard versions. This option
1475 is currently only supported when compiling C or C++.
1477 The compiler can accept several base standards, such as @samp{c90} or
1478 @samp{c++98}, and GNU dialects of those standards, such as
1479 @samp{gnu90} or @samp{gnu++98}. By specifying a base standard, the
1480 compiler will accept all programs following that standard and those
1481 using GNU extensions that do not contradict it. For example,
1482 @samp{-std=c90} turns off certain features of GCC that are
1483 incompatible with ISO C90, such as the @code{asm} and @code{typeof}
1484 keywords, but not other GNU extensions that do not have a meaning in
1485 ISO C90, such as omitting the middle term of a @code{?:}
1486 expression. On the other hand, by specifying a GNU dialect of a
1487 standard, all features the compiler support are enabled, even when
1488 those features change the meaning of the base standard and some
1489 strict-conforming programs may be rejected. The particular standard
1490 is used by @option{-pedantic} to identify which features are GNU
1491 extensions given that version of the standard. For example
1492 @samp{-std=gnu90 -pedantic} would warn about C++ style @samp{//}
1493 comments, while @samp{-std=gnu99 -pedantic} would not.
1495 A value for this option must be provided; possible values are
1501 Support all ISO C90 programs (certain GNU extensions that conflict
1502 with ISO C90 are disabled). Same as @option{-ansi} for C code.
1504 @item iso9899:199409
1505 ISO C90 as modified in amendment 1.
1511 ISO C99. Note that this standard is not yet fully supported; see
1512 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
1513 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
1516 ISO C1X, the draft of the next revision of the ISO C standard.
1517 Support is limited and experimental and features enabled by this
1518 option may be changed or removed if changed in or removed from the
1523 GNU dialect of ISO C90 (including some C99 features). This
1524 is the default for C code.
1528 GNU dialect of ISO C99. When ISO C99 is fully implemented in GCC,
1529 this will become the default. The name @samp{gnu9x} is deprecated.
1532 GNU dialect of ISO C1X. Support is limited and experimental and
1533 features enabled by this option may be changed or removed if changed
1534 in or removed from the standard draft.
1537 The 1998 ISO C++ standard plus amendments. Same as @option{-ansi} for
1541 GNU dialect of @option{-std=c++98}. This is the default for
1545 The working draft of the upcoming ISO C++0x standard. This option
1546 enables experimental features that are likely to be included in
1547 C++0x. The working draft is constantly changing, and any feature that is
1548 enabled by this flag may be removed from future versions of GCC if it is
1549 not part of the C++0x standard.
1552 GNU dialect of @option{-std=c++0x}. This option enables
1553 experimental features that may be removed in future versions of GCC.
1556 @item -fgnu89-inline
1557 @opindex fgnu89-inline
1558 The option @option{-fgnu89-inline} tells GCC to use the traditional
1559 GNU semantics for @code{inline} functions when in C99 mode.
1560 @xref{Inline,,An Inline Function is As Fast As a Macro}. This option
1561 is accepted and ignored by GCC versions 4.1.3 up to but not including
1562 4.3. In GCC versions 4.3 and later it changes the behavior of GCC in
1563 C99 mode. Using this option is roughly equivalent to adding the
1564 @code{gnu_inline} function attribute to all inline functions
1565 (@pxref{Function Attributes}).
1567 The option @option{-fno-gnu89-inline} explicitly tells GCC to use the
1568 C99 semantics for @code{inline} when in C99 or gnu99 mode (i.e., it
1569 specifies the default behavior). This option was first supported in
1570 GCC 4.3. This option is not supported in @option{-std=c90} or
1571 @option{-std=gnu90} mode.
1573 The preprocessor macros @code{__GNUC_GNU_INLINE__} and
1574 @code{__GNUC_STDC_INLINE__} may be used to check which semantics are
1575 in effect for @code{inline} functions. @xref{Common Predefined
1576 Macros,,,cpp,The C Preprocessor}.
1578 @item -aux-info @var{filename}
1580 Output to the given filename prototyped declarations for all functions
1581 declared and/or defined in a translation unit, including those in header
1582 files. This option is silently ignored in any language other than C@.
1584 Besides declarations, the file indicates, in comments, the origin of
1585 each declaration (source file and line), whether the declaration was
1586 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1587 @samp{O} for old, respectively, in the first character after the line
1588 number and the colon), and whether it came from a declaration or a
1589 definition (@samp{C} or @samp{F}, respectively, in the following
1590 character). In the case of function definitions, a K&R-style list of
1591 arguments followed by their declarations is also provided, inside
1592 comments, after the declaration.
1596 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1597 keyword, so that code can use these words as identifiers. You can use
1598 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1599 instead. @option{-ansi} implies @option{-fno-asm}.
1601 In C++, this switch only affects the @code{typeof} keyword, since
1602 @code{asm} and @code{inline} are standard keywords. You may want to
1603 use the @option{-fno-gnu-keywords} flag instead, which has the same
1604 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1605 switch only affects the @code{asm} and @code{typeof} keywords, since
1606 @code{inline} is a standard keyword in ISO C99.
1609 @itemx -fno-builtin-@var{function}
1610 @opindex fno-builtin
1611 @cindex built-in functions
1612 Don't recognize built-in functions that do not begin with
1613 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1614 functions provided by GCC}, for details of the functions affected,
1615 including those which are not built-in functions when @option{-ansi} or
1616 @option{-std} options for strict ISO C conformance are used because they
1617 do not have an ISO standard meaning.
1619 GCC normally generates special code to handle certain built-in functions
1620 more efficiently; for instance, calls to @code{alloca} may become single
1621 instructions that adjust the stack directly, and calls to @code{memcpy}
1622 may become inline copy loops. The resulting code is often both smaller
1623 and faster, but since the function calls no longer appear as such, you
1624 cannot set a breakpoint on those calls, nor can you change the behavior
1625 of the functions by linking with a different library. In addition,
1626 when a function is recognized as a built-in function, GCC may use
1627 information about that function to warn about problems with calls to
1628 that function, or to generate more efficient code, even if the
1629 resulting code still contains calls to that function. For example,
1630 warnings are given with @option{-Wformat} for bad calls to
1631 @code{printf}, when @code{printf} is built in, and @code{strlen} is
1632 known not to modify global memory.
1634 With the @option{-fno-builtin-@var{function}} option
1635 only the built-in function @var{function} is
1636 disabled. @var{function} must not begin with @samp{__builtin_}. If a
1637 function is named that is not built-in in this version of GCC, this
1638 option is ignored. There is no corresponding
1639 @option{-fbuiltin-@var{function}} option; if you wish to enable
1640 built-in functions selectively when using @option{-fno-builtin} or
1641 @option{-ffreestanding}, you may define macros such as:
1644 #define abs(n) __builtin_abs ((n))
1645 #define strcpy(d, s) __builtin_strcpy ((d), (s))
1650 @cindex hosted environment
1652 Assert that compilation takes place in a hosted environment. This implies
1653 @option{-fbuiltin}. A hosted environment is one in which the
1654 entire standard library is available, and in which @code{main} has a return
1655 type of @code{int}. Examples are nearly everything except a kernel.
1656 This is equivalent to @option{-fno-freestanding}.
1658 @item -ffreestanding
1659 @opindex ffreestanding
1660 @cindex hosted environment
1662 Assert that compilation takes place in a freestanding environment. This
1663 implies @option{-fno-builtin}. A freestanding environment
1664 is one in which the standard library may not exist, and program startup may
1665 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1666 This is equivalent to @option{-fno-hosted}.
1668 @xref{Standards,,Language Standards Supported by GCC}, for details of
1669 freestanding and hosted environments.
1673 @cindex openmp parallel
1674 Enable handling of OpenMP directives @code{#pragma omp} in C/C++ and
1675 @code{!$omp} in Fortran. When @option{-fopenmp} is specified, the
1676 compiler generates parallel code according to the OpenMP Application
1677 Program Interface v3.0 @w{@uref{http://www.openmp.org/}}. This option
1678 implies @option{-pthread}, and thus is only supported on targets that
1679 have support for @option{-pthread}.
1681 @item -fms-extensions
1682 @opindex fms-extensions
1683 Accept some non-standard constructs used in Microsoft header files.
1685 Some cases of unnamed fields in structures and unions are only
1686 accepted with this option. @xref{Unnamed Fields,,Unnamed struct/union
1687 fields within structs/unions}, for details.
1691 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1692 options for strict ISO C conformance) implies @option{-trigraphs}.
1694 @item -no-integrated-cpp
1695 @opindex no-integrated-cpp
1696 Performs a compilation in two passes: preprocessing and compiling. This
1697 option allows a user supplied "cc1", "cc1plus", or "cc1obj" via the
1698 @option{-B} option. The user supplied compilation step can then add in
1699 an additional preprocessing step after normal preprocessing but before
1700 compiling. The default is to use the integrated cpp (internal cpp)
1702 The semantics of this option will change if "cc1", "cc1plus", and
1703 "cc1obj" are merged.
1705 @cindex traditional C language
1706 @cindex C language, traditional
1708 @itemx -traditional-cpp
1709 @opindex traditional-cpp
1710 @opindex traditional
1711 Formerly, these options caused GCC to attempt to emulate a pre-standard
1712 C compiler. They are now only supported with the @option{-E} switch.
1713 The preprocessor continues to support a pre-standard mode. See the GNU
1714 CPP manual for details.
1716 @item -fcond-mismatch
1717 @opindex fcond-mismatch
1718 Allow conditional expressions with mismatched types in the second and
1719 third arguments. The value of such an expression is void. This option
1720 is not supported for C++.
1722 @item -flax-vector-conversions
1723 @opindex flax-vector-conversions
1724 Allow implicit conversions between vectors with differing numbers of
1725 elements and/or incompatible element types. This option should not be
1728 @item -funsigned-char
1729 @opindex funsigned-char
1730 Let the type @code{char} be unsigned, like @code{unsigned char}.
1732 Each kind of machine has a default for what @code{char} should
1733 be. It is either like @code{unsigned char} by default or like
1734 @code{signed char} by default.
1736 Ideally, a portable program should always use @code{signed char} or
1737 @code{unsigned char} when it depends on the signedness of an object.
1738 But many programs have been written to use plain @code{char} and
1739 expect it to be signed, or expect it to be unsigned, depending on the
1740 machines they were written for. This option, and its inverse, let you
1741 make such a program work with the opposite default.
1743 The type @code{char} is always a distinct type from each of
1744 @code{signed char} or @code{unsigned char}, even though its behavior
1745 is always just like one of those two.
1748 @opindex fsigned-char
1749 Let the type @code{char} be signed, like @code{signed char}.
1751 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1752 the negative form of @option{-funsigned-char}. Likewise, the option
1753 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1755 @item -fsigned-bitfields
1756 @itemx -funsigned-bitfields
1757 @itemx -fno-signed-bitfields
1758 @itemx -fno-unsigned-bitfields
1759 @opindex fsigned-bitfields
1760 @opindex funsigned-bitfields
1761 @opindex fno-signed-bitfields
1762 @opindex fno-unsigned-bitfields
1763 These options control whether a bit-field is signed or unsigned, when the
1764 declaration does not use either @code{signed} or @code{unsigned}. By
1765 default, such a bit-field is signed, because this is consistent: the
1766 basic integer types such as @code{int} are signed types.
1769 @node C++ Dialect Options
1770 @section Options Controlling C++ Dialect
1772 @cindex compiler options, C++
1773 @cindex C++ options, command line
1774 @cindex options, C++
1775 This section describes the command-line options that are only meaningful
1776 for C++ programs; but you can also use most of the GNU compiler options
1777 regardless of what language your program is in. For example, you
1778 might compile a file @code{firstClass.C} like this:
1781 g++ -g -frepo -O -c firstClass.C
1785 In this example, only @option{-frepo} is an option meant
1786 only for C++ programs; you can use the other options with any
1787 language supported by GCC@.
1789 Here is a list of options that are @emph{only} for compiling C++ programs:
1793 @item -fabi-version=@var{n}
1794 @opindex fabi-version
1795 Use version @var{n} of the C++ ABI@. Version 2 is the version of the
1796 C++ ABI that first appeared in G++ 3.4. Version 1 is the version of
1797 the C++ ABI that first appeared in G++ 3.2. Version 0 will always be
1798 the version that conforms most closely to the C++ ABI specification.
1799 Therefore, the ABI obtained using version 0 will change as ABI bugs
1802 The default is version 2.
1804 Version 3 corrects an error in mangling a constant address as a
1807 Version 4 implements a standard mangling for vector types.
1809 See also @option{-Wabi}.
1811 @item -fno-access-control
1812 @opindex fno-access-control
1813 Turn off all access checking. This switch is mainly useful for working
1814 around bugs in the access control code.
1818 Check that the pointer returned by @code{operator new} is non-null
1819 before attempting to modify the storage allocated. This check is
1820 normally unnecessary because the C++ standard specifies that
1821 @code{operator new} will only return @code{0} if it is declared
1822 @samp{throw()}, in which case the compiler will always check the
1823 return value even without this option. In all other cases, when
1824 @code{operator new} has a non-empty exception specification, memory
1825 exhaustion is signalled by throwing @code{std::bad_alloc}. See also
1826 @samp{new (nothrow)}.
1828 @item -fconserve-space
1829 @opindex fconserve-space
1830 Put uninitialized or runtime-initialized global variables into the
1831 common segment, as C does. This saves space in the executable at the
1832 cost of not diagnosing duplicate definitions. If you compile with this
1833 flag and your program mysteriously crashes after @code{main()} has
1834 completed, you may have an object that is being destroyed twice because
1835 two definitions were merged.
1837 This option is no longer useful on most targets, now that support has
1838 been added for putting variables into BSS without making them common.
1840 @item -fno-deduce-init-list
1841 @opindex fno-deduce-init-list
1842 Disable deduction of a template type parameter as
1843 std::initializer_list from a brace-enclosed initializer list, i.e.
1846 template <class T> auto forward(T t) -> decltype (realfn (t))
1853 forward(@{1,2@}); // call forward<std::initializer_list<int>>
1857 This option is present because this deduction is an extension to the
1858 current specification in the C++0x working draft, and there was
1859 some concern about potential overload resolution problems.
1861 @item -ffriend-injection
1862 @opindex ffriend-injection
1863 Inject friend functions into the enclosing namespace, so that they are
1864 visible outside the scope of the class in which they are declared.
1865 Friend functions were documented to work this way in the old Annotated
1866 C++ Reference Manual, and versions of G++ before 4.1 always worked
1867 that way. However, in ISO C++ a friend function which is not declared
1868 in an enclosing scope can only be found using argument dependent
1869 lookup. This option causes friends to be injected as they were in
1872 This option is for compatibility, and may be removed in a future
1875 @item -fno-elide-constructors
1876 @opindex fno-elide-constructors
1877 The C++ standard allows an implementation to omit creating a temporary
1878 which is only used to initialize another object of the same type.
1879 Specifying this option disables that optimization, and forces G++ to
1880 call the copy constructor in all cases.
1882 @item -fno-enforce-eh-specs
1883 @opindex fno-enforce-eh-specs
1884 Don't generate code to check for violation of exception specifications
1885 at runtime. This option violates the C++ standard, but may be useful
1886 for reducing code size in production builds, much like defining
1887 @samp{NDEBUG}. This does not give user code permission to throw
1888 exceptions in violation of the exception specifications; the compiler
1889 will still optimize based on the specifications, so throwing an
1890 unexpected exception will result in undefined behavior.
1893 @itemx -fno-for-scope
1895 @opindex fno-for-scope
1896 If @option{-ffor-scope} is specified, the scope of variables declared in
1897 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1898 as specified by the C++ standard.
1899 If @option{-fno-for-scope} is specified, the scope of variables declared in
1900 a @i{for-init-statement} extends to the end of the enclosing scope,
1901 as was the case in old versions of G++, and other (traditional)
1902 implementations of C++.
1904 The default if neither flag is given to follow the standard,
1905 but to allow and give a warning for old-style code that would
1906 otherwise be invalid, or have different behavior.
1908 @item -fno-gnu-keywords
1909 @opindex fno-gnu-keywords
1910 Do not recognize @code{typeof} as a keyword, so that code can use this
1911 word as an identifier. You can use the keyword @code{__typeof__} instead.
1912 @option{-ansi} implies @option{-fno-gnu-keywords}.
1914 @item -fno-implicit-templates
1915 @opindex fno-implicit-templates
1916 Never emit code for non-inline templates which are instantiated
1917 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1918 @xref{Template Instantiation}, for more information.
1920 @item -fno-implicit-inline-templates
1921 @opindex fno-implicit-inline-templates
1922 Don't emit code for implicit instantiations of inline templates, either.
1923 The default is to handle inlines differently so that compiles with and
1924 without optimization will need the same set of explicit instantiations.
1926 @item -fno-implement-inlines
1927 @opindex fno-implement-inlines
1928 To save space, do not emit out-of-line copies of inline functions
1929 controlled by @samp{#pragma implementation}. This will cause linker
1930 errors if these functions are not inlined everywhere they are called.
1932 @item -fms-extensions
1933 @opindex fms-extensions
1934 Disable pedantic warnings about constructs used in MFC, such as implicit
1935 int and getting a pointer to member function via non-standard syntax.
1937 @item -fno-nonansi-builtins
1938 @opindex fno-nonansi-builtins
1939 Disable built-in declarations of functions that are not mandated by
1940 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1941 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1944 @opindex fnothrow-opt
1945 Treat a @code{throw()} exception specification as though it were a
1946 @code{noexcept} specification to reduce or eliminate the text size
1947 overhead relative to a function with no exception specification. If
1948 the function has local variables of types with non-trivial
1949 destructors, the exception specification will actually make the
1950 function smaller because the EH cleanups for those variables can be
1951 optimized away. The semantic effect is that an exception thrown out of
1952 a function with such an exception specification will result in a call
1953 to @code{terminate} rather than @code{unexpected}.
1955 @item -fno-operator-names
1956 @opindex fno-operator-names
1957 Do not treat the operator name keywords @code{and}, @code{bitand},
1958 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1959 synonyms as keywords.
1961 @item -fno-optional-diags
1962 @opindex fno-optional-diags
1963 Disable diagnostics that the standard says a compiler does not need to
1964 issue. Currently, the only such diagnostic issued by G++ is the one for
1965 a name having multiple meanings within a class.
1968 @opindex fpermissive
1969 Downgrade some diagnostics about nonconformant code from errors to
1970 warnings. Thus, using @option{-fpermissive} will allow some
1971 nonconforming code to compile.
1973 @item -fno-pretty-templates
1974 @opindex fno-pretty-templates
1975 When an error message refers to a specialization of a function
1976 template, the compiler will normally print the signature of the
1977 template followed by the template arguments and any typedefs or
1978 typenames in the signature (e.g. @code{void f(T) [with T = int]}
1979 rather than @code{void f(int)}) so that it's clear which template is
1980 involved. When an error message refers to a specialization of a class
1981 template, the compiler will omit any template arguments which match
1982 the default template arguments for that template. If either of these
1983 behaviors make it harder to understand the error message rather than
1984 easier, using @option{-fno-pretty-templates} will disable them.
1988 Enable automatic template instantiation at link time. This option also
1989 implies @option{-fno-implicit-templates}. @xref{Template
1990 Instantiation}, for more information.
1994 Disable generation of information about every class with virtual
1995 functions for use by the C++ runtime type identification features
1996 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1997 of the language, you can save some space by using this flag. Note that
1998 exception handling uses the same information, but it will generate it as
1999 needed. The @samp{dynamic_cast} operator can still be used for casts that
2000 do not require runtime type information, i.e.@: casts to @code{void *} or to
2001 unambiguous base classes.
2005 Emit statistics about front-end processing at the end of the compilation.
2006 This information is generally only useful to the G++ development team.
2008 @item -fstrict-enums
2009 @opindex fstrict-enums
2010 Allow the compiler to optimize using the assumption that a value of
2011 enumeration type can only be one of the values of the enumeration (as
2012 defined in the C++ standard; basically, a value which can be
2013 represented in the minimum number of bits needed to represent all the
2014 enumerators). This assumption may not be valid if the program uses a
2015 cast to convert an arbitrary integer value to the enumeration type.
2017 @item -ftemplate-depth=@var{n}
2018 @opindex ftemplate-depth
2019 Set the maximum instantiation depth for template classes to @var{n}.
2020 A limit on the template instantiation depth is needed to detect
2021 endless recursions during template class instantiation. ANSI/ISO C++
2022 conforming programs must not rely on a maximum depth greater than 17
2023 (changed to 1024 in C++0x).
2025 @item -fno-threadsafe-statics
2026 @opindex fno-threadsafe-statics
2027 Do not emit the extra code to use the routines specified in the C++
2028 ABI for thread-safe initialization of local statics. You can use this
2029 option to reduce code size slightly in code that doesn't need to be
2032 @item -fuse-cxa-atexit
2033 @opindex fuse-cxa-atexit
2034 Register destructors for objects with static storage duration with the
2035 @code{__cxa_atexit} function rather than the @code{atexit} function.
2036 This option is required for fully standards-compliant handling of static
2037 destructors, but will only work if your C library supports
2038 @code{__cxa_atexit}.
2040 @item -fno-use-cxa-get-exception-ptr
2041 @opindex fno-use-cxa-get-exception-ptr
2042 Don't use the @code{__cxa_get_exception_ptr} runtime routine. This
2043 will cause @code{std::uncaught_exception} to be incorrect, but is necessary
2044 if the runtime routine is not available.
2046 @item -fvisibility-inlines-hidden
2047 @opindex fvisibility-inlines-hidden
2048 This switch declares that the user does not attempt to compare
2049 pointers to inline methods where the addresses of the two functions
2050 were taken in different shared objects.
2052 The effect of this is that GCC may, effectively, mark inline methods with
2053 @code{__attribute__ ((visibility ("hidden")))} so that they do not
2054 appear in the export table of a DSO and do not require a PLT indirection
2055 when used within the DSO@. Enabling this option can have a dramatic effect
2056 on load and link times of a DSO as it massively reduces the size of the
2057 dynamic export table when the library makes heavy use of templates.
2059 The behavior of this switch is not quite the same as marking the
2060 methods as hidden directly, because it does not affect static variables
2061 local to the function or cause the compiler to deduce that
2062 the function is defined in only one shared object.
2064 You may mark a method as having a visibility explicitly to negate the
2065 effect of the switch for that method. For example, if you do want to
2066 compare pointers to a particular inline method, you might mark it as
2067 having default visibility. Marking the enclosing class with explicit
2068 visibility will have no effect.
2070 Explicitly instantiated inline methods are unaffected by this option
2071 as their linkage might otherwise cross a shared library boundary.
2072 @xref{Template Instantiation}.
2074 @item -fvisibility-ms-compat
2075 @opindex fvisibility-ms-compat
2076 This flag attempts to use visibility settings to make GCC's C++
2077 linkage model compatible with that of Microsoft Visual Studio.
2079 The flag makes these changes to GCC's linkage model:
2083 It sets the default visibility to @code{hidden}, like
2084 @option{-fvisibility=hidden}.
2087 Types, but not their members, are not hidden by default.
2090 The One Definition Rule is relaxed for types without explicit
2091 visibility specifications which are defined in more than one different
2092 shared object: those declarations are permitted if they would have
2093 been permitted when this option was not used.
2096 In new code it is better to use @option{-fvisibility=hidden} and
2097 export those classes which are intended to be externally visible.
2098 Unfortunately it is possible for code to rely, perhaps accidentally,
2099 on the Visual Studio behavior.
2101 Among the consequences of these changes are that static data members
2102 of the same type with the same name but defined in different shared
2103 objects will be different, so changing one will not change the other;
2104 and that pointers to function members defined in different shared
2105 objects may not compare equal. When this flag is given, it is a
2106 violation of the ODR to define types with the same name differently.
2110 Do not use weak symbol support, even if it is provided by the linker.
2111 By default, G++ will use weak symbols if they are available. This
2112 option exists only for testing, and should not be used by end-users;
2113 it will result in inferior code and has no benefits. This option may
2114 be removed in a future release of G++.
2118 Do not search for header files in the standard directories specific to
2119 C++, but do still search the other standard directories. (This option
2120 is used when building the C++ library.)
2123 In addition, these optimization, warning, and code generation options
2124 have meanings only for C++ programs:
2127 @item -fno-default-inline
2128 @opindex fno-default-inline
2129 Do not assume @samp{inline} for functions defined inside a class scope.
2130 @xref{Optimize Options,,Options That Control Optimization}. Note that these
2131 functions will have linkage like inline functions; they just won't be
2134 @item -Wabi @r{(C, Objective-C, C++ and Objective-C++ only)}
2137 Warn when G++ generates code that is probably not compatible with the
2138 vendor-neutral C++ ABI@. Although an effort has been made to warn about
2139 all such cases, there are probably some cases that are not warned about,
2140 even though G++ is generating incompatible code. There may also be
2141 cases where warnings are emitted even though the code that is generated
2144 You should rewrite your code to avoid these warnings if you are
2145 concerned about the fact that code generated by G++ may not be binary
2146 compatible with code generated by other compilers.
2148 The known incompatibilities in @option{-fabi-version=2} (the default) include:
2153 A template with a non-type template parameter of reference type is
2154 mangled incorrectly:
2157 template <int &> struct S @{@};
2161 This is fixed in @option{-fabi-version=3}.
2164 SIMD vector types declared using @code{__attribute ((vector_size))} are
2165 mangled in a non-standard way that does not allow for overloading of
2166 functions taking vectors of different sizes.
2168 The mangling is changed in @option{-fabi-version=4}.
2171 The known incompatibilities in @option{-fabi-version=1} include:
2176 Incorrect handling of tail-padding for bit-fields. G++ may attempt to
2177 pack data into the same byte as a base class. For example:
2180 struct A @{ virtual void f(); int f1 : 1; @};
2181 struct B : public A @{ int f2 : 1; @};
2185 In this case, G++ will place @code{B::f2} into the same byte
2186 as@code{A::f1}; other compilers will not. You can avoid this problem
2187 by explicitly padding @code{A} so that its size is a multiple of the
2188 byte size on your platform; that will cause G++ and other compilers to
2189 layout @code{B} identically.
2192 Incorrect handling of tail-padding for virtual bases. G++ does not use
2193 tail padding when laying out virtual bases. For example:
2196 struct A @{ virtual void f(); char c1; @};
2197 struct B @{ B(); char c2; @};
2198 struct C : public A, public virtual B @{@};
2202 In this case, G++ will not place @code{B} into the tail-padding for
2203 @code{A}; other compilers will. You can avoid this problem by
2204 explicitly padding @code{A} so that its size is a multiple of its
2205 alignment (ignoring virtual base classes); that will cause G++ and other
2206 compilers to layout @code{C} identically.
2209 Incorrect handling of bit-fields with declared widths greater than that
2210 of their underlying types, when the bit-fields appear in a union. For
2214 union U @{ int i : 4096; @};
2218 Assuming that an @code{int} does not have 4096 bits, G++ will make the
2219 union too small by the number of bits in an @code{int}.
2222 Empty classes can be placed at incorrect offsets. For example:
2232 struct C : public B, public A @{@};
2236 G++ will place the @code{A} base class of @code{C} at a nonzero offset;
2237 it should be placed at offset zero. G++ mistakenly believes that the
2238 @code{A} data member of @code{B} is already at offset zero.
2241 Names of template functions whose types involve @code{typename} or
2242 template template parameters can be mangled incorrectly.
2245 template <typename Q>
2246 void f(typename Q::X) @{@}
2248 template <template <typename> class Q>
2249 void f(typename Q<int>::X) @{@}
2253 Instantiations of these templates may be mangled incorrectly.
2257 It also warns psABI related changes. The known psABI changes at this
2263 For SYSV/x86-64, when passing union with long double, it is changed to
2264 pass in memory as specified in psABI. For example:
2274 @code{union U} will always be passed in memory.
2278 @item -Wctor-dtor-privacy @r{(C++ and Objective-C++ only)}
2279 @opindex Wctor-dtor-privacy
2280 @opindex Wno-ctor-dtor-privacy
2281 Warn when a class seems unusable because all the constructors or
2282 destructors in that class are private, and it has neither friends nor
2283 public static member functions.
2285 @item -Wnoexcept @r{(C++ and Objective-C++ only)}
2287 @opindex Wno-noexcept
2288 Warn when a noexcept-expression evaluates to false because of a call
2289 to a function that does not have a non-throwing exception
2290 specification (i.e. @samp{throw()} or @samp{noexcept}) but is known by
2291 the compiler to never throw an exception.
2293 @item -Wnon-virtual-dtor @r{(C++ and Objective-C++ only)}
2294 @opindex Wnon-virtual-dtor
2295 @opindex Wno-non-virtual-dtor
2296 Warn when a class has virtual functions and accessible non-virtual
2297 destructor, in which case it would be possible but unsafe to delete
2298 an instance of a derived class through a pointer to the base class.
2299 This warning is also enabled if -Weffc++ is specified.
2301 @item -Wreorder @r{(C++ and Objective-C++ only)}
2303 @opindex Wno-reorder
2304 @cindex reordering, warning
2305 @cindex warning for reordering of member initializers
2306 Warn when the order of member initializers given in the code does not
2307 match the order in which they must be executed. For instance:
2313 A(): j (0), i (1) @{ @}
2317 The compiler will rearrange the member initializers for @samp{i}
2318 and @samp{j} to match the declaration order of the members, emitting
2319 a warning to that effect. This warning is enabled by @option{-Wall}.
2322 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
2325 @item -Weffc++ @r{(C++ and Objective-C++ only)}
2328 Warn about violations of the following style guidelines from Scott Meyers'
2329 @cite{Effective C++} book:
2333 Item 11: Define a copy constructor and an assignment operator for classes
2334 with dynamically allocated memory.
2337 Item 12: Prefer initialization to assignment in constructors.
2340 Item 14: Make destructors virtual in base classes.
2343 Item 15: Have @code{operator=} return a reference to @code{*this}.
2346 Item 23: Don't try to return a reference when you must return an object.
2350 Also warn about violations of the following style guidelines from
2351 Scott Meyers' @cite{More Effective C++} book:
2355 Item 6: Distinguish between prefix and postfix forms of increment and
2356 decrement operators.
2359 Item 7: Never overload @code{&&}, @code{||}, or @code{,}.
2363 When selecting this option, be aware that the standard library
2364 headers do not obey all of these guidelines; use @samp{grep -v}
2365 to filter out those warnings.
2367 @item -Wstrict-null-sentinel @r{(C++ and Objective-C++ only)}
2368 @opindex Wstrict-null-sentinel
2369 @opindex Wno-strict-null-sentinel
2370 Warn also about the use of an uncasted @code{NULL} as sentinel. When
2371 compiling only with GCC this is a valid sentinel, as @code{NULL} is defined
2372 to @code{__null}. Although it is a null pointer constant not a null pointer,
2373 it is guaranteed to be of the same size as a pointer. But this use is
2374 not portable across different compilers.
2376 @item -Wno-non-template-friend @r{(C++ and Objective-C++ only)}
2377 @opindex Wno-non-template-friend
2378 @opindex Wnon-template-friend
2379 Disable warnings when non-templatized friend functions are declared
2380 within a template. Since the advent of explicit template specification
2381 support in G++, if the name of the friend is an unqualified-id (i.e.,
2382 @samp{friend foo(int)}), the C++ language specification demands that the
2383 friend declare or define an ordinary, nontemplate function. (Section
2384 14.5.3). Before G++ implemented explicit specification, unqualified-ids
2385 could be interpreted as a particular specialization of a templatized
2386 function. Because this non-conforming behavior is no longer the default
2387 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
2388 check existing code for potential trouble spots and is on by default.
2389 This new compiler behavior can be turned off with
2390 @option{-Wno-non-template-friend} which keeps the conformant compiler code
2391 but disables the helpful warning.
2393 @item -Wold-style-cast @r{(C++ and Objective-C++ only)}
2394 @opindex Wold-style-cast
2395 @opindex Wno-old-style-cast
2396 Warn if an old-style (C-style) cast to a non-void type is used within
2397 a C++ program. The new-style casts (@samp{dynamic_cast},
2398 @samp{static_cast}, @samp{reinterpret_cast}, and @samp{const_cast}) are
2399 less vulnerable to unintended effects and much easier to search for.
2401 @item -Woverloaded-virtual @r{(C++ and Objective-C++ only)}
2402 @opindex Woverloaded-virtual
2403 @opindex Wno-overloaded-virtual
2404 @cindex overloaded virtual fn, warning
2405 @cindex warning for overloaded virtual fn
2406 Warn when a function declaration hides virtual functions from a
2407 base class. For example, in:
2414 struct B: public A @{
2419 the @code{A} class version of @code{f} is hidden in @code{B}, and code
2427 will fail to compile.
2429 @item -Wno-pmf-conversions @r{(C++ and Objective-C++ only)}
2430 @opindex Wno-pmf-conversions
2431 @opindex Wpmf-conversions
2432 Disable the diagnostic for converting a bound pointer to member function
2435 @item -Wsign-promo @r{(C++ and Objective-C++ only)}
2436 @opindex Wsign-promo
2437 @opindex Wno-sign-promo
2438 Warn when overload resolution chooses a promotion from unsigned or
2439 enumerated type to a signed type, over a conversion to an unsigned type of
2440 the same size. Previous versions of G++ would try to preserve
2441 unsignedness, but the standard mandates the current behavior.
2446 A& operator = (int);
2456 In this example, G++ will synthesize a default @samp{A& operator =
2457 (const A&);}, while cfront will use the user-defined @samp{operator =}.
2460 @node Objective-C and Objective-C++ Dialect Options
2461 @section Options Controlling Objective-C and Objective-C++ Dialects
2463 @cindex compiler options, Objective-C and Objective-C++
2464 @cindex Objective-C and Objective-C++ options, command line
2465 @cindex options, Objective-C and Objective-C++
2466 (NOTE: This manual does not describe the Objective-C and Objective-C++
2467 languages themselves. See @xref{Standards,,Language Standards
2468 Supported by GCC}, for references.)
2470 This section describes the command-line options that are only meaningful
2471 for Objective-C and Objective-C++ programs, but you can also use most of
2472 the language-independent GNU compiler options.
2473 For example, you might compile a file @code{some_class.m} like this:
2476 gcc -g -fgnu-runtime -O -c some_class.m
2480 In this example, @option{-fgnu-runtime} is an option meant only for
2481 Objective-C and Objective-C++ programs; you can use the other options with
2482 any language supported by GCC@.
2484 Note that since Objective-C is an extension of the C language, Objective-C
2485 compilations may also use options specific to the C front-end (e.g.,
2486 @option{-Wtraditional}). Similarly, Objective-C++ compilations may use
2487 C++-specific options (e.g., @option{-Wabi}).
2489 Here is a list of options that are @emph{only} for compiling Objective-C
2490 and Objective-C++ programs:
2493 @item -fconstant-string-class=@var{class-name}
2494 @opindex fconstant-string-class
2495 Use @var{class-name} as the name of the class to instantiate for each
2496 literal string specified with the syntax @code{@@"@dots{}"}. The default
2497 class name is @code{NXConstantString} if the GNU runtime is being used, and
2498 @code{NSConstantString} if the NeXT runtime is being used (see below). The
2499 @option{-fconstant-cfstrings} option, if also present, will override the
2500 @option{-fconstant-string-class} setting and cause @code{@@"@dots{}"} literals
2501 to be laid out as constant CoreFoundation strings.
2504 @opindex fgnu-runtime
2505 Generate object code compatible with the standard GNU Objective-C
2506 runtime. This is the default for most types of systems.
2508 @item -fnext-runtime
2509 @opindex fnext-runtime
2510 Generate output compatible with the NeXT runtime. This is the default
2511 for NeXT-based systems, including Darwin and Mac OS X@. The macro
2512 @code{__NEXT_RUNTIME__} is predefined if (and only if) this option is
2515 @item -fno-nil-receivers
2516 @opindex fno-nil-receivers
2517 Assume that all Objective-C message dispatches (e.g.,
2518 @code{[receiver message:arg]}) in this translation unit ensure that the receiver
2519 is not @code{nil}. This allows for more efficient entry points in the runtime
2520 to be used. Currently, this option is only available in conjunction with
2521 the NeXT runtime on Mac OS X 10.3 and later.
2523 @item -fobjc-call-cxx-cdtors
2524 @opindex fobjc-call-cxx-cdtors
2525 For each Objective-C class, check if any of its instance variables is a
2526 C++ object with a non-trivial default constructor. If so, synthesize a
2527 special @code{- (id) .cxx_construct} instance method that will run
2528 non-trivial default constructors on any such instance variables, in order,
2529 and then return @code{self}. Similarly, check if any instance variable
2530 is a C++ object with a non-trivial destructor, and if so, synthesize a
2531 special @code{- (void) .cxx_destruct} method that will run
2532 all such default destructors, in reverse order.
2534 The @code{- (id) .cxx_construct} and/or @code{- (void) .cxx_destruct} methods
2535 thusly generated will only operate on instance variables declared in the
2536 current Objective-C class, and not those inherited from superclasses. It
2537 is the responsibility of the Objective-C runtime to invoke all such methods
2538 in an object's inheritance hierarchy. The @code{- (id) .cxx_construct} methods
2539 will be invoked by the runtime immediately after a new object
2540 instance is allocated; the @code{- (void) .cxx_destruct} methods will
2541 be invoked immediately before the runtime deallocates an object instance.
2543 As of this writing, only the NeXT runtime on Mac OS X 10.4 and later has
2544 support for invoking the @code{- (id) .cxx_construct} and
2545 @code{- (void) .cxx_destruct} methods.
2547 @item -fobjc-direct-dispatch
2548 @opindex fobjc-direct-dispatch
2549 Allow fast jumps to the message dispatcher. On Darwin this is
2550 accomplished via the comm page.
2552 @item -fobjc-exceptions
2553 @opindex fobjc-exceptions
2554 Enable syntactic support for structured exception handling in Objective-C,
2555 similar to what is offered by C++ and Java. This option is
2556 unavailable in conjunction with the NeXT runtime on Mac OS X 10.2 and
2565 @@catch (AnObjCClass *exc) @{
2572 @@catch (AnotherClass *exc) @{
2575 @@catch (id allOthers) @{
2585 The @code{@@throw} statement may appear anywhere in an Objective-C or
2586 Objective-C++ program; when used inside of a @code{@@catch} block, the
2587 @code{@@throw} may appear without an argument (as shown above), in which case
2588 the object caught by the @code{@@catch} will be rethrown.
2590 Note that only (pointers to) Objective-C objects may be thrown and
2591 caught using this scheme. When an object is thrown, it will be caught
2592 by the nearest @code{@@catch} clause capable of handling objects of that type,
2593 analogously to how @code{catch} blocks work in C++ and Java. A
2594 @code{@@catch(id @dots{})} clause (as shown above) may also be provided to catch
2595 any and all Objective-C exceptions not caught by previous @code{@@catch}
2598 The @code{@@finally} clause, if present, will be executed upon exit from the
2599 immediately preceding @code{@@try @dots{} @@catch} section. This will happen
2600 regardless of whether any exceptions are thrown, caught or rethrown
2601 inside the @code{@@try @dots{} @@catch} section, analogously to the behavior
2602 of the @code{finally} clause in Java.
2604 There are several caveats to using the new exception mechanism:
2608 Although currently designed to be binary compatible with @code{NS_HANDLER}-style
2609 idioms provided by the @code{NSException} class, the new
2610 exceptions can only be used on Mac OS X 10.3 (Panther) and later
2611 systems, due to additional functionality needed in the (NeXT) Objective-C
2615 As mentioned above, the new exceptions do not support handling
2616 types other than Objective-C objects. Furthermore, when used from
2617 Objective-C++, the Objective-C exception model does not interoperate with C++
2618 exceptions at this time. This means you cannot @code{@@throw} an exception
2619 from Objective-C and @code{catch} it in C++, or vice versa
2620 (i.e., @code{throw @dots{} @@catch}).
2623 The @option{-fobjc-exceptions} switch also enables the use of synchronization
2624 blocks for thread-safe execution:
2627 @@synchronized (ObjCClass *guard) @{
2632 Upon entering the @code{@@synchronized} block, a thread of execution shall
2633 first check whether a lock has been placed on the corresponding @code{guard}
2634 object by another thread. If it has, the current thread shall wait until
2635 the other thread relinquishes its lock. Once @code{guard} becomes available,
2636 the current thread will place its own lock on it, execute the code contained in
2637 the @code{@@synchronized} block, and finally relinquish the lock (thereby
2638 making @code{guard} available to other threads).
2640 Unlike Java, Objective-C does not allow for entire methods to be marked
2641 @code{@@synchronized}. Note that throwing exceptions out of
2642 @code{@@synchronized} blocks is allowed, and will cause the guarding object
2643 to be unlocked properly.
2647 Enable garbage collection (GC) in Objective-C and Objective-C++ programs.
2649 @item -freplace-objc-classes
2650 @opindex freplace-objc-classes
2651 Emit a special marker instructing @command{ld(1)} not to statically link in
2652 the resulting object file, and allow @command{dyld(1)} to load it in at
2653 run time instead. This is used in conjunction with the Fix-and-Continue
2654 debugging mode, where the object file in question may be recompiled and
2655 dynamically reloaded in the course of program execution, without the need
2656 to restart the program itself. Currently, Fix-and-Continue functionality
2657 is only available in conjunction with the NeXT runtime on Mac OS X 10.3
2662 When compiling for the NeXT runtime, the compiler ordinarily replaces calls
2663 to @code{objc_getClass("@dots{}")} (when the name of the class is known at
2664 compile time) with static class references that get initialized at load time,
2665 which improves run-time performance. Specifying the @option{-fzero-link} flag
2666 suppresses this behavior and causes calls to @code{objc_getClass("@dots{}")}
2667 to be retained. This is useful in Zero-Link debugging mode, since it allows
2668 for individual class implementations to be modified during program execution.
2672 Dump interface declarations for all classes seen in the source file to a
2673 file named @file{@var{sourcename}.decl}.
2675 @item -Wassign-intercept @r{(Objective-C and Objective-C++ only)}
2676 @opindex Wassign-intercept
2677 @opindex Wno-assign-intercept
2678 Warn whenever an Objective-C assignment is being intercepted by the
2681 @item -Wno-protocol @r{(Objective-C and Objective-C++ only)}
2682 @opindex Wno-protocol
2684 If a class is declared to implement a protocol, a warning is issued for
2685 every method in the protocol that is not implemented by the class. The
2686 default behavior is to issue a warning for every method not explicitly
2687 implemented in the class, even if a method implementation is inherited
2688 from the superclass. If you use the @option{-Wno-protocol} option, then
2689 methods inherited from the superclass are considered to be implemented,
2690 and no warning is issued for them.
2692 @item -Wselector @r{(Objective-C and Objective-C++ only)}
2694 @opindex Wno-selector
2695 Warn if multiple methods of different types for the same selector are
2696 found during compilation. The check is performed on the list of methods
2697 in the final stage of compilation. Additionally, a check is performed
2698 for each selector appearing in a @code{@@selector(@dots{})}
2699 expression, and a corresponding method for that selector has been found
2700 during compilation. Because these checks scan the method table only at
2701 the end of compilation, these warnings are not produced if the final
2702 stage of compilation is not reached, for example because an error is
2703 found during compilation, or because the @option{-fsyntax-only} option is
2706 @item -Wstrict-selector-match @r{(Objective-C and Objective-C++ only)}
2707 @opindex Wstrict-selector-match
2708 @opindex Wno-strict-selector-match
2709 Warn if multiple methods with differing argument and/or return types are
2710 found for a given selector when attempting to send a message using this
2711 selector to a receiver of type @code{id} or @code{Class}. When this flag
2712 is off (which is the default behavior), the compiler will omit such warnings
2713 if any differences found are confined to types which share the same size
2716 @item -Wundeclared-selector @r{(Objective-C and Objective-C++ only)}
2717 @opindex Wundeclared-selector
2718 @opindex Wno-undeclared-selector
2719 Warn if a @code{@@selector(@dots{})} expression referring to an
2720 undeclared selector is found. A selector is considered undeclared if no
2721 method with that name has been declared before the
2722 @code{@@selector(@dots{})} expression, either explicitly in an
2723 @code{@@interface} or @code{@@protocol} declaration, or implicitly in
2724 an @code{@@implementation} section. This option always performs its
2725 checks as soon as a @code{@@selector(@dots{})} expression is found,
2726 while @option{-Wselector} only performs its checks in the final stage of
2727 compilation. This also enforces the coding style convention
2728 that methods and selectors must be declared before being used.
2730 @item -print-objc-runtime-info
2731 @opindex print-objc-runtime-info
2732 Generate C header describing the largest structure that is passed by
2737 @node Language Independent Options
2738 @section Options to Control Diagnostic Messages Formatting
2739 @cindex options to control diagnostics formatting
2740 @cindex diagnostic messages
2741 @cindex message formatting
2743 Traditionally, diagnostic messages have been formatted irrespective of
2744 the output device's aspect (e.g.@: its width, @dots{}). The options described
2745 below can be used to control the diagnostic messages formatting
2746 algorithm, e.g.@: how many characters per line, how often source location
2747 information should be reported. Right now, only the C++ front end can
2748 honor these options. However it is expected, in the near future, that
2749 the remaining front ends would be able to digest them correctly.
2752 @item -fmessage-length=@var{n}
2753 @opindex fmessage-length
2754 Try to format error messages so that they fit on lines of about @var{n}
2755 characters. The default is 72 characters for @command{g++} and 0 for the rest of
2756 the front ends supported by GCC@. If @var{n} is zero, then no
2757 line-wrapping will be done; each error message will appear on a single
2760 @opindex fdiagnostics-show-location
2761 @item -fdiagnostics-show-location=once
2762 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
2763 reporter to emit @emph{once} source location information; that is, in
2764 case the message is too long to fit on a single physical line and has to
2765 be wrapped, the source location won't be emitted (as prefix) again,
2766 over and over, in subsequent continuation lines. This is the default
2769 @item -fdiagnostics-show-location=every-line
2770 Only meaningful in line-wrapping mode. Instructs the diagnostic
2771 messages reporter to emit the same source location information (as
2772 prefix) for physical lines that result from the process of breaking
2773 a message which is too long to fit on a single line.
2775 @item -fdiagnostics-show-option
2776 @opindex fdiagnostics-show-option
2777 This option instructs the diagnostic machinery to add text to each
2778 diagnostic emitted, which indicates which command line option directly
2779 controls that diagnostic, when such an option is known to the
2780 diagnostic machinery.
2782 @item -Wcoverage-mismatch
2783 @opindex Wcoverage-mismatch
2784 Warn if feedback profiles do not match when using the
2785 @option{-fprofile-use} option.
2786 If a source file was changed between @option{-fprofile-gen} and
2787 @option{-fprofile-use}, the files with the profile feedback can fail
2788 to match the source file and GCC can not use the profile feedback
2789 information. By default, this warning is enabled and is treated as an
2790 error. @option{-Wno-coverage-mismatch} can be used to disable the
2791 warning or @option{-Wno-error=coverage-mismatch} can be used to
2792 disable the error. Disable the error for this warning can result in
2793 poorly optimized code, so disabling the error is useful only in the
2794 case of very minor changes such as bug fixes to an existing code-base.
2795 Completely disabling the warning is not recommended.
2799 @node Warning Options
2800 @section Options to Request or Suppress Warnings
2801 @cindex options to control warnings
2802 @cindex warning messages
2803 @cindex messages, warning
2804 @cindex suppressing warnings
2806 Warnings are diagnostic messages that report constructions which
2807 are not inherently erroneous but which are risky or suggest there
2808 may have been an error.
2810 The following language-independent options do not enable specific
2811 warnings but control the kinds of diagnostics produced by GCC.
2814 @cindex syntax checking
2816 @opindex fsyntax-only
2817 Check the code for syntax errors, but don't do anything beyond that.
2821 Inhibit all warning messages.
2826 Make all warnings into errors.
2831 Make the specified warning into an error. The specifier for a warning
2832 is appended, for example @option{-Werror=switch} turns the warnings
2833 controlled by @option{-Wswitch} into errors. This switch takes a
2834 negative form, to be used to negate @option{-Werror} for specific
2835 warnings, for example @option{-Wno-error=switch} makes
2836 @option{-Wswitch} warnings not be errors, even when @option{-Werror}
2837 is in effect. You can use the @option{-fdiagnostics-show-option}
2838 option to have each controllable warning amended with the option which
2839 controls it, to determine what to use with this option.
2841 Note that specifying @option{-Werror=}@var{foo} automatically implies
2842 @option{-W}@var{foo}. However, @option{-Wno-error=}@var{foo} does not
2845 @item -Wfatal-errors
2846 @opindex Wfatal-errors
2847 @opindex Wno-fatal-errors
2848 This option causes the compiler to abort compilation on the first error
2849 occurred rather than trying to keep going and printing further error
2854 You can request many specific warnings with options beginning
2855 @samp{-W}, for example @option{-Wimplicit} to request warnings on
2856 implicit declarations. Each of these specific warning options also
2857 has a negative form beginning @samp{-Wno-} to turn off warnings; for
2858 example, @option{-Wno-implicit}. This manual lists only one of the
2859 two forms, whichever is not the default. For further,
2860 language-specific options also refer to @ref{C++ Dialect Options} and
2861 @ref{Objective-C and Objective-C++ Dialect Options}.
2863 When an unrecognized warning label is requested (e.g.,
2864 @option{-Wunknown-warning}), GCC will emit a diagnostic stating
2865 that the option is not recognized. However, if the @samp{-Wno-} form
2866 is used, the behavior is slightly different: No diagnostic will be
2867 produced for @option{-Wno-unknown-warning} unless other diagnostics
2868 are being produced. This allows the use of new @option{-Wno-} options
2869 with old compilers, but if something goes wrong, the compiler will
2870 warn that an unrecognized option was used.
2875 Issue all the warnings demanded by strict ISO C and ISO C++;
2876 reject all programs that use forbidden extensions, and some other
2877 programs that do not follow ISO C and ISO C++. For ISO C, follows the
2878 version of the ISO C standard specified by any @option{-std} option used.
2880 Valid ISO C and ISO C++ programs should compile properly with or without
2881 this option (though a rare few will require @option{-ansi} or a
2882 @option{-std} option specifying the required version of ISO C)@. However,
2883 without this option, certain GNU extensions and traditional C and C++
2884 features are supported as well. With this option, they are rejected.
2886 @option{-pedantic} does not cause warning messages for use of the
2887 alternate keywords whose names begin and end with @samp{__}. Pedantic
2888 warnings are also disabled in the expression that follows
2889 @code{__extension__}. However, only system header files should use
2890 these escape routes; application programs should avoid them.
2891 @xref{Alternate Keywords}.
2893 Some users try to use @option{-pedantic} to check programs for strict ISO
2894 C conformance. They soon find that it does not do quite what they want:
2895 it finds some non-ISO practices, but not all---only those for which
2896 ISO C @emph{requires} a diagnostic, and some others for which
2897 diagnostics have been added.
2899 A feature to report any failure to conform to ISO C might be useful in
2900 some instances, but would require considerable additional work and would
2901 be quite different from @option{-pedantic}. We don't have plans to
2902 support such a feature in the near future.
2904 Where the standard specified with @option{-std} represents a GNU
2905 extended dialect of C, such as @samp{gnu90} or @samp{gnu99}, there is a
2906 corresponding @dfn{base standard}, the version of ISO C on which the GNU
2907 extended dialect is based. Warnings from @option{-pedantic} are given
2908 where they are required by the base standard. (It would not make sense
2909 for such warnings to be given only for features not in the specified GNU
2910 C dialect, since by definition the GNU dialects of C include all
2911 features the compiler supports with the given option, and there would be
2912 nothing to warn about.)
2914 @item -pedantic-errors
2915 @opindex pedantic-errors
2916 Like @option{-pedantic}, except that errors are produced rather than
2922 This enables all the warnings about constructions that some users
2923 consider questionable, and that are easy to avoid (or modify to
2924 prevent the warning), even in conjunction with macros. This also
2925 enables some language-specific warnings described in @ref{C++ Dialect
2926 Options} and @ref{Objective-C and Objective-C++ Dialect Options}.
2928 @option{-Wall} turns on the following warning flags:
2930 @gccoptlist{-Waddress @gol
2931 -Warray-bounds @r{(only with} @option{-O2}@r{)} @gol
2933 -Wchar-subscripts @gol
2934 -Wenum-compare @r{(in C/Objc; this is on by default in C++)} @gol
2935 -Wimplicit-int @r{(C and Objective-C only)} @gol
2936 -Wimplicit-function-declaration @r{(C and Objective-C only)} @gol
2939 -Wmain @r{(only for C/ObjC and unless} @option{-ffreestanding}@r{)} @gol
2940 -Wmissing-braces @gol
2946 -Wsequence-point @gol
2947 -Wsign-compare @r{(only in C++)} @gol
2948 -Wstrict-aliasing @gol
2949 -Wstrict-overflow=1 @gol
2952 -Wuninitialized @gol
2953 -Wunknown-pragmas @gol
2954 -Wunused-function @gol
2957 -Wunused-variable @gol
2958 -Wvolatile-register-var @gol
2961 Note that some warning flags are not implied by @option{-Wall}. Some of
2962 them warn about constructions that users generally do not consider
2963 questionable, but which occasionally you might wish to check for;
2964 others warn about constructions that are necessary or hard to avoid in
2965 some cases, and there is no simple way to modify the code to suppress
2966 the warning. Some of them are enabled by @option{-Wextra} but many of
2967 them must be enabled individually.
2973 This enables some extra warning flags that are not enabled by
2974 @option{-Wall}. (This option used to be called @option{-W}. The older
2975 name is still supported, but the newer name is more descriptive.)
2977 @gccoptlist{-Wclobbered @gol
2979 -Wignored-qualifiers @gol
2980 -Wmissing-field-initializers @gol
2981 -Wmissing-parameter-type @r{(C only)} @gol
2982 -Wold-style-declaration @r{(C only)} @gol
2983 -Woverride-init @gol
2986 -Wuninitialized @gol
2987 -Wunused-parameter @r{(only with} @option{-Wunused} @r{or} @option{-Wall}@r{)} @gol
2988 -Wunused-but-set-parameter @r{(only with} @option{-Wunused} @r{or} @option{-Wall}@r{)} @gol
2991 The option @option{-Wextra} also prints warning messages for the
2997 A pointer is compared against integer zero with @samp{<}, @samp{<=},
2998 @samp{>}, or @samp{>=}.
3001 (C++ only) An enumerator and a non-enumerator both appear in a
3002 conditional expression.
3005 (C++ only) Ambiguous virtual bases.
3008 (C++ only) Subscripting an array which has been declared @samp{register}.
3011 (C++ only) Taking the address of a variable which has been declared
3015 (C++ only) A base class is not initialized in a derived class' copy
3020 @item -Wchar-subscripts
3021 @opindex Wchar-subscripts
3022 @opindex Wno-char-subscripts
3023 Warn if an array subscript has type @code{char}. This is a common cause
3024 of error, as programmers often forget that this type is signed on some
3026 This warning is enabled by @option{-Wall}.
3030 @opindex Wno-comment
3031 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
3032 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
3033 This warning is enabled by @option{-Wall}.
3036 @r{(C, Objective-C, C++, Objective-C++ and Fortran only)}
3038 Suppress warning messages emitted by @code{#warning} directives.
3043 @opindex ffreestanding
3044 @opindex fno-builtin
3045 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
3046 the arguments supplied have types appropriate to the format string
3047 specified, and that the conversions specified in the format string make
3048 sense. This includes standard functions, and others specified by format
3049 attributes (@pxref{Function Attributes}), in the @code{printf},
3050 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
3051 not in the C standard) families (or other target-specific families).
3052 Which functions are checked without format attributes having been
3053 specified depends on the standard version selected, and such checks of
3054 functions without the attribute specified are disabled by
3055 @option{-ffreestanding} or @option{-fno-builtin}.
3057 The formats are checked against the format features supported by GNU
3058 libc version 2.2. These include all ISO C90 and C99 features, as well
3059 as features from the Single Unix Specification and some BSD and GNU
3060 extensions. Other library implementations may not support all these
3061 features; GCC does not support warning about features that go beyond a
3062 particular library's limitations. However, if @option{-pedantic} is used
3063 with @option{-Wformat}, warnings will be given about format features not
3064 in the selected standard version (but not for @code{strfmon} formats,
3065 since those are not in any version of the C standard). @xref{C Dialect
3066 Options,,Options Controlling C Dialect}.
3068 Since @option{-Wformat} also checks for null format arguments for
3069 several functions, @option{-Wformat} also implies @option{-Wnonnull}.
3071 @option{-Wformat} is included in @option{-Wall}. For more control over some
3072 aspects of format checking, the options @option{-Wformat-y2k},
3073 @option{-Wno-format-extra-args}, @option{-Wno-format-zero-length},
3074 @option{-Wformat-nonliteral}, @option{-Wformat-security}, and
3075 @option{-Wformat=2} are available, but are not included in @option{-Wall}.
3078 @opindex Wformat-y2k
3079 @opindex Wno-format-y2k
3080 If @option{-Wformat} is specified, also warn about @code{strftime}
3081 formats which may yield only a two-digit year.
3083 @item -Wno-format-contains-nul
3084 @opindex Wno-format-contains-nul
3085 @opindex Wformat-contains-nul
3086 If @option{-Wformat} is specified, do not warn about format strings that
3089 @item -Wno-format-extra-args
3090 @opindex Wno-format-extra-args
3091 @opindex Wformat-extra-args
3092 If @option{-Wformat} is specified, do not warn about excess arguments to a
3093 @code{printf} or @code{scanf} format function. The C standard specifies
3094 that such arguments are ignored.
3096 Where the unused arguments lie between used arguments that are
3097 specified with @samp{$} operand number specifications, normally
3098 warnings are still given, since the implementation could not know what
3099 type to pass to @code{va_arg} to skip the unused arguments. However,
3100 in the case of @code{scanf} formats, this option will suppress the
3101 warning if the unused arguments are all pointers, since the Single
3102 Unix Specification says that such unused arguments are allowed.
3104 @item -Wno-format-zero-length @r{(C and Objective-C only)}
3105 @opindex Wno-format-zero-length
3106 @opindex Wformat-zero-length
3107 If @option{-Wformat} is specified, do not warn about zero-length formats.
3108 The C standard specifies that zero-length formats are allowed.
3110 @item -Wformat-nonliteral
3111 @opindex Wformat-nonliteral
3112 @opindex Wno-format-nonliteral
3113 If @option{-Wformat} is specified, also warn if the format string is not a
3114 string literal and so cannot be checked, unless the format function
3115 takes its format arguments as a @code{va_list}.
3117 @item -Wformat-security
3118 @opindex Wformat-security
3119 @opindex Wno-format-security
3120 If @option{-Wformat} is specified, also warn about uses of format
3121 functions that represent possible security problems. At present, this
3122 warns about calls to @code{printf} and @code{scanf} functions where the
3123 format string is not a string literal and there are no format arguments,
3124 as in @code{printf (foo);}. This may be a security hole if the format
3125 string came from untrusted input and contains @samp{%n}. (This is
3126 currently a subset of what @option{-Wformat-nonliteral} warns about, but
3127 in future warnings may be added to @option{-Wformat-security} that are not
3128 included in @option{-Wformat-nonliteral}.)
3132 @opindex Wno-format=2
3133 Enable @option{-Wformat} plus format checks not included in
3134 @option{-Wformat}. Currently equivalent to @samp{-Wformat
3135 -Wformat-nonliteral -Wformat-security -Wformat-y2k}.
3137 @item -Wnonnull @r{(C and Objective-C only)}
3139 @opindex Wno-nonnull
3140 Warn about passing a null pointer for arguments marked as
3141 requiring a non-null value by the @code{nonnull} function attribute.
3143 @option{-Wnonnull} is included in @option{-Wall} and @option{-Wformat}. It
3144 can be disabled with the @option{-Wno-nonnull} option.
3146 @item -Winit-self @r{(C, C++, Objective-C and Objective-C++ only)}
3148 @opindex Wno-init-self
3149 Warn about uninitialized variables which are initialized with themselves.
3150 Note this option can only be used with the @option{-Wuninitialized} option.
3152 For example, GCC will warn about @code{i} being uninitialized in the
3153 following snippet only when @option{-Winit-self} has been specified:
3164 @item -Wimplicit-int @r{(C and Objective-C only)}
3165 @opindex Wimplicit-int
3166 @opindex Wno-implicit-int
3167 Warn when a declaration does not specify a type.
3168 This warning is enabled by @option{-Wall}.
3170 @item -Wimplicit-function-declaration @r{(C and Objective-C only)}
3171 @opindex Wimplicit-function-declaration
3172 @opindex Wno-implicit-function-declaration
3173 Give a warning whenever a function is used before being declared. In
3174 C99 mode (@option{-std=c99} or @option{-std=gnu99}), this warning is
3175 enabled by default and it is made into an error by
3176 @option{-pedantic-errors}. This warning is also enabled by
3179 @item -Wimplicit @r{(C and Objective-C only)}
3181 @opindex Wno-implicit
3182 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
3183 This warning is enabled by @option{-Wall}.
3185 @item -Wignored-qualifiers @r{(C and C++ only)}
3186 @opindex Wignored-qualifiers
3187 @opindex Wno-ignored-qualifiers
3188 Warn if the return type of a function has a type qualifier
3189 such as @code{const}. For ISO C such a type qualifier has no effect,
3190 since the value returned by a function is not an lvalue.
3191 For C++, the warning is only emitted for scalar types or @code{void}.
3192 ISO C prohibits qualified @code{void} return types on function
3193 definitions, so such return types always receive a warning
3194 even without this option.
3196 This warning is also enabled by @option{-Wextra}.
3201 Warn if the type of @samp{main} is suspicious. @samp{main} should be
3202 a function with external linkage, returning int, taking either zero
3203 arguments, two, or three arguments of appropriate types. This warning
3204 is enabled by default in C++ and is enabled by either @option{-Wall}
3205 or @option{-pedantic}.
3207 @item -Wmissing-braces
3208 @opindex Wmissing-braces
3209 @opindex Wno-missing-braces
3210 Warn if an aggregate or union initializer is not fully bracketed. In
3211 the following example, the initializer for @samp{a} is not fully
3212 bracketed, but that for @samp{b} is fully bracketed.
3215 int a[2][2] = @{ 0, 1, 2, 3 @};
3216 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
3219 This warning is enabled by @option{-Wall}.
3221 @item -Wmissing-include-dirs @r{(C, C++, Objective-C and Objective-C++ only)}
3222 @opindex Wmissing-include-dirs
3223 @opindex Wno-missing-include-dirs
3224 Warn if a user-supplied include directory does not exist.
3227 @opindex Wparentheses
3228 @opindex Wno-parentheses
3229 Warn if parentheses are omitted in certain contexts, such
3230 as when there is an assignment in a context where a truth value
3231 is expected, or when operators are nested whose precedence people
3232 often get confused about.
3234 Also warn if a comparison like @samp{x<=y<=z} appears; this is
3235 equivalent to @samp{(x<=y ? 1 : 0) <= z}, which is a different
3236 interpretation from that of ordinary mathematical notation.
3238 Also warn about constructions where there may be confusion to which
3239 @code{if} statement an @code{else} branch belongs. Here is an example of
3254 In C/C++, every @code{else} branch belongs to the innermost possible
3255 @code{if} statement, which in this example is @code{if (b)}. This is
3256 often not what the programmer expected, as illustrated in the above
3257 example by indentation the programmer chose. When there is the
3258 potential for this confusion, GCC will issue a warning when this flag
3259 is specified. To eliminate the warning, add explicit braces around
3260 the innermost @code{if} statement so there is no way the @code{else}
3261 could belong to the enclosing @code{if}. The resulting code would
3278 Also warn for dangerous uses of the
3279 ?: with omitted middle operand GNU extension. When the condition
3280 in the ?: operator is a boolean expression the omitted value will
3281 be always 1. Often the user expects it to be a value computed
3282 inside the conditional expression instead.
3284 This warning is enabled by @option{-Wall}.
3286 @item -Wsequence-point
3287 @opindex Wsequence-point
3288 @opindex Wno-sequence-point
3289 Warn about code that may have undefined semantics because of violations
3290 of sequence point rules in the C and C++ standards.
3292 The C and C++ standards defines the order in which expressions in a C/C++
3293 program are evaluated in terms of @dfn{sequence points}, which represent
3294 a partial ordering between the execution of parts of the program: those
3295 executed before the sequence point, and those executed after it. These
3296 occur after the evaluation of a full expression (one which is not part
3297 of a larger expression), after the evaluation of the first operand of a
3298 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
3299 function is called (but after the evaluation of its arguments and the
3300 expression denoting the called function), and in certain other places.
3301 Other than as expressed by the sequence point rules, the order of
3302 evaluation of subexpressions of an expression is not specified. All
3303 these rules describe only a partial order rather than a total order,
3304 since, for example, if two functions are called within one expression
3305 with no sequence point between them, the order in which the functions
3306 are called is not specified. However, the standards committee have
3307 ruled that function calls do not overlap.
3309 It is not specified when between sequence points modifications to the
3310 values of objects take effect. Programs whose behavior depends on this
3311 have undefined behavior; the C and C++ standards specify that ``Between
3312 the previous and next sequence point an object shall have its stored
3313 value modified at most once by the evaluation of an expression.
3314 Furthermore, the prior value shall be read only to determine the value
3315 to be stored.''. If a program breaks these rules, the results on any
3316 particular implementation are entirely unpredictable.
3318 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
3319 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
3320 diagnosed by this option, and it may give an occasional false positive
3321 result, but in general it has been found fairly effective at detecting
3322 this sort of problem in programs.
3324 The standard is worded confusingly, therefore there is some debate
3325 over the precise meaning of the sequence point rules in subtle cases.
3326 Links to discussions of the problem, including proposed formal
3327 definitions, may be found on the GCC readings page, at
3328 @w{@uref{http://gcc.gnu.org/readings.html}}.
3330 This warning is enabled by @option{-Wall} for C and C++.
3333 @opindex Wreturn-type
3334 @opindex Wno-return-type
3335 Warn whenever a function is defined with a return-type that defaults
3336 to @code{int}. Also warn about any @code{return} statement with no
3337 return-value in a function whose return-type is not @code{void}
3338 (falling off the end of the function body is considered returning
3339 without a value), and about a @code{return} statement with an
3340 expression in a function whose return-type is @code{void}.
3342 For C++, a function without return type always produces a diagnostic
3343 message, even when @option{-Wno-return-type} is specified. The only
3344 exceptions are @samp{main} and functions defined in system headers.
3346 This warning is enabled by @option{-Wall}.
3351 Warn whenever a @code{switch} statement has an index of enumerated type
3352 and lacks a @code{case} for one or more of the named codes of that
3353 enumeration. (The presence of a @code{default} label prevents this
3354 warning.) @code{case} labels outside the enumeration range also
3355 provoke warnings when this option is used (even if there is a
3356 @code{default} label).
3357 This warning is enabled by @option{-Wall}.
3359 @item -Wswitch-default
3360 @opindex Wswitch-default
3361 @opindex Wno-switch-default
3362 Warn whenever a @code{switch} statement does not have a @code{default}
3366 @opindex Wswitch-enum
3367 @opindex Wno-switch-enum
3368 Warn whenever a @code{switch} statement has an index of enumerated type
3369 and lacks a @code{case} for one or more of the named codes of that
3370 enumeration. @code{case} labels outside the enumeration range also
3371 provoke warnings when this option is used. The only difference
3372 between @option{-Wswitch} and this option is that this option gives a
3373 warning about an omitted enumeration code even if there is a
3374 @code{default} label.
3376 @item -Wsync-nand @r{(C and C++ only)}
3378 @opindex Wno-sync-nand
3379 Warn when @code{__sync_fetch_and_nand} and @code{__sync_nand_and_fetch}
3380 built-in functions are used. These functions changed semantics in GCC 4.4.
3384 @opindex Wno-trigraphs
3385 Warn if any trigraphs are encountered that might change the meaning of
3386 the program (trigraphs within comments are not warned about).
3387 This warning is enabled by @option{-Wall}.
3389 @item -Wunused-but-set-parameter
3390 @opindex Wunused-but-set-parameter
3391 @opindex Wno-unused-but-set-parameter
3392 Warn whenever a function parameter is assigned to, but otherwise unused
3393 (aside from its declaration).
3395 To suppress this warning use the @samp{unused} attribute
3396 (@pxref{Variable Attributes}).
3398 This warning is also enabled by @option{-Wunused} together with
3401 @item -Wunused-but-set-variable
3402 @opindex Wunused-but-set-variable
3403 @opindex Wno-unused-but-set-variable
3404 Warn whenever a local variable is assigned to, but otherwise unused
3405 (aside from its declaration).
3406 This warning is enabled by @option{-Wall}.
3408 To suppress this warning use the @samp{unused} attribute
3409 (@pxref{Variable Attributes}).
3411 This warning is also enabled by @option{-Wunused}, which is enabled
3414 @item -Wunused-function
3415 @opindex Wunused-function
3416 @opindex Wno-unused-function
3417 Warn whenever a static function is declared but not defined or a
3418 non-inline static function is unused.
3419 This warning is enabled by @option{-Wall}.
3421 @item -Wunused-label
3422 @opindex Wunused-label
3423 @opindex Wno-unused-label
3424 Warn whenever a label is declared but not used.
3425 This warning is enabled by @option{-Wall}.
3427 To suppress this warning use the @samp{unused} attribute
3428 (@pxref{Variable Attributes}).
3430 @item -Wunused-parameter
3431 @opindex Wunused-parameter
3432 @opindex Wno-unused-parameter
3433 Warn whenever a function parameter is unused aside from its declaration.
3435 To suppress this warning use the @samp{unused} attribute
3436 (@pxref{Variable Attributes}).
3438 @item -Wno-unused-result
3439 @opindex Wunused-result
3440 @opindex Wno-unused-result
3441 Do not warn if a caller of a function marked with attribute
3442 @code{warn_unused_result} (@pxref{Variable Attributes}) does not use
3443 its return value. The default is @option{-Wunused-result}.
3445 @item -Wunused-variable
3446 @opindex Wunused-variable
3447 @opindex Wno-unused-variable
3448 Warn whenever a local variable or non-constant static variable is unused
3449 aside from its declaration.
3450 This warning is enabled by @option{-Wall}.
3452 To suppress this warning use the @samp{unused} attribute
3453 (@pxref{Variable Attributes}).
3455 @item -Wunused-value
3456 @opindex Wunused-value
3457 @opindex Wno-unused-value
3458 Warn whenever a statement computes a result that is explicitly not
3459 used. To suppress this warning cast the unused expression to
3460 @samp{void}. This includes an expression-statement or the left-hand
3461 side of a comma expression that contains no side effects. For example,
3462 an expression such as @samp{x[i,j]} will cause a warning, while
3463 @samp{x[(void)i,j]} will not.
3465 This warning is enabled by @option{-Wall}.
3470 All the above @option{-Wunused} options combined.
3472 In order to get a warning about an unused function parameter, you must
3473 either specify @samp{-Wextra -Wunused} (note that @samp{-Wall} implies
3474 @samp{-Wunused}), or separately specify @option{-Wunused-parameter}.
3476 @item -Wuninitialized
3477 @opindex Wuninitialized
3478 @opindex Wno-uninitialized
3479 Warn if an automatic variable is used without first being initialized
3480 or if a variable may be clobbered by a @code{setjmp} call. In C++,
3481 warn if a non-static reference or non-static @samp{const} member
3482 appears in a class without constructors.
3484 If you want to warn about code which uses the uninitialized value of the
3485 variable in its own initializer, use the @option{-Winit-self} option.
3487 These warnings occur for individual uninitialized or clobbered
3488 elements of structure, union or array variables as well as for
3489 variables which are uninitialized or clobbered as a whole. They do
3490 not occur for variables or elements declared @code{volatile}. Because
3491 these warnings depend on optimization, the exact variables or elements
3492 for which there are warnings will depend on the precise optimization
3493 options and version of GCC used.
3495 Note that there may be no warning about a variable that is used only
3496 to compute a value that itself is never used, because such
3497 computations may be deleted by data flow analysis before the warnings
3500 These warnings are made optional because GCC is not smart
3501 enough to see all the reasons why the code might be correct
3502 despite appearing to have an error. Here is one example of how
3523 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
3524 always initialized, but GCC doesn't know this. Here is
3525 another common case:
3530 if (change_y) save_y = y, y = new_y;
3532 if (change_y) y = save_y;
3537 This has no bug because @code{save_y} is used only if it is set.
3539 @cindex @code{longjmp} warnings
3540 This option also warns when a non-volatile automatic variable might be
3541 changed by a call to @code{longjmp}. These warnings as well are possible
3542 only in optimizing compilation.
3544 The compiler sees only the calls to @code{setjmp}. It cannot know
3545 where @code{longjmp} will be called; in fact, a signal handler could
3546 call it at any point in the code. As a result, you may get a warning
3547 even when there is in fact no problem because @code{longjmp} cannot
3548 in fact be called at the place which would cause a problem.
3550 Some spurious warnings can be avoided if you declare all the functions
3551 you use that never return as @code{noreturn}. @xref{Function
3554 This warning is enabled by @option{-Wall} or @option{-Wextra}.
3556 @item -Wunknown-pragmas
3557 @opindex Wunknown-pragmas
3558 @opindex Wno-unknown-pragmas
3559 @cindex warning for unknown pragmas
3560 @cindex unknown pragmas, warning
3561 @cindex pragmas, warning of unknown
3562 Warn when a #pragma directive is encountered which is not understood by
3563 GCC@. If this command line option is used, warnings will even be issued
3564 for unknown pragmas in system header files. This is not the case if
3565 the warnings were only enabled by the @option{-Wall} command line option.
3568 @opindex Wno-pragmas
3570 Do not warn about misuses of pragmas, such as incorrect parameters,
3571 invalid syntax, or conflicts between pragmas. See also
3572 @samp{-Wunknown-pragmas}.
3574 @item -Wstrict-aliasing
3575 @opindex Wstrict-aliasing
3576 @opindex Wno-strict-aliasing
3577 This option is only active when @option{-fstrict-aliasing} is active.
3578 It warns about code which might break the strict aliasing rules that the
3579 compiler is using for optimization. The warning does not catch all
3580 cases, but does attempt to catch the more common pitfalls. It is
3581 included in @option{-Wall}.
3582 It is equivalent to @option{-Wstrict-aliasing=3}
3584 @item -Wstrict-aliasing=n
3585 @opindex Wstrict-aliasing=n
3586 @opindex Wno-strict-aliasing=n
3587 This option is only active when @option{-fstrict-aliasing} is active.
3588 It warns about code which might break the strict aliasing rules that the
3589 compiler is using for optimization.
3590 Higher levels correspond to higher accuracy (fewer false positives).
3591 Higher levels also correspond to more effort, similar to the way -O works.
3592 @option{-Wstrict-aliasing} is equivalent to @option{-Wstrict-aliasing=n},
3595 Level 1: Most aggressive, quick, least accurate.
3596 Possibly useful when higher levels
3597 do not warn but -fstrict-aliasing still breaks the code, as it has very few
3598 false negatives. However, it has many false positives.
3599 Warns for all pointer conversions between possibly incompatible types,
3600 even if never dereferenced. Runs in the frontend only.
3602 Level 2: Aggressive, quick, not too precise.
3603 May still have many false positives (not as many as level 1 though),
3604 and few false negatives (but possibly more than level 1).
3605 Unlike level 1, it only warns when an address is taken. Warns about
3606 incomplete types. Runs in the frontend only.
3608 Level 3 (default for @option{-Wstrict-aliasing}):
3609 Should have very few false positives and few false
3610 negatives. Slightly slower than levels 1 or 2 when optimization is enabled.
3611 Takes care of the common pun+dereference pattern in the frontend:
3612 @code{*(int*)&some_float}.
3613 If optimization is enabled, it also runs in the backend, where it deals
3614 with multiple statement cases using flow-sensitive points-to information.
3615 Only warns when the converted pointer is dereferenced.
3616 Does not warn about incomplete types.
3618 @item -Wstrict-overflow
3619 @itemx -Wstrict-overflow=@var{n}
3620 @opindex Wstrict-overflow
3621 @opindex Wno-strict-overflow
3622 This option is only active when @option{-fstrict-overflow} is active.
3623 It warns about cases where the compiler optimizes based on the
3624 assumption that signed overflow does not occur. Note that it does not
3625 warn about all cases where the code might overflow: it only warns
3626 about cases where the compiler implements some optimization. Thus
3627 this warning depends on the optimization level.
3629 An optimization which assumes that signed overflow does not occur is
3630 perfectly safe if the values of the variables involved are such that
3631 overflow never does, in fact, occur. Therefore this warning can
3632 easily give a false positive: a warning about code which is not
3633 actually a problem. To help focus on important issues, several
3634 warning levels are defined. No warnings are issued for the use of
3635 undefined signed overflow when estimating how many iterations a loop
3636 will require, in particular when determining whether a loop will be
3640 @item -Wstrict-overflow=1
3641 Warn about cases which are both questionable and easy to avoid. For
3642 example: @code{x + 1 > x}; with @option{-fstrict-overflow}, the
3643 compiler will simplify this to @code{1}. This level of
3644 @option{-Wstrict-overflow} is enabled by @option{-Wall}; higher levels
3645 are not, and must be explicitly requested.
3647 @item -Wstrict-overflow=2
3648 Also warn about other cases where a comparison is simplified to a
3649 constant. For example: @code{abs (x) >= 0}. This can only be
3650 simplified when @option{-fstrict-overflow} is in effect, because
3651 @code{abs (INT_MIN)} overflows to @code{INT_MIN}, which is less than
3652 zero. @option{-Wstrict-overflow} (with no level) is the same as
3653 @option{-Wstrict-overflow=2}.
3655 @item -Wstrict-overflow=3
3656 Also warn about other cases where a comparison is simplified. For
3657 example: @code{x + 1 > 1} will be simplified to @code{x > 0}.
3659 @item -Wstrict-overflow=4
3660 Also warn about other simplifications not covered by the above cases.
3661 For example: @code{(x * 10) / 5} will be simplified to @code{x * 2}.
3663 @item -Wstrict-overflow=5
3664 Also warn about cases where the compiler reduces the magnitude of a
3665 constant involved in a comparison. For example: @code{x + 2 > y} will
3666 be simplified to @code{x + 1 >= y}. This is reported only at the
3667 highest warning level because this simplification applies to many
3668 comparisons, so this warning level will give a very large number of
3672 @item -Wsuggest-attribute=@r{[}pure@r{|}const@r{|}noreturn@r{]}
3673 @opindex Wsuggest-attribute=
3674 @opindex Wno-suggest-attribute=
3675 Warn for cases where adding an attribute may be beneficial. The
3676 attributes currently supported are listed below.
3679 @item -Wsuggest-attribute=pure
3680 @itemx -Wsuggest-attribute=const
3681 @itemx -Wsuggest-attribute=noreturn
3682 @opindex Wsuggest-attribute=pure
3683 @opindex Wno-suggest-attribute=pure
3684 @opindex Wsuggest-attribute=const
3685 @opindex Wno-suggest-attribute=const
3686 @opindex Wsuggest-attribute=noreturn
3687 @opindex Wno-suggest-attribute=noreturn
3689 Warn about functions which might be candidates for attributes
3690 @code{pure}, @code{const} or @code{noreturn}. The compiler only warns for
3691 functions visible in other compilation units or (in the case of @code{pure} and
3692 @code{const}) if it cannot prove that the function returns normally. A function
3693 returns normally if it doesn't contain an infinite loop nor returns abnormally
3694 by throwing, calling @code{abort()} or trapping. This analysis requires option
3695 @option{-fipa-pure-const}, which is enabled by default at @option{-O} and
3696 higher. Higher optimization levels improve the accuracy of the analysis.
3699 @item -Warray-bounds
3700 @opindex Wno-array-bounds
3701 @opindex Warray-bounds
3702 This option is only active when @option{-ftree-vrp} is active
3703 (default for @option{-O2} and above). It warns about subscripts to arrays
3704 that are always out of bounds. This warning is enabled by @option{-Wall}.
3706 @item -Wno-div-by-zero
3707 @opindex Wno-div-by-zero
3708 @opindex Wdiv-by-zero
3709 Do not warn about compile-time integer division by zero. Floating point
3710 division by zero is not warned about, as it can be a legitimate way of
3711 obtaining infinities and NaNs.
3713 @item -Wsystem-headers
3714 @opindex Wsystem-headers
3715 @opindex Wno-system-headers
3716 @cindex warnings from system headers
3717 @cindex system headers, warnings from
3718 Print warning messages for constructs found in system header files.
3719 Warnings from system headers are normally suppressed, on the assumption
3720 that they usually do not indicate real problems and would only make the
3721 compiler output harder to read. Using this command line option tells
3722 GCC to emit warnings from system headers as if they occurred in user
3723 code. However, note that using @option{-Wall} in conjunction with this
3724 option will @emph{not} warn about unknown pragmas in system
3725 headers---for that, @option{-Wunknown-pragmas} must also be used.
3728 @opindex Wtrampolines
3729 @opindex Wno-trampolines
3730 Warn about trampolines generated for pointers to nested functions.
3732 A trampoline is a small piece of data or code that is created at run
3733 time on the stack when the address of a nested function is taken, and
3734 is used to call the nested function indirectly. For some targets, it
3735 is made up of data only and thus requires no special treatment. But,
3736 for most targets, it is made up of code and thus requires the stack
3737 to be made executable in order for the program to work properly.
3740 @opindex Wfloat-equal
3741 @opindex Wno-float-equal
3742 Warn if floating point values are used in equality comparisons.
3744 The idea behind this is that sometimes it is convenient (for the
3745 programmer) to consider floating-point values as approximations to
3746 infinitely precise real numbers. If you are doing this, then you need
3747 to compute (by analyzing the code, or in some other way) the maximum or
3748 likely maximum error that the computation introduces, and allow for it
3749 when performing comparisons (and when producing output, but that's a
3750 different problem). In particular, instead of testing for equality, you
3751 would check to see whether the two values have ranges that overlap; and
3752 this is done with the relational operators, so equality comparisons are
3755 @item -Wtraditional @r{(C and Objective-C only)}
3756 @opindex Wtraditional
3757 @opindex Wno-traditional
3758 Warn about certain constructs that behave differently in traditional and
3759 ISO C@. Also warn about ISO C constructs that have no traditional C
3760 equivalent, and/or problematic constructs which should be avoided.
3764 Macro parameters that appear within string literals in the macro body.
3765 In traditional C macro replacement takes place within string literals,
3766 but does not in ISO C@.
3769 In traditional C, some preprocessor directives did not exist.
3770 Traditional preprocessors would only consider a line to be a directive
3771 if the @samp{#} appeared in column 1 on the line. Therefore
3772 @option{-Wtraditional} warns about directives that traditional C
3773 understands but would ignore because the @samp{#} does not appear as the
3774 first character on the line. It also suggests you hide directives like
3775 @samp{#pragma} not understood by traditional C by indenting them. Some
3776 traditional implementations would not recognize @samp{#elif}, so it
3777 suggests avoiding it altogether.
3780 A function-like macro that appears without arguments.
3783 The unary plus operator.
3786 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
3787 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
3788 constants.) Note, these suffixes appear in macros defined in the system
3789 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
3790 Use of these macros in user code might normally lead to spurious
3791 warnings, however GCC's integrated preprocessor has enough context to
3792 avoid warning in these cases.
3795 A function declared external in one block and then used after the end of
3799 A @code{switch} statement has an operand of type @code{long}.
3802 A non-@code{static} function declaration follows a @code{static} one.
3803 This construct is not accepted by some traditional C compilers.
3806 The ISO type of an integer constant has a different width or
3807 signedness from its traditional type. This warning is only issued if
3808 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
3809 typically represent bit patterns, are not warned about.
3812 Usage of ISO string concatenation is detected.
3815 Initialization of automatic aggregates.
3818 Identifier conflicts with labels. Traditional C lacks a separate
3819 namespace for labels.
3822 Initialization of unions. If the initializer is zero, the warning is
3823 omitted. This is done under the assumption that the zero initializer in
3824 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
3825 initializer warnings and relies on default initialization to zero in the
3829 Conversions by prototypes between fixed/floating point values and vice
3830 versa. The absence of these prototypes when compiling with traditional
3831 C would cause serious problems. This is a subset of the possible
3832 conversion warnings, for the full set use @option{-Wtraditional-conversion}.
3835 Use of ISO C style function definitions. This warning intentionally is
3836 @emph{not} issued for prototype declarations or variadic functions
3837 because these ISO C features will appear in your code when using
3838 libiberty's traditional C compatibility macros, @code{PARAMS} and
3839 @code{VPARAMS}. This warning is also bypassed for nested functions
3840 because that feature is already a GCC extension and thus not relevant to
3841 traditional C compatibility.
3844 @item -Wtraditional-conversion @r{(C and Objective-C only)}
3845 @opindex Wtraditional-conversion
3846 @opindex Wno-traditional-conversion
3847 Warn if a prototype causes a type conversion that is different from what
3848 would happen to the same argument in the absence of a prototype. This
3849 includes conversions of fixed point to floating and vice versa, and
3850 conversions changing the width or signedness of a fixed point argument
3851 except when the same as the default promotion.
3853 @item -Wdeclaration-after-statement @r{(C and Objective-C only)}
3854 @opindex Wdeclaration-after-statement
3855 @opindex Wno-declaration-after-statement
3856 Warn when a declaration is found after a statement in a block. This
3857 construct, known from C++, was introduced with ISO C99 and is by default
3858 allowed in GCC@. It is not supported by ISO C90 and was not supported by
3859 GCC versions before GCC 3.0. @xref{Mixed Declarations}.
3864 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
3866 @item -Wno-endif-labels
3867 @opindex Wno-endif-labels
3868 @opindex Wendif-labels
3869 Do not warn whenever an @samp{#else} or an @samp{#endif} are followed by text.
3874 Warn whenever a local variable or type declaration shadows another variable,
3875 parameter, type, or class member (in C++), or whenever a built-in function
3876 is shadowed. Note that in C++, the compiler will not warn if a local variable
3877 shadows a struct/class/enum, but will warn if it shadows an explicit typedef.
3879 @item -Wlarger-than=@var{len}
3880 @opindex Wlarger-than=@var{len}
3881 @opindex Wlarger-than-@var{len}
3882 Warn whenever an object of larger than @var{len} bytes is defined.
3884 @item -Wframe-larger-than=@var{len}
3885 @opindex Wframe-larger-than
3886 Warn if the size of a function frame is larger than @var{len} bytes.
3887 The computation done to determine the stack frame size is approximate
3888 and not conservative.
3889 The actual requirements may be somewhat greater than @var{len}
3890 even if you do not get a warning. In addition, any space allocated
3891 via @code{alloca}, variable-length arrays, or related constructs
3892 is not included by the compiler when determining
3893 whether or not to issue a warning.
3895 @item -Wunsafe-loop-optimizations
3896 @opindex Wunsafe-loop-optimizations
3897 @opindex Wno-unsafe-loop-optimizations
3898 Warn if the loop cannot be optimized because the compiler could not
3899 assume anything on the bounds of the loop indices. With
3900 @option{-funsafe-loop-optimizations} warn if the compiler made
3903 @item -Wno-pedantic-ms-format @r{(MinGW targets only)}
3904 @opindex Wno-pedantic-ms-format
3905 @opindex Wpedantic-ms-format
3906 Disables the warnings about non-ISO @code{printf} / @code{scanf} format
3907 width specifiers @code{I32}, @code{I64}, and @code{I} used on Windows targets
3908 depending on the MS runtime, when you are using the options @option{-Wformat}
3909 and @option{-pedantic} without gnu-extensions.
3911 @item -Wpointer-arith
3912 @opindex Wpointer-arith
3913 @opindex Wno-pointer-arith
3914 Warn about anything that depends on the ``size of'' a function type or
3915 of @code{void}. GNU C assigns these types a size of 1, for
3916 convenience in calculations with @code{void *} pointers and pointers
3917 to functions. In C++, warn also when an arithmetic operation involves
3918 @code{NULL}. This warning is also enabled by @option{-pedantic}.
3921 @opindex Wtype-limits
3922 @opindex Wno-type-limits
3923 Warn if a comparison is always true or always false due to the limited
3924 range of the data type, but do not warn for constant expressions. For
3925 example, warn if an unsigned variable is compared against zero with
3926 @samp{<} or @samp{>=}. This warning is also enabled by
3929 @item -Wbad-function-cast @r{(C and Objective-C only)}
3930 @opindex Wbad-function-cast
3931 @opindex Wno-bad-function-cast
3932 Warn whenever a function call is cast to a non-matching type.
3933 For example, warn if @code{int malloc()} is cast to @code{anything *}.
3935 @item -Wc++-compat @r{(C and Objective-C only)}
3936 Warn about ISO C constructs that are outside of the common subset of
3937 ISO C and ISO C++, e.g.@: request for implicit conversion from
3938 @code{void *} to a pointer to non-@code{void} type.
3940 @item -Wc++0x-compat @r{(C++ and Objective-C++ only)}
3941 Warn about C++ constructs whose meaning differs between ISO C++ 1998 and
3942 ISO C++ 200x, e.g., identifiers in ISO C++ 1998 that will become keywords
3943 in ISO C++ 200x. This warning is enabled by @option{-Wall}.
3947 @opindex Wno-cast-qual
3948 Warn whenever a pointer is cast so as to remove a type qualifier from
3949 the target type. For example, warn if a @code{const char *} is cast
3950 to an ordinary @code{char *}.
3952 Also warn when making a cast which introduces a type qualifier in an
3953 unsafe way. For example, casting @code{char **} to @code{const char **}
3954 is unsafe, as in this example:
3957 /* p is char ** value. */
3958 const char **q = (const char **) p;
3959 /* Assignment of readonly string to const char * is OK. */
3961 /* Now char** pointer points to read-only memory. */
3966 @opindex Wcast-align
3967 @opindex Wno-cast-align
3968 Warn whenever a pointer is cast such that the required alignment of the
3969 target is increased. For example, warn if a @code{char *} is cast to
3970 an @code{int *} on machines where integers can only be accessed at
3971 two- or four-byte boundaries.
3973 @item -Wwrite-strings
3974 @opindex Wwrite-strings
3975 @opindex Wno-write-strings
3976 When compiling C, give string constants the type @code{const
3977 char[@var{length}]} so that copying the address of one into a
3978 non-@code{const} @code{char *} pointer will get a warning. These
3979 warnings will help you find at compile time code that can try to write
3980 into a string constant, but only if you have been very careful about
3981 using @code{const} in declarations and prototypes. Otherwise, it will
3982 just be a nuisance. This is why we did not make @option{-Wall} request
3985 When compiling C++, warn about the deprecated conversion from string
3986 literals to @code{char *}. This warning is enabled by default for C++
3991 @opindex Wno-clobbered
3992 Warn for variables that might be changed by @samp{longjmp} or
3993 @samp{vfork}. This warning is also enabled by @option{-Wextra}.
3996 @opindex Wconversion
3997 @opindex Wno-conversion
3998 Warn for implicit conversions that may alter a value. This includes
3999 conversions between real and integer, like @code{abs (x)} when
4000 @code{x} is @code{double}; conversions between signed and unsigned,
4001 like @code{unsigned ui = -1}; and conversions to smaller types, like
4002 @code{sqrtf (M_PI)}. Do not warn for explicit casts like @code{abs
4003 ((int) x)} and @code{ui = (unsigned) -1}, or if the value is not
4004 changed by the conversion like in @code{abs (2.0)}. Warnings about
4005 conversions between signed and unsigned integers can be disabled by
4006 using @option{-Wno-sign-conversion}.
4008 For C++, also warn for confusing overload resolution for user-defined
4009 conversions; and conversions that will never use a type conversion
4010 operator: conversions to @code{void}, the same type, a base class or a
4011 reference to them. Warnings about conversions between signed and
4012 unsigned integers are disabled by default in C++ unless
4013 @option{-Wsign-conversion} is explicitly enabled.
4015 @item -Wno-conversion-null @r{(C++ and Objective-C++ only)}
4016 @opindex Wconversion-null
4017 @opindex Wno-conversion-null
4018 Do not warn for conversions between @code{NULL} and non-pointer
4019 types. @option{-Wconversion-null} is enabled by default.
4022 @opindex Wempty-body
4023 @opindex Wno-empty-body
4024 Warn if an empty body occurs in an @samp{if}, @samp{else} or @samp{do
4025 while} statement. This warning is also enabled by @option{-Wextra}.
4027 @item -Wenum-compare
4028 @opindex Wenum-compare
4029 @opindex Wno-enum-compare
4030 Warn about a comparison between values of different enum types. In C++
4031 this warning is enabled by default. In C this warning is enabled by
4034 @item -Wjump-misses-init @r{(C, Objective-C only)}
4035 @opindex Wjump-misses-init
4036 @opindex Wno-jump-misses-init
4037 Warn if a @code{goto} statement or a @code{switch} statement jumps
4038 forward across the initialization of a variable, or jumps backward to a
4039 label after the variable has been initialized. This only warns about
4040 variables which are initialized when they are declared. This warning is
4041 only supported for C and Objective C; in C++ this sort of branch is an
4044 @option{-Wjump-misses-init} is included in @option{-Wc++-compat}. It
4045 can be disabled with the @option{-Wno-jump-misses-init} option.
4047 @item -Wsign-compare
4048 @opindex Wsign-compare
4049 @opindex Wno-sign-compare
4050 @cindex warning for comparison of signed and unsigned values
4051 @cindex comparison of signed and unsigned values, warning
4052 @cindex signed and unsigned values, comparison warning
4053 Warn when a comparison between signed and unsigned values could produce
4054 an incorrect result when the signed value is converted to unsigned.
4055 This warning is also enabled by @option{-Wextra}; to get the other warnings
4056 of @option{-Wextra} without this warning, use @samp{-Wextra -Wno-sign-compare}.
4058 @item -Wsign-conversion
4059 @opindex Wsign-conversion
4060 @opindex Wno-sign-conversion
4061 Warn for implicit conversions that may change the sign of an integer
4062 value, like assigning a signed integer expression to an unsigned
4063 integer variable. An explicit cast silences the warning. In C, this
4064 option is enabled also by @option{-Wconversion}.
4068 @opindex Wno-address
4069 Warn about suspicious uses of memory addresses. These include using
4070 the address of a function in a conditional expression, such as
4071 @code{void func(void); if (func)}, and comparisons against the memory
4072 address of a string literal, such as @code{if (x == "abc")}. Such
4073 uses typically indicate a programmer error: the address of a function
4074 always evaluates to true, so their use in a conditional usually
4075 indicate that the programmer forgot the parentheses in a function
4076 call; and comparisons against string literals result in unspecified
4077 behavior and are not portable in C, so they usually indicate that the
4078 programmer intended to use @code{strcmp}. This warning is enabled by
4082 @opindex Wlogical-op
4083 @opindex Wno-logical-op
4084 Warn about suspicious uses of logical operators in expressions.
4085 This includes using logical operators in contexts where a
4086 bit-wise operator is likely to be expected.
4088 @item -Waggregate-return
4089 @opindex Waggregate-return
4090 @opindex Wno-aggregate-return
4091 Warn if any functions that return structures or unions are defined or
4092 called. (In languages where you can return an array, this also elicits
4095 @item -Wno-attributes
4096 @opindex Wno-attributes
4097 @opindex Wattributes
4098 Do not warn if an unexpected @code{__attribute__} is used, such as
4099 unrecognized attributes, function attributes applied to variables,
4100 etc. This will not stop errors for incorrect use of supported
4103 @item -Wno-builtin-macro-redefined
4104 @opindex Wno-builtin-macro-redefined
4105 @opindex Wbuiltin-macro-redefined
4106 Do not warn if certain built-in macros are redefined. This suppresses
4107 warnings for redefinition of @code{__TIMESTAMP__}, @code{__TIME__},
4108 @code{__DATE__}, @code{__FILE__}, and @code{__BASE_FILE__}.
4110 @item -Wstrict-prototypes @r{(C and Objective-C only)}
4111 @opindex Wstrict-prototypes
4112 @opindex Wno-strict-prototypes
4113 Warn if a function is declared or defined without specifying the
4114 argument types. (An old-style function definition is permitted without
4115 a warning if preceded by a declaration which specifies the argument
4118 @item -Wold-style-declaration @r{(C and Objective-C only)}
4119 @opindex Wold-style-declaration
4120 @opindex Wno-old-style-declaration
4121 Warn for obsolescent usages, according to the C Standard, in a
4122 declaration. For example, warn if storage-class specifiers like
4123 @code{static} are not the first things in a declaration. This warning
4124 is also enabled by @option{-Wextra}.
4126 @item -Wold-style-definition @r{(C and Objective-C only)}
4127 @opindex Wold-style-definition
4128 @opindex Wno-old-style-definition
4129 Warn if an old-style function definition is used. A warning is given
4130 even if there is a previous prototype.
4132 @item -Wmissing-parameter-type @r{(C and Objective-C only)}
4133 @opindex Wmissing-parameter-type
4134 @opindex Wno-missing-parameter-type
4135 A function parameter is declared without a type specifier in K&R-style
4142 This warning is also enabled by @option{-Wextra}.
4144 @item -Wmissing-prototypes @r{(C and Objective-C only)}
4145 @opindex Wmissing-prototypes
4146 @opindex Wno-missing-prototypes
4147 Warn if a global function is defined without a previous prototype
4148 declaration. This warning is issued even if the definition itself
4149 provides a prototype. The aim is to detect global functions that fail
4150 to be declared in header files.
4152 @item -Wmissing-declarations
4153 @opindex Wmissing-declarations
4154 @opindex Wno-missing-declarations
4155 Warn if a global function is defined without a previous declaration.
4156 Do so even if the definition itself provides a prototype.
4157 Use this option to detect global functions that are not declared in
4158 header files. In C++, no warnings are issued for function templates,
4159 or for inline functions, or for functions in anonymous namespaces.
4161 @item -Wmissing-field-initializers
4162 @opindex Wmissing-field-initializers
4163 @opindex Wno-missing-field-initializers
4167 Warn if a structure's initializer has some fields missing. For
4168 example, the following code would cause such a warning, because
4169 @code{x.h} is implicitly zero:
4172 struct s @{ int f, g, h; @};
4173 struct s x = @{ 3, 4 @};
4176 This option does not warn about designated initializers, so the following
4177 modification would not trigger a warning:
4180 struct s @{ int f, g, h; @};
4181 struct s x = @{ .f = 3, .g = 4 @};
4184 This warning is included in @option{-Wextra}. To get other @option{-Wextra}
4185 warnings without this one, use @samp{-Wextra -Wno-missing-field-initializers}.
4187 @item -Wmissing-format-attribute
4188 @opindex Wmissing-format-attribute
4189 @opindex Wno-missing-format-attribute
4192 Warn about function pointers which might be candidates for @code{format}
4193 attributes. Note these are only possible candidates, not absolute ones.
4194 GCC will guess that function pointers with @code{format} attributes that
4195 are used in assignment, initialization, parameter passing or return
4196 statements should have a corresponding @code{format} attribute in the
4197 resulting type. I.e.@: the left-hand side of the assignment or
4198 initialization, the type of the parameter variable, or the return type
4199 of the containing function respectively should also have a @code{format}
4200 attribute to avoid the warning.
4202 GCC will also warn about function definitions which might be
4203 candidates for @code{format} attributes. Again, these are only
4204 possible candidates. GCC will guess that @code{format} attributes
4205 might be appropriate for any function that calls a function like
4206 @code{vprintf} or @code{vscanf}, but this might not always be the
4207 case, and some functions for which @code{format} attributes are
4208 appropriate may not be detected.
4210 @item -Wno-multichar
4211 @opindex Wno-multichar
4213 Do not warn if a multicharacter constant (@samp{'FOOF'}) is used.
4214 Usually they indicate a typo in the user's code, as they have
4215 implementation-defined values, and should not be used in portable code.
4217 @item -Wnormalized=<none|id|nfc|nfkc>
4218 @opindex Wnormalized=
4221 @cindex character set, input normalization
4222 In ISO C and ISO C++, two identifiers are different if they are
4223 different sequences of characters. However, sometimes when characters
4224 outside the basic ASCII character set are used, you can have two
4225 different character sequences that look the same. To avoid confusion,
4226 the ISO 10646 standard sets out some @dfn{normalization rules} which
4227 when applied ensure that two sequences that look the same are turned into
4228 the same sequence. GCC can warn you if you are using identifiers which
4229 have not been normalized; this option controls that warning.
4231 There are four levels of warning that GCC supports. The default is
4232 @option{-Wnormalized=nfc}, which warns about any identifier which is
4233 not in the ISO 10646 ``C'' normalized form, @dfn{NFC}. NFC is the
4234 recommended form for most uses.
4236 Unfortunately, there are some characters which ISO C and ISO C++ allow
4237 in identifiers that when turned into NFC aren't allowable as
4238 identifiers. That is, there's no way to use these symbols in portable
4239 ISO C or C++ and have all your identifiers in NFC@.
4240 @option{-Wnormalized=id} suppresses the warning for these characters.
4241 It is hoped that future versions of the standards involved will correct
4242 this, which is why this option is not the default.
4244 You can switch the warning off for all characters by writing
4245 @option{-Wnormalized=none}. You would only want to do this if you
4246 were using some other normalization scheme (like ``D''), because
4247 otherwise you can easily create bugs that are literally impossible to see.
4249 Some characters in ISO 10646 have distinct meanings but look identical
4250 in some fonts or display methodologies, especially once formatting has
4251 been applied. For instance @code{\u207F}, ``SUPERSCRIPT LATIN SMALL
4252 LETTER N'', will display just like a regular @code{n} which has been
4253 placed in a superscript. ISO 10646 defines the @dfn{NFKC}
4254 normalization scheme to convert all these into a standard form as
4255 well, and GCC will warn if your code is not in NFKC if you use
4256 @option{-Wnormalized=nfkc}. This warning is comparable to warning
4257 about every identifier that contains the letter O because it might be
4258 confused with the digit 0, and so is not the default, but may be
4259 useful as a local coding convention if the programming environment is
4260 unable to be fixed to display these characters distinctly.
4262 @item -Wno-deprecated
4263 @opindex Wno-deprecated
4264 @opindex Wdeprecated
4265 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
4267 @item -Wno-deprecated-declarations
4268 @opindex Wno-deprecated-declarations
4269 @opindex Wdeprecated-declarations
4270 Do not warn about uses of functions (@pxref{Function Attributes}),
4271 variables (@pxref{Variable Attributes}), and types (@pxref{Type
4272 Attributes}) marked as deprecated by using the @code{deprecated}
4276 @opindex Wno-overflow
4278 Do not warn about compile-time overflow in constant expressions.
4280 @item -Woverride-init @r{(C and Objective-C only)}
4281 @opindex Woverride-init
4282 @opindex Wno-override-init
4286 Warn if an initialized field without side effects is overridden when
4287 using designated initializers (@pxref{Designated Inits, , Designated
4290 This warning is included in @option{-Wextra}. To get other
4291 @option{-Wextra} warnings without this one, use @samp{-Wextra
4292 -Wno-override-init}.
4297 Warn if a structure is given the packed attribute, but the packed
4298 attribute has no effect on the layout or size of the structure.
4299 Such structures may be mis-aligned for little benefit. For
4300 instance, in this code, the variable @code{f.x} in @code{struct bar}
4301 will be misaligned even though @code{struct bar} does not itself
4302 have the packed attribute:
4309 @} __attribute__((packed));
4317 @item -Wpacked-bitfield-compat
4318 @opindex Wpacked-bitfield-compat
4319 @opindex Wno-packed-bitfield-compat
4320 The 4.1, 4.2 and 4.3 series of GCC ignore the @code{packed} attribute
4321 on bit-fields of type @code{char}. This has been fixed in GCC 4.4 but
4322 the change can lead to differences in the structure layout. GCC
4323 informs you when the offset of such a field has changed in GCC 4.4.
4324 For example there is no longer a 4-bit padding between field @code{a}
4325 and @code{b} in this structure:
4332 @} __attribute__ ((packed));
4335 This warning is enabled by default. Use
4336 @option{-Wno-packed-bitfield-compat} to disable this warning.
4341 Warn if padding is included in a structure, either to align an element
4342 of the structure or to align the whole structure. Sometimes when this
4343 happens it is possible to rearrange the fields of the structure to
4344 reduce the padding and so make the structure smaller.
4346 @item -Wredundant-decls
4347 @opindex Wredundant-decls
4348 @opindex Wno-redundant-decls
4349 Warn if anything is declared more than once in the same scope, even in
4350 cases where multiple declaration is valid and changes nothing.
4352 @item -Wnested-externs @r{(C and Objective-C only)}
4353 @opindex Wnested-externs
4354 @opindex Wno-nested-externs
4355 Warn if an @code{extern} declaration is encountered within a function.
4360 Warn if a function can not be inlined and it was declared as inline.
4361 Even with this option, the compiler will not warn about failures to
4362 inline functions declared in system headers.
4364 The compiler uses a variety of heuristics to determine whether or not
4365 to inline a function. For example, the compiler takes into account
4366 the size of the function being inlined and the amount of inlining
4367 that has already been done in the current function. Therefore,
4368 seemingly insignificant changes in the source program can cause the
4369 warnings produced by @option{-Winline} to appear or disappear.
4371 @item -Wno-invalid-offsetof @r{(C++ and Objective-C++ only)}
4372 @opindex Wno-invalid-offsetof
4373 @opindex Winvalid-offsetof
4374 Suppress warnings from applying the @samp{offsetof} macro to a non-POD
4375 type. According to the 1998 ISO C++ standard, applying @samp{offsetof}
4376 to a non-POD type is undefined. In existing C++ implementations,
4377 however, @samp{offsetof} typically gives meaningful results even when
4378 applied to certain kinds of non-POD types. (Such as a simple
4379 @samp{struct} that fails to be a POD type only by virtue of having a
4380 constructor.) This flag is for users who are aware that they are
4381 writing nonportable code and who have deliberately chosen to ignore the
4384 The restrictions on @samp{offsetof} may be relaxed in a future version
4385 of the C++ standard.
4387 @item -Wno-int-to-pointer-cast
4388 @opindex Wno-int-to-pointer-cast
4389 @opindex Wint-to-pointer-cast
4390 Suppress warnings from casts to pointer type of an integer of a
4391 different size. In C++, casting to a pointer type of smaller size is
4392 an error. @option{Wint-to-pointer-cast} is enabled by default.
4395 @item -Wno-pointer-to-int-cast @r{(C and Objective-C only)}
4396 @opindex Wno-pointer-to-int-cast
4397 @opindex Wpointer-to-int-cast
4398 Suppress warnings from casts from a pointer to an integer type of a
4402 @opindex Winvalid-pch
4403 @opindex Wno-invalid-pch
4404 Warn if a precompiled header (@pxref{Precompiled Headers}) is found in
4405 the search path but can't be used.
4409 @opindex Wno-long-long
4410 Warn if @samp{long long} type is used. This is enabled by either
4411 @option{-pedantic} or @option{-Wtraditional} in ISO C90 and C++98
4412 modes. To inhibit the warning messages, use @option{-Wno-long-long}.
4414 @item -Wvariadic-macros
4415 @opindex Wvariadic-macros
4416 @opindex Wno-variadic-macros
4417 Warn if variadic macros are used in pedantic ISO C90 mode, or the GNU
4418 alternate syntax when in pedantic ISO C99 mode. This is default.
4419 To inhibit the warning messages, use @option{-Wno-variadic-macros}.
4424 Warn if variable length array is used in the code.
4425 @option{-Wno-vla} will prevent the @option{-pedantic} warning of
4426 the variable length array.
4428 @item -Wvolatile-register-var
4429 @opindex Wvolatile-register-var
4430 @opindex Wno-volatile-register-var
4431 Warn if a register variable is declared volatile. The volatile
4432 modifier does not inhibit all optimizations that may eliminate reads
4433 and/or writes to register variables. This warning is enabled by
4436 @item -Wdisabled-optimization
4437 @opindex Wdisabled-optimization
4438 @opindex Wno-disabled-optimization
4439 Warn if a requested optimization pass is disabled. This warning does
4440 not generally indicate that there is anything wrong with your code; it
4441 merely indicates that GCC's optimizers were unable to handle the code
4442 effectively. Often, the problem is that your code is too big or too
4443 complex; GCC will refuse to optimize programs when the optimization
4444 itself is likely to take inordinate amounts of time.
4446 @item -Wpointer-sign @r{(C and Objective-C only)}
4447 @opindex Wpointer-sign
4448 @opindex Wno-pointer-sign
4449 Warn for pointer argument passing or assignment with different signedness.
4450 This option is only supported for C and Objective-C@. It is implied by
4451 @option{-Wall} and by @option{-pedantic}, which can be disabled with
4452 @option{-Wno-pointer-sign}.
4454 @item -Wstack-protector
4455 @opindex Wstack-protector
4456 @opindex Wno-stack-protector
4457 This option is only active when @option{-fstack-protector} is active. It
4458 warns about functions that will not be protected against stack smashing.
4461 @opindex Wno-mudflap
4462 Suppress warnings about constructs that cannot be instrumented by
4465 @item -Woverlength-strings
4466 @opindex Woverlength-strings
4467 @opindex Wno-overlength-strings
4468 Warn about string constants which are longer than the ``minimum
4469 maximum'' length specified in the C standard. Modern compilers
4470 generally allow string constants which are much longer than the
4471 standard's minimum limit, but very portable programs should avoid
4472 using longer strings.
4474 The limit applies @emph{after} string constant concatenation, and does
4475 not count the trailing NUL@. In C90, the limit was 509 characters; in
4476 C99, it was raised to 4095. C++98 does not specify a normative
4477 minimum maximum, so we do not diagnose overlength strings in C++@.
4479 This option is implied by @option{-pedantic}, and can be disabled with
4480 @option{-Wno-overlength-strings}.
4482 @item -Wunsuffixed-float-constants @r{(C and Objective-C only)}
4483 @opindex Wunsuffixed-float-constants
4485 GCC will issue a warning for any floating constant that does not have
4486 a suffix. When used together with @option{-Wsystem-headers} it will
4487 warn about such constants in system header files. This can be useful
4488 when preparing code to use with the @code{FLOAT_CONST_DECIMAL64} pragma
4489 from the decimal floating-point extension to C99.
4492 @node Debugging Options
4493 @section Options for Debugging Your Program or GCC
4494 @cindex options, debugging
4495 @cindex debugging information options
4497 GCC has various special options that are used for debugging
4498 either your program or GCC:
4503 Produce debugging information in the operating system's native format
4504 (stabs, COFF, XCOFF, or DWARF 2)@. GDB can work with this debugging
4507 On most systems that use stabs format, @option{-g} enables use of extra
4508 debugging information that only GDB can use; this extra information
4509 makes debugging work better in GDB but will probably make other debuggers
4511 refuse to read the program. If you want to control for certain whether
4512 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
4513 @option{-gxcoff+}, @option{-gxcoff}, or @option{-gvms} (see below).
4515 GCC allows you to use @option{-g} with
4516 @option{-O}. The shortcuts taken by optimized code may occasionally
4517 produce surprising results: some variables you declared may not exist
4518 at all; flow of control may briefly move where you did not expect it;
4519 some statements may not be executed because they compute constant
4520 results or their values were already at hand; some statements may
4521 execute in different places because they were moved out of loops.
4523 Nevertheless it proves possible to debug optimized output. This makes
4524 it reasonable to use the optimizer for programs that might have bugs.
4526 The following options are useful when GCC is generated with the
4527 capability for more than one debugging format.
4531 Produce debugging information for use by GDB@. This means to use the
4532 most expressive format available (DWARF 2, stabs, or the native format
4533 if neither of those are supported), including GDB extensions if at all
4538 Produce debugging information in stabs format (if that is supported),
4539 without GDB extensions. This is the format used by DBX on most BSD
4540 systems. On MIPS, Alpha and System V Release 4 systems this option
4541 produces stabs debugging output which is not understood by DBX or SDB@.
4542 On System V Release 4 systems this option requires the GNU assembler.
4544 @item -feliminate-unused-debug-symbols
4545 @opindex feliminate-unused-debug-symbols
4546 Produce debugging information in stabs format (if that is supported),
4547 for only symbols that are actually used.
4549 @item -femit-class-debug-always
4550 Instead of emitting debugging information for a C++ class in only one
4551 object file, emit it in all object files using the class. This option
4552 should be used only with debuggers that are unable to handle the way GCC
4553 normally emits debugging information for classes because using this
4554 option will increase the size of debugging information by as much as a
4559 Produce debugging information in stabs format (if that is supported),
4560 using GNU extensions understood only by the GNU debugger (GDB)@. The
4561 use of these extensions is likely to make other debuggers crash or
4562 refuse to read the program.
4566 Produce debugging information in COFF format (if that is supported).
4567 This is the format used by SDB on most System V systems prior to
4572 Produce debugging information in XCOFF format (if that is supported).
4573 This is the format used by the DBX debugger on IBM RS/6000 systems.
4577 Produce debugging information in XCOFF format (if that is supported),
4578 using GNU extensions understood only by the GNU debugger (GDB)@. The
4579 use of these extensions is likely to make other debuggers crash or
4580 refuse to read the program, and may cause assemblers other than the GNU
4581 assembler (GAS) to fail with an error.
4583 @item -gdwarf-@var{version}
4584 @opindex gdwarf-@var{version}
4585 Produce debugging information in DWARF format (if that is
4586 supported). This is the format used by DBX on IRIX 6. The value
4587 of @var{version} may be either 2, 3 or 4; the default version is 2.
4589 Note that with DWARF version 2 some ports require, and will always
4590 use, some non-conflicting DWARF 3 extensions in the unwind tables.
4592 Version 4 may require GDB 7.0 and @option{-fvar-tracking-assignments}
4593 for maximum benefit.
4595 @item -gstrict-dwarf
4596 @opindex gstrict-dwarf
4597 Disallow using extensions of later DWARF standard version than selected
4598 with @option{-gdwarf-@var{version}}. On most targets using non-conflicting
4599 DWARF extensions from later standard versions is allowed.
4601 @item -gno-strict-dwarf
4602 @opindex gno-strict-dwarf
4603 Allow using extensions of later DWARF standard version than selected with
4604 @option{-gdwarf-@var{version}}.
4608 Produce debugging information in VMS debug format (if that is
4609 supported). This is the format used by DEBUG on VMS systems.
4612 @itemx -ggdb@var{level}
4613 @itemx -gstabs@var{level}
4614 @itemx -gcoff@var{level}
4615 @itemx -gxcoff@var{level}
4616 @itemx -gvms@var{level}
4617 Request debugging information and also use @var{level} to specify how
4618 much information. The default level is 2.
4620 Level 0 produces no debug information at all. Thus, @option{-g0} negates
4623 Level 1 produces minimal information, enough for making backtraces in
4624 parts of the program that you don't plan to debug. This includes
4625 descriptions of functions and external variables, but no information
4626 about local variables and no line numbers.
4628 Level 3 includes extra information, such as all the macro definitions
4629 present in the program. Some debuggers support macro expansion when
4630 you use @option{-g3}.
4632 @option{-gdwarf-2} does not accept a concatenated debug level, because
4633 GCC used to support an option @option{-gdwarf} that meant to generate
4634 debug information in version 1 of the DWARF format (which is very
4635 different from version 2), and it would have been too confusing. That
4636 debug format is long obsolete, but the option cannot be changed now.
4637 Instead use an additional @option{-g@var{level}} option to change the
4638 debug level for DWARF.
4642 Turn off generation of debug info, if leaving out this option would have
4643 generated it, or turn it on at level 2 otherwise. The position of this
4644 argument in the command line does not matter, it takes effect after all
4645 other options are processed, and it does so only once, no matter how
4646 many times it is given. This is mainly intended to be used with
4647 @option{-fcompare-debug}.
4649 @item -fdump-final-insns@r{[}=@var{file}@r{]}
4650 @opindex fdump-final-insns
4651 Dump the final internal representation (RTL) to @var{file}. If the
4652 optional argument is omitted (or if @var{file} is @code{.}), the name
4653 of the dump file will be determined by appending @code{.gkd} to the
4654 compilation output file name.
4656 @item -fcompare-debug@r{[}=@var{opts}@r{]}
4657 @opindex fcompare-debug
4658 @opindex fno-compare-debug
4659 If no error occurs during compilation, run the compiler a second time,
4660 adding @var{opts} and @option{-fcompare-debug-second} to the arguments
4661 passed to the second compilation. Dump the final internal
4662 representation in both compilations, and print an error if they differ.
4664 If the equal sign is omitted, the default @option{-gtoggle} is used.
4666 The environment variable @env{GCC_COMPARE_DEBUG}, if defined, non-empty
4667 and nonzero, implicitly enables @option{-fcompare-debug}. If
4668 @env{GCC_COMPARE_DEBUG} is defined to a string starting with a dash,
4669 then it is used for @var{opts}, otherwise the default @option{-gtoggle}
4672 @option{-fcompare-debug=}, with the equal sign but without @var{opts},
4673 is equivalent to @option{-fno-compare-debug}, which disables the dumping
4674 of the final representation and the second compilation, preventing even
4675 @env{GCC_COMPARE_DEBUG} from taking effect.
4677 To verify full coverage during @option{-fcompare-debug} testing, set
4678 @env{GCC_COMPARE_DEBUG} to say @samp{-fcompare-debug-not-overridden},
4679 which GCC will reject as an invalid option in any actual compilation
4680 (rather than preprocessing, assembly or linking). To get just a
4681 warning, setting @env{GCC_COMPARE_DEBUG} to @samp{-w%n-fcompare-debug
4682 not overridden} will do.
4684 @item -fcompare-debug-second
4685 @opindex fcompare-debug-second
4686 This option is implicitly passed to the compiler for the second
4687 compilation requested by @option{-fcompare-debug}, along with options to
4688 silence warnings, and omitting other options that would cause
4689 side-effect compiler outputs to files or to the standard output. Dump
4690 files and preserved temporary files are renamed so as to contain the
4691 @code{.gk} additional extension during the second compilation, to avoid
4692 overwriting those generated by the first.
4694 When this option is passed to the compiler driver, it causes the
4695 @emph{first} compilation to be skipped, which makes it useful for little
4696 other than debugging the compiler proper.
4698 @item -feliminate-dwarf2-dups
4699 @opindex feliminate-dwarf2-dups
4700 Compress DWARF2 debugging information by eliminating duplicated
4701 information about each symbol. This option only makes sense when
4702 generating DWARF2 debugging information with @option{-gdwarf-2}.
4704 @item -femit-struct-debug-baseonly
4705 Emit debug information for struct-like types
4706 only when the base name of the compilation source file
4707 matches the base name of file in which the struct was defined.
4709 This option substantially reduces the size of debugging information,
4710 but at significant potential loss in type information to the debugger.
4711 See @option{-femit-struct-debug-reduced} for a less aggressive option.
4712 See @option{-femit-struct-debug-detailed} for more detailed control.
4714 This option works only with DWARF 2.
4716 @item -femit-struct-debug-reduced
4717 Emit debug information for struct-like types
4718 only when the base name of the compilation source file
4719 matches the base name of file in which the type was defined,
4720 unless the struct is a template or defined in a system header.
4722 This option significantly reduces the size of debugging information,
4723 with some potential loss in type information to the debugger.
4724 See @option{-femit-struct-debug-baseonly} for a more aggressive option.
4725 See @option{-femit-struct-debug-detailed} for more detailed control.
4727 This option works only with DWARF 2.
4729 @item -femit-struct-debug-detailed@r{[}=@var{spec-list}@r{]}
4730 Specify the struct-like types
4731 for which the compiler will generate debug information.
4732 The intent is to reduce duplicate struct debug information
4733 between different object files within the same program.
4735 This option is a detailed version of
4736 @option{-femit-struct-debug-reduced} and @option{-femit-struct-debug-baseonly},
4737 which will serve for most needs.
4739 A specification has the syntax
4740 [@samp{dir:}|@samp{ind:}][@samp{ord:}|@samp{gen:}](@samp{any}|@samp{sys}|@samp{base}|@samp{none})
4742 The optional first word limits the specification to
4743 structs that are used directly (@samp{dir:}) or used indirectly (@samp{ind:}).
4744 A struct type is used directly when it is the type of a variable, member.
4745 Indirect uses arise through pointers to structs.
4746 That is, when use of an incomplete struct would be legal, the use is indirect.
4748 @samp{struct one direct; struct two * indirect;}.
4750 The optional second word limits the specification to
4751 ordinary structs (@samp{ord:}) or generic structs (@samp{gen:}).
4752 Generic structs are a bit complicated to explain.
4753 For C++, these are non-explicit specializations of template classes,
4754 or non-template classes within the above.
4755 Other programming languages have generics,
4756 but @samp{-femit-struct-debug-detailed} does not yet implement them.
4758 The third word specifies the source files for those
4759 structs for which the compiler will emit debug information.
4760 The values @samp{none} and @samp{any} have the normal meaning.
4761 The value @samp{base} means that
4762 the base of name of the file in which the type declaration appears
4763 must match the base of the name of the main compilation file.
4764 In practice, this means that
4765 types declared in @file{foo.c} and @file{foo.h} will have debug information,
4766 but types declared in other header will not.
4767 The value @samp{sys} means those types satisfying @samp{base}
4768 or declared in system or compiler headers.
4770 You may need to experiment to determine the best settings for your application.
4772 The default is @samp{-femit-struct-debug-detailed=all}.
4774 This option works only with DWARF 2.
4776 @item -fenable-icf-debug
4777 @opindex fenable-icf-debug
4778 Generate additional debug information to support identical code folding (ICF).
4779 This option only works with DWARF version 2 or higher.
4781 @item -fno-merge-debug-strings
4782 @opindex fmerge-debug-strings
4783 @opindex fno-merge-debug-strings
4784 Direct the linker to not merge together strings in the debugging
4785 information which are identical in different object files. Merging is
4786 not supported by all assemblers or linkers. Merging decreases the size
4787 of the debug information in the output file at the cost of increasing
4788 link processing time. Merging is enabled by default.
4790 @item -fdebug-prefix-map=@var{old}=@var{new}
4791 @opindex fdebug-prefix-map
4792 When compiling files in directory @file{@var{old}}, record debugging
4793 information describing them as in @file{@var{new}} instead.
4795 @item -fno-dwarf2-cfi-asm
4796 @opindex fdwarf2-cfi-asm
4797 @opindex fno-dwarf2-cfi-asm
4798 Emit DWARF 2 unwind info as compiler generated @code{.eh_frame} section
4799 instead of using GAS @code{.cfi_*} directives.
4801 @cindex @command{prof}
4804 Generate extra code to write profile information suitable for the
4805 analysis program @command{prof}. You must use this option when compiling
4806 the source files you want data about, and you must also use it when
4809 @cindex @command{gprof}
4812 Generate extra code to write profile information suitable for the
4813 analysis program @command{gprof}. You must use this option when compiling
4814 the source files you want data about, and you must also use it when
4819 Makes the compiler print out each function name as it is compiled, and
4820 print some statistics about each pass when it finishes.
4823 @opindex ftime-report
4824 Makes the compiler print some statistics about the time consumed by each
4825 pass when it finishes.
4828 @opindex fmem-report
4829 Makes the compiler print some statistics about permanent memory
4830 allocation when it finishes.
4832 @item -fpre-ipa-mem-report
4833 @opindex fpre-ipa-mem-report
4834 @item -fpost-ipa-mem-report
4835 @opindex fpost-ipa-mem-report
4836 Makes the compiler print some statistics about permanent memory
4837 allocation before or after interprocedural optimization.
4839 @item -fprofile-arcs
4840 @opindex fprofile-arcs
4841 Add code so that program flow @dfn{arcs} are instrumented. During
4842 execution the program records how many times each branch and call is
4843 executed and how many times it is taken or returns. When the compiled
4844 program exits it saves this data to a file called
4845 @file{@var{auxname}.gcda} for each source file. The data may be used for
4846 profile-directed optimizations (@option{-fbranch-probabilities}), or for
4847 test coverage analysis (@option{-ftest-coverage}). Each object file's
4848 @var{auxname} is generated from the name of the output file, if
4849 explicitly specified and it is not the final executable, otherwise it is
4850 the basename of the source file. In both cases any suffix is removed
4851 (e.g.@: @file{foo.gcda} for input file @file{dir/foo.c}, or
4852 @file{dir/foo.gcda} for output file specified as @option{-o dir/foo.o}).
4853 @xref{Cross-profiling}.
4855 @cindex @command{gcov}
4859 This option is used to compile and link code instrumented for coverage
4860 analysis. The option is a synonym for @option{-fprofile-arcs}
4861 @option{-ftest-coverage} (when compiling) and @option{-lgcov} (when
4862 linking). See the documentation for those options for more details.
4867 Compile the source files with @option{-fprofile-arcs} plus optimization
4868 and code generation options. For test coverage analysis, use the
4869 additional @option{-ftest-coverage} option. You do not need to profile
4870 every source file in a program.
4873 Link your object files with @option{-lgcov} or @option{-fprofile-arcs}
4874 (the latter implies the former).
4877 Run the program on a representative workload to generate the arc profile
4878 information. This may be repeated any number of times. You can run
4879 concurrent instances of your program, and provided that the file system
4880 supports locking, the data files will be correctly updated. Also
4881 @code{fork} calls are detected and correctly handled (double counting
4885 For profile-directed optimizations, compile the source files again with
4886 the same optimization and code generation options plus
4887 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
4888 Control Optimization}).
4891 For test coverage analysis, use @command{gcov} to produce human readable
4892 information from the @file{.gcno} and @file{.gcda} files. Refer to the
4893 @command{gcov} documentation for further information.
4897 With @option{-fprofile-arcs}, for each function of your program GCC
4898 creates a program flow graph, then finds a spanning tree for the graph.
4899 Only arcs that are not on the spanning tree have to be instrumented: the
4900 compiler adds code to count the number of times that these arcs are
4901 executed. When an arc is the only exit or only entrance to a block, the
4902 instrumentation code can be added to the block; otherwise, a new basic
4903 block must be created to hold the instrumentation code.
4906 @item -ftest-coverage
4907 @opindex ftest-coverage
4908 Produce a notes file that the @command{gcov} code-coverage utility
4909 (@pxref{Gcov,, @command{gcov}---a Test Coverage Program}) can use to
4910 show program coverage. Each source file's note file is called
4911 @file{@var{auxname}.gcno}. Refer to the @option{-fprofile-arcs} option
4912 above for a description of @var{auxname} and instructions on how to
4913 generate test coverage data. Coverage data will match the source files
4914 more closely, if you do not optimize.
4916 @item -fdbg-cnt-list
4917 @opindex fdbg-cnt-list
4918 Print the name and the counter upperbound for all debug counters.
4920 @item -fdbg-cnt=@var{counter-value-list}
4922 Set the internal debug counter upperbound. @var{counter-value-list}
4923 is a comma-separated list of @var{name}:@var{value} pairs
4924 which sets the upperbound of each debug counter @var{name} to @var{value}.
4925 All debug counters have the initial upperbound of @var{UINT_MAX},
4926 thus dbg_cnt() returns true always unless the upperbound is set by this option.
4927 e.g. With -fdbg-cnt=dce:10,tail_call:0
4928 dbg_cnt(dce) will return true only for first 10 invocations
4929 and dbg_cnt(tail_call) will return false always.
4931 @item -d@var{letters}
4932 @itemx -fdump-rtl-@var{pass}
4934 Says to make debugging dumps during compilation at times specified by
4935 @var{letters}. This is used for debugging the RTL-based passes of the
4936 compiler. The file names for most of the dumps are made by appending
4937 a pass number and a word to the @var{dumpname}, and the files are
4938 created in the directory of the output file. @var{dumpname} is
4939 generated from the name of the output file, if explicitly specified
4940 and it is not an executable, otherwise it is the basename of the
4941 source file. These switches may have different effects when
4942 @option{-E} is used for preprocessing.
4944 Debug dumps can be enabled with a @option{-fdump-rtl} switch or some
4945 @option{-d} option @var{letters}. Here are the possible
4946 letters for use in @var{pass} and @var{letters}, and their meanings:
4950 @item -fdump-rtl-alignments
4951 @opindex fdump-rtl-alignments
4952 Dump after branch alignments have been computed.
4954 @item -fdump-rtl-asmcons
4955 @opindex fdump-rtl-asmcons
4956 Dump after fixing rtl statements that have unsatisfied in/out constraints.
4958 @item -fdump-rtl-auto_inc_dec
4959 @opindex fdump-rtl-auto_inc_dec
4960 Dump after auto-inc-dec discovery. This pass is only run on
4961 architectures that have auto inc or auto dec instructions.
4963 @item -fdump-rtl-barriers
4964 @opindex fdump-rtl-barriers
4965 Dump after cleaning up the barrier instructions.
4967 @item -fdump-rtl-bbpart
4968 @opindex fdump-rtl-bbpart
4969 Dump after partitioning hot and cold basic blocks.
4971 @item -fdump-rtl-bbro
4972 @opindex fdump-rtl-bbro
4973 Dump after block reordering.
4975 @item -fdump-rtl-btl1
4976 @itemx -fdump-rtl-btl2
4977 @opindex fdump-rtl-btl2
4978 @opindex fdump-rtl-btl2
4979 @option{-fdump-rtl-btl1} and @option{-fdump-rtl-btl2} enable dumping
4980 after the two branch
4981 target load optimization passes.
4983 @item -fdump-rtl-bypass
4984 @opindex fdump-rtl-bypass
4985 Dump after jump bypassing and control flow optimizations.
4987 @item -fdump-rtl-combine
4988 @opindex fdump-rtl-combine
4989 Dump after the RTL instruction combination pass.
4991 @item -fdump-rtl-compgotos
4992 @opindex fdump-rtl-compgotos
4993 Dump after duplicating the computed gotos.
4995 @item -fdump-rtl-ce1
4996 @itemx -fdump-rtl-ce2
4997 @itemx -fdump-rtl-ce3
4998 @opindex fdump-rtl-ce1
4999 @opindex fdump-rtl-ce2
5000 @opindex fdump-rtl-ce3
5001 @option{-fdump-rtl-ce1}, @option{-fdump-rtl-ce2}, and
5002 @option{-fdump-rtl-ce3} enable dumping after the three
5003 if conversion passes.
5005 @itemx -fdump-rtl-cprop_hardreg
5006 @opindex fdump-rtl-cprop_hardreg
5007 Dump after hard register copy propagation.
5009 @itemx -fdump-rtl-csa
5010 @opindex fdump-rtl-csa
5011 Dump after combining stack adjustments.
5013 @item -fdump-rtl-cse1
5014 @itemx -fdump-rtl-cse2
5015 @opindex fdump-rtl-cse1
5016 @opindex fdump-rtl-cse2
5017 @option{-fdump-rtl-cse1} and @option{-fdump-rtl-cse2} enable dumping after
5018 the two common sub-expression elimination passes.
5020 @itemx -fdump-rtl-dce
5021 @opindex fdump-rtl-dce
5022 Dump after the standalone dead code elimination passes.
5024 @itemx -fdump-rtl-dbr
5025 @opindex fdump-rtl-dbr
5026 Dump after delayed branch scheduling.
5028 @item -fdump-rtl-dce1
5029 @itemx -fdump-rtl-dce2
5030 @opindex fdump-rtl-dce1
5031 @opindex fdump-rtl-dce2
5032 @option{-fdump-rtl-dce1} and @option{-fdump-rtl-dce2} enable dumping after
5033 the two dead store elimination passes.
5036 @opindex fdump-rtl-eh
5037 Dump after finalization of EH handling code.
5039 @item -fdump-rtl-eh_ranges
5040 @opindex fdump-rtl-eh_ranges
5041 Dump after conversion of EH handling range regions.
5043 @item -fdump-rtl-expand
5044 @opindex fdump-rtl-expand
5045 Dump after RTL generation.
5047 @item -fdump-rtl-fwprop1
5048 @itemx -fdump-rtl-fwprop2
5049 @opindex fdump-rtl-fwprop1
5050 @opindex fdump-rtl-fwprop2
5051 @option{-fdump-rtl-fwprop1} and @option{-fdump-rtl-fwprop2} enable
5052 dumping after the two forward propagation passes.
5054 @item -fdump-rtl-gcse1
5055 @itemx -fdump-rtl-gcse2
5056 @opindex fdump-rtl-gcse1
5057 @opindex fdump-rtl-gcse2
5058 @option{-fdump-rtl-gcse1} and @option{-fdump-rtl-gcse2} enable dumping
5059 after global common subexpression elimination.
5061 @item -fdump-rtl-init-regs
5062 @opindex fdump-rtl-init-regs
5063 Dump after the initialization of the registers.
5065 @item -fdump-rtl-initvals
5066 @opindex fdump-rtl-initvals
5067 Dump after the computation of the initial value sets.
5069 @itemx -fdump-rtl-into_cfglayout
5070 @opindex fdump-rtl-into_cfglayout
5071 Dump after converting to cfglayout mode.
5073 @item -fdump-rtl-ira
5074 @opindex fdump-rtl-ira
5075 Dump after iterated register allocation.
5077 @item -fdump-rtl-jump
5078 @opindex fdump-rtl-jump
5079 Dump after the second jump optimization.
5081 @item -fdump-rtl-loop2
5082 @opindex fdump-rtl-loop2
5083 @option{-fdump-rtl-loop2} enables dumping after the rtl
5084 loop optimization passes.
5086 @item -fdump-rtl-mach
5087 @opindex fdump-rtl-mach
5088 Dump after performing the machine dependent reorganization pass, if that
5091 @item -fdump-rtl-mode_sw
5092 @opindex fdump-rtl-mode_sw
5093 Dump after removing redundant mode switches.
5095 @item -fdump-rtl-rnreg
5096 @opindex fdump-rtl-rnreg
5097 Dump after register renumbering.
5099 @itemx -fdump-rtl-outof_cfglayout
5100 @opindex fdump-rtl-outof_cfglayout
5101 Dump after converting from cfglayout mode.
5103 @item -fdump-rtl-peephole2
5104 @opindex fdump-rtl-peephole2
5105 Dump after the peephole pass.
5107 @item -fdump-rtl-postreload
5108 @opindex fdump-rtl-postreload
5109 Dump after post-reload optimizations.
5111 @itemx -fdump-rtl-pro_and_epilogue
5112 @opindex fdump-rtl-pro_and_epilogue
5113 Dump after generating the function pro and epilogues.
5115 @item -fdump-rtl-regmove
5116 @opindex fdump-rtl-regmove
5117 Dump after the register move pass.
5119 @item -fdump-rtl-sched1
5120 @itemx -fdump-rtl-sched2
5121 @opindex fdump-rtl-sched1
5122 @opindex fdump-rtl-sched2
5123 @option{-fdump-rtl-sched1} and @option{-fdump-rtl-sched2} enable dumping
5124 after the basic block scheduling passes.
5126 @item -fdump-rtl-see
5127 @opindex fdump-rtl-see
5128 Dump after sign extension elimination.
5130 @item -fdump-rtl-seqabstr
5131 @opindex fdump-rtl-seqabstr
5132 Dump after common sequence discovery.
5134 @item -fdump-rtl-shorten
5135 @opindex fdump-rtl-shorten
5136 Dump after shortening branches.
5138 @item -fdump-rtl-sibling
5139 @opindex fdump-rtl-sibling
5140 Dump after sibling call optimizations.
5142 @item -fdump-rtl-split1
5143 @itemx -fdump-rtl-split2
5144 @itemx -fdump-rtl-split3
5145 @itemx -fdump-rtl-split4
5146 @itemx -fdump-rtl-split5
5147 @opindex fdump-rtl-split1
5148 @opindex fdump-rtl-split2
5149 @opindex fdump-rtl-split3
5150 @opindex fdump-rtl-split4
5151 @opindex fdump-rtl-split5
5152 @option{-fdump-rtl-split1}, @option{-fdump-rtl-split2},
5153 @option{-fdump-rtl-split3}, @option{-fdump-rtl-split4} and
5154 @option{-fdump-rtl-split5} enable dumping after five rounds of
5155 instruction splitting.
5157 @item -fdump-rtl-sms
5158 @opindex fdump-rtl-sms
5159 Dump after modulo scheduling. This pass is only run on some
5162 @item -fdump-rtl-stack
5163 @opindex fdump-rtl-stack
5164 Dump after conversion from GCC's "flat register file" registers to the
5165 x87's stack-like registers. This pass is only run on x86 variants.
5167 @item -fdump-rtl-subreg1
5168 @itemx -fdump-rtl-subreg2
5169 @opindex fdump-rtl-subreg1
5170 @opindex fdump-rtl-subreg2
5171 @option{-fdump-rtl-subreg1} and @option{-fdump-rtl-subreg2} enable dumping after
5172 the two subreg expansion passes.
5174 @item -fdump-rtl-unshare
5175 @opindex fdump-rtl-unshare
5176 Dump after all rtl has been unshared.
5178 @item -fdump-rtl-vartrack
5179 @opindex fdump-rtl-vartrack
5180 Dump after variable tracking.
5182 @item -fdump-rtl-vregs
5183 @opindex fdump-rtl-vregs
5184 Dump after converting virtual registers to hard registers.
5186 @item -fdump-rtl-web
5187 @opindex fdump-rtl-web
5188 Dump after live range splitting.
5190 @item -fdump-rtl-regclass
5191 @itemx -fdump-rtl-subregs_of_mode_init
5192 @itemx -fdump-rtl-subregs_of_mode_finish
5193 @itemx -fdump-rtl-dfinit
5194 @itemx -fdump-rtl-dfinish
5195 @opindex fdump-rtl-regclass
5196 @opindex fdump-rtl-subregs_of_mode_init
5197 @opindex fdump-rtl-subregs_of_mode_finish
5198 @opindex fdump-rtl-dfinit
5199 @opindex fdump-rtl-dfinish
5200 These dumps are defined but always produce empty files.
5202 @item -fdump-rtl-all
5203 @opindex fdump-rtl-all
5204 Produce all the dumps listed above.
5208 Annotate the assembler output with miscellaneous debugging information.
5212 Dump all macro definitions, at the end of preprocessing, in addition to
5217 Produce a core dump whenever an error occurs.
5221 Print statistics on memory usage, at the end of the run, to
5226 Annotate the assembler output with a comment indicating which
5227 pattern and alternative was used. The length of each instruction is
5232 Dump the RTL in the assembler output as a comment before each instruction.
5233 Also turns on @option{-dp} annotation.
5237 For each of the other indicated dump files (@option{-fdump-rtl-@var{pass}}),
5238 dump a representation of the control flow graph suitable for viewing with VCG
5239 to @file{@var{file}.@var{pass}.vcg}.
5243 Just generate RTL for a function instead of compiling it. Usually used
5244 with @option{-fdump-rtl-expand}.
5248 Dump debugging information during parsing, to standard error.
5252 @opindex fdump-noaddr
5253 When doing debugging dumps, suppress address output. This makes it more
5254 feasible to use diff on debugging dumps for compiler invocations with
5255 different compiler binaries and/or different
5256 text / bss / data / heap / stack / dso start locations.
5258 @item -fdump-unnumbered
5259 @opindex fdump-unnumbered
5260 When doing debugging dumps, suppress instruction numbers and address output.
5261 This makes it more feasible to use diff on debugging dumps for compiler
5262 invocations with different options, in particular with and without
5265 @item -fdump-unnumbered-links
5266 @opindex fdump-unnumbered-links
5267 When doing debugging dumps (see @option{-d} option above), suppress
5268 instruction numbers for the links to the previous and next instructions
5271 @item -fdump-translation-unit @r{(C++ only)}
5272 @itemx -fdump-translation-unit-@var{options} @r{(C++ only)}
5273 @opindex fdump-translation-unit
5274 Dump a representation of the tree structure for the entire translation
5275 unit to a file. The file name is made by appending @file{.tu} to the
5276 source file name, and the file is created in the same directory as the
5277 output file. If the @samp{-@var{options}} form is used, @var{options}
5278 controls the details of the dump as described for the
5279 @option{-fdump-tree} options.
5281 @item -fdump-class-hierarchy @r{(C++ only)}
5282 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
5283 @opindex fdump-class-hierarchy
5284 Dump a representation of each class's hierarchy and virtual function
5285 table layout to a file. The file name is made by appending
5286 @file{.class} to the source file name, and the file is created in the
5287 same directory as the output file. If the @samp{-@var{options}} form
5288 is used, @var{options} controls the details of the dump as described
5289 for the @option{-fdump-tree} options.
5291 @item -fdump-ipa-@var{switch}
5293 Control the dumping at various stages of inter-procedural analysis
5294 language tree to a file. The file name is generated by appending a
5295 switch specific suffix to the source file name, and the file is created
5296 in the same directory as the output file. The following dumps are
5301 Enables all inter-procedural analysis dumps.
5304 Dumps information about call-graph optimization, unused function removal,
5305 and inlining decisions.
5308 Dump after function inlining.
5312 @item -fdump-statistics-@var{option}
5313 @opindex fdump-statistics
5314 Enable and control dumping of pass statistics in a separate file. The
5315 file name is generated by appending a suffix ending in
5316 @samp{.statistics} to the source file name, and the file is created in
5317 the same directory as the output file. If the @samp{-@var{option}}
5318 form is used, @samp{-stats} will cause counters to be summed over the
5319 whole compilation unit while @samp{-details} will dump every event as
5320 the passes generate them. The default with no option is to sum
5321 counters for each function compiled.
5323 @item -fdump-tree-@var{switch}
5324 @itemx -fdump-tree-@var{switch}-@var{options}
5326 Control the dumping at various stages of processing the intermediate
5327 language tree to a file. The file name is generated by appending a
5328 switch specific suffix to the source file name, and the file is
5329 created in the same directory as the output file. If the
5330 @samp{-@var{options}} form is used, @var{options} is a list of
5331 @samp{-} separated options that control the details of the dump. Not
5332 all options are applicable to all dumps, those which are not
5333 meaningful will be ignored. The following options are available
5337 Print the address of each node. Usually this is not meaningful as it
5338 changes according to the environment and source file. Its primary use
5339 is for tying up a dump file with a debug environment.
5341 If @code{DECL_ASSEMBLER_NAME} has been set for a given decl, use that
5342 in the dump instead of @code{DECL_NAME}. Its primary use is ease of
5343 use working backward from mangled names in the assembly file.
5345 Inhibit dumping of members of a scope or body of a function merely
5346 because that scope has been reached. Only dump such items when they
5347 are directly reachable by some other path. When dumping pretty-printed
5348 trees, this option inhibits dumping the bodies of control structures.
5350 Print a raw representation of the tree. By default, trees are
5351 pretty-printed into a C-like representation.
5353 Enable more detailed dumps (not honored by every dump option).
5355 Enable dumping various statistics about the pass (not honored by every dump
5358 Enable showing basic block boundaries (disabled in raw dumps).
5360 Enable showing virtual operands for every statement.
5362 Enable showing line numbers for statements.
5364 Enable showing the unique ID (@code{DECL_UID}) for each variable.
5366 Enable showing the tree dump for each statement.
5368 Enable showing the EH region number holding each statement.
5370 Turn on all options, except @option{raw}, @option{slim}, @option{verbose}
5371 and @option{lineno}.
5374 The following tree dumps are possible:
5378 @opindex fdump-tree-original
5379 Dump before any tree based optimization, to @file{@var{file}.original}.
5382 @opindex fdump-tree-optimized
5383 Dump after all tree based optimization, to @file{@var{file}.optimized}.
5386 @opindex fdump-tree-gimple
5387 Dump each function before and after the gimplification pass to a file. The
5388 file name is made by appending @file{.gimple} to the source file name.
5391 @opindex fdump-tree-cfg
5392 Dump the control flow graph of each function to a file. The file name is
5393 made by appending @file{.cfg} to the source file name.
5396 @opindex fdump-tree-vcg
5397 Dump the control flow graph of each function to a file in VCG format. The
5398 file name is made by appending @file{.vcg} to the source file name. Note
5399 that if the file contains more than one function, the generated file cannot
5400 be used directly by VCG@. You will need to cut and paste each function's
5401 graph into its own separate file first.
5404 @opindex fdump-tree-ch
5405 Dump each function after copying loop headers. The file name is made by
5406 appending @file{.ch} to the source file name.
5409 @opindex fdump-tree-ssa
5410 Dump SSA related information to a file. The file name is made by appending
5411 @file{.ssa} to the source file name.
5414 @opindex fdump-tree-alias
5415 Dump aliasing information for each function. The file name is made by
5416 appending @file{.alias} to the source file name.
5419 @opindex fdump-tree-ccp
5420 Dump each function after CCP@. The file name is made by appending
5421 @file{.ccp} to the source file name.
5424 @opindex fdump-tree-storeccp
5425 Dump each function after STORE-CCP@. The file name is made by appending
5426 @file{.storeccp} to the source file name.
5429 @opindex fdump-tree-pre
5430 Dump trees after partial redundancy elimination. The file name is made
5431 by appending @file{.pre} to the source file name.
5434 @opindex fdump-tree-fre
5435 Dump trees after full redundancy elimination. The file name is made
5436 by appending @file{.fre} to the source file name.
5439 @opindex fdump-tree-copyprop
5440 Dump trees after copy propagation. The file name is made
5441 by appending @file{.copyprop} to the source file name.
5443 @item store_copyprop
5444 @opindex fdump-tree-store_copyprop
5445 Dump trees after store copy-propagation. The file name is made
5446 by appending @file{.store_copyprop} to the source file name.
5449 @opindex fdump-tree-dce
5450 Dump each function after dead code elimination. The file name is made by
5451 appending @file{.dce} to the source file name.
5454 @opindex fdump-tree-mudflap
5455 Dump each function after adding mudflap instrumentation. The file name is
5456 made by appending @file{.mudflap} to the source file name.
5459 @opindex fdump-tree-sra
5460 Dump each function after performing scalar replacement of aggregates. The
5461 file name is made by appending @file{.sra} to the source file name.
5464 @opindex fdump-tree-sink
5465 Dump each function after performing code sinking. The file name is made
5466 by appending @file{.sink} to the source file name.
5469 @opindex fdump-tree-dom
5470 Dump each function after applying dominator tree optimizations. The file
5471 name is made by appending @file{.dom} to the source file name.
5474 @opindex fdump-tree-dse
5475 Dump each function after applying dead store elimination. The file
5476 name is made by appending @file{.dse} to the source file name.
5479 @opindex fdump-tree-phiopt
5480 Dump each function after optimizing PHI nodes into straightline code. The file
5481 name is made by appending @file{.phiopt} to the source file name.
5484 @opindex fdump-tree-forwprop
5485 Dump each function after forward propagating single use variables. The file
5486 name is made by appending @file{.forwprop} to the source file name.
5489 @opindex fdump-tree-copyrename
5490 Dump each function after applying the copy rename optimization. The file
5491 name is made by appending @file{.copyrename} to the source file name.
5494 @opindex fdump-tree-nrv
5495 Dump each function after applying the named return value optimization on
5496 generic trees. The file name is made by appending @file{.nrv} to the source
5500 @opindex fdump-tree-vect
5501 Dump each function after applying vectorization of loops. The file name is
5502 made by appending @file{.vect} to the source file name.
5505 @opindex fdump-tree-slp
5506 Dump each function after applying vectorization of basic blocks. The file name
5507 is made by appending @file{.slp} to the source file name.
5510 @opindex fdump-tree-vrp
5511 Dump each function after Value Range Propagation (VRP). The file name
5512 is made by appending @file{.vrp} to the source file name.
5515 @opindex fdump-tree-all
5516 Enable all the available tree dumps with the flags provided in this option.
5519 @item -ftree-vectorizer-verbose=@var{n}
5520 @opindex ftree-vectorizer-verbose
5521 This option controls the amount of debugging output the vectorizer prints.
5522 This information is written to standard error, unless
5523 @option{-fdump-tree-all} or @option{-fdump-tree-vect} is specified,
5524 in which case it is output to the usual dump listing file, @file{.vect}.
5525 For @var{n}=0 no diagnostic information is reported.
5526 If @var{n}=1 the vectorizer reports each loop that got vectorized,
5527 and the total number of loops that got vectorized.
5528 If @var{n}=2 the vectorizer also reports non-vectorized loops that passed
5529 the first analysis phase (vect_analyze_loop_form) - i.e.@: countable,
5530 inner-most, single-bb, single-entry/exit loops. This is the same verbosity
5531 level that @option{-fdump-tree-vect-stats} uses.
5532 Higher verbosity levels mean either more information dumped for each
5533 reported loop, or same amount of information reported for more loops:
5534 if @var{n}=3, vectorizer cost model information is reported.
5535 If @var{n}=4, alignment related information is added to the reports.
5536 If @var{n}=5, data-references related information (e.g.@: memory dependences,
5537 memory access-patterns) is added to the reports.
5538 If @var{n}=6, the vectorizer reports also non-vectorized inner-most loops
5539 that did not pass the first analysis phase (i.e., may not be countable, or
5540 may have complicated control-flow).
5541 If @var{n}=7, the vectorizer reports also non-vectorized nested loops.
5542 If @var{n}=8, SLP related information is added to the reports.
5543 For @var{n}=9, all the information the vectorizer generates during its
5544 analysis and transformation is reported. This is the same verbosity level
5545 that @option{-fdump-tree-vect-details} uses.
5547 @item -frandom-seed=@var{string}
5548 @opindex frandom-seed
5549 This option provides a seed that GCC uses when it would otherwise use
5550 random numbers. It is used to generate certain symbol names
5551 that have to be different in every compiled file. It is also used to
5552 place unique stamps in coverage data files and the object files that
5553 produce them. You can use the @option{-frandom-seed} option to produce
5554 reproducibly identical object files.
5556 The @var{string} should be different for every file you compile.
5558 @item -fsched-verbose=@var{n}
5559 @opindex fsched-verbose
5560 On targets that use instruction scheduling, this option controls the
5561 amount of debugging output the scheduler prints. This information is
5562 written to standard error, unless @option{-fdump-rtl-sched1} or
5563 @option{-fdump-rtl-sched2} is specified, in which case it is output
5564 to the usual dump listing file, @file{.sched1} or @file{.sched2}
5565 respectively. However for @var{n} greater than nine, the output is
5566 always printed to standard error.
5568 For @var{n} greater than zero, @option{-fsched-verbose} outputs the
5569 same information as @option{-fdump-rtl-sched1} and @option{-fdump-rtl-sched2}.
5570 For @var{n} greater than one, it also output basic block probabilities,
5571 detailed ready list information and unit/insn info. For @var{n} greater
5572 than two, it includes RTL at abort point, control-flow and regions info.
5573 And for @var{n} over four, @option{-fsched-verbose} also includes
5577 @itemx -save-temps=cwd
5579 Store the usual ``temporary'' intermediate files permanently; place them
5580 in the current directory and name them based on the source file. Thus,
5581 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
5582 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
5583 preprocessed @file{foo.i} output file even though the compiler now
5584 normally uses an integrated preprocessor.
5586 When used in combination with the @option{-x} command line option,
5587 @option{-save-temps} is sensible enough to avoid over writing an
5588 input source file with the same extension as an intermediate file.
5589 The corresponding intermediate file may be obtained by renaming the
5590 source file before using @option{-save-temps}.
5592 If you invoke GCC in parallel, compiling several different source
5593 files that share a common base name in different subdirectories or the
5594 same source file compiled for multiple output destinations, it is
5595 likely that the different parallel compilers will interfere with each
5596 other, and overwrite the temporary files. For instance:
5599 gcc -save-temps -o outdir1/foo.o indir1/foo.c&
5600 gcc -save-temps -o outdir2/foo.o indir2/foo.c&
5603 may result in @file{foo.i} and @file{foo.o} being written to
5604 simultaneously by both compilers.
5606 @item -save-temps=obj
5607 @opindex save-temps=obj
5608 Store the usual ``temporary'' intermediate files permanently. If the
5609 @option{-o} option is used, the temporary files are based on the
5610 object file. If the @option{-o} option is not used, the
5611 @option{-save-temps=obj} switch behaves like @option{-save-temps}.
5616 gcc -save-temps=obj -c foo.c
5617 gcc -save-temps=obj -c bar.c -o dir/xbar.o
5618 gcc -save-temps=obj foobar.c -o dir2/yfoobar
5621 would create @file{foo.i}, @file{foo.s}, @file{dir/xbar.i},
5622 @file{dir/xbar.s}, @file{dir2/yfoobar.i}, @file{dir2/yfoobar.s}, and
5623 @file{dir2/yfoobar.o}.
5625 @item -time@r{[}=@var{file}@r{]}
5627 Report the CPU time taken by each subprocess in the compilation
5628 sequence. For C source files, this is the compiler proper and assembler
5629 (plus the linker if linking is done).
5631 Without the specification of an output file, the output looks like this:
5638 The first number on each line is the ``user time'', that is time spent
5639 executing the program itself. The second number is ``system time'',
5640 time spent executing operating system routines on behalf of the program.
5641 Both numbers are in seconds.
5643 With the specification of an output file, the output is appended to the
5644 named file, and it looks like this:
5647 0.12 0.01 cc1 @var{options}
5648 0.00 0.01 as @var{options}
5651 The ``user time'' and the ``system time'' are moved before the program
5652 name, and the options passed to the program are displayed, so that one
5653 can later tell what file was being compiled, and with which options.
5655 @item -fvar-tracking
5656 @opindex fvar-tracking
5657 Run variable tracking pass. It computes where variables are stored at each
5658 position in code. Better debugging information is then generated
5659 (if the debugging information format supports this information).
5661 It is enabled by default when compiling with optimization (@option{-Os},
5662 @option{-O}, @option{-O2}, @dots{}), debugging information (@option{-g}) and
5663 the debug info format supports it.
5665 @item -fvar-tracking-assignments
5666 @opindex fvar-tracking-assignments
5667 @opindex fno-var-tracking-assignments
5668 Annotate assignments to user variables early in the compilation and
5669 attempt to carry the annotations over throughout the compilation all the
5670 way to the end, in an attempt to improve debug information while
5671 optimizing. Use of @option{-gdwarf-4} is recommended along with it.
5673 It can be enabled even if var-tracking is disabled, in which case
5674 annotations will be created and maintained, but discarded at the end.
5676 @item -fvar-tracking-assignments-toggle
5677 @opindex fvar-tracking-assignments-toggle
5678 @opindex fno-var-tracking-assignments-toggle
5679 Toggle @option{-fvar-tracking-assignments}, in the same way that
5680 @option{-gtoggle} toggles @option{-g}.
5682 @item -print-file-name=@var{library}
5683 @opindex print-file-name
5684 Print the full absolute name of the library file @var{library} that
5685 would be used when linking---and don't do anything else. With this
5686 option, GCC does not compile or link anything; it just prints the
5689 @item -print-multi-directory
5690 @opindex print-multi-directory
5691 Print the directory name corresponding to the multilib selected by any
5692 other switches present in the command line. This directory is supposed
5693 to exist in @env{GCC_EXEC_PREFIX}.
5695 @item -print-multi-lib
5696 @opindex print-multi-lib
5697 Print the mapping from multilib directory names to compiler switches
5698 that enable them. The directory name is separated from the switches by
5699 @samp{;}, and each switch starts with an @samp{@@} instead of the
5700 @samp{-}, without spaces between multiple switches. This is supposed to
5701 ease shell-processing.
5703 @item -print-multi-os-directory
5704 @opindex print-multi-os-directory
5705 Print the path to OS libraries for the selected
5706 multilib, relative to some @file{lib} subdirectory. If OS libraries are
5707 present in the @file{lib} subdirectory and no multilibs are used, this is
5708 usually just @file{.}, if OS libraries are present in @file{lib@var{suffix}}
5709 sibling directories this prints e.g.@: @file{../lib64}, @file{../lib} or
5710 @file{../lib32}, or if OS libraries are present in @file{lib/@var{subdir}}
5711 subdirectories it prints e.g.@: @file{amd64}, @file{sparcv9} or @file{ev6}.
5713 @item -print-prog-name=@var{program}
5714 @opindex print-prog-name
5715 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
5717 @item -print-libgcc-file-name
5718 @opindex print-libgcc-file-name
5719 Same as @option{-print-file-name=libgcc.a}.
5721 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
5722 but you do want to link with @file{libgcc.a}. You can do
5725 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
5728 @item -print-search-dirs
5729 @opindex print-search-dirs
5730 Print the name of the configured installation directory and a list of
5731 program and library directories @command{gcc} will search---and don't do anything else.
5733 This is useful when @command{gcc} prints the error message
5734 @samp{installation problem, cannot exec cpp0: No such file or directory}.
5735 To resolve this you either need to put @file{cpp0} and the other compiler
5736 components where @command{gcc} expects to find them, or you can set the environment
5737 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
5738 Don't forget the trailing @samp{/}.
5739 @xref{Environment Variables}.
5741 @item -print-sysroot
5742 @opindex print-sysroot
5743 Print the target sysroot directory that will be used during
5744 compilation. This is the target sysroot specified either at configure
5745 time or using the @option{--sysroot} option, possibly with an extra
5746 suffix that depends on compilation options. If no target sysroot is
5747 specified, the option prints nothing.
5749 @item -print-sysroot-headers-suffix
5750 @opindex print-sysroot-headers-suffix
5751 Print the suffix added to the target sysroot when searching for
5752 headers, or give an error if the compiler is not configured with such
5753 a suffix---and don't do anything else.
5756 @opindex dumpmachine
5757 Print the compiler's target machine (for example,
5758 @samp{i686-pc-linux-gnu})---and don't do anything else.
5761 @opindex dumpversion
5762 Print the compiler version (for example, @samp{3.0})---and don't do
5767 Print the compiler's built-in specs---and don't do anything else. (This
5768 is used when GCC itself is being built.) @xref{Spec Files}.
5770 @item -feliminate-unused-debug-types
5771 @opindex feliminate-unused-debug-types
5772 Normally, when producing DWARF2 output, GCC will emit debugging
5773 information for all types declared in a compilation
5774 unit, regardless of whether or not they are actually used
5775 in that compilation unit. Sometimes this is useful, such as
5776 if, in the debugger, you want to cast a value to a type that is
5777 not actually used in your program (but is declared). More often,
5778 however, this results in a significant amount of wasted space.
5779 With this option, GCC will avoid producing debug symbol output
5780 for types that are nowhere used in the source file being compiled.
5783 @node Optimize Options
5784 @section Options That Control Optimization
5785 @cindex optimize options
5786 @cindex options, optimization
5788 These options control various sorts of optimizations.
5790 Without any optimization option, the compiler's goal is to reduce the
5791 cost of compilation and to make debugging produce the expected
5792 results. Statements are independent: if you stop the program with a
5793 breakpoint between statements, you can then assign a new value to any
5794 variable or change the program counter to any other statement in the
5795 function and get exactly the results you would expect from the source
5798 Turning on optimization flags makes the compiler attempt to improve
5799 the performance and/or code size at the expense of compilation time
5800 and possibly the ability to debug the program.
5802 The compiler performs optimization based on the knowledge it has of the
5803 program. Compiling multiple files at once to a single output file mode allows
5804 the compiler to use information gained from all of the files when compiling
5807 Not all optimizations are controlled directly by a flag. Only
5808 optimizations that have a flag are listed in this section.
5810 Most optimizations are only enabled if an @option{-O} level is set on
5811 the command line. Otherwise they are disabled, even if individual
5812 optimization flags are specified.
5814 Depending on the target and how GCC was configured, a slightly different
5815 set of optimizations may be enabled at each @option{-O} level than
5816 those listed here. You can invoke GCC with @samp{-Q --help=optimizers}
5817 to find out the exact set of optimizations that are enabled at each level.
5818 @xref{Overall Options}, for examples.
5825 Optimize. Optimizing compilation takes somewhat more time, and a lot
5826 more memory for a large function.
5828 With @option{-O}, the compiler tries to reduce code size and execution
5829 time, without performing any optimizations that take a great deal of
5832 @option{-O} turns on the following optimization flags:
5835 -fcprop-registers @gol
5838 -fdelayed-branch @gol
5840 -fguess-branch-probability @gol
5841 -fif-conversion2 @gol
5842 -fif-conversion @gol
5843 -fipa-pure-const @gol
5845 -fipa-reference @gol
5847 -fsplit-wide-types @gol
5848 -ftree-builtin-call-dce @gol
5851 -ftree-copyrename @gol
5853 -ftree-dominator-opts @gol
5855 -ftree-forwprop @gol
5863 @option{-O} also turns on @option{-fomit-frame-pointer} on machines
5864 where doing so does not interfere with debugging.
5868 Optimize even more. GCC performs nearly all supported optimizations
5869 that do not involve a space-speed tradeoff.
5870 As compared to @option{-O}, this option increases both compilation time
5871 and the performance of the generated code.
5873 @option{-O2} turns on all optimization flags specified by @option{-O}. It
5874 also turns on the following optimization flags:
5875 @gccoptlist{-fthread-jumps @gol
5876 -falign-functions -falign-jumps @gol
5877 -falign-loops -falign-labels @gol
5880 -fcse-follow-jumps -fcse-skip-blocks @gol
5881 -fdelete-null-pointer-checks @gol
5882 -fexpensive-optimizations @gol
5883 -fgcse -fgcse-lm @gol
5884 -finline-small-functions @gol
5885 -findirect-inlining @gol
5887 -foptimize-sibling-calls @gol
5888 -fpartial-inlining @gol
5891 -freorder-blocks -freorder-functions @gol
5892 -frerun-cse-after-loop @gol
5893 -fsched-interblock -fsched-spec @gol
5894 -fschedule-insns -fschedule-insns2 @gol
5895 -fstrict-aliasing -fstrict-overflow @gol
5896 -ftree-switch-conversion @gol
5900 Please note the warning under @option{-fgcse} about
5901 invoking @option{-O2} on programs that use computed gotos.
5905 Optimize yet more. @option{-O3} turns on all optimizations specified
5906 by @option{-O2} and also turns on the @option{-finline-functions},
5907 @option{-funswitch-loops}, @option{-fpredictive-commoning},
5908 @option{-fgcse-after-reload} and @option{-ftree-vectorize} options.
5912 Reduce compilation time and make debugging produce the expected
5913 results. This is the default.
5917 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
5918 do not typically increase code size. It also performs further
5919 optimizations designed to reduce code size.
5921 @option{-Os} disables the following optimization flags:
5922 @gccoptlist{-falign-functions -falign-jumps -falign-loops @gol
5923 -falign-labels -freorder-blocks -freorder-blocks-and-partition @gol
5924 -fprefetch-loop-arrays -ftree-vect-loop-version}
5928 Disregard strict standards compliance. @option{-Ofast} enables all
5929 @option{-O3} optimizations. It also enables optimizations that are not
5930 valid for all standard compliant programs.
5931 It turns on @option{-ffast-math}.
5933 If you use multiple @option{-O} options, with or without level numbers,
5934 the last such option is the one that is effective.
5937 Options of the form @option{-f@var{flag}} specify machine-independent
5938 flags. Most flags have both positive and negative forms; the negative
5939 form of @option{-ffoo} would be @option{-fno-foo}. In the table
5940 below, only one of the forms is listed---the one you typically will
5941 use. You can figure out the other form by either removing @samp{no-}
5944 The following options control specific optimizations. They are either
5945 activated by @option{-O} options or are related to ones that are. You
5946 can use the following flags in the rare cases when ``fine-tuning'' of
5947 optimizations to be performed is desired.
5950 @item -fno-default-inline
5951 @opindex fno-default-inline
5952 Do not make member functions inline by default merely because they are
5953 defined inside the class scope (C++ only). Otherwise, when you specify
5954 @w{@option{-O}}, member functions defined inside class scope are compiled
5955 inline by default; i.e., you don't need to add @samp{inline} in front of
5956 the member function name.
5958 @item -fno-defer-pop
5959 @opindex fno-defer-pop
5960 Always pop the arguments to each function call as soon as that function
5961 returns. For machines which must pop arguments after a function call,
5962 the compiler normally lets arguments accumulate on the stack for several
5963 function calls and pops them all at once.
5965 Disabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5967 @item -fforward-propagate
5968 @opindex fforward-propagate
5969 Perform a forward propagation pass on RTL@. The pass tries to combine two
5970 instructions and checks if the result can be simplified. If loop unrolling
5971 is active, two passes are performed and the second is scheduled after
5974 This option is enabled by default at optimization levels @option{-O},
5975 @option{-O2}, @option{-O3}, @option{-Os}.
5977 @item -fomit-frame-pointer
5978 @opindex fomit-frame-pointer
5979 Don't keep the frame pointer in a register for functions that
5980 don't need one. This avoids the instructions to save, set up and
5981 restore frame pointers; it also makes an extra register available
5982 in many functions. @strong{It also makes debugging impossible on
5985 On some machines, such as the VAX, this flag has no effect, because
5986 the standard calling sequence automatically handles the frame pointer
5987 and nothing is saved by pretending it doesn't exist. The
5988 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
5989 whether a target machine supports this flag. @xref{Registers,,Register
5990 Usage, gccint, GNU Compiler Collection (GCC) Internals}.
5992 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
5994 @item -foptimize-sibling-calls
5995 @opindex foptimize-sibling-calls
5996 Optimize sibling and tail recursive calls.
5998 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6002 Don't pay attention to the @code{inline} keyword. Normally this option
6003 is used to keep the compiler from expanding any functions inline.
6004 Note that if you are not optimizing, no functions can be expanded inline.
6006 @item -finline-small-functions
6007 @opindex finline-small-functions
6008 Integrate functions into their callers when their body is smaller than expected
6009 function call code (so overall size of program gets smaller). The compiler
6010 heuristically decides which functions are simple enough to be worth integrating
6013 Enabled at level @option{-O2}.
6015 @item -findirect-inlining
6016 @opindex findirect-inlining
6017 Inline also indirect calls that are discovered to be known at compile
6018 time thanks to previous inlining. This option has any effect only
6019 when inlining itself is turned on by the @option{-finline-functions}
6020 or @option{-finline-small-functions} options.
6022 Enabled at level @option{-O2}.
6024 @item -finline-functions
6025 @opindex finline-functions
6026 Integrate all simple functions into their callers. The compiler
6027 heuristically decides which functions are simple enough to be worth
6028 integrating in this way.
6030 If all calls to a given function are integrated, and the function is
6031 declared @code{static}, then the function is normally not output as
6032 assembler code in its own right.
6034 Enabled at level @option{-O3}.
6036 @item -finline-functions-called-once
6037 @opindex finline-functions-called-once
6038 Consider all @code{static} functions called once for inlining into their
6039 caller even if they are not marked @code{inline}. If a call to a given
6040 function is integrated, then the function is not output as assembler code
6043 Enabled at levels @option{-O1}, @option{-O2}, @option{-O3} and @option{-Os}.
6045 @item -fearly-inlining
6046 @opindex fearly-inlining
6047 Inline functions marked by @code{always_inline} and functions whose body seems
6048 smaller than the function call overhead early before doing
6049 @option{-fprofile-generate} instrumentation and real inlining pass. Doing so
6050 makes profiling significantly cheaper and usually inlining faster on programs
6051 having large chains of nested wrapper functions.
6057 Perform interprocedural scalar replacement of aggregates, removal of
6058 unused parameters and replacement of parameters passed by reference
6059 by parameters passed by value.
6061 Enabled at levels @option{-O2}, @option{-O3} and @option{-Os}.
6063 @item -finline-limit=@var{n}
6064 @opindex finline-limit
6065 By default, GCC limits the size of functions that can be inlined. This flag
6066 allows coarse control of this limit. @var{n} is the size of functions that
6067 can be inlined in number of pseudo instructions.
6069 Inlining is actually controlled by a number of parameters, which may be
6070 specified individually by using @option{--param @var{name}=@var{value}}.
6071 The @option{-finline-limit=@var{n}} option sets some of these parameters
6075 @item max-inline-insns-single
6076 is set to @var{n}/2.
6077 @item max-inline-insns-auto
6078 is set to @var{n}/2.
6081 See below for a documentation of the individual
6082 parameters controlling inlining and for the defaults of these parameters.
6084 @emph{Note:} there may be no value to @option{-finline-limit} that results
6085 in default behavior.
6087 @emph{Note:} pseudo instruction represents, in this particular context, an
6088 abstract measurement of function's size. In no way does it represent a count
6089 of assembly instructions and as such its exact meaning might change from one
6090 release to an another.
6092 @item -fkeep-inline-functions
6093 @opindex fkeep-inline-functions
6094 In C, emit @code{static} functions that are declared @code{inline}
6095 into the object file, even if the function has been inlined into all
6096 of its callers. This switch does not affect functions using the
6097 @code{extern inline} extension in GNU C90@. In C++, emit any and all
6098 inline functions into the object file.
6100 @item -fkeep-static-consts
6101 @opindex fkeep-static-consts
6102 Emit variables declared @code{static const} when optimization isn't turned
6103 on, even if the variables aren't referenced.
6105 GCC enables this option by default. If you want to force the compiler to
6106 check if the variable was referenced, regardless of whether or not
6107 optimization is turned on, use the @option{-fno-keep-static-consts} option.
6109 @item -fmerge-constants
6110 @opindex fmerge-constants
6111 Attempt to merge identical constants (string constants and floating point
6112 constants) across compilation units.
6114 This option is the default for optimized compilation if the assembler and
6115 linker support it. Use @option{-fno-merge-constants} to inhibit this
6118 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6120 @item -fmerge-all-constants
6121 @opindex fmerge-all-constants
6122 Attempt to merge identical constants and identical variables.
6124 This option implies @option{-fmerge-constants}. In addition to
6125 @option{-fmerge-constants} this considers e.g.@: even constant initialized
6126 arrays or initialized constant variables with integral or floating point
6127 types. Languages like C or C++ require each variable, including multiple
6128 instances of the same variable in recursive calls, to have distinct locations,
6129 so using this option will result in non-conforming
6132 @item -fmodulo-sched
6133 @opindex fmodulo-sched
6134 Perform swing modulo scheduling immediately before the first scheduling
6135 pass. This pass looks at innermost loops and reorders their
6136 instructions by overlapping different iterations.
6138 @item -fmodulo-sched-allow-regmoves
6139 @opindex fmodulo-sched-allow-regmoves
6140 Perform more aggressive SMS based modulo scheduling with register moves
6141 allowed. By setting this flag certain anti-dependences edges will be
6142 deleted which will trigger the generation of reg-moves based on the
6143 life-range analysis. This option is effective only with
6144 @option{-fmodulo-sched} enabled.
6146 @item -fno-branch-count-reg
6147 @opindex fno-branch-count-reg
6148 Do not use ``decrement and branch'' instructions on a count register,
6149 but instead generate a sequence of instructions that decrement a
6150 register, compare it against zero, then branch based upon the result.
6151 This option is only meaningful on architectures that support such
6152 instructions, which include x86, PowerPC, IA-64 and S/390.
6154 The default is @option{-fbranch-count-reg}.
6156 @item -fno-function-cse
6157 @opindex fno-function-cse
6158 Do not put function addresses in registers; make each instruction that
6159 calls a constant function contain the function's address explicitly.
6161 This option results in less efficient code, but some strange hacks
6162 that alter the assembler output may be confused by the optimizations
6163 performed when this option is not used.
6165 The default is @option{-ffunction-cse}
6167 @item -fno-zero-initialized-in-bss
6168 @opindex fno-zero-initialized-in-bss
6169 If the target supports a BSS section, GCC by default puts variables that
6170 are initialized to zero into BSS@. This can save space in the resulting
6173 This option turns off this behavior because some programs explicitly
6174 rely on variables going to the data section. E.g., so that the
6175 resulting executable can find the beginning of that section and/or make
6176 assumptions based on that.
6178 The default is @option{-fzero-initialized-in-bss}.
6180 @item -fmudflap -fmudflapth -fmudflapir
6184 @cindex bounds checking
6186 For front-ends that support it (C and C++), instrument all risky
6187 pointer/array dereferencing operations, some standard library
6188 string/heap functions, and some other associated constructs with
6189 range/validity tests. Modules so instrumented should be immune to
6190 buffer overflows, invalid heap use, and some other classes of C/C++
6191 programming errors. The instrumentation relies on a separate runtime
6192 library (@file{libmudflap}), which will be linked into a program if
6193 @option{-fmudflap} is given at link time. Run-time behavior of the
6194 instrumented program is controlled by the @env{MUDFLAP_OPTIONS}
6195 environment variable. See @code{env MUDFLAP_OPTIONS=-help a.out}
6198 Use @option{-fmudflapth} instead of @option{-fmudflap} to compile and to
6199 link if your program is multi-threaded. Use @option{-fmudflapir}, in
6200 addition to @option{-fmudflap} or @option{-fmudflapth}, if
6201 instrumentation should ignore pointer reads. This produces less
6202 instrumentation (and therefore faster execution) and still provides
6203 some protection against outright memory corrupting writes, but allows
6204 erroneously read data to propagate within a program.
6206 @item -fthread-jumps
6207 @opindex fthread-jumps
6208 Perform optimizations where we check to see if a jump branches to a
6209 location where another comparison subsumed by the first is found. If
6210 so, the first branch is redirected to either the destination of the
6211 second branch or a point immediately following it, depending on whether
6212 the condition is known to be true or false.
6214 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6216 @item -fsplit-wide-types
6217 @opindex fsplit-wide-types
6218 When using a type that occupies multiple registers, such as @code{long
6219 long} on a 32-bit system, split the registers apart and allocate them
6220 independently. This normally generates better code for those types,
6221 but may make debugging more difficult.
6223 Enabled at levels @option{-O}, @option{-O2}, @option{-O3},
6226 @item -fcse-follow-jumps
6227 @opindex fcse-follow-jumps
6228 In common subexpression elimination (CSE), scan through jump instructions
6229 when the target of the jump is not reached by any other path. For
6230 example, when CSE encounters an @code{if} statement with an
6231 @code{else} clause, CSE will follow the jump when the condition
6234 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6236 @item -fcse-skip-blocks
6237 @opindex fcse-skip-blocks
6238 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
6239 follow jumps which conditionally skip over blocks. When CSE
6240 encounters a simple @code{if} statement with no else clause,
6241 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
6242 body of the @code{if}.
6244 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6246 @item -frerun-cse-after-loop
6247 @opindex frerun-cse-after-loop
6248 Re-run common subexpression elimination after loop optimizations has been
6251 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6255 Perform a global common subexpression elimination pass.
6256 This pass also performs global constant and copy propagation.
6258 @emph{Note:} When compiling a program using computed gotos, a GCC
6259 extension, you may get better runtime performance if you disable
6260 the global common subexpression elimination pass by adding
6261 @option{-fno-gcse} to the command line.
6263 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6267 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
6268 attempt to move loads which are only killed by stores into themselves. This
6269 allows a loop containing a load/store sequence to be changed to a load outside
6270 the loop, and a copy/store within the loop.
6272 Enabled by default when gcse is enabled.
6276 When @option{-fgcse-sm} is enabled, a store motion pass is run after
6277 global common subexpression elimination. This pass will attempt to move
6278 stores out of loops. When used in conjunction with @option{-fgcse-lm},
6279 loops containing a load/store sequence can be changed to a load before
6280 the loop and a store after the loop.
6282 Not enabled at any optimization level.
6286 When @option{-fgcse-las} is enabled, the global common subexpression
6287 elimination pass eliminates redundant loads that come after stores to the
6288 same memory location (both partial and full redundancies).
6290 Not enabled at any optimization level.
6292 @item -fgcse-after-reload
6293 @opindex fgcse-after-reload
6294 When @option{-fgcse-after-reload} is enabled, a redundant load elimination
6295 pass is performed after reload. The purpose of this pass is to cleanup
6298 @item -funsafe-loop-optimizations
6299 @opindex funsafe-loop-optimizations
6300 If given, the loop optimizer will assume that loop indices do not
6301 overflow, and that the loops with nontrivial exit condition are not
6302 infinite. This enables a wider range of loop optimizations even if
6303 the loop optimizer itself cannot prove that these assumptions are valid.
6304 Using @option{-Wunsafe-loop-optimizations}, the compiler will warn you
6305 if it finds this kind of loop.
6307 @item -fcrossjumping
6308 @opindex fcrossjumping
6309 Perform cross-jumping transformation. This transformation unifies equivalent code and save code size. The
6310 resulting code may or may not perform better than without cross-jumping.
6312 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6314 @item -fauto-inc-dec
6315 @opindex fauto-inc-dec
6316 Combine increments or decrements of addresses with memory accesses.
6317 This pass is always skipped on architectures that do not have
6318 instructions to support this. Enabled by default at @option{-O} and
6319 higher on architectures that support this.
6323 Perform dead code elimination (DCE) on RTL@.
6324 Enabled by default at @option{-O} and higher.
6328 Perform dead store elimination (DSE) on RTL@.
6329 Enabled by default at @option{-O} and higher.
6331 @item -fif-conversion
6332 @opindex fif-conversion
6333 Attempt to transform conditional jumps into branch-less equivalents. This
6334 include use of conditional moves, min, max, set flags and abs instructions, and
6335 some tricks doable by standard arithmetics. The use of conditional execution
6336 on chips where it is available is controlled by @code{if-conversion2}.
6338 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6340 @item -fif-conversion2
6341 @opindex fif-conversion2
6342 Use conditional execution (where available) to transform conditional jumps into
6343 branch-less equivalents.
6345 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6347 @item -fdelete-null-pointer-checks
6348 @opindex fdelete-null-pointer-checks
6349 Assume that programs cannot safely dereference null pointers, and that
6350 no code or data element resides there. This enables simple constant
6351 folding optimizations at all optimization levels. In addition, other
6352 optimization passes in GCC use this flag to control global dataflow
6353 analyses that eliminate useless checks for null pointers; these assume
6354 that if a pointer is checked after it has already been dereferenced,
6357 Note however that in some environments this assumption is not true.
6358 Use @option{-fno-delete-null-pointer-checks} to disable this optimization
6359 for programs which depend on that behavior.
6361 Some targets, especially embedded ones, disable this option at all levels.
6362 Otherwise it is enabled at all levels: @option{-O0}, @option{-O1},
6363 @option{-O2}, @option{-O3}, @option{-Os}. Passes that use the information
6364 are enabled independently at different optimization levels.
6366 @item -fexpensive-optimizations
6367 @opindex fexpensive-optimizations
6368 Perform a number of minor optimizations that are relatively expensive.
6370 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6372 @item -foptimize-register-move
6374 @opindex foptimize-register-move
6376 Attempt to reassign register numbers in move instructions and as
6377 operands of other simple instructions in order to maximize the amount of
6378 register tying. This is especially helpful on machines with two-operand
6381 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
6384 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6386 @item -fira-algorithm=@var{algorithm}
6387 Use specified coloring algorithm for the integrated register
6388 allocator. The @var{algorithm} argument should be @code{priority} or
6389 @code{CB}. The first algorithm specifies Chow's priority coloring,
6390 the second one specifies Chaitin-Briggs coloring. The second
6391 algorithm can be unimplemented for some architectures. If it is
6392 implemented, it is the default because Chaitin-Briggs coloring as a
6393 rule generates a better code.
6395 @item -fira-region=@var{region}
6396 Use specified regions for the integrated register allocator. The
6397 @var{region} argument should be one of @code{all}, @code{mixed}, or
6398 @code{one}. The first value means using all loops as register
6399 allocation regions, the second value which is the default means using
6400 all loops except for loops with small register pressure as the
6401 regions, and third one means using all function as a single region.
6402 The first value can give best result for machines with small size and
6403 irregular register set, the third one results in faster and generates
6404 decent code and the smallest size code, and the default value usually
6405 give the best results in most cases and for most architectures.
6407 @item -fira-coalesce
6408 @opindex fira-coalesce
6409 Do optimistic register coalescing. This option might be profitable for
6410 architectures with big regular register files.
6412 @item -fira-loop-pressure
6413 @opindex fira-loop-pressure
6414 Use IRA to evaluate register pressure in loops for decision to move
6415 loop invariants. Usage of this option usually results in generation
6416 of faster and smaller code on machines with big register files (>= 32
6417 registers) but it can slow compiler down.
6419 This option is enabled at level @option{-O3} for some targets.
6421 @item -fno-ira-share-save-slots
6422 @opindex fno-ira-share-save-slots
6423 Switch off sharing stack slots used for saving call used hard
6424 registers living through a call. Each hard register will get a
6425 separate stack slot and as a result function stack frame will be
6428 @item -fno-ira-share-spill-slots
6429 @opindex fno-ira-share-spill-slots
6430 Switch off sharing stack slots allocated for pseudo-registers. Each
6431 pseudo-register which did not get a hard register will get a separate
6432 stack slot and as a result function stack frame will be bigger.
6434 @item -fira-verbose=@var{n}
6435 @opindex fira-verbose
6436 Set up how verbose dump file for the integrated register allocator
6437 will be. Default value is 5. If the value is greater or equal to 10,
6438 the dump file will be stderr as if the value were @var{n} minus 10.
6440 @item -fdelayed-branch
6441 @opindex fdelayed-branch
6442 If supported for the target machine, attempt to reorder instructions
6443 to exploit instruction slots available after delayed branch
6446 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
6448 @item -fschedule-insns
6449 @opindex fschedule-insns
6450 If supported for the target machine, attempt to reorder instructions to
6451 eliminate execution stalls due to required data being unavailable. This
6452 helps machines that have slow floating point or memory load instructions
6453 by allowing other instructions to be issued until the result of the load
6454 or floating point instruction is required.
6456 Enabled at levels @option{-O2}, @option{-O3}.
6458 @item -fschedule-insns2
6459 @opindex fschedule-insns2
6460 Similar to @option{-fschedule-insns}, but requests an additional pass of
6461 instruction scheduling after register allocation has been done. This is
6462 especially useful on machines with a relatively small number of
6463 registers and where memory load instructions take more than one cycle.
6465 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6467 @item -fno-sched-interblock
6468 @opindex fno-sched-interblock
6469 Don't schedule instructions across basic blocks. This is normally
6470 enabled by default when scheduling before register allocation, i.e.@:
6471 with @option{-fschedule-insns} or at @option{-O2} or higher.
6473 @item -fno-sched-spec
6474 @opindex fno-sched-spec
6475 Don't allow speculative motion of non-load instructions. This is normally
6476 enabled by default when scheduling before register allocation, i.e.@:
6477 with @option{-fschedule-insns} or at @option{-O2} or higher.
6479 @item -fsched-pressure
6480 @opindex fsched-pressure
6481 Enable register pressure sensitive insn scheduling before the register
6482 allocation. This only makes sense when scheduling before register
6483 allocation is enabled, i.e.@: with @option{-fschedule-insns} or at
6484 @option{-O2} or higher. Usage of this option can improve the
6485 generated code and decrease its size by preventing register pressure
6486 increase above the number of available hard registers and as a
6487 consequence register spills in the register allocation.
6489 @item -fsched-spec-load
6490 @opindex fsched-spec-load
6491 Allow speculative motion of some load instructions. This only makes
6492 sense when scheduling before register allocation, i.e.@: with
6493 @option{-fschedule-insns} or at @option{-O2} or higher.
6495 @item -fsched-spec-load-dangerous
6496 @opindex fsched-spec-load-dangerous
6497 Allow speculative motion of more load instructions. This only makes
6498 sense when scheduling before register allocation, i.e.@: with
6499 @option{-fschedule-insns} or at @option{-O2} or higher.
6501 @item -fsched-stalled-insns
6502 @itemx -fsched-stalled-insns=@var{n}
6503 @opindex fsched-stalled-insns
6504 Define how many insns (if any) can be moved prematurely from the queue
6505 of stalled insns into the ready list, during the second scheduling pass.
6506 @option{-fno-sched-stalled-insns} means that no insns will be moved
6507 prematurely, @option{-fsched-stalled-insns=0} means there is no limit
6508 on how many queued insns can be moved prematurely.
6509 @option{-fsched-stalled-insns} without a value is equivalent to
6510 @option{-fsched-stalled-insns=1}.
6512 @item -fsched-stalled-insns-dep
6513 @itemx -fsched-stalled-insns-dep=@var{n}
6514 @opindex fsched-stalled-insns-dep
6515 Define how many insn groups (cycles) will be examined for a dependency
6516 on a stalled insn that is candidate for premature removal from the queue
6517 of stalled insns. This has an effect only during the second scheduling pass,
6518 and only if @option{-fsched-stalled-insns} is used.
6519 @option{-fno-sched-stalled-insns-dep} is equivalent to
6520 @option{-fsched-stalled-insns-dep=0}.
6521 @option{-fsched-stalled-insns-dep} without a value is equivalent to
6522 @option{-fsched-stalled-insns-dep=1}.
6524 @item -fsched2-use-superblocks
6525 @opindex fsched2-use-superblocks
6526 When scheduling after register allocation, do use superblock scheduling
6527 algorithm. Superblock scheduling allows motion across basic block boundaries
6528 resulting on faster schedules. This option is experimental, as not all machine
6529 descriptions used by GCC model the CPU closely enough to avoid unreliable
6530 results from the algorithm.
6532 This only makes sense when scheduling after register allocation, i.e.@: with
6533 @option{-fschedule-insns2} or at @option{-O2} or higher.
6535 @item -fsched-group-heuristic
6536 @opindex fsched-group-heuristic
6537 Enable the group heuristic in the scheduler. This heuristic favors
6538 the instruction that belongs to a schedule group. This is enabled
6539 by default when scheduling is enabled, i.e.@: with @option{-fschedule-insns}
6540 or @option{-fschedule-insns2} or at @option{-O2} or higher.
6542 @item -fsched-critical-path-heuristic
6543 @opindex fsched-critical-path-heuristic
6544 Enable the critical-path heuristic in the scheduler. This heuristic favors
6545 instructions on the critical path. This is enabled by default when
6546 scheduling is enabled, i.e.@: with @option{-fschedule-insns}
6547 or @option{-fschedule-insns2} or at @option{-O2} or higher.
6549 @item -fsched-spec-insn-heuristic
6550 @opindex fsched-spec-insn-heuristic
6551 Enable the speculative instruction heuristic in the scheduler. This
6552 heuristic favors speculative instructions with greater dependency weakness.
6553 This is enabled by default when scheduling is enabled, i.e.@:
6554 with @option{-fschedule-insns} or @option{-fschedule-insns2}
6555 or at @option{-O2} or higher.
6557 @item -fsched-rank-heuristic
6558 @opindex fsched-rank-heuristic
6559 Enable the rank heuristic in the scheduler. This heuristic favors
6560 the instruction belonging to a basic block with greater size or frequency.
6561 This is enabled by default when scheduling is enabled, i.e.@:
6562 with @option{-fschedule-insns} or @option{-fschedule-insns2} or
6563 at @option{-O2} or higher.
6565 @item -fsched-last-insn-heuristic
6566 @opindex fsched-last-insn-heuristic
6567 Enable the last-instruction heuristic in the scheduler. This heuristic
6568 favors the instruction that is less dependent on the last instruction
6569 scheduled. This is enabled by default when scheduling is enabled,
6570 i.e.@: with @option{-fschedule-insns} or @option{-fschedule-insns2} or
6571 at @option{-O2} or higher.
6573 @item -fsched-dep-count-heuristic
6574 @opindex fsched-dep-count-heuristic
6575 Enable the dependent-count heuristic in the scheduler. This heuristic
6576 favors the instruction that has more instructions depending on it.
6577 This is enabled by default when scheduling is enabled, i.e.@:
6578 with @option{-fschedule-insns} or @option{-fschedule-insns2} or
6579 at @option{-O2} or higher.
6581 @item -freschedule-modulo-scheduled-loops
6582 @opindex freschedule-modulo-scheduled-loops
6583 The modulo scheduling comes before the traditional scheduling, if a loop
6584 was modulo scheduled we may want to prevent the later scheduling passes
6585 from changing its schedule, we use this option to control that.
6587 @item -fselective-scheduling
6588 @opindex fselective-scheduling
6589 Schedule instructions using selective scheduling algorithm. Selective
6590 scheduling runs instead of the first scheduler pass.
6592 @item -fselective-scheduling2
6593 @opindex fselective-scheduling2
6594 Schedule instructions using selective scheduling algorithm. Selective
6595 scheduling runs instead of the second scheduler pass.
6597 @item -fsel-sched-pipelining
6598 @opindex fsel-sched-pipelining
6599 Enable software pipelining of innermost loops during selective scheduling.
6600 This option has no effect until one of @option{-fselective-scheduling} or
6601 @option{-fselective-scheduling2} is turned on.
6603 @item -fsel-sched-pipelining-outer-loops
6604 @opindex fsel-sched-pipelining-outer-loops
6605 When pipelining loops during selective scheduling, also pipeline outer loops.
6606 This option has no effect until @option{-fsel-sched-pipelining} is turned on.
6608 @item -fcaller-saves
6609 @opindex fcaller-saves
6610 Enable values to be allocated in registers that will be clobbered by
6611 function calls, by emitting extra instructions to save and restore the
6612 registers around such calls. Such allocation is done only when it
6613 seems to result in better code than would otherwise be produced.
6615 This option is always enabled by default on certain machines, usually
6616 those which have no call-preserved registers to use instead.
6618 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
6620 @item -fconserve-stack
6621 @opindex fconserve-stack
6622 Attempt to minimize stack usage. The compiler will attempt to use less
6623 stack space, even if that makes the program slower. This option
6624 implies setting the @option{large-stack-frame} parameter to 100
6625 and the @option{large-stack-frame-growth} parameter to 400.
6627 @item -ftree-reassoc
6628 @opindex ftree-reassoc
6629 Perform reassociation on trees. This flag is enabled by default
6630 at @option{-O} and higher.
6634 Perform partial redundancy elimination (PRE) on trees. This flag is
6635 enabled by default at @option{-O2} and @option{-O3}.
6637 @item -ftree-forwprop
6638 @opindex ftree-forwprop
6639 Perform forward propagation on trees. This flag is enabled by default
6640 at @option{-O} and higher.
6644 Perform full redundancy elimination (FRE) on trees. The difference
6645 between FRE and PRE is that FRE only considers expressions
6646 that are computed on all paths leading to the redundant computation.
6647 This analysis is faster than PRE, though it exposes fewer redundancies.
6648 This flag is enabled by default at @option{-O} and higher.
6650 @item -ftree-phiprop
6651 @opindex ftree-phiprop
6652 Perform hoisting of loads from conditional pointers on trees. This
6653 pass is enabled by default at @option{-O} and higher.
6655 @item -ftree-copy-prop
6656 @opindex ftree-copy-prop
6657 Perform copy propagation on trees. This pass eliminates unnecessary
6658 copy operations. This flag is enabled by default at @option{-O} and
6661 @item -fipa-pure-const
6662 @opindex fipa-pure-const
6663 Discover which functions are pure or constant.
6664 Enabled by default at @option{-O} and higher.
6666 @item -fipa-reference
6667 @opindex fipa-reference
6668 Discover which static variables do not escape cannot escape the
6670 Enabled by default at @option{-O} and higher.
6672 @item -fipa-struct-reorg
6673 @opindex fipa-struct-reorg
6674 Perform structure reorganization optimization, that change C-like structures
6675 layout in order to better utilize spatial locality. This transformation is
6676 affective for programs containing arrays of structures. Available in two
6677 compilation modes: profile-based (enabled with @option{-fprofile-generate})
6678 or static (which uses built-in heuristics). It works only in whole program
6679 mode, so it requires @option{-fwhole-program} and @option{-combine} to be
6680 enabled. Structures considered @samp{cold} by this transformation are not
6681 affected (see @option{--param struct-reorg-cold-struct-ratio=@var{value}}).
6683 With this flag, the program debug info reflects a new structure layout.
6687 Perform interprocedural pointer analysis and interprocedural modification
6688 and reference analysis. This option can cause excessive memory and
6689 compile-time usage on large compilation units. It is not enabled by
6690 default at any optimization level.
6693 @opindex fipa-profile
6694 Perform interprocedural profile propagation. The functions called only from
6695 cold functions are marked as cold. Also functions executed once (such as
6696 @code{cold}, @code{noreturn}, static constructors or destructors) are identified. Cold
6697 functions and loop less parts of functions executed once are then optimized for
6699 Enabled by default at @option{-O} and higher.
6703 Perform interprocedural constant propagation.
6704 This optimization analyzes the program to determine when values passed
6705 to functions are constants and then optimizes accordingly.
6706 This optimization can substantially increase performance
6707 if the application has constants passed to functions.
6708 This flag is enabled by default at @option{-O2}, @option{-Os} and @option{-O3}.
6710 @item -fipa-cp-clone
6711 @opindex fipa-cp-clone
6712 Perform function cloning to make interprocedural constant propagation stronger.
6713 When enabled, interprocedural constant propagation will perform function cloning
6714 when externally visible function can be called with constant arguments.
6715 Because this optimization can create multiple copies of functions,
6716 it may significantly increase code size
6717 (see @option{--param ipcp-unit-growth=@var{value}}).
6718 This flag is enabled by default at @option{-O3}.
6720 @item -fipa-matrix-reorg
6721 @opindex fipa-matrix-reorg
6722 Perform matrix flattening and transposing.
6723 Matrix flattening tries to replace an @math{m}-dimensional matrix
6724 with its equivalent @math{n}-dimensional matrix, where @math{n < m}.
6725 This reduces the level of indirection needed for accessing the elements
6726 of the matrix. The second optimization is matrix transposing that
6727 attempts to change the order of the matrix's dimensions in order to
6728 improve cache locality.
6729 Both optimizations need the @option{-fwhole-program} flag.
6730 Transposing is enabled only if profiling information is available.
6734 Perform forward store motion on trees. This flag is
6735 enabled by default at @option{-O} and higher.
6739 Perform sparse conditional constant propagation (CCP) on trees. This
6740 pass only operates on local scalar variables and is enabled by default
6741 at @option{-O} and higher.
6743 @item -ftree-switch-conversion
6744 Perform conversion of simple initializations in a switch to
6745 initializations from a scalar array. This flag is enabled by default
6746 at @option{-O2} and higher.
6750 Perform dead code elimination (DCE) on trees. This flag is enabled by
6751 default at @option{-O} and higher.
6753 @item -ftree-builtin-call-dce
6754 @opindex ftree-builtin-call-dce
6755 Perform conditional dead code elimination (DCE) for calls to builtin functions
6756 that may set @code{errno} but are otherwise side-effect free. This flag is
6757 enabled by default at @option{-O2} and higher if @option{-Os} is not also
6760 @item -ftree-dominator-opts
6761 @opindex ftree-dominator-opts
6762 Perform a variety of simple scalar cleanups (constant/copy
6763 propagation, redundancy elimination, range propagation and expression
6764 simplification) based on a dominator tree traversal. This also
6765 performs jump threading (to reduce jumps to jumps). This flag is
6766 enabled by default at @option{-O} and higher.
6770 Perform dead store elimination (DSE) on trees. A dead store is a store into
6771 a memory location which will later be overwritten by another store without
6772 any intervening loads. In this case the earlier store can be deleted. This
6773 flag is enabled by default at @option{-O} and higher.
6777 Perform loop header copying on trees. This is beneficial since it increases
6778 effectiveness of code motion optimizations. It also saves one jump. This flag
6779 is enabled by default at @option{-O} and higher. It is not enabled
6780 for @option{-Os}, since it usually increases code size.
6782 @item -ftree-loop-optimize
6783 @opindex ftree-loop-optimize
6784 Perform loop optimizations on trees. This flag is enabled by default
6785 at @option{-O} and higher.
6787 @item -ftree-loop-linear
6788 @opindex ftree-loop-linear
6789 Perform linear loop transformations on tree. This flag can improve cache
6790 performance and allow further loop optimizations to take place.
6792 @item -floop-interchange
6793 Perform loop interchange transformations on loops. Interchanging two
6794 nested loops switches the inner and outer loops. For example, given a
6799 A(J, I) = A(J, I) * C
6803 loop interchange will transform the loop as if the user had written:
6807 A(J, I) = A(J, I) * C
6811 which can be beneficial when @code{N} is larger than the caches,
6812 because in Fortran, the elements of an array are stored in memory
6813 contiguously by column, and the original loop iterates over rows,
6814 potentially creating at each access a cache miss. This optimization
6815 applies to all the languages supported by GCC and is not limited to
6816 Fortran. To use this code transformation, GCC has to be configured
6817 with @option{--with-ppl} and @option{--with-cloog} to enable the
6818 Graphite loop transformation infrastructure.
6820 @item -floop-strip-mine
6821 Perform loop strip mining transformations on loops. Strip mining
6822 splits a loop into two nested loops. The outer loop has strides
6823 equal to the strip size and the inner loop has strides of the
6824 original loop within a strip. The strip length can be changed
6825 using the @option{loop-block-tile-size} parameter. For example,
6832 loop strip mining will transform the loop as if the user had written:
6835 DO I = II, min (II + 50, N)
6840 This optimization applies to all the languages supported by GCC and is
6841 not limited to Fortran. To use this code transformation, GCC has to
6842 be configured with @option{--with-ppl} and @option{--with-cloog} to
6843 enable the Graphite loop transformation infrastructure.
6846 Perform loop blocking transformations on loops. Blocking strip mines
6847 each loop in the loop nest such that the memory accesses of the
6848 element loops fit inside caches. The strip length can be changed
6849 using the @option{loop-block-tile-size} parameter. For example, given
6854 A(J, I) = B(I) + C(J)
6858 loop blocking will transform the loop as if the user had written:
6862 DO I = II, min (II + 50, N)
6863 DO J = JJ, min (JJ + 50, M)
6864 A(J, I) = B(I) + C(J)
6870 which can be beneficial when @code{M} is larger than the caches,
6871 because the innermost loop will iterate over a smaller amount of data
6872 that can be kept in the caches. This optimization applies to all the
6873 languages supported by GCC and is not limited to Fortran. To use this
6874 code transformation, GCC has to be configured with @option{--with-ppl}
6875 and @option{--with-cloog} to enable the Graphite loop transformation
6878 @item -fgraphite-identity
6879 @opindex fgraphite-identity
6880 Enable the identity transformation for graphite. For every SCoP we generate
6881 the polyhedral representation and transform it back to gimple. Using
6882 @option{-fgraphite-identity} we can check the costs or benefits of the
6883 GIMPLE -> GRAPHITE -> GIMPLE transformation. Some minimal optimizations
6884 are also performed by the code generator CLooG, like index splitting and
6885 dead code elimination in loops.
6887 @item -floop-parallelize-all
6888 Use the Graphite data dependence analysis to identify loops that can
6889 be parallelized. Parallelize all the loops that can be analyzed to
6890 not contain loop carried dependences without checking that it is
6891 profitable to parallelize the loops.
6893 @item -fcheck-data-deps
6894 @opindex fcheck-data-deps
6895 Compare the results of several data dependence analyzers. This option
6896 is used for debugging the data dependence analyzers.
6898 @item -ftree-loop-if-convert
6899 Attempt to transform conditional jumps in the innermost loops to
6900 branch-less equivalents. The intent is to remove control-flow from
6901 the innermost loops in order to improve the ability of the
6902 vectorization pass to handle these loops. This is enabled by default
6903 if vectorization is enabled.
6905 @item -ftree-loop-distribution
6906 Perform loop distribution. This flag can improve cache performance on
6907 big loop bodies and allow further loop optimizations, like
6908 parallelization or vectorization, to take place. For example, the loop
6925 @item -ftree-loop-im
6926 @opindex ftree-loop-im
6927 Perform loop invariant motion on trees. This pass moves only invariants that
6928 would be hard to handle at RTL level (function calls, operations that expand to
6929 nontrivial sequences of insns). With @option{-funswitch-loops} it also moves
6930 operands of conditions that are invariant out of the loop, so that we can use
6931 just trivial invariantness analysis in loop unswitching. The pass also includes
6934 @item -ftree-loop-ivcanon
6935 @opindex ftree-loop-ivcanon
6936 Create a canonical counter for number of iterations in the loop for that
6937 determining number of iterations requires complicated analysis. Later
6938 optimizations then may determine the number easily. Useful especially
6939 in connection with unrolling.
6943 Perform induction variable optimizations (strength reduction, induction
6944 variable merging and induction variable elimination) on trees.
6946 @item -ftree-parallelize-loops=n
6947 @opindex ftree-parallelize-loops
6948 Parallelize loops, i.e., split their iteration space to run in n threads.
6949 This is only possible for loops whose iterations are independent
6950 and can be arbitrarily reordered. The optimization is only
6951 profitable on multiprocessor machines, for loops that are CPU-intensive,
6952 rather than constrained e.g.@: by memory bandwidth. This option
6953 implies @option{-pthread}, and thus is only supported on targets
6954 that have support for @option{-pthread}.
6958 Perform function-local points-to analysis on trees. This flag is
6959 enabled by default at @option{-O} and higher.
6963 Perform scalar replacement of aggregates. This pass replaces structure
6964 references with scalars to prevent committing structures to memory too
6965 early. This flag is enabled by default at @option{-O} and higher.
6967 @item -ftree-copyrename
6968 @opindex ftree-copyrename
6969 Perform copy renaming on trees. This pass attempts to rename compiler
6970 temporaries to other variables at copy locations, usually resulting in
6971 variable names which more closely resemble the original variables. This flag
6972 is enabled by default at @option{-O} and higher.
6976 Perform temporary expression replacement during the SSA->normal phase. Single
6977 use/single def temporaries are replaced at their use location with their
6978 defining expression. This results in non-GIMPLE code, but gives the expanders
6979 much more complex trees to work on resulting in better RTL generation. This is
6980 enabled by default at @option{-O} and higher.
6982 @item -ftree-vectorize
6983 @opindex ftree-vectorize
6984 Perform loop vectorization on trees. This flag is enabled by default at
6987 @item -ftree-slp-vectorize
6988 @opindex ftree-slp-vectorize
6989 Perform basic block vectorization on trees. This flag is enabled by default at
6990 @option{-O3} and when @option{-ftree-vectorize} is enabled.
6992 @item -ftree-vect-loop-version
6993 @opindex ftree-vect-loop-version
6994 Perform loop versioning when doing loop vectorization on trees. When a loop
6995 appears to be vectorizable except that data alignment or data dependence cannot
6996 be determined at compile time then vectorized and non-vectorized versions of
6997 the loop are generated along with runtime checks for alignment or dependence
6998 to control which version is executed. This option is enabled by default
6999 except at level @option{-Os} where it is disabled.
7001 @item -fvect-cost-model
7002 @opindex fvect-cost-model
7003 Enable cost model for vectorization.
7007 Perform Value Range Propagation on trees. This is similar to the
7008 constant propagation pass, but instead of values, ranges of values are
7009 propagated. This allows the optimizers to remove unnecessary range
7010 checks like array bound checks and null pointer checks. This is
7011 enabled by default at @option{-O2} and higher. Null pointer check
7012 elimination is only done if @option{-fdelete-null-pointer-checks} is
7017 Perform tail duplication to enlarge superblock size. This transformation
7018 simplifies the control flow of the function allowing other optimizations to do
7021 @item -funroll-loops
7022 @opindex funroll-loops
7023 Unroll loops whose number of iterations can be determined at compile
7024 time or upon entry to the loop. @option{-funroll-loops} implies
7025 @option{-frerun-cse-after-loop}. This option makes code larger,
7026 and may or may not make it run faster.
7028 @item -funroll-all-loops
7029 @opindex funroll-all-loops
7030 Unroll all loops, even if their number of iterations is uncertain when
7031 the loop is entered. This usually makes programs run more slowly.
7032 @option{-funroll-all-loops} implies the same options as
7033 @option{-funroll-loops},
7035 @item -fsplit-ivs-in-unroller
7036 @opindex fsplit-ivs-in-unroller
7037 Enables expressing of values of induction variables in later iterations
7038 of the unrolled loop using the value in the first iteration. This breaks
7039 long dependency chains, thus improving efficiency of the scheduling passes.
7041 Combination of @option{-fweb} and CSE is often sufficient to obtain the
7042 same effect. However in cases the loop body is more complicated than
7043 a single basic block, this is not reliable. It also does not work at all
7044 on some of the architectures due to restrictions in the CSE pass.
7046 This optimization is enabled by default.
7048 @item -fvariable-expansion-in-unroller
7049 @opindex fvariable-expansion-in-unroller
7050 With this option, the compiler will create multiple copies of some
7051 local variables when unrolling a loop which can result in superior code.
7053 @item -fpartial-inlining
7054 @opindex fpartial-inlining
7055 Inline parts of functions. This option has any effect only
7056 when inlining itself is turned on by the @option{-finline-functions}
7057 or @option{-finline-small-functions} options.
7059 Enabled at level @option{-O2}.
7061 @item -fpredictive-commoning
7062 @opindex fpredictive-commoning
7063 Perform predictive commoning optimization, i.e., reusing computations
7064 (especially memory loads and stores) performed in previous
7065 iterations of loops.
7067 This option is enabled at level @option{-O3}.
7069 @item -fprefetch-loop-arrays
7070 @opindex fprefetch-loop-arrays
7071 If supported by the target machine, generate instructions to prefetch
7072 memory to improve the performance of loops that access large arrays.
7074 This option may generate better or worse code; results are highly
7075 dependent on the structure of loops within the source code.
7077 Disabled at level @option{-Os}.
7080 @itemx -fno-peephole2
7081 @opindex fno-peephole
7082 @opindex fno-peephole2
7083 Disable any machine-specific peephole optimizations. The difference
7084 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
7085 are implemented in the compiler; some targets use one, some use the
7086 other, a few use both.
7088 @option{-fpeephole} is enabled by default.
7089 @option{-fpeephole2} enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
7091 @item -fno-guess-branch-probability
7092 @opindex fno-guess-branch-probability
7093 Do not guess branch probabilities using heuristics.
7095 GCC will use heuristics to guess branch probabilities if they are
7096 not provided by profiling feedback (@option{-fprofile-arcs}). These
7097 heuristics are based on the control flow graph. If some branch probabilities
7098 are specified by @samp{__builtin_expect}, then the heuristics will be
7099 used to guess branch probabilities for the rest of the control flow graph,
7100 taking the @samp{__builtin_expect} info into account. The interactions
7101 between the heuristics and @samp{__builtin_expect} can be complex, and in
7102 some cases, it may be useful to disable the heuristics so that the effects
7103 of @samp{__builtin_expect} are easier to understand.
7105 The default is @option{-fguess-branch-probability} at levels
7106 @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
7108 @item -freorder-blocks
7109 @opindex freorder-blocks
7110 Reorder basic blocks in the compiled function in order to reduce number of
7111 taken branches and improve code locality.
7113 Enabled at levels @option{-O2}, @option{-O3}.
7115 @item -freorder-blocks-and-partition
7116 @opindex freorder-blocks-and-partition
7117 In addition to reordering basic blocks in the compiled function, in order
7118 to reduce number of taken branches, partitions hot and cold basic blocks
7119 into separate sections of the assembly and .o files, to improve
7120 paging and cache locality performance.
7122 This optimization is automatically turned off in the presence of
7123 exception handling, for linkonce sections, for functions with a user-defined
7124 section attribute and on any architecture that does not support named
7127 @item -freorder-functions
7128 @opindex freorder-functions
7129 Reorder functions in the object file in order to
7130 improve code locality. This is implemented by using special
7131 subsections @code{.text.hot} for most frequently executed functions and
7132 @code{.text.unlikely} for unlikely executed functions. Reordering is done by
7133 the linker so object file format must support named sections and linker must
7134 place them in a reasonable way.
7136 Also profile feedback must be available in to make this option effective. See
7137 @option{-fprofile-arcs} for details.
7139 Enabled at levels @option{-O2}, @option{-O3}, @option{-Os}.
7141 @item -fstrict-aliasing
7142 @opindex fstrict-aliasing
7143 Allow the compiler to assume the strictest aliasing rules applicable to
7144 the language being compiled. For C (and C++), this activates
7145 optimizations based on the type of expressions. In particular, an
7146 object of one type is assumed never to reside at the same address as an
7147 object of a different type, unless the types are almost the same. For
7148 example, an @code{unsigned int} can alias an @code{int}, but not a
7149 @code{void*} or a @code{double}. A character type may alias any other
7152 @anchor{Type-punning}Pay special attention to code like this:
7165 The practice of reading from a different union member than the one most
7166 recently written to (called ``type-punning'') is common. Even with
7167 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
7168 is accessed through the union type. So, the code above will work as
7169 expected. @xref{Structures unions enumerations and bit-fields
7170 implementation}. However, this code might not:
7181 Similarly, access by taking the address, casting the resulting pointer
7182 and dereferencing the result has undefined behavior, even if the cast
7183 uses a union type, e.g.:
7187 return ((union a_union *) &d)->i;
7191 The @option{-fstrict-aliasing} option is enabled at levels
7192 @option{-O2}, @option{-O3}, @option{-Os}.
7194 @item -fstrict-overflow
7195 @opindex fstrict-overflow
7196 Allow the compiler to assume strict signed overflow rules, depending
7197 on the language being compiled. For C (and C++) this means that
7198 overflow when doing arithmetic with signed numbers is undefined, which
7199 means that the compiler may assume that it will not happen. This
7200 permits various optimizations. For example, the compiler will assume
7201 that an expression like @code{i + 10 > i} will always be true for
7202 signed @code{i}. This assumption is only valid if signed overflow is
7203 undefined, as the expression is false if @code{i + 10} overflows when
7204 using twos complement arithmetic. When this option is in effect any
7205 attempt to determine whether an operation on signed numbers will
7206 overflow must be written carefully to not actually involve overflow.
7208 This option also allows the compiler to assume strict pointer
7209 semantics: given a pointer to an object, if adding an offset to that
7210 pointer does not produce a pointer to the same object, the addition is
7211 undefined. This permits the compiler to conclude that @code{p + u >
7212 p} is always true for a pointer @code{p} and unsigned integer
7213 @code{u}. This assumption is only valid because pointer wraparound is
7214 undefined, as the expression is false if @code{p + u} overflows using
7215 twos complement arithmetic.
7217 See also the @option{-fwrapv} option. Using @option{-fwrapv} means
7218 that integer signed overflow is fully defined: it wraps. When
7219 @option{-fwrapv} is used, there is no difference between
7220 @option{-fstrict-overflow} and @option{-fno-strict-overflow} for
7221 integers. With @option{-fwrapv} certain types of overflow are
7222 permitted. For example, if the compiler gets an overflow when doing
7223 arithmetic on constants, the overflowed value can still be used with
7224 @option{-fwrapv}, but not otherwise.
7226 The @option{-fstrict-overflow} option is enabled at levels
7227 @option{-O2}, @option{-O3}, @option{-Os}.
7229 @item -falign-functions
7230 @itemx -falign-functions=@var{n}
7231 @opindex falign-functions
7232 Align the start of functions to the next power-of-two greater than
7233 @var{n}, skipping up to @var{n} bytes. For instance,
7234 @option{-falign-functions=32} aligns functions to the next 32-byte
7235 boundary, but @option{-falign-functions=24} would align to the next
7236 32-byte boundary only if this can be done by skipping 23 bytes or less.
7238 @option{-fno-align-functions} and @option{-falign-functions=1} are
7239 equivalent and mean that functions will not be aligned.
7241 Some assemblers only support this flag when @var{n} is a power of two;
7242 in that case, it is rounded up.
7244 If @var{n} is not specified or is zero, use a machine-dependent default.
7246 Enabled at levels @option{-O2}, @option{-O3}.
7248 @item -falign-labels
7249 @itemx -falign-labels=@var{n}
7250 @opindex falign-labels
7251 Align all branch targets to a power-of-two boundary, skipping up to
7252 @var{n} bytes like @option{-falign-functions}. This option can easily
7253 make code slower, because it must insert dummy operations for when the
7254 branch target is reached in the usual flow of the code.
7256 @option{-fno-align-labels} and @option{-falign-labels=1} are
7257 equivalent and mean that labels will not be aligned.
7259 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
7260 are greater than this value, then their values are used instead.
7262 If @var{n} is not specified or is zero, use a machine-dependent default
7263 which is very likely to be @samp{1}, meaning no alignment.
7265 Enabled at levels @option{-O2}, @option{-O3}.
7268 @itemx -falign-loops=@var{n}
7269 @opindex falign-loops
7270 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
7271 like @option{-falign-functions}. The hope is that the loop will be
7272 executed many times, which will make up for any execution of the dummy
7275 @option{-fno-align-loops} and @option{-falign-loops=1} are
7276 equivalent and mean that loops will not be aligned.
7278 If @var{n} is not specified or is zero, use a machine-dependent default.
7280 Enabled at levels @option{-O2}, @option{-O3}.
7283 @itemx -falign-jumps=@var{n}
7284 @opindex falign-jumps
7285 Align branch targets to a power-of-two boundary, for branch targets
7286 where the targets can only be reached by jumping, skipping up to @var{n}
7287 bytes like @option{-falign-functions}. In this case, no dummy operations
7290 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
7291 equivalent and mean that loops will not be aligned.
7293 If @var{n} is not specified or is zero, use a machine-dependent default.
7295 Enabled at levels @option{-O2}, @option{-O3}.
7297 @item -funit-at-a-time
7298 @opindex funit-at-a-time
7299 This option is left for compatibility reasons. @option{-funit-at-a-time}
7300 has no effect, while @option{-fno-unit-at-a-time} implies
7301 @option{-fno-toplevel-reorder} and @option{-fno-section-anchors}.
7305 @item -fno-toplevel-reorder
7306 @opindex fno-toplevel-reorder
7307 Do not reorder top-level functions, variables, and @code{asm}
7308 statements. Output them in the same order that they appear in the
7309 input file. When this option is used, unreferenced static variables
7310 will not be removed. This option is intended to support existing code
7311 which relies on a particular ordering. For new code, it is better to
7314 Enabled at level @option{-O0}. When disabled explicitly, it also imply
7315 @option{-fno-section-anchors} that is otherwise enabled at @option{-O0} on some
7320 Constructs webs as commonly used for register allocation purposes and assign
7321 each web individual pseudo register. This allows the register allocation pass
7322 to operate on pseudos directly, but also strengthens several other optimization
7323 passes, such as CSE, loop optimizer and trivial dead code remover. It can,
7324 however, make debugging impossible, since variables will no longer stay in a
7327 Enabled by default with @option{-funroll-loops}.
7329 @item -fwhole-program
7330 @opindex fwhole-program
7331 Assume that the current compilation unit represents the whole program being
7332 compiled. All public functions and variables with the exception of @code{main}
7333 and those merged by attribute @code{externally_visible} become static functions
7334 and in effect are optimized more aggressively by interprocedural optimizers. If @command{gold} is used as the linker plugin, @code{externally_visible} attributes are automatically added to functions (not variable yet due to a current @command{gold} issue) that are accessed outside of LTO objects according to resolution file produced by @command{gold}. For other linkers that cannot generate resolution file, explicit @code{externally_visible} attributes are still necessary.
7335 While this option is equivalent to proper use of the @code{static} keyword for
7336 programs consisting of a single file, in combination with option
7337 @option{-combine}, @option{-flto} or @option{-fwhopr} this flag can be used to
7338 compile many smaller scale programs since the functions and variables become
7339 local for the whole combined compilation unit, not for the single source file
7342 This option implies @option{-fwhole-file} for Fortran programs.
7346 This option runs the standard link-time optimizer. When invoked
7347 with source code, it generates GIMPLE (one of GCC's internal
7348 representations) and writes it to special ELF sections in the object
7349 file. When the object files are linked together, all the function
7350 bodies are read from these ELF sections and instantiated as if they
7351 had been part of the same translation unit.
7353 To use the link-timer optimizer, @option{-flto} needs to be specified at
7354 compile time and during the final link. For example,
7357 gcc -c -O2 -flto foo.c
7358 gcc -c -O2 -flto bar.c
7359 gcc -o myprog -flto -O2 foo.o bar.o
7362 The first two invocations to GCC will save a bytecode representation
7363 of GIMPLE into special ELF sections inside @file{foo.o} and
7364 @file{bar.o}. The final invocation will read the GIMPLE bytecode from
7365 @file{foo.o} and @file{bar.o}, merge the two files into a single
7366 internal image, and compile the result as usual. Since both
7367 @file{foo.o} and @file{bar.o} are merged into a single image, this
7368 causes all the inter-procedural analyses and optimizations in GCC to
7369 work across the two files as if they were a single one. This means,
7370 for example, that the inliner will be able to inline functions in
7371 @file{bar.o} into functions in @file{foo.o} and vice-versa.
7373 Another (simpler) way to enable link-time optimization is,
7376 gcc -o myprog -flto -O2 foo.c bar.c
7379 The above will generate bytecode for @file{foo.c} and @file{bar.c},
7380 merge them together into a single GIMPLE representation and optimize
7381 them as usual to produce @file{myprog}.
7383 The only important thing to keep in mind is that to enable link-time
7384 optimizations the @option{-flto} flag needs to be passed to both the
7385 compile and the link commands.
7387 Note that when a file is compiled with @option{-flto}, the generated
7388 object file will be larger than a regular object file because it will
7389 contain GIMPLE bytecodes and the usual final code. This means that
7390 object files with LTO information can be linked as a normal object
7391 file. So, in the previous example, if the final link is done with
7394 gcc -o myprog foo.o bar.o
7397 The only difference will be that no inter-procedural optimizations
7398 will be applied to produce @file{myprog}. The two object files
7399 @file{foo.o} and @file{bar.o} will be simply sent to the regular
7402 Additionally, the optimization flags used to compile individual files
7403 are not necessarily related to those used at link-time. For instance,
7406 gcc -c -O0 -flto foo.c
7407 gcc -c -O0 -flto bar.c
7408 gcc -o myprog -flto -O3 foo.o bar.o
7411 This will produce individual object files with unoptimized assembler
7412 code, but the resulting binary @file{myprog} will be optimized at
7413 @option{-O3}. Now, if the final binary is generated without
7414 @option{-flto}, then @file{myprog} will not be optimized.
7416 When producing the final binary with @option{-flto}, GCC will only
7417 apply link-time optimizations to those files that contain bytecode.
7418 Therefore, you can mix and match object files and libraries with
7419 GIMPLE bytecodes and final object code. GCC will automatically select
7420 which files to optimize in LTO mode and which files to link without
7423 There are some code generation flags that GCC will preserve when
7424 generating bytecodes, as they need to be used during the final link
7425 stage. Currently, the following options are saved into the GIMPLE
7426 bytecode files: @option{-fPIC}, @option{-fcommon} and all the
7427 @option{-m} target flags.
7429 At link time, these options are read-in and reapplied. Note that the
7430 current implementation makes no attempt at recognizing conflicting
7431 values for these options. If two or more files have a conflicting
7432 value (e.g., one file is compiled with @option{-fPIC} and another
7433 isn't), the compiler will simply use the last value read from the
7434 bytecode files. It is recommended, then, that all the files
7435 participating in the same link be compiled with the same options.
7437 Another feature of LTO is that it is possible to apply interprocedural
7438 optimizations on files written in different languages. This requires
7439 some support in the language front end. Currently, the C, C++ and
7440 Fortran front ends are capable of emitting GIMPLE bytecodes, so
7441 something like this should work
7446 gfortran -c -flto baz.f90
7447 g++ -o myprog -flto -O3 foo.o bar.o baz.o -lgfortran
7450 Notice that the final link is done with @command{g++} to get the C++
7451 runtime libraries and @option{-lgfortran} is added to get the Fortran
7452 runtime libraries. In general, when mixing languages in LTO mode, you
7453 should use the same link command used when mixing languages in a
7454 regular (non-LTO) compilation. This means that if your build process
7455 was mixing languages before, all you need to add is @option{-flto} to
7456 all the compile and link commands.
7458 If LTO encounters objects with C linkage declared with incompatible
7459 types in separate translation units to be linked together (undefined
7460 behavior according to ISO C99 6.2.7), a non-fatal diagnostic may be
7461 issued. The behavior is still undefined at runtime.
7463 If object files containing GIMPLE bytecode are stored in a library
7464 archive, say @file{libfoo.a}, it is possible to extract and use them
7465 in an LTO link if you are using @command{gold} as the linker (which,
7466 in turn requires GCC to be configured with @option{--enable-gold}).
7467 To enable this feature, use the flag @option{-fuse-linker-plugin} at
7471 gcc -o myprog -O2 -flto -fuse-linker-plugin a.o b.o -lfoo
7474 With the linker plugin enabled, @command{gold} will extract the needed
7475 GIMPLE files from @file{libfoo.a} and pass them on to the running GCC
7476 to make them part of the aggregated GIMPLE image to be optimized.
7478 If you are not using @command{gold} and/or do not specify
7479 @option{-fuse-linker-plugin} then the objects inside @file{libfoo.a}
7480 will be extracted and linked as usual, but they will not participate
7481 in the LTO optimization process.
7483 Link time optimizations do not require the presence of the whole
7484 program to operate. If the program does not require any symbols to
7485 be exported, it is possible to combine @option{-flto} and
7486 @option{-fwhopr} with @option{-fwhole-program} to allow the
7487 interprocedural optimizers to use more aggressive assumptions which
7488 may lead to improved optimization opportunities.
7490 Regarding portability: the current implementation of LTO makes no
7491 attempt at generating bytecode that can be ported between different
7492 types of hosts. The bytecode files are versioned and there is a
7493 strict version check, so bytecode files generated in one version of
7494 GCC will not work with an older/newer version of GCC.
7496 Link time optimization does not play well with generating debugging
7497 information. Combining @option{-flto} or @option{-fwhopr} with
7498 @option{-g} is experimental.
7500 This option is disabled by default.
7502 @item -fwhopr[=@var{n}]
7504 This option is identical in functionality to @option{-flto} but it
7505 differs in how the final link stage is executed. Instead of loading
7506 all the function bodies in memory, the callgraph is analyzed and
7507 optimization decisions are made (whole program analysis or WPA). Once
7508 optimization decisions are made, the callgraph is partitioned and the
7509 different sections are compiled separately (local transformations or
7510 LTRANS)@. This process allows optimizations on very large programs
7511 that otherwise would not fit in memory. This option enables
7512 @option{-fwpa} and @option{-fltrans} automatically.
7514 If you specify the optional @var{n} the link stage is executed in
7515 parallel using @var{n} parallel jobs by utilizing an installed
7516 @command{make} program. The environment variable @env{MAKE} may be
7517 used to override the program used.
7519 Disabled by default.
7523 This is an internal option used by GCC when compiling with
7524 @option{-fwhopr}. You should never need to use it.
7526 This option runs the link-time optimizer in the whole-program-analysis
7527 (WPA) mode, which reads in summary information from all inputs and
7528 performs a whole-program analysis based on summary information only.
7529 It generates object files for subsequent runs of the link-time
7530 optimizer where individual object files are optimized using both
7531 summary information from the WPA mode and the actual function bodies.
7532 It then drives the LTRANS phase.
7534 Disabled by default.
7538 This is an internal option used by GCC when compiling with
7539 @option{-fwhopr}. You should never need to use it.
7541 This option runs the link-time optimizer in the local-transformation (LTRANS)
7542 mode, which reads in output from a previous run of the LTO in WPA mode.
7543 In the LTRANS mode, LTO optimizes an object and produces the final assembly.
7545 Disabled by default.
7547 @item -fltrans-output-list=@var{file}
7548 @opindex fltrans-output-list
7549 This is an internal option used by GCC when compiling with
7550 @option{-fwhopr}. You should never need to use it.
7552 This option specifies a file to which the names of LTRANS output files are
7553 written. This option is only meaningful in conjunction with @option{-fwpa}.
7555 Disabled by default.
7557 @item -flto-compression-level=@var{n}
7558 This option specifies the level of compression used for intermediate
7559 language written to LTO object files, and is only meaningful in
7560 conjunction with LTO mode (@option{-fwhopr}, @option{-flto}). Valid
7561 values are 0 (no compression) to 9 (maximum compression). Values
7562 outside this range are clamped to either 0 or 9. If the option is not
7563 given, a default balanced compression setting is used.
7566 Prints a report with internal details on the workings of the link-time
7567 optimizer. The contents of this report vary from version to version,
7568 it is meant to be useful to GCC developers when processing object
7569 files in LTO mode (via @option{-fwhopr} or @option{-flto}).
7571 Disabled by default.
7573 @item -fuse-linker-plugin
7574 Enables the extraction of objects with GIMPLE bytecode information
7575 from library archives. This option relies on features available only
7576 in @command{gold}, so to use this you must configure GCC with
7577 @option{--enable-gold}. See @option{-flto} for a description on the
7578 effect of this flag and how to use it.
7580 Disabled by default.
7582 @item -fcprop-registers
7583 @opindex fcprop-registers
7584 After register allocation and post-register allocation instruction splitting,
7585 we perform a copy-propagation pass to try to reduce scheduling dependencies
7586 and occasionally eliminate the copy.
7588 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
7590 @item -fprofile-correction
7591 @opindex fprofile-correction
7592 Profiles collected using an instrumented binary for multi-threaded programs may
7593 be inconsistent due to missed counter updates. When this option is specified,
7594 GCC will use heuristics to correct or smooth out such inconsistencies. By
7595 default, GCC will emit an error message when an inconsistent profile is detected.
7597 @item -fprofile-dir=@var{path}
7598 @opindex fprofile-dir
7600 Set the directory to search the profile data files in to @var{path}.
7601 This option affects only the profile data generated by
7602 @option{-fprofile-generate}, @option{-ftest-coverage}, @option{-fprofile-arcs}
7603 and used by @option{-fprofile-use} and @option{-fbranch-probabilities}
7604 and its related options.
7605 By default, GCC will use the current directory as @var{path}
7606 thus the profile data file will appear in the same directory as the object file.
7608 @item -fprofile-generate
7609 @itemx -fprofile-generate=@var{path}
7610 @opindex fprofile-generate
7612 Enable options usually used for instrumenting application to produce
7613 profile useful for later recompilation with profile feedback based
7614 optimization. You must use @option{-fprofile-generate} both when
7615 compiling and when linking your program.
7617 The following options are enabled: @code{-fprofile-arcs}, @code{-fprofile-values}, @code{-fvpt}.
7619 If @var{path} is specified, GCC will look at the @var{path} to find
7620 the profile feedback data files. See @option{-fprofile-dir}.
7623 @itemx -fprofile-use=@var{path}
7624 @opindex fprofile-use
7625 Enable profile feedback directed optimizations, and optimizations
7626 generally profitable only with profile feedback available.
7628 The following options are enabled: @code{-fbranch-probabilities}, @code{-fvpt},
7629 @code{-funroll-loops}, @code{-fpeel-loops}, @code{-ftracer}
7631 By default, GCC emits an error message if the feedback profiles do not
7632 match the source code. This error can be turned into a warning by using
7633 @option{-Wcoverage-mismatch}. Note this may result in poorly optimized
7636 If @var{path} is specified, GCC will look at the @var{path} to find
7637 the profile feedback data files. See @option{-fprofile-dir}.
7640 The following options control compiler behavior regarding floating
7641 point arithmetic. These options trade off between speed and
7642 correctness. All must be specifically enabled.
7646 @opindex ffloat-store
7647 Do not store floating point variables in registers, and inhibit other
7648 options that might change whether a floating point value is taken from a
7651 @cindex floating point precision
7652 This option prevents undesirable excess precision on machines such as
7653 the 68000 where the floating registers (of the 68881) keep more
7654 precision than a @code{double} is supposed to have. Similarly for the
7655 x86 architecture. For most programs, the excess precision does only
7656 good, but a few programs rely on the precise definition of IEEE floating
7657 point. Use @option{-ffloat-store} for such programs, after modifying
7658 them to store all pertinent intermediate computations into variables.
7660 @item -fexcess-precision=@var{style}
7661 @opindex fexcess-precision
7662 This option allows further control over excess precision on machines
7663 where floating-point registers have more precision than the IEEE
7664 @code{float} and @code{double} types and the processor does not
7665 support operations rounding to those types. By default,
7666 @option{-fexcess-precision=fast} is in effect; this means that
7667 operations are carried out in the precision of the registers and that
7668 it is unpredictable when rounding to the types specified in the source
7669 code takes place. When compiling C, if
7670 @option{-fexcess-precision=standard} is specified then excess
7671 precision will follow the rules specified in ISO C99; in particular,
7672 both casts and assignments cause values to be rounded to their
7673 semantic types (whereas @option{-ffloat-store} only affects
7674 assignments). This option is enabled by default for C if a strict
7675 conformance option such as @option{-std=c99} is used.
7678 @option{-fexcess-precision=standard} is not implemented for languages
7679 other than C, and has no effect if
7680 @option{-funsafe-math-optimizations} or @option{-ffast-math} is
7681 specified. On the x86, it also has no effect if @option{-mfpmath=sse}
7682 or @option{-mfpmath=sse+387} is specified; in the former case, IEEE
7683 semantics apply without excess precision, and in the latter, rounding
7688 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations},
7689 @option{-ffinite-math-only}, @option{-fno-rounding-math},
7690 @option{-fno-signaling-nans} and @option{-fcx-limited-range}.
7692 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
7694 This option is not turned on by any @option{-O} option since
7695 it can result in incorrect output for programs which depend on
7696 an exact implementation of IEEE or ISO rules/specifications for
7697 math functions. It may, however, yield faster code for programs
7698 that do not require the guarantees of these specifications.
7700 @item -fno-math-errno
7701 @opindex fno-math-errno
7702 Do not set ERRNO after calling math functions that are executed
7703 with a single instruction, e.g., sqrt. A program that relies on
7704 IEEE exceptions for math error handling may want to use this flag
7705 for speed while maintaining IEEE arithmetic compatibility.
7707 This option is not turned on by any @option{-O} option since
7708 it can result in incorrect output for programs which depend on
7709 an exact implementation of IEEE or ISO rules/specifications for
7710 math functions. It may, however, yield faster code for programs
7711 that do not require the guarantees of these specifications.
7713 The default is @option{-fmath-errno}.
7715 On Darwin systems, the math library never sets @code{errno}. There is
7716 therefore no reason for the compiler to consider the possibility that
7717 it might, and @option{-fno-math-errno} is the default.
7719 @item -funsafe-math-optimizations
7720 @opindex funsafe-math-optimizations
7722 Allow optimizations for floating-point arithmetic that (a) assume
7723 that arguments and results are valid and (b) may violate IEEE or
7724 ANSI standards. When used at link-time, it may include libraries
7725 or startup files that change the default FPU control word or other
7726 similar optimizations.
7728 This option is not turned on by any @option{-O} option since
7729 it can result in incorrect output for programs which depend on
7730 an exact implementation of IEEE or ISO rules/specifications for
7731 math functions. It may, however, yield faster code for programs
7732 that do not require the guarantees of these specifications.
7733 Enables @option{-fno-signed-zeros}, @option{-fno-trapping-math},
7734 @option{-fassociative-math} and @option{-freciprocal-math}.
7736 The default is @option{-fno-unsafe-math-optimizations}.
7738 @item -fassociative-math
7739 @opindex fassociative-math
7741 Allow re-association of operands in series of floating-point operations.
7742 This violates the ISO C and C++ language standard by possibly changing
7743 computation result. NOTE: re-ordering may change the sign of zero as
7744 well as ignore NaNs and inhibit or create underflow or overflow (and
7745 thus cannot be used on a code which relies on rounding behavior like
7746 @code{(x + 2**52) - 2**52)}. May also reorder floating-point comparisons
7747 and thus may not be used when ordered comparisons are required.
7748 This option requires that both @option{-fno-signed-zeros} and
7749 @option{-fno-trapping-math} be in effect. Moreover, it doesn't make
7750 much sense with @option{-frounding-math}. For Fortran the option
7751 is automatically enabled when both @option{-fno-signed-zeros} and
7752 @option{-fno-trapping-math} are in effect.
7754 The default is @option{-fno-associative-math}.
7756 @item -freciprocal-math
7757 @opindex freciprocal-math
7759 Allow the reciprocal of a value to be used instead of dividing by
7760 the value if this enables optimizations. For example @code{x / y}
7761 can be replaced with @code{x * (1/y)} which is useful if @code{(1/y)}
7762 is subject to common subexpression elimination. Note that this loses
7763 precision and increases the number of flops operating on the value.
7765 The default is @option{-fno-reciprocal-math}.
7767 @item -ffinite-math-only
7768 @opindex ffinite-math-only
7769 Allow optimizations for floating-point arithmetic that assume
7770 that arguments and results are not NaNs or +-Infs.
7772 This option is not turned on by any @option{-O} option since
7773 it can result in incorrect output for programs which depend on
7774 an exact implementation of IEEE or ISO rules/specifications for
7775 math functions. It may, however, yield faster code for programs
7776 that do not require the guarantees of these specifications.
7778 The default is @option{-fno-finite-math-only}.
7780 @item -fno-signed-zeros
7781 @opindex fno-signed-zeros
7782 Allow optimizations for floating point arithmetic that ignore the
7783 signedness of zero. IEEE arithmetic specifies the behavior of
7784 distinct +0.0 and @minus{}0.0 values, which then prohibits simplification
7785 of expressions such as x+0.0 or 0.0*x (even with @option{-ffinite-math-only}).
7786 This option implies that the sign of a zero result isn't significant.
7788 The default is @option{-fsigned-zeros}.
7790 @item -fno-trapping-math
7791 @opindex fno-trapping-math
7792 Compile code assuming that floating-point operations cannot generate
7793 user-visible traps. These traps include division by zero, overflow,
7794 underflow, inexact result and invalid operation. This option requires
7795 that @option{-fno-signaling-nans} be in effect. Setting this option may
7796 allow faster code if one relies on ``non-stop'' IEEE arithmetic, for example.
7798 This option should never be turned on by any @option{-O} option since
7799 it can result in incorrect output for programs which depend on
7800 an exact implementation of IEEE or ISO rules/specifications for
7803 The default is @option{-ftrapping-math}.
7805 @item -frounding-math
7806 @opindex frounding-math
7807 Disable transformations and optimizations that assume default floating
7808 point rounding behavior. This is round-to-zero for all floating point
7809 to integer conversions, and round-to-nearest for all other arithmetic
7810 truncations. This option should be specified for programs that change
7811 the FP rounding mode dynamically, or that may be executed with a
7812 non-default rounding mode. This option disables constant folding of
7813 floating point expressions at compile-time (which may be affected by
7814 rounding mode) and arithmetic transformations that are unsafe in the
7815 presence of sign-dependent rounding modes.
7817 The default is @option{-fno-rounding-math}.
7819 This option is experimental and does not currently guarantee to
7820 disable all GCC optimizations that are affected by rounding mode.
7821 Future versions of GCC may provide finer control of this setting
7822 using C99's @code{FENV_ACCESS} pragma. This command line option
7823 will be used to specify the default state for @code{FENV_ACCESS}.
7825 @item -fsignaling-nans
7826 @opindex fsignaling-nans
7827 Compile code assuming that IEEE signaling NaNs may generate user-visible
7828 traps during floating-point operations. Setting this option disables
7829 optimizations that may change the number of exceptions visible with
7830 signaling NaNs. This option implies @option{-ftrapping-math}.
7832 This option causes the preprocessor macro @code{__SUPPORT_SNAN__} to
7835 The default is @option{-fno-signaling-nans}.
7837 This option is experimental and does not currently guarantee to
7838 disable all GCC optimizations that affect signaling NaN behavior.
7840 @item -fsingle-precision-constant
7841 @opindex fsingle-precision-constant
7842 Treat floating point constant as single precision constant instead of
7843 implicitly converting it to double precision constant.
7845 @item -fcx-limited-range
7846 @opindex fcx-limited-range
7847 When enabled, this option states that a range reduction step is not
7848 needed when performing complex division. Also, there is no checking
7849 whether the result of a complex multiplication or division is @code{NaN
7850 + I*NaN}, with an attempt to rescue the situation in that case. The
7851 default is @option{-fno-cx-limited-range}, but is enabled by
7852 @option{-ffast-math}.
7854 This option controls the default setting of the ISO C99
7855 @code{CX_LIMITED_RANGE} pragma. Nevertheless, the option applies to
7858 @item -fcx-fortran-rules
7859 @opindex fcx-fortran-rules
7860 Complex multiplication and division follow Fortran rules. Range
7861 reduction is done as part of complex division, but there is no checking
7862 whether the result of a complex multiplication or division is @code{NaN
7863 + I*NaN}, with an attempt to rescue the situation in that case.
7865 The default is @option{-fno-cx-fortran-rules}.
7869 The following options control optimizations that may improve
7870 performance, but are not enabled by any @option{-O} options. This
7871 section includes experimental options that may produce broken code.
7874 @item -fbranch-probabilities
7875 @opindex fbranch-probabilities
7876 After running a program compiled with @option{-fprofile-arcs}
7877 (@pxref{Debugging Options,, Options for Debugging Your Program or
7878 @command{gcc}}), you can compile it a second time using
7879 @option{-fbranch-probabilities}, to improve optimizations based on
7880 the number of times each branch was taken. When the program
7881 compiled with @option{-fprofile-arcs} exits it saves arc execution
7882 counts to a file called @file{@var{sourcename}.gcda} for each source
7883 file. The information in this data file is very dependent on the
7884 structure of the generated code, so you must use the same source code
7885 and the same optimization options for both compilations.
7887 With @option{-fbranch-probabilities}, GCC puts a
7888 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
7889 These can be used to improve optimization. Currently, they are only
7890 used in one place: in @file{reorg.c}, instead of guessing which path a
7891 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
7892 exactly determine which path is taken more often.
7894 @item -fprofile-values
7895 @opindex fprofile-values
7896 If combined with @option{-fprofile-arcs}, it adds code so that some
7897 data about values of expressions in the program is gathered.
7899 With @option{-fbranch-probabilities}, it reads back the data gathered
7900 from profiling values of expressions and adds @samp{REG_VALUE_PROFILE}
7901 notes to instructions for their later usage in optimizations.
7903 Enabled with @option{-fprofile-generate} and @option{-fprofile-use}.
7907 If combined with @option{-fprofile-arcs}, it instructs the compiler to add
7908 a code to gather information about values of expressions.
7910 With @option{-fbranch-probabilities}, it reads back the data gathered
7911 and actually performs the optimizations based on them.
7912 Currently the optimizations include specialization of division operation
7913 using the knowledge about the value of the denominator.
7915 @item -frename-registers
7916 @opindex frename-registers
7917 Attempt to avoid false dependencies in scheduled code by making use
7918 of registers left over after register allocation. This optimization
7919 will most benefit processors with lots of registers. Depending on the
7920 debug information format adopted by the target, however, it can
7921 make debugging impossible, since variables will no longer stay in
7922 a ``home register''.
7924 Enabled by default with @option{-funroll-loops} and @option{-fpeel-loops}.
7928 Perform tail duplication to enlarge superblock size. This transformation
7929 simplifies the control flow of the function allowing other optimizations to do
7932 Enabled with @option{-fprofile-use}.
7934 @item -funroll-loops
7935 @opindex funroll-loops
7936 Unroll loops whose number of iterations can be determined at compile time or
7937 upon entry to the loop. @option{-funroll-loops} implies
7938 @option{-frerun-cse-after-loop}, @option{-fweb} and @option{-frename-registers}.
7939 It also turns on complete loop peeling (i.e.@: complete removal of loops with
7940 small constant number of iterations). This option makes code larger, and may
7941 or may not make it run faster.
7943 Enabled with @option{-fprofile-use}.
7945 @item -funroll-all-loops
7946 @opindex funroll-all-loops
7947 Unroll all loops, even if their number of iterations is uncertain when
7948 the loop is entered. This usually makes programs run more slowly.
7949 @option{-funroll-all-loops} implies the same options as
7950 @option{-funroll-loops}.
7953 @opindex fpeel-loops
7954 Peels the loops for that there is enough information that they do not
7955 roll much (from profile feedback). It also turns on complete loop peeling
7956 (i.e.@: complete removal of loops with small constant number of iterations).
7958 Enabled with @option{-fprofile-use}.
7960 @item -fmove-loop-invariants
7961 @opindex fmove-loop-invariants
7962 Enables the loop invariant motion pass in the RTL loop optimizer. Enabled
7963 at level @option{-O1}
7965 @item -funswitch-loops
7966 @opindex funswitch-loops
7967 Move branches with loop invariant conditions out of the loop, with duplicates
7968 of the loop on both branches (modified according to result of the condition).
7970 @item -ffunction-sections
7971 @itemx -fdata-sections
7972 @opindex ffunction-sections
7973 @opindex fdata-sections
7974 Place each function or data item into its own section in the output
7975 file if the target supports arbitrary sections. The name of the
7976 function or the name of the data item determines the section's name
7979 Use these options on systems where the linker can perform optimizations
7980 to improve locality of reference in the instruction space. Most systems
7981 using the ELF object format and SPARC processors running Solaris 2 have
7982 linkers with such optimizations. AIX may have these optimizations in
7985 Only use these options when there are significant benefits from doing
7986 so. When you specify these options, the assembler and linker will
7987 create larger object and executable files and will also be slower.
7988 You will not be able to use @code{gprof} on all systems if you
7989 specify this option and you may have problems with debugging if
7990 you specify both this option and @option{-g}.
7992 @item -fbranch-target-load-optimize
7993 @opindex fbranch-target-load-optimize
7994 Perform branch target register load optimization before prologue / epilogue
7996 The use of target registers can typically be exposed only during reload,
7997 thus hoisting loads out of loops and doing inter-block scheduling needs
7998 a separate optimization pass.
8000 @item -fbranch-target-load-optimize2
8001 @opindex fbranch-target-load-optimize2
8002 Perform branch target register load optimization after prologue / epilogue
8005 @item -fbtr-bb-exclusive
8006 @opindex fbtr-bb-exclusive
8007 When performing branch target register load optimization, don't reuse
8008 branch target registers in within any basic block.
8010 @item -fstack-protector
8011 @opindex fstack-protector
8012 Emit extra code to check for buffer overflows, such as stack smashing
8013 attacks. This is done by adding a guard variable to functions with
8014 vulnerable objects. This includes functions that call alloca, and
8015 functions with buffers larger than 8 bytes. The guards are initialized
8016 when a function is entered and then checked when the function exits.
8017 If a guard check fails, an error message is printed and the program exits.
8019 @item -fstack-protector-all
8020 @opindex fstack-protector-all
8021 Like @option{-fstack-protector} except that all functions are protected.
8023 @item -fsection-anchors
8024 @opindex fsection-anchors
8025 Try to reduce the number of symbolic address calculations by using
8026 shared ``anchor'' symbols to address nearby objects. This transformation
8027 can help to reduce the number of GOT entries and GOT accesses on some
8030 For example, the implementation of the following function @code{foo}:
8034 int foo (void) @{ return a + b + c; @}
8037 would usually calculate the addresses of all three variables, but if you
8038 compile it with @option{-fsection-anchors}, it will access the variables
8039 from a common anchor point instead. The effect is similar to the
8040 following pseudocode (which isn't valid C):
8045 register int *xr = &x;
8046 return xr[&a - &x] + xr[&b - &x] + xr[&c - &x];
8050 Not all targets support this option.
8052 @item --param @var{name}=@var{value}
8054 In some places, GCC uses various constants to control the amount of
8055 optimization that is done. For example, GCC will not inline functions
8056 that contain more that a certain number of instructions. You can
8057 control some of these constants on the command-line using the
8058 @option{--param} option.
8060 The names of specific parameters, and the meaning of the values, are
8061 tied to the internals of the compiler, and are subject to change
8062 without notice in future releases.
8064 In each case, the @var{value} is an integer. The allowable choices for
8065 @var{name} are given in the following table:
8068 @item struct-reorg-cold-struct-ratio
8069 The threshold ratio (as a percentage) between a structure frequency
8070 and the frequency of the hottest structure in the program. This parameter
8071 is used by struct-reorg optimization enabled by @option{-fipa-struct-reorg}.
8072 We say that if the ratio of a structure frequency, calculated by profiling,
8073 to the hottest structure frequency in the program is less than this
8074 parameter, then structure reorganization is not applied to this structure.
8077 @item predictable-branch-outcome
8078 When branch is predicted to be taken with probability lower than this threshold
8079 (in percent), then it is considered well predictable. The default is 10.
8081 @item max-crossjump-edges
8082 The maximum number of incoming edges to consider for crossjumping.
8083 The algorithm used by @option{-fcrossjumping} is @math{O(N^2)} in
8084 the number of edges incoming to each block. Increasing values mean
8085 more aggressive optimization, making the compile time increase with
8086 probably small improvement in executable size.
8088 @item min-crossjump-insns
8089 The minimum number of instructions which must be matched at the end
8090 of two blocks before crossjumping will be performed on them. This
8091 value is ignored in the case where all instructions in the block being
8092 crossjumped from are matched. The default value is 5.
8094 @item max-grow-copy-bb-insns
8095 The maximum code size expansion factor when copying basic blocks
8096 instead of jumping. The expansion is relative to a jump instruction.
8097 The default value is 8.
8099 @item max-goto-duplication-insns
8100 The maximum number of instructions to duplicate to a block that jumps
8101 to a computed goto. To avoid @math{O(N^2)} behavior in a number of
8102 passes, GCC factors computed gotos early in the compilation process,
8103 and unfactors them as late as possible. Only computed jumps at the
8104 end of a basic blocks with no more than max-goto-duplication-insns are
8105 unfactored. The default value is 8.
8107 @item max-delay-slot-insn-search
8108 The maximum number of instructions to consider when looking for an
8109 instruction to fill a delay slot. If more than this arbitrary number of
8110 instructions is searched, the time savings from filling the delay slot
8111 will be minimal so stop searching. Increasing values mean more
8112 aggressive optimization, making the compile time increase with probably
8113 small improvement in executable run time.
8115 @item max-delay-slot-live-search
8116 When trying to fill delay slots, the maximum number of instructions to
8117 consider when searching for a block with valid live register
8118 information. Increasing this arbitrarily chosen value means more
8119 aggressive optimization, increasing the compile time. This parameter
8120 should be removed when the delay slot code is rewritten to maintain the
8123 @item max-gcse-memory
8124 The approximate maximum amount of memory that will be allocated in
8125 order to perform the global common subexpression elimination
8126 optimization. If more memory than specified is required, the
8127 optimization will not be done.
8129 @item max-pending-list-length
8130 The maximum number of pending dependencies scheduling will allow
8131 before flushing the current state and starting over. Large functions
8132 with few branches or calls can create excessively large lists which
8133 needlessly consume memory and resources.
8135 @item max-inline-insns-single
8136 Several parameters control the tree inliner used in gcc.
8137 This number sets the maximum number of instructions (counted in GCC's
8138 internal representation) in a single function that the tree inliner
8139 will consider for inlining. This only affects functions declared
8140 inline and methods implemented in a class declaration (C++).
8141 The default value is 300.
8143 @item max-inline-insns-auto
8144 When you use @option{-finline-functions} (included in @option{-O3}),
8145 a lot of functions that would otherwise not be considered for inlining
8146 by the compiler will be investigated. To those functions, a different
8147 (more restrictive) limit compared to functions declared inline can
8149 The default value is 40.
8151 @item large-function-insns
8152 The limit specifying really large functions. For functions larger than this
8153 limit after inlining, inlining is constrained by
8154 @option{--param large-function-growth}. This parameter is useful primarily
8155 to avoid extreme compilation time caused by non-linear algorithms used by the
8157 The default value is 2700.
8159 @item large-function-growth
8160 Specifies maximal growth of large function caused by inlining in percents.
8161 The default value is 100 which limits large function growth to 2.0 times
8164 @item large-unit-insns
8165 The limit specifying large translation unit. Growth caused by inlining of
8166 units larger than this limit is limited by @option{--param inline-unit-growth}.
8167 For small units this might be too tight (consider unit consisting of function A
8168 that is inline and B that just calls A three time. If B is small relative to
8169 A, the growth of unit is 300\% and yet such inlining is very sane. For very
8170 large units consisting of small inlineable functions however the overall unit
8171 growth limit is needed to avoid exponential explosion of code size. Thus for
8172 smaller units, the size is increased to @option{--param large-unit-insns}
8173 before applying @option{--param inline-unit-growth}. The default is 10000
8175 @item inline-unit-growth
8176 Specifies maximal overall growth of the compilation unit caused by inlining.
8177 The default value is 30 which limits unit growth to 1.3 times the original
8180 @item ipcp-unit-growth
8181 Specifies maximal overall growth of the compilation unit caused by
8182 interprocedural constant propagation. The default value is 10 which limits
8183 unit growth to 1.1 times the original size.
8185 @item large-stack-frame
8186 The limit specifying large stack frames. While inlining the algorithm is trying
8187 to not grow past this limit too much. Default value is 256 bytes.
8189 @item large-stack-frame-growth
8190 Specifies maximal growth of large stack frames caused by inlining in percents.
8191 The default value is 1000 which limits large stack frame growth to 11 times
8194 @item max-inline-insns-recursive
8195 @itemx max-inline-insns-recursive-auto
8196 Specifies maximum number of instructions out-of-line copy of self recursive inline
8197 function can grow into by performing recursive inlining.
8199 For functions declared inline @option{--param max-inline-insns-recursive} is
8200 taken into account. For function not declared inline, recursive inlining
8201 happens only when @option{-finline-functions} (included in @option{-O3}) is
8202 enabled and @option{--param max-inline-insns-recursive-auto} is used. The
8203 default value is 450.
8205 @item max-inline-recursive-depth
8206 @itemx max-inline-recursive-depth-auto
8207 Specifies maximum recursion depth used by the recursive inlining.
8209 For functions declared inline @option{--param max-inline-recursive-depth} is
8210 taken into account. For function not declared inline, recursive inlining
8211 happens only when @option{-finline-functions} (included in @option{-O3}) is
8212 enabled and @option{--param max-inline-recursive-depth-auto} is used. The
8215 @item min-inline-recursive-probability
8216 Recursive inlining is profitable only for function having deep recursion
8217 in average and can hurt for function having little recursion depth by
8218 increasing the prologue size or complexity of function body to other
8221 When profile feedback is available (see @option{-fprofile-generate}) the actual
8222 recursion depth can be guessed from probability that function will recurse via
8223 given call expression. This parameter limits inlining only to call expression
8224 whose probability exceeds given threshold (in percents). The default value is
8227 @item early-inlining-insns
8228 Specify growth that early inliner can make. In effect it increases amount of
8229 inlining for code having large abstraction penalty. The default value is 8.
8231 @item max-early-inliner-iterations
8232 @itemx max-early-inliner-iterations
8233 Limit of iterations of early inliner. This basically bounds number of nested
8234 indirect calls early inliner can resolve. Deeper chains are still handled by
8237 @item min-vect-loop-bound
8238 The minimum number of iterations under which a loop will not get vectorized
8239 when @option{-ftree-vectorize} is used. The number of iterations after
8240 vectorization needs to be greater than the value specified by this option
8241 to allow vectorization. The default value is 0.
8243 @item max-unrolled-insns
8244 The maximum number of instructions that a loop should have if that loop
8245 is unrolled, and if the loop is unrolled, it determines how many times
8246 the loop code is unrolled.
8248 @item max-average-unrolled-insns
8249 The maximum number of instructions biased by probabilities of their execution
8250 that a loop should have if that loop is unrolled, and if the loop is unrolled,
8251 it determines how many times the loop code is unrolled.
8253 @item max-unroll-times
8254 The maximum number of unrollings of a single loop.
8256 @item max-peeled-insns
8257 The maximum number of instructions that a loop should have if that loop
8258 is peeled, and if the loop is peeled, it determines how many times
8259 the loop code is peeled.
8261 @item max-peel-times
8262 The maximum number of peelings of a single loop.
8264 @item max-completely-peeled-insns
8265 The maximum number of insns of a completely peeled loop.
8267 @item max-completely-peel-times
8268 The maximum number of iterations of a loop to be suitable for complete peeling.
8270 @item max-completely-peel-loop-nest-depth
8271 The maximum depth of a loop nest suitable for complete peeling.
8273 @item max-unswitch-insns
8274 The maximum number of insns of an unswitched loop.
8276 @item max-unswitch-level
8277 The maximum number of branches unswitched in a single loop.
8280 The minimum cost of an expensive expression in the loop invariant motion.
8282 @item iv-consider-all-candidates-bound
8283 Bound on number of candidates for induction variables below that
8284 all candidates are considered for each use in induction variable
8285 optimizations. Only the most relevant candidates are considered
8286 if there are more candidates, to avoid quadratic time complexity.
8288 @item iv-max-considered-uses
8289 The induction variable optimizations give up on loops that contain more
8290 induction variable uses.
8292 @item iv-always-prune-cand-set-bound
8293 If number of candidates in the set is smaller than this value,
8294 we always try to remove unnecessary ivs from the set during its
8295 optimization when a new iv is added to the set.
8297 @item scev-max-expr-size
8298 Bound on size of expressions used in the scalar evolutions analyzer.
8299 Large expressions slow the analyzer.
8301 @item omega-max-vars
8302 The maximum number of variables in an Omega constraint system.
8303 The default value is 128.
8305 @item omega-max-geqs
8306 The maximum number of inequalities in an Omega constraint system.
8307 The default value is 256.
8310 The maximum number of equalities in an Omega constraint system.
8311 The default value is 128.
8313 @item omega-max-wild-cards
8314 The maximum number of wildcard variables that the Omega solver will
8315 be able to insert. The default value is 18.
8317 @item omega-hash-table-size
8318 The size of the hash table in the Omega solver. The default value is
8321 @item omega-max-keys
8322 The maximal number of keys used by the Omega solver. The default
8325 @item omega-eliminate-redundant-constraints
8326 When set to 1, use expensive methods to eliminate all redundant
8327 constraints. The default value is 0.
8329 @item vect-max-version-for-alignment-checks
8330 The maximum number of runtime checks that can be performed when
8331 doing loop versioning for alignment in the vectorizer. See option
8332 ftree-vect-loop-version for more information.
8334 @item vect-max-version-for-alias-checks
8335 The maximum number of runtime checks that can be performed when
8336 doing loop versioning for alias in the vectorizer. See option
8337 ftree-vect-loop-version for more information.
8339 @item max-iterations-to-track
8341 The maximum number of iterations of a loop the brute force algorithm
8342 for analysis of # of iterations of the loop tries to evaluate.
8344 @item hot-bb-count-fraction
8345 Select fraction of the maximal count of repetitions of basic block in program
8346 given basic block needs to have to be considered hot.
8348 @item hot-bb-frequency-fraction
8349 Select fraction of the maximal frequency of executions of basic block in
8350 function given basic block needs to have to be considered hot
8352 @item max-predicted-iterations
8353 The maximum number of loop iterations we predict statically. This is useful
8354 in cases where function contain single loop with known bound and other loop
8355 with unknown. We predict the known number of iterations correctly, while
8356 the unknown number of iterations average to roughly 10. This means that the
8357 loop without bounds would appear artificially cold relative to the other one.
8359 @item align-threshold
8361 Select fraction of the maximal frequency of executions of basic block in
8362 function given basic block will get aligned.
8364 @item align-loop-iterations
8366 A loop expected to iterate at lest the selected number of iterations will get
8369 @item tracer-dynamic-coverage
8370 @itemx tracer-dynamic-coverage-feedback
8372 This value is used to limit superblock formation once the given percentage of
8373 executed instructions is covered. This limits unnecessary code size
8376 The @option{tracer-dynamic-coverage-feedback} is used only when profile
8377 feedback is available. The real profiles (as opposed to statically estimated
8378 ones) are much less balanced allowing the threshold to be larger value.
8380 @item tracer-max-code-growth
8381 Stop tail duplication once code growth has reached given percentage. This is
8382 rather hokey argument, as most of the duplicates will be eliminated later in
8383 cross jumping, so it may be set to much higher values than is the desired code
8386 @item tracer-min-branch-ratio
8388 Stop reverse growth when the reverse probability of best edge is less than this
8389 threshold (in percent).
8391 @item tracer-min-branch-ratio
8392 @itemx tracer-min-branch-ratio-feedback
8394 Stop forward growth if the best edge do have probability lower than this
8397 Similarly to @option{tracer-dynamic-coverage} two values are present, one for
8398 compilation for profile feedback and one for compilation without. The value
8399 for compilation with profile feedback needs to be more conservative (higher) in
8400 order to make tracer effective.
8402 @item max-cse-path-length
8404 Maximum number of basic blocks on path that cse considers. The default is 10.
8407 The maximum instructions CSE process before flushing. The default is 1000.
8409 @item ggc-min-expand
8411 GCC uses a garbage collector to manage its own memory allocation. This
8412 parameter specifies the minimum percentage by which the garbage
8413 collector's heap should be allowed to expand between collections.
8414 Tuning this may improve compilation speed; it has no effect on code
8417 The default is 30% + 70% * (RAM/1GB) with an upper bound of 100% when
8418 RAM >= 1GB@. If @code{getrlimit} is available, the notion of "RAM" is
8419 the smallest of actual RAM and @code{RLIMIT_DATA} or @code{RLIMIT_AS}. If
8420 GCC is not able to calculate RAM on a particular platform, the lower
8421 bound of 30% is used. Setting this parameter and
8422 @option{ggc-min-heapsize} to zero causes a full collection to occur at
8423 every opportunity. This is extremely slow, but can be useful for
8426 @item ggc-min-heapsize
8428 Minimum size of the garbage collector's heap before it begins bothering
8429 to collect garbage. The first collection occurs after the heap expands
8430 by @option{ggc-min-expand}% beyond @option{ggc-min-heapsize}. Again,
8431 tuning this may improve compilation speed, and has no effect on code
8434 The default is the smaller of RAM/8, RLIMIT_RSS, or a limit which
8435 tries to ensure that RLIMIT_DATA or RLIMIT_AS are not exceeded, but
8436 with a lower bound of 4096 (four megabytes) and an upper bound of
8437 131072 (128 megabytes). If GCC is not able to calculate RAM on a
8438 particular platform, the lower bound is used. Setting this parameter
8439 very large effectively disables garbage collection. Setting this
8440 parameter and @option{ggc-min-expand} to zero causes a full collection
8441 to occur at every opportunity.
8443 @item max-reload-search-insns
8444 The maximum number of instruction reload should look backward for equivalent
8445 register. Increasing values mean more aggressive optimization, making the
8446 compile time increase with probably slightly better performance. The default
8449 @item max-cselib-memory-locations
8450 The maximum number of memory locations cselib should take into account.
8451 Increasing values mean more aggressive optimization, making the compile time
8452 increase with probably slightly better performance. The default value is 500.
8454 @item reorder-blocks-duplicate
8455 @itemx reorder-blocks-duplicate-feedback
8457 Used by basic block reordering pass to decide whether to use unconditional
8458 branch or duplicate the code on its destination. Code is duplicated when its
8459 estimated size is smaller than this value multiplied by the estimated size of
8460 unconditional jump in the hot spots of the program.
8462 The @option{reorder-block-duplicate-feedback} is used only when profile
8463 feedback is available and may be set to higher values than
8464 @option{reorder-block-duplicate} since information about the hot spots is more
8467 @item max-sched-ready-insns
8468 The maximum number of instructions ready to be issued the scheduler should
8469 consider at any given time during the first scheduling pass. Increasing
8470 values mean more thorough searches, making the compilation time increase
8471 with probably little benefit. The default value is 100.
8473 @item max-sched-region-blocks
8474 The maximum number of blocks in a region to be considered for
8475 interblock scheduling. The default value is 10.
8477 @item max-pipeline-region-blocks
8478 The maximum number of blocks in a region to be considered for
8479 pipelining in the selective scheduler. The default value is 15.
8481 @item max-sched-region-insns
8482 The maximum number of insns in a region to be considered for
8483 interblock scheduling. The default value is 100.
8485 @item max-pipeline-region-insns
8486 The maximum number of insns in a region to be considered for
8487 pipelining in the selective scheduler. The default value is 200.
8490 The minimum probability (in percents) of reaching a source block
8491 for interblock speculative scheduling. The default value is 40.
8493 @item max-sched-extend-regions-iters
8494 The maximum number of iterations through CFG to extend regions.
8495 0 - disable region extension,
8496 N - do at most N iterations.
8497 The default value is 0.
8499 @item max-sched-insn-conflict-delay
8500 The maximum conflict delay for an insn to be considered for speculative motion.
8501 The default value is 3.
8503 @item sched-spec-prob-cutoff
8504 The minimal probability of speculation success (in percents), so that
8505 speculative insn will be scheduled.
8506 The default value is 40.
8508 @item sched-mem-true-dep-cost
8509 Minimal distance (in CPU cycles) between store and load targeting same
8510 memory locations. The default value is 1.
8512 @item selsched-max-lookahead
8513 The maximum size of the lookahead window of selective scheduling. It is a
8514 depth of search for available instructions.
8515 The default value is 50.
8517 @item selsched-max-sched-times
8518 The maximum number of times that an instruction will be scheduled during
8519 selective scheduling. This is the limit on the number of iterations
8520 through which the instruction may be pipelined. The default value is 2.
8522 @item selsched-max-insns-to-rename
8523 The maximum number of best instructions in the ready list that are considered
8524 for renaming in the selective scheduler. The default value is 2.
8526 @item max-last-value-rtl
8527 The maximum size measured as number of RTLs that can be recorded in an expression
8528 in combiner for a pseudo register as last known value of that register. The default
8531 @item integer-share-limit
8532 Small integer constants can use a shared data structure, reducing the
8533 compiler's memory usage and increasing its speed. This sets the maximum
8534 value of a shared integer constant. The default value is 256.
8536 @item min-virtual-mappings
8537 Specifies the minimum number of virtual mappings in the incremental
8538 SSA updater that should be registered to trigger the virtual mappings
8539 heuristic defined by virtual-mappings-ratio. The default value is
8542 @item virtual-mappings-ratio
8543 If the number of virtual mappings is virtual-mappings-ratio bigger
8544 than the number of virtual symbols to be updated, then the incremental
8545 SSA updater switches to a full update for those symbols. The default
8548 @item ssp-buffer-size
8549 The minimum size of buffers (i.e.@: arrays) that will receive stack smashing
8550 protection when @option{-fstack-protection} is used.
8552 @item max-jump-thread-duplication-stmts
8553 Maximum number of statements allowed in a block that needs to be
8554 duplicated when threading jumps.
8556 @item max-fields-for-field-sensitive
8557 Maximum number of fields in a structure we will treat in
8558 a field sensitive manner during pointer analysis. The default is zero
8559 for -O0, and -O1 and 100 for -Os, -O2, and -O3.
8561 @item prefetch-latency
8562 Estimate on average number of instructions that are executed before
8563 prefetch finishes. The distance we prefetch ahead is proportional
8564 to this constant. Increasing this number may also lead to less
8565 streams being prefetched (see @option{simultaneous-prefetches}).
8567 @item simultaneous-prefetches
8568 Maximum number of prefetches that can run at the same time.
8570 @item l1-cache-line-size
8571 The size of cache line in L1 cache, in bytes.
8574 The size of L1 cache, in kilobytes.
8577 The size of L2 cache, in kilobytes.
8579 @item min-insn-to-prefetch-ratio
8580 The minimum ratio between the number of instructions and the
8581 number of prefetches to enable prefetching in a loop.
8583 @item prefetch-min-insn-to-mem-ratio
8584 The minimum ratio between the number of instructions and the
8585 number of memory references to enable prefetching in a loop.
8587 @item use-canonical-types
8588 Whether the compiler should use the ``canonical'' type system. By
8589 default, this should always be 1, which uses a more efficient internal
8590 mechanism for comparing types in C++ and Objective-C++. However, if
8591 bugs in the canonical type system are causing compilation failures,
8592 set this value to 0 to disable canonical types.
8594 @item switch-conversion-max-branch-ratio
8595 Switch initialization conversion will refuse to create arrays that are
8596 bigger than @option{switch-conversion-max-branch-ratio} times the number of
8597 branches in the switch.
8599 @item max-partial-antic-length
8600 Maximum length of the partial antic set computed during the tree
8601 partial redundancy elimination optimization (@option{-ftree-pre}) when
8602 optimizing at @option{-O3} and above. For some sorts of source code
8603 the enhanced partial redundancy elimination optimization can run away,
8604 consuming all of the memory available on the host machine. This
8605 parameter sets a limit on the length of the sets that are computed,
8606 which prevents the runaway behavior. Setting a value of 0 for
8607 this parameter will allow an unlimited set length.
8609 @item sccvn-max-scc-size
8610 Maximum size of a strongly connected component (SCC) during SCCVN
8611 processing. If this limit is hit, SCCVN processing for the whole
8612 function will not be done and optimizations depending on it will
8613 be disabled. The default maximum SCC size is 10000.
8615 @item ira-max-loops-num
8616 IRA uses a regional register allocation by default. If a function
8617 contains loops more than number given by the parameter, only at most
8618 given number of the most frequently executed loops will form regions
8619 for the regional register allocation. The default value of the
8622 @item ira-max-conflict-table-size
8623 Although IRA uses a sophisticated algorithm of compression conflict
8624 table, the table can be still big for huge functions. If the conflict
8625 table for a function could be more than size in MB given by the
8626 parameter, the conflict table is not built and faster, simpler, and
8627 lower quality register allocation algorithm will be used. The
8628 algorithm do not use pseudo-register conflicts. The default value of
8629 the parameter is 2000.
8631 @item ira-loop-reserved-regs
8632 IRA can be used to evaluate more accurate register pressure in loops
8633 for decision to move loop invariants (see @option{-O3}). The number
8634 of available registers reserved for some other purposes is described
8635 by this parameter. The default value of the parameter is 2 which is
8636 minimal number of registers needed for execution of typical
8637 instruction. This value is the best found from numerous experiments.
8639 @item loop-invariant-max-bbs-in-loop
8640 Loop invariant motion can be very expensive, both in compile time and
8641 in amount of needed compile time memory, with very large loops. Loops
8642 with more basic blocks than this parameter won't have loop invariant
8643 motion optimization performed on them. The default value of the
8644 parameter is 1000 for -O1 and 10000 for -O2 and above.
8646 @item max-vartrack-size
8647 Sets a maximum number of hash table slots to use during variable
8648 tracking dataflow analysis of any function. If this limit is exceeded
8649 with variable tracking at assignments enabled, analysis for that
8650 function is retried without it, after removing all debug insns from
8651 the function. If the limit is exceeded even without debug insns, var
8652 tracking analysis is completely disabled for the function. Setting
8653 the parameter to zero makes it unlimited.
8655 @item min-nondebug-insn-uid
8656 Use uids starting at this parameter for nondebug insns. The range below
8657 the parameter is reserved exclusively for debug insns created by
8658 @option{-fvar-tracking-assignments}, but debug insns may get
8659 (non-overlapping) uids above it if the reserved range is exhausted.
8661 @item ipa-sra-ptr-growth-factor
8662 IPA-SRA will replace a pointer to an aggregate with one or more new
8663 parameters only when their cumulative size is less or equal to
8664 @option{ipa-sra-ptr-growth-factor} times the size of the original
8667 @item graphite-max-nb-scop-params
8668 To avoid exponential effects in the Graphite loop transforms, the
8669 number of parameters in a Static Control Part (SCoP) is bounded. The
8670 default value is 10 parameters. A variable whose value is unknown at
8671 compile time and defined outside a SCoP is a parameter of the SCoP.
8673 @item graphite-max-bbs-per-function
8674 To avoid exponential effects in the detection of SCoPs, the size of
8675 the functions analyzed by Graphite is bounded. The default value is
8678 @item loop-block-tile-size
8679 Loop blocking or strip mining transforms, enabled with
8680 @option{-floop-block} or @option{-floop-strip-mine}, strip mine each
8681 loop in the loop nest by a given number of iterations. The strip
8682 length can be changed using the @option{loop-block-tile-size}
8683 parameter. The default value is 51 iterations.
8688 @node Preprocessor Options
8689 @section Options Controlling the Preprocessor
8690 @cindex preprocessor options
8691 @cindex options, preprocessor
8693 These options control the C preprocessor, which is run on each C source
8694 file before actual compilation.
8696 If you use the @option{-E} option, nothing is done except preprocessing.
8697 Some of these options make sense only together with @option{-E} because
8698 they cause the preprocessor output to be unsuitable for actual
8702 @item -Wp,@var{option}
8704 You can use @option{-Wp,@var{option}} to bypass the compiler driver
8705 and pass @var{option} directly through to the preprocessor. If
8706 @var{option} contains commas, it is split into multiple options at the
8707 commas. However, many options are modified, translated or interpreted
8708 by the compiler driver before being passed to the preprocessor, and
8709 @option{-Wp} forcibly bypasses this phase. The preprocessor's direct
8710 interface is undocumented and subject to change, so whenever possible
8711 you should avoid using @option{-Wp} and let the driver handle the
8714 @item -Xpreprocessor @var{option}
8715 @opindex Xpreprocessor
8716 Pass @var{option} as an option to the preprocessor. You can use this to
8717 supply system-specific preprocessor options which GCC does not know how to
8720 If you want to pass an option that takes an argument, you must use
8721 @option{-Xpreprocessor} twice, once for the option and once for the argument.
8724 @include cppopts.texi
8726 @node Assembler Options
8727 @section Passing Options to the Assembler
8729 @c prevent bad page break with this line
8730 You can pass options to the assembler.
8733 @item -Wa,@var{option}
8735 Pass @var{option} as an option to the assembler. If @var{option}
8736 contains commas, it is split into multiple options at the commas.
8738 @item -Xassembler @var{option}
8740 Pass @var{option} as an option to the assembler. You can use this to
8741 supply system-specific assembler options which GCC does not know how to
8744 If you want to pass an option that takes an argument, you must use
8745 @option{-Xassembler} twice, once for the option and once for the argument.
8750 @section Options for Linking
8751 @cindex link options
8752 @cindex options, linking
8754 These options come into play when the compiler links object files into
8755 an executable output file. They are meaningless if the compiler is
8756 not doing a link step.
8760 @item @var{object-file-name}
8761 A file name that does not end in a special recognized suffix is
8762 considered to name an object file or library. (Object files are
8763 distinguished from libraries by the linker according to the file
8764 contents.) If linking is done, these object files are used as input
8773 If any of these options is used, then the linker is not run, and
8774 object file names should not be used as arguments. @xref{Overall
8778 @item -l@var{library}
8779 @itemx -l @var{library}
8781 Search the library named @var{library} when linking. (The second
8782 alternative with the library as a separate argument is only for
8783 POSIX compliance and is not recommended.)
8785 It makes a difference where in the command you write this option; the
8786 linker searches and processes libraries and object files in the order they
8787 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
8788 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
8789 to functions in @samp{z}, those functions may not be loaded.
8791 The linker searches a standard list of directories for the library,
8792 which is actually a file named @file{lib@var{library}.a}. The linker
8793 then uses this file as if it had been specified precisely by name.
8795 The directories searched include several standard system directories
8796 plus any that you specify with @option{-L}.
8798 Normally the files found this way are library files---archive files
8799 whose members are object files. The linker handles an archive file by
8800 scanning through it for members which define symbols that have so far
8801 been referenced but not defined. But if the file that is found is an
8802 ordinary object file, it is linked in the usual fashion. The only
8803 difference between using an @option{-l} option and specifying a file name
8804 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
8805 and searches several directories.
8809 You need this special case of the @option{-l} option in order to
8810 link an Objective-C or Objective-C++ program.
8813 @opindex nostartfiles
8814 Do not use the standard system startup files when linking.
8815 The standard system libraries are used normally, unless @option{-nostdlib}
8816 or @option{-nodefaultlibs} is used.
8818 @item -nodefaultlibs
8819 @opindex nodefaultlibs
8820 Do not use the standard system libraries when linking.
8821 Only the libraries you specify will be passed to the linker, options
8822 specifying linkage of the system libraries, such as @code{-static-libgcc}
8823 or @code{-shared-libgcc}, will be ignored.
8824 The standard startup files are used normally, unless @option{-nostartfiles}
8825 is used. The compiler may generate calls to @code{memcmp},
8826 @code{memset}, @code{memcpy} and @code{memmove}.
8827 These entries are usually resolved by entries in
8828 libc. These entry points should be supplied through some other
8829 mechanism when this option is specified.
8833 Do not use the standard system startup files or libraries when linking.
8834 No startup files and only the libraries you specify will be passed to
8835 the linker, options specifying linkage of the system libraries, such as
8836 @code{-static-libgcc} or @code{-shared-libgcc}, will be ignored.
8837 The compiler may generate calls to @code{memcmp}, @code{memset},
8838 @code{memcpy} and @code{memmove}.
8839 These entries are usually resolved by entries in
8840 libc. These entry points should be supplied through some other
8841 mechanism when this option is specified.
8843 @cindex @option{-lgcc}, use with @option{-nostdlib}
8844 @cindex @option{-nostdlib} and unresolved references
8845 @cindex unresolved references and @option{-nostdlib}
8846 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
8847 @cindex @option{-nodefaultlibs} and unresolved references
8848 @cindex unresolved references and @option{-nodefaultlibs}
8849 One of the standard libraries bypassed by @option{-nostdlib} and
8850 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
8851 that GCC uses to overcome shortcomings of particular machines, or special
8852 needs for some languages.
8853 (@xref{Interface,,Interfacing to GCC Output,gccint,GNU Compiler
8854 Collection (GCC) Internals},
8855 for more discussion of @file{libgcc.a}.)
8856 In most cases, you need @file{libgcc.a} even when you want to avoid
8857 other standard libraries. In other words, when you specify @option{-nostdlib}
8858 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
8859 This ensures that you have no unresolved references to internal GCC
8860 library subroutines. (For example, @samp{__main}, used to ensure C++
8861 constructors will be called; @pxref{Collect2,,@code{collect2}, gccint,
8862 GNU Compiler Collection (GCC) Internals}.)
8866 Produce a position independent executable on targets which support it.
8867 For predictable results, you must also specify the same set of options
8868 that were used to generate code (@option{-fpie}, @option{-fPIE},
8869 or model suboptions) when you specify this option.
8873 Pass the flag @option{-export-dynamic} to the ELF linker, on targets
8874 that support it. This instructs the linker to add all symbols, not
8875 only used ones, to the dynamic symbol table. This option is needed
8876 for some uses of @code{dlopen} or to allow obtaining backtraces
8877 from within a program.
8881 Remove all symbol table and relocation information from the executable.
8885 On systems that support dynamic linking, this prevents linking with the shared
8886 libraries. On other systems, this option has no effect.
8890 Produce a shared object which can then be linked with other objects to
8891 form an executable. Not all systems support this option. For predictable
8892 results, you must also specify the same set of options that were used to
8893 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
8894 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
8895 needs to build supplementary stub code for constructors to work. On
8896 multi-libbed systems, @samp{gcc -shared} must select the correct support
8897 libraries to link against. Failing to supply the correct flags may lead
8898 to subtle defects. Supplying them in cases where they are not necessary
8901 @item -shared-libgcc
8902 @itemx -static-libgcc
8903 @opindex shared-libgcc
8904 @opindex static-libgcc
8905 On systems that provide @file{libgcc} as a shared library, these options
8906 force the use of either the shared or static version respectively.
8907 If no shared version of @file{libgcc} was built when the compiler was
8908 configured, these options have no effect.
8910 There are several situations in which an application should use the
8911 shared @file{libgcc} instead of the static version. The most common
8912 of these is when the application wishes to throw and catch exceptions
8913 across different shared libraries. In that case, each of the libraries
8914 as well as the application itself should use the shared @file{libgcc}.
8916 Therefore, the G++ and GCJ drivers automatically add
8917 @option{-shared-libgcc} whenever you build a shared library or a main
8918 executable, because C++ and Java programs typically use exceptions, so
8919 this is the right thing to do.
8921 If, instead, you use the GCC driver to create shared libraries, you may
8922 find that they will not always be linked with the shared @file{libgcc}.
8923 If GCC finds, at its configuration time, that you have a non-GNU linker
8924 or a GNU linker that does not support option @option{--eh-frame-hdr},
8925 it will link the shared version of @file{libgcc} into shared libraries
8926 by default. Otherwise, it will take advantage of the linker and optimize
8927 away the linking with the shared version of @file{libgcc}, linking with
8928 the static version of libgcc by default. This allows exceptions to
8929 propagate through such shared libraries, without incurring relocation
8930 costs at library load time.
8932 However, if a library or main executable is supposed to throw or catch
8933 exceptions, you must link it using the G++ or GCJ driver, as appropriate
8934 for the languages used in the program, or using the option
8935 @option{-shared-libgcc}, such that it is linked with the shared
8938 @item -static-libstdc++
8939 When the @command{g++} program is used to link a C++ program, it will
8940 normally automatically link against @option{libstdc++}. If
8941 @file{libstdc++} is available as a shared library, and the
8942 @option{-static} option is not used, then this will link against the
8943 shared version of @file{libstdc++}. That is normally fine. However, it
8944 is sometimes useful to freeze the version of @file{libstdc++} used by
8945 the program without going all the way to a fully static link. The
8946 @option{-static-libstdc++} option directs the @command{g++} driver to
8947 link @file{libstdc++} statically, without necessarily linking other
8948 libraries statically.
8952 Bind references to global symbols when building a shared object. Warn
8953 about any unresolved references (unless overridden by the link editor
8954 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
8957 @item -T @var{script}
8959 @cindex linker script
8960 Use @var{script} as the linker script. This option is supported by most
8961 systems using the GNU linker. On some targets, such as bare-board
8962 targets without an operating system, the @option{-T} option may be required
8963 when linking to avoid references to undefined symbols.
8965 @item -Xlinker @var{option}
8967 Pass @var{option} as an option to the linker. You can use this to
8968 supply system-specific linker options which GCC does not know how to
8971 If you want to pass an option that takes a separate argument, you must use
8972 @option{-Xlinker} twice, once for the option and once for the argument.
8973 For example, to pass @option{-assert definitions}, you must write
8974 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
8975 @option{-Xlinker "-assert definitions"}, because this passes the entire
8976 string as a single argument, which is not what the linker expects.
8978 When using the GNU linker, it is usually more convenient to pass
8979 arguments to linker options using the @option{@var{option}=@var{value}}
8980 syntax than as separate arguments. For example, you can specify
8981 @samp{-Xlinker -Map=output.map} rather than
8982 @samp{-Xlinker -Map -Xlinker output.map}. Other linkers may not support
8983 this syntax for command-line options.
8985 @item -Wl,@var{option}
8987 Pass @var{option} as an option to the linker. If @var{option} contains
8988 commas, it is split into multiple options at the commas. You can use this
8989 syntax to pass an argument to the option.
8990 For example, @samp{-Wl,-Map,output.map} passes @samp{-Map output.map} to the
8991 linker. When using the GNU linker, you can also get the same effect with
8992 @samp{-Wl,-Map=output.map}.
8994 @item -u @var{symbol}
8996 Pretend the symbol @var{symbol} is undefined, to force linking of
8997 library modules to define it. You can use @option{-u} multiple times with
8998 different symbols to force loading of additional library modules.
9001 @node Directory Options
9002 @section Options for Directory Search
9003 @cindex directory options
9004 @cindex options, directory search
9007 These options specify directories to search for header files, for
9008 libraries and for parts of the compiler:
9013 Add the directory @var{dir} to the head of the list of directories to be
9014 searched for header files. This can be used to override a system header
9015 file, substituting your own version, since these directories are
9016 searched before the system header file directories. However, you should
9017 not use this option to add directories that contain vendor-supplied
9018 system header files (use @option{-isystem} for that). If you use more than
9019 one @option{-I} option, the directories are scanned in left-to-right
9020 order; the standard system directories come after.
9022 If a standard system include directory, or a directory specified with
9023 @option{-isystem}, is also specified with @option{-I}, the @option{-I}
9024 option will be ignored. The directory will still be searched but as a
9025 system directory at its normal position in the system include chain.
9026 This is to ensure that GCC's procedure to fix buggy system headers and
9027 the ordering for the include_next directive are not inadvertently changed.
9028 If you really need to change the search order for system directories,
9029 use the @option{-nostdinc} and/or @option{-isystem} options.
9031 @item -iplugindir=@var{dir}
9032 Set the directory to search for plugins which are passed
9033 by @option{-fplugin=@var{name}} instead of
9034 @option{-fplugin=@var{path}/@var{name}.so}. This option is not meant
9035 to be used by the user, but only passed by the driver.
9037 @item -iquote@var{dir}
9039 Add the directory @var{dir} to the head of the list of directories to
9040 be searched for header files only for the case of @samp{#include
9041 "@var{file}"}; they are not searched for @samp{#include <@var{file}>},
9042 otherwise just like @option{-I}.
9046 Add directory @var{dir} to the list of directories to be searched
9049 @item -B@var{prefix}
9051 This option specifies where to find the executables, libraries,
9052 include files, and data files of the compiler itself.
9054 The compiler driver program runs one or more of the subprograms
9055 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
9056 @var{prefix} as a prefix for each program it tries to run, both with and
9057 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
9059 For each subprogram to be run, the compiler driver first tries the
9060 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
9061 was not specified, the driver tries two standard prefixes, which are
9062 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc/}. If neither of
9063 those results in a file name that is found, the unmodified program
9064 name is searched for using the directories specified in your
9065 @env{PATH} environment variable.
9067 The compiler will check to see if the path provided by the @option{-B}
9068 refers to a directory, and if necessary it will add a directory
9069 separator character at the end of the path.
9071 @option{-B} prefixes that effectively specify directory names also apply
9072 to libraries in the linker, because the compiler translates these
9073 options into @option{-L} options for the linker. They also apply to
9074 includes files in the preprocessor, because the compiler translates these
9075 options into @option{-isystem} options for the preprocessor. In this case,
9076 the compiler appends @samp{include} to the prefix.
9078 The run-time support file @file{libgcc.a} can also be searched for using
9079 the @option{-B} prefix, if needed. If it is not found there, the two
9080 standard prefixes above are tried, and that is all. The file is left
9081 out of the link if it is not found by those means.
9083 Another way to specify a prefix much like the @option{-B} prefix is to use
9084 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
9087 As a special kludge, if the path provided by @option{-B} is
9088 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
9089 9, then it will be replaced by @file{[dir/]include}. This is to help
9090 with boot-strapping the compiler.
9092 @item -specs=@var{file}
9094 Process @var{file} after the compiler reads in the standard @file{specs}
9095 file, in order to override the defaults that the @file{gcc} driver
9096 program uses when determining what switches to pass to @file{cc1},
9097 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
9098 @option{-specs=@var{file}} can be specified on the command line, and they
9099 are processed in order, from left to right.
9101 @item --sysroot=@var{dir}
9103 Use @var{dir} as the logical root directory for headers and libraries.
9104 For example, if the compiler would normally search for headers in
9105 @file{/usr/include} and libraries in @file{/usr/lib}, it will instead
9106 search @file{@var{dir}/usr/include} and @file{@var{dir}/usr/lib}.
9108 If you use both this option and the @option{-isysroot} option, then
9109 the @option{--sysroot} option will apply to libraries, but the
9110 @option{-isysroot} option will apply to header files.
9112 The GNU linker (beginning with version 2.16) has the necessary support
9113 for this option. If your linker does not support this option, the
9114 header file aspect of @option{--sysroot} will still work, but the
9115 library aspect will not.
9119 This option has been deprecated. Please use @option{-iquote} instead for
9120 @option{-I} directories before the @option{-I-} and remove the @option{-I-}.
9121 Any directories you specify with @option{-I} options before the @option{-I-}
9122 option are searched only for the case of @samp{#include "@var{file}"};
9123 they are not searched for @samp{#include <@var{file}>}.
9125 If additional directories are specified with @option{-I} options after
9126 the @option{-I-}, these directories are searched for all @samp{#include}
9127 directives. (Ordinarily @emph{all} @option{-I} directories are used
9130 In addition, the @option{-I-} option inhibits the use of the current
9131 directory (where the current input file came from) as the first search
9132 directory for @samp{#include "@var{file}"}. There is no way to
9133 override this effect of @option{-I-}. With @option{-I.} you can specify
9134 searching the directory which was current when the compiler was
9135 invoked. That is not exactly the same as what the preprocessor does
9136 by default, but it is often satisfactory.
9138 @option{-I-} does not inhibit the use of the standard system directories
9139 for header files. Thus, @option{-I-} and @option{-nostdinc} are
9146 @section Specifying subprocesses and the switches to pass to them
9149 @command{gcc} is a driver program. It performs its job by invoking a
9150 sequence of other programs to do the work of compiling, assembling and
9151 linking. GCC interprets its command-line parameters and uses these to
9152 deduce which programs it should invoke, and which command-line options
9153 it ought to place on their command lines. This behavior is controlled
9154 by @dfn{spec strings}. In most cases there is one spec string for each
9155 program that GCC can invoke, but a few programs have multiple spec
9156 strings to control their behavior. The spec strings built into GCC can
9157 be overridden by using the @option{-specs=} command-line switch to specify
9160 @dfn{Spec files} are plaintext files that are used to construct spec
9161 strings. They consist of a sequence of directives separated by blank
9162 lines. The type of directive is determined by the first non-whitespace
9163 character on the line and it can be one of the following:
9166 @item %@var{command}
9167 Issues a @var{command} to the spec file processor. The commands that can
9171 @item %include <@var{file}>
9173 Search for @var{file} and insert its text at the current point in the
9176 @item %include_noerr <@var{file}>
9177 @cindex %include_noerr
9178 Just like @samp{%include}, but do not generate an error message if the include
9179 file cannot be found.
9181 @item %rename @var{old_name} @var{new_name}
9183 Rename the spec string @var{old_name} to @var{new_name}.
9187 @item *[@var{spec_name}]:
9188 This tells the compiler to create, override or delete the named spec
9189 string. All lines after this directive up to the next directive or
9190 blank line are considered to be the text for the spec string. If this
9191 results in an empty string then the spec will be deleted. (Or, if the
9192 spec did not exist, then nothing will happened.) Otherwise, if the spec
9193 does not currently exist a new spec will be created. If the spec does
9194 exist then its contents will be overridden by the text of this
9195 directive, unless the first character of that text is the @samp{+}
9196 character, in which case the text will be appended to the spec.
9198 @item [@var{suffix}]:
9199 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
9200 and up to the next directive or blank line are considered to make up the
9201 spec string for the indicated suffix. When the compiler encounters an
9202 input file with the named suffix, it will processes the spec string in
9203 order to work out how to compile that file. For example:
9210 This says that any input file whose name ends in @samp{.ZZ} should be
9211 passed to the program @samp{z-compile}, which should be invoked with the
9212 command-line switch @option{-input} and with the result of performing the
9213 @samp{%i} substitution. (See below.)
9215 As an alternative to providing a spec string, the text that follows a
9216 suffix directive can be one of the following:
9219 @item @@@var{language}
9220 This says that the suffix is an alias for a known @var{language}. This is
9221 similar to using the @option{-x} command-line switch to GCC to specify a
9222 language explicitly. For example:
9229 Says that .ZZ files are, in fact, C++ source files.
9232 This causes an error messages saying:
9235 @var{name} compiler not installed on this system.
9239 GCC already has an extensive list of suffixes built into it.
9240 This directive will add an entry to the end of the list of suffixes, but
9241 since the list is searched from the end backwards, it is effectively
9242 possible to override earlier entries using this technique.
9246 GCC has the following spec strings built into it. Spec files can
9247 override these strings or create their own. Note that individual
9248 targets can also add their own spec strings to this list.
9251 asm Options to pass to the assembler
9252 asm_final Options to pass to the assembler post-processor
9253 cpp Options to pass to the C preprocessor
9254 cc1 Options to pass to the C compiler
9255 cc1plus Options to pass to the C++ compiler
9256 endfile Object files to include at the end of the link
9257 link Options to pass to the linker
9258 lib Libraries to include on the command line to the linker
9259 libgcc Decides which GCC support library to pass to the linker
9260 linker Sets the name of the linker
9261 predefines Defines to be passed to the C preprocessor
9262 signed_char Defines to pass to CPP to say whether @code{char} is signed
9264 startfile Object files to include at the start of the link
9267 Here is a small example of a spec file:
9273 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
9276 This example renames the spec called @samp{lib} to @samp{old_lib} and
9277 then overrides the previous definition of @samp{lib} with a new one.
9278 The new definition adds in some extra command-line options before
9279 including the text of the old definition.
9281 @dfn{Spec strings} are a list of command-line options to be passed to their
9282 corresponding program. In addition, the spec strings can contain
9283 @samp{%}-prefixed sequences to substitute variable text or to
9284 conditionally insert text into the command line. Using these constructs
9285 it is possible to generate quite complex command lines.
9287 Here is a table of all defined @samp{%}-sequences for spec
9288 strings. Note that spaces are not generated automatically around the
9289 results of expanding these sequences. Therefore you can concatenate them
9290 together or combine them with constant text in a single argument.
9294 Substitute one @samp{%} into the program name or argument.
9297 Substitute the name of the input file being processed.
9300 Substitute the basename of the input file being processed.
9301 This is the substring up to (and not including) the last period
9302 and not including the directory.
9305 This is the same as @samp{%b}, but include the file suffix (text after
9309 Marks the argument containing or following the @samp{%d} as a
9310 temporary file name, so that that file will be deleted if GCC exits
9311 successfully. Unlike @samp{%g}, this contributes no text to the
9314 @item %g@var{suffix}
9315 Substitute a file name that has suffix @var{suffix} and is chosen
9316 once per compilation, and mark the argument in the same way as
9317 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
9318 name is now chosen in a way that is hard to predict even when previously
9319 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
9320 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
9321 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
9322 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
9323 was simply substituted with a file name chosen once per compilation,
9324 without regard to any appended suffix (which was therefore treated
9325 just like ordinary text), making such attacks more likely to succeed.
9327 @item %u@var{suffix}
9328 Like @samp{%g}, but generates a new temporary file name even if
9329 @samp{%u@var{suffix}} was already seen.
9331 @item %U@var{suffix}
9332 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
9333 new one if there is no such last file name. In the absence of any
9334 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
9335 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
9336 would involve the generation of two distinct file names, one
9337 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
9338 simply substituted with a file name chosen for the previous @samp{%u},
9339 without regard to any appended suffix.
9341 @item %j@var{suffix}
9342 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
9343 writable, and if save-temps is off; otherwise, substitute the name
9344 of a temporary file, just like @samp{%u}. This temporary file is not
9345 meant for communication between processes, but rather as a junk
9348 @item %|@var{suffix}
9349 @itemx %m@var{suffix}
9350 Like @samp{%g}, except if @option{-pipe} is in effect. In that case
9351 @samp{%|} substitutes a single dash and @samp{%m} substitutes nothing at
9352 all. These are the two most common ways to instruct a program that it
9353 should read from standard input or write to standard output. If you
9354 need something more elaborate you can use an @samp{%@{pipe:@code{X}@}}
9355 construct: see for example @file{f/lang-specs.h}.
9357 @item %.@var{SUFFIX}
9358 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
9359 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
9360 terminated by the next space or %.
9363 Marks the argument containing or following the @samp{%w} as the
9364 designated output file of this compilation. This puts the argument
9365 into the sequence of arguments that @samp{%o} will substitute later.
9368 Substitutes the names of all the output files, with spaces
9369 automatically placed around them. You should write spaces
9370 around the @samp{%o} as well or the results are undefined.
9371 @samp{%o} is for use in the specs for running the linker.
9372 Input files whose names have no recognized suffix are not compiled
9373 at all, but they are included among the output files, so they will
9377 Substitutes the suffix for object files. Note that this is
9378 handled specially when it immediately follows @samp{%g, %u, or %U},
9379 because of the need for those to form complete file names. The
9380 handling is such that @samp{%O} is treated exactly as if it had already
9381 been substituted, except that @samp{%g, %u, and %U} do not currently
9382 support additional @var{suffix} characters following @samp{%O} as they would
9383 following, for example, @samp{.o}.
9386 Substitutes the standard macro predefinitions for the
9387 current target machine. Use this when running @code{cpp}.
9390 Like @samp{%p}, but puts @samp{__} before and after the name of each
9391 predefined macro, except for macros that start with @samp{__} or with
9392 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
9396 Substitute any of @option{-iprefix} (made from @env{GCC_EXEC_PREFIX}),
9397 @option{-isysroot} (made from @env{TARGET_SYSTEM_ROOT}),
9398 @option{-isystem} (made from @env{COMPILER_PATH} and @option{-B} options)
9399 and @option{-imultilib} as necessary.
9402 Current argument is the name of a library or startup file of some sort.
9403 Search for that file in a standard list of directories and substitute
9404 the full name found. The current working directory is included in the
9405 list of directories scanned.
9408 Current argument is the name of a linker script. Search for that file
9409 in the current list of directories to scan for libraries. If the file
9410 is located insert a @option{--script} option into the command line
9411 followed by the full path name found. If the file is not found then
9412 generate an error message. Note: the current working directory is not
9416 Print @var{str} as an error message. @var{str} is terminated by a newline.
9417 Use this when inconsistent options are detected.
9420 Substitute the contents of spec string @var{name} at this point.
9423 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
9425 @item %x@{@var{option}@}
9426 Accumulate an option for @samp{%X}.
9429 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
9433 Output the accumulated assembler options specified by @option{-Wa}.
9436 Output the accumulated preprocessor options specified by @option{-Wp}.
9439 Process the @code{asm} spec. This is used to compute the
9440 switches to be passed to the assembler.
9443 Process the @code{asm_final} spec. This is a spec string for
9444 passing switches to an assembler post-processor, if such a program is
9448 Process the @code{link} spec. This is the spec for computing the
9449 command line passed to the linker. Typically it will make use of the
9450 @samp{%L %G %S %D and %E} sequences.
9453 Dump out a @option{-L} option for each directory that GCC believes might
9454 contain startup files. If the target supports multilibs then the
9455 current multilib directory will be prepended to each of these paths.
9458 Process the @code{lib} spec. This is a spec string for deciding which
9459 libraries should be included on the command line to the linker.
9462 Process the @code{libgcc} spec. This is a spec string for deciding
9463 which GCC support library should be included on the command line to the linker.
9466 Process the @code{startfile} spec. This is a spec for deciding which
9467 object files should be the first ones passed to the linker. Typically
9468 this might be a file named @file{crt0.o}.
9471 Process the @code{endfile} spec. This is a spec string that specifies
9472 the last object files that will be passed to the linker.
9475 Process the @code{cpp} spec. This is used to construct the arguments
9476 to be passed to the C preprocessor.
9479 Process the @code{cc1} spec. This is used to construct the options to be
9480 passed to the actual C compiler (@samp{cc1}).
9483 Process the @code{cc1plus} spec. This is used to construct the options to be
9484 passed to the actual C++ compiler (@samp{cc1plus}).
9487 Substitute the variable part of a matched option. See below.
9488 Note that each comma in the substituted string is replaced by
9492 Remove all occurrences of @code{-S} from the command line. Note---this
9493 command is position dependent. @samp{%} commands in the spec string
9494 before this one will see @code{-S}, @samp{%} commands in the spec string
9495 after this one will not.
9497 @item %:@var{function}(@var{args})
9498 Call the named function @var{function}, passing it @var{args}.
9499 @var{args} is first processed as a nested spec string, then split
9500 into an argument vector in the usual fashion. The function returns
9501 a string which is processed as if it had appeared literally as part
9502 of the current spec.
9504 The following built-in spec functions are provided:
9508 The @code{getenv} spec function takes two arguments: an environment
9509 variable name and a string. If the environment variable is not
9510 defined, a fatal error is issued. Otherwise, the return value is the
9511 value of the environment variable concatenated with the string. For
9512 example, if @env{TOPDIR} is defined as @file{/path/to/top}, then:
9515 %:getenv(TOPDIR /include)
9518 expands to @file{/path/to/top/include}.
9520 @item @code{if-exists}
9521 The @code{if-exists} spec function takes one argument, an absolute
9522 pathname to a file. If the file exists, @code{if-exists} returns the
9523 pathname. Here is a small example of its usage:
9527 crt0%O%s %:if-exists(crti%O%s) crtbegin%O%s
9530 @item @code{if-exists-else}
9531 The @code{if-exists-else} spec function is similar to the @code{if-exists}
9532 spec function, except that it takes two arguments. The first argument is
9533 an absolute pathname to a file. If the file exists, @code{if-exists-else}
9534 returns the pathname. If it does not exist, it returns the second argument.
9535 This way, @code{if-exists-else} can be used to select one file or another,
9536 based on the existence of the first. Here is a small example of its usage:
9540 crt0%O%s %:if-exists(crti%O%s) \
9541 %:if-exists-else(crtbeginT%O%s crtbegin%O%s)
9544 @item @code{replace-outfile}
9545 The @code{replace-outfile} spec function takes two arguments. It looks for the
9546 first argument in the outfiles array and replaces it with the second argument. Here
9547 is a small example of its usage:
9550 %@{fgnu-runtime:%:replace-outfile(-lobjc -lobjc-gnu)@}
9553 @item @code{print-asm-header}
9554 The @code{print-asm-header} function takes no arguments and simply
9555 prints a banner like:
9561 Use "-Wa,OPTION" to pass "OPTION" to the assembler.
9564 It is used to separate compiler options from assembler options
9565 in the @option{--target-help} output.
9569 Substitutes the @code{-S} switch, if that switch was given to GCC@.
9570 If that switch was not specified, this substitutes nothing. Note that
9571 the leading dash is omitted when specifying this option, and it is
9572 automatically inserted if the substitution is performed. Thus the spec
9573 string @samp{%@{foo@}} would match the command-line option @option{-foo}
9574 and would output the command line option @option{-foo}.
9576 @item %W@{@code{S}@}
9577 Like %@{@code{S}@} but mark last argument supplied within as a file to be
9580 @item %@{@code{S}*@}
9581 Substitutes all the switches specified to GCC whose names start
9582 with @code{-S}, but which also take an argument. This is used for
9583 switches like @option{-o}, @option{-D}, @option{-I}, etc.
9584 GCC considers @option{-o foo} as being
9585 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
9586 text, including the space. Thus two arguments would be generated.
9588 @item %@{@code{S}*&@code{T}*@}
9589 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
9590 (the order of @code{S} and @code{T} in the spec is not significant).
9591 There can be any number of ampersand-separated variables; for each the
9592 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
9594 @item %@{@code{S}:@code{X}@}
9595 Substitutes @code{X}, if the @samp{-S} switch was given to GCC@.
9597 @item %@{!@code{S}:@code{X}@}
9598 Substitutes @code{X}, if the @samp{-S} switch was @emph{not} given to GCC@.
9600 @item %@{@code{S}*:@code{X}@}
9601 Substitutes @code{X} if one or more switches whose names start with
9602 @code{-S} are specified to GCC@. Normally @code{X} is substituted only
9603 once, no matter how many such switches appeared. However, if @code{%*}
9604 appears somewhere in @code{X}, then @code{X} will be substituted once
9605 for each matching switch, with the @code{%*} replaced by the part of
9606 that switch that matched the @code{*}.
9608 @item %@{.@code{S}:@code{X}@}
9609 Substitutes @code{X}, if processing a file with suffix @code{S}.
9611 @item %@{!.@code{S}:@code{X}@}
9612 Substitutes @code{X}, if @emph{not} processing a file with suffix @code{S}.
9614 @item %@{,@code{S}:@code{X}@}
9615 Substitutes @code{X}, if processing a file for language @code{S}.
9617 @item %@{!,@code{S}:@code{X}@}
9618 Substitutes @code{X}, if not processing a file for language @code{S}.
9620 @item %@{@code{S}|@code{P}:@code{X}@}
9621 Substitutes @code{X} if either @code{-S} or @code{-P} was given to
9622 GCC@. This may be combined with @samp{!}, @samp{.}, @samp{,}, and
9623 @code{*} sequences as well, although they have a stronger binding than
9624 the @samp{|}. If @code{%*} appears in @code{X}, all of the
9625 alternatives must be starred, and only the first matching alternative
9628 For example, a spec string like this:
9631 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
9634 will output the following command-line options from the following input
9635 command-line options:
9640 -d fred.c -foo -baz -boggle
9641 -d jim.d -bar -baz -boggle
9644 @item %@{S:X; T:Y; :D@}
9646 If @code{S} was given to GCC, substitutes @code{X}; else if @code{T} was
9647 given to GCC, substitutes @code{Y}; else substitutes @code{D}. There can
9648 be as many clauses as you need. This may be combined with @code{.},
9649 @code{,}, @code{!}, @code{|}, and @code{*} as needed.
9654 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or similar
9655 construct may contain other nested @samp{%} constructs or spaces, or
9656 even newlines. They are processed as usual, as described above.
9657 Trailing white space in @code{X} is ignored. White space may also
9658 appear anywhere on the left side of the colon in these constructs,
9659 except between @code{.} or @code{*} and the corresponding word.
9661 The @option{-O}, @option{-f}, @option{-m}, and @option{-W} switches are
9662 handled specifically in these constructs. If another value of
9663 @option{-O} or the negated form of a @option{-f}, @option{-m}, or
9664 @option{-W} switch is found later in the command line, the earlier
9665 switch value is ignored, except with @{@code{S}*@} where @code{S} is
9666 just one letter, which passes all matching options.
9668 The character @samp{|} at the beginning of the predicate text is used to
9669 indicate that a command should be piped to the following command, but
9670 only if @option{-pipe} is specified.
9672 It is built into GCC which switches take arguments and which do not.
9673 (You might think it would be useful to generalize this to allow each
9674 compiler's spec to say which switches take arguments. But this cannot
9675 be done in a consistent fashion. GCC cannot even decide which input
9676 files have been specified without knowing which switches take arguments,
9677 and it must know which input files to compile in order to tell which
9680 GCC also knows implicitly that arguments starting in @option{-l} are to be
9681 treated as compiler output files, and passed to the linker in their
9682 proper position among the other output files.
9684 @c man begin OPTIONS
9686 @node Target Options
9687 @section Specifying Target Machine and Compiler Version
9688 @cindex target options
9689 @cindex cross compiling
9690 @cindex specifying machine version
9691 @cindex specifying compiler version and target machine
9692 @cindex compiler version, specifying
9693 @cindex target machine, specifying
9695 The usual way to run GCC is to run the executable called @file{gcc}, or
9696 @file{<machine>-gcc} when cross-compiling, or
9697 @file{<machine>-gcc-<version>} to run a version other than the one that
9700 @node Submodel Options
9701 @section Hardware Models and Configurations
9702 @cindex submodel options
9703 @cindex specifying hardware config
9704 @cindex hardware models and configurations, specifying
9705 @cindex machine dependent options
9707 Each target machine types can have its own
9708 special options, starting with @samp{-m}, to choose among various
9709 hardware models or configurations---for example, 68010 vs 68020,
9710 floating coprocessor or none. A single installed version of the
9711 compiler can compile for any model or configuration, according to the
9714 Some configurations of the compiler also support additional special
9715 options, usually for compatibility with other compilers on the same
9718 @c This list is ordered alphanumerically by subsection name.
9719 @c It should be the same order and spelling as these options are listed
9720 @c in Machine Dependent Options
9726 * Blackfin Options::
9730 * DEC Alpha Options::
9731 * DEC Alpha/VMS Options::
9734 * GNU/Linux Options::
9737 * i386 and x86-64 Options::
9738 * i386 and x86-64 Windows Options::
9740 * IA-64/VMS Options::
9752 * picoChip Options::
9754 * RS/6000 and PowerPC Options::
9756 * S/390 and zSeries Options::
9761 * System V Options::
9766 * Xstormy16 Options::
9772 @subsection ARC Options
9775 These options are defined for ARC implementations:
9780 Compile code for little endian mode. This is the default.
9784 Compile code for big endian mode.
9787 @opindex mmangle-cpu
9788 Prepend the name of the cpu to all public symbol names.
9789 In multiple-processor systems, there are many ARC variants with different
9790 instruction and register set characteristics. This flag prevents code
9791 compiled for one cpu to be linked with code compiled for another.
9792 No facility exists for handling variants that are ``almost identical''.
9793 This is an all or nothing option.
9795 @item -mcpu=@var{cpu}
9797 Compile code for ARC variant @var{cpu}.
9798 Which variants are supported depend on the configuration.
9799 All variants support @option{-mcpu=base}, this is the default.
9801 @item -mtext=@var{text-section}
9802 @itemx -mdata=@var{data-section}
9803 @itemx -mrodata=@var{readonly-data-section}
9807 Put functions, data, and readonly data in @var{text-section},
9808 @var{data-section}, and @var{readonly-data-section} respectively
9809 by default. This can be overridden with the @code{section} attribute.
9810 @xref{Variable Attributes}.
9815 @subsection ARM Options
9818 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
9822 @item -mabi=@var{name}
9824 Generate code for the specified ABI@. Permissible values are: @samp{apcs-gnu},
9825 @samp{atpcs}, @samp{aapcs}, @samp{aapcs-linux} and @samp{iwmmxt}.
9828 @opindex mapcs-frame
9829 Generate a stack frame that is compliant with the ARM Procedure Call
9830 Standard for all functions, even if this is not strictly necessary for
9831 correct execution of the code. Specifying @option{-fomit-frame-pointer}
9832 with this option will cause the stack frames not to be generated for
9833 leaf functions. The default is @option{-mno-apcs-frame}.
9837 This is a synonym for @option{-mapcs-frame}.
9840 @c not currently implemented
9841 @item -mapcs-stack-check
9842 @opindex mapcs-stack-check
9843 Generate code to check the amount of stack space available upon entry to
9844 every function (that actually uses some stack space). If there is
9845 insufficient space available then either the function
9846 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
9847 called, depending upon the amount of stack space required. The run time
9848 system is required to provide these functions. The default is
9849 @option{-mno-apcs-stack-check}, since this produces smaller code.
9851 @c not currently implemented
9853 @opindex mapcs-float
9854 Pass floating point arguments using the float point registers. This is
9855 one of the variants of the APCS@. This option is recommended if the
9856 target hardware has a floating point unit or if a lot of floating point
9857 arithmetic is going to be performed by the code. The default is
9858 @option{-mno-apcs-float}, since integer only code is slightly increased in
9859 size if @option{-mapcs-float} is used.
9861 @c not currently implemented
9862 @item -mapcs-reentrant
9863 @opindex mapcs-reentrant
9864 Generate reentrant, position independent code. The default is
9865 @option{-mno-apcs-reentrant}.
9868 @item -mthumb-interwork
9869 @opindex mthumb-interwork
9870 Generate code which supports calling between the ARM and Thumb
9871 instruction sets. Without this option the two instruction sets cannot
9872 be reliably used inside one program. The default is
9873 @option{-mno-thumb-interwork}, since slightly larger code is generated
9874 when @option{-mthumb-interwork} is specified.
9876 @item -mno-sched-prolog
9877 @opindex mno-sched-prolog
9878 Prevent the reordering of instructions in the function prolog, or the
9879 merging of those instruction with the instructions in the function's
9880 body. This means that all functions will start with a recognizable set
9881 of instructions (or in fact one of a choice from a small set of
9882 different function prologues), and this information can be used to
9883 locate the start if functions inside an executable piece of code. The
9884 default is @option{-msched-prolog}.
9886 @item -mfloat-abi=@var{name}
9888 Specifies which floating-point ABI to use. Permissible values
9889 are: @samp{soft}, @samp{softfp} and @samp{hard}.
9891 Specifying @samp{soft} causes GCC to generate output containing
9892 library calls for floating-point operations.
9893 @samp{softfp} allows the generation of code using hardware floating-point
9894 instructions, but still uses the soft-float calling conventions.
9895 @samp{hard} allows generation of floating-point instructions
9896 and uses FPU-specific calling conventions.
9898 The default depends on the specific target configuration. Note that
9899 the hard-float and soft-float ABIs are not link-compatible; you must
9900 compile your entire program with the same ABI, and link with a
9901 compatible set of libraries.
9904 @opindex mhard-float
9905 Equivalent to @option{-mfloat-abi=hard}.
9908 @opindex msoft-float
9909 Equivalent to @option{-mfloat-abi=soft}.
9911 @item -mlittle-endian
9912 @opindex mlittle-endian
9913 Generate code for a processor running in little-endian mode. This is
9914 the default for all standard configurations.
9917 @opindex mbig-endian
9918 Generate code for a processor running in big-endian mode; the default is
9919 to compile code for a little-endian processor.
9921 @item -mwords-little-endian
9922 @opindex mwords-little-endian
9923 This option only applies when generating code for big-endian processors.
9924 Generate code for a little-endian word order but a big-endian byte
9925 order. That is, a byte order of the form @samp{32107654}. Note: this
9926 option should only be used if you require compatibility with code for
9927 big-endian ARM processors generated by versions of the compiler prior to
9930 @item -mcpu=@var{name}
9932 This specifies the name of the target ARM processor. GCC uses this name
9933 to determine what kind of instructions it can emit when generating
9934 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
9935 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
9936 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
9937 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
9938 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
9940 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm7tdmi-s},
9941 @samp{arm710t}, @samp{arm720t}, @samp{arm740t},
9942 @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
9943 @samp{strongarm1110},
9944 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
9945 @samp{arm920t}, @samp{arm922t}, @samp{arm946e-s}, @samp{arm966e-s},
9946 @samp{arm968e-s}, @samp{arm926ej-s}, @samp{arm940t}, @samp{arm9tdmi},
9947 @samp{arm10tdmi}, @samp{arm1020t}, @samp{arm1026ej-s},
9948 @samp{arm10e}, @samp{arm1020e}, @samp{arm1022e},
9949 @samp{arm1136j-s}, @samp{arm1136jf-s}, @samp{mpcore}, @samp{mpcorenovfp},
9950 @samp{arm1156t2-s}, @samp{arm1156t2f-s}, @samp{arm1176jz-s}, @samp{arm1176jzf-s},
9951 @samp{cortex-a5}, @samp{cortex-a8}, @samp{cortex-a9},
9952 @samp{cortex-r4}, @samp{cortex-r4f}, @samp{cortex-m4}, @samp{cortex-m3},
9955 @samp{xscale}, @samp{iwmmxt}, @samp{iwmmxt2}, @samp{ep9312}.
9957 @item -mtune=@var{name}
9959 This option is very similar to the @option{-mcpu=} option, except that
9960 instead of specifying the actual target processor type, and hence
9961 restricting which instructions can be used, it specifies that GCC should
9962 tune the performance of the code as if the target were of the type
9963 specified in this option, but still choosing the instructions that it
9964 will generate based on the cpu specified by a @option{-mcpu=} option.
9965 For some ARM implementations better performance can be obtained by using
9968 @item -march=@var{name}
9970 This specifies the name of the target ARM architecture. GCC uses this
9971 name to determine what kind of instructions it can emit when generating
9972 assembly code. This option can be used in conjunction with or instead
9973 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
9974 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
9975 @samp{armv5}, @samp{armv5t}, @samp{armv5e}, @samp{armv5te},
9976 @samp{armv6}, @samp{armv6j},
9977 @samp{armv6t2}, @samp{armv6z}, @samp{armv6zk}, @samp{armv6-m},
9978 @samp{armv7}, @samp{armv7-a}, @samp{armv7-r}, @samp{armv7-m},
9979 @samp{iwmmxt}, @samp{iwmmxt2}, @samp{ep9312}.
9981 @item -mfpu=@var{name}
9982 @itemx -mfpe=@var{number}
9983 @itemx -mfp=@var{number}
9987 This specifies what floating point hardware (or hardware emulation) is
9988 available on the target. Permissible names are: @samp{fpa}, @samp{fpe2},
9989 @samp{fpe3}, @samp{maverick}, @samp{vfp}, @samp{vfpv3}, @samp{vfpv3-fp16},
9990 @samp{vfpv3-d16}, @samp{vfpv3-d16-fp16}, @samp{vfpv3xd}, @samp{vfpv3xd-fp16},
9991 @samp{neon}, @samp{neon-fp16}, @samp{vfpv4}, @samp{vfpv4-d16},
9992 @samp{fpv4-sp-d16} and @samp{neon-vfpv4}.
9993 @option{-mfp} and @option{-mfpe} are synonyms for
9994 @option{-mfpu}=@samp{fpe}@var{number}, for compatibility with older versions
9997 If @option{-msoft-float} is specified this specifies the format of
9998 floating point values.
10000 If the selected floating-point hardware includes the NEON extension
10001 (e.g. @option{-mfpu}=@samp{neon}), note that floating-point
10002 operations will not be used by GCC's auto-vectorization pass unless
10003 @option{-funsafe-math-optimizations} is also specified. This is
10004 because NEON hardware does not fully implement the IEEE 754 standard for
10005 floating-point arithmetic (in particular denormal values are treated as
10006 zero), so the use of NEON instructions may lead to a loss of precision.
10008 @item -mfp16-format=@var{name}
10009 @opindex mfp16-format
10010 Specify the format of the @code{__fp16} half-precision floating-point type.
10011 Permissible names are @samp{none}, @samp{ieee}, and @samp{alternative};
10012 the default is @samp{none}, in which case the @code{__fp16} type is not
10013 defined. @xref{Half-Precision}, for more information.
10015 @item -mstructure-size-boundary=@var{n}
10016 @opindex mstructure-size-boundary
10017 The size of all structures and unions will be rounded up to a multiple
10018 of the number of bits set by this option. Permissible values are 8, 32
10019 and 64. The default value varies for different toolchains. For the COFF
10020 targeted toolchain the default value is 8. A value of 64 is only allowed
10021 if the underlying ABI supports it.
10023 Specifying the larger number can produce faster, more efficient code, but
10024 can also increase the size of the program. Different values are potentially
10025 incompatible. Code compiled with one value cannot necessarily expect to
10026 work with code or libraries compiled with another value, if they exchange
10027 information using structures or unions.
10029 @item -mabort-on-noreturn
10030 @opindex mabort-on-noreturn
10031 Generate a call to the function @code{abort} at the end of a
10032 @code{noreturn} function. It will be executed if the function tries to
10036 @itemx -mno-long-calls
10037 @opindex mlong-calls
10038 @opindex mno-long-calls
10039 Tells the compiler to perform function calls by first loading the
10040 address of the function into a register and then performing a subroutine
10041 call on this register. This switch is needed if the target function
10042 will lie outside of the 64 megabyte addressing range of the offset based
10043 version of subroutine call instruction.
10045 Even if this switch is enabled, not all function calls will be turned
10046 into long calls. The heuristic is that static functions, functions
10047 which have the @samp{short-call} attribute, functions that are inside
10048 the scope of a @samp{#pragma no_long_calls} directive and functions whose
10049 definitions have already been compiled within the current compilation
10050 unit, will not be turned into long calls. The exception to this rule is
10051 that weak function definitions, functions with the @samp{long-call}
10052 attribute or the @samp{section} attribute, and functions that are within
10053 the scope of a @samp{#pragma long_calls} directive, will always be
10054 turned into long calls.
10056 This feature is not enabled by default. Specifying
10057 @option{-mno-long-calls} will restore the default behavior, as will
10058 placing the function calls within the scope of a @samp{#pragma
10059 long_calls_off} directive. Note these switches have no effect on how
10060 the compiler generates code to handle function calls via function
10063 @item -msingle-pic-base
10064 @opindex msingle-pic-base
10065 Treat the register used for PIC addressing as read-only, rather than
10066 loading it in the prologue for each function. The run-time system is
10067 responsible for initializing this register with an appropriate value
10068 before execution begins.
10070 @item -mpic-register=@var{reg}
10071 @opindex mpic-register
10072 Specify the register to be used for PIC addressing. The default is R10
10073 unless stack-checking is enabled, when R9 is used.
10075 @item -mcirrus-fix-invalid-insns
10076 @opindex mcirrus-fix-invalid-insns
10077 @opindex mno-cirrus-fix-invalid-insns
10078 Insert NOPs into the instruction stream to in order to work around
10079 problems with invalid Maverick instruction combinations. This option
10080 is only valid if the @option{-mcpu=ep9312} option has been used to
10081 enable generation of instructions for the Cirrus Maverick floating
10082 point co-processor. This option is not enabled by default, since the
10083 problem is only present in older Maverick implementations. The default
10084 can be re-enabled by use of the @option{-mno-cirrus-fix-invalid-insns}
10087 @item -mpoke-function-name
10088 @opindex mpoke-function-name
10089 Write the name of each function into the text section, directly
10090 preceding the function prologue. The generated code is similar to this:
10094 .ascii "arm_poke_function_name", 0
10097 .word 0xff000000 + (t1 - t0)
10098 arm_poke_function_name
10100 stmfd sp!, @{fp, ip, lr, pc@}
10104 When performing a stack backtrace, code can inspect the value of
10105 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
10106 location @code{pc - 12} and the top 8 bits are set, then we know that
10107 there is a function name embedded immediately preceding this location
10108 and has length @code{((pc[-3]) & 0xff000000)}.
10112 Generate code for the Thumb instruction set. The default is to
10113 use the 32-bit ARM instruction set.
10114 This option automatically enables either 16-bit Thumb-1 or
10115 mixed 16/32-bit Thumb-2 instructions based on the @option{-mcpu=@var{name}}
10116 and @option{-march=@var{name}} options. This option is not passed to the
10117 assembler. If you want to force assembler files to be interpreted as Thumb code,
10118 either add a @samp{.thumb} directive to the source or pass the @option{-mthumb}
10119 option directly to the assembler by prefixing it with @option{-Wa}.
10122 @opindex mtpcs-frame
10123 Generate a stack frame that is compliant with the Thumb Procedure Call
10124 Standard for all non-leaf functions. (A leaf function is one that does
10125 not call any other functions.) The default is @option{-mno-tpcs-frame}.
10127 @item -mtpcs-leaf-frame
10128 @opindex mtpcs-leaf-frame
10129 Generate a stack frame that is compliant with the Thumb Procedure Call
10130 Standard for all leaf functions. (A leaf function is one that does
10131 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
10133 @item -mcallee-super-interworking
10134 @opindex mcallee-super-interworking
10135 Gives all externally visible functions in the file being compiled an ARM
10136 instruction set header which switches to Thumb mode before executing the
10137 rest of the function. This allows these functions to be called from
10138 non-interworking code. This option is not valid in AAPCS configurations
10139 because interworking is enabled by default.
10141 @item -mcaller-super-interworking
10142 @opindex mcaller-super-interworking
10143 Allows calls via function pointers (including virtual functions) to
10144 execute correctly regardless of whether the target code has been
10145 compiled for interworking or not. There is a small overhead in the cost
10146 of executing a function pointer if this option is enabled. This option
10147 is not valid in AAPCS configurations because interworking is enabled
10150 @item -mtp=@var{name}
10152 Specify the access model for the thread local storage pointer. The valid
10153 models are @option{soft}, which generates calls to @code{__aeabi_read_tp},
10154 @option{cp15}, which fetches the thread pointer from @code{cp15} directly
10155 (supported in the arm6k architecture), and @option{auto}, which uses the
10156 best available method for the selected processor. The default setting is
10159 @item -mword-relocations
10160 @opindex mword-relocations
10161 Only generate absolute relocations on word sized values (i.e. R_ARM_ABS32).
10162 This is enabled by default on targets (uClinux, SymbianOS) where the runtime
10163 loader imposes this restriction, and when @option{-fpic} or @option{-fPIC}
10166 @item -mfix-cortex-m3-ldrd
10167 @opindex mfix-cortex-m3-ldrd
10168 Some Cortex-M3 cores can cause data corruption when @code{ldrd} instructions
10169 with overlapping destination and base registers are used. This option avoids
10170 generating these instructions. This option is enabled by default when
10171 @option{-mcpu=cortex-m3} is specified.
10176 @subsection AVR Options
10177 @cindex AVR Options
10179 These options are defined for AVR implementations:
10182 @item -mmcu=@var{mcu}
10184 Specify ATMEL AVR instruction set or MCU type.
10186 Instruction set avr1 is for the minimal AVR core, not supported by the C
10187 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
10188 attiny11, attiny12, attiny15, attiny28).
10190 Instruction set avr2 (default) is for the classic AVR core with up to
10191 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
10192 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
10193 at90c8534, at90s8535).
10195 Instruction set avr3 is for the classic AVR core with up to 128K program
10196 memory space (MCU types: atmega103, atmega603, at43usb320, at76c711).
10198 Instruction set avr4 is for the enhanced AVR core with up to 8K program
10199 memory space (MCU types: atmega8, atmega83, atmega85).
10201 Instruction set avr5 is for the enhanced AVR core with up to 128K program
10202 memory space (MCU types: atmega16, atmega161, atmega163, atmega32, atmega323,
10203 atmega64, atmega128, at43usb355, at94k).
10205 @item -mno-interrupts
10206 @opindex mno-interrupts
10207 Generated code is not compatible with hardware interrupts.
10208 Code size will be smaller.
10210 @item -mcall-prologues
10211 @opindex mcall-prologues
10212 Functions prologues/epilogues expanded as call to appropriate
10213 subroutines. Code size will be smaller.
10216 @opindex mtiny-stack
10217 Change only the low 8 bits of the stack pointer.
10221 Assume int to be 8 bit integer. This affects the sizes of all types: A
10222 char will be 1 byte, an int will be 1 byte, a long will be 2 bytes
10223 and long long will be 4 bytes. Please note that this option does not
10224 comply to the C standards, but it will provide you with smaller code
10228 @node Blackfin Options
10229 @subsection Blackfin Options
10230 @cindex Blackfin Options
10233 @item -mcpu=@var{cpu}@r{[}-@var{sirevision}@r{]}
10235 Specifies the name of the target Blackfin processor. Currently, @var{cpu}
10236 can be one of @samp{bf512}, @samp{bf514}, @samp{bf516}, @samp{bf518},
10237 @samp{bf522}, @samp{bf523}, @samp{bf524}, @samp{bf525}, @samp{bf526},
10238 @samp{bf527}, @samp{bf531}, @samp{bf532}, @samp{bf533},
10239 @samp{bf534}, @samp{bf536}, @samp{bf537}, @samp{bf538}, @samp{bf539},
10240 @samp{bf542}, @samp{bf544}, @samp{bf547}, @samp{bf548}, @samp{bf549},
10241 @samp{bf542m}, @samp{bf544m}, @samp{bf547m}, @samp{bf548m}, @samp{bf549m},
10243 The optional @var{sirevision} specifies the silicon revision of the target
10244 Blackfin processor. Any workarounds available for the targeted silicon revision
10245 will be enabled. If @var{sirevision} is @samp{none}, no workarounds are enabled.
10246 If @var{sirevision} is @samp{any}, all workarounds for the targeted processor
10247 will be enabled. The @code{__SILICON_REVISION__} macro is defined to two
10248 hexadecimal digits representing the major and minor numbers in the silicon
10249 revision. If @var{sirevision} is @samp{none}, the @code{__SILICON_REVISION__}
10250 is not defined. If @var{sirevision} is @samp{any}, the
10251 @code{__SILICON_REVISION__} is defined to be @code{0xffff}.
10252 If this optional @var{sirevision} is not used, GCC assumes the latest known
10253 silicon revision of the targeted Blackfin processor.
10255 Support for @samp{bf561} is incomplete. For @samp{bf561},
10256 Only the processor macro is defined.
10257 Without this option, @samp{bf532} is used as the processor by default.
10258 The corresponding predefined processor macros for @var{cpu} is to
10259 be defined. And for @samp{bfin-elf} toolchain, this causes the hardware BSP
10260 provided by libgloss to be linked in if @option{-msim} is not given.
10264 Specifies that the program will be run on the simulator. This causes
10265 the simulator BSP provided by libgloss to be linked in. This option
10266 has effect only for @samp{bfin-elf} toolchain.
10267 Certain other options, such as @option{-mid-shared-library} and
10268 @option{-mfdpic}, imply @option{-msim}.
10270 @item -momit-leaf-frame-pointer
10271 @opindex momit-leaf-frame-pointer
10272 Don't keep the frame pointer in a register for leaf functions. This
10273 avoids the instructions to save, set up and restore frame pointers and
10274 makes an extra register available in leaf functions. The option
10275 @option{-fomit-frame-pointer} removes the frame pointer for all functions
10276 which might make debugging harder.
10278 @item -mspecld-anomaly
10279 @opindex mspecld-anomaly
10280 When enabled, the compiler will ensure that the generated code does not
10281 contain speculative loads after jump instructions. If this option is used,
10282 @code{__WORKAROUND_SPECULATIVE_LOADS} is defined.
10284 @item -mno-specld-anomaly
10285 @opindex mno-specld-anomaly
10286 Don't generate extra code to prevent speculative loads from occurring.
10288 @item -mcsync-anomaly
10289 @opindex mcsync-anomaly
10290 When enabled, the compiler will ensure that the generated code does not
10291 contain CSYNC or SSYNC instructions too soon after conditional branches.
10292 If this option is used, @code{__WORKAROUND_SPECULATIVE_SYNCS} is defined.
10294 @item -mno-csync-anomaly
10295 @opindex mno-csync-anomaly
10296 Don't generate extra code to prevent CSYNC or SSYNC instructions from
10297 occurring too soon after a conditional branch.
10301 When enabled, the compiler is free to take advantage of the knowledge that
10302 the entire program fits into the low 64k of memory.
10305 @opindex mno-low-64k
10306 Assume that the program is arbitrarily large. This is the default.
10308 @item -mstack-check-l1
10309 @opindex mstack-check-l1
10310 Do stack checking using information placed into L1 scratchpad memory by the
10313 @item -mid-shared-library
10314 @opindex mid-shared-library
10315 Generate code that supports shared libraries via the library ID method.
10316 This allows for execute in place and shared libraries in an environment
10317 without virtual memory management. This option implies @option{-fPIC}.
10318 With a @samp{bfin-elf} target, this option implies @option{-msim}.
10320 @item -mno-id-shared-library
10321 @opindex mno-id-shared-library
10322 Generate code that doesn't assume ID based shared libraries are being used.
10323 This is the default.
10325 @item -mleaf-id-shared-library
10326 @opindex mleaf-id-shared-library
10327 Generate code that supports shared libraries via the library ID method,
10328 but assumes that this library or executable won't link against any other
10329 ID shared libraries. That allows the compiler to use faster code for jumps
10332 @item -mno-leaf-id-shared-library
10333 @opindex mno-leaf-id-shared-library
10334 Do not assume that the code being compiled won't link against any ID shared
10335 libraries. Slower code will be generated for jump and call insns.
10337 @item -mshared-library-id=n
10338 @opindex mshared-library-id
10339 Specified the identification number of the ID based shared library being
10340 compiled. Specifying a value of 0 will generate more compact code, specifying
10341 other values will force the allocation of that number to the current
10342 library but is no more space or time efficient than omitting this option.
10346 Generate code that allows the data segment to be located in a different
10347 area of memory from the text segment. This allows for execute in place in
10348 an environment without virtual memory management by eliminating relocations
10349 against the text section.
10351 @item -mno-sep-data
10352 @opindex mno-sep-data
10353 Generate code that assumes that the data segment follows the text segment.
10354 This is the default.
10357 @itemx -mno-long-calls
10358 @opindex mlong-calls
10359 @opindex mno-long-calls
10360 Tells the compiler to perform function calls by first loading the
10361 address of the function into a register and then performing a subroutine
10362 call on this register. This switch is needed if the target function
10363 will lie outside of the 24 bit addressing range of the offset based
10364 version of subroutine call instruction.
10366 This feature is not enabled by default. Specifying
10367 @option{-mno-long-calls} will restore the default behavior. Note these
10368 switches have no effect on how the compiler generates code to handle
10369 function calls via function pointers.
10373 Link with the fast floating-point library. This library relaxes some of
10374 the IEEE floating-point standard's rules for checking inputs against
10375 Not-a-Number (NAN), in the interest of performance.
10378 @opindex minline-plt
10379 Enable inlining of PLT entries in function calls to functions that are
10380 not known to bind locally. It has no effect without @option{-mfdpic}.
10383 @opindex mmulticore
10384 Build standalone application for multicore Blackfin processor. Proper
10385 start files and link scripts will be used to support multicore.
10386 This option defines @code{__BFIN_MULTICORE}. It can only be used with
10387 @option{-mcpu=bf561@r{[}-@var{sirevision}@r{]}}. It can be used with
10388 @option{-mcorea} or @option{-mcoreb}. If it's used without
10389 @option{-mcorea} or @option{-mcoreb}, single application/dual core
10390 programming model is used. In this model, the main function of Core B
10391 should be named as coreb_main. If it's used with @option{-mcorea} or
10392 @option{-mcoreb}, one application per core programming model is used.
10393 If this option is not used, single core application programming
10398 Build standalone application for Core A of BF561 when using
10399 one application per core programming model. Proper start files
10400 and link scripts will be used to support Core A. This option
10401 defines @code{__BFIN_COREA}. It must be used with @option{-mmulticore}.
10405 Build standalone application for Core B of BF561 when using
10406 one application per core programming model. Proper start files
10407 and link scripts will be used to support Core B. This option
10408 defines @code{__BFIN_COREB}. When this option is used, coreb_main
10409 should be used instead of main. It must be used with
10410 @option{-mmulticore}.
10414 Build standalone application for SDRAM. Proper start files and
10415 link scripts will be used to put the application into SDRAM.
10416 Loader should initialize SDRAM before loading the application
10417 into SDRAM. This option defines @code{__BFIN_SDRAM}.
10421 Assume that ICPLBs are enabled at runtime. This has an effect on certain
10422 anomaly workarounds. For Linux targets, the default is to assume ICPLBs
10423 are enabled; for standalone applications the default is off.
10427 @subsection CRIS Options
10428 @cindex CRIS Options
10430 These options are defined specifically for the CRIS ports.
10433 @item -march=@var{architecture-type}
10434 @itemx -mcpu=@var{architecture-type}
10437 Generate code for the specified architecture. The choices for
10438 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
10439 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX@.
10440 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
10443 @item -mtune=@var{architecture-type}
10445 Tune to @var{architecture-type} everything applicable about the generated
10446 code, except for the ABI and the set of available instructions. The
10447 choices for @var{architecture-type} are the same as for
10448 @option{-march=@var{architecture-type}}.
10450 @item -mmax-stack-frame=@var{n}
10451 @opindex mmax-stack-frame
10452 Warn when the stack frame of a function exceeds @var{n} bytes.
10458 The options @option{-metrax4} and @option{-metrax100} are synonyms for
10459 @option{-march=v3} and @option{-march=v8} respectively.
10461 @item -mmul-bug-workaround
10462 @itemx -mno-mul-bug-workaround
10463 @opindex mmul-bug-workaround
10464 @opindex mno-mul-bug-workaround
10465 Work around a bug in the @code{muls} and @code{mulu} instructions for CPU
10466 models where it applies. This option is active by default.
10470 Enable CRIS-specific verbose debug-related information in the assembly
10471 code. This option also has the effect to turn off the @samp{#NO_APP}
10472 formatted-code indicator to the assembler at the beginning of the
10477 Do not use condition-code results from previous instruction; always emit
10478 compare and test instructions before use of condition codes.
10480 @item -mno-side-effects
10481 @opindex mno-side-effects
10482 Do not emit instructions with side-effects in addressing modes other than
10485 @item -mstack-align
10486 @itemx -mno-stack-align
10487 @itemx -mdata-align
10488 @itemx -mno-data-align
10489 @itemx -mconst-align
10490 @itemx -mno-const-align
10491 @opindex mstack-align
10492 @opindex mno-stack-align
10493 @opindex mdata-align
10494 @opindex mno-data-align
10495 @opindex mconst-align
10496 @opindex mno-const-align
10497 These options (no-options) arranges (eliminate arrangements) for the
10498 stack-frame, individual data and constants to be aligned for the maximum
10499 single data access size for the chosen CPU model. The default is to
10500 arrange for 32-bit alignment. ABI details such as structure layout are
10501 not affected by these options.
10509 Similar to the stack- data- and const-align options above, these options
10510 arrange for stack-frame, writable data and constants to all be 32-bit,
10511 16-bit or 8-bit aligned. The default is 32-bit alignment.
10513 @item -mno-prologue-epilogue
10514 @itemx -mprologue-epilogue
10515 @opindex mno-prologue-epilogue
10516 @opindex mprologue-epilogue
10517 With @option{-mno-prologue-epilogue}, the normal function prologue and
10518 epilogue that sets up the stack-frame are omitted and no return
10519 instructions or return sequences are generated in the code. Use this
10520 option only together with visual inspection of the compiled code: no
10521 warnings or errors are generated when call-saved registers must be saved,
10522 or storage for local variable needs to be allocated.
10526 @opindex mno-gotplt
10528 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
10529 instruction sequences that load addresses for functions from the PLT part
10530 of the GOT rather than (traditional on other architectures) calls to the
10531 PLT@. The default is @option{-mgotplt}.
10535 Legacy no-op option only recognized with the cris-axis-elf and
10536 cris-axis-linux-gnu targets.
10540 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
10544 This option, recognized for the cris-axis-elf arranges
10545 to link with input-output functions from a simulator library. Code,
10546 initialized data and zero-initialized data are allocated consecutively.
10550 Like @option{-sim}, but pass linker options to locate initialized data at
10551 0x40000000 and zero-initialized data at 0x80000000.
10555 @subsection CRX Options
10556 @cindex CRX Options
10558 These options are defined specifically for the CRX ports.
10564 Enable the use of multiply-accumulate instructions. Disabled by default.
10567 @opindex mpush-args
10568 Push instructions will be used to pass outgoing arguments when functions
10569 are called. Enabled by default.
10572 @node Darwin Options
10573 @subsection Darwin Options
10574 @cindex Darwin options
10576 These options are defined for all architectures running the Darwin operating
10579 FSF GCC on Darwin does not create ``fat'' object files; it will create
10580 an object file for the single architecture that it was built to
10581 target. Apple's GCC on Darwin does create ``fat'' files if multiple
10582 @option{-arch} options are used; it does so by running the compiler or
10583 linker multiple times and joining the results together with
10586 The subtype of the file created (like @samp{ppc7400} or @samp{ppc970} or
10587 @samp{i686}) is determined by the flags that specify the ISA
10588 that GCC is targetting, like @option{-mcpu} or @option{-march}. The
10589 @option{-force_cpusubtype_ALL} option can be used to override this.
10591 The Darwin tools vary in their behavior when presented with an ISA
10592 mismatch. The assembler, @file{as}, will only permit instructions to
10593 be used that are valid for the subtype of the file it is generating,
10594 so you cannot put 64-bit instructions in a @samp{ppc750} object file.
10595 The linker for shared libraries, @file{/usr/bin/libtool}, will fail
10596 and print an error if asked to create a shared library with a less
10597 restrictive subtype than its input files (for instance, trying to put
10598 a @samp{ppc970} object file in a @samp{ppc7400} library). The linker
10599 for executables, @file{ld}, will quietly give the executable the most
10600 restrictive subtype of any of its input files.
10605 Add the framework directory @var{dir} to the head of the list of
10606 directories to be searched for header files. These directories are
10607 interleaved with those specified by @option{-I} options and are
10608 scanned in a left-to-right order.
10610 A framework directory is a directory with frameworks in it. A
10611 framework is a directory with a @samp{"Headers"} and/or
10612 @samp{"PrivateHeaders"} directory contained directly in it that ends
10613 in @samp{".framework"}. The name of a framework is the name of this
10614 directory excluding the @samp{".framework"}. Headers associated with
10615 the framework are found in one of those two directories, with
10616 @samp{"Headers"} being searched first. A subframework is a framework
10617 directory that is in a framework's @samp{"Frameworks"} directory.
10618 Includes of subframework headers can only appear in a header of a
10619 framework that contains the subframework, or in a sibling subframework
10620 header. Two subframeworks are siblings if they occur in the same
10621 framework. A subframework should not have the same name as a
10622 framework, a warning will be issued if this is violated. Currently a
10623 subframework cannot have subframeworks, in the future, the mechanism
10624 may be extended to support this. The standard frameworks can be found
10625 in @samp{"/System/Library/Frameworks"} and
10626 @samp{"/Library/Frameworks"}. An example include looks like
10627 @code{#include <Framework/header.h>}, where @samp{Framework} denotes
10628 the name of the framework and header.h is found in the
10629 @samp{"PrivateHeaders"} or @samp{"Headers"} directory.
10631 @item -iframework@var{dir}
10632 @opindex iframework
10633 Like @option{-F} except the directory is a treated as a system
10634 directory. The main difference between this @option{-iframework} and
10635 @option{-F} is that with @option{-iframework} the compiler does not
10636 warn about constructs contained within header files found via
10637 @var{dir}. This option is valid only for the C family of languages.
10641 Emit debugging information for symbols that are used. For STABS
10642 debugging format, this enables @option{-feliminate-unused-debug-symbols}.
10643 This is by default ON@.
10647 Emit debugging information for all symbols and types.
10649 @item -mmacosx-version-min=@var{version}
10650 The earliest version of MacOS X that this executable will run on
10651 is @var{version}. Typical values of @var{version} include @code{10.1},
10652 @code{10.2}, and @code{10.3.9}.
10654 If the compiler was built to use the system's headers by default,
10655 then the default for this option is the system version on which the
10656 compiler is running, otherwise the default is to make choices which
10657 are compatible with as many systems and code bases as possible.
10661 Enable kernel development mode. The @option{-mkernel} option sets
10662 @option{-static}, @option{-fno-common}, @option{-fno-cxa-atexit},
10663 @option{-fno-exceptions}, @option{-fno-non-call-exceptions},
10664 @option{-fapple-kext}, @option{-fno-weak} and @option{-fno-rtti} where
10665 applicable. This mode also sets @option{-mno-altivec},
10666 @option{-msoft-float}, @option{-fno-builtin} and
10667 @option{-mlong-branch} for PowerPC targets.
10669 @item -mone-byte-bool
10670 @opindex mone-byte-bool
10671 Override the defaults for @samp{bool} so that @samp{sizeof(bool)==1}.
10672 By default @samp{sizeof(bool)} is @samp{4} when compiling for
10673 Darwin/PowerPC and @samp{1} when compiling for Darwin/x86, so this
10674 option has no effect on x86.
10676 @strong{Warning:} The @option{-mone-byte-bool} switch causes GCC
10677 to generate code that is not binary compatible with code generated
10678 without that switch. Using this switch may require recompiling all
10679 other modules in a program, including system libraries. Use this
10680 switch to conform to a non-default data model.
10682 @item -mfix-and-continue
10683 @itemx -ffix-and-continue
10684 @itemx -findirect-data
10685 @opindex mfix-and-continue
10686 @opindex ffix-and-continue
10687 @opindex findirect-data
10688 Generate code suitable for fast turn around development. Needed to
10689 enable gdb to dynamically load @code{.o} files into already running
10690 programs. @option{-findirect-data} and @option{-ffix-and-continue}
10691 are provided for backwards compatibility.
10695 Loads all members of static archive libraries.
10696 See man ld(1) for more information.
10698 @item -arch_errors_fatal
10699 @opindex arch_errors_fatal
10700 Cause the errors having to do with files that have the wrong architecture
10703 @item -bind_at_load
10704 @opindex bind_at_load
10705 Causes the output file to be marked such that the dynamic linker will
10706 bind all undefined references when the file is loaded or launched.
10710 Produce a Mach-o bundle format file.
10711 See man ld(1) for more information.
10713 @item -bundle_loader @var{executable}
10714 @opindex bundle_loader
10715 This option specifies the @var{executable} that will be loading the build
10716 output file being linked. See man ld(1) for more information.
10719 @opindex dynamiclib
10720 When passed this option, GCC will produce a dynamic library instead of
10721 an executable when linking, using the Darwin @file{libtool} command.
10723 @item -force_cpusubtype_ALL
10724 @opindex force_cpusubtype_ALL
10725 This causes GCC's output file to have the @var{ALL} subtype, instead of
10726 one controlled by the @option{-mcpu} or @option{-march} option.
10728 @item -allowable_client @var{client_name}
10729 @itemx -client_name
10730 @itemx -compatibility_version
10731 @itemx -current_version
10733 @itemx -dependency-file
10735 @itemx -dylinker_install_name
10737 @itemx -exported_symbols_list
10739 @itemx -flat_namespace
10740 @itemx -force_flat_namespace
10741 @itemx -headerpad_max_install_names
10744 @itemx -install_name
10745 @itemx -keep_private_externs
10746 @itemx -multi_module
10747 @itemx -multiply_defined
10748 @itemx -multiply_defined_unused
10750 @itemx -no_dead_strip_inits_and_terms
10751 @itemx -nofixprebinding
10752 @itemx -nomultidefs
10754 @itemx -noseglinkedit
10755 @itemx -pagezero_size
10757 @itemx -prebind_all_twolevel_modules
10758 @itemx -private_bundle
10759 @itemx -read_only_relocs
10761 @itemx -sectobjectsymbols
10765 @itemx -sectobjectsymbols
10768 @itemx -segs_read_only_addr
10769 @itemx -segs_read_write_addr
10770 @itemx -seg_addr_table
10771 @itemx -seg_addr_table_filename
10772 @itemx -seglinkedit
10774 @itemx -segs_read_only_addr
10775 @itemx -segs_read_write_addr
10776 @itemx -single_module
10778 @itemx -sub_library
10779 @itemx -sub_umbrella
10780 @itemx -twolevel_namespace
10783 @itemx -unexported_symbols_list
10784 @itemx -weak_reference_mismatches
10785 @itemx -whatsloaded
10786 @opindex allowable_client
10787 @opindex client_name
10788 @opindex compatibility_version
10789 @opindex current_version
10790 @opindex dead_strip
10791 @opindex dependency-file
10792 @opindex dylib_file
10793 @opindex dylinker_install_name
10795 @opindex exported_symbols_list
10797 @opindex flat_namespace
10798 @opindex force_flat_namespace
10799 @opindex headerpad_max_install_names
10800 @opindex image_base
10802 @opindex install_name
10803 @opindex keep_private_externs
10804 @opindex multi_module
10805 @opindex multiply_defined
10806 @opindex multiply_defined_unused
10807 @opindex noall_load
10808 @opindex no_dead_strip_inits_and_terms
10809 @opindex nofixprebinding
10810 @opindex nomultidefs
10812 @opindex noseglinkedit
10813 @opindex pagezero_size
10815 @opindex prebind_all_twolevel_modules
10816 @opindex private_bundle
10817 @opindex read_only_relocs
10819 @opindex sectobjectsymbols
10822 @opindex sectcreate
10823 @opindex sectobjectsymbols
10826 @opindex segs_read_only_addr
10827 @opindex segs_read_write_addr
10828 @opindex seg_addr_table
10829 @opindex seg_addr_table_filename
10830 @opindex seglinkedit
10832 @opindex segs_read_only_addr
10833 @opindex segs_read_write_addr
10834 @opindex single_module
10836 @opindex sub_library
10837 @opindex sub_umbrella
10838 @opindex twolevel_namespace
10841 @opindex unexported_symbols_list
10842 @opindex weak_reference_mismatches
10843 @opindex whatsloaded
10844 These options are passed to the Darwin linker. The Darwin linker man page
10845 describes them in detail.
10848 @node DEC Alpha Options
10849 @subsection DEC Alpha Options
10851 These @samp{-m} options are defined for the DEC Alpha implementations:
10854 @item -mno-soft-float
10855 @itemx -msoft-float
10856 @opindex mno-soft-float
10857 @opindex msoft-float
10858 Use (do not use) the hardware floating-point instructions for
10859 floating-point operations. When @option{-msoft-float} is specified,
10860 functions in @file{libgcc.a} will be used to perform floating-point
10861 operations. Unless they are replaced by routines that emulate the
10862 floating-point operations, or compiled in such a way as to call such
10863 emulations routines, these routines will issue floating-point
10864 operations. If you are compiling for an Alpha without floating-point
10865 operations, you must ensure that the library is built so as not to call
10868 Note that Alpha implementations without floating-point operations are
10869 required to have floating-point registers.
10872 @itemx -mno-fp-regs
10874 @opindex mno-fp-regs
10875 Generate code that uses (does not use) the floating-point register set.
10876 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
10877 register set is not used, floating point operands are passed in integer
10878 registers as if they were integers and floating-point results are passed
10879 in @code{$0} instead of @code{$f0}. This is a non-standard calling sequence,
10880 so any function with a floating-point argument or return value called by code
10881 compiled with @option{-mno-fp-regs} must also be compiled with that
10884 A typical use of this option is building a kernel that does not use,
10885 and hence need not save and restore, any floating-point registers.
10889 The Alpha architecture implements floating-point hardware optimized for
10890 maximum performance. It is mostly compliant with the IEEE floating
10891 point standard. However, for full compliance, software assistance is
10892 required. This option generates code fully IEEE compliant code
10893 @emph{except} that the @var{inexact-flag} is not maintained (see below).
10894 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
10895 defined during compilation. The resulting code is less efficient but is
10896 able to correctly support denormalized numbers and exceptional IEEE
10897 values such as not-a-number and plus/minus infinity. Other Alpha
10898 compilers call this option @option{-ieee_with_no_inexact}.
10900 @item -mieee-with-inexact
10901 @opindex mieee-with-inexact
10902 This is like @option{-mieee} except the generated code also maintains
10903 the IEEE @var{inexact-flag}. Turning on this option causes the
10904 generated code to implement fully-compliant IEEE math. In addition to
10905 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
10906 macro. On some Alpha implementations the resulting code may execute
10907 significantly slower than the code generated by default. Since there is
10908 very little code that depends on the @var{inexact-flag}, you should
10909 normally not specify this option. Other Alpha compilers call this
10910 option @option{-ieee_with_inexact}.
10912 @item -mfp-trap-mode=@var{trap-mode}
10913 @opindex mfp-trap-mode
10914 This option controls what floating-point related traps are enabled.
10915 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
10916 The trap mode can be set to one of four values:
10920 This is the default (normal) setting. The only traps that are enabled
10921 are the ones that cannot be disabled in software (e.g., division by zero
10925 In addition to the traps enabled by @samp{n}, underflow traps are enabled
10929 Like @samp{u}, but the instructions are marked to be safe for software
10930 completion (see Alpha architecture manual for details).
10933 Like @samp{su}, but inexact traps are enabled as well.
10936 @item -mfp-rounding-mode=@var{rounding-mode}
10937 @opindex mfp-rounding-mode
10938 Selects the IEEE rounding mode. Other Alpha compilers call this option
10939 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
10944 Normal IEEE rounding mode. Floating point numbers are rounded towards
10945 the nearest machine number or towards the even machine number in case
10949 Round towards minus infinity.
10952 Chopped rounding mode. Floating point numbers are rounded towards zero.
10955 Dynamic rounding mode. A field in the floating point control register
10956 (@var{fpcr}, see Alpha architecture reference manual) controls the
10957 rounding mode in effect. The C library initializes this register for
10958 rounding towards plus infinity. Thus, unless your program modifies the
10959 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
10962 @item -mtrap-precision=@var{trap-precision}
10963 @opindex mtrap-precision
10964 In the Alpha architecture, floating point traps are imprecise. This
10965 means without software assistance it is impossible to recover from a
10966 floating trap and program execution normally needs to be terminated.
10967 GCC can generate code that can assist operating system trap handlers
10968 in determining the exact location that caused a floating point trap.
10969 Depending on the requirements of an application, different levels of
10970 precisions can be selected:
10974 Program precision. This option is the default and means a trap handler
10975 can only identify which program caused a floating point exception.
10978 Function precision. The trap handler can determine the function that
10979 caused a floating point exception.
10982 Instruction precision. The trap handler can determine the exact
10983 instruction that caused a floating point exception.
10986 Other Alpha compilers provide the equivalent options called
10987 @option{-scope_safe} and @option{-resumption_safe}.
10989 @item -mieee-conformant
10990 @opindex mieee-conformant
10991 This option marks the generated code as IEEE conformant. You must not
10992 use this option unless you also specify @option{-mtrap-precision=i} and either
10993 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
10994 is to emit the line @samp{.eflag 48} in the function prologue of the
10995 generated assembly file. Under DEC Unix, this has the effect that
10996 IEEE-conformant math library routines will be linked in.
10998 @item -mbuild-constants
10999 @opindex mbuild-constants
11000 Normally GCC examines a 32- or 64-bit integer constant to
11001 see if it can construct it from smaller constants in two or three
11002 instructions. If it cannot, it will output the constant as a literal and
11003 generate code to load it from the data segment at runtime.
11005 Use this option to require GCC to construct @emph{all} integer constants
11006 using code, even if it takes more instructions (the maximum is six).
11008 You would typically use this option to build a shared library dynamic
11009 loader. Itself a shared library, it must relocate itself in memory
11010 before it can find the variables and constants in its own data segment.
11016 Select whether to generate code to be assembled by the vendor-supplied
11017 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
11035 Indicate whether GCC should generate code to use the optional BWX,
11036 CIX, FIX and MAX instruction sets. The default is to use the instruction
11037 sets supported by the CPU type specified via @option{-mcpu=} option or that
11038 of the CPU on which GCC was built if none was specified.
11041 @itemx -mfloat-ieee
11042 @opindex mfloat-vax
11043 @opindex mfloat-ieee
11044 Generate code that uses (does not use) VAX F and G floating point
11045 arithmetic instead of IEEE single and double precision.
11047 @item -mexplicit-relocs
11048 @itemx -mno-explicit-relocs
11049 @opindex mexplicit-relocs
11050 @opindex mno-explicit-relocs
11051 Older Alpha assemblers provided no way to generate symbol relocations
11052 except via assembler macros. Use of these macros does not allow
11053 optimal instruction scheduling. GNU binutils as of version 2.12
11054 supports a new syntax that allows the compiler to explicitly mark
11055 which relocations should apply to which instructions. This option
11056 is mostly useful for debugging, as GCC detects the capabilities of
11057 the assembler when it is built and sets the default accordingly.
11060 @itemx -mlarge-data
11061 @opindex msmall-data
11062 @opindex mlarge-data
11063 When @option{-mexplicit-relocs} is in effect, static data is
11064 accessed via @dfn{gp-relative} relocations. When @option{-msmall-data}
11065 is used, objects 8 bytes long or smaller are placed in a @dfn{small data area}
11066 (the @code{.sdata} and @code{.sbss} sections) and are accessed via
11067 16-bit relocations off of the @code{$gp} register. This limits the
11068 size of the small data area to 64KB, but allows the variables to be
11069 directly accessed via a single instruction.
11071 The default is @option{-mlarge-data}. With this option the data area
11072 is limited to just below 2GB@. Programs that require more than 2GB of
11073 data must use @code{malloc} or @code{mmap} to allocate the data in the
11074 heap instead of in the program's data segment.
11076 When generating code for shared libraries, @option{-fpic} implies
11077 @option{-msmall-data} and @option{-fPIC} implies @option{-mlarge-data}.
11080 @itemx -mlarge-text
11081 @opindex msmall-text
11082 @opindex mlarge-text
11083 When @option{-msmall-text} is used, the compiler assumes that the
11084 code of the entire program (or shared library) fits in 4MB, and is
11085 thus reachable with a branch instruction. When @option{-msmall-data}
11086 is used, the compiler can assume that all local symbols share the
11087 same @code{$gp} value, and thus reduce the number of instructions
11088 required for a function call from 4 to 1.
11090 The default is @option{-mlarge-text}.
11092 @item -mcpu=@var{cpu_type}
11094 Set the instruction set and instruction scheduling parameters for
11095 machine type @var{cpu_type}. You can specify either the @samp{EV}
11096 style name or the corresponding chip number. GCC supports scheduling
11097 parameters for the EV4, EV5 and EV6 family of processors and will
11098 choose the default values for the instruction set from the processor
11099 you specify. If you do not specify a processor type, GCC will default
11100 to the processor on which the compiler was built.
11102 Supported values for @var{cpu_type} are
11108 Schedules as an EV4 and has no instruction set extensions.
11112 Schedules as an EV5 and has no instruction set extensions.
11116 Schedules as an EV5 and supports the BWX extension.
11121 Schedules as an EV5 and supports the BWX and MAX extensions.
11125 Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
11129 Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
11132 Native Linux/GNU toolchains also support the value @samp{native},
11133 which selects the best architecture option for the host processor.
11134 @option{-mcpu=native} has no effect if GCC does not recognize
11137 @item -mtune=@var{cpu_type}
11139 Set only the instruction scheduling parameters for machine type
11140 @var{cpu_type}. The instruction set is not changed.
11142 Native Linux/GNU toolchains also support the value @samp{native},
11143 which selects the best architecture option for the host processor.
11144 @option{-mtune=native} has no effect if GCC does not recognize
11147 @item -mmemory-latency=@var{time}
11148 @opindex mmemory-latency
11149 Sets the latency the scheduler should assume for typical memory
11150 references as seen by the application. This number is highly
11151 dependent on the memory access patterns used by the application
11152 and the size of the external cache on the machine.
11154 Valid options for @var{time} are
11158 A decimal number representing clock cycles.
11164 The compiler contains estimates of the number of clock cycles for
11165 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
11166 (also called Dcache, Scache, and Bcache), as well as to main memory.
11167 Note that L3 is only valid for EV5.
11172 @node DEC Alpha/VMS Options
11173 @subsection DEC Alpha/VMS Options
11175 These @samp{-m} options are defined for the DEC Alpha/VMS implementations:
11178 @item -mvms-return-codes
11179 @opindex mvms-return-codes
11180 Return VMS condition codes from main. The default is to return POSIX
11181 style condition (e.g.@: error) codes.
11183 @item -mdebug-main=@var{prefix}
11184 @opindex mdebug-main=@var{prefix}
11185 Flag the first routine whose name starts with @var{prefix} as the main
11186 routine for the debugger.
11190 Default to 64bit memory allocation routines.
11194 @subsection FR30 Options
11195 @cindex FR30 Options
11197 These options are defined specifically for the FR30 port.
11201 @item -msmall-model
11202 @opindex msmall-model
11203 Use the small address space model. This can produce smaller code, but
11204 it does assume that all symbolic values and addresses will fit into a
11209 Assume that run-time support has been provided and so there is no need
11210 to include the simulator library (@file{libsim.a}) on the linker
11216 @subsection FRV Options
11217 @cindex FRV Options
11223 Only use the first 32 general purpose registers.
11228 Use all 64 general purpose registers.
11233 Use only the first 32 floating point registers.
11238 Use all 64 floating point registers
11241 @opindex mhard-float
11243 Use hardware instructions for floating point operations.
11246 @opindex msoft-float
11248 Use library routines for floating point operations.
11253 Dynamically allocate condition code registers.
11258 Do not try to dynamically allocate condition code registers, only
11259 use @code{icc0} and @code{fcc0}.
11264 Change ABI to use double word insns.
11269 Do not use double word instructions.
11274 Use floating point double instructions.
11277 @opindex mno-double
11279 Do not use floating point double instructions.
11284 Use media instructions.
11289 Do not use media instructions.
11294 Use multiply and add/subtract instructions.
11297 @opindex mno-muladd
11299 Do not use multiply and add/subtract instructions.
11304 Select the FDPIC ABI, that uses function descriptors to represent
11305 pointers to functions. Without any PIC/PIE-related options, it
11306 implies @option{-fPIE}. With @option{-fpic} or @option{-fpie}, it
11307 assumes GOT entries and small data are within a 12-bit range from the
11308 GOT base address; with @option{-fPIC} or @option{-fPIE}, GOT offsets
11309 are computed with 32 bits.
11310 With a @samp{bfin-elf} target, this option implies @option{-msim}.
11313 @opindex minline-plt
11315 Enable inlining of PLT entries in function calls to functions that are
11316 not known to bind locally. It has no effect without @option{-mfdpic}.
11317 It's enabled by default if optimizing for speed and compiling for
11318 shared libraries (i.e., @option{-fPIC} or @option{-fpic}), or when an
11319 optimization option such as @option{-O3} or above is present in the
11325 Assume a large TLS segment when generating thread-local code.
11330 Do not assume a large TLS segment when generating thread-local code.
11335 Enable the use of @code{GPREL} relocations in the FDPIC ABI for data
11336 that is known to be in read-only sections. It's enabled by default,
11337 except for @option{-fpic} or @option{-fpie}: even though it may help
11338 make the global offset table smaller, it trades 1 instruction for 4.
11339 With @option{-fPIC} or @option{-fPIE}, it trades 3 instructions for 4,
11340 one of which may be shared by multiple symbols, and it avoids the need
11341 for a GOT entry for the referenced symbol, so it's more likely to be a
11342 win. If it is not, @option{-mno-gprel-ro} can be used to disable it.
11344 @item -multilib-library-pic
11345 @opindex multilib-library-pic
11347 Link with the (library, not FD) pic libraries. It's implied by
11348 @option{-mlibrary-pic}, as well as by @option{-fPIC} and
11349 @option{-fpic} without @option{-mfdpic}. You should never have to use
11353 @opindex mlinked-fp
11355 Follow the EABI requirement of always creating a frame pointer whenever
11356 a stack frame is allocated. This option is enabled by default and can
11357 be disabled with @option{-mno-linked-fp}.
11360 @opindex mlong-calls
11362 Use indirect addressing to call functions outside the current
11363 compilation unit. This allows the functions to be placed anywhere
11364 within the 32-bit address space.
11366 @item -malign-labels
11367 @opindex malign-labels
11369 Try to align labels to an 8-byte boundary by inserting nops into the
11370 previous packet. This option only has an effect when VLIW packing
11371 is enabled. It doesn't create new packets; it merely adds nops to
11374 @item -mlibrary-pic
11375 @opindex mlibrary-pic
11377 Generate position-independent EABI code.
11382 Use only the first four media accumulator registers.
11387 Use all eight media accumulator registers.
11392 Pack VLIW instructions.
11397 Do not pack VLIW instructions.
11400 @opindex mno-eflags
11402 Do not mark ABI switches in e_flags.
11405 @opindex mcond-move
11407 Enable the use of conditional-move instructions (default).
11409 This switch is mainly for debugging the compiler and will likely be removed
11410 in a future version.
11412 @item -mno-cond-move
11413 @opindex mno-cond-move
11415 Disable the use of conditional-move instructions.
11417 This switch is mainly for debugging the compiler and will likely be removed
11418 in a future version.
11423 Enable the use of conditional set instructions (default).
11425 This switch is mainly for debugging the compiler and will likely be removed
11426 in a future version.
11431 Disable the use of conditional set instructions.
11433 This switch is mainly for debugging the compiler and will likely be removed
11434 in a future version.
11437 @opindex mcond-exec
11439 Enable the use of conditional execution (default).
11441 This switch is mainly for debugging the compiler and will likely be removed
11442 in a future version.
11444 @item -mno-cond-exec
11445 @opindex mno-cond-exec
11447 Disable the use of conditional execution.
11449 This switch is mainly for debugging the compiler and will likely be removed
11450 in a future version.
11452 @item -mvliw-branch
11453 @opindex mvliw-branch
11455 Run a pass to pack branches into VLIW instructions (default).
11457 This switch is mainly for debugging the compiler and will likely be removed
11458 in a future version.
11460 @item -mno-vliw-branch
11461 @opindex mno-vliw-branch
11463 Do not run a pass to pack branches into VLIW instructions.
11465 This switch is mainly for debugging the compiler and will likely be removed
11466 in a future version.
11468 @item -mmulti-cond-exec
11469 @opindex mmulti-cond-exec
11471 Enable optimization of @code{&&} and @code{||} in conditional execution
11474 This switch is mainly for debugging the compiler and will likely be removed
11475 in a future version.
11477 @item -mno-multi-cond-exec
11478 @opindex mno-multi-cond-exec
11480 Disable optimization of @code{&&} and @code{||} in conditional execution.
11482 This switch is mainly for debugging the compiler and will likely be removed
11483 in a future version.
11485 @item -mnested-cond-exec
11486 @opindex mnested-cond-exec
11488 Enable nested conditional execution optimizations (default).
11490 This switch is mainly for debugging the compiler and will likely be removed
11491 in a future version.
11493 @item -mno-nested-cond-exec
11494 @opindex mno-nested-cond-exec
11496 Disable nested conditional execution optimizations.
11498 This switch is mainly for debugging the compiler and will likely be removed
11499 in a future version.
11501 @item -moptimize-membar
11502 @opindex moptimize-membar
11504 This switch removes redundant @code{membar} instructions from the
11505 compiler generated code. It is enabled by default.
11507 @item -mno-optimize-membar
11508 @opindex mno-optimize-membar
11510 This switch disables the automatic removal of redundant @code{membar}
11511 instructions from the generated code.
11513 @item -mtomcat-stats
11514 @opindex mtomcat-stats
11516 Cause gas to print out tomcat statistics.
11518 @item -mcpu=@var{cpu}
11521 Select the processor type for which to generate code. Possible values are
11522 @samp{frv}, @samp{fr550}, @samp{tomcat}, @samp{fr500}, @samp{fr450},
11523 @samp{fr405}, @samp{fr400}, @samp{fr300} and @samp{simple}.
11527 @node GNU/Linux Options
11528 @subsection GNU/Linux Options
11530 These @samp{-m} options are defined for GNU/Linux targets:
11535 Use the GNU C library. This is the default except
11536 on @samp{*-*-linux-*uclibc*} and @samp{*-*-linux-*android*} targets.
11540 Use uClibc C library. This is the default on
11541 @samp{*-*-linux-*uclibc*} targets.
11545 Use Bionic C library. This is the default on
11546 @samp{*-*-linux-*android*} targets.
11550 Compile code compatible with Android platform. This is the default on
11551 @samp{*-*-linux-*android*} targets.
11553 When compiling, this option enables @option{-mbionic}, @option{-fPIC},
11554 @option{-fno-exceptions} and @option{-fno-rtti} by default. When linking,
11555 this option makes the GCC driver pass Android-specific options to the linker.
11556 Finally, this option causes the preprocessor macro @code{__ANDROID__}
11559 @item -tno-android-cc
11560 @opindex tno-android-cc
11561 Disable compilation effects of @option{-mandroid}, i.e., do not enable
11562 @option{-mbionic}, @option{-fPIC}, @option{-fno-exceptions} and
11563 @option{-fno-rtti} by default.
11565 @item -tno-android-ld
11566 @opindex tno-android-ld
11567 Disable linking effects of @option{-mandroid}, i.e., pass standard Linux
11568 linking options to the linker.
11572 @node H8/300 Options
11573 @subsection H8/300 Options
11575 These @samp{-m} options are defined for the H8/300 implementations:
11580 Shorten some address references at link time, when possible; uses the
11581 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
11582 ld, Using ld}, for a fuller description.
11586 Generate code for the H8/300H@.
11590 Generate code for the H8S@.
11594 Generate code for the H8S and H8/300H in the normal mode. This switch
11595 must be used either with @option{-mh} or @option{-ms}.
11599 Generate code for the H8S/2600. This switch must be used with @option{-ms}.
11603 Make @code{int} data 32 bits by default.
11606 @opindex malign-300
11607 On the H8/300H and H8S, use the same alignment rules as for the H8/300.
11608 The default for the H8/300H and H8S is to align longs and floats on 4
11610 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
11611 This option has no effect on the H8/300.
11615 @subsection HPPA Options
11616 @cindex HPPA Options
11618 These @samp{-m} options are defined for the HPPA family of computers:
11621 @item -march=@var{architecture-type}
11623 Generate code for the specified architecture. The choices for
11624 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
11625 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
11626 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
11627 architecture option for your machine. Code compiled for lower numbered
11628 architectures will run on higher numbered architectures, but not the
11631 @item -mpa-risc-1-0
11632 @itemx -mpa-risc-1-1
11633 @itemx -mpa-risc-2-0
11634 @opindex mpa-risc-1-0
11635 @opindex mpa-risc-1-1
11636 @opindex mpa-risc-2-0
11637 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
11640 @opindex mbig-switch
11641 Generate code suitable for big switch tables. Use this option only if
11642 the assembler/linker complain about out of range branches within a switch
11645 @item -mjump-in-delay
11646 @opindex mjump-in-delay
11647 Fill delay slots of function calls with unconditional jump instructions
11648 by modifying the return pointer for the function call to be the target
11649 of the conditional jump.
11651 @item -mdisable-fpregs
11652 @opindex mdisable-fpregs
11653 Prevent floating point registers from being used in any manner. This is
11654 necessary for compiling kernels which perform lazy context switching of
11655 floating point registers. If you use this option and attempt to perform
11656 floating point operations, the compiler will abort.
11658 @item -mdisable-indexing
11659 @opindex mdisable-indexing
11660 Prevent the compiler from using indexing address modes. This avoids some
11661 rather obscure problems when compiling MIG generated code under MACH@.
11663 @item -mno-space-regs
11664 @opindex mno-space-regs
11665 Generate code that assumes the target has no space registers. This allows
11666 GCC to generate faster indirect calls and use unscaled index address modes.
11668 Such code is suitable for level 0 PA systems and kernels.
11670 @item -mfast-indirect-calls
11671 @opindex mfast-indirect-calls
11672 Generate code that assumes calls never cross space boundaries. This
11673 allows GCC to emit code which performs faster indirect calls.
11675 This option will not work in the presence of shared libraries or nested
11678 @item -mfixed-range=@var{register-range}
11679 @opindex mfixed-range
11680 Generate code treating the given register range as fixed registers.
11681 A fixed register is one that the register allocator can not use. This is
11682 useful when compiling kernel code. A register range is specified as
11683 two registers separated by a dash. Multiple register ranges can be
11684 specified separated by a comma.
11686 @item -mlong-load-store
11687 @opindex mlong-load-store
11688 Generate 3-instruction load and store sequences as sometimes required by
11689 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
11692 @item -mportable-runtime
11693 @opindex mportable-runtime
11694 Use the portable calling conventions proposed by HP for ELF systems.
11698 Enable the use of assembler directives only GAS understands.
11700 @item -mschedule=@var{cpu-type}
11702 Schedule code according to the constraints for the machine type
11703 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
11704 @samp{7100}, @samp{7100LC}, @samp{7200}, @samp{7300} and @samp{8000}. Refer
11705 to @file{/usr/lib/sched.models} on an HP-UX system to determine the
11706 proper scheduling option for your machine. The default scheduling is
11710 @opindex mlinker-opt
11711 Enable the optimization pass in the HP-UX linker. Note this makes symbolic
11712 debugging impossible. It also triggers a bug in the HP-UX 8 and HP-UX 9
11713 linkers in which they give bogus error messages when linking some programs.
11716 @opindex msoft-float
11717 Generate output containing library calls for floating point.
11718 @strong{Warning:} the requisite libraries are not available for all HPPA
11719 targets. Normally the facilities of the machine's usual C compiler are
11720 used, but this cannot be done directly in cross-compilation. You must make
11721 your own arrangements to provide suitable library functions for
11724 @option{-msoft-float} changes the calling convention in the output file;
11725 therefore, it is only useful if you compile @emph{all} of a program with
11726 this option. In particular, you need to compile @file{libgcc.a}, the
11727 library that comes with GCC, with @option{-msoft-float} in order for
11732 Generate the predefine, @code{_SIO}, for server IO@. The default is
11733 @option{-mwsio}. This generates the predefines, @code{__hp9000s700},
11734 @code{__hp9000s700__} and @code{_WSIO}, for workstation IO@. These
11735 options are available under HP-UX and HI-UX@.
11739 Use GNU ld specific options. This passes @option{-shared} to ld when
11740 building a shared library. It is the default when GCC is configured,
11741 explicitly or implicitly, with the GNU linker. This option does not
11742 have any affect on which ld is called, it only changes what parameters
11743 are passed to that ld. The ld that is called is determined by the
11744 @option{--with-ld} configure option, GCC's program search path, and
11745 finally by the user's @env{PATH}. The linker used by GCC can be printed
11746 using @samp{which `gcc -print-prog-name=ld`}. This option is only available
11747 on the 64 bit HP-UX GCC, i.e.@: configured with @samp{hppa*64*-*-hpux*}.
11751 Use HP ld specific options. This passes @option{-b} to ld when building
11752 a shared library and passes @option{+Accept TypeMismatch} to ld on all
11753 links. It is the default when GCC is configured, explicitly or
11754 implicitly, with the HP linker. This option does not have any affect on
11755 which ld is called, it only changes what parameters are passed to that
11756 ld. The ld that is called is determined by the @option{--with-ld}
11757 configure option, GCC's program search path, and finally by the user's
11758 @env{PATH}. The linker used by GCC can be printed using @samp{which
11759 `gcc -print-prog-name=ld`}. This option is only available on the 64 bit
11760 HP-UX GCC, i.e.@: configured with @samp{hppa*64*-*-hpux*}.
11763 @opindex mno-long-calls
11764 Generate code that uses long call sequences. This ensures that a call
11765 is always able to reach linker generated stubs. The default is to generate
11766 long calls only when the distance from the call site to the beginning
11767 of the function or translation unit, as the case may be, exceeds a
11768 predefined limit set by the branch type being used. The limits for
11769 normal calls are 7,600,000 and 240,000 bytes, respectively for the
11770 PA 2.0 and PA 1.X architectures. Sibcalls are always limited at
11773 Distances are measured from the beginning of functions when using the
11774 @option{-ffunction-sections} option, or when using the @option{-mgas}
11775 and @option{-mno-portable-runtime} options together under HP-UX with
11778 It is normally not desirable to use this option as it will degrade
11779 performance. However, it may be useful in large applications,
11780 particularly when partial linking is used to build the application.
11782 The types of long calls used depends on the capabilities of the
11783 assembler and linker, and the type of code being generated. The
11784 impact on systems that support long absolute calls, and long pic
11785 symbol-difference or pc-relative calls should be relatively small.
11786 However, an indirect call is used on 32-bit ELF systems in pic code
11787 and it is quite long.
11789 @item -munix=@var{unix-std}
11791 Generate compiler predefines and select a startfile for the specified
11792 UNIX standard. The choices for @var{unix-std} are @samp{93}, @samp{95}
11793 and @samp{98}. @samp{93} is supported on all HP-UX versions. @samp{95}
11794 is available on HP-UX 10.10 and later. @samp{98} is available on HP-UX
11795 11.11 and later. The default values are @samp{93} for HP-UX 10.00,
11796 @samp{95} for HP-UX 10.10 though to 11.00, and @samp{98} for HP-UX 11.11
11799 @option{-munix=93} provides the same predefines as GCC 3.3 and 3.4.
11800 @option{-munix=95} provides additional predefines for @code{XOPEN_UNIX}
11801 and @code{_XOPEN_SOURCE_EXTENDED}, and the startfile @file{unix95.o}.
11802 @option{-munix=98} provides additional predefines for @code{_XOPEN_UNIX},
11803 @code{_XOPEN_SOURCE_EXTENDED}, @code{_INCLUDE__STDC_A1_SOURCE} and
11804 @code{_INCLUDE_XOPEN_SOURCE_500}, and the startfile @file{unix98.o}.
11806 It is @emph{important} to note that this option changes the interfaces
11807 for various library routines. It also affects the operational behavior
11808 of the C library. Thus, @emph{extreme} care is needed in using this
11811 Library code that is intended to operate with more than one UNIX
11812 standard must test, set and restore the variable @var{__xpg4_extended_mask}
11813 as appropriate. Most GNU software doesn't provide this capability.
11817 Suppress the generation of link options to search libdld.sl when the
11818 @option{-static} option is specified on HP-UX 10 and later.
11822 The HP-UX implementation of setlocale in libc has a dependency on
11823 libdld.sl. There isn't an archive version of libdld.sl. Thus,
11824 when the @option{-static} option is specified, special link options
11825 are needed to resolve this dependency.
11827 On HP-UX 10 and later, the GCC driver adds the necessary options to
11828 link with libdld.sl when the @option{-static} option is specified.
11829 This causes the resulting binary to be dynamic. On the 64-bit port,
11830 the linkers generate dynamic binaries by default in any case. The
11831 @option{-nolibdld} option can be used to prevent the GCC driver from
11832 adding these link options.
11836 Add support for multithreading with the @dfn{dce thread} library
11837 under HP-UX@. This option sets flags for both the preprocessor and
11841 @node i386 and x86-64 Options
11842 @subsection Intel 386 and AMD x86-64 Options
11843 @cindex i386 Options
11844 @cindex x86-64 Options
11845 @cindex Intel 386 Options
11846 @cindex AMD x86-64 Options
11848 These @samp{-m} options are defined for the i386 and x86-64 family of
11852 @item -mtune=@var{cpu-type}
11854 Tune to @var{cpu-type} everything applicable about the generated code, except
11855 for the ABI and the set of available instructions. The choices for
11856 @var{cpu-type} are:
11859 Produce code optimized for the most common IA32/AMD64/EM64T processors.
11860 If you know the CPU on which your code will run, then you should use
11861 the corresponding @option{-mtune} option instead of
11862 @option{-mtune=generic}. But, if you do not know exactly what CPU users
11863 of your application will have, then you should use this option.
11865 As new processors are deployed in the marketplace, the behavior of this
11866 option will change. Therefore, if you upgrade to a newer version of
11867 GCC, the code generated option will change to reflect the processors
11868 that were most common when that version of GCC was released.
11870 There is no @option{-march=generic} option because @option{-march}
11871 indicates the instruction set the compiler can use, and there is no
11872 generic instruction set applicable to all processors. In contrast,
11873 @option{-mtune} indicates the processor (or, in this case, collection of
11874 processors) for which the code is optimized.
11876 This selects the CPU to tune for at compilation time by determining
11877 the processor type of the compiling machine. Using @option{-mtune=native}
11878 will produce code optimized for the local machine under the constraints
11879 of the selected instruction set. Using @option{-march=native} will
11880 enable all instruction subsets supported by the local machine (hence
11881 the result might not run on different machines).
11883 Original Intel's i386 CPU@.
11885 Intel's i486 CPU@. (No scheduling is implemented for this chip.)
11886 @item i586, pentium
11887 Intel Pentium CPU with no MMX support.
11889 Intel PentiumMMX CPU based on Pentium core with MMX instruction set support.
11891 Intel PentiumPro CPU@.
11893 Same as @code{generic}, but when used as @code{march} option, PentiumPro
11894 instruction set will be used, so the code will run on all i686 family chips.
11896 Intel Pentium2 CPU based on PentiumPro core with MMX instruction set support.
11897 @item pentium3, pentium3m
11898 Intel Pentium3 CPU based on PentiumPro core with MMX and SSE instruction set
11901 Low power version of Intel Pentium3 CPU with MMX, SSE and SSE2 instruction set
11902 support. Used by Centrino notebooks.
11903 @item pentium4, pentium4m
11904 Intel Pentium4 CPU with MMX, SSE and SSE2 instruction set support.
11906 Improved version of Intel Pentium4 CPU with MMX, SSE, SSE2 and SSE3 instruction
11909 Improved version of Intel Pentium4 CPU with 64-bit extensions, MMX, SSE,
11910 SSE2 and SSE3 instruction set support.
11912 Intel Core2 CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3 and SSSE3
11913 instruction set support.
11915 Intel Atom CPU with 64-bit extensions, MMX, SSE, SSE2, SSE3 and SSSE3
11916 instruction set support.
11918 AMD K6 CPU with MMX instruction set support.
11920 Improved versions of AMD K6 CPU with MMX and 3DNow!@: instruction set support.
11921 @item athlon, athlon-tbird
11922 AMD Athlon CPU with MMX, 3dNOW!, enhanced 3DNow!@: and SSE prefetch instructions
11924 @item athlon-4, athlon-xp, athlon-mp
11925 Improved AMD Athlon CPU with MMX, 3DNow!, enhanced 3DNow!@: and full SSE
11926 instruction set support.
11927 @item k8, opteron, athlon64, athlon-fx
11928 AMD K8 core based CPUs with x86-64 instruction set support. (This supersets
11929 MMX, SSE, SSE2, 3DNow!, enhanced 3DNow!@: and 64-bit instruction set extensions.)
11930 @item k8-sse3, opteron-sse3, athlon64-sse3
11931 Improved versions of k8, opteron and athlon64 with SSE3 instruction set support.
11932 @item amdfam10, barcelona
11933 AMD Family 10h core based CPUs with x86-64 instruction set support. (This
11934 supersets MMX, SSE, SSE2, SSE3, SSE4A, 3DNow!, enhanced 3DNow!, ABM and 64-bit
11935 instruction set extensions.)
11937 IDT Winchip C6 CPU, dealt in same way as i486 with additional MMX instruction
11940 IDT Winchip2 CPU, dealt in same way as i486 with additional MMX and 3DNow!@:
11941 instruction set support.
11943 Via C3 CPU with MMX and 3DNow!@: instruction set support. (No scheduling is
11944 implemented for this chip.)
11946 Via C3-2 CPU with MMX and SSE instruction set support. (No scheduling is
11947 implemented for this chip.)
11949 Embedded AMD CPU with MMX and 3DNow!@: instruction set support.
11952 While picking a specific @var{cpu-type} will schedule things appropriately
11953 for that particular chip, the compiler will not generate any code that
11954 does not run on the i386 without the @option{-march=@var{cpu-type}} option
11957 @item -march=@var{cpu-type}
11959 Generate instructions for the machine type @var{cpu-type}. The choices
11960 for @var{cpu-type} are the same as for @option{-mtune}. Moreover,
11961 specifying @option{-march=@var{cpu-type}} implies @option{-mtune=@var{cpu-type}}.
11963 @item -mcpu=@var{cpu-type}
11965 A deprecated synonym for @option{-mtune}.
11967 @item -mfpmath=@var{unit}
11969 Generate floating point arithmetics for selected unit @var{unit}. The choices
11970 for @var{unit} are:
11974 Use the standard 387 floating point coprocessor present majority of chips and
11975 emulated otherwise. Code compiled with this option will run almost everywhere.
11976 The temporary results are computed in 80bit precision instead of precision
11977 specified by the type resulting in slightly different results compared to most
11978 of other chips. See @option{-ffloat-store} for more detailed description.
11980 This is the default choice for i386 compiler.
11983 Use scalar floating point instructions present in the SSE instruction set.
11984 This instruction set is supported by Pentium3 and newer chips, in the AMD line
11985 by Athlon-4, Athlon-xp and Athlon-mp chips. The earlier version of SSE
11986 instruction set supports only single precision arithmetics, thus the double and
11987 extended precision arithmetics is still done using 387. Later version, present
11988 only in Pentium4 and the future AMD x86-64 chips supports double precision
11991 For the i386 compiler, you need to use @option{-march=@var{cpu-type}}, @option{-msse}
11992 or @option{-msse2} switches to enable SSE extensions and make this option
11993 effective. For the x86-64 compiler, these extensions are enabled by default.
11995 The resulting code should be considerably faster in the majority of cases and avoid
11996 the numerical instability problems of 387 code, but may break some existing
11997 code that expects temporaries to be 80bit.
11999 This is the default choice for the x86-64 compiler.
12004 Attempt to utilize both instruction sets at once. This effectively double the
12005 amount of available registers and on chips with separate execution units for
12006 387 and SSE the execution resources too. Use this option with care, as it is
12007 still experimental, because the GCC register allocator does not model separate
12008 functional units well resulting in instable performance.
12011 @item -masm=@var{dialect}
12012 @opindex masm=@var{dialect}
12013 Output asm instructions using selected @var{dialect}. Supported
12014 choices are @samp{intel} or @samp{att} (the default one). Darwin does
12015 not support @samp{intel}.
12018 @itemx -mno-ieee-fp
12020 @opindex mno-ieee-fp
12021 Control whether or not the compiler uses IEEE floating point
12022 comparisons. These handle correctly the case where the result of a
12023 comparison is unordered.
12026 @opindex msoft-float
12027 Generate output containing library calls for floating point.
12028 @strong{Warning:} the requisite libraries are not part of GCC@.
12029 Normally the facilities of the machine's usual C compiler are used, but
12030 this can't be done directly in cross-compilation. You must make your
12031 own arrangements to provide suitable library functions for
12034 On machines where a function returns floating point results in the 80387
12035 register stack, some floating point opcodes may be emitted even if
12036 @option{-msoft-float} is used.
12038 @item -mno-fp-ret-in-387
12039 @opindex mno-fp-ret-in-387
12040 Do not use the FPU registers for return values of functions.
12042 The usual calling convention has functions return values of types
12043 @code{float} and @code{double} in an FPU register, even if there
12044 is no FPU@. The idea is that the operating system should emulate
12047 The option @option{-mno-fp-ret-in-387} causes such values to be returned
12048 in ordinary CPU registers instead.
12050 @item -mno-fancy-math-387
12051 @opindex mno-fancy-math-387
12052 Some 387 emulators do not support the @code{sin}, @code{cos} and
12053 @code{sqrt} instructions for the 387. Specify this option to avoid
12054 generating those instructions. This option is the default on FreeBSD,
12055 OpenBSD and NetBSD@. This option is overridden when @option{-march}
12056 indicates that the target cpu will always have an FPU and so the
12057 instruction will not need emulation. As of revision 2.6.1, these
12058 instructions are not generated unless you also use the
12059 @option{-funsafe-math-optimizations} switch.
12061 @item -malign-double
12062 @itemx -mno-align-double
12063 @opindex malign-double
12064 @opindex mno-align-double
12065 Control whether GCC aligns @code{double}, @code{long double}, and
12066 @code{long long} variables on a two word boundary or a one word
12067 boundary. Aligning @code{double} variables on a two word boundary will
12068 produce code that runs somewhat faster on a @samp{Pentium} at the
12069 expense of more memory.
12071 On x86-64, @option{-malign-double} is enabled by default.
12073 @strong{Warning:} if you use the @option{-malign-double} switch,
12074 structures containing the above types will be aligned differently than
12075 the published application binary interface specifications for the 386
12076 and will not be binary compatible with structures in code compiled
12077 without that switch.
12079 @item -m96bit-long-double
12080 @itemx -m128bit-long-double
12081 @opindex m96bit-long-double
12082 @opindex m128bit-long-double
12083 These switches control the size of @code{long double} type. The i386
12084 application binary interface specifies the size to be 96 bits,
12085 so @option{-m96bit-long-double} is the default in 32 bit mode.
12087 Modern architectures (Pentium and newer) would prefer @code{long double}
12088 to be aligned to an 8 or 16 byte boundary. In arrays or structures
12089 conforming to the ABI, this would not be possible. So specifying a
12090 @option{-m128bit-long-double} will align @code{long double}
12091 to a 16 byte boundary by padding the @code{long double} with an additional
12094 In the x86-64 compiler, @option{-m128bit-long-double} is the default choice as
12095 its ABI specifies that @code{long double} is to be aligned on 16 byte boundary.
12097 Notice that neither of these options enable any extra precision over the x87
12098 standard of 80 bits for a @code{long double}.
12100 @strong{Warning:} if you override the default value for your target ABI, the
12101 structures and arrays containing @code{long double} variables will change
12102 their size as well as function calling convention for function taking
12103 @code{long double} will be modified. Hence they will not be binary
12104 compatible with arrays or structures in code compiled without that switch.
12106 @item -mlarge-data-threshold=@var{number}
12107 @opindex mlarge-data-threshold=@var{number}
12108 When @option{-mcmodel=medium} is specified, the data greater than
12109 @var{threshold} are placed in large data section. This value must be the
12110 same across all object linked into the binary and defaults to 65535.
12114 Use a different function-calling convention, in which functions that
12115 take a fixed number of arguments return with the @code{ret} @var{num}
12116 instruction, which pops their arguments while returning. This saves one
12117 instruction in the caller since there is no need to pop the arguments
12120 You can specify that an individual function is called with this calling
12121 sequence with the function attribute @samp{stdcall}. You can also
12122 override the @option{-mrtd} option by using the function attribute
12123 @samp{cdecl}. @xref{Function Attributes}.
12125 @strong{Warning:} this calling convention is incompatible with the one
12126 normally used on Unix, so you cannot use it if you need to call
12127 libraries compiled with the Unix compiler.
12129 Also, you must provide function prototypes for all functions that
12130 take variable numbers of arguments (including @code{printf});
12131 otherwise incorrect code will be generated for calls to those
12134 In addition, seriously incorrect code will result if you call a
12135 function with too many arguments. (Normally, extra arguments are
12136 harmlessly ignored.)
12138 @item -mregparm=@var{num}
12140 Control how many registers are used to pass integer arguments. By
12141 default, no registers are used to pass arguments, and at most 3
12142 registers can be used. You can control this behavior for a specific
12143 function by using the function attribute @samp{regparm}.
12144 @xref{Function Attributes}.
12146 @strong{Warning:} if you use this switch, and
12147 @var{num} is nonzero, then you must build all modules with the same
12148 value, including any libraries. This includes the system libraries and
12152 @opindex msseregparm
12153 Use SSE register passing conventions for float and double arguments
12154 and return values. You can control this behavior for a specific
12155 function by using the function attribute @samp{sseregparm}.
12156 @xref{Function Attributes}.
12158 @strong{Warning:} if you use this switch then you must build all
12159 modules with the same value, including any libraries. This includes
12160 the system libraries and startup modules.
12169 Set 80387 floating-point precision to 32, 64 or 80 bits. When @option{-mpc32}
12170 is specified, the significands of results of floating-point operations are
12171 rounded to 24 bits (single precision); @option{-mpc64} rounds the
12172 significands of results of floating-point operations to 53 bits (double
12173 precision) and @option{-mpc80} rounds the significands of results of
12174 floating-point operations to 64 bits (extended double precision), which is
12175 the default. When this option is used, floating-point operations in higher
12176 precisions are not available to the programmer without setting the FPU
12177 control word explicitly.
12179 Setting the rounding of floating-point operations to less than the default
12180 80 bits can speed some programs by 2% or more. Note that some mathematical
12181 libraries assume that extended precision (80 bit) floating-point operations
12182 are enabled by default; routines in such libraries could suffer significant
12183 loss of accuracy, typically through so-called "catastrophic cancellation",
12184 when this option is used to set the precision to less than extended precision.
12186 @item -mstackrealign
12187 @opindex mstackrealign
12188 Realign the stack at entry. On the Intel x86, the @option{-mstackrealign}
12189 option will generate an alternate prologue and epilogue that realigns the
12190 runtime stack if necessary. This supports mixing legacy codes that keep
12191 a 4-byte aligned stack with modern codes that keep a 16-byte stack for
12192 SSE compatibility. See also the attribute @code{force_align_arg_pointer},
12193 applicable to individual functions.
12195 @item -mpreferred-stack-boundary=@var{num}
12196 @opindex mpreferred-stack-boundary
12197 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
12198 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
12199 the default is 4 (16 bytes or 128 bits).
12201 @item -mincoming-stack-boundary=@var{num}
12202 @opindex mincoming-stack-boundary
12203 Assume the incoming stack is aligned to a 2 raised to @var{num} byte
12204 boundary. If @option{-mincoming-stack-boundary} is not specified,
12205 the one specified by @option{-mpreferred-stack-boundary} will be used.
12207 On Pentium and PentiumPro, @code{double} and @code{long double} values
12208 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
12209 suffer significant run time performance penalties. On Pentium III, the
12210 Streaming SIMD Extension (SSE) data type @code{__m128} may not work
12211 properly if it is not 16 byte aligned.
12213 To ensure proper alignment of this values on the stack, the stack boundary
12214 must be as aligned as that required by any value stored on the stack.
12215 Further, every function must be generated such that it keeps the stack
12216 aligned. Thus calling a function compiled with a higher preferred
12217 stack boundary from a function compiled with a lower preferred stack
12218 boundary will most likely misalign the stack. It is recommended that
12219 libraries that use callbacks always use the default setting.
12221 This extra alignment does consume extra stack space, and generally
12222 increases code size. Code that is sensitive to stack space usage, such
12223 as embedded systems and operating system kernels, may want to reduce the
12224 preferred alignment to @option{-mpreferred-stack-boundary=2}.
12249 @itemx -mno-fsgsbase
12274 These switches enable or disable the use of instructions in the MMX,
12275 SSE, SSE2, SSE3, SSSE3, SSE4.1, AVX, AES, PCLMUL, FSGSBASE, RDRND,
12276 F16C, SSE4A, FMA4, XOP, LWP, ABM or 3DNow!@: extended instruction sets.
12277 These extensions are also available as built-in functions: see
12278 @ref{X86 Built-in Functions}, for details of the functions enabled and
12279 disabled by these switches.
12281 To have SSE/SSE2 instructions generated automatically from floating-point
12282 code (as opposed to 387 instructions), see @option{-mfpmath=sse}.
12284 GCC depresses SSEx instructions when @option{-mavx} is used. Instead, it
12285 generates new AVX instructions or AVX equivalence for all SSEx instructions
12288 These options will enable GCC to use these extended instructions in
12289 generated code, even without @option{-mfpmath=sse}. Applications which
12290 perform runtime CPU detection must compile separate files for each
12291 supported architecture, using the appropriate flags. In particular,
12292 the file containing the CPU detection code should be compiled without
12296 @itemx -mno-fused-madd
12297 @opindex mfused-madd
12298 @opindex mno-fused-madd
12299 Do (don't) generate code that uses the fused multiply/add or multiply/subtract
12300 instructions. The default is to use these instructions.
12304 This option instructs GCC to emit a @code{cld} instruction in the prologue
12305 of functions that use string instructions. String instructions depend on
12306 the DF flag to select between autoincrement or autodecrement mode. While the
12307 ABI specifies the DF flag to be cleared on function entry, some operating
12308 systems violate this specification by not clearing the DF flag in their
12309 exception dispatchers. The exception handler can be invoked with the DF flag
12310 set which leads to wrong direction mode, when string instructions are used.
12311 This option can be enabled by default on 32-bit x86 targets by configuring
12312 GCC with the @option{--enable-cld} configure option. Generation of @code{cld}
12313 instructions can be suppressed with the @option{-mno-cld} compiler option
12318 This option will enable GCC to use CMPXCHG16B instruction in generated code.
12319 CMPXCHG16B allows for atomic operations on 128-bit double quadword (or oword)
12320 data types. This is useful for high resolution counters that could be updated
12321 by multiple processors (or cores). This instruction is generated as part of
12322 atomic built-in functions: see @ref{Atomic Builtins} for details.
12326 This option will enable GCC to use SAHF instruction in generated 64-bit code.
12327 Early Intel CPUs with Intel 64 lacked LAHF and SAHF instructions supported
12328 by AMD64 until introduction of Pentium 4 G1 step in December 2005. LAHF and
12329 SAHF are load and store instructions, respectively, for certain status flags.
12330 In 64-bit mode, SAHF instruction is used to optimize @code{fmod}, @code{drem}
12331 or @code{remainder} built-in functions: see @ref{Other Builtins} for details.
12335 This option will enable GCC to use movbe instruction to implement
12336 @code{__builtin_bswap32} and @code{__builtin_bswap64}.
12340 This option will enable built-in functions, @code{__builtin_ia32_crc32qi},
12341 @code{__builtin_ia32_crc32hi}. @code{__builtin_ia32_crc32si} and
12342 @code{__builtin_ia32_crc32di} to generate the crc32 machine instruction.
12346 This option will enable GCC to use RCPSS and RSQRTSS instructions (and their
12347 vectorized variants RCPPS and RSQRTPS) with an additional Newton-Raphson step
12348 to increase precision instead of DIVSS and SQRTSS (and their vectorized
12349 variants) for single precision floating point arguments. These instructions
12350 are generated only when @option{-funsafe-math-optimizations} is enabled
12351 together with @option{-finite-math-only} and @option{-fno-trapping-math}.
12352 Note that while the throughput of the sequence is higher than the throughput
12353 of the non-reciprocal instruction, the precision of the sequence can be
12354 decreased by up to 2 ulp (i.e. the inverse of 1.0 equals 0.99999994).
12356 Note that GCC implements 1.0f/sqrtf(x) in terms of RSQRTSS (or RSQRTPS)
12357 already with @option{-ffast-math} (or the above option combination), and
12358 doesn't need @option{-mrecip}.
12360 @item -mveclibabi=@var{type}
12361 @opindex mveclibabi
12362 Specifies the ABI type to use for vectorizing intrinsics using an
12363 external library. Supported types are @code{svml} for the Intel short
12364 vector math library and @code{acml} for the AMD math core library style
12365 of interfacing. GCC will currently emit calls to @code{vmldExp2},
12366 @code{vmldLn2}, @code{vmldLog102}, @code{vmldLog102}, @code{vmldPow2},
12367 @code{vmldTanh2}, @code{vmldTan2}, @code{vmldAtan2}, @code{vmldAtanh2},
12368 @code{vmldCbrt2}, @code{vmldSinh2}, @code{vmldSin2}, @code{vmldAsinh2},
12369 @code{vmldAsin2}, @code{vmldCosh2}, @code{vmldCos2}, @code{vmldAcosh2},
12370 @code{vmldAcos2}, @code{vmlsExp4}, @code{vmlsLn4}, @code{vmlsLog104},
12371 @code{vmlsLog104}, @code{vmlsPow4}, @code{vmlsTanh4}, @code{vmlsTan4},
12372 @code{vmlsAtan4}, @code{vmlsAtanh4}, @code{vmlsCbrt4}, @code{vmlsSinh4},
12373 @code{vmlsSin4}, @code{vmlsAsinh4}, @code{vmlsAsin4}, @code{vmlsCosh4},
12374 @code{vmlsCos4}, @code{vmlsAcosh4} and @code{vmlsAcos4} for corresponding
12375 function type when @option{-mveclibabi=svml} is used and @code{__vrd2_sin},
12376 @code{__vrd2_cos}, @code{__vrd2_exp}, @code{__vrd2_log}, @code{__vrd2_log2},
12377 @code{__vrd2_log10}, @code{__vrs4_sinf}, @code{__vrs4_cosf},
12378 @code{__vrs4_expf}, @code{__vrs4_logf}, @code{__vrs4_log2f},
12379 @code{__vrs4_log10f} and @code{__vrs4_powf} for corresponding function type
12380 when @option{-mveclibabi=acml} is used. Both @option{-ftree-vectorize} and
12381 @option{-funsafe-math-optimizations} have to be enabled. A SVML or ACML ABI
12382 compatible library will have to be specified at link time.
12384 @item -mabi=@var{name}
12386 Generate code for the specified calling convention. Permissible values
12387 are: @samp{sysv} for the ABI used on GNU/Linux and other systems and
12388 @samp{ms} for the Microsoft ABI. The default is to use the Microsoft
12389 ABI when targeting Windows. On all other systems, the default is the
12390 SYSV ABI. You can control this behavior for a specific function by
12391 using the function attribute @samp{ms_abi}/@samp{sysv_abi}.
12392 @xref{Function Attributes}.
12395 @itemx -mno-push-args
12396 @opindex mpush-args
12397 @opindex mno-push-args
12398 Use PUSH operations to store outgoing parameters. This method is shorter
12399 and usually equally fast as method using SUB/MOV operations and is enabled
12400 by default. In some cases disabling it may improve performance because of
12401 improved scheduling and reduced dependencies.
12403 @item -maccumulate-outgoing-args
12404 @opindex maccumulate-outgoing-args
12405 If enabled, the maximum amount of space required for outgoing arguments will be
12406 computed in the function prologue. This is faster on most modern CPUs
12407 because of reduced dependencies, improved scheduling and reduced stack usage
12408 when preferred stack boundary is not equal to 2. The drawback is a notable
12409 increase in code size. This switch implies @option{-mno-push-args}.
12413 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
12414 on thread-safe exception handling must compile and link all code with the
12415 @option{-mthreads} option. When compiling, @option{-mthreads} defines
12416 @option{-D_MT}; when linking, it links in a special thread helper library
12417 @option{-lmingwthrd} which cleans up per thread exception handling data.
12419 @item -mno-align-stringops
12420 @opindex mno-align-stringops
12421 Do not align destination of inlined string operations. This switch reduces
12422 code size and improves performance in case the destination is already aligned,
12423 but GCC doesn't know about it.
12425 @item -minline-all-stringops
12426 @opindex minline-all-stringops
12427 By default GCC inlines string operations only when destination is known to be
12428 aligned at least to 4 byte boundary. This enables more inlining, increase code
12429 size, but may improve performance of code that depends on fast memcpy, strlen
12430 and memset for short lengths.
12432 @item -minline-stringops-dynamically
12433 @opindex minline-stringops-dynamically
12434 For string operation of unknown size, inline runtime checks so for small
12435 blocks inline code is used, while for large blocks library call is used.
12437 @item -mstringop-strategy=@var{alg}
12438 @opindex mstringop-strategy=@var{alg}
12439 Overwrite internal decision heuristic about particular algorithm to inline
12440 string operation with. The allowed values are @code{rep_byte},
12441 @code{rep_4byte}, @code{rep_8byte} for expanding using i386 @code{rep} prefix
12442 of specified size, @code{byte_loop}, @code{loop}, @code{unrolled_loop} for
12443 expanding inline loop, @code{libcall} for always expanding library call.
12445 @item -momit-leaf-frame-pointer
12446 @opindex momit-leaf-frame-pointer
12447 Don't keep the frame pointer in a register for leaf functions. This
12448 avoids the instructions to save, set up and restore frame pointers and
12449 makes an extra register available in leaf functions. The option
12450 @option{-fomit-frame-pointer} removes the frame pointer for all functions
12451 which might make debugging harder.
12453 @item -mtls-direct-seg-refs
12454 @itemx -mno-tls-direct-seg-refs
12455 @opindex mtls-direct-seg-refs
12456 Controls whether TLS variables may be accessed with offsets from the
12457 TLS segment register (@code{%gs} for 32-bit, @code{%fs} for 64-bit),
12458 or whether the thread base pointer must be added. Whether or not this
12459 is legal depends on the operating system, and whether it maps the
12460 segment to cover the entire TLS area.
12462 For systems that use GNU libc, the default is on.
12465 @itemx -mno-sse2avx
12467 Specify that the assembler should encode SSE instructions with VEX
12468 prefix. The option @option{-mavx} turns this on by default.
12471 These @samp{-m} switches are supported in addition to the above
12472 on AMD x86-64 processors in 64-bit environments.
12479 Generate code for a 32-bit or 64-bit environment.
12480 The 32-bit environment sets int, long and pointer to 32 bits and
12481 generates code that runs on any i386 system.
12482 The 64-bit environment sets int to 32 bits and long and pointer
12483 to 64 bits and generates code for AMD's x86-64 architecture. For
12484 darwin only the -m64 option turns off the @option{-fno-pic} and
12485 @option{-mdynamic-no-pic} options.
12487 @item -mno-red-zone
12488 @opindex mno-red-zone
12489 Do not use a so called red zone for x86-64 code. The red zone is mandated
12490 by the x86-64 ABI, it is a 128-byte area beyond the location of the
12491 stack pointer that will not be modified by signal or interrupt handlers
12492 and therefore can be used for temporary data without adjusting the stack
12493 pointer. The flag @option{-mno-red-zone} disables this red zone.
12495 @item -mcmodel=small
12496 @opindex mcmodel=small
12497 Generate code for the small code model: the program and its symbols must
12498 be linked in the lower 2 GB of the address space. Pointers are 64 bits.
12499 Programs can be statically or dynamically linked. This is the default
12502 @item -mcmodel=kernel
12503 @opindex mcmodel=kernel
12504 Generate code for the kernel code model. The kernel runs in the
12505 negative 2 GB of the address space.
12506 This model has to be used for Linux kernel code.
12508 @item -mcmodel=medium
12509 @opindex mcmodel=medium
12510 Generate code for the medium model: The program is linked in the lower 2
12511 GB of the address space. Small symbols are also placed there. Symbols
12512 with sizes larger than @option{-mlarge-data-threshold} are put into
12513 large data or bss sections and can be located above 2GB. Programs can
12514 be statically or dynamically linked.
12516 @item -mcmodel=large
12517 @opindex mcmodel=large
12518 Generate code for the large model: This model makes no assumptions
12519 about addresses and sizes of sections.
12522 @node IA-64 Options
12523 @subsection IA-64 Options
12524 @cindex IA-64 Options
12526 These are the @samp{-m} options defined for the Intel IA-64 architecture.
12530 @opindex mbig-endian
12531 Generate code for a big endian target. This is the default for HP-UX@.
12533 @item -mlittle-endian
12534 @opindex mlittle-endian
12535 Generate code for a little endian target. This is the default for AIX5
12541 @opindex mno-gnu-as
12542 Generate (or don't) code for the GNU assembler. This is the default.
12543 @c Also, this is the default if the configure option @option{--with-gnu-as}
12549 @opindex mno-gnu-ld
12550 Generate (or don't) code for the GNU linker. This is the default.
12551 @c Also, this is the default if the configure option @option{--with-gnu-ld}
12556 Generate code that does not use a global pointer register. The result
12557 is not position independent code, and violates the IA-64 ABI@.
12559 @item -mvolatile-asm-stop
12560 @itemx -mno-volatile-asm-stop
12561 @opindex mvolatile-asm-stop
12562 @opindex mno-volatile-asm-stop
12563 Generate (or don't) a stop bit immediately before and after volatile asm
12566 @item -mregister-names
12567 @itemx -mno-register-names
12568 @opindex mregister-names
12569 @opindex mno-register-names
12570 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
12571 the stacked registers. This may make assembler output more readable.
12577 Disable (or enable) optimizations that use the small data section. This may
12578 be useful for working around optimizer bugs.
12580 @item -mconstant-gp
12581 @opindex mconstant-gp
12582 Generate code that uses a single constant global pointer value. This is
12583 useful when compiling kernel code.
12587 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
12588 This is useful when compiling firmware code.
12590 @item -minline-float-divide-min-latency
12591 @opindex minline-float-divide-min-latency
12592 Generate code for inline divides of floating point values
12593 using the minimum latency algorithm.
12595 @item -minline-float-divide-max-throughput
12596 @opindex minline-float-divide-max-throughput
12597 Generate code for inline divides of floating point values
12598 using the maximum throughput algorithm.
12600 @item -mno-inline-float-divide
12601 @opindex mno-inline-float-divide
12602 Do not generate inline code for divides of floating point values.
12604 @item -minline-int-divide-min-latency
12605 @opindex minline-int-divide-min-latency
12606 Generate code for inline divides of integer values
12607 using the minimum latency algorithm.
12609 @item -minline-int-divide-max-throughput
12610 @opindex minline-int-divide-max-throughput
12611 Generate code for inline divides of integer values
12612 using the maximum throughput algorithm.
12614 @item -mno-inline-int-divide
12615 @opindex mno-inline-int-divide
12616 Do not generate inline code for divides of integer values.
12618 @item -minline-sqrt-min-latency
12619 @opindex minline-sqrt-min-latency
12620 Generate code for inline square roots
12621 using the minimum latency algorithm.
12623 @item -minline-sqrt-max-throughput
12624 @opindex minline-sqrt-max-throughput
12625 Generate code for inline square roots
12626 using the maximum throughput algorithm.
12628 @item -mno-inline-sqrt
12629 @opindex mno-inline-sqrt
12630 Do not generate inline code for sqrt.
12633 @itemx -mno-fused-madd
12634 @opindex mfused-madd
12635 @opindex mno-fused-madd
12636 Do (don't) generate code that uses the fused multiply/add or multiply/subtract
12637 instructions. The default is to use these instructions.
12639 @item -mno-dwarf2-asm
12640 @itemx -mdwarf2-asm
12641 @opindex mno-dwarf2-asm
12642 @opindex mdwarf2-asm
12643 Don't (or do) generate assembler code for the DWARF2 line number debugging
12644 info. This may be useful when not using the GNU assembler.
12646 @item -mearly-stop-bits
12647 @itemx -mno-early-stop-bits
12648 @opindex mearly-stop-bits
12649 @opindex mno-early-stop-bits
12650 Allow stop bits to be placed earlier than immediately preceding the
12651 instruction that triggered the stop bit. This can improve instruction
12652 scheduling, but does not always do so.
12654 @item -mfixed-range=@var{register-range}
12655 @opindex mfixed-range
12656 Generate code treating the given register range as fixed registers.
12657 A fixed register is one that the register allocator can not use. This is
12658 useful when compiling kernel code. A register range is specified as
12659 two registers separated by a dash. Multiple register ranges can be
12660 specified separated by a comma.
12662 @item -mtls-size=@var{tls-size}
12664 Specify bit size of immediate TLS offsets. Valid values are 14, 22, and
12667 @item -mtune=@var{cpu-type}
12669 Tune the instruction scheduling for a particular CPU, Valid values are
12670 itanium, itanium1, merced, itanium2, and mckinley.
12676 Generate code for a 32-bit or 64-bit environment.
12677 The 32-bit environment sets int, long and pointer to 32 bits.
12678 The 64-bit environment sets int to 32 bits and long and pointer
12679 to 64 bits. These are HP-UX specific flags.
12681 @item -mno-sched-br-data-spec
12682 @itemx -msched-br-data-spec
12683 @opindex mno-sched-br-data-spec
12684 @opindex msched-br-data-spec
12685 (Dis/En)able data speculative scheduling before reload.
12686 This will result in generation of the ld.a instructions and
12687 the corresponding check instructions (ld.c / chk.a).
12688 The default is 'disable'.
12690 @item -msched-ar-data-spec
12691 @itemx -mno-sched-ar-data-spec
12692 @opindex msched-ar-data-spec
12693 @opindex mno-sched-ar-data-spec
12694 (En/Dis)able data speculative scheduling after reload.
12695 This will result in generation of the ld.a instructions and
12696 the corresponding check instructions (ld.c / chk.a).
12697 The default is 'enable'.
12699 @item -mno-sched-control-spec
12700 @itemx -msched-control-spec
12701 @opindex mno-sched-control-spec
12702 @opindex msched-control-spec
12703 (Dis/En)able control speculative scheduling. This feature is
12704 available only during region scheduling (i.e.@: before reload).
12705 This will result in generation of the ld.s instructions and
12706 the corresponding check instructions chk.s .
12707 The default is 'disable'.
12709 @item -msched-br-in-data-spec
12710 @itemx -mno-sched-br-in-data-spec
12711 @opindex msched-br-in-data-spec
12712 @opindex mno-sched-br-in-data-spec
12713 (En/Dis)able speculative scheduling of the instructions that
12714 are dependent on the data speculative loads before reload.
12715 This is effective only with @option{-msched-br-data-spec} enabled.
12716 The default is 'enable'.
12718 @item -msched-ar-in-data-spec
12719 @itemx -mno-sched-ar-in-data-spec
12720 @opindex msched-ar-in-data-spec
12721 @opindex mno-sched-ar-in-data-spec
12722 (En/Dis)able speculative scheduling of the instructions that
12723 are dependent on the data speculative loads after reload.
12724 This is effective only with @option{-msched-ar-data-spec} enabled.
12725 The default is 'enable'.
12727 @item -msched-in-control-spec
12728 @itemx -mno-sched-in-control-spec
12729 @opindex msched-in-control-spec
12730 @opindex mno-sched-in-control-spec
12731 (En/Dis)able speculative scheduling of the instructions that
12732 are dependent on the control speculative loads.
12733 This is effective only with @option{-msched-control-spec} enabled.
12734 The default is 'enable'.
12736 @item -mno-sched-prefer-non-data-spec-insns
12737 @itemx -msched-prefer-non-data-spec-insns
12738 @opindex mno-sched-prefer-non-data-spec-insns
12739 @opindex msched-prefer-non-data-spec-insns
12740 If enabled, data speculative instructions will be chosen for schedule
12741 only if there are no other choices at the moment. This will make
12742 the use of the data speculation much more conservative.
12743 The default is 'disable'.
12745 @item -mno-sched-prefer-non-control-spec-insns
12746 @itemx -msched-prefer-non-control-spec-insns
12747 @opindex mno-sched-prefer-non-control-spec-insns
12748 @opindex msched-prefer-non-control-spec-insns
12749 If enabled, control speculative instructions will be chosen for schedule
12750 only if there are no other choices at the moment. This will make
12751 the use of the control speculation much more conservative.
12752 The default is 'disable'.
12754 @item -mno-sched-count-spec-in-critical-path
12755 @itemx -msched-count-spec-in-critical-path
12756 @opindex mno-sched-count-spec-in-critical-path
12757 @opindex msched-count-spec-in-critical-path
12758 If enabled, speculative dependencies will be considered during
12759 computation of the instructions priorities. This will make the use of the
12760 speculation a bit more conservative.
12761 The default is 'disable'.
12763 @item -msched-spec-ldc
12764 @opindex msched-spec-ldc
12765 Use a simple data speculation check. This option is on by default.
12767 @item -msched-control-spec-ldc
12768 @opindex msched-spec-ldc
12769 Use a simple check for control speculation. This option is on by default.
12771 @item -msched-stop-bits-after-every-cycle
12772 @opindex msched-stop-bits-after-every-cycle
12773 Place a stop bit after every cycle when scheduling. This option is on
12776 @item -msched-fp-mem-deps-zero-cost
12777 @opindex msched-fp-mem-deps-zero-cost
12778 Assume that floating-point stores and loads are not likely to cause a conflict
12779 when placed into the same instruction group. This option is disabled by
12782 @item -msel-sched-dont-check-control-spec
12783 @opindex msel-sched-dont-check-control-spec
12784 Generate checks for control speculation in selective scheduling.
12785 This flag is disabled by default.
12787 @item -msched-max-memory-insns=@var{max-insns}
12788 @opindex msched-max-memory-insns
12789 Limit on the number of memory insns per instruction group, giving lower
12790 priority to subsequent memory insns attempting to schedule in the same
12791 instruction group. Frequently useful to prevent cache bank conflicts.
12792 The default value is 1.
12794 @item -msched-max-memory-insns-hard-limit
12795 @opindex msched-max-memory-insns-hard-limit
12796 Disallow more than `msched-max-memory-insns' in instruction group.
12797 Otherwise, limit is `soft' meaning that we would prefer non-memory operations
12798 when limit is reached but may still schedule memory operations.
12802 @node IA-64/VMS Options
12803 @subsection IA-64/VMS Options
12805 These @samp{-m} options are defined for the IA-64/VMS implementations:
12808 @item -mvms-return-codes
12809 @opindex mvms-return-codes
12810 Return VMS condition codes from main. The default is to return POSIX
12811 style condition (e.g.@ error) codes.
12813 @item -mdebug-main=@var{prefix}
12814 @opindex mdebug-main=@var{prefix}
12815 Flag the first routine whose name starts with @var{prefix} as the main
12816 routine for the debugger.
12820 Default to 64bit memory allocation routines.
12824 @subsection LM32 Options
12825 @cindex LM32 options
12827 These @option{-m} options are defined for the Lattice Mico32 architecture:
12830 @item -mbarrel-shift-enabled
12831 @opindex mbarrel-shift-enabled
12832 Enable barrel-shift instructions.
12834 @item -mdivide-enabled
12835 @opindex mdivide-enabled
12836 Enable divide and modulus instructions.
12838 @item -mmultiply-enabled
12839 @opindex multiply-enabled
12840 Enable multiply instructions.
12842 @item -msign-extend-enabled
12843 @opindex msign-extend-enabled
12844 Enable sign extend instructions.
12846 @item -muser-enabled
12847 @opindex muser-enabled
12848 Enable user-defined instructions.
12853 @subsection M32C Options
12854 @cindex M32C options
12857 @item -mcpu=@var{name}
12859 Select the CPU for which code is generated. @var{name} may be one of
12860 @samp{r8c} for the R8C/Tiny series, @samp{m16c} for the M16C (up to
12861 /60) series, @samp{m32cm} for the M16C/80 series, or @samp{m32c} for
12862 the M32C/80 series.
12866 Specifies that the program will be run on the simulator. This causes
12867 an alternate runtime library to be linked in which supports, for
12868 example, file I/O@. You must not use this option when generating
12869 programs that will run on real hardware; you must provide your own
12870 runtime library for whatever I/O functions are needed.
12872 @item -memregs=@var{number}
12874 Specifies the number of memory-based pseudo-registers GCC will use
12875 during code generation. These pseudo-registers will be used like real
12876 registers, so there is a tradeoff between GCC's ability to fit the
12877 code into available registers, and the performance penalty of using
12878 memory instead of registers. Note that all modules in a program must
12879 be compiled with the same value for this option. Because of that, you
12880 must not use this option with the default runtime libraries gcc
12885 @node M32R/D Options
12886 @subsection M32R/D Options
12887 @cindex M32R/D options
12889 These @option{-m} options are defined for Renesas M32R/D architectures:
12894 Generate code for the M32R/2@.
12898 Generate code for the M32R/X@.
12902 Generate code for the M32R@. This is the default.
12904 @item -mmodel=small
12905 @opindex mmodel=small
12906 Assume all objects live in the lower 16MB of memory (so that their addresses
12907 can be loaded with the @code{ld24} instruction), and assume all subroutines
12908 are reachable with the @code{bl} instruction.
12909 This is the default.
12911 The addressability of a particular object can be set with the
12912 @code{model} attribute.
12914 @item -mmodel=medium
12915 @opindex mmodel=medium
12916 Assume objects may be anywhere in the 32-bit address space (the compiler
12917 will generate @code{seth/add3} instructions to load their addresses), and
12918 assume all subroutines are reachable with the @code{bl} instruction.
12920 @item -mmodel=large
12921 @opindex mmodel=large
12922 Assume objects may be anywhere in the 32-bit address space (the compiler
12923 will generate @code{seth/add3} instructions to load their addresses), and
12924 assume subroutines may not be reachable with the @code{bl} instruction
12925 (the compiler will generate the much slower @code{seth/add3/jl}
12926 instruction sequence).
12929 @opindex msdata=none
12930 Disable use of the small data area. Variables will be put into
12931 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
12932 @code{section} attribute has been specified).
12933 This is the default.
12935 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
12936 Objects may be explicitly put in the small data area with the
12937 @code{section} attribute using one of these sections.
12939 @item -msdata=sdata
12940 @opindex msdata=sdata
12941 Put small global and static data in the small data area, but do not
12942 generate special code to reference them.
12945 @opindex msdata=use
12946 Put small global and static data in the small data area, and generate
12947 special instructions to reference them.
12951 @cindex smaller data references
12952 Put global and static objects less than or equal to @var{num} bytes
12953 into the small data or bss sections instead of the normal data or bss
12954 sections. The default value of @var{num} is 8.
12955 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
12956 for this option to have any effect.
12958 All modules should be compiled with the same @option{-G @var{num}} value.
12959 Compiling with different values of @var{num} may or may not work; if it
12960 doesn't the linker will give an error message---incorrect code will not be
12965 Makes the M32R specific code in the compiler display some statistics
12966 that might help in debugging programs.
12968 @item -malign-loops
12969 @opindex malign-loops
12970 Align all loops to a 32-byte boundary.
12972 @item -mno-align-loops
12973 @opindex mno-align-loops
12974 Do not enforce a 32-byte alignment for loops. This is the default.
12976 @item -missue-rate=@var{number}
12977 @opindex missue-rate=@var{number}
12978 Issue @var{number} instructions per cycle. @var{number} can only be 1
12981 @item -mbranch-cost=@var{number}
12982 @opindex mbranch-cost=@var{number}
12983 @var{number} can only be 1 or 2. If it is 1 then branches will be
12984 preferred over conditional code, if it is 2, then the opposite will
12987 @item -mflush-trap=@var{number}
12988 @opindex mflush-trap=@var{number}
12989 Specifies the trap number to use to flush the cache. The default is
12990 12. Valid numbers are between 0 and 15 inclusive.
12992 @item -mno-flush-trap
12993 @opindex mno-flush-trap
12994 Specifies that the cache cannot be flushed by using a trap.
12996 @item -mflush-func=@var{name}
12997 @opindex mflush-func=@var{name}
12998 Specifies the name of the operating system function to call to flush
12999 the cache. The default is @emph{_flush_cache}, but a function call
13000 will only be used if a trap is not available.
13002 @item -mno-flush-func
13003 @opindex mno-flush-func
13004 Indicates that there is no OS function for flushing the cache.
13008 @node M680x0 Options
13009 @subsection M680x0 Options
13010 @cindex M680x0 options
13012 These are the @samp{-m} options defined for M680x0 and ColdFire processors.
13013 The default settings depend on which architecture was selected when
13014 the compiler was configured; the defaults for the most common choices
13018 @item -march=@var{arch}
13020 Generate code for a specific M680x0 or ColdFire instruction set
13021 architecture. Permissible values of @var{arch} for M680x0
13022 architectures are: @samp{68000}, @samp{68010}, @samp{68020},
13023 @samp{68030}, @samp{68040}, @samp{68060} and @samp{cpu32}. ColdFire
13024 architectures are selected according to Freescale's ISA classification
13025 and the permissible values are: @samp{isaa}, @samp{isaaplus},
13026 @samp{isab} and @samp{isac}.
13028 gcc defines a macro @samp{__mcf@var{arch}__} whenever it is generating
13029 code for a ColdFire target. The @var{arch} in this macro is one of the
13030 @option{-march} arguments given above.
13032 When used together, @option{-march} and @option{-mtune} select code
13033 that runs on a family of similar processors but that is optimized
13034 for a particular microarchitecture.
13036 @item -mcpu=@var{cpu}
13038 Generate code for a specific M680x0 or ColdFire processor.
13039 The M680x0 @var{cpu}s are: @samp{68000}, @samp{68010}, @samp{68020},
13040 @samp{68030}, @samp{68040}, @samp{68060}, @samp{68302}, @samp{68332}
13041 and @samp{cpu32}. The ColdFire @var{cpu}s are given by the table
13042 below, which also classifies the CPUs into families:
13044 @multitable @columnfractions 0.20 0.80
13045 @item @strong{Family} @tab @strong{@samp{-mcpu} arguments}
13046 @item @samp{51} @tab @samp{51} @samp{51ac} @samp{51cn} @samp{51em} @samp{51qe}
13047 @item @samp{5206} @tab @samp{5202} @samp{5204} @samp{5206}
13048 @item @samp{5206e} @tab @samp{5206e}
13049 @item @samp{5208} @tab @samp{5207} @samp{5208}
13050 @item @samp{5211a} @tab @samp{5210a} @samp{5211a}
13051 @item @samp{5213} @tab @samp{5211} @samp{5212} @samp{5213}
13052 @item @samp{5216} @tab @samp{5214} @samp{5216}
13053 @item @samp{52235} @tab @samp{52230} @samp{52231} @samp{52232} @samp{52233} @samp{52234} @samp{52235}
13054 @item @samp{5225} @tab @samp{5224} @samp{5225}
13055 @item @samp{52259} @tab @samp{52252} @samp{52254} @samp{52255} @samp{52256} @samp{52258} @samp{52259}
13056 @item @samp{5235} @tab @samp{5232} @samp{5233} @samp{5234} @samp{5235} @samp{523x}
13057 @item @samp{5249} @tab @samp{5249}
13058 @item @samp{5250} @tab @samp{5250}
13059 @item @samp{5271} @tab @samp{5270} @samp{5271}
13060 @item @samp{5272} @tab @samp{5272}
13061 @item @samp{5275} @tab @samp{5274} @samp{5275}
13062 @item @samp{5282} @tab @samp{5280} @samp{5281} @samp{5282} @samp{528x}
13063 @item @samp{53017} @tab @samp{53011} @samp{53012} @samp{53013} @samp{53014} @samp{53015} @samp{53016} @samp{53017}
13064 @item @samp{5307} @tab @samp{5307}
13065 @item @samp{5329} @tab @samp{5327} @samp{5328} @samp{5329} @samp{532x}
13066 @item @samp{5373} @tab @samp{5372} @samp{5373} @samp{537x}
13067 @item @samp{5407} @tab @samp{5407}
13068 @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}
13071 @option{-mcpu=@var{cpu}} overrides @option{-march=@var{arch}} if
13072 @var{arch} is compatible with @var{cpu}. Other combinations of
13073 @option{-mcpu} and @option{-march} are rejected.
13075 gcc defines the macro @samp{__mcf_cpu_@var{cpu}} when ColdFire target
13076 @var{cpu} is selected. It also defines @samp{__mcf_family_@var{family}},
13077 where the value of @var{family} is given by the table above.
13079 @item -mtune=@var{tune}
13081 Tune the code for a particular microarchitecture, within the
13082 constraints set by @option{-march} and @option{-mcpu}.
13083 The M680x0 microarchitectures are: @samp{68000}, @samp{68010},
13084 @samp{68020}, @samp{68030}, @samp{68040}, @samp{68060}
13085 and @samp{cpu32}. The ColdFire microarchitectures
13086 are: @samp{cfv1}, @samp{cfv2}, @samp{cfv3}, @samp{cfv4} and @samp{cfv4e}.
13088 You can also use @option{-mtune=68020-40} for code that needs
13089 to run relatively well on 68020, 68030 and 68040 targets.
13090 @option{-mtune=68020-60} is similar but includes 68060 targets
13091 as well. These two options select the same tuning decisions as
13092 @option{-m68020-40} and @option{-m68020-60} respectively.
13094 gcc defines the macros @samp{__mc@var{arch}} and @samp{__mc@var{arch}__}
13095 when tuning for 680x0 architecture @var{arch}. It also defines
13096 @samp{mc@var{arch}} unless either @option{-ansi} or a non-GNU @option{-std}
13097 option is used. If gcc is tuning for a range of architectures,
13098 as selected by @option{-mtune=68020-40} or @option{-mtune=68020-60},
13099 it defines the macros for every architecture in the range.
13101 gcc also defines the macro @samp{__m@var{uarch}__} when tuning for
13102 ColdFire microarchitecture @var{uarch}, where @var{uarch} is one
13103 of the arguments given above.
13109 Generate output for a 68000. This is the default
13110 when the compiler is configured for 68000-based systems.
13111 It is equivalent to @option{-march=68000}.
13113 Use this option for microcontrollers with a 68000 or EC000 core,
13114 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
13118 Generate output for a 68010. This is the default
13119 when the compiler is configured for 68010-based systems.
13120 It is equivalent to @option{-march=68010}.
13126 Generate output for a 68020. This is the default
13127 when the compiler is configured for 68020-based systems.
13128 It is equivalent to @option{-march=68020}.
13132 Generate output for a 68030. This is the default when the compiler is
13133 configured for 68030-based systems. It is equivalent to
13134 @option{-march=68030}.
13138 Generate output for a 68040. This is the default when the compiler is
13139 configured for 68040-based systems. It is equivalent to
13140 @option{-march=68040}.
13142 This option inhibits the use of 68881/68882 instructions that have to be
13143 emulated by software on the 68040. Use this option if your 68040 does not
13144 have code to emulate those instructions.
13148 Generate output for a 68060. This is the default when the compiler is
13149 configured for 68060-based systems. It is equivalent to
13150 @option{-march=68060}.
13152 This option inhibits the use of 68020 and 68881/68882 instructions that
13153 have to be emulated by software on the 68060. Use this option if your 68060
13154 does not have code to emulate those instructions.
13158 Generate output for a CPU32. This is the default
13159 when the compiler is configured for CPU32-based systems.
13160 It is equivalent to @option{-march=cpu32}.
13162 Use this option for microcontrollers with a
13163 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
13164 68336, 68340, 68341, 68349 and 68360.
13168 Generate output for a 520X ColdFire CPU@. This is the default
13169 when the compiler is configured for 520X-based systems.
13170 It is equivalent to @option{-mcpu=5206}, and is now deprecated
13171 in favor of that option.
13173 Use this option for microcontroller with a 5200 core, including
13174 the MCF5202, MCF5203, MCF5204 and MCF5206.
13178 Generate output for a 5206e ColdFire CPU@. The option is now
13179 deprecated in favor of the equivalent @option{-mcpu=5206e}.
13183 Generate output for a member of the ColdFire 528X family.
13184 The option is now deprecated in favor of the equivalent
13185 @option{-mcpu=528x}.
13189 Generate output for a ColdFire 5307 CPU@. The option is now deprecated
13190 in favor of the equivalent @option{-mcpu=5307}.
13194 Generate output for a ColdFire 5407 CPU@. The option is now deprecated
13195 in favor of the equivalent @option{-mcpu=5407}.
13199 Generate output for a ColdFire V4e family CPU (e.g.@: 547x/548x).
13200 This includes use of hardware floating point instructions.
13201 The option is equivalent to @option{-mcpu=547x}, and is now
13202 deprecated in favor of that option.
13206 Generate output for a 68040, without using any of the new instructions.
13207 This results in code which can run relatively efficiently on either a
13208 68020/68881 or a 68030 or a 68040. The generated code does use the
13209 68881 instructions that are emulated on the 68040.
13211 The option is equivalent to @option{-march=68020} @option{-mtune=68020-40}.
13215 Generate output for a 68060, without using any of the new instructions.
13216 This results in code which can run relatively efficiently on either a
13217 68020/68881 or a 68030 or a 68040. The generated code does use the
13218 68881 instructions that are emulated on the 68060.
13220 The option is equivalent to @option{-march=68020} @option{-mtune=68020-60}.
13224 @opindex mhard-float
13226 Generate floating-point instructions. This is the default for 68020
13227 and above, and for ColdFire devices that have an FPU@. It defines the
13228 macro @samp{__HAVE_68881__} on M680x0 targets and @samp{__mcffpu__}
13229 on ColdFire targets.
13232 @opindex msoft-float
13233 Do not generate floating-point instructions; use library calls instead.
13234 This is the default for 68000, 68010, and 68832 targets. It is also
13235 the default for ColdFire devices that have no FPU.
13241 Generate (do not generate) ColdFire hardware divide and remainder
13242 instructions. If @option{-march} is used without @option{-mcpu},
13243 the default is ``on'' for ColdFire architectures and ``off'' for M680x0
13244 architectures. Otherwise, the default is taken from the target CPU
13245 (either the default CPU, or the one specified by @option{-mcpu}). For
13246 example, the default is ``off'' for @option{-mcpu=5206} and ``on'' for
13247 @option{-mcpu=5206e}.
13249 gcc defines the macro @samp{__mcfhwdiv__} when this option is enabled.
13253 Consider type @code{int} to be 16 bits wide, like @code{short int}.
13254 Additionally, parameters passed on the stack are also aligned to a
13255 16-bit boundary even on targets whose API mandates promotion to 32-bit.
13259 Do not consider type @code{int} to be 16 bits wide. This is the default.
13262 @itemx -mno-bitfield
13263 @opindex mnobitfield
13264 @opindex mno-bitfield
13265 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
13266 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
13270 Do use the bit-field instructions. The @option{-m68020} option implies
13271 @option{-mbitfield}. This is the default if you use a configuration
13272 designed for a 68020.
13276 Use a different function-calling convention, in which functions
13277 that take a fixed number of arguments return with the @code{rtd}
13278 instruction, which pops their arguments while returning. This
13279 saves one instruction in the caller since there is no need to pop
13280 the arguments there.
13282 This calling convention is incompatible with the one normally
13283 used on Unix, so you cannot use it if you need to call libraries
13284 compiled with the Unix compiler.
13286 Also, you must provide function prototypes for all functions that
13287 take variable numbers of arguments (including @code{printf});
13288 otherwise incorrect code will be generated for calls to those
13291 In addition, seriously incorrect code will result if you call a
13292 function with too many arguments. (Normally, extra arguments are
13293 harmlessly ignored.)
13295 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
13296 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
13300 Do not use the calling conventions selected by @option{-mrtd}.
13301 This is the default.
13304 @itemx -mno-align-int
13305 @opindex malign-int
13306 @opindex mno-align-int
13307 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
13308 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
13309 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
13310 Aligning variables on 32-bit boundaries produces code that runs somewhat
13311 faster on processors with 32-bit busses at the expense of more memory.
13313 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
13314 align structures containing the above types differently than
13315 most published application binary interface specifications for the m68k.
13319 Use the pc-relative addressing mode of the 68000 directly, instead of
13320 using a global offset table. At present, this option implies @option{-fpic},
13321 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
13322 not presently supported with @option{-mpcrel}, though this could be supported for
13323 68020 and higher processors.
13325 @item -mno-strict-align
13326 @itemx -mstrict-align
13327 @opindex mno-strict-align
13328 @opindex mstrict-align
13329 Do not (do) assume that unaligned memory references will be handled by
13333 Generate code that allows the data segment to be located in a different
13334 area of memory from the text segment. This allows for execute in place in
13335 an environment without virtual memory management. This option implies
13338 @item -mno-sep-data
13339 Generate code that assumes that the data segment follows the text segment.
13340 This is the default.
13342 @item -mid-shared-library
13343 Generate code that supports shared libraries via the library ID method.
13344 This allows for execute in place and shared libraries in an environment
13345 without virtual memory management. This option implies @option{-fPIC}.
13347 @item -mno-id-shared-library
13348 Generate code that doesn't assume ID based shared libraries are being used.
13349 This is the default.
13351 @item -mshared-library-id=n
13352 Specified the identification number of the ID based shared library being
13353 compiled. Specifying a value of 0 will generate more compact code, specifying
13354 other values will force the allocation of that number to the current
13355 library but is no more space or time efficient than omitting this option.
13361 When generating position-independent code for ColdFire, generate code
13362 that works if the GOT has more than 8192 entries. This code is
13363 larger and slower than code generated without this option. On M680x0
13364 processors, this option is not needed; @option{-fPIC} suffices.
13366 GCC normally uses a single instruction to load values from the GOT@.
13367 While this is relatively efficient, it only works if the GOT
13368 is smaller than about 64k. Anything larger causes the linker
13369 to report an error such as:
13371 @cindex relocation truncated to fit (ColdFire)
13373 relocation truncated to fit: R_68K_GOT16O foobar
13376 If this happens, you should recompile your code with @option{-mxgot}.
13377 It should then work with very large GOTs. However, code generated with
13378 @option{-mxgot} is less efficient, since it takes 4 instructions to fetch
13379 the value of a global symbol.
13381 Note that some linkers, including newer versions of the GNU linker,
13382 can create multiple GOTs and sort GOT entries. If you have such a linker,
13383 you should only need to use @option{-mxgot} when compiling a single
13384 object file that accesses more than 8192 GOT entries. Very few do.
13386 These options have no effect unless GCC is generating
13387 position-independent code.
13391 @node M68hc1x Options
13392 @subsection M68hc1x Options
13393 @cindex M68hc1x options
13395 These are the @samp{-m} options defined for the 68hc11 and 68hc12
13396 microcontrollers. The default values for these options depends on
13397 which style of microcontroller was selected when the compiler was configured;
13398 the defaults for the most common choices are given below.
13405 Generate output for a 68HC11. This is the default
13406 when the compiler is configured for 68HC11-based systems.
13412 Generate output for a 68HC12. This is the default
13413 when the compiler is configured for 68HC12-based systems.
13419 Generate output for a 68HCS12.
13421 @item -mauto-incdec
13422 @opindex mauto-incdec
13423 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
13430 Enable the use of 68HC12 min and max instructions.
13433 @itemx -mno-long-calls
13434 @opindex mlong-calls
13435 @opindex mno-long-calls
13436 Treat all calls as being far away (near). If calls are assumed to be
13437 far away, the compiler will use the @code{call} instruction to
13438 call a function and the @code{rtc} instruction for returning.
13442 Consider type @code{int} to be 16 bits wide, like @code{short int}.
13444 @item -msoft-reg-count=@var{count}
13445 @opindex msoft-reg-count
13446 Specify the number of pseudo-soft registers which are used for the
13447 code generation. The maximum number is 32. Using more pseudo-soft
13448 register may or may not result in better code depending on the program.
13449 The default is 4 for 68HC11 and 2 for 68HC12.
13453 @node MCore Options
13454 @subsection MCore Options
13455 @cindex MCore options
13457 These are the @samp{-m} options defined for the Motorola M*Core
13463 @itemx -mno-hardlit
13465 @opindex mno-hardlit
13466 Inline constants into the code stream if it can be done in two
13467 instructions or less.
13473 Use the divide instruction. (Enabled by default).
13475 @item -mrelax-immediate
13476 @itemx -mno-relax-immediate
13477 @opindex mrelax-immediate
13478 @opindex mno-relax-immediate
13479 Allow arbitrary sized immediates in bit operations.
13481 @item -mwide-bitfields
13482 @itemx -mno-wide-bitfields
13483 @opindex mwide-bitfields
13484 @opindex mno-wide-bitfields
13485 Always treat bit-fields as int-sized.
13487 @item -m4byte-functions
13488 @itemx -mno-4byte-functions
13489 @opindex m4byte-functions
13490 @opindex mno-4byte-functions
13491 Force all functions to be aligned to a four byte boundary.
13493 @item -mcallgraph-data
13494 @itemx -mno-callgraph-data
13495 @opindex mcallgraph-data
13496 @opindex mno-callgraph-data
13497 Emit callgraph information.
13500 @itemx -mno-slow-bytes
13501 @opindex mslow-bytes
13502 @opindex mno-slow-bytes
13503 Prefer word access when reading byte quantities.
13505 @item -mlittle-endian
13506 @itemx -mbig-endian
13507 @opindex mlittle-endian
13508 @opindex mbig-endian
13509 Generate code for a little endian target.
13515 Generate code for the 210 processor.
13519 Assume that run-time support has been provided and so omit the
13520 simulator library (@file{libsim.a)} from the linker command line.
13522 @item -mstack-increment=@var{size}
13523 @opindex mstack-increment
13524 Set the maximum amount for a single stack increment operation. Large
13525 values can increase the speed of programs which contain functions
13526 that need a large amount of stack space, but they can also trigger a
13527 segmentation fault if the stack is extended too much. The default
13533 @subsection MeP Options
13534 @cindex MeP options
13540 Enables the @code{abs} instruction, which is the absolute difference
13541 between two registers.
13545 Enables all the optional instructions - average, multiply, divide, bit
13546 operations, leading zero, absolute difference, min/max, clip, and
13552 Enables the @code{ave} instruction, which computes the average of two
13555 @item -mbased=@var{n}
13557 Variables of size @var{n} bytes or smaller will be placed in the
13558 @code{.based} section by default. Based variables use the @code{$tp}
13559 register as a base register, and there is a 128 byte limit to the
13560 @code{.based} section.
13564 Enables the bit operation instructions - bit test (@code{btstm}), set
13565 (@code{bsetm}), clear (@code{bclrm}), invert (@code{bnotm}), and
13566 test-and-set (@code{tas}).
13568 @item -mc=@var{name}
13570 Selects which section constant data will be placed in. @var{name} may
13571 be @code{tiny}, @code{near}, or @code{far}.
13575 Enables the @code{clip} instruction. Note that @code{-mclip} is not
13576 useful unless you also provide @code{-mminmax}.
13578 @item -mconfig=@var{name}
13580 Selects one of the build-in core configurations. Each MeP chip has
13581 one or more modules in it; each module has a core CPU and a variety of
13582 coprocessors, optional instructions, and peripherals. The
13583 @code{MeP-Integrator} tool, not part of GCC, provides these
13584 configurations through this option; using this option is the same as
13585 using all the corresponding command line options. The default
13586 configuration is @code{default}.
13590 Enables the coprocessor instructions. By default, this is a 32-bit
13591 coprocessor. Note that the coprocessor is normally enabled via the
13592 @code{-mconfig=} option.
13596 Enables the 32-bit coprocessor's instructions.
13600 Enables the 64-bit coprocessor's instructions.
13604 Enables IVC2 scheduling. IVC2 is a 64-bit VLIW coprocessor.
13608 Causes constant variables to be placed in the @code{.near} section.
13612 Enables the @code{div} and @code{divu} instructions.
13616 Generate big-endian code.
13620 Generate little-endian code.
13622 @item -mio-volatile
13623 @opindex mio-volatile
13624 Tells the compiler that any variable marked with the @code{io}
13625 attribute is to be considered volatile.
13629 Causes variables to be assigned to the @code{.far} section by default.
13633 Enables the @code{leadz} (leading zero) instruction.
13637 Causes variables to be assigned to the @code{.near} section by default.
13641 Enables the @code{min} and @code{max} instructions.
13645 Enables the multiplication and multiply-accumulate instructions.
13649 Disables all the optional instructions enabled by @code{-mall-opts}.
13653 Enables the @code{repeat} and @code{erepeat} instructions, used for
13654 low-overhead looping.
13658 Causes all variables to default to the @code{.tiny} section. Note
13659 that there is a 65536 byte limit to this section. Accesses to these
13660 variables use the @code{%gp} base register.
13664 Enables the saturation instructions. Note that the compiler does not
13665 currently generate these itself, but this option is included for
13666 compatibility with other tools, like @code{as}.
13670 Link the SDRAM-based runtime instead of the default ROM-based runtime.
13674 Link the simulator runtime libraries.
13678 Link the simulator runtime libraries, excluding built-in support
13679 for reset and exception vectors and tables.
13683 Causes all functions to default to the @code{.far} section. Without
13684 this option, functions default to the @code{.near} section.
13686 @item -mtiny=@var{n}
13688 Variables that are @var{n} bytes or smaller will be allocated to the
13689 @code{.tiny} section. These variables use the @code{$gp} base
13690 register. The default for this option is 4, but note that there's a
13691 65536 byte limit to the @code{.tiny} section.
13696 @subsection MIPS Options
13697 @cindex MIPS options
13703 Generate big-endian code.
13707 Generate little-endian code. This is the default for @samp{mips*el-*-*}
13710 @item -march=@var{arch}
13712 Generate code that will run on @var{arch}, which can be the name of a
13713 generic MIPS ISA, or the name of a particular processor.
13715 @samp{mips1}, @samp{mips2}, @samp{mips3}, @samp{mips4},
13716 @samp{mips32}, @samp{mips32r2}, @samp{mips64} and @samp{mips64r2}.
13717 The processor names are:
13718 @samp{4kc}, @samp{4km}, @samp{4kp}, @samp{4ksc},
13719 @samp{4kec}, @samp{4kem}, @samp{4kep}, @samp{4ksd},
13720 @samp{5kc}, @samp{5kf},
13722 @samp{24kc}, @samp{24kf2_1}, @samp{24kf1_1},
13723 @samp{24kec}, @samp{24kef2_1}, @samp{24kef1_1},
13724 @samp{34kc}, @samp{34kf2_1}, @samp{34kf1_1},
13725 @samp{74kc}, @samp{74kf2_1}, @samp{74kf1_1}, @samp{74kf3_2},
13726 @samp{1004kc}, @samp{1004kf2_1}, @samp{1004kf1_1},
13727 @samp{loongson2e}, @samp{loongson2f},
13731 @samp{r2000}, @samp{r3000}, @samp{r3900}, @samp{r4000}, @samp{r4400},
13732 @samp{r4600}, @samp{r4650}, @samp{r6000}, @samp{r8000},
13733 @samp{rm7000}, @samp{rm9000},
13734 @samp{r10000}, @samp{r12000}, @samp{r14000}, @samp{r16000},
13737 @samp{vr4100}, @samp{vr4111}, @samp{vr4120}, @samp{vr4130}, @samp{vr4300},
13738 @samp{vr5000}, @samp{vr5400}, @samp{vr5500}
13740 The special value @samp{from-abi} selects the
13741 most compatible architecture for the selected ABI (that is,
13742 @samp{mips1} for 32-bit ABIs and @samp{mips3} for 64-bit ABIs)@.
13744 Native Linux/GNU toolchains also support the value @samp{native},
13745 which selects the best architecture option for the host processor.
13746 @option{-march=native} has no effect if GCC does not recognize
13749 In processor names, a final @samp{000} can be abbreviated as @samp{k}
13750 (for example, @samp{-march=r2k}). Prefixes are optional, and
13751 @samp{vr} may be written @samp{r}.
13753 Names of the form @samp{@var{n}f2_1} refer to processors with
13754 FPUs clocked at half the rate of the core, names of the form
13755 @samp{@var{n}f1_1} refer to processors with FPUs clocked at the same
13756 rate as the core, and names of the form @samp{@var{n}f3_2} refer to
13757 processors with FPUs clocked a ratio of 3:2 with respect to the core.
13758 For compatibility reasons, @samp{@var{n}f} is accepted as a synonym
13759 for @samp{@var{n}f2_1} while @samp{@var{n}x} and @samp{@var{b}fx} are
13760 accepted as synonyms for @samp{@var{n}f1_1}.
13762 GCC defines two macros based on the value of this option. The first
13763 is @samp{_MIPS_ARCH}, which gives the name of target architecture, as
13764 a string. The second has the form @samp{_MIPS_ARCH_@var{foo}},
13765 where @var{foo} is the capitalized value of @samp{_MIPS_ARCH}@.
13766 For example, @samp{-march=r2000} will set @samp{_MIPS_ARCH}
13767 to @samp{"r2000"} and define the macro @samp{_MIPS_ARCH_R2000}.
13769 Note that the @samp{_MIPS_ARCH} macro uses the processor names given
13770 above. In other words, it will have the full prefix and will not
13771 abbreviate @samp{000} as @samp{k}. In the case of @samp{from-abi},
13772 the macro names the resolved architecture (either @samp{"mips1"} or
13773 @samp{"mips3"}). It names the default architecture when no
13774 @option{-march} option is given.
13776 @item -mtune=@var{arch}
13778 Optimize for @var{arch}. Among other things, this option controls
13779 the way instructions are scheduled, and the perceived cost of arithmetic
13780 operations. The list of @var{arch} values is the same as for
13783 When this option is not used, GCC will optimize for the processor
13784 specified by @option{-march}. By using @option{-march} and
13785 @option{-mtune} together, it is possible to generate code that will
13786 run on a family of processors, but optimize the code for one
13787 particular member of that family.
13789 @samp{-mtune} defines the macros @samp{_MIPS_TUNE} and
13790 @samp{_MIPS_TUNE_@var{foo}}, which work in the same way as the
13791 @samp{-march} ones described above.
13795 Equivalent to @samp{-march=mips1}.
13799 Equivalent to @samp{-march=mips2}.
13803 Equivalent to @samp{-march=mips3}.
13807 Equivalent to @samp{-march=mips4}.
13811 Equivalent to @samp{-march=mips32}.
13815 Equivalent to @samp{-march=mips32r2}.
13819 Equivalent to @samp{-march=mips64}.
13823 Equivalent to @samp{-march=mips64r2}.
13828 @opindex mno-mips16
13829 Generate (do not generate) MIPS16 code. If GCC is targetting a
13830 MIPS32 or MIPS64 architecture, it will make use of the MIPS16e ASE@.
13832 MIPS16 code generation can also be controlled on a per-function basis
13833 by means of @code{mips16} and @code{nomips16} attributes.
13834 @xref{Function Attributes}, for more information.
13836 @item -mflip-mips16
13837 @opindex mflip-mips16
13838 Generate MIPS16 code on alternating functions. This option is provided
13839 for regression testing of mixed MIPS16/non-MIPS16 code generation, and is
13840 not intended for ordinary use in compiling user code.
13842 @item -minterlink-mips16
13843 @itemx -mno-interlink-mips16
13844 @opindex minterlink-mips16
13845 @opindex mno-interlink-mips16
13846 Require (do not require) that non-MIPS16 code be link-compatible with
13849 For example, non-MIPS16 code cannot jump directly to MIPS16 code;
13850 it must either use a call or an indirect jump. @option{-minterlink-mips16}
13851 therefore disables direct jumps unless GCC knows that the target of the
13852 jump is not MIPS16.
13864 Generate code for the given ABI@.
13866 Note that the EABI has a 32-bit and a 64-bit variant. GCC normally
13867 generates 64-bit code when you select a 64-bit architecture, but you
13868 can use @option{-mgp32} to get 32-bit code instead.
13870 For information about the O64 ABI, see
13871 @w{@uref{http://gcc.gnu.org/projects/mipso64-abi.html}}.
13873 GCC supports a variant of the o32 ABI in which floating-point registers
13874 are 64 rather than 32 bits wide. You can select this combination with
13875 @option{-mabi=32} @option{-mfp64}. This ABI relies on the @samp{mthc1}
13876 and @samp{mfhc1} instructions and is therefore only supported for
13877 MIPS32R2 processors.
13879 The register assignments for arguments and return values remain the
13880 same, but each scalar value is passed in a single 64-bit register
13881 rather than a pair of 32-bit registers. For example, scalar
13882 floating-point values are returned in @samp{$f0} only, not a
13883 @samp{$f0}/@samp{$f1} pair. The set of call-saved registers also
13884 remains the same, but all 64 bits are saved.
13887 @itemx -mno-abicalls
13889 @opindex mno-abicalls
13890 Generate (do not generate) code that is suitable for SVR4-style
13891 dynamic objects. @option{-mabicalls} is the default for SVR4-based
13896 Generate (do not generate) code that is fully position-independent,
13897 and that can therefore be linked into shared libraries. This option
13898 only affects @option{-mabicalls}.
13900 All @option{-mabicalls} code has traditionally been position-independent,
13901 regardless of options like @option{-fPIC} and @option{-fpic}. However,
13902 as an extension, the GNU toolchain allows executables to use absolute
13903 accesses for locally-binding symbols. It can also use shorter GP
13904 initialization sequences and generate direct calls to locally-defined
13905 functions. This mode is selected by @option{-mno-shared}.
13907 @option{-mno-shared} depends on binutils 2.16 or higher and generates
13908 objects that can only be linked by the GNU linker. However, the option
13909 does not affect the ABI of the final executable; it only affects the ABI
13910 of relocatable objects. Using @option{-mno-shared} will generally make
13911 executables both smaller and quicker.
13913 @option{-mshared} is the default.
13919 Assume (do not assume) that the static and dynamic linkers
13920 support PLTs and copy relocations. This option only affects
13921 @samp{-mno-shared -mabicalls}. For the n64 ABI, this option
13922 has no effect without @samp{-msym32}.
13924 You can make @option{-mplt} the default by configuring
13925 GCC with @option{--with-mips-plt}. The default is
13926 @option{-mno-plt} otherwise.
13932 Lift (do not lift) the usual restrictions on the size of the global
13935 GCC normally uses a single instruction to load values from the GOT@.
13936 While this is relatively efficient, it will only work if the GOT
13937 is smaller than about 64k. Anything larger will cause the linker
13938 to report an error such as:
13940 @cindex relocation truncated to fit (MIPS)
13942 relocation truncated to fit: R_MIPS_GOT16 foobar
13945 If this happens, you should recompile your code with @option{-mxgot}.
13946 It should then work with very large GOTs, although it will also be
13947 less efficient, since it will take three instructions to fetch the
13948 value of a global symbol.
13950 Note that some linkers can create multiple GOTs. If you have such a
13951 linker, you should only need to use @option{-mxgot} when a single object
13952 file accesses more than 64k's worth of GOT entries. Very few do.
13954 These options have no effect unless GCC is generating position
13959 Assume that general-purpose registers are 32 bits wide.
13963 Assume that general-purpose registers are 64 bits wide.
13967 Assume that floating-point registers are 32 bits wide.
13971 Assume that floating-point registers are 64 bits wide.
13974 @opindex mhard-float
13975 Use floating-point coprocessor instructions.
13978 @opindex msoft-float
13979 Do not use floating-point coprocessor instructions. Implement
13980 floating-point calculations using library calls instead.
13982 @item -msingle-float
13983 @opindex msingle-float
13984 Assume that the floating-point coprocessor only supports single-precision
13987 @item -mdouble-float
13988 @opindex mdouble-float
13989 Assume that the floating-point coprocessor supports double-precision
13990 operations. This is the default.
13996 Use (do not use) @samp{ll}, @samp{sc}, and @samp{sync} instructions to
13997 implement atomic memory built-in functions. When neither option is
13998 specified, GCC will use the instructions if the target architecture
14001 @option{-mllsc} is useful if the runtime environment can emulate the
14002 instructions and @option{-mno-llsc} can be useful when compiling for
14003 nonstandard ISAs. You can make either option the default by
14004 configuring GCC with @option{--with-llsc} and @option{--without-llsc}
14005 respectively. @option{--with-llsc} is the default for some
14006 configurations; see the installation documentation for details.
14012 Use (do not use) revision 1 of the MIPS DSP ASE@.
14013 @xref{MIPS DSP Built-in Functions}. This option defines the
14014 preprocessor macro @samp{__mips_dsp}. It also defines
14015 @samp{__mips_dsp_rev} to 1.
14021 Use (do not use) revision 2 of the MIPS DSP ASE@.
14022 @xref{MIPS DSP Built-in Functions}. This option defines the
14023 preprocessor macros @samp{__mips_dsp} and @samp{__mips_dspr2}.
14024 It also defines @samp{__mips_dsp_rev} to 2.
14027 @itemx -mno-smartmips
14028 @opindex msmartmips
14029 @opindex mno-smartmips
14030 Use (do not use) the MIPS SmartMIPS ASE.
14032 @item -mpaired-single
14033 @itemx -mno-paired-single
14034 @opindex mpaired-single
14035 @opindex mno-paired-single
14036 Use (do not use) paired-single floating-point instructions.
14037 @xref{MIPS Paired-Single Support}. This option requires
14038 hardware floating-point support to be enabled.
14044 Use (do not use) MIPS Digital Media Extension instructions.
14045 This option can only be used when generating 64-bit code and requires
14046 hardware floating-point support to be enabled.
14051 @opindex mno-mips3d
14052 Use (do not use) the MIPS-3D ASE@. @xref{MIPS-3D Built-in Functions}.
14053 The option @option{-mips3d} implies @option{-mpaired-single}.
14059 Use (do not use) MT Multithreading instructions.
14063 Force @code{long} types to be 64 bits wide. See @option{-mlong32} for
14064 an explanation of the default and the way that the pointer size is
14069 Force @code{long}, @code{int}, and pointer types to be 32 bits wide.
14071 The default size of @code{int}s, @code{long}s and pointers depends on
14072 the ABI@. All the supported ABIs use 32-bit @code{int}s. The n64 ABI
14073 uses 64-bit @code{long}s, as does the 64-bit EABI; the others use
14074 32-bit @code{long}s. Pointers are the same size as @code{long}s,
14075 or the same size as integer registers, whichever is smaller.
14081 Assume (do not assume) that all symbols have 32-bit values, regardless
14082 of the selected ABI@. This option is useful in combination with
14083 @option{-mabi=64} and @option{-mno-abicalls} because it allows GCC
14084 to generate shorter and faster references to symbolic addresses.
14088 Put definitions of externally-visible data in a small data section
14089 if that data is no bigger than @var{num} bytes. GCC can then access
14090 the data more efficiently; see @option{-mgpopt} for details.
14092 The default @option{-G} option depends on the configuration.
14094 @item -mlocal-sdata
14095 @itemx -mno-local-sdata
14096 @opindex mlocal-sdata
14097 @opindex mno-local-sdata
14098 Extend (do not extend) the @option{-G} behavior to local data too,
14099 such as to static variables in C@. @option{-mlocal-sdata} is the
14100 default for all configurations.
14102 If the linker complains that an application is using too much small data,
14103 you might want to try rebuilding the less performance-critical parts with
14104 @option{-mno-local-sdata}. You might also want to build large
14105 libraries with @option{-mno-local-sdata}, so that the libraries leave
14106 more room for the main program.
14108 @item -mextern-sdata
14109 @itemx -mno-extern-sdata
14110 @opindex mextern-sdata
14111 @opindex mno-extern-sdata
14112 Assume (do not assume) that externally-defined data will be in
14113 a small data section if that data is within the @option{-G} limit.
14114 @option{-mextern-sdata} is the default for all configurations.
14116 If you compile a module @var{Mod} with @option{-mextern-sdata} @option{-G
14117 @var{num}} @option{-mgpopt}, and @var{Mod} references a variable @var{Var}
14118 that is no bigger than @var{num} bytes, you must make sure that @var{Var}
14119 is placed in a small data section. If @var{Var} is defined by another
14120 module, you must either compile that module with a high-enough
14121 @option{-G} setting or attach a @code{section} attribute to @var{Var}'s
14122 definition. If @var{Var} is common, you must link the application
14123 with a high-enough @option{-G} setting.
14125 The easiest way of satisfying these restrictions is to compile
14126 and link every module with the same @option{-G} option. However,
14127 you may wish to build a library that supports several different
14128 small data limits. You can do this by compiling the library with
14129 the highest supported @option{-G} setting and additionally using
14130 @option{-mno-extern-sdata} to stop the library from making assumptions
14131 about externally-defined data.
14137 Use (do not use) GP-relative accesses for symbols that are known to be
14138 in a small data section; see @option{-G}, @option{-mlocal-sdata} and
14139 @option{-mextern-sdata}. @option{-mgpopt} is the default for all
14142 @option{-mno-gpopt} is useful for cases where the @code{$gp} register
14143 might not hold the value of @code{_gp}. For example, if the code is
14144 part of a library that might be used in a boot monitor, programs that
14145 call boot monitor routines will pass an unknown value in @code{$gp}.
14146 (In such situations, the boot monitor itself would usually be compiled
14147 with @option{-G0}.)
14149 @option{-mno-gpopt} implies @option{-mno-local-sdata} and
14150 @option{-mno-extern-sdata}.
14152 @item -membedded-data
14153 @itemx -mno-embedded-data
14154 @opindex membedded-data
14155 @opindex mno-embedded-data
14156 Allocate variables to the read-only data section first if possible, then
14157 next in the small data section if possible, otherwise in data. This gives
14158 slightly slower code than the default, but reduces the amount of RAM required
14159 when executing, and thus may be preferred for some embedded systems.
14161 @item -muninit-const-in-rodata
14162 @itemx -mno-uninit-const-in-rodata
14163 @opindex muninit-const-in-rodata
14164 @opindex mno-uninit-const-in-rodata
14165 Put uninitialized @code{const} variables in the read-only data section.
14166 This option is only meaningful in conjunction with @option{-membedded-data}.
14168 @item -mcode-readable=@var{setting}
14169 @opindex mcode-readable
14170 Specify whether GCC may generate code that reads from executable sections.
14171 There are three possible settings:
14174 @item -mcode-readable=yes
14175 Instructions may freely access executable sections. This is the
14178 @item -mcode-readable=pcrel
14179 MIPS16 PC-relative load instructions can access executable sections,
14180 but other instructions must not do so. This option is useful on 4KSc
14181 and 4KSd processors when the code TLBs have the Read Inhibit bit set.
14182 It is also useful on processors that can be configured to have a dual
14183 instruction/data SRAM interface and that, like the M4K, automatically
14184 redirect PC-relative loads to the instruction RAM.
14186 @item -mcode-readable=no
14187 Instructions must not access executable sections. This option can be
14188 useful on targets that are configured to have a dual instruction/data
14189 SRAM interface but that (unlike the M4K) do not automatically redirect
14190 PC-relative loads to the instruction RAM.
14193 @item -msplit-addresses
14194 @itemx -mno-split-addresses
14195 @opindex msplit-addresses
14196 @opindex mno-split-addresses
14197 Enable (disable) use of the @code{%hi()} and @code{%lo()} assembler
14198 relocation operators. This option has been superseded by
14199 @option{-mexplicit-relocs} but is retained for backwards compatibility.
14201 @item -mexplicit-relocs
14202 @itemx -mno-explicit-relocs
14203 @opindex mexplicit-relocs
14204 @opindex mno-explicit-relocs
14205 Use (do not use) assembler relocation operators when dealing with symbolic
14206 addresses. The alternative, selected by @option{-mno-explicit-relocs},
14207 is to use assembler macros instead.
14209 @option{-mexplicit-relocs} is the default if GCC was configured
14210 to use an assembler that supports relocation operators.
14212 @item -mcheck-zero-division
14213 @itemx -mno-check-zero-division
14214 @opindex mcheck-zero-division
14215 @opindex mno-check-zero-division
14216 Trap (do not trap) on integer division by zero.
14218 The default is @option{-mcheck-zero-division}.
14220 @item -mdivide-traps
14221 @itemx -mdivide-breaks
14222 @opindex mdivide-traps
14223 @opindex mdivide-breaks
14224 MIPS systems check for division by zero by generating either a
14225 conditional trap or a break instruction. Using traps results in
14226 smaller code, but is only supported on MIPS II and later. Also, some
14227 versions of the Linux kernel have a bug that prevents trap from
14228 generating the proper signal (@code{SIGFPE}). Use @option{-mdivide-traps} to
14229 allow conditional traps on architectures that support them and
14230 @option{-mdivide-breaks} to force the use of breaks.
14232 The default is usually @option{-mdivide-traps}, but this can be
14233 overridden at configure time using @option{--with-divide=breaks}.
14234 Divide-by-zero checks can be completely disabled using
14235 @option{-mno-check-zero-division}.
14240 @opindex mno-memcpy
14241 Force (do not force) the use of @code{memcpy()} for non-trivial block
14242 moves. The default is @option{-mno-memcpy}, which allows GCC to inline
14243 most constant-sized copies.
14246 @itemx -mno-long-calls
14247 @opindex mlong-calls
14248 @opindex mno-long-calls
14249 Disable (do not disable) use of the @code{jal} instruction. Calling
14250 functions using @code{jal} is more efficient but requires the caller
14251 and callee to be in the same 256 megabyte segment.
14253 This option has no effect on abicalls code. The default is
14254 @option{-mno-long-calls}.
14260 Enable (disable) use of the @code{mad}, @code{madu} and @code{mul}
14261 instructions, as provided by the R4650 ISA@.
14264 @itemx -mno-fused-madd
14265 @opindex mfused-madd
14266 @opindex mno-fused-madd
14267 Enable (disable) use of the floating point multiply-accumulate
14268 instructions, when they are available. The default is
14269 @option{-mfused-madd}.
14271 When multiply-accumulate instructions are used, the intermediate
14272 product is calculated to infinite precision and is not subject to
14273 the FCSR Flush to Zero bit. This may be undesirable in some
14278 Tell the MIPS assembler to not run its preprocessor over user
14279 assembler files (with a @samp{.s} suffix) when assembling them.
14282 @itemx -mno-fix-r4000
14283 @opindex mfix-r4000
14284 @opindex mno-fix-r4000
14285 Work around certain R4000 CPU errata:
14288 A double-word or a variable shift may give an incorrect result if executed
14289 immediately after starting an integer division.
14291 A double-word or a variable shift may give an incorrect result if executed
14292 while an integer multiplication is in progress.
14294 An integer division may give an incorrect result if started in a delay slot
14295 of a taken branch or a jump.
14299 @itemx -mno-fix-r4400
14300 @opindex mfix-r4400
14301 @opindex mno-fix-r4400
14302 Work around certain R4400 CPU errata:
14305 A double-word or a variable shift may give an incorrect result if executed
14306 immediately after starting an integer division.
14310 @itemx -mno-fix-r10000
14311 @opindex mfix-r10000
14312 @opindex mno-fix-r10000
14313 Work around certain R10000 errata:
14316 @code{ll}/@code{sc} sequences may not behave atomically on revisions
14317 prior to 3.0. They may deadlock on revisions 2.6 and earlier.
14320 This option can only be used if the target architecture supports
14321 branch-likely instructions. @option{-mfix-r10000} is the default when
14322 @option{-march=r10000} is used; @option{-mno-fix-r10000} is the default
14326 @itemx -mno-fix-vr4120
14327 @opindex mfix-vr4120
14328 Work around certain VR4120 errata:
14331 @code{dmultu} does not always produce the correct result.
14333 @code{div} and @code{ddiv} do not always produce the correct result if one
14334 of the operands is negative.
14336 The workarounds for the division errata rely on special functions in
14337 @file{libgcc.a}. At present, these functions are only provided by
14338 the @code{mips64vr*-elf} configurations.
14340 Other VR4120 errata require a nop to be inserted between certain pairs of
14341 instructions. These errata are handled by the assembler, not by GCC itself.
14344 @opindex mfix-vr4130
14345 Work around the VR4130 @code{mflo}/@code{mfhi} errata. The
14346 workarounds are implemented by the assembler rather than by GCC,
14347 although GCC will avoid using @code{mflo} and @code{mfhi} if the
14348 VR4130 @code{macc}, @code{macchi}, @code{dmacc} and @code{dmacchi}
14349 instructions are available instead.
14352 @itemx -mno-fix-sb1
14354 Work around certain SB-1 CPU core errata.
14355 (This flag currently works around the SB-1 revision 2
14356 ``F1'' and ``F2'' floating point errata.)
14358 @item -mr10k-cache-barrier=@var{setting}
14359 @opindex mr10k-cache-barrier
14360 Specify whether GCC should insert cache barriers to avoid the
14361 side-effects of speculation on R10K processors.
14363 In common with many processors, the R10K tries to predict the outcome
14364 of a conditional branch and speculatively executes instructions from
14365 the ``taken'' branch. It later aborts these instructions if the
14366 predicted outcome was wrong. However, on the R10K, even aborted
14367 instructions can have side effects.
14369 This problem only affects kernel stores and, depending on the system,
14370 kernel loads. As an example, a speculatively-executed store may load
14371 the target memory into cache and mark the cache line as dirty, even if
14372 the store itself is later aborted. If a DMA operation writes to the
14373 same area of memory before the ``dirty'' line is flushed, the cached
14374 data will overwrite the DMA-ed data. See the R10K processor manual
14375 for a full description, including other potential problems.
14377 One workaround is to insert cache barrier instructions before every memory
14378 access that might be speculatively executed and that might have side
14379 effects even if aborted. @option{-mr10k-cache-barrier=@var{setting}}
14380 controls GCC's implementation of this workaround. It assumes that
14381 aborted accesses to any byte in the following regions will not have
14386 the memory occupied by the current function's stack frame;
14389 the memory occupied by an incoming stack argument;
14392 the memory occupied by an object with a link-time-constant address.
14395 It is the kernel's responsibility to ensure that speculative
14396 accesses to these regions are indeed safe.
14398 If the input program contains a function declaration such as:
14404 then the implementation of @code{foo} must allow @code{j foo} and
14405 @code{jal foo} to be executed speculatively. GCC honors this
14406 restriction for functions it compiles itself. It expects non-GCC
14407 functions (such as hand-written assembly code) to do the same.
14409 The option has three forms:
14412 @item -mr10k-cache-barrier=load-store
14413 Insert a cache barrier before a load or store that might be
14414 speculatively executed and that might have side effects even
14417 @item -mr10k-cache-barrier=store
14418 Insert a cache barrier before a store that might be speculatively
14419 executed and that might have side effects even if aborted.
14421 @item -mr10k-cache-barrier=none
14422 Disable the insertion of cache barriers. This is the default setting.
14425 @item -mflush-func=@var{func}
14426 @itemx -mno-flush-func
14427 @opindex mflush-func
14428 Specifies the function to call to flush the I and D caches, or to not
14429 call any such function. If called, the function must take the same
14430 arguments as the common @code{_flush_func()}, that is, the address of the
14431 memory range for which the cache is being flushed, the size of the
14432 memory range, and the number 3 (to flush both caches). The default
14433 depends on the target GCC was configured for, but commonly is either
14434 @samp{_flush_func} or @samp{__cpu_flush}.
14436 @item mbranch-cost=@var{num}
14437 @opindex mbranch-cost
14438 Set the cost of branches to roughly @var{num} ``simple'' instructions.
14439 This cost is only a heuristic and is not guaranteed to produce
14440 consistent results across releases. A zero cost redundantly selects
14441 the default, which is based on the @option{-mtune} setting.
14443 @item -mbranch-likely
14444 @itemx -mno-branch-likely
14445 @opindex mbranch-likely
14446 @opindex mno-branch-likely
14447 Enable or disable use of Branch Likely instructions, regardless of the
14448 default for the selected architecture. By default, Branch Likely
14449 instructions may be generated if they are supported by the selected
14450 architecture. An exception is for the MIPS32 and MIPS64 architectures
14451 and processors which implement those architectures; for those, Branch
14452 Likely instructions will not be generated by default because the MIPS32
14453 and MIPS64 architectures specifically deprecate their use.
14455 @item -mfp-exceptions
14456 @itemx -mno-fp-exceptions
14457 @opindex mfp-exceptions
14458 Specifies whether FP exceptions are enabled. This affects how we schedule
14459 FP instructions for some processors. The default is that FP exceptions are
14462 For instance, on the SB-1, if FP exceptions are disabled, and we are emitting
14463 64-bit code, then we can use both FP pipes. Otherwise, we can only use one
14466 @item -mvr4130-align
14467 @itemx -mno-vr4130-align
14468 @opindex mvr4130-align
14469 The VR4130 pipeline is two-way superscalar, but can only issue two
14470 instructions together if the first one is 8-byte aligned. When this
14471 option is enabled, GCC will align pairs of instructions that it
14472 thinks should execute in parallel.
14474 This option only has an effect when optimizing for the VR4130.
14475 It normally makes code faster, but at the expense of making it bigger.
14476 It is enabled by default at optimization level @option{-O3}.
14481 Enable (disable) generation of @code{synci} instructions on
14482 architectures that support it. The @code{synci} instructions (if
14483 enabled) will be generated when @code{__builtin___clear_cache()} is
14486 This option defaults to @code{-mno-synci}, but the default can be
14487 overridden by configuring with @code{--with-synci}.
14489 When compiling code for single processor systems, it is generally safe
14490 to use @code{synci}. However, on many multi-core (SMP) systems, it
14491 will not invalidate the instruction caches on all cores and may lead
14492 to undefined behavior.
14494 @item -mrelax-pic-calls
14495 @itemx -mno-relax-pic-calls
14496 @opindex mrelax-pic-calls
14497 Try to turn PIC calls that are normally dispatched via register
14498 @code{$25} into direct calls. This is only possible if the linker can
14499 resolve the destination at link-time and if the destination is within
14500 range for a direct call.
14502 @option{-mrelax-pic-calls} is the default if GCC was configured to use
14503 an assembler and a linker that supports the @code{.reloc} assembly
14504 directive and @code{-mexplicit-relocs} is in effect. With
14505 @code{-mno-explicit-relocs}, this optimization can be performed by the
14506 assembler and the linker alone without help from the compiler.
14508 @item -mmcount-ra-address
14509 @itemx -mno-mcount-ra-address
14510 @opindex mmcount-ra-address
14511 @opindex mno-mcount-ra-address
14512 Emit (do not emit) code that allows @code{_mcount} to modify the
14513 calling function's return address. When enabled, this option extends
14514 the usual @code{_mcount} interface with a new @var{ra-address}
14515 parameter, which has type @code{intptr_t *} and is passed in register
14516 @code{$12}. @code{_mcount} can then modify the return address by
14517 doing both of the following:
14520 Returning the new address in register @code{$31}.
14522 Storing the new address in @code{*@var{ra-address}},
14523 if @var{ra-address} is nonnull.
14526 The default is @option{-mno-mcount-ra-address}.
14531 @subsection MMIX Options
14532 @cindex MMIX Options
14534 These options are defined for the MMIX:
14538 @itemx -mno-libfuncs
14540 @opindex mno-libfuncs
14541 Specify that intrinsic library functions are being compiled, passing all
14542 values in registers, no matter the size.
14545 @itemx -mno-epsilon
14547 @opindex mno-epsilon
14548 Generate floating-point comparison instructions that compare with respect
14549 to the @code{rE} epsilon register.
14551 @item -mabi=mmixware
14553 @opindex mabi=mmixware
14555 Generate code that passes function parameters and return values that (in
14556 the called function) are seen as registers @code{$0} and up, as opposed to
14557 the GNU ABI which uses global registers @code{$231} and up.
14559 @item -mzero-extend
14560 @itemx -mno-zero-extend
14561 @opindex mzero-extend
14562 @opindex mno-zero-extend
14563 When reading data from memory in sizes shorter than 64 bits, use (do not
14564 use) zero-extending load instructions by default, rather than
14565 sign-extending ones.
14568 @itemx -mno-knuthdiv
14570 @opindex mno-knuthdiv
14571 Make the result of a division yielding a remainder have the same sign as
14572 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
14573 remainder follows the sign of the dividend. Both methods are
14574 arithmetically valid, the latter being almost exclusively used.
14576 @item -mtoplevel-symbols
14577 @itemx -mno-toplevel-symbols
14578 @opindex mtoplevel-symbols
14579 @opindex mno-toplevel-symbols
14580 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
14581 code can be used with the @code{PREFIX} assembly directive.
14585 Generate an executable in the ELF format, rather than the default
14586 @samp{mmo} format used by the @command{mmix} simulator.
14588 @item -mbranch-predict
14589 @itemx -mno-branch-predict
14590 @opindex mbranch-predict
14591 @opindex mno-branch-predict
14592 Use (do not use) the probable-branch instructions, when static branch
14593 prediction indicates a probable branch.
14595 @item -mbase-addresses
14596 @itemx -mno-base-addresses
14597 @opindex mbase-addresses
14598 @opindex mno-base-addresses
14599 Generate (do not generate) code that uses @emph{base addresses}. Using a
14600 base address automatically generates a request (handled by the assembler
14601 and the linker) for a constant to be set up in a global register. The
14602 register is used for one or more base address requests within the range 0
14603 to 255 from the value held in the register. The generally leads to short
14604 and fast code, but the number of different data items that can be
14605 addressed is limited. This means that a program that uses lots of static
14606 data may require @option{-mno-base-addresses}.
14608 @item -msingle-exit
14609 @itemx -mno-single-exit
14610 @opindex msingle-exit
14611 @opindex mno-single-exit
14612 Force (do not force) generated code to have a single exit point in each
14616 @node MN10300 Options
14617 @subsection MN10300 Options
14618 @cindex MN10300 options
14620 These @option{-m} options are defined for Matsushita MN10300 architectures:
14625 Generate code to avoid bugs in the multiply instructions for the MN10300
14626 processors. This is the default.
14628 @item -mno-mult-bug
14629 @opindex mno-mult-bug
14630 Do not generate code to avoid bugs in the multiply instructions for the
14631 MN10300 processors.
14635 Generate code which uses features specific to the AM33 processor.
14639 Do not generate code which uses features specific to the AM33 processor. This
14642 @item -mreturn-pointer-on-d0
14643 @opindex mreturn-pointer-on-d0
14644 When generating a function which returns a pointer, return the pointer
14645 in both @code{a0} and @code{d0}. Otherwise, the pointer is returned
14646 only in a0, and attempts to call such functions without a prototype
14647 would result in errors. Note that this option is on by default; use
14648 @option{-mno-return-pointer-on-d0} to disable it.
14652 Do not link in the C run-time initialization object file.
14656 Indicate to the linker that it should perform a relaxation optimization pass
14657 to shorten branches, calls and absolute memory addresses. This option only
14658 has an effect when used on the command line for the final link step.
14660 This option makes symbolic debugging impossible.
14663 @node PDP-11 Options
14664 @subsection PDP-11 Options
14665 @cindex PDP-11 Options
14667 These options are defined for the PDP-11:
14672 Use hardware FPP floating point. This is the default. (FIS floating
14673 point on the PDP-11/40 is not supported.)
14676 @opindex msoft-float
14677 Do not use hardware floating point.
14681 Return floating-point results in ac0 (fr0 in Unix assembler syntax).
14685 Return floating-point results in memory. This is the default.
14689 Generate code for a PDP-11/40.
14693 Generate code for a PDP-11/45. This is the default.
14697 Generate code for a PDP-11/10.
14699 @item -mbcopy-builtin
14700 @opindex mbcopy-builtin
14701 Use inline @code{movmemhi} patterns for copying memory. This is the
14706 Do not use inline @code{movmemhi} patterns for copying memory.
14712 Use 16-bit @code{int}. This is the default.
14718 Use 32-bit @code{int}.
14721 @itemx -mno-float32
14723 @opindex mno-float32
14724 Use 64-bit @code{float}. This is the default.
14727 @itemx -mno-float64
14729 @opindex mno-float64
14730 Use 32-bit @code{float}.
14734 Use @code{abshi2} pattern. This is the default.
14738 Do not use @code{abshi2} pattern.
14740 @item -mbranch-expensive
14741 @opindex mbranch-expensive
14742 Pretend that branches are expensive. This is for experimenting with
14743 code generation only.
14745 @item -mbranch-cheap
14746 @opindex mbranch-cheap
14747 Do not pretend that branches are expensive. This is the default.
14751 Generate code for a system with split I&D@.
14755 Generate code for a system without split I&D@. This is the default.
14759 Use Unix assembler syntax. This is the default when configured for
14760 @samp{pdp11-*-bsd}.
14764 Use DEC assembler syntax. This is the default when configured for any
14765 PDP-11 target other than @samp{pdp11-*-bsd}.
14768 @node picoChip Options
14769 @subsection picoChip Options
14770 @cindex picoChip options
14772 These @samp{-m} options are defined for picoChip implementations:
14776 @item -mae=@var{ae_type}
14778 Set the instruction set, register set, and instruction scheduling
14779 parameters for array element type @var{ae_type}. Supported values
14780 for @var{ae_type} are @samp{ANY}, @samp{MUL}, and @samp{MAC}.
14782 @option{-mae=ANY} selects a completely generic AE type. Code
14783 generated with this option will run on any of the other AE types. The
14784 code will not be as efficient as it would be if compiled for a specific
14785 AE type, and some types of operation (e.g., multiplication) will not
14786 work properly on all types of AE.
14788 @option{-mae=MUL} selects a MUL AE type. This is the most useful AE type
14789 for compiled code, and is the default.
14791 @option{-mae=MAC} selects a DSP-style MAC AE. Code compiled with this
14792 option may suffer from poor performance of byte (char) manipulation,
14793 since the DSP AE does not provide hardware support for byte load/stores.
14795 @item -msymbol-as-address
14796 Enable the compiler to directly use a symbol name as an address in a
14797 load/store instruction, without first loading it into a
14798 register. Typically, the use of this option will generate larger
14799 programs, which run faster than when the option isn't used. However, the
14800 results vary from program to program, so it is left as a user option,
14801 rather than being permanently enabled.
14803 @item -mno-inefficient-warnings
14804 Disables warnings about the generation of inefficient code. These
14805 warnings can be generated, for example, when compiling code which
14806 performs byte-level memory operations on the MAC AE type. The MAC AE has
14807 no hardware support for byte-level memory operations, so all byte
14808 load/stores must be synthesized from word load/store operations. This is
14809 inefficient and a warning will be generated indicating to the programmer
14810 that they should rewrite the code to avoid byte operations, or to target
14811 an AE type which has the necessary hardware support. This option enables
14812 the warning to be turned off.
14816 @node PowerPC Options
14817 @subsection PowerPC Options
14818 @cindex PowerPC options
14820 These are listed under @xref{RS/6000 and PowerPC Options}.
14822 @node RS/6000 and PowerPC Options
14823 @subsection IBM RS/6000 and PowerPC Options
14824 @cindex RS/6000 and PowerPC Options
14825 @cindex IBM RS/6000 and PowerPC Options
14827 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
14834 @itemx -mno-powerpc
14835 @itemx -mpowerpc-gpopt
14836 @itemx -mno-powerpc-gpopt
14837 @itemx -mpowerpc-gfxopt
14838 @itemx -mno-powerpc-gfxopt
14840 @itemx -mno-powerpc64
14844 @itemx -mno-popcntb
14846 @itemx -mno-popcntd
14854 @itemx -mno-hard-dfp
14858 @opindex mno-power2
14860 @opindex mno-powerpc
14861 @opindex mpowerpc-gpopt
14862 @opindex mno-powerpc-gpopt
14863 @opindex mpowerpc-gfxopt
14864 @opindex mno-powerpc-gfxopt
14865 @opindex mpowerpc64
14866 @opindex mno-powerpc64
14870 @opindex mno-popcntb
14872 @opindex mno-popcntd
14878 @opindex mno-mfpgpr
14880 @opindex mno-hard-dfp
14881 GCC supports two related instruction set architectures for the
14882 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
14883 instructions supported by the @samp{rios} chip set used in the original
14884 RS/6000 systems and the @dfn{PowerPC} instruction set is the
14885 architecture of the Freescale MPC5xx, MPC6xx, MPC8xx microprocessors, and
14886 the IBM 4xx, 6xx, and follow-on microprocessors.
14888 Neither architecture is a subset of the other. However there is a
14889 large common subset of instructions supported by both. An MQ
14890 register is included in processors supporting the POWER architecture.
14892 You use these options to specify which instructions are available on the
14893 processor you are using. The default value of these options is
14894 determined when configuring GCC@. Specifying the
14895 @option{-mcpu=@var{cpu_type}} overrides the specification of these
14896 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
14897 rather than the options listed above.
14899 The @option{-mpower} option allows GCC to generate instructions that
14900 are found only in the POWER architecture and to use the MQ register.
14901 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
14902 to generate instructions that are present in the POWER2 architecture but
14903 not the original POWER architecture.
14905 The @option{-mpowerpc} option allows GCC to generate instructions that
14906 are found only in the 32-bit subset of the PowerPC architecture.
14907 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
14908 GCC to use the optional PowerPC architecture instructions in the
14909 General Purpose group, including floating-point square root. Specifying
14910 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
14911 use the optional PowerPC architecture instructions in the Graphics
14912 group, including floating-point select.
14914 The @option{-mmfcrf} option allows GCC to generate the move from
14915 condition register field instruction implemented on the POWER4
14916 processor and other processors that support the PowerPC V2.01
14918 The @option{-mpopcntb} option allows GCC to generate the popcount and
14919 double precision FP reciprocal estimate instruction implemented on the
14920 POWER5 processor and other processors that support the PowerPC V2.02
14922 The @option{-mpopcntd} option allows GCC to generate the popcount
14923 instruction implemented on the POWER7 processor and other processors
14924 that support the PowerPC V2.06 architecture.
14925 The @option{-mfprnd} option allows GCC to generate the FP round to
14926 integer instructions implemented on the POWER5+ processor and other
14927 processors that support the PowerPC V2.03 architecture.
14928 The @option{-mcmpb} option allows GCC to generate the compare bytes
14929 instruction implemented on the POWER6 processor and other processors
14930 that support the PowerPC V2.05 architecture.
14931 The @option{-mmfpgpr} option allows GCC to generate the FP move to/from
14932 general purpose register instructions implemented on the POWER6X
14933 processor and other processors that support the extended PowerPC V2.05
14935 The @option{-mhard-dfp} option allows GCC to generate the decimal floating
14936 point instructions implemented on some POWER processors.
14938 The @option{-mpowerpc64} option allows GCC to generate the additional
14939 64-bit instructions that are found in the full PowerPC64 architecture
14940 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
14941 @option{-mno-powerpc64}.
14943 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
14944 will use only the instructions in the common subset of both
14945 architectures plus some special AIX common-mode calls, and will not use
14946 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
14947 permits GCC to use any instruction from either architecture and to
14948 allow use of the MQ register; specify this for the Motorola MPC601.
14950 @item -mnew-mnemonics
14951 @itemx -mold-mnemonics
14952 @opindex mnew-mnemonics
14953 @opindex mold-mnemonics
14954 Select which mnemonics to use in the generated assembler code. With
14955 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
14956 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
14957 assembler mnemonics defined for the POWER architecture. Instructions
14958 defined in only one architecture have only one mnemonic; GCC uses that
14959 mnemonic irrespective of which of these options is specified.
14961 GCC defaults to the mnemonics appropriate for the architecture in
14962 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
14963 value of these option. Unless you are building a cross-compiler, you
14964 should normally not specify either @option{-mnew-mnemonics} or
14965 @option{-mold-mnemonics}, but should instead accept the default.
14967 @item -mcpu=@var{cpu_type}
14969 Set architecture type, register usage, choice of mnemonics, and
14970 instruction scheduling parameters for machine type @var{cpu_type}.
14971 Supported values for @var{cpu_type} are @samp{401}, @samp{403},
14972 @samp{405}, @samp{405fp}, @samp{440}, @samp{440fp}, @samp{464}, @samp{464fp},
14973 @samp{476}, @samp{476fp}, @samp{505}, @samp{601}, @samp{602}, @samp{603},
14974 @samp{603e}, @samp{604}, @samp{604e}, @samp{620}, @samp{630}, @samp{740},
14975 @samp{7400}, @samp{7450}, @samp{750}, @samp{801}, @samp{821}, @samp{823},
14976 @samp{860}, @samp{970}, @samp{8540}, @samp{a2}, @samp{e300c2},
14977 @samp{e300c3}, @samp{e500mc}, @samp{e500mc64}, @samp{ec603e}, @samp{G3},
14978 @samp{G4}, @samp{G5}, @samp{titan}, @samp{power}, @samp{power2}, @samp{power3},
14979 @samp{power4}, @samp{power5}, @samp{power5+}, @samp{power6}, @samp{power6x},
14980 @samp{power7}, @samp{common}, @samp{powerpc}, @samp{powerpc64}, @samp{rios},
14981 @samp{rios1}, @samp{rios2}, @samp{rsc}, and @samp{rs64}.
14983 @option{-mcpu=common} selects a completely generic processor. Code
14984 generated under this option will run on any POWER or PowerPC processor.
14985 GCC will use only the instructions in the common subset of both
14986 architectures, and will not use the MQ register. GCC assumes a generic
14987 processor model for scheduling purposes.
14989 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
14990 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
14991 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
14992 types, with an appropriate, generic processor model assumed for
14993 scheduling purposes.
14995 The other options specify a specific processor. Code generated under
14996 those options will run best on that processor, and may not run at all on
14999 The @option{-mcpu} options automatically enable or disable the
15002 @gccoptlist{-maltivec -mfprnd -mhard-float -mmfcrf -mmultiple @gol
15003 -mnew-mnemonics -mpopcntb -mpopcntd -mpower -mpower2 -mpowerpc64 @gol
15004 -mpowerpc-gpopt -mpowerpc-gfxopt -msingle-float -mdouble-float @gol
15005 -msimple-fpu -mstring -mmulhw -mdlmzb -mmfpgpr -mvsx}
15007 The particular options set for any particular CPU will vary between
15008 compiler versions, depending on what setting seems to produce optimal
15009 code for that CPU; it doesn't necessarily reflect the actual hardware's
15010 capabilities. If you wish to set an individual option to a particular
15011 value, you may specify it after the @option{-mcpu} option, like
15012 @samp{-mcpu=970 -mno-altivec}.
15014 On AIX, the @option{-maltivec} and @option{-mpowerpc64} options are
15015 not enabled or disabled by the @option{-mcpu} option at present because
15016 AIX does not have full support for these options. You may still
15017 enable or disable them individually if you're sure it'll work in your
15020 @item -mtune=@var{cpu_type}
15022 Set the instruction scheduling parameters for machine type
15023 @var{cpu_type}, but do not set the architecture type, register usage, or
15024 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
15025 values for @var{cpu_type} are used for @option{-mtune} as for
15026 @option{-mcpu}. If both are specified, the code generated will use the
15027 architecture, registers, and mnemonics set by @option{-mcpu}, but the
15028 scheduling parameters set by @option{-mtune}.
15030 @item -mcmodel=small
15031 @opindex mcmodel=small
15032 Generate PowerPC64 code for the small model: The TOC is limited to
15035 @item -mcmodel=large
15036 @opindex mcmodel=large
15037 Generate PowerPC64 code for the large model: The TOC may be up to 4G
15038 in size. Other data and code is only limited by the 64-bit address
15042 @itemx -mno-altivec
15044 @opindex mno-altivec
15045 Generate code that uses (does not use) AltiVec instructions, and also
15046 enable the use of built-in functions that allow more direct access to
15047 the AltiVec instruction set. You may also need to set
15048 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
15054 @opindex mno-vrsave
15055 Generate VRSAVE instructions when generating AltiVec code.
15057 @item -mgen-cell-microcode
15058 @opindex mgen-cell-microcode
15059 Generate Cell microcode instructions
15061 @item -mwarn-cell-microcode
15062 @opindex mwarn-cell-microcode
15063 Warning when a Cell microcode instruction is going to emitted. An example
15064 of a Cell microcode instruction is a variable shift.
15067 @opindex msecure-plt
15068 Generate code that allows ld and ld.so to build executables and shared
15069 libraries with non-exec .plt and .got sections. This is a PowerPC
15070 32-bit SYSV ABI option.
15074 Generate code that uses a BSS .plt section that ld.so fills in, and
15075 requires .plt and .got sections that are both writable and executable.
15076 This is a PowerPC 32-bit SYSV ABI option.
15082 This switch enables or disables the generation of ISEL instructions.
15084 @item -misel=@var{yes/no}
15085 This switch has been deprecated. Use @option{-misel} and
15086 @option{-mno-isel} instead.
15092 This switch enables or disables the generation of SPE simd
15098 @opindex mno-paired
15099 This switch enables or disables the generation of PAIRED simd
15102 @item -mspe=@var{yes/no}
15103 This option has been deprecated. Use @option{-mspe} and
15104 @option{-mno-spe} instead.
15110 Generate code that uses (does not use) vector/scalar (VSX)
15111 instructions, and also enable the use of built-in functions that allow
15112 more direct access to the VSX instruction set.
15114 @item -mfloat-gprs=@var{yes/single/double/no}
15115 @itemx -mfloat-gprs
15116 @opindex mfloat-gprs
15117 This switch enables or disables the generation of floating point
15118 operations on the general purpose registers for architectures that
15121 The argument @var{yes} or @var{single} enables the use of
15122 single-precision floating point operations.
15124 The argument @var{double} enables the use of single and
15125 double-precision floating point operations.
15127 The argument @var{no} disables floating point operations on the
15128 general purpose registers.
15130 This option is currently only available on the MPC854x.
15136 Generate code for 32-bit or 64-bit environments of Darwin and SVR4
15137 targets (including GNU/Linux). The 32-bit environment sets int, long
15138 and pointer to 32 bits and generates code that runs on any PowerPC
15139 variant. The 64-bit environment sets int to 32 bits and long and
15140 pointer to 64 bits, and generates code for PowerPC64, as for
15141 @option{-mpowerpc64}.
15144 @itemx -mno-fp-in-toc
15145 @itemx -mno-sum-in-toc
15146 @itemx -mminimal-toc
15148 @opindex mno-fp-in-toc
15149 @opindex mno-sum-in-toc
15150 @opindex mminimal-toc
15151 Modify generation of the TOC (Table Of Contents), which is created for
15152 every executable file. The @option{-mfull-toc} option is selected by
15153 default. In that case, GCC will allocate at least one TOC entry for
15154 each unique non-automatic variable reference in your program. GCC
15155 will also place floating-point constants in the TOC@. However, only
15156 16,384 entries are available in the TOC@.
15158 If you receive a linker error message that saying you have overflowed
15159 the available TOC space, you can reduce the amount of TOC space used
15160 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
15161 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
15162 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
15163 generate code to calculate the sum of an address and a constant at
15164 run-time instead of putting that sum into the TOC@. You may specify one
15165 or both of these options. Each causes GCC to produce very slightly
15166 slower and larger code at the expense of conserving TOC space.
15168 If you still run out of space in the TOC even when you specify both of
15169 these options, specify @option{-mminimal-toc} instead. This option causes
15170 GCC to make only one TOC entry for every file. When you specify this
15171 option, GCC will produce code that is slower and larger but which
15172 uses extremely little TOC space. You may wish to use this option
15173 only on files that contain less frequently executed code.
15179 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
15180 @code{long} type, and the infrastructure needed to support them.
15181 Specifying @option{-maix64} implies @option{-mpowerpc64} and
15182 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
15183 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
15186 @itemx -mno-xl-compat
15187 @opindex mxl-compat
15188 @opindex mno-xl-compat
15189 Produce code that conforms more closely to IBM XL compiler semantics
15190 when using AIX-compatible ABI@. Pass floating-point arguments to
15191 prototyped functions beyond the register save area (RSA) on the stack
15192 in addition to argument FPRs. Do not assume that most significant
15193 double in 128-bit long double value is properly rounded when comparing
15194 values and converting to double. Use XL symbol names for long double
15197 The AIX calling convention was extended but not initially documented to
15198 handle an obscure K&R C case of calling a function that takes the
15199 address of its arguments with fewer arguments than declared. IBM XL
15200 compilers access floating point arguments which do not fit in the
15201 RSA from the stack when a subroutine is compiled without
15202 optimization. Because always storing floating-point arguments on the
15203 stack is inefficient and rarely needed, this option is not enabled by
15204 default and only is necessary when calling subroutines compiled by IBM
15205 XL compilers without optimization.
15209 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
15210 application written to use message passing with special startup code to
15211 enable the application to run. The system must have PE installed in the
15212 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
15213 must be overridden with the @option{-specs=} option to specify the
15214 appropriate directory location. The Parallel Environment does not
15215 support threads, so the @option{-mpe} option and the @option{-pthread}
15216 option are incompatible.
15218 @item -malign-natural
15219 @itemx -malign-power
15220 @opindex malign-natural
15221 @opindex malign-power
15222 On AIX, 32-bit Darwin, and 64-bit PowerPC GNU/Linux, the option
15223 @option{-malign-natural} overrides the ABI-defined alignment of larger
15224 types, such as floating-point doubles, on their natural size-based boundary.
15225 The option @option{-malign-power} instructs GCC to follow the ABI-specified
15226 alignment rules. GCC defaults to the standard alignment defined in the ABI@.
15228 On 64-bit Darwin, natural alignment is the default, and @option{-malign-power}
15232 @itemx -mhard-float
15233 @opindex msoft-float
15234 @opindex mhard-float
15235 Generate code that does not use (uses) the floating-point register set.
15236 Software floating point emulation is provided if you use the
15237 @option{-msoft-float} option, and pass the option to GCC when linking.
15239 @item -msingle-float
15240 @itemx -mdouble-float
15241 @opindex msingle-float
15242 @opindex mdouble-float
15243 Generate code for single or double-precision floating point operations.
15244 @option{-mdouble-float} implies @option{-msingle-float}.
15247 @opindex msimple-fpu
15248 Do not generate sqrt and div instructions for hardware floating point unit.
15252 Specify type of floating point unit. Valid values are @var{sp_lite}
15253 (equivalent to -msingle-float -msimple-fpu), @var{dp_lite} (equivalent
15254 to -mdouble-float -msimple-fpu), @var{sp_full} (equivalent to -msingle-float),
15255 and @var{dp_full} (equivalent to -mdouble-float).
15258 @opindex mxilinx-fpu
15259 Perform optimizations for floating point unit on Xilinx PPC 405/440.
15262 @itemx -mno-multiple
15264 @opindex mno-multiple
15265 Generate code that uses (does not use) the load multiple word
15266 instructions and the store multiple word instructions. These
15267 instructions are generated by default on POWER systems, and not
15268 generated on PowerPC systems. Do not use @option{-mmultiple} on little
15269 endian PowerPC systems, since those instructions do not work when the
15270 processor is in little endian mode. The exceptions are PPC740 and
15271 PPC750 which permit the instructions usage in little endian mode.
15276 @opindex mno-string
15277 Generate code that uses (does not use) the load string instructions
15278 and the store string word instructions to save multiple registers and
15279 do small block moves. These instructions are generated by default on
15280 POWER systems, and not generated on PowerPC systems. Do not use
15281 @option{-mstring} on little endian PowerPC systems, since those
15282 instructions do not work when the processor is in little endian mode.
15283 The exceptions are PPC740 and PPC750 which permit the instructions
15284 usage in little endian mode.
15289 @opindex mno-update
15290 Generate code that uses (does not use) the load or store instructions
15291 that update the base register to the address of the calculated memory
15292 location. These instructions are generated by default. If you use
15293 @option{-mno-update}, there is a small window between the time that the
15294 stack pointer is updated and the address of the previous frame is
15295 stored, which means code that walks the stack frame across interrupts or
15296 signals may get corrupted data.
15298 @item -mavoid-indexed-addresses
15299 @itemx -mno-avoid-indexed-addresses
15300 @opindex mavoid-indexed-addresses
15301 @opindex mno-avoid-indexed-addresses
15302 Generate code that tries to avoid (not avoid) the use of indexed load
15303 or store instructions. These instructions can incur a performance
15304 penalty on Power6 processors in certain situations, such as when
15305 stepping through large arrays that cross a 16M boundary. This option
15306 is enabled by default when targetting Power6 and disabled otherwise.
15309 @itemx -mno-fused-madd
15310 @opindex mfused-madd
15311 @opindex mno-fused-madd
15312 Generate code that uses (does not use) the floating point multiply and
15313 accumulate instructions. These instructions are generated by default if
15314 hardware floating is used.
15320 Generate code that uses (does not use) the half-word multiply and
15321 multiply-accumulate instructions on the IBM 405, 440, 464 and 476 processors.
15322 These instructions are generated by default when targetting those
15329 Generate code that uses (does not use) the string-search @samp{dlmzb}
15330 instruction on the IBM 405, 440, 464 and 476 processors. This instruction is
15331 generated by default when targetting those processors.
15333 @item -mno-bit-align
15335 @opindex mno-bit-align
15336 @opindex mbit-align
15337 On System V.4 and embedded PowerPC systems do not (do) force structures
15338 and unions that contain bit-fields to be aligned to the base type of the
15341 For example, by default a structure containing nothing but 8
15342 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
15343 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
15344 the structure would be aligned to a 1 byte boundary and be one byte in
15347 @item -mno-strict-align
15348 @itemx -mstrict-align
15349 @opindex mno-strict-align
15350 @opindex mstrict-align
15351 On System V.4 and embedded PowerPC systems do not (do) assume that
15352 unaligned memory references will be handled by the system.
15354 @item -mrelocatable
15355 @itemx -mno-relocatable
15356 @opindex mrelocatable
15357 @opindex mno-relocatable
15358 On embedded PowerPC systems generate code that allows (does not allow)
15359 the program to be relocated to a different address at runtime. If you
15360 use @option{-mrelocatable} on any module, all objects linked together must
15361 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
15363 @item -mrelocatable-lib
15364 @itemx -mno-relocatable-lib
15365 @opindex mrelocatable-lib
15366 @opindex mno-relocatable-lib
15367 On embedded PowerPC systems generate code that allows (does not allow)
15368 the program to be relocated to a different address at runtime. Modules
15369 compiled with @option{-mrelocatable-lib} can be linked with either modules
15370 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
15371 with modules compiled with the @option{-mrelocatable} options.
15377 On System V.4 and embedded PowerPC systems do not (do) assume that
15378 register 2 contains a pointer to a global area pointing to the addresses
15379 used in the program.
15382 @itemx -mlittle-endian
15384 @opindex mlittle-endian
15385 On System V.4 and embedded PowerPC systems compile code for the
15386 processor in little endian mode. The @option{-mlittle-endian} option is
15387 the same as @option{-mlittle}.
15390 @itemx -mbig-endian
15392 @opindex mbig-endian
15393 On System V.4 and embedded PowerPC systems compile code for the
15394 processor in big endian mode. The @option{-mbig-endian} option is
15395 the same as @option{-mbig}.
15397 @item -mdynamic-no-pic
15398 @opindex mdynamic-no-pic
15399 On Darwin and Mac OS X systems, compile code so that it is not
15400 relocatable, but that its external references are relocatable. The
15401 resulting code is suitable for applications, but not shared
15404 @item -mprioritize-restricted-insns=@var{priority}
15405 @opindex mprioritize-restricted-insns
15406 This option controls the priority that is assigned to
15407 dispatch-slot restricted instructions during the second scheduling
15408 pass. The argument @var{priority} takes the value @var{0/1/2} to assign
15409 @var{no/highest/second-highest} priority to dispatch slot restricted
15412 @item -msched-costly-dep=@var{dependence_type}
15413 @opindex msched-costly-dep
15414 This option controls which dependences are considered costly
15415 by the target during instruction scheduling. The argument
15416 @var{dependence_type} takes one of the following values:
15417 @var{no}: no dependence is costly,
15418 @var{all}: all dependences are costly,
15419 @var{true_store_to_load}: a true dependence from store to load is costly,
15420 @var{store_to_load}: any dependence from store to load is costly,
15421 @var{number}: any dependence which latency >= @var{number} is costly.
15423 @item -minsert-sched-nops=@var{scheme}
15424 @opindex minsert-sched-nops
15425 This option controls which nop insertion scheme will be used during
15426 the second scheduling pass. The argument @var{scheme} takes one of the
15428 @var{no}: Don't insert nops.
15429 @var{pad}: Pad with nops any dispatch group which has vacant issue slots,
15430 according to the scheduler's grouping.
15431 @var{regroup_exact}: Insert nops to force costly dependent insns into
15432 separate groups. Insert exactly as many nops as needed to force an insn
15433 to a new group, according to the estimated processor grouping.
15434 @var{number}: Insert nops to force costly dependent insns into
15435 separate groups. Insert @var{number} nops to force an insn to a new group.
15438 @opindex mcall-sysv
15439 On System V.4 and embedded PowerPC systems compile code using calling
15440 conventions that adheres to the March 1995 draft of the System V
15441 Application Binary Interface, PowerPC processor supplement. This is the
15442 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
15444 @item -mcall-sysv-eabi
15446 @opindex mcall-sysv-eabi
15447 @opindex mcall-eabi
15448 Specify both @option{-mcall-sysv} and @option{-meabi} options.
15450 @item -mcall-sysv-noeabi
15451 @opindex mcall-sysv-noeabi
15452 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
15454 @item -mcall-aixdesc
15456 On System V.4 and embedded PowerPC systems compile code for the AIX
15460 @opindex mcall-linux
15461 On System V.4 and embedded PowerPC systems compile code for the
15462 Linux-based GNU system.
15466 On System V.4 and embedded PowerPC systems compile code for the
15467 Hurd-based GNU system.
15469 @item -mcall-freebsd
15470 @opindex mcall-freebsd
15471 On System V.4 and embedded PowerPC systems compile code for the
15472 FreeBSD operating system.
15474 @item -mcall-netbsd
15475 @opindex mcall-netbsd
15476 On System V.4 and embedded PowerPC systems compile code for the
15477 NetBSD operating system.
15479 @item -mcall-openbsd
15480 @opindex mcall-netbsd
15481 On System V.4 and embedded PowerPC systems compile code for the
15482 OpenBSD operating system.
15484 @item -maix-struct-return
15485 @opindex maix-struct-return
15486 Return all structures in memory (as specified by the AIX ABI)@.
15488 @item -msvr4-struct-return
15489 @opindex msvr4-struct-return
15490 Return structures smaller than 8 bytes in registers (as specified by the
15493 @item -mabi=@var{abi-type}
15495 Extend the current ABI with a particular extension, or remove such extension.
15496 Valid values are @var{altivec}, @var{no-altivec}, @var{spe},
15497 @var{no-spe}, @var{ibmlongdouble}, @var{ieeelongdouble}@.
15501 Extend the current ABI with SPE ABI extensions. This does not change
15502 the default ABI, instead it adds the SPE ABI extensions to the current
15506 @opindex mabi=no-spe
15507 Disable Booke SPE ABI extensions for the current ABI@.
15509 @item -mabi=ibmlongdouble
15510 @opindex mabi=ibmlongdouble
15511 Change the current ABI to use IBM extended precision long double.
15512 This is a PowerPC 32-bit SYSV ABI option.
15514 @item -mabi=ieeelongdouble
15515 @opindex mabi=ieeelongdouble
15516 Change the current ABI to use IEEE extended precision long double.
15517 This is a PowerPC 32-bit Linux ABI option.
15520 @itemx -mno-prototype
15521 @opindex mprototype
15522 @opindex mno-prototype
15523 On System V.4 and embedded PowerPC systems assume that all calls to
15524 variable argument functions are properly prototyped. Otherwise, the
15525 compiler must insert an instruction before every non prototyped call to
15526 set or clear bit 6 of the condition code register (@var{CR}) to
15527 indicate whether floating point values were passed in the floating point
15528 registers in case the function takes a variable arguments. With
15529 @option{-mprototype}, only calls to prototyped variable argument functions
15530 will set or clear the bit.
15534 On embedded PowerPC systems, assume that the startup module is called
15535 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
15536 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}
15541 On embedded PowerPC systems, assume that the startup module is called
15542 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
15547 On embedded PowerPC systems, assume that the startup module is called
15548 @file{crt0.o} and the standard C libraries are @file{libads.a} and
15551 @item -myellowknife
15552 @opindex myellowknife
15553 On embedded PowerPC systems, assume that the startup module is called
15554 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
15559 On System V.4 and embedded PowerPC systems, specify that you are
15560 compiling for a VxWorks system.
15564 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
15565 header to indicate that @samp{eabi} extended relocations are used.
15571 On System V.4 and embedded PowerPC systems do (do not) adhere to the
15572 Embedded Applications Binary Interface (eabi) which is a set of
15573 modifications to the System V.4 specifications. Selecting @option{-meabi}
15574 means that the stack is aligned to an 8 byte boundary, a function
15575 @code{__eabi} is called to from @code{main} to set up the eabi
15576 environment, and the @option{-msdata} option can use both @code{r2} and
15577 @code{r13} to point to two separate small data areas. Selecting
15578 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
15579 do not call an initialization function from @code{main}, and the
15580 @option{-msdata} option will only use @code{r13} to point to a single
15581 small data area. The @option{-meabi} option is on by default if you
15582 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
15585 @opindex msdata=eabi
15586 On System V.4 and embedded PowerPC systems, put small initialized
15587 @code{const} global and static data in the @samp{.sdata2} section, which
15588 is pointed to by register @code{r2}. Put small initialized
15589 non-@code{const} global and static data in the @samp{.sdata} section,
15590 which is pointed to by register @code{r13}. Put small uninitialized
15591 global and static data in the @samp{.sbss} section, which is adjacent to
15592 the @samp{.sdata} section. The @option{-msdata=eabi} option is
15593 incompatible with the @option{-mrelocatable} option. The
15594 @option{-msdata=eabi} option also sets the @option{-memb} option.
15597 @opindex msdata=sysv
15598 On System V.4 and embedded PowerPC systems, put small global and static
15599 data in the @samp{.sdata} section, which is pointed to by register
15600 @code{r13}. Put small uninitialized global and static data in the
15601 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
15602 The @option{-msdata=sysv} option is incompatible with the
15603 @option{-mrelocatable} option.
15605 @item -msdata=default
15607 @opindex msdata=default
15609 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
15610 compile code the same as @option{-msdata=eabi}, otherwise compile code the
15611 same as @option{-msdata=sysv}.
15614 @opindex msdata=data
15615 On System V.4 and embedded PowerPC systems, put small global
15616 data in the @samp{.sdata} section. Put small uninitialized global
15617 data in the @samp{.sbss} section. Do not use register @code{r13}
15618 to address small data however. This is the default behavior unless
15619 other @option{-msdata} options are used.
15623 @opindex msdata=none
15625 On embedded PowerPC systems, put all initialized global and static data
15626 in the @samp{.data} section, and all uninitialized data in the
15627 @samp{.bss} section.
15631 @cindex smaller data references (PowerPC)
15632 @cindex .sdata/.sdata2 references (PowerPC)
15633 On embedded PowerPC systems, put global and static items less than or
15634 equal to @var{num} bytes into the small data or bss sections instead of
15635 the normal data or bss section. By default, @var{num} is 8. The
15636 @option{-G @var{num}} switch is also passed to the linker.
15637 All modules should be compiled with the same @option{-G @var{num}} value.
15640 @itemx -mno-regnames
15642 @opindex mno-regnames
15643 On System V.4 and embedded PowerPC systems do (do not) emit register
15644 names in the assembly language output using symbolic forms.
15647 @itemx -mno-longcall
15649 @opindex mno-longcall
15650 By default assume that all calls are far away so that a longer more
15651 expensive calling sequence is required. This is required for calls
15652 further than 32 megabytes (33,554,432 bytes) from the current location.
15653 A short call will be generated if the compiler knows
15654 the call cannot be that far away. This setting can be overridden by
15655 the @code{shortcall} function attribute, or by @code{#pragma
15658 Some linkers are capable of detecting out-of-range calls and generating
15659 glue code on the fly. On these systems, long calls are unnecessary and
15660 generate slower code. As of this writing, the AIX linker can do this,
15661 as can the GNU linker for PowerPC/64. It is planned to add this feature
15662 to the GNU linker for 32-bit PowerPC systems as well.
15664 On Darwin/PPC systems, @code{#pragma longcall} will generate ``jbsr
15665 callee, L42'', plus a ``branch island'' (glue code). The two target
15666 addresses represent the callee and the ``branch island''. The
15667 Darwin/PPC linker will prefer the first address and generate a ``bl
15668 callee'' if the PPC ``bl'' instruction will reach the callee directly;
15669 otherwise, the linker will generate ``bl L42'' to call the ``branch
15670 island''. The ``branch island'' is appended to the body of the
15671 calling function; it computes the full 32-bit address of the callee
15674 On Mach-O (Darwin) systems, this option directs the compiler emit to
15675 the glue for every direct call, and the Darwin linker decides whether
15676 to use or discard it.
15678 In the future, we may cause GCC to ignore all longcall specifications
15679 when the linker is known to generate glue.
15681 @item -mtls-markers
15682 @itemx -mno-tls-markers
15683 @opindex mtls-markers
15684 @opindex mno-tls-markers
15685 Mark (do not mark) calls to @code{__tls_get_addr} with a relocation
15686 specifying the function argument. The relocation allows ld to
15687 reliably associate function call with argument setup instructions for
15688 TLS optimization, which in turn allows gcc to better schedule the
15693 Adds support for multithreading with the @dfn{pthreads} library.
15694 This option sets flags for both the preprocessor and linker.
15699 This option will enable GCC to use the reciprocal estimate and
15700 reciprocal square root estimate instructions with additional
15701 Newton-Raphson steps to increase precision instead of doing a divide or
15702 square root and divide for floating point arguments. You should use
15703 the @option{-ffast-math} option when using @option{-mrecip} (or at
15704 least @option{-funsafe-math-optimizations},
15705 @option{-finite-math-only}, @option{-freciprocal-math} and
15706 @option{-fno-trapping-math}). Note that while the throughput of the
15707 sequence is generally higher than the throughput of the non-reciprocal
15708 instruction, the precision of the sequence can be decreased by up to 2
15709 ulp (i.e. the inverse of 1.0 equals 0.99999994) for reciprocal square
15712 @item -mrecip=@var{opt}
15713 @opindex mrecip=opt
15714 This option allows to control which reciprocal estimate instructions
15715 may be used. @var{opt} is a comma separated list of options, that may
15716 be preceeded by a @code{!} to invert the option:
15717 @code{all}: enable all estimate instructions,
15718 @code{default}: enable the default instructions, equvalent to @option{-mrecip},
15719 @code{none}: disable all estimate instructions, equivalent to @option{-mno-recip};
15720 @code{div}: enable the reciprocal approximation instructions for both single and double precision;
15721 @code{divf}: enable the single precision reciprocal approximation instructions;
15722 @code{divd}: enable the double precision reciprocal approximation instructions;
15723 @code{rsqrt}: enable the reciprocal square root approximation instructions for both single and double precision;
15724 @code{rsqrtf}: enable the single precision reciprocal square root approximation instructions;
15725 @code{rsqrtd}: enable the double precision reciprocal square root approximation instructions;
15727 So for example, @option{-mrecip=all,!rsqrtd} would enable the
15728 all of the reciprocal estimate instructions, except for the
15729 @code{FRSQRTE}, @code{XSRSQRTEDP}, and @code{XVRSQRTEDP} instructions
15730 which handle the double precision reciprocal square root calculations.
15732 @item -mrecip-precision
15733 @itemx -mno-recip-precision
15734 @opindex mrecip-precision
15735 Assume (do not assume) that the reciprocal estimate instructions
15736 provide higher precision estimates than is mandated by the powerpc
15737 ABI. Selecting @option{-mcpu=power6} or @option{-mcpu=power7}
15738 automatically selects @option{-mrecip-precision}. The double
15739 precision square root estimate instructions are not generated by
15740 default on low precision machines, since they do not provide an
15741 estimate that converges after three steps.
15745 @subsection RX Options
15748 These command line options are defined for RX targets:
15751 @item -m64bit-doubles
15752 @itemx -m32bit-doubles
15753 @opindex m64bit-doubles
15754 @opindex m32bit-doubles
15755 Make the @code{double} data type be 64-bits (@option{-m64bit-doubles})
15756 or 32-bits (@option{-m32bit-doubles}) in size. The default is
15757 @option{-m32bit-doubles}. @emph{Note} RX floating point hardware only
15758 works on 32-bit values, which is why the default is
15759 @option{-m32bit-doubles}.
15765 Enables (@option{-fpu}) or disables (@option{-nofpu}) the use of RX
15766 floating point hardware. The default is enabled for the @var{RX600}
15767 series and disabled for the @var{RX200} series.
15769 Floating point instructions will only be generated for 32-bit floating
15770 point values however, so if the @option{-m64bit-doubles} option is in
15771 use then the FPU hardware will not be used for doubles.
15773 @emph{Note} If the @option{-fpu} option is enabled then
15774 @option{-funsafe-math-optimizations} is also enabled automatically.
15775 This is because the RX FPU instructions are themselves unsafe.
15777 @item -mcpu=@var{name}
15778 @itemx -patch=@var{name}
15781 Selects the type of RX CPU to be targeted. Currently three types are
15782 supported, the generic @var{RX600} and @var{RX200} series hardware and
15783 the specific @var{RX610} cpu. The default is @var{RX600}.
15785 The only difference between @var{RX600} and @var{RX610} is that the
15786 @var{RX610} does not support the @code{MVTIPL} instruction.
15788 The @var{RX200} series does not have a hardware floating point unit
15789 and so @option{-nofpu} is enabled by default when this type is
15792 @item -mbig-endian-data
15793 @itemx -mlittle-endian-data
15794 @opindex mbig-endian-data
15795 @opindex mlittle-endian-data
15796 Store data (but not code) in the big-endian format. The default is
15797 @option{-mlittle-endian-data}, ie to store data in the little endian
15800 @item -msmall-data-limit=@var{N}
15801 @opindex msmall-data-limit
15802 Specifies the maximum size in bytes of global and static variables
15803 which can be placed into the small data area. Using the small data
15804 area can lead to smaller and faster code, but the size of area is
15805 limited and it is up to the programmer to ensure that the area does
15806 not overflow. Also when the small data area is used one of the RX's
15807 registers (@code{r13}) is reserved for use pointing to this area, so
15808 it is no longer available for use by the compiler. This could result
15809 in slower and/or larger code if variables which once could have been
15810 held in @code{r13} are now pushed onto the stack.
15812 Note, common variables (variables which have not been initialised) and
15813 constants are not placed into the small data area as they are assigned
15814 to other sections in the output executable.
15816 The default value is zero, which disables this feature. Note, this
15817 feature is not enabled by default with higher optimization levels
15818 (@option{-O2} etc) because of the potentially detrimental effects of
15819 reserving register @code{r13}. It is up to the programmer to
15820 experiment and discover whether this feature is of benefit to their
15827 Use the simulator runtime. The default is to use the libgloss board
15830 @item -mas100-syntax
15831 @itemx -mno-as100-syntax
15832 @opindex mas100-syntax
15833 @opindex mno-as100-syntax
15834 When generating assembler output use a syntax that is compatible with
15835 Renesas's AS100 assembler. This syntax can also be handled by the GAS
15836 assembler but it has some restrictions so generating it is not the
15839 @item -mmax-constant-size=@var{N}
15840 @opindex mmax-constant-size
15841 Specifies the maximum size, in bytes, of a constant that can be used as
15842 an operand in a RX instruction. Although the RX instruction set does
15843 allow constants of up to 4 bytes in length to be used in instructions,
15844 a longer value equates to a longer instruction. Thus in some
15845 circumstances it can be beneficial to restrict the size of constants
15846 that are used in instructions. Constants that are too big are instead
15847 placed into a constant pool and referenced via register indirection.
15849 The value @var{N} can be between 0 and 4. A value of 0 (the default)
15850 or 4 means that constants of any size are allowed.
15854 Enable linker relaxation. Linker relaxation is a process whereby the
15855 linker will attempt to reduce the size of a program by finding shorter
15856 versions of various instructions. Disabled by default.
15858 @item -mint-register=@var{N}
15859 @opindex mint-register
15860 Specify the number of registers to reserve for fast interrupt handler
15861 functions. The value @var{N} can be between 0 and 4. A value of 1
15862 means that register @code{r13} will be reserved for the exclusive use
15863 of fast interrupt handlers. A value of 2 reserves @code{r13} and
15864 @code{r12}. A value of 3 reserves @code{r13}, @code{r12} and
15865 @code{r11}, and a value of 4 reserves @code{r13} through @code{r10}.
15866 A value of 0, the default, does not reserve any registers.
15868 @item -msave-acc-in-interrupts
15869 @opindex msave-acc-in-interrupts
15870 Specifies that interrupt handler functions should preserve the
15871 accumulator register. This is only necessary if normal code might use
15872 the accumulator register, for example because it performs 64-bit
15873 multiplications. The default is to ignore the accumulator as this
15874 makes the interrupt handlers faster.
15878 @emph{Note:} The generic GCC command line @option{-ffixed-@var{reg}}
15879 has special significance to the RX port when used with the
15880 @code{interrupt} function attribute. This attribute indicates a
15881 function intended to process fast interrupts. GCC will will ensure
15882 that it only uses the registers @code{r10}, @code{r11}, @code{r12}
15883 and/or @code{r13} and only provided that the normal use of the
15884 corresponding registers have been restricted via the
15885 @option{-ffixed-@var{reg}} or @option{-mint-register} command line
15888 @node S/390 and zSeries Options
15889 @subsection S/390 and zSeries Options
15890 @cindex S/390 and zSeries Options
15892 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
15896 @itemx -msoft-float
15897 @opindex mhard-float
15898 @opindex msoft-float
15899 Use (do not use) the hardware floating-point instructions and registers
15900 for floating-point operations. When @option{-msoft-float} is specified,
15901 functions in @file{libgcc.a} will be used to perform floating-point
15902 operations. When @option{-mhard-float} is specified, the compiler
15903 generates IEEE floating-point instructions. This is the default.
15906 @itemx -mno-hard-dfp
15908 @opindex mno-hard-dfp
15909 Use (do not use) the hardware decimal-floating-point instructions for
15910 decimal-floating-point operations. When @option{-mno-hard-dfp} is
15911 specified, functions in @file{libgcc.a} will be used to perform
15912 decimal-floating-point operations. When @option{-mhard-dfp} is
15913 specified, the compiler generates decimal-floating-point hardware
15914 instructions. This is the default for @option{-march=z9-ec} or higher.
15916 @item -mlong-double-64
15917 @itemx -mlong-double-128
15918 @opindex mlong-double-64
15919 @opindex mlong-double-128
15920 These switches control the size of @code{long double} type. A size
15921 of 64bit makes the @code{long double} type equivalent to the @code{double}
15922 type. This is the default.
15925 @itemx -mno-backchain
15926 @opindex mbackchain
15927 @opindex mno-backchain
15928 Store (do not store) the address of the caller's frame as backchain pointer
15929 into the callee's stack frame.
15930 A backchain may be needed to allow debugging using tools that do not understand
15931 DWARF-2 call frame information.
15932 When @option{-mno-packed-stack} is in effect, the backchain pointer is stored
15933 at the bottom of the stack frame; when @option{-mpacked-stack} is in effect,
15934 the backchain is placed into the topmost word of the 96/160 byte register
15937 In general, code compiled with @option{-mbackchain} is call-compatible with
15938 code compiled with @option{-mmo-backchain}; however, use of the backchain
15939 for debugging purposes usually requires that the whole binary is built with
15940 @option{-mbackchain}. Note that the combination of @option{-mbackchain},
15941 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
15942 to build a linux kernel use @option{-msoft-float}.
15944 The default is to not maintain the backchain.
15946 @item -mpacked-stack
15947 @itemx -mno-packed-stack
15948 @opindex mpacked-stack
15949 @opindex mno-packed-stack
15950 Use (do not use) the packed stack layout. When @option{-mno-packed-stack} is
15951 specified, the compiler uses the all fields of the 96/160 byte register save
15952 area only for their default purpose; unused fields still take up stack space.
15953 When @option{-mpacked-stack} is specified, register save slots are densely
15954 packed at the top of the register save area; unused space is reused for other
15955 purposes, allowing for more efficient use of the available stack space.
15956 However, when @option{-mbackchain} is also in effect, the topmost word of
15957 the save area is always used to store the backchain, and the return address
15958 register is always saved two words below the backchain.
15960 As long as the stack frame backchain is not used, code generated with
15961 @option{-mpacked-stack} is call-compatible with code generated with
15962 @option{-mno-packed-stack}. Note that some non-FSF releases of GCC 2.95 for
15963 S/390 or zSeries generated code that uses the stack frame backchain at run
15964 time, not just for debugging purposes. Such code is not call-compatible
15965 with code compiled with @option{-mpacked-stack}. Also, note that the
15966 combination of @option{-mbackchain},
15967 @option{-mpacked-stack} and @option{-mhard-float} is not supported. In order
15968 to build a linux kernel use @option{-msoft-float}.
15970 The default is to not use the packed stack layout.
15973 @itemx -mno-small-exec
15974 @opindex msmall-exec
15975 @opindex mno-small-exec
15976 Generate (or do not generate) code using the @code{bras} instruction
15977 to do subroutine calls.
15978 This only works reliably if the total executable size does not
15979 exceed 64k. The default is to use the @code{basr} instruction instead,
15980 which does not have this limitation.
15986 When @option{-m31} is specified, generate code compliant to the
15987 GNU/Linux for S/390 ABI@. When @option{-m64} is specified, generate
15988 code compliant to the GNU/Linux for zSeries ABI@. This allows GCC in
15989 particular to generate 64-bit instructions. For the @samp{s390}
15990 targets, the default is @option{-m31}, while the @samp{s390x}
15991 targets default to @option{-m64}.
15997 When @option{-mzarch} is specified, generate code using the
15998 instructions available on z/Architecture.
15999 When @option{-mesa} is specified, generate code using the
16000 instructions available on ESA/390. Note that @option{-mesa} is
16001 not possible with @option{-m64}.
16002 When generating code compliant to the GNU/Linux for S/390 ABI,
16003 the default is @option{-mesa}. When generating code compliant
16004 to the GNU/Linux for zSeries ABI, the default is @option{-mzarch}.
16010 Generate (or do not generate) code using the @code{mvcle} instruction
16011 to perform block moves. When @option{-mno-mvcle} is specified,
16012 use a @code{mvc} loop instead. This is the default unless optimizing for
16019 Print (or do not print) additional debug information when compiling.
16020 The default is to not print debug information.
16022 @item -march=@var{cpu-type}
16024 Generate code that will run on @var{cpu-type}, which is the name of a system
16025 representing a certain processor type. Possible values for
16026 @var{cpu-type} are @samp{g5}, @samp{g6}, @samp{z900}, @samp{z990},
16027 @samp{z9-109}, @samp{z9-ec} and @samp{z10}.
16028 When generating code using the instructions available on z/Architecture,
16029 the default is @option{-march=z900}. Otherwise, the default is
16030 @option{-march=g5}.
16032 @item -mtune=@var{cpu-type}
16034 Tune to @var{cpu-type} everything applicable about the generated code,
16035 except for the ABI and the set of available instructions.
16036 The list of @var{cpu-type} values is the same as for @option{-march}.
16037 The default is the value used for @option{-march}.
16040 @itemx -mno-tpf-trace
16041 @opindex mtpf-trace
16042 @opindex mno-tpf-trace
16043 Generate code that adds (does not add) in TPF OS specific branches to trace
16044 routines in the operating system. This option is off by default, even
16045 when compiling for the TPF OS@.
16048 @itemx -mno-fused-madd
16049 @opindex mfused-madd
16050 @opindex mno-fused-madd
16051 Generate code that uses (does not use) the floating point multiply and
16052 accumulate instructions. These instructions are generated by default if
16053 hardware floating point is used.
16055 @item -mwarn-framesize=@var{framesize}
16056 @opindex mwarn-framesize
16057 Emit a warning if the current function exceeds the given frame size. Because
16058 this is a compile time check it doesn't need to be a real problem when the program
16059 runs. It is intended to identify functions which most probably cause
16060 a stack overflow. It is useful to be used in an environment with limited stack
16061 size e.g.@: the linux kernel.
16063 @item -mwarn-dynamicstack
16064 @opindex mwarn-dynamicstack
16065 Emit a warning if the function calls alloca or uses dynamically
16066 sized arrays. This is generally a bad idea with a limited stack size.
16068 @item -mstack-guard=@var{stack-guard}
16069 @itemx -mstack-size=@var{stack-size}
16070 @opindex mstack-guard
16071 @opindex mstack-size
16072 If these options are provided the s390 back end emits additional instructions in
16073 the function prologue which trigger a trap if the stack size is @var{stack-guard}
16074 bytes above the @var{stack-size} (remember that the stack on s390 grows downward).
16075 If the @var{stack-guard} option is omitted the smallest power of 2 larger than
16076 the frame size of the compiled function is chosen.
16077 These options are intended to be used to help debugging stack overflow problems.
16078 The additionally emitted code causes only little overhead and hence can also be
16079 used in production like systems without greater performance degradation. The given
16080 values have to be exact powers of 2 and @var{stack-size} has to be greater than
16081 @var{stack-guard} without exceeding 64k.
16082 In order to be efficient the extra code makes the assumption that the stack starts
16083 at an address aligned to the value given by @var{stack-size}.
16084 The @var{stack-guard} option can only be used in conjunction with @var{stack-size}.
16087 @node Score Options
16088 @subsection Score Options
16089 @cindex Score Options
16091 These options are defined for Score implementations:
16096 Compile code for big endian mode. This is the default.
16100 Compile code for little endian mode.
16104 Disable generate bcnz instruction.
16108 Enable generate unaligned load and store instruction.
16112 Enable the use of multiply-accumulate instructions. Disabled by default.
16116 Specify the SCORE5 as the target architecture.
16120 Specify the SCORE5U of the target architecture.
16124 Specify the SCORE7 as the target architecture. This is the default.
16128 Specify the SCORE7D as the target architecture.
16132 @subsection SH Options
16134 These @samp{-m} options are defined for the SH implementations:
16139 Generate code for the SH1.
16143 Generate code for the SH2.
16146 Generate code for the SH2e.
16150 Generate code for the SH2a without FPU, or for a SH2a-FPU in such a way
16151 that the floating-point unit is not used.
16153 @item -m2a-single-only
16154 @opindex m2a-single-only
16155 Generate code for the SH2a-FPU, in such a way that no double-precision
16156 floating point operations are used.
16159 @opindex m2a-single
16160 Generate code for the SH2a-FPU assuming the floating-point unit is in
16161 single-precision mode by default.
16165 Generate code for the SH2a-FPU assuming the floating-point unit is in
16166 double-precision mode by default.
16170 Generate code for the SH3.
16174 Generate code for the SH3e.
16178 Generate code for the SH4 without a floating-point unit.
16180 @item -m4-single-only
16181 @opindex m4-single-only
16182 Generate code for the SH4 with a floating-point unit that only
16183 supports single-precision arithmetic.
16187 Generate code for the SH4 assuming the floating-point unit is in
16188 single-precision mode by default.
16192 Generate code for the SH4.
16196 Generate code for the SH4al-dsp, or for a SH4a in such a way that the
16197 floating-point unit is not used.
16199 @item -m4a-single-only
16200 @opindex m4a-single-only
16201 Generate code for the SH4a, in such a way that no double-precision
16202 floating point operations are used.
16205 @opindex m4a-single
16206 Generate code for the SH4a assuming the floating-point unit is in
16207 single-precision mode by default.
16211 Generate code for the SH4a.
16215 Same as @option{-m4a-nofpu}, except that it implicitly passes
16216 @option{-dsp} to the assembler. GCC doesn't generate any DSP
16217 instructions at the moment.
16221 Compile code for the processor in big endian mode.
16225 Compile code for the processor in little endian mode.
16229 Align doubles at 64-bit boundaries. Note that this changes the calling
16230 conventions, and thus some functions from the standard C library will
16231 not work unless you recompile it first with @option{-mdalign}.
16235 Shorten some address references at link time, when possible; uses the
16236 linker option @option{-relax}.
16240 Use 32-bit offsets in @code{switch} tables. The default is to use
16245 Enable the use of bit manipulation instructions on SH2A.
16249 Enable the use of the instruction @code{fmovd}. Check @option{-mdalign} for
16250 alignment constraints.
16254 Comply with the calling conventions defined by Renesas.
16258 Comply with the calling conventions defined by Renesas.
16262 Comply with the calling conventions defined for GCC before the Renesas
16263 conventions were available. This option is the default for all
16264 targets of the SH toolchain except for @samp{sh-symbianelf}.
16267 @opindex mnomacsave
16268 Mark the @code{MAC} register as call-clobbered, even if
16269 @option{-mhitachi} is given.
16273 Increase IEEE-compliance of floating-point code.
16274 At the moment, this is equivalent to @option{-fno-finite-math-only}.
16275 When generating 16 bit SH opcodes, getting IEEE-conforming results for
16276 comparisons of NANs / infinities incurs extra overhead in every
16277 floating point comparison, therefore the default is set to
16278 @option{-ffinite-math-only}.
16280 @item -minline-ic_invalidate
16281 @opindex minline-ic_invalidate
16282 Inline code to invalidate instruction cache entries after setting up
16283 nested function trampolines.
16284 This option has no effect if -musermode is in effect and the selected
16285 code generation option (e.g. -m4) does not allow the use of the icbi
16287 If the selected code generation option does not allow the use of the icbi
16288 instruction, and -musermode is not in effect, the inlined code will
16289 manipulate the instruction cache address array directly with an associative
16290 write. This not only requires privileged mode, but it will also
16291 fail if the cache line had been mapped via the TLB and has become unmapped.
16295 Dump instruction size and location in the assembly code.
16298 @opindex mpadstruct
16299 This option is deprecated. It pads structures to multiple of 4 bytes,
16300 which is incompatible with the SH ABI@.
16304 Optimize for space instead of speed. Implied by @option{-Os}.
16307 @opindex mprefergot
16308 When generating position-independent code, emit function calls using
16309 the Global Offset Table instead of the Procedure Linkage Table.
16313 Don't generate privileged mode only code; implies -mno-inline-ic_invalidate
16314 if the inlined code would not work in user mode.
16315 This is the default when the target is @code{sh-*-linux*}.
16317 @item -multcost=@var{number}
16318 @opindex multcost=@var{number}
16319 Set the cost to assume for a multiply insn.
16321 @item -mdiv=@var{strategy}
16322 @opindex mdiv=@var{strategy}
16323 Set the division strategy to use for SHmedia code. @var{strategy} must be
16324 one of: call, call2, fp, inv, inv:minlat, inv20u, inv20l, inv:call,
16325 inv:call2, inv:fp .
16326 "fp" performs the operation in floating point. This has a very high latency,
16327 but needs only a few instructions, so it might be a good choice if
16328 your code has enough easily exploitable ILP to allow the compiler to
16329 schedule the floating point instructions together with other instructions.
16330 Division by zero causes a floating point exception.
16331 "inv" uses integer operations to calculate the inverse of the divisor,
16332 and then multiplies the dividend with the inverse. This strategy allows
16333 cse and hoisting of the inverse calculation. Division by zero calculates
16334 an unspecified result, but does not trap.
16335 "inv:minlat" is a variant of "inv" where if no cse / hoisting opportunities
16336 have been found, or if the entire operation has been hoisted to the same
16337 place, the last stages of the inverse calculation are intertwined with the
16338 final multiply to reduce the overall latency, at the expense of using a few
16339 more instructions, and thus offering fewer scheduling opportunities with
16341 "call" calls a library function that usually implements the inv:minlat
16343 This gives high code density for m5-*media-nofpu compilations.
16344 "call2" uses a different entry point of the same library function, where it
16345 assumes that a pointer to a lookup table has already been set up, which
16346 exposes the pointer load to cse / code hoisting optimizations.
16347 "inv:call", "inv:call2" and "inv:fp" all use the "inv" algorithm for initial
16348 code generation, but if the code stays unoptimized, revert to the "call",
16349 "call2", or "fp" strategies, respectively. Note that the
16350 potentially-trapping side effect of division by zero is carried by a
16351 separate instruction, so it is possible that all the integer instructions
16352 are hoisted out, but the marker for the side effect stays where it is.
16353 A recombination to fp operations or a call is not possible in that case.
16354 "inv20u" and "inv20l" are variants of the "inv:minlat" strategy. In the case
16355 that the inverse calculation was nor separated from the multiply, they speed
16356 up division where the dividend fits into 20 bits (plus sign where applicable),
16357 by inserting a test to skip a number of operations in this case; this test
16358 slows down the case of larger dividends. inv20u assumes the case of a such
16359 a small dividend to be unlikely, and inv20l assumes it to be likely.
16361 @item -maccumulate-outgoing-args
16362 @opindex maccumulate-outgoing-args
16363 Reserve space once for outgoing arguments in the function prologue rather
16364 than around each call. Generally beneficial for performance and size. Also
16365 needed for unwinding to avoid changing the stack frame around conditional code.
16367 @item -mdivsi3_libfunc=@var{name}
16368 @opindex mdivsi3_libfunc=@var{name}
16369 Set the name of the library function used for 32 bit signed division to
16370 @var{name}. This only affect the name used in the call and inv:call
16371 division strategies, and the compiler will still expect the same
16372 sets of input/output/clobbered registers as if this option was not present.
16374 @item -mfixed-range=@var{register-range}
16375 @opindex mfixed-range
16376 Generate code treating the given register range as fixed registers.
16377 A fixed register is one that the register allocator can not use. This is
16378 useful when compiling kernel code. A register range is specified as
16379 two registers separated by a dash. Multiple register ranges can be
16380 specified separated by a comma.
16382 @item -madjust-unroll
16383 @opindex madjust-unroll
16384 Throttle unrolling to avoid thrashing target registers.
16385 This option only has an effect if the gcc code base supports the
16386 TARGET_ADJUST_UNROLL_MAX target hook.
16388 @item -mindexed-addressing
16389 @opindex mindexed-addressing
16390 Enable the use of the indexed addressing mode for SHmedia32/SHcompact.
16391 This is only safe if the hardware and/or OS implement 32 bit wrap-around
16392 semantics for the indexed addressing mode. The architecture allows the
16393 implementation of processors with 64 bit MMU, which the OS could use to
16394 get 32 bit addressing, but since no current hardware implementation supports
16395 this or any other way to make the indexed addressing mode safe to use in
16396 the 32 bit ABI, the default is -mno-indexed-addressing.
16398 @item -mgettrcost=@var{number}
16399 @opindex mgettrcost=@var{number}
16400 Set the cost assumed for the gettr instruction to @var{number}.
16401 The default is 2 if @option{-mpt-fixed} is in effect, 100 otherwise.
16405 Assume pt* instructions won't trap. This will generally generate better
16406 scheduled code, but is unsafe on current hardware. The current architecture
16407 definition says that ptabs and ptrel trap when the target anded with 3 is 3.
16408 This has the unintentional effect of making it unsafe to schedule ptabs /
16409 ptrel before a branch, or hoist it out of a loop. For example,
16410 __do_global_ctors, a part of libgcc that runs constructors at program
16411 startup, calls functions in a list which is delimited by @minus{}1. With the
16412 -mpt-fixed option, the ptabs will be done before testing against @minus{}1.
16413 That means that all the constructors will be run a bit quicker, but when
16414 the loop comes to the end of the list, the program crashes because ptabs
16415 loads @minus{}1 into a target register. Since this option is unsafe for any
16416 hardware implementing the current architecture specification, the default
16417 is -mno-pt-fixed. Unless the user specifies a specific cost with
16418 @option{-mgettrcost}, -mno-pt-fixed also implies @option{-mgettrcost=100};
16419 this deters register allocation using target registers for storing
16422 @item -minvalid-symbols
16423 @opindex minvalid-symbols
16424 Assume symbols might be invalid. Ordinary function symbols generated by
16425 the compiler will always be valid to load with movi/shori/ptabs or
16426 movi/shori/ptrel, but with assembler and/or linker tricks it is possible
16427 to generate symbols that will cause ptabs / ptrel to trap.
16428 This option is only meaningful when @option{-mno-pt-fixed} is in effect.
16429 It will then prevent cross-basic-block cse, hoisting and most scheduling
16430 of symbol loads. The default is @option{-mno-invalid-symbols}.
16433 @node SPARC Options
16434 @subsection SPARC Options
16435 @cindex SPARC options
16437 These @samp{-m} options are supported on the SPARC:
16440 @item -mno-app-regs
16442 @opindex mno-app-regs
16444 Specify @option{-mapp-regs} to generate output using the global registers
16445 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
16448 To be fully SVR4 ABI compliant at the cost of some performance loss,
16449 specify @option{-mno-app-regs}. You should compile libraries and system
16450 software with this option.
16453 @itemx -mhard-float
16455 @opindex mhard-float
16456 Generate output containing floating point instructions. This is the
16460 @itemx -msoft-float
16462 @opindex msoft-float
16463 Generate output containing library calls for floating point.
16464 @strong{Warning:} the requisite libraries are not available for all SPARC
16465 targets. Normally the facilities of the machine's usual C compiler are
16466 used, but this cannot be done directly in cross-compilation. You must make
16467 your own arrangements to provide suitable library functions for
16468 cross-compilation. The embedded targets @samp{sparc-*-aout} and
16469 @samp{sparclite-*-*} do provide software floating point support.
16471 @option{-msoft-float} changes the calling convention in the output file;
16472 therefore, it is only useful if you compile @emph{all} of a program with
16473 this option. In particular, you need to compile @file{libgcc.a}, the
16474 library that comes with GCC, with @option{-msoft-float} in order for
16477 @item -mhard-quad-float
16478 @opindex mhard-quad-float
16479 Generate output containing quad-word (long double) floating point
16482 @item -msoft-quad-float
16483 @opindex msoft-quad-float
16484 Generate output containing library calls for quad-word (long double)
16485 floating point instructions. The functions called are those specified
16486 in the SPARC ABI@. This is the default.
16488 As of this writing, there are no SPARC implementations that have hardware
16489 support for the quad-word floating point instructions. They all invoke
16490 a trap handler for one of these instructions, and then the trap handler
16491 emulates the effect of the instruction. Because of the trap handler overhead,
16492 this is much slower than calling the ABI library routines. Thus the
16493 @option{-msoft-quad-float} option is the default.
16495 @item -mno-unaligned-doubles
16496 @itemx -munaligned-doubles
16497 @opindex mno-unaligned-doubles
16498 @opindex munaligned-doubles
16499 Assume that doubles have 8 byte alignment. This is the default.
16501 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
16502 alignment only if they are contained in another type, or if they have an
16503 absolute address. Otherwise, it assumes they have 4 byte alignment.
16504 Specifying this option avoids some rare compatibility problems with code
16505 generated by other compilers. It is not the default because it results
16506 in a performance loss, especially for floating point code.
16508 @item -mno-faster-structs
16509 @itemx -mfaster-structs
16510 @opindex mno-faster-structs
16511 @opindex mfaster-structs
16512 With @option{-mfaster-structs}, the compiler assumes that structures
16513 should have 8 byte alignment. This enables the use of pairs of
16514 @code{ldd} and @code{std} instructions for copies in structure
16515 assignment, in place of twice as many @code{ld} and @code{st} pairs.
16516 However, the use of this changed alignment directly violates the SPARC
16517 ABI@. Thus, it's intended only for use on targets where the developer
16518 acknowledges that their resulting code will not be directly in line with
16519 the rules of the ABI@.
16521 @item -mimpure-text
16522 @opindex mimpure-text
16523 @option{-mimpure-text}, used in addition to @option{-shared}, tells
16524 the compiler to not pass @option{-z text} to the linker when linking a
16525 shared object. Using this option, you can link position-dependent
16526 code into a shared object.
16528 @option{-mimpure-text} suppresses the ``relocations remain against
16529 allocatable but non-writable sections'' linker error message.
16530 However, the necessary relocations will trigger copy-on-write, and the
16531 shared object is not actually shared across processes. Instead of
16532 using @option{-mimpure-text}, you should compile all source code with
16533 @option{-fpic} or @option{-fPIC}.
16535 This option is only available on SunOS and Solaris.
16537 @item -mcpu=@var{cpu_type}
16539 Set the instruction set, register set, and instruction scheduling parameters
16540 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
16541 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
16542 @samp{f930}, @samp{f934}, @samp{hypersparc}, @samp{sparclite86x},
16543 @samp{sparclet}, @samp{tsc701}, @samp{v9}, @samp{ultrasparc},
16544 @samp{ultrasparc3}, @samp{niagara} and @samp{niagara2}.
16546 Default instruction scheduling parameters are used for values that select
16547 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
16548 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
16550 Here is a list of each supported architecture and their supported
16555 v8: supersparc, hypersparc
16556 sparclite: f930, f934, sparclite86x
16558 v9: ultrasparc, ultrasparc3, niagara, niagara2
16561 By default (unless configured otherwise), GCC generates code for the V7
16562 variant of the SPARC architecture. With @option{-mcpu=cypress}, the compiler
16563 additionally optimizes it for the Cypress CY7C602 chip, as used in the
16564 SPARCStation/SPARCServer 3xx series. This is also appropriate for the older
16565 SPARCStation 1, 2, IPX etc.
16567 With @option{-mcpu=v8}, GCC generates code for the V8 variant of the SPARC
16568 architecture. The only difference from V7 code is that the compiler emits
16569 the integer multiply and integer divide instructions which exist in SPARC-V8
16570 but not in SPARC-V7. With @option{-mcpu=supersparc}, the compiler additionally
16571 optimizes it for the SuperSPARC chip, as used in the SPARCStation 10, 1000 and
16574 With @option{-mcpu=sparclite}, GCC generates code for the SPARClite variant of
16575 the SPARC architecture. This adds the integer multiply, integer divide step
16576 and scan (@code{ffs}) instructions which exist in SPARClite but not in SPARC-V7.
16577 With @option{-mcpu=f930}, the compiler additionally optimizes it for the
16578 Fujitsu MB86930 chip, which is the original SPARClite, with no FPU@. With
16579 @option{-mcpu=f934}, the compiler additionally optimizes it for the Fujitsu
16580 MB86934 chip, which is the more recent SPARClite with FPU@.
16582 With @option{-mcpu=sparclet}, GCC generates code for the SPARClet variant of
16583 the SPARC architecture. This adds the integer multiply, multiply/accumulate,
16584 integer divide step and scan (@code{ffs}) instructions which exist in SPARClet
16585 but not in SPARC-V7. With @option{-mcpu=tsc701}, the compiler additionally
16586 optimizes it for the TEMIC SPARClet chip.
16588 With @option{-mcpu=v9}, GCC generates code for the V9 variant of the SPARC
16589 architecture. This adds 64-bit integer and floating-point move instructions,
16590 3 additional floating-point condition code registers and conditional move
16591 instructions. With @option{-mcpu=ultrasparc}, the compiler additionally
16592 optimizes it for the Sun UltraSPARC I/II/IIi chips. With
16593 @option{-mcpu=ultrasparc3}, the compiler additionally optimizes it for the
16594 Sun UltraSPARC III/III+/IIIi/IIIi+/IV/IV+ chips. With
16595 @option{-mcpu=niagara}, the compiler additionally optimizes it for
16596 Sun UltraSPARC T1 chips. With @option{-mcpu=niagara2}, the compiler
16597 additionally optimizes it for Sun UltraSPARC T2 chips.
16599 @item -mtune=@var{cpu_type}
16601 Set the instruction scheduling parameters for machine type
16602 @var{cpu_type}, but do not set the instruction set or register set that the
16603 option @option{-mcpu=@var{cpu_type}} would.
16605 The same values for @option{-mcpu=@var{cpu_type}} can be used for
16606 @option{-mtune=@var{cpu_type}}, but the only useful values are those
16607 that select a particular cpu implementation. Those are @samp{cypress},
16608 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
16609 @samp{sparclite86x}, @samp{tsc701}, @samp{ultrasparc},
16610 @samp{ultrasparc3}, @samp{niagara}, and @samp{niagara2}.
16615 @opindex mno-v8plus
16616 With @option{-mv8plus}, GCC generates code for the SPARC-V8+ ABI@. The
16617 difference from the V8 ABI is that the global and out registers are
16618 considered 64-bit wide. This is enabled by default on Solaris in 32-bit
16619 mode for all SPARC-V9 processors.
16625 With @option{-mvis}, GCC generates code that takes advantage of the UltraSPARC
16626 Visual Instruction Set extensions. The default is @option{-mno-vis}.
16629 These @samp{-m} options are supported in addition to the above
16630 on SPARC-V9 processors in 64-bit environments:
16633 @item -mlittle-endian
16634 @opindex mlittle-endian
16635 Generate code for a processor running in little-endian mode. It is only
16636 available for a few configurations and most notably not on Solaris and Linux.
16642 Generate code for a 32-bit or 64-bit environment.
16643 The 32-bit environment sets int, long and pointer to 32 bits.
16644 The 64-bit environment sets int to 32 bits and long and pointer
16647 @item -mcmodel=medlow
16648 @opindex mcmodel=medlow
16649 Generate code for the Medium/Low code model: 64-bit addresses, programs
16650 must be linked in the low 32 bits of memory. Programs can be statically
16651 or dynamically linked.
16653 @item -mcmodel=medmid
16654 @opindex mcmodel=medmid
16655 Generate code for the Medium/Middle code model: 64-bit addresses, programs
16656 must be linked in the low 44 bits of memory, the text and data segments must
16657 be less than 2GB in size and the data segment must be located within 2GB of
16660 @item -mcmodel=medany
16661 @opindex mcmodel=medany
16662 Generate code for the Medium/Anywhere code model: 64-bit addresses, programs
16663 may be linked anywhere in memory, the text and data segments must be less
16664 than 2GB in size and the data segment must be located within 2GB of the
16667 @item -mcmodel=embmedany
16668 @opindex mcmodel=embmedany
16669 Generate code for the Medium/Anywhere code model for embedded systems:
16670 64-bit addresses, the text and data segments must be less than 2GB in
16671 size, both starting anywhere in memory (determined at link time). The
16672 global register %g4 points to the base of the data segment. Programs
16673 are statically linked and PIC is not supported.
16676 @itemx -mno-stack-bias
16677 @opindex mstack-bias
16678 @opindex mno-stack-bias
16679 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
16680 frame pointer if present, are offset by @minus{}2047 which must be added back
16681 when making stack frame references. This is the default in 64-bit mode.
16682 Otherwise, assume no such offset is present.
16685 These switches are supported in addition to the above on Solaris:
16690 Add support for multithreading using the Solaris threads library. This
16691 option sets flags for both the preprocessor and linker. This option does
16692 not affect the thread safety of object code produced by the compiler or
16693 that of libraries supplied with it.
16697 Add support for multithreading using the POSIX threads library. This
16698 option sets flags for both the preprocessor and linker. This option does
16699 not affect the thread safety of object code produced by the compiler or
16700 that of libraries supplied with it.
16704 This is a synonym for @option{-pthreads}.
16708 @subsection SPU Options
16709 @cindex SPU options
16711 These @samp{-m} options are supported on the SPU:
16715 @itemx -merror-reloc
16716 @opindex mwarn-reloc
16717 @opindex merror-reloc
16719 The loader for SPU does not handle dynamic relocations. By default, GCC
16720 will give an error when it generates code that requires a dynamic
16721 relocation. @option{-mno-error-reloc} disables the error,
16722 @option{-mwarn-reloc} will generate a warning instead.
16725 @itemx -munsafe-dma
16727 @opindex munsafe-dma
16729 Instructions which initiate or test completion of DMA must not be
16730 reordered with respect to loads and stores of the memory which is being
16731 accessed. Users typically address this problem using the volatile
16732 keyword, but that can lead to inefficient code in places where the
16733 memory is known to not change. Rather than mark the memory as volatile
16734 we treat the DMA instructions as potentially effecting all memory. With
16735 @option{-munsafe-dma} users must use the volatile keyword to protect
16738 @item -mbranch-hints
16739 @opindex mbranch-hints
16741 By default, GCC will generate a branch hint instruction to avoid
16742 pipeline stalls for always taken or probably taken branches. A hint
16743 will not be generated closer than 8 instructions away from its branch.
16744 There is little reason to disable them, except for debugging purposes,
16745 or to make an object a little bit smaller.
16749 @opindex msmall-mem
16750 @opindex mlarge-mem
16752 By default, GCC generates code assuming that addresses are never larger
16753 than 18 bits. With @option{-mlarge-mem} code is generated that assumes
16754 a full 32 bit address.
16759 By default, GCC links against startup code that assumes the SPU-style
16760 main function interface (which has an unconventional parameter list).
16761 With @option{-mstdmain}, GCC will link your program against startup
16762 code that assumes a C99-style interface to @code{main}, including a
16763 local copy of @code{argv} strings.
16765 @item -mfixed-range=@var{register-range}
16766 @opindex mfixed-range
16767 Generate code treating the given register range as fixed registers.
16768 A fixed register is one that the register allocator can not use. This is
16769 useful when compiling kernel code. A register range is specified as
16770 two registers separated by a dash. Multiple register ranges can be
16771 specified separated by a comma.
16777 Compile code assuming that pointers to the PPU address space accessed
16778 via the @code{__ea} named address space qualifier are either 32 or 64
16779 bits wide. The default is 32 bits. As this is an ABI changing option,
16780 all object code in an executable must be compiled with the same setting.
16782 @item -maddress-space-conversion
16783 @itemx -mno-address-space-conversion
16784 @opindex maddress-space-conversion
16785 @opindex mno-address-space-conversion
16786 Allow/disallow treating the @code{__ea} address space as superset
16787 of the generic address space. This enables explicit type casts
16788 between @code{__ea} and generic pointer as well as implicit
16789 conversions of generic pointers to @code{__ea} pointers. The
16790 default is to allow address space pointer conversions.
16792 @item -mcache-size=@var{cache-size}
16793 @opindex mcache-size
16794 This option controls the version of libgcc that the compiler links to an
16795 executable and selects a software-managed cache for accessing variables
16796 in the @code{__ea} address space with a particular cache size. Possible
16797 options for @var{cache-size} are @samp{8}, @samp{16}, @samp{32}, @samp{64}
16798 and @samp{128}. The default cache size is 64KB.
16800 @item -matomic-updates
16801 @itemx -mno-atomic-updates
16802 @opindex matomic-updates
16803 @opindex mno-atomic-updates
16804 This option controls the version of libgcc that the compiler links to an
16805 executable and selects whether atomic updates to the software-managed
16806 cache of PPU-side variables are used. If you use atomic updates, changes
16807 to a PPU variable from SPU code using the @code{__ea} named address space
16808 qualifier will not interfere with changes to other PPU variables residing
16809 in the same cache line from PPU code. If you do not use atomic updates,
16810 such interference may occur; however, writing back cache lines will be
16811 more efficient. The default behavior is to use atomic updates.
16814 @itemx -mdual-nops=@var{n}
16815 @opindex mdual-nops
16816 By default, GCC will insert nops to increase dual issue when it expects
16817 it to increase performance. @var{n} can be a value from 0 to 10. A
16818 smaller @var{n} will insert fewer nops. 10 is the default, 0 is the
16819 same as @option{-mno-dual-nops}. Disabled with @option{-Os}.
16821 @item -mhint-max-nops=@var{n}
16822 @opindex mhint-max-nops
16823 Maximum number of nops to insert for a branch hint. A branch hint must
16824 be at least 8 instructions away from the branch it is effecting. GCC
16825 will insert up to @var{n} nops to enforce this, otherwise it will not
16826 generate the branch hint.
16828 @item -mhint-max-distance=@var{n}
16829 @opindex mhint-max-distance
16830 The encoding of the branch hint instruction limits the hint to be within
16831 256 instructions of the branch it is effecting. By default, GCC makes
16832 sure it is within 125.
16835 @opindex msafe-hints
16836 Work around a hardware bug which causes the SPU to stall indefinitely.
16837 By default, GCC will insert the @code{hbrp} instruction to make sure
16838 this stall won't happen.
16842 @node System V Options
16843 @subsection Options for System V
16845 These additional options are available on System V Release 4 for
16846 compatibility with other compilers on those systems:
16851 Create a shared object.
16852 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
16856 Identify the versions of each tool used by the compiler, in a
16857 @code{.ident} assembler directive in the output.
16861 Refrain from adding @code{.ident} directives to the output file (this is
16864 @item -YP,@var{dirs}
16866 Search the directories @var{dirs}, and no others, for libraries
16867 specified with @option{-l}.
16869 @item -Ym,@var{dir}
16871 Look in the directory @var{dir} to find the M4 preprocessor.
16872 The assembler uses this option.
16873 @c This is supposed to go with a -Yd for predefined M4 macro files, but
16874 @c the generic assembler that comes with Solaris takes just -Ym.
16878 @subsection V850 Options
16879 @cindex V850 Options
16881 These @samp{-m} options are defined for V850 implementations:
16885 @itemx -mno-long-calls
16886 @opindex mlong-calls
16887 @opindex mno-long-calls
16888 Treat all calls as being far away (near). If calls are assumed to be
16889 far away, the compiler will always load the functions address up into a
16890 register, and call indirect through the pointer.
16896 Do not optimize (do optimize) basic blocks that use the same index
16897 pointer 4 or more times to copy pointer into the @code{ep} register, and
16898 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
16899 option is on by default if you optimize.
16901 @item -mno-prolog-function
16902 @itemx -mprolog-function
16903 @opindex mno-prolog-function
16904 @opindex mprolog-function
16905 Do not use (do use) external functions to save and restore registers
16906 at the prologue and epilogue of a function. The external functions
16907 are slower, but use less code space if more than one function saves
16908 the same number of registers. The @option{-mprolog-function} option
16909 is on by default if you optimize.
16913 Try to make the code as small as possible. At present, this just turns
16914 on the @option{-mep} and @option{-mprolog-function} options.
16916 @item -mtda=@var{n}
16918 Put static or global variables whose size is @var{n} bytes or less into
16919 the tiny data area that register @code{ep} points to. The tiny data
16920 area can hold up to 256 bytes in total (128 bytes for byte references).
16922 @item -msda=@var{n}
16924 Put static or global variables whose size is @var{n} bytes or less into
16925 the small data area that register @code{gp} points to. The small data
16926 area can hold up to 64 kilobytes.
16928 @item -mzda=@var{n}
16930 Put static or global variables whose size is @var{n} bytes or less into
16931 the first 32 kilobytes of memory.
16935 Specify that the target processor is the V850.
16938 @opindex mbig-switch
16939 Generate code suitable for big switch tables. Use this option only if
16940 the assembler/linker complain about out of range branches within a switch
16945 This option will cause r2 and r5 to be used in the code generated by
16946 the compiler. This setting is the default.
16948 @item -mno-app-regs
16949 @opindex mno-app-regs
16950 This option will cause r2 and r5 to be treated as fixed registers.
16954 Specify that the target processor is the V850E1. The preprocessor
16955 constants @samp{__v850e1__} and @samp{__v850e__} will be defined if
16956 this option is used.
16960 Specify that the target processor is the V850E@. The preprocessor
16961 constant @samp{__v850e__} will be defined if this option is used.
16963 If neither @option{-mv850} nor @option{-mv850e} nor @option{-mv850e1}
16964 are defined then a default target processor will be chosen and the
16965 relevant @samp{__v850*__} preprocessor constant will be defined.
16967 The preprocessor constants @samp{__v850} and @samp{__v851__} are always
16968 defined, regardless of which processor variant is the target.
16970 @item -mdisable-callt
16971 @opindex mdisable-callt
16972 This option will suppress generation of the CALLT instruction for the
16973 v850e and v850e1 flavors of the v850 architecture. The default is
16974 @option{-mno-disable-callt} which allows the CALLT instruction to be used.
16979 @subsection VAX Options
16980 @cindex VAX options
16982 These @samp{-m} options are defined for the VAX:
16987 Do not output certain jump instructions (@code{aobleq} and so on)
16988 that the Unix assembler for the VAX cannot handle across long
16993 Do output those jump instructions, on the assumption that you
16994 will assemble with the GNU assembler.
16998 Output code for g-format floating point numbers instead of d-format.
17001 @node VxWorks Options
17002 @subsection VxWorks Options
17003 @cindex VxWorks Options
17005 The options in this section are defined for all VxWorks targets.
17006 Options specific to the target hardware are listed with the other
17007 options for that target.
17012 GCC can generate code for both VxWorks kernels and real time processes
17013 (RTPs). This option switches from the former to the latter. It also
17014 defines the preprocessor macro @code{__RTP__}.
17017 @opindex non-static
17018 Link an RTP executable against shared libraries rather than static
17019 libraries. The options @option{-static} and @option{-shared} can
17020 also be used for RTPs (@pxref{Link Options}); @option{-static}
17027 These options are passed down to the linker. They are defined for
17028 compatibility with Diab.
17031 @opindex Xbind-lazy
17032 Enable lazy binding of function calls. This option is equivalent to
17033 @option{-Wl,-z,now} and is defined for compatibility with Diab.
17037 Disable lazy binding of function calls. This option is the default and
17038 is defined for compatibility with Diab.
17041 @node x86-64 Options
17042 @subsection x86-64 Options
17043 @cindex x86-64 options
17045 These are listed under @xref{i386 and x86-64 Options}.
17047 @node i386 and x86-64 Windows Options
17048 @subsection i386 and x86-64 Windows Options
17049 @cindex i386 and x86-64 Windows Options
17051 These additional options are available for Windows targets:
17056 This option is available for Cygwin and MinGW targets. It
17057 specifies that a console application is to be generated, by
17058 instructing the linker to set the PE header subsystem type
17059 required for console applications.
17060 This is the default behavior for Cygwin and MinGW targets.
17064 This option is available for Cygwin targets. It specifies that
17065 the Cygwin internal interface is to be used for predefined
17066 preprocessor macros, C runtime libraries and related linker
17067 paths and options. For Cygwin targets this is the default behavior.
17068 This option is deprecated and will be removed in a future release.
17071 @opindex mno-cygwin
17072 This option is available for Cygwin targets. It specifies that
17073 the MinGW internal interface is to be used instead of Cygwin's, by
17074 setting MinGW-related predefined macros and linker paths and default
17076 This option is deprecated and will be removed in a future release.
17080 This option is available for Cygwin and MinGW targets. It
17081 specifies that a DLL - a dynamic link library - is to be
17082 generated, enabling the selection of the required runtime
17083 startup object and entry point.
17085 @item -mnop-fun-dllimport
17086 @opindex mnop-fun-dllimport
17087 This option is available for Cygwin and MinGW targets. It
17088 specifies that the dllimport attribute should be ignored.
17092 This option is available for MinGW targets. It specifies
17093 that MinGW-specific thread support is to be used.
17097 This option is available for mingw-w64 targets. It specifies
17098 that the UNICODE macro is getting pre-defined and that the
17099 unicode capable runtime startup code is chosen.
17103 This option is available for Cygwin and MinGW targets. It
17104 specifies that the typical Windows pre-defined macros are to
17105 be set in the pre-processor, but does not influence the choice
17106 of runtime library/startup code.
17110 This option is available for Cygwin and MinGW targets. It
17111 specifies that a GUI application is to be generated by
17112 instructing the linker to set the PE header subsystem type
17115 @item -fno-set-stack-executable
17116 @opindex fno-set-stack-executable
17117 This option is available for MinGW targets. It specifies that
17118 the executable flag for stack used by nested functions isn't
17119 set. This is necessary for binaries running in kernel mode of
17120 Windows, as there the user32 API, which is used to set executable
17121 privileges, isn't available.
17123 @item -mpe-aligned-commons
17124 @opindex mpe-aligned-commons
17125 This option is available for Cygwin and MinGW targets. It
17126 specifies that the GNU extension to the PE file format that
17127 permits the correct alignment of COMMON variables should be
17128 used when generating code. It will be enabled by default if
17129 GCC detects that the target assembler found during configuration
17130 supports the feature.
17133 See also under @ref{i386 and x86-64 Options} for standard options.
17135 @node Xstormy16 Options
17136 @subsection Xstormy16 Options
17137 @cindex Xstormy16 Options
17139 These options are defined for Xstormy16:
17144 Choose startup files and linker script suitable for the simulator.
17147 @node Xtensa Options
17148 @subsection Xtensa Options
17149 @cindex Xtensa Options
17151 These options are supported for Xtensa targets:
17155 @itemx -mno-const16
17157 @opindex mno-const16
17158 Enable or disable use of @code{CONST16} instructions for loading
17159 constant values. The @code{CONST16} instruction is currently not a
17160 standard option from Tensilica. When enabled, @code{CONST16}
17161 instructions are always used in place of the standard @code{L32R}
17162 instructions. The use of @code{CONST16} is enabled by default only if
17163 the @code{L32R} instruction is not available.
17166 @itemx -mno-fused-madd
17167 @opindex mfused-madd
17168 @opindex mno-fused-madd
17169 Enable or disable use of fused multiply/add and multiply/subtract
17170 instructions in the floating-point option. This has no effect if the
17171 floating-point option is not also enabled. Disabling fused multiply/add
17172 and multiply/subtract instructions forces the compiler to use separate
17173 instructions for the multiply and add/subtract operations. This may be
17174 desirable in some cases where strict IEEE 754-compliant results are
17175 required: the fused multiply add/subtract instructions do not round the
17176 intermediate result, thereby producing results with @emph{more} bits of
17177 precision than specified by the IEEE standard. Disabling fused multiply
17178 add/subtract instructions also ensures that the program output is not
17179 sensitive to the compiler's ability to combine multiply and add/subtract
17182 @item -mserialize-volatile
17183 @itemx -mno-serialize-volatile
17184 @opindex mserialize-volatile
17185 @opindex mno-serialize-volatile
17186 When this option is enabled, GCC inserts @code{MEMW} instructions before
17187 @code{volatile} memory references to guarantee sequential consistency.
17188 The default is @option{-mserialize-volatile}. Use
17189 @option{-mno-serialize-volatile} to omit the @code{MEMW} instructions.
17191 @item -mforce-no-pic
17192 @opindex mforce-no-pic
17193 For targets, like GNU/Linux, where all user-mode Xtensa code must be
17194 position-independent code (PIC), this option disables PIC for compiling
17197 @item -mtext-section-literals
17198 @itemx -mno-text-section-literals
17199 @opindex mtext-section-literals
17200 @opindex mno-text-section-literals
17201 Control the treatment of literal pools. The default is
17202 @option{-mno-text-section-literals}, which places literals in a separate
17203 section in the output file. This allows the literal pool to be placed
17204 in a data RAM/ROM, and it also allows the linker to combine literal
17205 pools from separate object files to remove redundant literals and
17206 improve code size. With @option{-mtext-section-literals}, the literals
17207 are interspersed in the text section in order to keep them as close as
17208 possible to their references. This may be necessary for large assembly
17211 @item -mtarget-align
17212 @itemx -mno-target-align
17213 @opindex mtarget-align
17214 @opindex mno-target-align
17215 When this option is enabled, GCC instructs the assembler to
17216 automatically align instructions to reduce branch penalties at the
17217 expense of some code density. The assembler attempts to widen density
17218 instructions to align branch targets and the instructions following call
17219 instructions. If there are not enough preceding safe density
17220 instructions to align a target, no widening will be performed. The
17221 default is @option{-mtarget-align}. These options do not affect the
17222 treatment of auto-aligned instructions like @code{LOOP}, which the
17223 assembler will always align, either by widening density instructions or
17224 by inserting no-op instructions.
17227 @itemx -mno-longcalls
17228 @opindex mlongcalls
17229 @opindex mno-longcalls
17230 When this option is enabled, GCC instructs the assembler to translate
17231 direct calls to indirect calls unless it can determine that the target
17232 of a direct call is in the range allowed by the call instruction. This
17233 translation typically occurs for calls to functions in other source
17234 files. Specifically, the assembler translates a direct @code{CALL}
17235 instruction into an @code{L32R} followed by a @code{CALLX} instruction.
17236 The default is @option{-mno-longcalls}. This option should be used in
17237 programs where the call target can potentially be out of range. This
17238 option is implemented in the assembler, not the compiler, so the
17239 assembly code generated by GCC will still show direct call
17240 instructions---look at the disassembled object code to see the actual
17241 instructions. Note that the assembler will use an indirect call for
17242 every cross-file call, not just those that really will be out of range.
17245 @node zSeries Options
17246 @subsection zSeries Options
17247 @cindex zSeries options
17249 These are listed under @xref{S/390 and zSeries Options}.
17251 @node Code Gen Options
17252 @section Options for Code Generation Conventions
17253 @cindex code generation conventions
17254 @cindex options, code generation
17255 @cindex run-time options
17257 These machine-independent options control the interface conventions
17258 used in code generation.
17260 Most of them have both positive and negative forms; the negative form
17261 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
17262 one of the forms is listed---the one which is not the default. You
17263 can figure out the other form by either removing @samp{no-} or adding
17267 @item -fbounds-check
17268 @opindex fbounds-check
17269 For front-ends that support it, generate additional code to check that
17270 indices used to access arrays are within the declared range. This is
17271 currently only supported by the Java and Fortran front-ends, where
17272 this option defaults to true and false respectively.
17276 This option generates traps for signed overflow on addition, subtraction,
17277 multiplication operations.
17281 This option instructs the compiler to assume that signed arithmetic
17282 overflow of addition, subtraction and multiplication wraps around
17283 using twos-complement representation. This flag enables some optimizations
17284 and disables others. This option is enabled by default for the Java
17285 front-end, as required by the Java language specification.
17288 @opindex fexceptions
17289 Enable exception handling. Generates extra code needed to propagate
17290 exceptions. For some targets, this implies GCC will generate frame
17291 unwind information for all functions, which can produce significant data
17292 size overhead, although it does not affect execution. If you do not
17293 specify this option, GCC will enable it by default for languages like
17294 C++ which normally require exception handling, and disable it for
17295 languages like C that do not normally require it. However, you may need
17296 to enable this option when compiling C code that needs to interoperate
17297 properly with exception handlers written in C++. You may also wish to
17298 disable this option if you are compiling older C++ programs that don't
17299 use exception handling.
17301 @item -fnon-call-exceptions
17302 @opindex fnon-call-exceptions
17303 Generate code that allows trapping instructions to throw exceptions.
17304 Note that this requires platform-specific runtime support that does
17305 not exist everywhere. Moreover, it only allows @emph{trapping}
17306 instructions to throw exceptions, i.e.@: memory references or floating
17307 point instructions. It does not allow exceptions to be thrown from
17308 arbitrary signal handlers such as @code{SIGALRM}.
17310 @item -funwind-tables
17311 @opindex funwind-tables
17312 Similar to @option{-fexceptions}, except that it will just generate any needed
17313 static data, but will not affect the generated code in any other way.
17314 You will normally not enable this option; instead, a language processor
17315 that needs this handling would enable it on your behalf.
17317 @item -fasynchronous-unwind-tables
17318 @opindex fasynchronous-unwind-tables
17319 Generate unwind table in dwarf2 format, if supported by target machine. The
17320 table is exact at each instruction boundary, so it can be used for stack
17321 unwinding from asynchronous events (such as debugger or garbage collector).
17323 @item -fpcc-struct-return
17324 @opindex fpcc-struct-return
17325 Return ``short'' @code{struct} and @code{union} values in memory like
17326 longer ones, rather than in registers. This convention is less
17327 efficient, but it has the advantage of allowing intercallability between
17328 GCC-compiled files and files compiled with other compilers, particularly
17329 the Portable C Compiler (pcc).
17331 The precise convention for returning structures in memory depends
17332 on the target configuration macros.
17334 Short structures and unions are those whose size and alignment match
17335 that of some integer type.
17337 @strong{Warning:} code compiled with the @option{-fpcc-struct-return}
17338 switch is not binary compatible with code compiled with the
17339 @option{-freg-struct-return} switch.
17340 Use it to conform to a non-default application binary interface.
17342 @item -freg-struct-return
17343 @opindex freg-struct-return
17344 Return @code{struct} and @code{union} values in registers when possible.
17345 This is more efficient for small structures than
17346 @option{-fpcc-struct-return}.
17348 If you specify neither @option{-fpcc-struct-return} nor
17349 @option{-freg-struct-return}, GCC defaults to whichever convention is
17350 standard for the target. If there is no standard convention, GCC
17351 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
17352 the principal compiler. In those cases, we can choose the standard, and
17353 we chose the more efficient register return alternative.
17355 @strong{Warning:} code compiled with the @option{-freg-struct-return}
17356 switch is not binary compatible with code compiled with the
17357 @option{-fpcc-struct-return} switch.
17358 Use it to conform to a non-default application binary interface.
17360 @item -fshort-enums
17361 @opindex fshort-enums
17362 Allocate to an @code{enum} type only as many bytes as it needs for the
17363 declared range of possible values. Specifically, the @code{enum} type
17364 will be equivalent to the smallest integer type which has enough room.
17366 @strong{Warning:} the @option{-fshort-enums} switch causes GCC to generate
17367 code that is not binary compatible with code generated without that switch.
17368 Use it to conform to a non-default application binary interface.
17370 @item -fshort-double
17371 @opindex fshort-double
17372 Use the same size for @code{double} as for @code{float}.
17374 @strong{Warning:} the @option{-fshort-double} switch causes GCC to generate
17375 code that is not binary compatible with code generated without that switch.
17376 Use it to conform to a non-default application binary interface.
17378 @item -fshort-wchar
17379 @opindex fshort-wchar
17380 Override the underlying type for @samp{wchar_t} to be @samp{short
17381 unsigned int} instead of the default for the target. This option is
17382 useful for building programs to run under WINE@.
17384 @strong{Warning:} the @option{-fshort-wchar} switch causes GCC to generate
17385 code that is not binary compatible with code generated without that switch.
17386 Use it to conform to a non-default application binary interface.
17389 @opindex fno-common
17390 In C code, controls the placement of uninitialized global variables.
17391 Unix C compilers have traditionally permitted multiple definitions of
17392 such variables in different compilation units by placing the variables
17394 This is the behavior specified by @option{-fcommon}, and is the default
17395 for GCC on most targets.
17396 On the other hand, this behavior is not required by ISO C, and on some
17397 targets may carry a speed or code size penalty on variable references.
17398 The @option{-fno-common} option specifies that the compiler should place
17399 uninitialized global variables in the data section of the object file,
17400 rather than generating them as common blocks.
17401 This has the effect that if the same variable is declared
17402 (without @code{extern}) in two different compilations,
17403 you will get a multiple-definition error when you link them.
17404 In this case, you must compile with @option{-fcommon} instead.
17405 Compiling with @option{-fno-common} is useful on targets for which
17406 it provides better performance, or if you wish to verify that the
17407 program will work on other systems which always treat uninitialized
17408 variable declarations this way.
17412 Ignore the @samp{#ident} directive.
17414 @item -finhibit-size-directive
17415 @opindex finhibit-size-directive
17416 Don't output a @code{.size} assembler directive, or anything else that
17417 would cause trouble if the function is split in the middle, and the
17418 two halves are placed at locations far apart in memory. This option is
17419 used when compiling @file{crtstuff.c}; you should not need to use it
17422 @item -fverbose-asm
17423 @opindex fverbose-asm
17424 Put extra commentary information in the generated assembly code to
17425 make it more readable. This option is generally only of use to those
17426 who actually need to read the generated assembly code (perhaps while
17427 debugging the compiler itself).
17429 @option{-fno-verbose-asm}, the default, causes the
17430 extra information to be omitted and is useful when comparing two assembler
17433 @item -frecord-gcc-switches
17434 @opindex frecord-gcc-switches
17435 This switch causes the command line that was used to invoke the
17436 compiler to be recorded into the object file that is being created.
17437 This switch is only implemented on some targets and the exact format
17438 of the recording is target and binary file format dependent, but it
17439 usually takes the form of a section containing ASCII text. This
17440 switch is related to the @option{-fverbose-asm} switch, but that
17441 switch only records information in the assembler output file as
17442 comments, so it never reaches the object file.
17446 @cindex global offset table
17448 Generate position-independent code (PIC) suitable for use in a shared
17449 library, if supported for the target machine. Such code accesses all
17450 constant addresses through a global offset table (GOT)@. The dynamic
17451 loader resolves the GOT entries when the program starts (the dynamic
17452 loader is not part of GCC; it is part of the operating system). If
17453 the GOT size for the linked executable exceeds a machine-specific
17454 maximum size, you get an error message from the linker indicating that
17455 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
17456 instead. (These maximums are 8k on the SPARC and 32k
17457 on the m68k and RS/6000. The 386 has no such limit.)
17459 Position-independent code requires special support, and therefore works
17460 only on certain machines. For the 386, GCC supports PIC for System V
17461 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
17462 position-independent.
17464 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
17469 If supported for the target machine, emit position-independent code,
17470 suitable for dynamic linking and avoiding any limit on the size of the
17471 global offset table. This option makes a difference on the m68k,
17472 PowerPC and SPARC@.
17474 Position-independent code requires special support, and therefore works
17475 only on certain machines.
17477 When this flag is set, the macros @code{__pic__} and @code{__PIC__}
17484 These options are similar to @option{-fpic} and @option{-fPIC}, but
17485 generated position independent code can be only linked into executables.
17486 Usually these options are used when @option{-pie} GCC option will be
17487 used during linking.
17489 @option{-fpie} and @option{-fPIE} both define the macros
17490 @code{__pie__} and @code{__PIE__}. The macros have the value 1
17491 for @option{-fpie} and 2 for @option{-fPIE}.
17493 @item -fno-jump-tables
17494 @opindex fno-jump-tables
17495 Do not use jump tables for switch statements even where it would be
17496 more efficient than other code generation strategies. This option is
17497 of use in conjunction with @option{-fpic} or @option{-fPIC} for
17498 building code which forms part of a dynamic linker and cannot
17499 reference the address of a jump table. On some targets, jump tables
17500 do not require a GOT and this option is not needed.
17502 @item -ffixed-@var{reg}
17504 Treat the register named @var{reg} as a fixed register; generated code
17505 should never refer to it (except perhaps as a stack pointer, frame
17506 pointer or in some other fixed role).
17508 @var{reg} must be the name of a register. The register names accepted
17509 are machine-specific and are defined in the @code{REGISTER_NAMES}
17510 macro in the machine description macro file.
17512 This flag does not have a negative form, because it specifies a
17515 @item -fcall-used-@var{reg}
17516 @opindex fcall-used
17517 Treat the register named @var{reg} as an allocable register that is
17518 clobbered by function calls. It may be allocated for temporaries or
17519 variables that do not live across a call. Functions compiled this way
17520 will not save and restore the register @var{reg}.
17522 It is an error to used this flag with the frame pointer or stack pointer.
17523 Use of this flag for other registers that have fixed pervasive roles in
17524 the machine's execution model will produce disastrous results.
17526 This flag does not have a negative form, because it specifies a
17529 @item -fcall-saved-@var{reg}
17530 @opindex fcall-saved
17531 Treat the register named @var{reg} as an allocable register saved by
17532 functions. It may be allocated even for temporaries or variables that
17533 live across a call. Functions compiled this way will save and restore
17534 the register @var{reg} if they use it.
17536 It is an error to used this flag with the frame pointer or stack pointer.
17537 Use of this flag for other registers that have fixed pervasive roles in
17538 the machine's execution model will produce disastrous results.
17540 A different sort of disaster will result from the use of this flag for
17541 a register in which function values may be returned.
17543 This flag does not have a negative form, because it specifies a
17546 @item -fpack-struct[=@var{n}]
17547 @opindex fpack-struct
17548 Without a value specified, pack all structure members together without
17549 holes. When a value is specified (which must be a small power of two), pack
17550 structure members according to this value, representing the maximum
17551 alignment (that is, objects with default alignment requirements larger than
17552 this will be output potentially unaligned at the next fitting location.
17554 @strong{Warning:} the @option{-fpack-struct} switch causes GCC to generate
17555 code that is not binary compatible with code generated without that switch.
17556 Additionally, it makes the code suboptimal.
17557 Use it to conform to a non-default application binary interface.
17559 @item -finstrument-functions
17560 @opindex finstrument-functions
17561 Generate instrumentation calls for entry and exit to functions. Just
17562 after function entry and just before function exit, the following
17563 profiling functions will be called with the address of the current
17564 function and its call site. (On some platforms,
17565 @code{__builtin_return_address} does not work beyond the current
17566 function, so the call site information may not be available to the
17567 profiling functions otherwise.)
17570 void __cyg_profile_func_enter (void *this_fn,
17572 void __cyg_profile_func_exit (void *this_fn,
17576 The first argument is the address of the start of the current function,
17577 which may be looked up exactly in the symbol table.
17579 This instrumentation is also done for functions expanded inline in other
17580 functions. The profiling calls will indicate where, conceptually, the
17581 inline function is entered and exited. This means that addressable
17582 versions of such functions must be available. If all your uses of a
17583 function are expanded inline, this may mean an additional expansion of
17584 code size. If you use @samp{extern inline} in your C code, an
17585 addressable version of such functions must be provided. (This is
17586 normally the case anyways, but if you get lucky and the optimizer always
17587 expands the functions inline, you might have gotten away without
17588 providing static copies.)
17590 A function may be given the attribute @code{no_instrument_function}, in
17591 which case this instrumentation will not be done. This can be used, for
17592 example, for the profiling functions listed above, high-priority
17593 interrupt routines, and any functions from which the profiling functions
17594 cannot safely be called (perhaps signal handlers, if the profiling
17595 routines generate output or allocate memory).
17597 @item -finstrument-functions-exclude-file-list=@var{file},@var{file},@dots{}
17598 @opindex finstrument-functions-exclude-file-list
17600 Set the list of functions that are excluded from instrumentation (see
17601 the description of @code{-finstrument-functions}). If the file that
17602 contains a function definition matches with one of @var{file}, then
17603 that function is not instrumented. The match is done on substrings:
17604 if the @var{file} parameter is a substring of the file name, it is
17605 considered to be a match.
17608 @code{-finstrument-functions-exclude-file-list=/bits/stl,include/sys}
17609 will exclude any inline function defined in files whose pathnames
17610 contain @code{/bits/stl} or @code{include/sys}.
17612 If, for some reason, you want to include letter @code{','} in one of
17613 @var{sym}, write @code{'\,'}. For example,
17614 @code{-finstrument-functions-exclude-file-list='\,\,tmp'}
17615 (note the single quote surrounding the option).
17617 @item -finstrument-functions-exclude-function-list=@var{sym},@var{sym},@dots{}
17618 @opindex finstrument-functions-exclude-function-list
17620 This is similar to @code{-finstrument-functions-exclude-file-list},
17621 but this option sets the list of function names to be excluded from
17622 instrumentation. The function name to be matched is its user-visible
17623 name, such as @code{vector<int> blah(const vector<int> &)}, not the
17624 internal mangled name (e.g., @code{_Z4blahRSt6vectorIiSaIiEE}). The
17625 match is done on substrings: if the @var{sym} parameter is a substring
17626 of the function name, it is considered to be a match. For C99 and C++
17627 extended identifiers, the function name must be given in UTF-8, not
17628 using universal character names.
17630 @item -fstack-check
17631 @opindex fstack-check
17632 Generate code to verify that you do not go beyond the boundary of the
17633 stack. You should specify this flag if you are running in an
17634 environment with multiple threads, but only rarely need to specify it in
17635 a single-threaded environment since stack overflow is automatically
17636 detected on nearly all systems if there is only one stack.
17638 Note that this switch does not actually cause checking to be done; the
17639 operating system or the language runtime must do that. The switch causes
17640 generation of code to ensure that they see the stack being extended.
17642 You can additionally specify a string parameter: @code{no} means no
17643 checking, @code{generic} means force the use of old-style checking,
17644 @code{specific} means use the best checking method and is equivalent
17645 to bare @option{-fstack-check}.
17647 Old-style checking is a generic mechanism that requires no specific
17648 target support in the compiler but comes with the following drawbacks:
17652 Modified allocation strategy for large objects: they will always be
17653 allocated dynamically if their size exceeds a fixed threshold.
17656 Fixed limit on the size of the static frame of functions: when it is
17657 topped by a particular function, stack checking is not reliable and
17658 a warning is issued by the compiler.
17661 Inefficiency: because of both the modified allocation strategy and the
17662 generic implementation, the performances of the code are hampered.
17665 Note that old-style stack checking is also the fallback method for
17666 @code{specific} if no target support has been added in the compiler.
17668 @item -fstack-limit-register=@var{reg}
17669 @itemx -fstack-limit-symbol=@var{sym}
17670 @itemx -fno-stack-limit
17671 @opindex fstack-limit-register
17672 @opindex fstack-limit-symbol
17673 @opindex fno-stack-limit
17674 Generate code to ensure that the stack does not grow beyond a certain value,
17675 either the value of a register or the address of a symbol. If the stack
17676 would grow beyond the value, a signal is raised. For most targets,
17677 the signal is raised before the stack overruns the boundary, so
17678 it is possible to catch the signal without taking special precautions.
17680 For instance, if the stack starts at absolute address @samp{0x80000000}
17681 and grows downwards, you can use the flags
17682 @option{-fstack-limit-symbol=__stack_limit} and
17683 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
17684 of 128KB@. Note that this may only work with the GNU linker.
17686 @item -fleading-underscore
17687 @opindex fleading-underscore
17688 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
17689 change the way C symbols are represented in the object file. One use
17690 is to help link with legacy assembly code.
17692 @strong{Warning:} the @option{-fleading-underscore} switch causes GCC to
17693 generate code that is not binary compatible with code generated without that
17694 switch. Use it to conform to a non-default application binary interface.
17695 Not all targets provide complete support for this switch.
17697 @item -ftls-model=@var{model}
17698 @opindex ftls-model
17699 Alter the thread-local storage model to be used (@pxref{Thread-Local}).
17700 The @var{model} argument should be one of @code{global-dynamic},
17701 @code{local-dynamic}, @code{initial-exec} or @code{local-exec}.
17703 The default without @option{-fpic} is @code{initial-exec}; with
17704 @option{-fpic} the default is @code{global-dynamic}.
17706 @item -fvisibility=@var{default|internal|hidden|protected}
17707 @opindex fvisibility
17708 Set the default ELF image symbol visibility to the specified option---all
17709 symbols will be marked with this unless overridden within the code.
17710 Using this feature can very substantially improve linking and
17711 load times of shared object libraries, produce more optimized
17712 code, provide near-perfect API export and prevent symbol clashes.
17713 It is @strong{strongly} recommended that you use this in any shared objects
17716 Despite the nomenclature, @code{default} always means public ie;
17717 available to be linked against from outside the shared object.
17718 @code{protected} and @code{internal} are pretty useless in real-world
17719 usage so the only other commonly used option will be @code{hidden}.
17720 The default if @option{-fvisibility} isn't specified is
17721 @code{default}, i.e., make every
17722 symbol public---this causes the same behavior as previous versions of
17725 A good explanation of the benefits offered by ensuring ELF
17726 symbols have the correct visibility is given by ``How To Write
17727 Shared Libraries'' by Ulrich Drepper (which can be found at
17728 @w{@uref{http://people.redhat.com/~drepper/}})---however a superior
17729 solution made possible by this option to marking things hidden when
17730 the default is public is to make the default hidden and mark things
17731 public. This is the norm with DLL's on Windows and with @option{-fvisibility=hidden}
17732 and @code{__attribute__ ((visibility("default")))} instead of
17733 @code{__declspec(dllexport)} you get almost identical semantics with
17734 identical syntax. This is a great boon to those working with
17735 cross-platform projects.
17737 For those adding visibility support to existing code, you may find
17738 @samp{#pragma GCC visibility} of use. This works by you enclosing
17739 the declarations you wish to set visibility for with (for example)
17740 @samp{#pragma GCC visibility push(hidden)} and
17741 @samp{#pragma GCC visibility pop}.
17742 Bear in mind that symbol visibility should be viewed @strong{as
17743 part of the API interface contract} and thus all new code should
17744 always specify visibility when it is not the default ie; declarations
17745 only for use within the local DSO should @strong{always} be marked explicitly
17746 as hidden as so to avoid PLT indirection overheads---making this
17747 abundantly clear also aids readability and self-documentation of the code.
17748 Note that due to ISO C++ specification requirements, operator new and
17749 operator delete must always be of default visibility.
17751 Be aware that headers from outside your project, in particular system
17752 headers and headers from any other library you use, may not be
17753 expecting to be compiled with visibility other than the default. You
17754 may need to explicitly say @samp{#pragma GCC visibility push(default)}
17755 before including any such headers.
17757 @samp{extern} declarations are not affected by @samp{-fvisibility}, so
17758 a lot of code can be recompiled with @samp{-fvisibility=hidden} with
17759 no modifications. However, this means that calls to @samp{extern}
17760 functions with no explicit visibility will use the PLT, so it is more
17761 effective to use @samp{__attribute ((visibility))} and/or
17762 @samp{#pragma GCC visibility} to tell the compiler which @samp{extern}
17763 declarations should be treated as hidden.
17765 Note that @samp{-fvisibility} does affect C++ vague linkage
17766 entities. This means that, for instance, an exception class that will
17767 be thrown between DSOs must be explicitly marked with default
17768 visibility so that the @samp{type_info} nodes will be unified between
17771 An overview of these techniques, their benefits and how to use them
17772 is at @w{@uref{http://gcc.gnu.org/wiki/Visibility}}.
17774 @item -fstrict-volatile-bitfields
17775 This option should be used if accesses to volatile bitfields (or other
17776 structure fields, although the compiler usually honors those types
17777 anyway) should use a single access in a mode of the same size as the
17778 container's type, aligned to a natural alignment if possible. For
17779 example, targets with memory-mapped peripheral registers might require
17780 all such accesses to be 16 bits wide; with this flag the user could
17781 declare all peripheral bitfields as ``unsigned short'' (assuming short
17782 is 16 bits on these targets) to force GCC to use 16 bit accesses
17783 instead of, perhaps, a more efficient 32 bit access.
17785 If this option is disabled, the compiler will use the most efficient
17786 instruction. In the previous example, that might be a 32-bit load
17787 instruction, even though that will access bytes that do not contain
17788 any portion of the bitfield, or memory-mapped registers unrelated to
17789 the one being updated.
17791 If the target requires strict alignment, and honoring the container
17792 type would require violating this alignment, a warning is issued.
17793 However, the access happens as the user requested, under the
17794 assumption that the user knows something about the target hardware
17795 that GCC is unaware of.
17797 The default value of this option is determined by the application binary
17798 interface for the target processor.
17804 @node Environment Variables
17805 @section Environment Variables Affecting GCC
17806 @cindex environment variables
17808 @c man begin ENVIRONMENT
17809 This section describes several environment variables that affect how GCC
17810 operates. Some of them work by specifying directories or prefixes to use
17811 when searching for various kinds of files. Some are used to specify other
17812 aspects of the compilation environment.
17814 Note that you can also specify places to search using options such as
17815 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
17816 take precedence over places specified using environment variables, which
17817 in turn take precedence over those specified by the configuration of GCC@.
17818 @xref{Driver,, Controlling the Compilation Driver @file{gcc}, gccint,
17819 GNU Compiler Collection (GCC) Internals}.
17824 @c @itemx LC_COLLATE
17826 @c @itemx LC_MONETARY
17827 @c @itemx LC_NUMERIC
17832 @c @findex LC_COLLATE
17833 @findex LC_MESSAGES
17834 @c @findex LC_MONETARY
17835 @c @findex LC_NUMERIC
17839 These environment variables control the way that GCC uses
17840 localization information that allow GCC to work with different
17841 national conventions. GCC inspects the locale categories
17842 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
17843 so. These locale categories can be set to any value supported by your
17844 installation. A typical value is @samp{en_GB.UTF-8} for English in the United
17845 Kingdom encoded in UTF-8.
17847 The @env{LC_CTYPE} environment variable specifies character
17848 classification. GCC uses it to determine the character boundaries in
17849 a string; this is needed for some multibyte encodings that contain quote
17850 and escape characters that would otherwise be interpreted as a string
17853 The @env{LC_MESSAGES} environment variable specifies the language to
17854 use in diagnostic messages.
17856 If the @env{LC_ALL} environment variable is set, it overrides the value
17857 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
17858 and @env{LC_MESSAGES} default to the value of the @env{LANG}
17859 environment variable. If none of these variables are set, GCC
17860 defaults to traditional C English behavior.
17864 If @env{TMPDIR} is set, it specifies the directory to use for temporary
17865 files. GCC uses temporary files to hold the output of one stage of
17866 compilation which is to be used as input to the next stage: for example,
17867 the output of the preprocessor, which is the input to the compiler
17870 @item GCC_EXEC_PREFIX
17871 @findex GCC_EXEC_PREFIX
17872 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
17873 names of the subprograms executed by the compiler. No slash is added
17874 when this prefix is combined with the name of a subprogram, but you can
17875 specify a prefix that ends with a slash if you wish.
17877 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
17878 an appropriate prefix to use based on the pathname it was invoked with.
17880 If GCC cannot find the subprogram using the specified prefix, it
17881 tries looking in the usual places for the subprogram.
17883 The default value of @env{GCC_EXEC_PREFIX} is
17884 @file{@var{prefix}/lib/gcc/} where @var{prefix} is the prefix to
17885 the installed compiler. In many cases @var{prefix} is the value
17886 of @code{prefix} when you ran the @file{configure} script.
17888 Other prefixes specified with @option{-B} take precedence over this prefix.
17890 This prefix is also used for finding files such as @file{crt0.o} that are
17893 In addition, the prefix is used in an unusual way in finding the
17894 directories to search for header files. For each of the standard
17895 directories whose name normally begins with @samp{/usr/local/lib/gcc}
17896 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
17897 replacing that beginning with the specified prefix to produce an
17898 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
17899 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
17900 These alternate directories are searched first; the standard directories
17901 come next. If a standard directory begins with the configured
17902 @var{prefix} then the value of @var{prefix} is replaced by
17903 @env{GCC_EXEC_PREFIX} when looking for header files.
17905 @item COMPILER_PATH
17906 @findex COMPILER_PATH
17907 The value of @env{COMPILER_PATH} is a colon-separated list of
17908 directories, much like @env{PATH}. GCC tries the directories thus
17909 specified when searching for subprograms, if it can't find the
17910 subprograms using @env{GCC_EXEC_PREFIX}.
17913 @findex LIBRARY_PATH
17914 The value of @env{LIBRARY_PATH} is a colon-separated list of
17915 directories, much like @env{PATH}. When configured as a native compiler,
17916 GCC tries the directories thus specified when searching for special
17917 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
17918 using GCC also uses these directories when searching for ordinary
17919 libraries for the @option{-l} option (but directories specified with
17920 @option{-L} come first).
17924 @cindex locale definition
17925 This variable is used to pass locale information to the compiler. One way in
17926 which this information is used is to determine the character set to be used
17927 when character literals, string literals and comments are parsed in C and C++.
17928 When the compiler is configured to allow multibyte characters,
17929 the following values for @env{LANG} are recognized:
17933 Recognize JIS characters.
17935 Recognize SJIS characters.
17937 Recognize EUCJP characters.
17940 If @env{LANG} is not defined, or if it has some other value, then the
17941 compiler will use mblen and mbtowc as defined by the default locale to
17942 recognize and translate multibyte characters.
17946 Some additional environments variables affect the behavior of the
17949 @include cppenv.texi
17953 @node Precompiled Headers
17954 @section Using Precompiled Headers
17955 @cindex precompiled headers
17956 @cindex speed of compilation
17958 Often large projects have many header files that are included in every
17959 source file. The time the compiler takes to process these header files
17960 over and over again can account for nearly all of the time required to
17961 build the project. To make builds faster, GCC allows users to
17962 `precompile' a header file; then, if builds can use the precompiled
17963 header file they will be much faster.
17965 To create a precompiled header file, simply compile it as you would any
17966 other file, if necessary using the @option{-x} option to make the driver
17967 treat it as a C or C++ header file. You will probably want to use a
17968 tool like @command{make} to keep the precompiled header up-to-date when
17969 the headers it contains change.
17971 A precompiled header file will be searched for when @code{#include} is
17972 seen in the compilation. As it searches for the included file
17973 (@pxref{Search Path,,Search Path,cpp,The C Preprocessor}) the
17974 compiler looks for a precompiled header in each directory just before it
17975 looks for the include file in that directory. The name searched for is
17976 the name specified in the @code{#include} with @samp{.gch} appended. If
17977 the precompiled header file can't be used, it is ignored.
17979 For instance, if you have @code{#include "all.h"}, and you have
17980 @file{all.h.gch} in the same directory as @file{all.h}, then the
17981 precompiled header file will be used if possible, and the original
17982 header will be used otherwise.
17984 Alternatively, you might decide to put the precompiled header file in a
17985 directory and use @option{-I} to ensure that directory is searched
17986 before (or instead of) the directory containing the original header.
17987 Then, if you want to check that the precompiled header file is always
17988 used, you can put a file of the same name as the original header in this
17989 directory containing an @code{#error} command.
17991 This also works with @option{-include}. So yet another way to use
17992 precompiled headers, good for projects not designed with precompiled
17993 header files in mind, is to simply take most of the header files used by
17994 a project, include them from another header file, precompile that header
17995 file, and @option{-include} the precompiled header. If the header files
17996 have guards against multiple inclusion, they will be skipped because
17997 they've already been included (in the precompiled header).
17999 If you need to precompile the same header file for different
18000 languages, targets, or compiler options, you can instead make a
18001 @emph{directory} named like @file{all.h.gch}, and put each precompiled
18002 header in the directory, perhaps using @option{-o}. It doesn't matter
18003 what you call the files in the directory, every precompiled header in
18004 the directory will be considered. The first precompiled header
18005 encountered in the directory that is valid for this compilation will
18006 be used; they're searched in no particular order.
18008 There are many other possibilities, limited only by your imagination,
18009 good sense, and the constraints of your build system.
18011 A precompiled header file can be used only when these conditions apply:
18015 Only one precompiled header can be used in a particular compilation.
18018 A precompiled header can't be used once the first C token is seen. You
18019 can have preprocessor directives before a precompiled header; you can
18020 even include a precompiled header from inside another header, so long as
18021 there are no C tokens before the @code{#include}.
18024 The precompiled header file must be produced for the same language as
18025 the current compilation. You can't use a C precompiled header for a C++
18029 The precompiled header file must have been produced by the same compiler
18030 binary as the current compilation is using.
18033 Any macros defined before the precompiled header is included must
18034 either be defined in the same way as when the precompiled header was
18035 generated, or must not affect the precompiled header, which usually
18036 means that they don't appear in the precompiled header at all.
18038 The @option{-D} option is one way to define a macro before a
18039 precompiled header is included; using a @code{#define} can also do it.
18040 There are also some options that define macros implicitly, like
18041 @option{-O} and @option{-Wdeprecated}; the same rule applies to macros
18044 @item If debugging information is output when using the precompiled
18045 header, using @option{-g} or similar, the same kind of debugging information
18046 must have been output when building the precompiled header. However,
18047 a precompiled header built using @option{-g} can be used in a compilation
18048 when no debugging information is being output.
18050 @item The same @option{-m} options must generally be used when building
18051 and using the precompiled header. @xref{Submodel Options},
18052 for any cases where this rule is relaxed.
18054 @item Each of the following options must be the same when building and using
18055 the precompiled header:
18057 @gccoptlist{-fexceptions}
18060 Some other command-line options starting with @option{-f},
18061 @option{-p}, or @option{-O} must be defined in the same way as when
18062 the precompiled header was generated. At present, it's not clear
18063 which options are safe to change and which are not; the safest choice
18064 is to use exactly the same options when generating and using the
18065 precompiled header. The following are known to be safe:
18067 @gccoptlist{-fmessage-length= -fpreprocessed -fsched-interblock @gol
18068 -fsched-spec -fsched-spec-load -fsched-spec-load-dangerous @gol
18069 -fsched-verbose=<number> -fschedule-insns -fvisibility= @gol
18074 For all of these except the last, the compiler will automatically
18075 ignore the precompiled header if the conditions aren't met. If you
18076 find an option combination that doesn't work and doesn't cause the
18077 precompiled header to be ignored, please consider filing a bug report,
18080 If you do use differing options when generating and using the
18081 precompiled header, the actual behavior will be a mixture of the
18082 behavior for the options. For instance, if you use @option{-g} to
18083 generate the precompiled header but not when using it, you may or may
18084 not get debugging information for routines in the precompiled header.